Experimental Innovations in Surface Science: John T. Yates JR
Experimental Innovations in Surface Science: John T. Yates JR
Experimental Innovations in Surface Science: John T. Yates JR
Experimental
Innovations in
Surface Science
A Guide to Practical Laboratory
Methods and Instruments
Second Edition
Experimental Innovations in Surface Science
John T. Yates Jr.
Experimental Innovations
in Surface Science
A Guide to Practical Laboratory Methods
and Instruments
Second Edition
123
John T. Yates Jr.
Department of Chemistry
University of Virginia
Charlottesville, VA
USA
v
Preface to the First Edition
The modern era of surface science, developed in the last 30 years, features hundreds
of new physical and chemical measurement methods such as low-energy electron
diffraction and Auger electron spectroscopy, which form a vital foundation for the
field. These measurement methods constitute the basis for a number of books used
by beginning students as their introduction to experimental research.
This book is not about this particular experimental core of the field. Instead, it is
about those practical laboratory methods used throughout the field, but seldom
communicated coherently in writing and drawings to beginning students. These
practical methods of design, construction, and measurement also form their own
foundation for making possible significant research. They constitute a gold mine of
clever ideas and “tricks of the trade” about how to use basic materials and com-
ponents effectively for research and measurement in surface science.
Approximately 250 practical methods are described in the book in enough detail
that one could design a new experimental device from the book alone.
This book will be useful to beginning research workers in surface science as well
as to those who work with similar methods in fields such as semiconductor tech-
nology, materials science, and physical electronics.
The contents of the book have been selected from unpublished information
supplied by more than 100 workers in the field as well as from the literature, dating
back to the mid-1960s. I have attempted to communicate the core ideas from
published and unpublished accounts of research in easy-to-comprehend pictorial
form, believing that pictures are the most effective mode for communication of
information of this type. Thus, the reader can learn much just by looking at the
pictures. The text associated with each drawing gives more details. In many cases,
the text and its citations indicate the originators and subsequent contributors to the
innovation, but in other cases the origin of the key ideas is unknown, making
accurate historical citations impossible.
There are some topics that have not been addressed in the book, primarily
because they are currently highly specialized and fast-moving activities that are best
learned by advanced research students who have previously mastered some of the
topics covered in this book. Untreated topics include many details of the scanning
vii
viii Preface to the First Edition
ix
x Contents
3 Electrical Connections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.1 Tungsten Ribbon—Attachment to Power Leads . . . . . . . . . . 61
3.2 Wrap Connection—Tungsten to Tungsten . . . . . . . . . . . . . . 62
3.3 Cold-Formed Wire Connector . . . . . . . . . . . . . . . . . . . . . . . 64
3.4 Ultrahigh Resistance Vacuum Feedthroughs . . . . . . . . . . . . . 64
3.5 Sliding Metal Electrical Contact . . . . . . . . . . . . . . . . . . . . . 66
3.6 Low-Profile Electrical Lead Clamp . . . . . . . . . . . . . . . . . . . 67
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5 Bakeout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.1 Bakeout of Metal Ultrahigh Vacuum Systems. . . . . . . . . . . . 83
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Contents xi
8 Gasket Seals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
8.1 Gasket Seals for Ultrahigh Vacuum Systems . . . . . . . . . . . . 99
8.2 Cryogenic Gasket Seals . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.3 Copper Gasket Removal Devices . . . . . . . . . . . . . . . . . . . . 102
8.4 Unconventional Compression Seals for Ultrahigh Vacuum . . . 103
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
11 Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
11.1 Hemispherical Grids—Formation and Piercing . . . . . . . . . . . 119
11.2 Making Flat Mesh Grids of Large Size . . . . . . . . . . . . . . . . 120
11.3 Grid Fabrication Techniques—Conical Grid . . . . . . . . . . . . . 121
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
18 Spot-Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
18.1 Spot-Welding Difficult Junctions. . . . . . . . . . . . . . . . . . . . . 175
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
39 Ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
39.1 Solid State Cesium Ion Gun Source. . . . . . . . . . . . . . . . . . . 471
39.2 Alkaline Earth Metal Ion Sources . . . . . . . . . . . . . . . . . . . . 473
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
48 Adsorption/Desorption—Thermal . . . . . . . . . . . . . . . . . . . . . . . . 569
48.1 Adsorption and Desorption from High-Area Solids . . . . . . . . 569
48.2 Thermal Desorption from High Area Materials . . . . . . . . . . . 571
48.3 Microcalorimetric Studies of Adsorption Heat on Powders . . . 574
48.4 Chemical Vapor Synthesis of Oxide Nanoparticles . . . . . . . . 576
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
Part IX Safety
Appendix A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643
Part I
Vacuum System Technology
Chapter 1
Motion in Vacuum
This design [1] is shown in two parts to separate the azimuthal motion (Fig. 1.1)
device from the tilt motion device (Fig. 1.2). A central rod extends from a typical
x, y, z-plus rotary motion manipulator. The azimuthal and tilt motion (pushing and
pulling) devices are connected to the same central rod and are free to slide up and
down on this rod. The central rod is slotted to keep the crystal mount linking fittings
from rotating with respect to the rod. The two push-and-pull actuators permit the
central rod that passes through them to rotate in two ball-bearing (BB) mounts.
Azimuthal crystal motion is transmitted by the push-and-pull actuator, which,
through two coupled universal joints, turns the crystal mounting platform on its
own ball bearing mount behind. Tilt motion is transmitted by the push-and-pull
actuator, which tilts the crystal mounting platform about the axis of the hexagonal
nut shown in the Tilt Motion diagram. The degrees of freedom for motion at various
points are shown by double-headed arrows. This design does not result in a
decoupling of tilt motion from other degrees of freedom, which is a disadvantage in
some cases, but the design has been used for more than ten years with success.
A more recent design uses a circular slot to guide a pin in the crystal tilt mechanism
that eliminates the crystal translational motion associated with the tilt motion. It is
of interest to note that in this design, ordinary unlubricated ball bearings are used in
ultrahigh vacuum with success and without seizure or binding.
The crystal is mounted on electrical and thermal isolation standoffs on the crystal
mounting platform and may be heated by electron bombardment or by ohmic
heating. Various thermocouples may be connected to the crystal, and cooling to
about 40 K is achieved using a copper braid connected between the crystal
mounting assembly and a liquid He cryostat. Appropriate provisions are made to
shield the mounting assembly from emitted radiation generated during crystal
heating so that the desorption of impurities from the mounting is minimized.
reduced by using a heater tape around the brass housing and possibly a fan to
evaporate liquid water runoff. The apparatus can be inefficiently used with the
gaseous discharge from a liquid He Dewar to achieve temperatures near 30 K, and
the momentary transfer of liquid He brings the temperature to 15 K [14, 16].
Another rotary seal based on this principle has been reported [17].
Figure 1.4 shows a second design for a rotary seal in which O-rings are
employed as the sealing elements, with one stage of differential pumping between
the O-rings [18]. A polished rotary-seal tube rotates against the slightly compressed
8 1 Motion in Vacuum
Viton O-rings, which are held in place by tightening the flange on the compression
pieces, which are made of copper and which slide in between the outer wall and the
rotary seal. These two compression pieces contain a number of holes to provide a
pumping opportunity between the inside and outside of the cylindrical pieces. The
O-rings are lubricated with a trace of graphite powder. A different lubricant reported
to be satisfactory for Viton O-rings consists of a mixture of graphite powder and
Santovac 5 diffusion pump oil (a polyfluo-roether) [19]. Differential pumping with a
rotary pump or a small ion pump is used for the inner section holding the lower
compression piece. A Teflon spacer at the top supplies a bearing surface for the
rotary seal. Inside the rotary seal, a cryostat tube similar to that in Fig. 1.3 may be
inserted. A pressure rise of about 3 × 10−11 Torr was observed on rotation [18].
1.2 Rotation in Ultrahigh Vacuum 9
Fig. 1.5 Differentially pumped rotary seal—spring-loaded Teflon seals spring-loaded Teflon
pumps the space above the inner seal and the center seal. The system is baked to
200 °C, and the pressure rise upon rotation is less than 10−10 Torr and usually
unobservable. The pressure at the small ion pump is in the 10−8 Torr range. The
estimated gas leakage under static conditions is 10−13 Torr L/s [20]. Seals based on
this principle are now commercially available. Others have reported the use of this
type of seal for rotary manipulators [22–24].
Sometimes rapid rotation in ultrahigh vacuum is desired, where rather massive
components must turn in bearings. Figure 1.6 shows a magnetically driven rotation
device that drives an inner cylinder suspended in vacuum. Two vacuum ball
bearings are used at either end of the inner cylinder. These are ordinary ball
bearings that are unlubricated, and they contain large cross section ball races. These
bearings are routinely baked at 220 °C, and proved satisfactory for 18 months,
having undergone 5 × 106 rotations without any degradation in performance. The
evolution of gases is proportional to the rotation speed as shown in the lower part of
Fig. 1.6, and the rise in total pressure is about 2 × 10−11 mbar at 100 rpm. Most of
the gas evolution was due to H2O and H2 liberation [25].
Other designs for achieving independent rotational motions around two axes are
described in the literature [26–28]. Reference [26] describes an electrical readout
for the angular motions accurate to 0.5° and based on motion of a wire-wound
potentiometer.
The ability to position a single crystal precisely can be useful in many measurement
procedures in ultrahigh vacuum. This is a problem when the positioning must be
done at accuracies beyond the capability of micrometer drives on manipulators.
Factors such as thermal expansion effects, backlash in precision screw drives, and
vibration interfere with accurate positioning methods.
The topic of capacitive positioning of a single crystal is treated in the context of
locating the crystal accurately in front of a mass spectrometer that is housed in an
apertured shield for accurate and reproducible temperature programmed desorption
measurements [29]. This connects to the topic on p. 259.
Figure 1.7a shows a glass shield [29, 30] mounted over a quadrupole mass
spectrometer ionization source. A Ta ring is rigidly supported near the aperture of
the shield and acts as the capacitive pickup terminal. The capacitance is of the order
of 0.1 pF for a 1-mm spacing between the Ta ring and the crystal. The capacitance
measuring circuit, to be described below, can measure capacitance changes in the
range of 0.1 %, meaning that the crystal can be positioned to 1 μm reproducibility,
corresponding to an error in the measurement of the desorption rate from the crystal
of only about 0.05 %. Small tilting of the crystal will reduce this reproducibility,
and the demonstrated reproducibility of actual measurements was better than 1 %
[30].
Figure 1.7b shows a slight variation on the design. Here, an Ag ring is employed
as the capacitance pickup terminal, and it is fixed on the end of the glass shield as
shown. The Ag plate, (12 mm OD; 6 mm ID, 1 mm thick) is held in place by use of
3 Ta wires that are spotwelded to the plate and wound around three glass eyelets on
the shield. The Ag plate is connected through a Kapton-insulated and shielded lead
to a BNC connector that exits the vacuum system. To avoid static charging, the
glass shield is surrounded by a very thin grounded metal mesh [31]. An Aquadag
[32] (colloidal graphite) coating can also be used for this purpose, or a metal shield
like that shown on p. 260 may be used. The spacing employed between the crystal
12 1 Motion in Vacuum
Fig. 1.7 Crystal positioning using capacitance. a Ring capacitance. b Affixed capacitance.
c Capacitance measurement—block diagram
1.3 Crystal Positioning Using Capacitance 13
and the aperture is reported to be only 10 μm in this design, and this maybe
reproduced to 1 μm [31].
The capacitance measuring circuit [30, 31] is shown in schematic block diagram
form in Fig. 1.7c. It is a phase-shifted bridge circuit, working at 20 kHz [30].
A sinusoidal signal (20 Vpp) is applied to the crystal and the capacitive pickup is
made to the ring aperture, since the crystal is often connected to unshielded leads
and electronics outside of the vacuum system. The detection system is used as a
simple lock-in amplifier.
A rapid linear motion inside an ultrahigh vacuum system may be achieved by using
a solenoid-power-driven actuator like that shown in Fig. 1.9. All moving parts are
inside the vacuum system, and the external solenoids do not require vacuum
encapsulation. The actuator is a barrel-shaped plunger made of high-purity iron
sized to make a loose fit inside the stainless steel tube. In this device a stroke of
3 cm was achieved. Each solenoid consists of 1400 turns of #24 magnet wire
wound on an aluminum spool and giving a dc resistance of 14 Ω. Short electrical
pulses are used on either solenoid to move the actuator in or out; the current is then
turned off to prevent overheating or stray magnetic fields. In operation, the actuator
will tilt as shown in the drawing when it moves inward. Measurements made on the
actuator show that a torque of 500 g cm can be achieved at the internal pivot point.
The closing time was measured to be 50 ms. The original paper [34] also provides a
design for either an automatic or manual actuator circuit.
The solenoids are removed for system bakeout by releasing the lock ring on the
back end. A similar magnetically driven shutter assembly using a Nd Fe-B mag-
netic encapsulated slug and an external ceramic-ring magnet has also been
described [35].
A homemade linear motion device may be constructed quite easily from an inex-
pensive machinist’s device as shown in Fig. 1.10. The vise is made of high-grade
gray iron and the jaws as received from the manufacturer are precision ground for
squareness and parallelism to ±0.0001 cm per cm. The hardened jaw pieces are
16 1 Motion in Vacuum
removed from the vise and replaced with aluminum clamps that have an inner lip
machined in them to prevent the Conflat flanges from slipping through. These
clamps are mounted on the vise and then tightened around the flanges holding the
welded bellows. Machinist’s vises giving up to 7.5 cm linear travel are available
[36] at a cost of about $200–$300, and this device, while heavy, gives smooth and
repeatable translation. The precise position can be determined using a vernier
caliper between the clamps, or repeatable positions can be established by using
simple spacer bars. The overall price is about a quarter that of a comparable
commercial bellows manipulator [37].
Figure 1.11 shows a design [38] for translating a cylindrical mirror analyzer
(CMA) Auger spectrometer over a 14-cm distance. This is done using a l/2-in.-
diameter linear ball bearing [39] that slides on a polished stainless support rod at the
top, and is guided along a 1/4-in.-diameter rod-linear bearing assembly on the
bottom. The use of the smaller rod on the bottom allows the system to have a little
flexibility to prevent seizure due to slight misalignment. The first upper support
bushing is clamped with a stainless steel ring to the μ-metal shield of the CMA
spectrometer using three set screws at 120° to each other. The second support
bearing is attached to a baseplate that rigidly supports the CMA by rods. The entire
assembly is mounted on an 8-in.-diameter Conflat flange. Translation is achieved
using a small linear translator, mounted on a small flange, which pushes and pulls
on the baseplate at a point close to the upper support rod to avoid torque on the
upper bushing during motion.
The flexible electrical connecting wires are looped to give the needed extension
and are made of Teflon-coated coaxial cable [40]. The coaxial cable is made UHV
compatible by stripping off the outer cover, leaving the cable shield, and contributes
no significant outgassing problem.
Another design for a translating Auger spectrometer may be found in [41]. In
this case, the miniature instrument (electron optics 30 mm diameter; 45 mm length)
is held on a single linear translator mounted on a 2 3/4-in.-diameter Conflat flange.
The resolution of this instrument is R = AE/Eo = 1.5 % [41].
18 1 Motion in Vacuum
The provision of various degrees of motional freedom for samples, apertures, etc.,
in ultrahigh vacuum systems is often accomplished with complex manipulators
such as that shown on p. 3. When complex motion is required and the measurement
of the position of the sample is not required, more simple devices are possible. In
Fig. 1.12, a device involving the combination of a port aligner and a linear trans-
lator is used to give multiple degrees of freedom for a rod. The port aligner allows
±10° motion within a solid cone, which is schematically shaded. The linear
translator permits motion back and forth within this solid cone [42].
The rotation of gratings and other optical components through small angles in ultra
high vacuum for the purpose of alignment is not straightforward. This device
permits the adjustment of a grating through an angular range of less than 1° with
ease [43]. Adjustments in one direction or in two are possible, depending on the
structure used.
Figure 1.13a shows a side view of a device permitting one-dimensional rotation
about the x-axis. Two planar broad-leaf springs are connected to a base support, and
to a movable platform. A plane extrapolation of the two springs intercepts at the
origin, o, which is the center of the grating. Here the x-, y-, and z-axes are defined.
An adjustment rod is connected to the base support, and to an external adjustment
screw. Bellows, as shown, isolate the mounting system from the atmosphere.
Adjustment of the screw in the z-direction rotates the grating about the x-axis.
1.9 Simple Device for Small Rotations in Ultrahigh Vacuum—Grating Application 19
Fig. 1.13 Small rotations in UHV. a One-axis device. b Two-axis device. c Result—Z rotation
20 1 Motion in Vacuum
The principle can be extended to rotation about two axes, as shown in the
schematic (Fig. 1.13b). Here, two sets of orthogonal leaf springs are used, and
screws for rotation of the adjustment rod operate in the z- and x-directions, leading
to grating rotation about the x- and z-axis.
Figure 1.13c shows the use of the two-axis device for optimization of the res-
olution of a UHV spectrometer. In this case, an angular resolution of less than 0.1°
was achieved for optimum spectrometer resolution adjustment [43].
Figure 1.14 shows a turntable for the rotation of heavy equipment in UHV with
high angular precision and repeatability, and without the need for coated or
lubricated bearings. Rotation is achieved using a rotary feedthrough to move the
axle of the turntable. The flat bottom of the turntable rests on three spheres made of
a metal carbide mounted in a V-groove. The spheres are of 10-mm diameter and the
three define the plane on which the turnable rotates. The mechanical stability is
excellent, and several tens of kilograms may be supported. In addition, friction
is low in ultrahigh vacuum. The danger of the spheres coming close to each other is
small, especially when forward and backward motion is used alternatively. The
system is uninfluenced by bakeouts at 200 °C [44].
because shape memory alloy (SMA) wire helixes can be made to change from the
“deformed” cold shape to the “preferred” hot shape reversibly. The restoring force.
is quite large and may be used to achieve small motions as illustrated in the two
examples shown here. There are numerous alloys that exhibit the SME, and a
favorite SMA for UHV applications is NiTi. This material is easily formed into the
hot shape and has a convenient transformation temperature. The shape change is
caused by a transformation from a martensitic to an austenitic phase, with the high
temperature austenitic material being produced at about 60 °C, while the reverse
transformation occurs at about 30 °C. Deformations producing less that 2 % local
strain in the martensitic state can be reversed by heating through the transformation
temperature. By setting the hot shape in a helical wire it is possible to get extensions
of several hundred percent without exceeding this strain limit.
Figure 1.15a shows a shutter arrangement made with a 0.45-mm-diameter NiTi
wire [45]. The hot shape of the left-hand helix was set by winding it on a cylindrical
mandrel and then heating to 500 °C for 1 h in a vacuum annealing oven to prevent
oxidation. The helix was then stretched between two stainless steel posts and pinch
clamped to each. The helix was therefore stretched from its preferred hot shape.
Upon heating with a current of 2.5 A, the left segment of the helix contracts to its
preferred hot shape, moving the attached shutter to the left in a period of less than
0.5 s. This operation stretches the right-hand cold helix beyond its plastic limit,
which means that it has little restoring force. Thus, the heating current can be turned
off after the shutter has moved to the left. To restore the shutter to its original
position, one heats the right-hand portion of the helix.
The effect of bakeouts at 215 °C on a severely stretched (300 %) helix was tested
over a 24-h period, and it was found that this treatment did not affect the hot shape
memory [45].
A second example in which the SME is used to open and close a clamp on a
manipulator is shown in Fig. 1.15b [46]. Here, the jaws of the clamp are opened
when the NiTi SMA wire is heated. The particular wire [47] used here is of small
diameter (0.15 mm) and has been treated in manufacture to exhibit a shorter length
at high temperature and a longer length at low temperature. For strong clamping,
four such wires were braided together and connected to a copper wire by a simple
overhand knot above the ceramic insulator tube, which passes through the back end
of the stationary upper jaw. The other end of the wire bundle is grounded through
the jaw. A restoring force to close the jaws when the heating current is off is
supplied by a spring mounted on a screw that is fixed in the upper jaw but that does
not touch the lower jaw. When the lower jaw opens, this spring is compressed.
Electrical heating to a wire temperature of 115 °C causes the jaws of the clamp to
open up by about 2 mm when 0.21 A is passed through the four wires in vacuum. If
the current is taken too high (0.23 A in vacuum), the annealing temperature will be
exceeded, and the SMA wires will lose their pulling ability. The reader is referred to
texts on the SME for more details [48, 49]. Also engineering studies of SME test
structures are described in [50].
22 1 Motion in Vacuum
Fig. 1.15 Small motions using shape memory effect. a Shape Memory effect—Shutter. b shape
memory effect—Clamp
The use of magnetic eddy currents in an ultrahigh vacuum system to drive a low
friction shutter is illustrated in Fig. 1.16. The eddy current magnet, shown without
its housing in the upper part of the Figure, drives the stepped copper rotor at
1.12 Rotary Shutter Device Driven by Magnetic Eddy Current 23
300 rpm when ball bearings are used, and 60-Hz power is applied to the coil. The
copper disk is 0.8 mm in thickness. The stepped openings on the perimeter permit
the operator to vary the open duty cycle for evaporations through the openings from
an evaporation source mounted above the disk; operation near the perimeter gives
larger fractional transmission than with the substrate placed more toward the axis.
The lower end of the iron cylinder (5 cm dia. × 7.5 cm long) contains a semi-
circular piece of high-purity copper, 0.6 cm thick. It is wrapped with 150 turns of
#14 (1.6 mm dia.) magnet wire with Formvar insulation. This is housed in a
stainless steel can to prevent outgassing, and the can is welded shut using the
ceramic feedthrough for electrical contact to the coil inside.
The device works on the basis of the lag of the magnetic field at position b from
that at position a. This means that when the magnetic field is maximized during the
24 1 Motion in Vacuum
cycle at a, the eddy currents centered at b will also be maximized. A force on the
volume element of the copper disk near a will be in the direction from a to b, and
the disk will move. One quarter cycle later, the magnetic field at b will interact with
the eddy currents centered at a and the force is again in the direction from a to
b. During the next half cycle, both the magnetic field and eddy current directions
are reversed, again giving a force in the same direction. A mechanical brake is
employed to stop the rotation.
The shutter device is used to reduce the flux of evaporating material arriving at
the substrate by a precise geometrical method involving only the dimensions of the
transmission slots relative to the circumference of the sector disk. Thus, thick films
deposited near the outer edge of the disk are proportionately reduced to thinner
films deposited at a known radius under the disk. Measurements of the film
thickness deposited above the shutter can be converted to sector thicknesses below
the shutter [51]. A similar device, driven conventionally, for extending the range of
a quartz microbalance has been described previously [52].
The use of boron nitride, BN, as a bearing lubricant has been demonstrated for light
loads in a combination rotary-translator bearing [60]. This is especially convenient
for magnetically driven shafts where only small forces may be applied.
Figure 1.19 shows an exploded view of the bearing assembly. The rotary and
translation shaft, made of stainless steel, is supported at some distance from the
1.15 Self-lubricating Bearings in UHV 27
bearing by V-blocks that can be moved into contact with the shaft, and these are not
shown here. The shaft is held by WC bearings (0.48 cm diameter) which them-
selves are held in a BN ball separator cylinder containing holes that accommodate
the WC balls. This BN ball separator provides continuous lubrication to the WC
balls and the rod, since BN is a dry lubricant. The balls are held in their place by a
type 6061 Al alloy retainer which slips over the top of the balls keeping them from
touching each other. An outer-bearing sleeve then slips over all of the components
and is stabilized by contact with the inside of the vacuum wall tubulation. The
bearing assembly is loosely retained between the magnetic drive mechanism and a
stop on its other end, not shown.
The system can be repeatedly baked at 275 °C. Since BN is hygroscopic, when
the chamber is open to atmosphere, the bearing is kept at 80–100 °C using heating
tape on the outside of the tubulation. No contamination of samples by BN has been
detected using various surface analysis methods.
Fig. 1.20 Heavy load lubrication in UHV. a Ballscrew test device. b Wear track patterns
10−10 Torr. The vacuum condition was achieved after a bakeout at 140 °C. The
motion drove the mass upward and downward 45 mm, at a speed of 2.25 mm/s. The
screws and nuts of the ballscrews are made of 440C stainless steel, finished using
electrochemical buffing to a roughness of about 0.1 μm, and then coated as
described below.
The ballscrew was coated with a TiC film, 1.5 μm in thickness, and with a
micro-Vickers hardness of about Hv 1100. An Ag coating of 0.3-μm thickness was
then applied. The ball bearings mating to the ballscrew were also made of 440C
stainless steel, coated with an Ag film of 0.3-μm thickness. The unlubricated sur-
face, used as a control, was not treated with either TiC or Ag.
1.16 Lubrication for Heavy Sliding Loads in Ultrahigh Vacuum 29
Figure 1.20b shows the wear track observed visually after 200,000 cycles. The
thin-film Ag coating is still retained on sections of the screw that did not undergo
contact with the balls. The TiC underlayer is visible in the contact regions and was
not worn through. Surprisingly, when the device was operated in air, the untreated
ballscrew performed well but the coated ballscrew did not move smoothly.
Plasma-coated Ag films on stainless steel screws for use in UHV have been
found to resist cold welding [62]. Also, ordinary Ag plating of either the bolt
threads or the nut threads of stainless steel bolts used internally or externally in
baked UHV systems is reported to be an excellent lubricant using coatings of
10-μm thickness [63]. It has been found that coating metals such as Ta, Cu, Mo, and
stainless steel with an oxide film, by heating to redness in air, will provide pro-
tection against galling during motion in UHV [64].
A good review on tribology and lubrication in the vacuum environment in space
is found in [65].
Steel ball bearings are often used unlubricated in ultrahigh vacuum. At high rota-
tional speeds, over time, these bearings tend to become sticky before final seizure.
The development shown in Fig. 1.21 uses polytetrafluoroethylene (PTFE or Teflon)
as a bearing material for continuous rotation at 10–1000 rpm. The bearing is
machined using sharp oil-free tools, and care is taken not to introduce metallic
particles into the PTFE. Cleaning is carried out in organic solvents, and traces of
metal and oil may be removed by use of hot nitric acid solution, followed by
soaking in distilled water.
The design has three essential components. The first is the use of relief slots
made by a small, round file at 90° intervals around the inside of the bearing.
Removal of about one-third of the contact area was found to be essential. The
second feature is the placing of pumping holes in the PTFE bearing, and more of
these can be supplied on the upper housing. The third feature of the design is the
use of an 0.25–0.50-mm-diameter wire squeezed between the bearing and the
housing as a friction point to prevent the bearing from rotating when torque is
supplied from the shaft.
Bakeout below 200 °C is appropriate, and temperatures above about 250 °C are
to be avoided. The bearing should be operated near room temperature and the
mechanical admission to the bearing surface of flakes of metal from thin film
deposition in the chamber should be avoided. Small partial pressures of fluorine and
chlorine were seen initially, and these decreased with time. Initial use also produced
small partial pressures of methane and carbon monoxide which soon disappeared.
The rotary bearing was used with a 400 rpm shaft for a million rotations without
any apparent degradation [66].
30 1 Motion in Vacuum
The need for an internal aperture that can be adjusted in ultrahigh vacuum exists
both in optical experiments and also in vacuum systems where variable pumping
speed control is desired. An inexpensive aperture device is shown in Fig. 1.22. It
consists of a variable iris aperture [67] that may be purchased as a component,
mounted inside a 2 3/4-in.-diameter Conflat flange. The aperture is mounted on an
1.18 Flange-Mounted UHV Variable Aperture 31
adapter ring that is tack welded into the flange to avoid distortion. The aperture is
fixed from rotation by mounting pins that fit into the adapter ring and through the
wall of the aperture housing. The aperture is tightened into place by a threaded ring
containing holes for a spanner wrench. The original Conflat flange has had an inside
slot milled around its inner circumference for the motion of the aperture control.
This is accomplished by a linear actuator coupled to the aperture control lever by a
Cu-Be strip (0.3 mm thick) that extends around the aperture housing inside the
milled slot. The aperture is degreased before use and relubricated with a tiny
amount of MoS2 applied lightly with a cloth.
The iris opening ranges from 1 to 25 mm and can be reproduced to a precision of
2 % if one is careful to avoid backlash by moving always in the same direction for
final adjustment [68].
A clever mechanical design to convert a linear screw motion into a linear platform
motion with reduced amplitude has been described by Dr. H.P. Rust and
Mr. L. Cramer [69]. The design is used as an approach mechanism for an STM tip
in a low-temperature instrument [70].
32 1 Motion in Vacuum
Fig. 1.23 Linear movable platform. a 3D-sectional view of portion of LMP. b Top and sectional
view (x–x)
The device shown in Fig. 1.23a, b is made entirely from a single piece of bronze,
which, after fabrication, is rigidly attached to a base plate also made of bronze. The
active LMP part is milled with three orthogonal slots in the form of a C, and then
6 “milled hinge” cavities are also produced as shown in Fig. 1.23b by dashed lines.
These milled hinge cavities have only a thin section of bronze remaining at the top
1.19 Linear Motion Platform (LMP) 33
of each cavity, giving excellent bending response through a few degrees. The LMP
is then rigidly attached to the two projecting flats from the baseplate, and a screw
motion, schematically shown as a thumbscrew, is added on the left.
The device works as follows. As the screw is moved up or down, the plate
section above it moves up or down, and we will assign a length of motion a as
shown in Fig. 1.23a. This motion is reduced at point b by virtue of the relative
length of the arms involved as the milled hinge is bent near the point b. The section
of the LMP device labeled c is rigidly connected to the outer plate section con-
taining the a and b motions, and there is therefore an equivalent c motion as shown,
i.e., b = c. As section c moves up, a milled hinge, midway between c and c’ flexes
as a result of the forces it experiences. This causes a symmetrical distortion in the
bridged section c–c’ and leads to equal angular displacement on both ends of the
platform, which occurs as a result of distortions in the region of the pair of milled
hinges on each end of the platform. The small motion of the platform is sche-
matically indicated by d. Because of the symmetry, the motion of the platform up
and down will be in the normal direction without any tilting, which is the desired
result.
The function of the device can best be understood by remembering that the light
gray region, shown in Fig. 1.23b, and containing a, b, and c and the symmetrically
similar regions, is rigid and undergoes identical angular distortion away from the
baseplate when the screw is advanced. The mechanical advantage is given by the
ratio of the lever arms, l1 and l2.
The transport of samples from one vacuum system to another, where different
preparation and measurement methods are used, is a common necessity in UHV
measurements, as only so many techniques may be used in a single vacuum
chamber. A trolley system can be used to make this transport, and the system shown
has the capacity to transport 4 samples at one time, dropping them off amongst
5 vacuum systems. Magnetic transfer is used to insert a sample holder into the
trolley system from the loading chamber at the third position from the left. A screw
advance mechanism moves the samples laterally in 4 special sample holders which
are designed as locking devices. The locking sample holders, shown schematically
on the trolley in the insert to Fig. 1.24, are used to hold the samples. Engagement of
each sample holder by the magnetic transfer device allows the sample holder to be
inserted and locked by rotation on centrally-located sample stages in each system.
While this device has maximum versatility for simple analysis at various locations,
heating of the samples inside the vacuum chambers will require reversible electrical
connections for both electrical heating power and thermocouples as described
elsewhere in this book [71].
34 1 Motion in Vacuum
Two problems common to most STM designs have to do with the thermal drift of
the relative position of the STM tip and the sample and the inability to replace a
sample in the same position so that the STM tip can return to the same image point.
Both of these problems have been solved by a design that addresses these issues
[72].
Figure 1.25a shows a side view of the STM design. Two piezotubes are mounted
on the same base. The inner tube (scan piezotube) is used for scanning the tip, while
the outer tube (translation piezotube) connects to the sample support assembly and
is used to support and translate this assembly. Both of these piezotubes are of
exactly the same length, so that thermal drifts are compensated in the z-direction;
their circular symmetry eliminates thermal drifts in the x, y-directions. The outer
translation piezotube is surrounded by a thermal shield, made of oxygen-free high
conductivity (OFHC) Cu to eliminate temperature gradients, which would defeat
the principle of having two identical-length piezotubes for compensation of thermal
drift effects. The outer tube is connected to the sample holder through a quartz
mounting assembly that consists of an outer quartz tube and a quartz or
metal/ceramic sample holder. The sample holder slides (by means of five spherical
bearing contacts) on two metallized quartz rails that are mounted inside the outer
quartz tube as shown in Fig. 1.25b. Operation of the translator piezodrive causes the
sample to jump toward the tip as a result of the friction and inertia effects in the
sample holder, and adjustment of the piezo activation waveform can be used to
1.21 Thermally Compensated STM with Repeatable Sample Positioning 35
Fig. 1.25 Thermally compensated STM—repeatable sample positioning. a Side view. b Sample
holder. c Detail, probe translator
36 1 Motion in Vacuum
produce jumps from about 104 Å to less than 5 Å. By working in the kHz frequency
domain, translation rates up to 1 mm/s are possible.
The sample holder may be removed from the outer quartz tube containing the
rods and the sample may be transferred and heated in a furnace or in a load lock.
Upon return of the sample holder to the quartz rods, the design permits repositioning
to an accuracy of several thousand Å, depending upon how much the sample is
bumped during transfer. This is accomplished by using gravity to align the sample
on the rails, which are tilted with respect to the horizontal plane. The repositioning
scheme, using gravity to set the sample on the rails in the same position, is based on
kinematic design principles that are well described elsewhere [73].
The details of the construction of the probe translator are shown in Fig. 1.25c.
The thermal drift experienced by this design is about 1 Å/h.
A piezo-driven translator for the tip, which moves in the x, y-directions over
several millimeters by a stepping and gripping action, thereby eliminating
mechanical positioning devices, has been described as a modification of this device
[74].
References
1. P. Toennies, Max Planck Institut Stroemungsf, Bunsenstr. 20, D-37073 Göttingen, Germany
(private communication)
2. D. Venus, Rev. Sci. Instrum. 62, 1361 (1991)
3. Y. Dai, H. Li, F. Jona, Rev. Sci. Instrum. 61, 1724 (1990)
4. T. Engel, D. Braid, E.H. Conrad, Rev. Sci. Instrum. 57, 487 (1986)
5. J.J. Zinck, W.H. Weinberg, Rev. Sci. Instrum. 56, 1285 (1985)
6. K.D. Jamison, F.B. Dunning, Rev. Sci. Instrum. 55, 1509 (1984)
7. N.J. Wu, A. Ignatiev, Rev. Sci. Instrum. 56, 752 (1985)
8. J.A. Stroscio, W. Ho, Rev. Sci. Instrum. 55, 1672 (1984)
9. J. Larscheid, J. Kirschner, Rev. Sci. Instrum. 49, 1486 (1978)
10. K.D. Jamison, F.B. Dunning, Rev. Sci. Instrum. 55, 1509 (1984)
11. H. Durr, Th Fauster, R. Schneider, J. Vac. Sci. Technol. A8, 145 (1990)
12. M. Michaud, P. Cloutier, L. Sanche, Rev. Sci. Instrum. 66, 2661 (1995)
13. R.R. Wilson, Rev. Sci. Inst. 12, 91 (1941)
14. V.R. Unwin, K. Horn, P. Geng, Vakuum Tek 29, 149 (1980)
15. M.F. Zabielski, R.R. Blaszuk, J. Vac. Sci. Tech. 13, 644 (1976)
16. J.T. Yates, Jr. (unpublished)
17. J.J. Zinck, W.H. Weinberg, Rev. Sci. Inst. 56, 1285 (1985)
18. A. Pararas, S.T. Ceyer, J.T. Yates Jr, J. Vac. Sci. Tech. 21, 1031 (1982)
19. E. Puckrin, J.K. Fowler, A.J. Slavin, J. Vac. Sci. Tech. A7, 2818 (1989)
20. D.J. Auerbach, C.A. Becker, J.P. Cowin, L. Wharton, Rev. Sci. Inst. 49, 1518 (1978)
21. The spring-loaded Teflon seals are available in many sizes from the Fluorocarbon Corporation,
10871 Kyle Street, P.O. Box 520, Los Alamitos, CA 90720. Their descriptive literature gives
many details about the seals and the preparation of the metal surfaces on which they ride
22. S.M. George, J. Vac. Sci. Tech. A4, 2394 (1986)
23. J.A. Stroscio, W. Ho, Rev. Sci. Inst. 55, 1672 (1984)
24. P.A. Thiel, J.W. Anderegg, Rev. Sci. Instrum. 55, 1669 (1984) (A particularly nice design of a
manipulator with an easily demountable cold finger end for crystal replacement can be found)
References 37
25. R.A. Kubiak, P. Stonestreet, S.M. Newstead, E.H.C. Parker, T. Whall, N. Naylor, J. Vac. Sci.
Technol. A9, 2797 (1991)
26. J. Larscheid, J. Kirschner, Rev. Sci. Instrum. 49, 1486 (1978)
27. K.D. Jamison, F.B. Dunning, Rev. Sci. Instrum. 55, 1509 (1984)
28. H. Durr, Th Fauster, R. Schneider, J. Vac. Sci. Technol. A8, 145 (1990)
29. P. Feulner, D. Menzel, J. Vac. Sci. Technol. 17, 662 (1980)
30. H. Schlichting, D. Menzel, Surface Sci. 285, 209 (1993)
31. H. Schlichting (private communication)
32. Obtainable from Acheson Colloids, 1607 Washington Avenue, P.O. Drawer 611747, Port
Huron, MI 48061-1747
33. T. Gog, S.M. Durbin, Rev. Sci. Instrum. 60, 3030 (1989)
34. R.W. Bertram, J. Vac. Sci. Tech. 7, 612 (1970)
35. C.M. Rouleau, R.M. Park, J. Vac. Sci. Tech. A11, 464 (1993)
36. MSC, 151 Sunnyside Boulevard, Plainview, Long Island, New York, 11803
37. H. Gutleben, J.T. Yates Jr, J. Vac. Sci. Tech. A9, 170 (1991)
38. M.C. Reuter, R.M. Tromp, J. Vac. Sci. Technol. A6, 2575 (1988)
39. 1/2 in. -Thompson A-81420-SS; 1/4 in.—Thompson A-4812-SS
40. Custom Cable Corp., 242 Butler Street, P.O. 1050, Westbury, NY 11590-3193. Cable type:
RG 316/V
41. D.N. Mcllroy, P.A. Dowben, A. Knop, E. Rühl, J. Vac. Sci. Technol. B13, 2142 (1995)
42. Dr. Oliver Schaff, Fritz Haber Institut der Max Planck Gessellschaft, Faraday Weg 4-6,
D-14195 Berlin (private communication)
43. W.A.M. van Bers, A.J.J. Franken, P.K. Larsen, Rev. Sci. Instrum. 54, 637 (1983)
44. W. Stocker, K.H. Rieder, J. Vac. Sci. Tech. A2, 1606 (1984)
45. A.P. Jardine, M. Ahmad, R.J. McClelland, J.M. Blakely, J. Vac. Sci. Tech. A6, 3017 (1988);
This idea is based on an earlier paper which also characterizes the behavior of a similar device
as a function of repeated cycling; see B.J. Mulder, Vacuum 26, 31 (1976)
46. W.F. Smith, E.E. Ehrichs, A.L. de Lozanne, J. Vac. Sci. Tech. A10, 576 (1992) (This article
contains additional references to other devices constructed with SME alloys)
47. The thicker SME wire used in Figure 11A was obtained from RayChem, 300 Constitution
Drive, Menlo Park, CA 94025
48. J. Perkins (ed.) Shape Memory Effects in Alloys (Plenum, New York, 1975)
49. T.W. Duerig, K.N. Melton, Proc. Mater. Res. Soc, Int. Meet. Adv. Mater. 9, 581 (1989)
50. M. Nishikawa, M. Kawai, T. Yokoyama, T. Hoshiya, M. Naganuma, M. Kondo,
K. Yoshikawa, K. Watanabe, J. Nucl. Matls. 179–181, 1115 (1991)
51. J.S. Brooks, R. Meservey, R.K. MacNabb, J. Vac. Sci. Tech. 20, 243 (1982)
52. F.C. Wilson, D.B. Sandstrom, H.B. Serreze, Rev. Sci. Instrum. 46, 1593 (1975)
53. T. Engel, D. Braid, Rev. Sci. Instrum. 56, 1668 (1985)
54. Barden Bartemp bearings, SR 188 SSTB
55. Globe motor, model 75 A 1003-2
56. Professor Thomas Engel, Department of Chemistry, BG-10 University of Washington, Seattle,
WA 98195 (private communication)
57. T. Engel, Rev. Sci. Instrum. 52, 301 (1981)
58. The bearings were from Barden Bartemp, Type SR 188 SSTB
59. The rotor and stator were obtained from KaVo Electrotechnisches Werk GmbH D-7970
Leutkirch, Germany, Type EV 48-15
60. G.J. Collet, E.L. Garwin, R.E. Kirby, Rev. Sci. Instrum. 58, 479 (1987)
61. H. Saeki, J. Ikeda, I. Kohzu, H. Ishimaru, J. Vac. Sci. Technol. A8, 3360 (1990)
62. R.A. MacGill, R.A. Castro, X.Y. Yao, I.G. Brown, J. Vac. Sci. Technol. A12, 601 (1994)
63. A. Contaldo, Rev. Sci. Instrum. 36, 1510 (1965)
64. W.F. Egelhoff, National Institute of Standards and Technology, Gaithersburg, MD
20899-0001 (private communication)
65. E.W. Roberts, M.J. Todd, Wear 136, 157 (1990)
66. R.A.A. Kubiak, P. Driscoll, V. Manning, R. Houghton, J. Vac. Sci. Tech. A4, 1951 (1986)
38 1 Motion in Vacuum
67. Edmund Scientific Company, 101 East Gloucester Pike, Barrington, NJ 08007-1380
68. F. Homan, J. Vac. Sci. Tech. 17, 664 (1980)
69. Registered design: H.-P. Rust, L. Cramer, Deutsches Patentamt Gebrauchsmuster No. G94 12
112.5, (27 July 1994)
70. H.-P. Rust, J. Buisset, E.K. Schweizer, L. Cramer, Rev. Sci. Instrum. 68(1), Jan 1997
71. J. Szuber, private communication
72. J.W. Lyding, S. Skala, J.S. Hubacek, R. Brockenbrough, G. Gammie, Rev. Sci. Instrum. 59,
1897 (1988); also, U.S. Patent 4,841,148
73. E.B. Wilson, An Introduction to Scientific Research, 1st edn. (McGraw Hill, NY, 1952)
74. R.T. Brockenbrough, J.W. Lyding, Rev. Sci. Instrum. 64, 2225 (1993)
Chapter 2
Sample Transfer
The ability to transfer samples from high-pressure reaction conditions into ultrahigh
vacuum for surface analysis or for crystal preparation has its origin in the work of
the Somorjai group at Berkeley [1], who devised apparatus to do this for studies of
the kinetics of catalytic reactions on metal single crystal surfaces. Figure 2.1 shows
an early version of this scheme. A single crystal mounted on a flange is surrounded
by a movable reactor cup that seals the crystal into a small cavity capable of being
pressurized. The reactor cup is sealed under strong force by a drive rod that operates
inside a stainless steel welded bellows. Because of the strong sealing forces
involved, the body of the ultrahigh vacuum chamber, shown cross-hatched in
Fig. 2.1, is made of extra thick stainless steel to avoid significant distortion when
the sealing force is applied. The crystal can be heated electrically in the reactor cup
and its temperature measured with a thermocouple welded to the crystal edge. In
some cases the inside of the high-pressure chamber is coated with a gold film to
reduce wall reactivity. Gas circulation pumps cause mixing of the high-pressure gas
in order to mimic stirred flow conditions. The design shown in Fig. 2.1 would not
permit Auger analysis of the crystal surface because of its fixed geometry, but later
work permitted the crystal itself to be moved inside the ultrahigh vacuum system
after the reactor cup was retracted [2–5].
Figure 2.2 shows an alternate design that has been widely used [6]. The crystal is
mounted on a ceramic feedthrough at the end of a transfer tube housed inside a
welded stainless steel bellows. The bellows conveniently moves on an external
translation carriage containing two parallel stainless rods that accommodate linear
bearings connected to the outer flange (not shown) of the retraction device. When
retracted, the crystal sits in a reaction chamber which can be pressurized to about
two atmospheres. The reaction chamber is closed off by two all metal sealed valves,
one leading to the surface analysis/preparation chamber, and the second leading to
the high-pressure chamber pumps. As in Fig. 2.1, the entire apparatus can be
brought into ultrahigh vacuum condition by means of bakeout. The crystal can be
translated into the analysis/surface preparation chamber where surface analysis and
gas dosing can be carried out under well-controlled conditions. In the high-pressure
reactor chamber, gas analysis is made by extraction of a small quantity of the gas
either into the sampling loop of a flame ionization gas chromatograph, or into the
mass spectrometer of the main ultrahigh vacuum chamber. For many analyses, the
gas chromatograph is superior in sensitivity and selectivity to the mass spectrom-
eter. Compared to the reactor cup design, the internal volume of the high-pressure
cell shown in Fig. 2.2 is large and wall effects may in principle be larger. For certain
reactions, the wall effect has been measured and found to be small [7]. Viton O-ring
sealing valves are likely to give poorer results because of absorption and liberation
of gases from the Viton. Often, wall cooling is necessary when the single crystal
operates at high pressure and at elevated temperature because of efficient heat
transport to the s through the dense gas. It may be necessary to cool the outer wall
of the reactor chamber in these cases. Such cooling is not easy to achieve in the
reactor cup design of Fig. 2.1. High-pressure chambers are often useful in chemical
2.1 Sample Transfer from High Pressure to Ultrahigh Vacuum 41
cleaning of crystals because of the high gas density achievable, let us say for
oxygen or hydrogen treatment of a metal single crystal. There is a danger with all
stainless steel reactor chambers when operated at high pressure with carbon mon-
oxide gas. The carbon monoxide will produce nickel and iron carbonyls from the
stainless walls that transfer their metal atoms to the hot metal surface (see pp. 89
and 505).
Another type of sample transfer device was marketed by Leybold-Heraeus and is
schematically shown in Fig. 2.3. A 2-cm-diameter polished stainless steel rod is
arranged to move laterally by sliding through a number of differentially pumped
O-ring seals that permit rod translation rapidly from a high-pressure reactor into the
ultrahigh vacuum system. Differential pumping occurs by means of a mechanical
pump in the first stage between the reactor cell and the ultrahigh vacuum system,
and high-pressure gases used in the reactor cell would be carried away here. The
second stage is pumped with a turbomolecular pump, and achieves a pressure of the
order of 10−8 mbar [8, 9]. This design was used by others [10, 11] with minor
modifications, and in the lower section of Fig. 2.3 can be seen the manner in which
a metal single crystal is mounted [10]. The high-pressure reactor is only a few cubic
centimeters in volume and was entirely made of stainless steel and gold-plated pan
bronze, except for Viton/Teflon seals. The small volume allows an enhanced sen-
sitivity for product gases evolved in catalytic reactions at high pressures [10, 11].
42 2 Sample Transfer
Leaks due to poor O-ring lubrication or to scratching of the stainless rod is avoided
to some degree by providing Teflon liners, in the form of tubes which slide on the
rod between the O-rings [12].
A more recent type of arrangement for sample transfer between an ultrahigh
vacuum system and a high-pressure chamber is called the “rod-through-chamber”
geometry. As shown in Fig. 2.4, a single crystal is mounted on a long feedthrough
that contains a special conical shoulder that mates to a Teflon gasket when the
crystal has been advanced into the high-pressure chamber. The conical contour
shoulder is polished to a mirror finish, and acts like a plunger when closing the
2.1 Sample Transfer from High Pressure to Ultrahigh Vacuum 43
apparatus to isolate the two sections. This Teflon closure can seal against 1 atm
pressure with a leak rate less than 10−9 Torr L/s. When the high-pressure chamber is
evacuated and the probe is withdrawn, a gate valve is closed to completely isolate
the ultrahigh vacuum system from the high-pressure chamber. While bakeout
temperature limits are not specified in the original paper [13], temperatures in the
150–200 °C range would not be expected to disturb the Teflon. This type of seal
requires that a rather large force be applied to the rod in order to make a gas tight
junction with the Teflon gasket.
In some cases, it is not appropriate to employ high forces along the rod to make a
metal-to-Teflon seal. Fig. 2.5 shows a method for making a high-force seal using
another approach, on a crystal holder that moves vertically and that is not so robust as
that shown in Fig. 2.4. Here the crystal is mounted on electrical leads that are cooled
in a reservoir containing liquid nitrogen supplied through a tube entering from the
left-hand side. The gas tight seal is again made by contact between a polished
stainless steel surface on the bottom of the plunger, which is of spherical contour
mating up with a square-edge Teflon gasket. The upper portion of the plunger is also
of spherical contour, and when the plunger is seated properly on the Teflon gasket,
tapered steel wedges are able to be brought in from the side to mate with the upper
spherical surface. This gives a strong downward force, making the lower
Teflon-stainless steel junction gas tight. The steel wedges are driven laterally by two
separate linear motion feedthroughs. They enter through heavy guide blocks to
provide strength needed to make the compression seal. In order to provide the
nonrigid character needed to accommodate the downward motion produced by the
44 2 Sample Transfer
tapered wedges, the plunger assembly is coupled to the liquid nitrogen reservoir by
means of an internal bellows. This seal also has a leak rate less that 10−9 Torr L/s [13].
A number of other designs for transferring samples between vacuum systems
have been devised and are discussed in [14–19].
A novel design for a continuous stirred flow reactor for use in studying surface
reaction kinetics at high pressure, but which cannot be used for placing the catalyst
surface in a position for surface analysis, is discussed in [20]. The criteria for
deter-mining that the reactor is behaving as a stirred flow reactor are given in [20].
Another catalytic reactor, which permits the measurement of the spatial distribution
of reactant and product gases, has been described in [21]. This reactor also does not
permit the analysis of the catalyst surface composition by electron spectroscopies.
In this design, by measuring the leak rate for the reactor through a capillary,
the temperature of the gas in the vicinity of the catalyst surface can be measured
[21, 22].
2.2 Novel Sample Transfer from UHV Chamber to External Cell 45
Fig. 2.6 Novel sample transfer to external cell. a UHV chamber. b Sample transport device
2.3 UHV Sample Transfer Device with Low-Temperature Capability 47
Here is described the provision of an internal cup high-pressure cell that accom-
modates a sample mounted to move in a normal fashion to various analytical ports
in an ultrahigh vacuum chamber [31]. A schematic diagram of the apparatus is
shown in Fig. 2.8a, where a sample, mounted in a special cup, can rotate to various
analytical ports. One of its positions is opposite a reactor cup assembly that can be
Fig. 2.8 Internal cup high-pressure cell. a Schematic of rotating sample + cup assembly.
b Schematic of high-pressure reaction cell
50 2 Sample Transfer
moved laterally by hydraulic pressure. The sample cup is rigidly held by a retainer
assembly, connected to the reactor cup assembly, which prevents the application of
large forces to the manipulator shaft when the seal is made using hydraulic pressure.
Figure 2.8b shows an enlarged schematic view of the reactor cup assembly and its
hydraulic cylinder. When the sample has been rotated into its retainer assembly, the
reactor cup assembly is pressed against the sample cup, and an annealed Cu gasket
produces the seal at a mini-Conflat sealing surface. This operation is assisted by an
internally tapered guiding surface on the outer section of the reactor cup assembly.
This method of sealing allows repetitive use of the same annealed gasket. The
background pressure in the UHV system was found to rise into the low 10−9 Torr
range when the cell was pressurized to 100–2500 Torr. The reactor cell (volume
45 cm3), could be evacuated through gas lines passing through the heavy stainless
steel reactor cup assembly, or more rapid evacuation could be achieved by opening
the pressurized cell inside the UHV system, causing a pressure surge that decreased
into the 10−9 Torr range in 1–2 min. The original paper [31] contains a brief review
of other methods for building high-pressure cells.
A different design, permitting a single crystal catalyst sample to be moved into a
small reactor cup, sealed with a Viton O-ring, has been described previously [32–
34]. In this design also, the single crystal may be moved in an ordinary fashion to
various positions to face different analytical probes mounted radially in an ultrahigh
vacuum system. The design in [33] cleverly permits the reactor cup to be sealed off
from the UHV system when the sample is removed, preventing outgassing effects
from the inner walls from contaminating the UHV system.
A high-pressure transfer cell that is designed to have a small (30 cm3) volume and
provision for sample cooling is shown in Figs. 2.9a and b. Pressures as high as
15 bar are able to be sustained in the high-pressure cell. Sealing is accomplished by
a stainless steel/Teflon seat. Impurities on the crystal that might possibly come from
the Teflon under compression are eliminated in this design [35].
Figure 2.9a shows a schematic design for the transfer rod. The sample is elec-
trically mounted on Mo connector rods with W, Ta, or Mo wires. A thermocouple is
also welded to the sample. The transfer rod enters the chamber through differen-
tially pumped spring loaded Teflon seals (not shown) such as those mentioned on
pp. 6–11 [36]. These seals permit both translation and rotation of the sample when
it is in the UHV chamber. The transfer rod is self aligning as it enters the Teflon
sealing position. The compression of a spring permits the force on the Teflon seal to
be measured and reproduced at about 400 lb. Heating of the sample in vacuum to
1300 K is possible by resistive heating. Heating of the sample to about 700 K is
possible in 10 bar of H2 if cooling is achieved by using flowing air rather than liquid
nitrogen. Cooling to about 130 K can be achieved with liquid nitrogen coolant. In
order to prevent extreme cooling of the transfer rod when liquid nitrogen is used, a
2.5 High-Pressure Transfer Cell 51
Fig. 2.9 High-pressure transfer-cell. a Transfer rod for UHV. b High-pressure cell
52 2 Sample Transfer
U-shaped Cu tube is inserted inside the transfer rod wall so that it is held com-
pressively against the inner wall, and warm, dry air is continuously circulated
through this tube when liquid nitrogen is used. This prevents failure of the sliding
seal mechanism owing to cooling. Teflon spacers are inserted equidistantly inside
the transfer rod using a pusher rod with calibrated markings. The outside of the
transfer rod is polished to the correct rms roughness following the directions of the
spring-loaded Teflon seal manufacturer [36]. The inside of the transfer rod is
evacuated with a dual-stage mechanical pump when liquid nitrogen cooling is
employed. The outside of the rod is protected with a hypalon/nylon bellows [37]
and this bellows is purged continuously with nitrogen gas.
Figure 2.9b shows the transfer rod sealed into the high-pressure chamber. An
alignment insert, H, is brazed between the vacuum housing and the high-pressure
cell. It is made of a bakeable and UHV-compatible soft aluminized bronze [38] so
that it does not scratch or damage the transfer rod. The annular space between the
Teflon seat and the high-pressure chamber is of the order of 0.05 cm, and its length
is of the order of 2.5 cm. Thus, at high pressure, possible diffusion of impurities
from the Teflon to the sample is minimized by the long diffusion time. Calculations
of the diffusion time indicate that at 10 bar of gas pressure, diffusion times of
greater than 30 min are involved over the 6.5 cm distance from the Teflon to the
sample.
The high-pressure cell is directly attached to a gas handling manifold that serves
to introduce gases as well as to extract them for gas chromatographic analysis. In
addition, direct pumping of the high-pressure cell by a small turbopump may be
carried out.
Proof of the performance was made in two ways. First it was shown that a
pressure of 10 bar of H2 led to an increase of pressure of only 1 × 10−9 Torr in the
UHV chamber. Secondly, storage of an atomically clean Cu crystal for up to 30 min
in the evacuated high-pressure cell does not lead to measurable contamination as
measured by XPS or TPD methods. Also, long reaction periods at high pressure
occur with constant reaction kinetics on a Cu single crystal surface [35].
Often it may be desirable to transfer a sample from one chamber to another and to
disconnect the sample from the transfer device after transfer. In Fig. 2.10 one may
see that transfer from a reaction chamber to an ultrahigh vacuum chamber is
accomplished by use of a magnetic transfer rod, which is commercially available.
The sample, mounted on a ceramic sample holder ring at the end of the transfer rod,
sits between three tines on the fork at the end of the transfer arm. The hooks of the
tines lock into the sample holder ring by a slight twist, which may be accomplished
2.6 Sample Transfer with Disconnect-I 53
with the transfer magnet. The electrical connecting pins are machined from 6–32
screws as shown in the central Figure, and these pins insert into homemade sockets
made if necessary from alloys designed for thermocouple connections, etc. It is
stated that the electrical connections do not have to be particularly tight and that
ordinary misalignments in the pins supply the friction level needed [39]. The
54 2 Sample Transfer
ceramic in this case was cast from machinable ceramic [40] and fired in a hobbyist’s
ceramic kiln.
The sample is transferred to a manipulator that has a central translating rod used
to drive the tilt mechanism. This tilting action is carried out by moving the central
manipulator rod up and down, causing the base holding the sample to tilt backward
and forward as desired. A locking nut on this rod prevents overdriving the tilting
mechanism. In the bottom portion of Fig. 2.10, two types of samples are shown. On
the left, a metallic foil substrate is welded to the screwhead ends of the mounting
pins in the base. On the right, a single crystal of a metal is welded to a circular piece
of foil which is welded to the mounting pins. A circular hole in the backing foil will
permit electron bombardment of the back of the crystal. One might provide a
radiation shield between the emitter filament and the ceramic base to reduce heating
of the ceramic and outgassing.
The proper location of viewports for monitoring the sample transfer is recom-
mended, or else two persons may be needed to transfer the sample.
The special considerations needed for the transfer of samples being used for scan-
ning tunneling microscopy (STM) studies as well as other more conventional surface
science studies have been reported by Sullivan and Kummel [41]. Here, a design
coupling the STM sample to a cold bath, electron beam heating, two degrees of
rotation, and electrical isolation, along with sample transfer are described, and a
review of other manipulator designs is given [41]. Noteworthy is a simple design for
transferring sample holders between forklike manipulators [42]. (See also p. 45.)
if this is unacceptable, then a mechanical modification to the design near the banana
plugs can be added.
In cases where ohmic heating of the sample is inappropriate, it may be backed
with a section of a Si wafer (approximately 0.1 Ω cm) for indirect resistive heating.
detector head (consisting of a polymer pyroelectric foil material) to the back side of
a sample experiencing adsorption from a molecular beam, as for example in a
microcalorimeter as described in Chap. 163. The sample is held in an
electrically-isolated Mo sample holder and the movable detector head may be
adjusted on a Cu platform to make contact with the back of the sample, using an
Allen wrench to engage a translation screw. The Allen wrench can be rotated using
a rotatable wobble stick. The entire assembly is mounted on a vibration damping
stack which rests on a plate on the bottom of the chamber. Thus, when the trans-
lation has been achieved and the Allen wrench is disengaged, the detector and the
attached sample is free of vibration [45].
The entry into an ultrahigh vacuum chamber can often be achieved through a
flanged entry port that contains a standard copper gasket. For large ports that permit
access to the entire chamber, one wall of the system may be designed to be
reversibly removable without a need for gasket replacement [47].
The design for a removable wall for a rectangular access port is shown in
Fig. 2.14. The port contains spring-loaded Teflon [48] seals in separate wells. The
regions between the Teflon seals are differentially pumped to prevent small leakage
from influencing the vacuum conditions in the chamber. Thus, an ion pump is used
on the pumping region closest to the chamber, and a roughing pump is used on the
pumping region closest to the atmosphere. The access port is arranged, by means of
a recessed section, to be pulled down tightly against the mating surface, providing
some further sealing at the very inner section of the seal.
2.10 UHV Access Port 57
References
1. D.R. Kahn, E.E. Petersen, G.A. Somorjai, J. Catal. 34, 294 (1974)
2. D.W. Blakely, E.I. Kozak, B.A. Sexton, G.A. Somorjai, J. Vac. Sci. Tech. 13, 1091 (1976)
3. G.A. Somorjai, Surf. Sci., 89, 496 (1979)
4. A.L. Cabrera, N.D. Spencer, E. Kozak, P.W Davies, G.A. Somorjai, Rev. Sci. Instrum. 53,
1888 (1982)
5. T.G. Rucker, K. Franck, D. Colomb, M.A. Logan, G.A. Somorjai, Rev. Sci. Instrum. 58, 2292
(1987)
6. D.W. Goodman, R.D. Kelley, T.E. Madey, J.T. Yates, Jr., J. Catal. 63,226 (1980); see also, D.
W. Goodman, J. Vac. Sci. Technol. 20,522 (1982)
7. D.W. Goodman, J.T. Yates Jr, J. Catal. 82, 255 (1983)
8. H.J. Krebs, H.P. Bonzel, G. Gamer, Surf. Sci. 88, 269 (1979)
9. H.P. Bonzel, G. Broden, H.J. Krebs, Appl. Surf. Sci. 16, 373 (1983)
10. C.T. Campbell, M.T. Paffett, Surf. Sci. 139, 396 (1984)
11. C.T. Campbell, J. Vac. Sci. Tech. A2, 1024 (1984)
12. Dr. Renato Zenobi, Analytical Chemistry, Universitätstr. 16, ETH Zentrum CHN C59,
CH-8092 Zurich, Switzerland (private communication)
13. T. Yamada, T. Misono, K.I. Tanaka, Y. Murata, J. Vac. Sci. Tech. A7, 2808 (1989)
14. C.A. Crider, G. Cisneros, P. Mark, J.D. Levine, J. Vac. Sci. Technol. 13, 1202 (1976)
15. J.P. Hobson, E.V. Kornelsen, J. Vac. Sci. Technol. 16, 701 (1979)
16. R.E. Clausing, L. Heatherly, L.C. Emerson, J. Vac. Sci. Technol. 16, 708 (1979)
17. J. Klebanoff, V.H. Ritz, R.E. Thomas, J. Vac. Sci. Technol. A2, 1396 (1984)
18. G.S. Chottiner, W.D. Jennings, K.I. Pandya, J. Vac. Sci. Technol. A5, 2970 (1987)
19. R. Raval, M.A. Harrison, D.A. King, G. Caine, J. Vac. Sci. Technol. A9, 345 (1991)
20. J.J. Vajo, W. Tsai, W.H. Weinberg, Rev. Sci. Instrum. 56, 1439 (1985)
21. S. Lundgren, K.E. Keck, B. Kasemo, Rev. Sci. Instrum. 65, 2696 (1994)
22. B. Kasemo, Rev. Sci. Instrum. 50, 1602 (1979)
23. (a) R.A. Campbell and D.W. Goodman, Rev. Sci. Instrum. 63, 172 (1992). (b) J. Szanyi and
D.W. Goodman, Rev. Sci. Instrum. 64, 2350 (1993). These articles also contain useful
references to other external high-pressure cell designs
24. Furon Co., 3336 E. Lapaima Ave., Anaheim, CA 92806
25. L.-W.H. Leung, T.W. Gregg, D.W. Goodman, Rev. Sci. Instrum. 62, 1857 (1991)
26. N.J. DiNardo, J.E. Demuth, W.A. Thompson, P.G. Ledermann, Rev. Sci. Instrum. 55, 1492
(1984)
27. Custom as well as standard heaters are available from Spectra-Mat, Inc., 100 Westgate Drive,
Watsonville, CA 95076. The version used here has W/Re alloy wires embedded in an alumina
matrix and housed in a Mo can
28. Aquadag is a colloidal graphite paint, which can be brushed onto a surface and then outgassed
by bakeout. It provides lubrication and is electrically conductive. Obtain Dag 154 from
Ache-son Colloids, 1607 Washington Avenue, P.O. Drawer 611747, Port Huron, MI
48061-1747
29. G.S. Chottiner, W.D. Jennings, K.I. Pandya, J. Vac. Sci. Technol. A5, 2970 (1987)
30. R. Raval, M.A. Harrison, D.A. King, G. Caine, J. Vac. Sci. Technol. A9, 345 (1991)
31. Y.-N. Wang, R. McAllister, R.G. Herman, G.W. Simmons, K. Klier, Rev. Sci. Instrum. 63,
5767 (1992)
32. J.J. Weimer, F.A. Putnam, Rev. Sci. Instrum. 55, 238 (1984)
33. K.-E. Keck, B. Kasemo, A. Hoglund, Rev. Sci. Instrum. 54, 574 (1983)
34. M.J. Mummey, L.D. Schmidt, Surf. Sci. 91, 301 (1980)
35. A. Ludviksson, J. Yoshihara, C.T. Campbell, Rev. Sci. Instrum. 66, 4370 (1995). This paper
also contains references to many other transfer devices in the literature
36. Obtained from: Fluorocarbon Co., 10871 Kyle Street, P.O. Box 520, Los Alamitos, CA 90720
37. Gortiflex Protector Bellows
References 59
38. No. UNS C95400, Copper and Brass Sales, 17401 Ten Mile, East Detroit, MI 48021
39. A. Garcia, M.E. Kordesch, J. Vac. Sci. Tech. A11, 461 (1993)
40. Aremco, P.O. Box 429,23 Snowden Avenue, Ossining, NY 10562-0429
41. D.J.D. Sullivan, A.C. Kummel, Rev. Sci. Instrum. 63, 4285 (1992)
42. B.J. Mulder, J. Phys. E12, 908 (1979)
43. Dr. Ray Kaplan, Mail Code 6863, Naval Research Laboratory, Washington, DC 20375
(private communication)
44. UHV banana plugs (Model Nos. BPBPM and BPBPF) can be obtained from: Surface/Interface
Inc., 110 Pioneer Way, Suite H, Mountain View, CA 94041
45. S. Schauermann, private communication
46. Kejing Li, private communication
47. D.J. Auerbach, C.A. Becker, J.P. Cowin, L. Wharton, Rev. Sci. Instrum. 49, 1518 (1978)
48. The spring-loaded Teflon seals are available from the Fluorocarbon Co., 10871 Kyle Street,
RO. Box 520, Los Alamitos, CA 90720. Many sizes of the seals are available, and instructions
for their use are given by the company
Chapter 3
Electrical Connections
Heated tungsten ribbons are used in many experiments as radiation sources, metal
deposition sources, electron bombardment sources, and molecular cracking sources.
Most commonly, the ribbons are attached to heating leads by spot welding. This
method requires that the mounted power leads be removed from the vacuum system
for ribbon replacement, a time-consuming procedure. An alternative method for
ribbon attachment shown in Fig. 3.1 may be carried out inside the vacuum system
without welding [1].
A Ta feedthrough lead has a slot milled into it as shown in Fig. 3.1a. The length
of the slot is about 1.5 times the ribbon width, and the width of the slot is about 1.5
times the ribbon thickness. A small cap is machined from Ta, for example, as
shown in Fig. 3.1b. The inside diameter of the cap is larger than that of the
feedthrough lead by one thickness of the ribbon. The cap length is three to four
times that of the ribbon width, and the depth of the bore is 2.5 times the ribbon
width, leaving plenty of material in the slotted upper region. The cutout on the cap
is centered on the cap so as to allow the ribbon to pass as shown in Fig. 3.1c. The
end of the lead extends internally above the cutout by about 0.5 mm, as shown by
the dotted lines in Fig. 3.1c, preventing the cap from tilting when rotated.
To assemble the device, the ribbon is passed through the slot in the lead with the
cap in place, and extends about one-half the lead diameter beyond the cap. To
secure the ribbon, the cap is rotated one-half turn as shown, wrapping the ribbon
around the lead under the cap, and locking it in place.
This method will work on very thin W ribbon and has been extended to
0.010-mm-thick ribbon material, 1.2 mm wide; this particular mounting was done
on 1.5-mm-diameter Ta leads. To keep the ribbon straight after it has been used, it
is common to strain the two leads toward each other before making the second cap
attachment. The strain in the two supporting leads will place tension on the ribbon
and keep it straight.
Fig. 3.2 Wrap connection—Tungsten to Tungsten. a Starting coil. b During winding. c Final coil
—slightly bent in middle
axis of rotation. A wood-forming clamp is next placed on the wrapping wire coil,
and while under pressure, the coil is formed to the desired length. The friction
between the wood and the W will prevent fracture. The ends of the wrap wire coil
can now be cut to the desired length using tungsten carbide end nippers and
dressing on a grinding wheel if necessary.
The coil is removed from the core wire, and a slight bend is placed in the center
of the coil, using either two pairs of pliers or a knife blade inserted partially into the
coil. This slight bend allows the coil to be forced onto a 0.5 mm W wire and to
remain fixed owing to the strain induced in the coil as it is straightened by the inner
wire. The coil will not move at high temperature either during ohmic heating or
bakeout.
64 3 Electrical Connections
The fabrication of electrical connections that are conductive for both electricity and
heat may be carried out using silver soldering, welding, or screw-type connectors.
Another method, involving a press fit of a hard wire into a soft material such as Cu
or Al, is shown in Fig. 3.3. Here a vise is used to press the junction, and the various
stages of compression are illustrated. The last stage (d) involves deformation of the
soft metal and enclosure of the hard wire. The resistance of these junctions between
tungsten and copper is typically about 0.05 Ω [3].
sapphire. The assembly on the vacuum side is provided with a second shield as
shown, and also shielding of the outer air side of the feedthrough is carried out as
shown.
The impedance to ground of the sapphire insulator approached 1016 Ω when the
sapphire was thoroughly cleaned [4]. Spectroscopic grade alcohol, followed by
distilled water spraying through a fine glass nozzle, is recommended, followed by
blotting the excess water and drying in an oven.
A need for a sliding electrical contact can occur when flexible coils of ceramic
bead-insulated wire are unsatisfactory, perhaps because of the formation of ceramic
dust as motion occurs. Figure 3.5 shows a sliding stainless steel rod that moves
through a Teflon bearing that is mounted inside a re-entrant stainless tube connected
to a vacuum linear drive of the bellows or magnetic type. Electrical contact to the
rod is made by a phosphor bronze spring contact rigidly mounted on a stainless
support assembly, and electrically isolated by ceramic insulator cylinders that
contain small bolts for holding the spring contact strip. The phosphor bronze spring
strip is electrically connected to an electrical feedthrough as shown. The holes
drilled in the Teflon-bearing should be about 5 % larger than the sliding rod to
allow for deformation of the Teflon as well as for its large thermal expansion
coefficient. The contact can pass a 2-A current. The stainless steel rod can be moved
in and out at 5 cm/s without difficulty, and cold welding in ultrahigh vacuum does
not occur [1]. The authors also report that a Ta wire spring may be used success-
fully against a Mo slider in a rotary device [6].
There is often a need for a band clamp arrangement to anchor electrical leads to a
manipulator rod or other structure. Often, narrow clearance spacing is available
around the rod so that the binding on the electrical leads to the rod must be achieved
in a limited annular space. Figure 3.6 illustrates one solution to this problem, using
half-hard Cu-Be spring stock. This material is formed as desired by using
steel-forming wires of the correct diameter to produce the rounded grooves in the
Cu-Be stock by hammering the forming wire into the stock with the forming jig as a
guide to fabrication. An extra channel is provided as an additional spring expansion
channel to give the band extra elasticity. The tabs at the end of the band are each
about 3–4 mm in length. To install, the ceramic insulated wires are placed into
position on the rod and temporarily anchored. Overlapping the Cu-Be tabs and
folding them into each other will result in a tension holding the band and the
ceramic insulated wires in place. The band will not soften or lose its spring quality
from −200 °C to bakeout temperatures [7].
References
1. Mr. Gerald L. Fowler, 6816 Kelly Ave., N.E., Albuquerque, NM 87109 (private
communication)
2. Mr. G.L. Fowler, 6816 Kelly Ave., N.E., Albuquerque, NM 87109 (private communication)
3. D. Edwards Jr, J. Vac. Sci. Tech. 16, 970 (1979)
4. A.H. Madjid, J.M. Martinez, A.A. Vaala, Rev. Sci. Instrum. 43, 1049 (1972)
5. Obtained from: Cory Lab, Inc., 823 5th Street, Menominee, MI 49858
6. S.M. Newstead, R.A.A. Kubiak, E.H.C. Parker, J. Vac. Sci. Tech. A2, 1603 (1984)
7. G.L. Fowler, J. Vac. Sci. Tech. A5, 2978 (1987)
Chapter 4
Pumping and Trapping
Getter films of active metals have been used for many years in ultrahigh vacuum
technology, especially in the days when work was carried out in glass vacuum
systems that were sealed off from dynamic pumping [4]. This method of pumping
chemically active gases is superior to all other methods insofar as secondary effects
are concerned. The principle of getter film pumping is the strong irreversible
chemisorption of chemically active gases by clean active metal surfaces. Simple
considerations of monolayer capacity indicate that 1 cm2 of a getter surface will
pump of the order of 1–3 × 10−5 Torr L of an active gas. Film roughness factors of
2 or 3 will increase the gettering capacity accordingly.
One example of the use of a Ti sublimation pump (TSP) for pumping low flux
gases from a differentially pumped mass spectrometer is shown in Fig. 4.3. Here, a
thin film of Ti is deposited on the stainless steel wall of a cylindrical appendage
each morning before beginning work. The Ti film is kept at 77 K to maintain its
pumping efficiency to the highest level. A single Ti film will remain active all day
during use of the mass spectrometer at low gas loads. This particular apparatus also
contains a 60-L/s diode ion pump for maintenance of vacuum quality in the mass
spectrometer overnight when the titanium pump is not being used [5].
Careful studies, in which the Ti getter film pump has been compared with ion
pumping, show that regurgitation effects [6] involving methane liberation from the
ion pump are completely eliminated when the Ti pump is employed [5]. This may
be seen by looking at the thermal desorption spectra of methyl chloride from an
inert surface. The use of the TSP eliminates the CH4 evolution originating in the ion
pump. However, wall displacement effects caused by molecules entering the mass
spectrometer region from the single crystal when it is heated in thermal desorption
experiments can still be detected. These displaced molecules were H2, CO, and H2O
in the particular experiments carried out [5].
The detailed behavior of the residual gases in a liquid nitrogen-cooled titanium
sublimation pumping system has been reported [7].
Viton O-ring seal rings. The O-ring seal ring is brazed to a disk containing two
holes, and the thin metal sheets are mounted on opposite sides of the disk in order to
relieve pressure differences in either direction.
74 4 Pumping and Trapping
There is sometimes a need to fill a vacuum system with a high-pressure inert purge
gas, and then to recover ultrahigh vacuum. Diode ion pumps have only a limited
pumping speed for inert gases such as He, and their performance is of interest in
this application; triode ion pumps are designed to have good pumping speed for
inert gases at a small sacrifice in total pumping speed.
In Fig. 4.5, a series of pumping tests were performed on an ultrahigh vacuum
system that could be filled with high-purity He at 1 atm pressure. The He was
pumped out with a trapped mechanical pump to about 1 × 10−5 Torr, and then the
diode ion pump was employed. Figure 4.5 shows selected pumping performance
curves after the first He filling and the fifth. The performance degrades in successive
tests because of traces of water vapor admitted with the high-purity He and also
because of a drop in efficiency of the diode ion pump owing to the regurgitation
effect in which previously pumped gas is partially evolved in successive pumping
periods. Baking out (BO) the ion pump (IP) only for 4 h at 250 °C restores some of
the original pumping speed; baking out the ion pump plus the vacuum system
restores all of the original performance seen in the first test.
The behavior shown in Fig. 4.5 indicates that He can be pumped well by a diode
ion pump from ultrahigh vacuum systems made of stainless steel using highly
purified He as a purge gas, and that short bakeouts at 250 °C restore system
pumping performance. It was suspected that residual water in the He (1.9 ppm) was
partially responsible for degradation of the pumping behavior in successive tests
[9].
Copper [10], nickel [10] and aluminum [11] foams are commercially available and
may be used in a number of vacuum applications [12]. The foams are highly
porous, they are easily shaped and machined, and they have some inherent strength.
They also can be easily brazed or soldered.
Four vacuum applications of foamed metals are illustrated in Fig. 4.6. In panel
A, a plug of foamed Cu is used as a hydrocarbon trap between a forepump and a
vacuum system. The high surface area and the torturous pathways in the foam
impede hydrocarbon passage through the trap. In panel B, a zeolite trap is retained
between porous Cu plugs, and serves also as a hydrocarbon trap. The zeolite must
be baked out prior to becoming active and acts as a molecular sieve for hydrocarbon
molecules. Panel C shows a plug of Cu foam used as a conductance limiter under
molecular flow conditions. To adjust the conductance accurately, additional Cu
foam segments can be added as needed. Finally, in panel D, a cylindrical piece of
metal foam is used as a diffuser for injection of gases in random directions into a
vacuum system.
76 4 Pumping and Trapping
Fig. 4.5 Test chamber and pumping behavior after He purge (1 atm)
4.6 Vacuum Applications of Metal Foams 77
Fig. 4.6 Vacuum applications of metal foams. a Foreline hydrocarbon trap. b Foreline zeolite
trap. c Molecular flow conductance limiter. d Gas diffuser
Metal foam has a cellular internal structure. Low-density foam has cell diameters
in the range 0.5–1.5 μm, whereas high density foam has cell diameters from 9 to
40 μm. Typical surface areas range from 0.025 to 0.038 m2/g. For low-density foam
the density is typically from 2 to 5.5 % of that of the parent metal. For high-density
foam, the density ranges from 20 to 60 % of the bulk metal. Low-density foam can
be compressed by up to a factor of 7.
Measurements of the conductance of 2.4 mm thick disks of 5 % density Cu
foams made by compression of 1.9 % density foam gave a conductance (for air?) of
3.2 L/s cm2.
78 4 Pumping and Trapping
There are a number of companies that will take in used ion pumps for recondi-
tioning at a fraction of the cost of purchasing a new ion pump [13]. In cases where
the pump is a large one, this may be done by disassembly and cleaning or
replacement of the individual components. In the case of small ion pumps, it may
be necessary to open the pump casing to get at the inner components, and then to
reweld the reconditioned components back into the cleaned casing.
The reconditioning of a dirty ion pump can also be done by controlled chemical
etching procedures as described for two cases below.
Case A: Chemical Etching After Disassembly
The pump casing is separated from the Ti cathode assembly as shown in Fig. 4.7.
The surface region of the Ti containing implanted impurities is removed by etching
the assembly, containing electrode connectors and supports, in a water solution
containing 9 % HNO3, 12 % HF(aq), and 0.4 % H2O2 by weight. A 3-s dip,
followed by repeated rinsing in deionized water, should produce shiny Ti surfaces.
The pump casing is etched in a standard solution (Varian) containing 30 %
Liquid nitrogen traps are highly inefficient because of the discharge of liquid
nitrogen out of the trap when the trap is being refilled periodically. A design, shown
made of glass in Fig. 4.8, eliminates this problem [15], and reduces liquid nitrogen
consumption by severalfold, resulting in great economies over long periods of time.
The basic concept embodied in this small glass trap is transferable to large metal
traps.
The trap consists of an inner trap of conventional design, labeled the inner
chamber, surrounded by a second outer chamber, which is in turn surrounded by a
silvered Dewar flask. All of these are sealed together by glassblowing in this design.
A resistance temperature sensor is immersed in the second outer chamber.
When the trap is at a condition where refilling is to occur, all liquid nitrogen in
the Dewar has been lost, and the level of nitrogen in the outer chamber has dropped
to a level causing the sensor to turn on the liquid nitrogen supply, which is driven
by compressed nitrogen gas. Liquid nitrogen flows into the Dewar and spills into
the outer chamber through the lower hole; it then fills the outer chamber until the
sensor delivers a shutoff signal.
The avoidance of liquid nitrogen loss is attributed to the fact that the Dewar does
not contain liquid nitrogen at the time of filling, so that the burst of nitrogen gas at
the beginning of the fill sequence does not eject liquid nitrogen from the trap. In
addition, ice buildup does not occur at the inlet and outlet to the trap [15].
80 4 Pumping and Trapping
References
6. P.A. Redhead, J.P. Hobson, E.V. Kornelsen, The Physical Basis of Ultrahigh Vacuum (AIP,
New York, 1993), pp. 403–406
7. L.Y.L. Shen, Rev. Sci. Instrum. 43, 1301 (1972)
8. Dr. R. David, Institut fur Grenzflächenforschung und Vakuum physik, Forschungszentrum
Jülich GmbH, Postfach 1913, D-52425 Jülich, Germany (private communication)
9. P.M. Danielson, J. Vac. Sci. Technol. 5, 89 (1968)
10. Foametal, Inc., 37645 Vine Street, Willoughby, OH 44904
11. Energy Research and Generation, Inc., 900 Stanford Avenue, Oakland, CA 94608
12. B.R.F. Kendall, J. Vac. Sci. Technol. 17, 1385 (1980)
13. One company offering this service: Duniway Stockroom, Inc., 1600 N. Shoreline Boulevard,
Mountain View, CA 94043
14. P.H. Schmidt, L.Y.L. Shen, Rev. Sci. Instrum. 44, 1788 (1973)
15. R.S. Timsit, A.K.C. Kiang, Rev. Sci. Instrum. 44, 770 (1973)
Chapter 5
Bakeout
The removal of adsorbed impurities from the walls of an ultrahigh, vacuum system
is carried out by heating the walls uniformly while pumping the system. Generally,
for stainless steel systems this is done in the temperature range 150–200 °C. For
preservation of effective ion pumping, bakeout into a turbopump is recommended
to avoid introducing a high gas load into the ion pump. Ion pumps have only a finite
capacity, and also suffer from regurgitation effects in which previously pumped gas
is evolved when another gas is being pumped-as would occur in an experiment in
which gas is admitted to the chamber.
Many methods are used to bake out metal systems. Heating tapes, IR radiation
heaters on the outside, clamp-on band heaters, glue-on rubberized heaters, and
internal light-bulb heaters have all been used. In order to lose as little heat as
possible, it is appropriate to surround the system with some sort of reflective
bakeout shield [1]. Figure 5.1 shows a schematic system in which aluminized
fiberglass [2] is used to shield the main system. The fiberglass tent is supported
internally by temporary metal supports. Radiative heaters are uniformly distributed
over the system inside the reflective shield, and for appendages such as the bellows
on the top, lightweight reflective metal cylinders may be used for radiation capture.
The system should be monitored over many positions with thermocouples to assure
uniform heating, and the wattage output of the various radiative heaters may be
varied to achieve uniformity. Experience in monitoring the pressure during bakeout
will supply information about the bakeout time needed to reach the ultimate low
pressure [3]. It is reported that external IR lamps bring the vacuum system to
bakeout temperatures much faster than other methods [1].
Quite nice tents made, of the aluminized fiberglass may be constructed, and a
convenient method for joining the sections is to use a poprivet device to make
excellent lap seams. The aluminum surface is placed on the inside of the tent. Even
nonaluminized fiber glass curtain material may be used by sandwiching aluminum
foil between the layers. A small internal circulating fan inside the oven will lead to
improved temperature uniformity.
Experience indicates that clamp-on band heaters applied to metal flanges can
thermally shock the flanges to a point where a leak develops. In the event that these
heaters are used, a gradual increase in the applied voltage over time is recom-
mended. Internal lamp heaters, working inside the vacuum system, would seem to
be ideal for bakeout because of efficient radiation capture by the system walls, but if
breakage of the glass envelope occurs, the system will have to be completely
cleaned of glass slivers before continued use. Such slivers are very detrimental to
welded bellows and to the sealing surfaces of valves.
Turbopumps should be baked at the end of the bakeout period (for about 4 h).
Titanium sublimation pumps should be outgassed until the pressure reading reaches
a maximum; long outgassing at high temperature will oxidize the components near
the hot Ti. It has been found that cooling the room during measurements will cause
a substantial improvement in the limiting pressure under ultrahigh vacuum condi-
tions owing to the reduction in the outgassing rate from the system walls [4].
5.1 Bakeout of Metal Ultrahigh Vacuum Systems 85
Often it is appropriate to bake, out a new system, with all heat sensitive com-
ponents removed, to temperatures between 300 and 450 °C to remove impurities
left from chemical cleaning procedures.
References
1. B.V. Hess, T.E. Felter, J. Vac. Sci. Tech. A3, 2218 (1985)
2. Fyrepel, P.O. Box 518, Newark, OH 43055
3. R. David, Institut fur Grenzflächenforschung und Vakuum-physik, Forschungszentrum Jülich
GmbH, Postfach 1913, D-5170 Jülich, Germany (private communication)
4. Pittsburgh Surface Science Center
Chapter 6
Behavior of UHV Systems
Fig. 6.1 Nitric oxide surface reactivity for passivation of vacuum system walls
The purity of gases used in surface science adsorption and surface reaction studies
is of paramount concern just as the production of ultrahigh vacuum is directed
toward the same end. Often gas dosers are employed to deliver pure gases to
surfaces, and for some gases, the displacement of adsorbed impurities, or the
reaction of the dosing gas with the internal surfaces of the gas dosing system itself
can be a problem. In the case of a surface reaction, it is sometimes possible to
passivate the internal surface of a gas doser to reduce the rate of the surface
reaction. Figure 6.2a shows a typical gas doser made mainly of stainless steel. The
construction of this doser and its calibration are discussed elsewhere (p. 421);
however, it may be seen that the conductance is controlled by a laser-drilled
aperture, and collimation of the exiting gas is achieved using a microcapillary array
made of glass. Both the internal metal surface and the internal surface of the
microcapillary array (which is many tens of cm2) provide reactive surface area that
can evolve impurities by displacement or reaction [3], Displacement processes,
such as the exchange of surface OH groups with OD groups in deuterated alcohols,
can often best be handled by overnight flushing of the doser with the isotopic gas.
In extreme cases of reaction with acid sites on the internal surfaces of the capillaries
[3], one must remove the glass microcapillary array from the doser [3, 4]. Reaction
processes with the stainless steel walls of the doser system can sometimes be
reduced in magnitude by passivation of the internal surfaces. This is vividly seen in
the behavior of the metal carbonyls, where decomposition of Fe(CO)5 on the inner
stainless steel walls has been shown to produce CO(g), leaving iron on the wall [4].
The surface reaction was studied, and it was found that predosing the doser and its
stainless steel delivery system with high fluxes of molecular oxygen resulted in
significant reduction of the decomposition reaction, as judged by mass spectro-
metric studies of the gas exiting the doser, line-of-sight, into the mass spectrometer
[4]. The internal walls of the ultrahigh vacuum system were found to have little
effect on the metal carbonyl, probably because it was exposed to oxygen each day
[4]. For the experiments, it was found that the glass microcapillary usually
employed with this design of gas doser had to be removed because of deleterious
effects on Fe(CO)5.
Another solution to this problem is to make the doser system from materials
other than stainless steel [5]. Figure 6.2b shows such a system, in which the glass
tube doser contains a capillary for the control of the flow rate from the connecting
tube into the ultrahigh vacuum system through the Pyrex doser tube. The
high-pressure (several Torr) section beyond the Teflon-glass valve is made of PFA
(polyfluoroalkoxy) plastic and fittings are readily available commercially [6]. The
paper describing this system indicates that the capillary tube was produced by hand
and was drawn down to a fine diameter. This particular design was used for the
delivery of hydrazine (N2H4). Thermometer tubing, made generally of lead glass,
has a *100–150-μm diameter capillary for sensitive thermometers (1 K/3.8 cm
90 6 Behavior of UHV Systems
Fig. 6.2 Gas doser designs. a Stainless doser+microcapillary array collimator. b Pyrex—plastic
doser. c Cleaning clogged glass capillaries
capillary length) and may be used if a lead glass to Pyrex glass graded seal is
employed. A Teflon doser for ultrahigh vacuum use also has been described [7], as
well as a heatable glass and Teflon doser for admission of organometallics of low
vapor pressure [8].
It is unfortunate that glass capillaries often clog up with traces of dust or grease
that are present within the vacuum system. A clever method to restore the capillary
to a clean, unblocked condition is shown in Fig. 6.2c [9]. Here, the capillary section
is surrounded on both ends of the capillary with pointed tungsten electrodes, easily
installed by the glassblower. When the capillary becomes blocked, micron pres-
sures are established on the high-pressure side of the capillary, and the vacuum-side
6.2 Minimizing Wall Reactions in Ultrahigh Vacuum Systems … 91
References
1. P.A. Redhead, J.R. Hobson, E.V. Kornelsen, The Physical Basis of Ultrahigh Vacuum (AIP,
New York, 1993)
2. M. Grunze, G. Strasser, O. Eishazly, J. Vac. Sci. Technol. A4, 2396 (1986)
3. P.K. Leavitt, P.A. Thiel, J. Vac. Sci. Technol. A8, 148 (1990)
4. M.A. Henderson, R.D. Ramsier, J.T. Yates Jr, J. Vac. Sci. Technol. A9, 2785 (1991)
5. E. Apen, R. Wentz, F. Pompei, J.L. Gland, J. Vac. Sci. Technol. A12, 2946 (1994)
6. Swagelok fittings are available from Swagelok, 31400 Aurora Road, Solon, OH 44139
7. M.A. Mendicino, E.G. Seebauer, J. Vac. Sci. Technol. A10, 3590 (1992)
8. L.A. Jones, F.-X. Wei, D.F. Thomas, Rev. Sci. Instrum. 66, 4981 (1995)
9. D.A. King, (1970) (private communication)
10. E.C. Henn, M.E. Bussell, C.T. Campbell, J. Vac. Sci. Technol. A9, 10 (1991)
11. W. Muller-Markgraf, M.J. Rossi, Rev. Sci. Instrum. 61, 1217 (1990)
12. Professor E.G. Seebauer, Department of Chemical Engineering, University of Illinois, Urbana,
IL 61801. From thesis by M.A. Mendicino, Ph.D (1994) (private communication)
13. Teflon or PTFE suspension is obtainable from: Intek Services, Inc., 7 McMillan Way, Suite
101, Newark, DE 19713
Chapter 7
Mechanical Action on Samples
Often one of the best ways to produce an atomically clean surface is to expose an
inner cleavage plane by cleaving the crystal in ultrahigh vacuum. A simple
mechanical cleaver design is shown in Fig. 7.1a [1]. Here, a MgO single crystal is
mounted on a coolable manipulator inside a small rectangular compartment, held by
a screw that presses on an internal plate that distributes the screw pressure over the
side of the crystal. A portion of the crystal projects upward, ready for the cleaver
blade.
The cleaver is a spring-loaded tungsten carbide blade (25° cutter angle) that
slides in a stainless steel channel. Two springs are compressed by turning a pair of
screws prior to attaching the cleaver to the manipulator. The manipulator/cleaver
assembly is loaded into the vacuum system, and after system processing, the trigger
holding the blade back is released by moving the manipulator against the side of the
manipulator port. The blade cleaves the crystal, and caution should be exercised not
to allow the cleaved-off portion of the crystal to enter fragile apparatus as it bounces
around the chamber.
Using a closed cycle He cryocooler, this assembly can reach 44 K. Studies on
the freshly cleaved MgO(100) surface, involving Kr adsorption, are shown in
Fig. 7.1b, and the results are compared to MgO that has been cleaved outside the
vacuum system, transferred, and thermally processed. The sharp vertical step
indicating a 2D gas → 2D solid transformation, as studied by LEED, is charac-
teristic of the UHV-cleaved crystal; a less distinct transition is seen on the exter-
nally cleaved crystal because of defects produced by water adsorption that cannot
be removed by thermal processing. Indeed, after days of storage in UHV, the
UHV-cleaved crystal begins to yield an isotherm resembling that of the defective
crystal because of slow impurity adsorption.
Fig. 7.1 Simple UHV cleaver. a Cleaver design. b Comparison of Kr isotherms on MgO(100)
using LEED
7.1 Simple UHV Crystal Cleaver 95
The authors report that the only disadvantage of this design is that only one
cleave can be achieved per loading into the vacuum system because of the necessity
to compress the springs outside the vacuum system.
A cleavage device for mica has been described that can be used in ultrahigh
vacuum [2].
There is much interest in the role of adsorption on the rate of fatigue of metals
undergoing periodic stress conditions. Usually these measurements are performed
under cycle rates of about 1 Hz. Hydraulic cycling methods often impart noise to
the stress-strain measurement system. A piezoelectric drive mechanism, capable of
working at 100 Hz frequency in UHV, is described below [3].
Figure 7.2a shows a schematic drawing of the apparatus. The specimen, coupled
to an extensomer, is clamped between two grips that have serrated clamping sur-
faces to prevent misalignment during testing. A load cell is placed above the
moving grip and below the piezoelectric actuator. The translator used has a capacity
of 10 kN in compression, and 3.5 kN in tension [4]. It is capable of translation
through about 40 μm over an applied voltage range of 0–1500 V. The load is
measured with a load cell having a range of 0–2000 N [5]. A strain of about 0.3 %
may be measured using an extensometer [6] in a gauge section of 5 mm, and the
specimen tested in this case was polycrystalline Ag. The control electronics are
described in [7]. Figure 7.2b shows a hysteresis curve measured for Ag in which the
stress and strain are measured during compression and expansion. The hysteresis
loop changes as the number of cycles increases, and is a measure of the specimen
fatigue. The data obtained with the piezodriver exhibits a better signal-to-noise ratio
than data taken with hydraulic devices.
The main disadvantage of the device is the small translation range of the driver.
The device must account for all of the translations taken up by the components
along the axis of the load. This limits the machine to small specimens, and for
high-strength materials, smaller gauge lengths and more narrow cross sections
would be needed to obtain sufficient strains for accurate measurement [3].
Another design, providing fatigue to a thin metal specimen in ultrahigh vacuum,
has been described [8]. It was found that clean Al(111) single crystals exhibited
changes in their ultraviolet photoelectron spectrum after 1000 fatigue cycles at
0.2 % strain. The involvement of chemisorbed species containing oxygen was
eliminated on the basis of the absence of O(2p) emission in the photoelectron
spectrum, and no explanation was given for this effect [8].
96 7 Mechanical Action on Samples
Fig. 7.2 Piezoelectric fatigue apparatus—UHV. a Schematic design. b Typical stress-strain curve
—polycrystalline Ag
References 97
References
1. T. Angot, J. Suzanne, J.Y. Hoarau, Rev. Sci. Instrum. 62, 1865 (1991)
2. K. Reichelt, Rev. Sci. Instrum. 44, 243 (1973)
3. T.S. Sriram, M.E. Fine, Y.-W. Chung, Rev. Sci. Instrum. 62, 2008 (1991)
4. Physik Instrumente, Model 243.37, Option 703.20, sold by PolyTec Optronics, 3001 Redhill
Ave., Costa Mesa, CA 92626
5. Model 31, SENSOTE Inc., 1202 Chesapeake Ave., Columbus, OH 43212
6. Model 632.29C, Option 011, MTS Systems Corporation, Box 24012, Minneapolis, MN 55424
7. W.J. Lee, Y.-W. Chung, M.E. Fine, J.P. Baker, Rev. Sci. Instrum. 57, 2854 (1986)
8. T.K.G. Swami, Y.W. Chung, Rev. Sci. Instrum. 51, 842 (1980)
Chapter 8
Gasket Seals
The primary ultrahigh vacuum flange design used universally is the stainless steel
Conflat sexless flange, which operates with two opposed knife edges that seal a
metallic gasket by deformation when the flange is tightened. A cross-sectional view
of the flange and a confined gasket is shown in Fig. 8.1.
In many cases where it may be difficult to seal the gasket by applying the
necessary torque to the flange bolts, or where tightening down to the maximum
force level is not desired, it is useful to use annealed copper gaskets. These maybe
purchased from the suppliers, but they can be made quite simply in the laboratory
beginning with a nonannealed gasket. The gasket is heated while rotating it around
its circumference in an oxygen-gas torch (cool flame) until dull red color is emitted
by the hot copper. The gasket is then quenched in water quickly, resulting in
softening to almost the softness of lead or tin. This gasket will deform in the seal
region with less torque on the bolts and also with less deformation of the flange
during tightening. Often, gaskets of this type are needed for spectroscopic windows
mounted in stainless steel flanges in order to prevent window cracking during
tightening owing to flange flexure.
The annealing process results in considerable oxidation of the copper gasket, and
this oxide layer must be removed before use. Two methods are available to do this.
In the first, the hot gasket, immediately after annealing, is dropped into methanol
instead of water. The methanol, a reducing agent, immediately strips off the oxide
from the hot metal, leaving a clean copper surface. Care should be exercised to
avoid damage if the methanol should ignite. The second method involves dropping
the cool gasket into an aqueous solution of citric acid. Here the citric acid forms a
soluble complex with the copper ions in the oxide film, and again clean copper
surfaces are produced.
Certain coatings are employed to prevent the oxidation of copper gaskets fol-
lowing relatively high temperature bakeouts. Heavy amounts of copper oxide can
Fig. 8.1 Gasket designs. a Conflat flange + Cu gasket. b Gold wire—shear seal
fall into the vacuum system when the flange is opened, and also seizure of the
gasket in its recess in the stainless steel flanges can occur. Silver plating has been
used to prevent this effect [1]. A minimum 5-μm-thick silver film, made by plating,
is required. The silver will stand repeated bakeouts at 300–350 °C. Contamination
of the silver film by sulfur from either the gasket or from sulfur-containing lubri-
cants is to be avoided as this causes discoloration and peeling of the silver film.
Thus, bolt lubricants such as molybdenum disulfide cannot be used and graphite or
other lubricants must be employed.
Another useful copper gasket coating is titanium nitride, TiN [2]. Using reactive
magnetron sputtering from a Ti target in an N2 + Ar mixture (20 % N2), films from
50 to 150 Å were deposited on copper gaskets. In tests, ten flanges were baked at
550 °C in a vacuum of 10−5 Torr for up to 200 h, and were then tested for leak
8.1 Gasket Seals for Ultrahigh Vacuum Systems 101
tightness with a He leak detector, finding no leaks. The flanges could then be
disassembled without seizure. Without the TiN film, separation must be done with
jacking screws and copper is left on the sealing surfaces, requiring abrasives for
removal. A similar coating has also been prepared using diamondlike carbon
coatings prepared by RF discharge techniques in methane gas [3].
One of the early methods of sealing stainless steel flanges employed gold wire
gaskets [4]. For flat flanges without a knife edge or for a stepped seal used for
crushing a metal gasket, gold wire will make an excellent seal. The wire, of about
0.25–0.5 mm diameter, is made into a circular loop on a forming mandrel; then a
microtorch is used to butt weld the ends together when the loop is removed from the
mandrel. This process is best accomplished by gently clamping together the squared
off ends of the gold wire and slowly melting the gold so that surface tension causes
liquid gold flow to even out the diameter of the heated region. Too much heating
will result in separation of the gold wires with the formation of balls of gold on the
ends of each. Practice is, therefore, necessary. The wire once formed is placed
between two smooth flange faces, and the bolts are tightened, making an ultrahigh
vacuum seal. A stepped seal may also be used with either gold or copper, and here
the seal produces a shear seal within the gasket as shown in Fig. 8.1. In the case of
gold wire shear flanges, the wire must be butt-welded to fairly exact dimensions
using a mandrel of the same size as the male flange for accurately cutting the gold
wire to appropriate length. A nice description of the method to make custom copper
wire gaskets is given in [5].
The reuse of large Cu gaskets for many cycles has been described [6]. Here, a
“feeler gauge” such as that used to measure spark plug gaps is employed, and
successive seals with the original gasket are made by diminishing the
flange-to-flange clearance in a systematic manner.
The specially designed stainless steel flange has a precision entrance guide to
prevent damage to the aluminum knife edge and it is also machined to be very
smooth on its sealing surface. In the particular case shown here, the stainless steel,
exhibiting poor heat conduction, was needed to reduce the thermal conduction of
heat to the aluminum section therefore preserving liquid nitrogen which was housed
in the aluminum section.
There are three published methods that are directed toward easy removal of copper
gaskets from Conflat flanges after disassembly [8–10]. These methods are all
designed to prevent damage to the sharp sealing knife edges by tools being used to
pry up the gaskets. The first of these [8], shown in Fig. 8.3a, employs Conflat
flanges that have two opposed slots machined on the inside of the flange.
A screwdriver blade can be inserted into the slot and rotated to pop out the gasket
and the boundary of the slot prevents the blade from coming into contact with the
knife edge. The disadvantage of this design is that one must machine every flange
with the two slots [9]. A second design [9] (Fig. 8.3b), uses the leak-check groove
as a slot to control the motion of a specially wedge-shaped and hardened steel
blade, shown. The third design (Fig. 8.3c), shows a modified “parrot-nose” set of
8.3 Copper Gasket Removal Devices 103
Fig. 8.3 Tools for copper gasket removal. a Inside slot design. b Leak-check groove tool.
c Custom pliers
ignition pliers that can safely grip the gasket [10]. This design allows only upward
force on the gasket and also grips the gasket during removal to prevent it from
falling. In certain instances where there is insufficient clearance for the pliers on the
outside of the copper gasket, one must resort to a tool of the second design [9, 10].
Several unconventional compression seals have been designed for ultrahigh vac-
uum systems where flange mounting was inappropriate. These seals all have in
common the use of a material softer than stainless steel that is deformed under
pressure supplied by screwing down the re-entrant seal from the outside of the
vacuum system. This forms the vacuum seal, which is always on the inside of the
104 8 Gasket Seals
Fig. 8.4 Unconventional compression seals. a Hollow bolt seal. b Mini-Conflat seal. c Gauge
tube seal
8.4 Unconventional Compression Seals for Ultrahigh Vacuum 105
produce the seal at the copper gasket is applied externally by means of jacking
screws. The assembled seal was designed to apply 2000 lb/in. along the copper
gasket seal path. The seal is assembled from inside and initially tightened using the
threaded jack ring, which pushes down on a chrome plated thrust washer needed to
prevent distortion of the main sealing flange that contains the hole and also to
prevent galling of the jackbolts under torque. First torquing of ten bolts to 20 in. lb
resulted in a leak tight seal. Further torquing to 40 in. lb reached the plastic-flow
limited torque for the No. 8 stainless bolts. Disassembly and reassembly at 40 in. lb
torque produced a He-leak-tight seal, which withstood baking to 250 °C [12].
A third design, solving a major vacuum problem for sealing thermocouple
gauges into ultrahigh vacuum systems, is shown in Fig. 8.4c. For most applications
under non-stringent vacuum conditions, thermocouple gauges are sealed at the pipe
thread joint by use of Teflon tape, vacuum cements, etc. This is unacceptable for
ultrahigh vacuum operation owing to the possibility of gas trapping in the pipe
thread or cement. The design shown in Fig. 8.1c uses the pipe thread to compress
either a Teflon or aluminum or OFHC copper gasket at the bottom of the tubulation.
The end of the stainless steel gauge stem must be smoothed with 600-grit paper,
using a removable plug inside the tubulation to prevent entry of debris. Similarly,
the sealing surface in the receptor hole must also be polished. This method permits
thermocouple gauges to be used on ultrahigh vacuum systems and to withstand
normal bakeout temperatures [13].
References
He-leak checking is commonly used, either in the shop with a portable leak
detector, or in the laboratory where a mass spectrometer inside the vacuum system
is used for detection. A common problem involves the detection of a small leak at
some point, where subsequent investigation will show that there is really no leak at
that point. This confusing result is caused by the diffusion through the atmosphere
of He from the suspect point to the real leak, which is some distance away. The
device shown in Fig. 9.1 eliminates this effect by sweeping He away from the test
point using a coaxial vacuum sleeve.
Figure 9.1 shows the device, and the reader is referred to the original publication
for construction and assembly instructions [1]. The He flow rate is adjusted to one
or two bubbles of He per second as determined by placing the probe under water
with the valve-controlled vacuum pump turned off. The rate of pumping should be
adjusted so that the sensitivity for He is not diminished by too rapid pumping.
A pump with a displacement of 5–10 L/min is satisfactory, with a throttle valve for
pumping speed adjustment.
Once a leak is noted, a second test should be performed to verify its general
location. This consists of placing a plastic bag around the suspected part, sealing the
bag with tape, and then filling the bag with He, and observing the leak again. Using
this coaxial probe and then using the bag verification test will eliminate carrying
non-leaking components to the shop for repair.
In the event that leak sealing compounds are used for temporary repairs, only
very small quantities of the sealer should be applied at the leak location, and a tag
should be placed on the affected component telling the location of the leak and the
date of temporary repair. The leak should be repaired correctly as soon as possible
after the temporary repair. The tag is a guide for the memory when the vacuum
system is ready to be disassembled.
Viton [3] seals are used in ultrahigh vacuum applications where deformable met-
alseals cannot be used. Viton seals may be baked out at 200 °C and after a 24 h
bake-out an outgassing rate of the order of 2 × 10−9 Torr L/s cm2 (nitrogen
equivalent) has been reported [4]. After bakeout, the main gas evolved was water.
9.3 Atmospheric Permeation—Viton O-Ring 109
Fig. 9.3 Atmospheric permeation—Viton O-Ring. a Apparatus for O-Ring studies. b O-Ring test
schedule. c Low humidity—permeation. d High humidity—permeation
These results suggest that the best application for Viton seals in ultrahigh vac-
uum systems is in cases where exposure of the seal to the atmosphere does not
occur, i.e., where the seal is used internally for isolation of one part of the vacuum
system from another. This is the case with Viton sealed gate valves mounted
between sections of the vacuum system, where the seal is retained in vacuum during
the time when the valve is opened.
A number of studies of outgassing and permeation of vacuum materials have
been carried out [6–8].
References
3. Viton A is a copolymer between vinylidine fluoride and hexafluoropropylene with the formula
[-FC(CF3)-CF2-CH2-CF2-]n. Initial outgassing evolves water, oxygen, hydrogen, nitrogen, and
some hydrocarbons (most prominent mass peak at 43 amu, probably from C2F+). After bakeout,
the main gas observed was water [2]
4. R.S. Barton, R.R. Govier, J. Vac. Sci. Technol. 2, 113 (1965)
5. N. Yoshimura, J. Vac. Sci. Technol. A7, 110 (1989)
6. W.G. Perkins, J. Vac. Sci. Technol. 10, 543 (1973)
7. B.B. Dayton, Transaction of the American Vacuum Society 6th Vacuum Symposium, 101 (1959)
8. R.A. Redhead, J.R Hobson, E.V. Kornelson, The Physical Basis of Ultrahigh Vacuum
(American Institute of Physics, New York, NY, 1993), chapter 10
Chapter 10
Specialized UHV Systems
There are advantages in using aluminum UHV systems. For proton synchrotrons,
the residual radioactivity produced in stainless systems may be reduced by one or
two orders of magnitude, 100 h after machine shutdown, by employing Al instead
of stainless steel. For electron synchrotron machines and electron storage rings,
aluminum is preferred because of good thermal conductivity, ease of manufacture
by extrusion, low residual radioactivity, nonmagnetic character, and low cost. The
design here is an all aluminum system made of two types of aluminum alloy;
aluminum flanges and gaskets are specially designed [1, 2]1.
A special aluminum alloy, 2219-T87, is found to have the necessary mechanical
properties for making the sealing flanges and nut/bolt sets. As shown in Fig. 10.1,
the sealing surface consists of a circular slot on each flange that houses a Helicoflex
[3] O-ring containing an elastic spring core inside a pure aluminum sheath. The
fixed depth of the O-ring groove was adjusted to correspond to the specified
compression radius of the Helicoflex element—a 4.5-mm diameter Helicoflex
required two 1.5-mm deep grooves.
The chamber body was made of a 6063-T alloy, which could be welded by an
AC TIG welding process. Distortion after welding was negligible and remachining
was not necessary. The entire chamber, except the O-ring sealing surfaces, was
anodized to a 30-μm thickness. Bellows made of 6063-T6 alloy could be made by a
rolling process yielding 0.5 mm thick walls. The bellows are employed for align-
ment purposes, not for dynamic motion. They also compensate for thermal
expansion effects.
The O-rings were sealed by torquing the nut/bolt combinations, bringing the
aluminum flange surfaces together. The torque (per unit area of the O-ring) used for
tightening was between 140 and 215 kg/cm, with 200 kg/cm being normal.
1
Which contains references to various papers on this subject.
Following bakeout at 200 °C, the leak rate at the Helicoflex seals was measured
with a He leak detector. It is reported that the leak rate is less than 10−11 atm cm/s,
where the units on this number suggest that a leak rate per cm2 of sealing surface
was calculated. The ultimate pressure in the system was 3 × 10−10 Torr after
bakeout and the residual gases consisted mainly of H2, CO, and CO2. The 38-L
system was pumped by a 220-L/s turbopump and a 160-L/s sputter ion pump. It is
indicated that the practical upper bakeout temperature is limited to 170 °C by the
6063-T6 aluminum alloy.
A study of the outgassing rate of an aluminum vacuum system during bakeout is
given in [4].
References
Formed grids are employed in ion and electron optical devices. Often these grids must
be made of highly transparent woven stock, either of stainless steel or tungsten [1].
Woven grid material may be formed into a hemispherical shape using a mold or
mandrel, as shown in Fig. 11.1a. Here the grid material is stretched over the mandrel
and an annealed aluminum ring is slipped over the stretched grid, capturing it near
the base of the mandrel. A steel pressure ring is then seated on the aluminum ring and
pressure is applied causing the soft aluminum ring to distort into an L-shaped cross
section as shown. This also encases the grid material, fixing it in the aluminum [2].
Another method for attaching a stretched grid to a support ring is shown in
Fig. 11.1b. Here a mesh grid is stretched over a brass mandrel, and a stainless steel
plate is clamped in place to hold the grid on the mandrel. The plate is tack-welded
to the grid in about ten places and the clamps are released and the plate turned over.
The grid wires are trimmed and individually spot-welded [3]. A more complex
variation on this method is shown in Fig. 11.1c where the grid is captured between
two shaped rings, R, which are bolted inside a jig. Spot-welding takes place through
the holes, H, after the grid is stretched over the mandrel [4].
Grids must often be pierced in specific locations for the passage of hardware or
source beams through the grid. Figure 11.1d shows a nice way to do this job
accurately. The brass mandrel is punched with the location of the center of the hole.
A scribe is then used to scratch a circular shape on the mandrel locating the cutting
line on the grid. The grid is then stretched and spot welded, and the assembly is
immersed into aqueous NaCl solution (50 g/L). An electrode to a spot welder (S1) is
attached to the assembly, and a second electrode, S2, made of tungsten wire
(0.5 mm) is sharpened to a radius about twice the diameter of the mesh wires to be
cut. An electrical arc from the spot welder will cut the wires individually, usually
between adjacent wires perpendicularly crossing the wire to be cut. The salt water
prevents pitting on the brass mandrel by the electric arc [3].
Fig. 11.1 Hemispherical grids—formation and Piercing. a Grid capture—Al Ring. b Grid capture
—spot welding. c Grid capture—spot welding. d Piercing grid
This stretching method will not make perfect hemispherical grids, since some
relaxation will occur upon removal from the mandrel. If more rigid grid material is
needed, the stretched grid may be electroplated on the mandrel [5] or it may be fired
in an oven. The electroplating stiffens the contact points between the meshes while
firing will stiffen the wire itself. In the case of firing, a brass mandrel cannot be
used, and release agents on the mandrel will be needed. Either Aquadag (colloidal
graphite) or MoS2 will work as release agents.
Figure 11.2 shows a convenient method [7] for using a Viton O-ring to produce
uniform stresses at the boundaries of electroformed grids, making them flat. The
lower plate has a special groove cut in it with the inner wall beveled 30° with
respect to the normal. When the sandwich involving the two plates, the grid, and the
O-ring is compressed by adjusting bolts around the perimeter, the motion of the
O-ring gradually pulls the mesh tight as the O-ring expands outward in the beveled
groove under pressure. This produces the desired flat grid.
Fig. 11.3 Grid fabrication techniques—conical grid. a Conical grid. b Mandrel + spot-welding
References
1. Unique Wire Weaving Co., Inc., 762 Ramsey Avenue, Hillside, NI07205-1094; Electronic
Space Products International, 1050 Bensen Way, Ashland, OR 97520
2. D.A. Shaw, D.A. Bennett, Rev. Sci. Instrum. 43, 1706 (1972)
3. R.A. Taylor, J. Vac. Sci. Technol. A6, 2583 (1988)
4. J. Koch, Rev. Sci. Instrum. 45, 1212 (1974)
5. P.W. Palmberg, D.C. Johnson, H.J. Boll, Rev. Sci. Instrum. 35, 244 (1964)
6. Electroformed mesh is a product of Buckbee Mears, 245 E. Sixth Street, St. Paul, MN 55101,
and it may be obtained in various materials, with various thicknesses and optical transmission
7. R.J. Noll, L.L. Ochalla, J.C. Weisshaar, Rev. Sci. Instrum. 62, 246 (1991)
8. K. Kuhnke, K. Kern, R. David, G. Comsa, Rev. Sci. Instrum. 65, 3458 (1994)
9. Dr. R. David, Institut fur Grenzflächenforschung und Vakuumphysik, Forschungszentrum
Jülich GmbH, Postfach 1913, D-52425 Jülich, Germany (private communication)
Chapter 12
Conductive Coatings
Many devices used in surface science research require the deposition of electrically
conductive films. These films are often used as the backing for phosphor screens in
LEED apparatus and other electron beam display techniques.
Figure 12.1 shows the method used in many laboratories for the deposition of
semiconducting SnO2 films on Pyrex glass surfaces. A glass surface, appropriately
shielded on the inside and outside with Aquadag (colloidal graphite) [1] to prevent
unwanted deposition, is placed into a small furnace or glassblower’s annealing
oven. The oven is slowly brought to the annealing temperature of Pyrex, 550 °C.
A saturated solution of SnCl2 in isopropanol is made up and loaded into the
reservoir of an atomizer, as shown. Since we wish to deposit SnO2, the atomizer is
connected to a regulated oxygen tank, and oxygen gas is used to oxidize the tin
from Sn(II) to Sn(IV) upon deposition of the SnCl2 uniformly to the hot surface of
the glass. The spraying is carried out in short bursts, and the oven door is then
closed between bursts in order to rewarm the glass to the deposition temperature.
An experienced operator can judge the conductivity of the film from its color. Blue
and barely visible coatings possess approximately kΩ resistances between two
points a few centimeters apart, while yellow or brown coatings give less than 1 kΩ
resistance.
Caution should be taken not to breath in during the film deposition with the
SnCl2, since pungent HCl gas is a major product of the pyrolysis chemistry leading
to SnO2 deposition.
Following film deposition, the Aquadag can be removed using an abrasive
kitchen cleanser and a cotton swab in water, or by heating in a glass annealing oven
for long times, followed by washing.
The electrical contact to the conductive film can easily be made using a
glass-metal feedthrough fabricated by the glassblower. To the inner end of the metal
feedthrough, a small piece of platinum ribbon or wire is welded in such a manner
that it makes good contact with the inner surface of the glass. The contact is made
permanent by applying either platinum bright or gold bright paint [2] with a small
camel’s hair brush, followed by firing in the oven to produce a shiny film of metal
which is both adherent and conductive. Alternate methods for making electrical
contacts through glass to a conductive SnO2 film are described in [3].
The work function of SnO2 films prepared in this way has been measured using a
retarding potential method for an incident electron beam. The coatings were found
to have a work function of 4.79 ± 0.07 eV. Spatial variations in the work function
averaged ±10 meV, and work function differences between samples were less than
100 meV [4].
The original scientific report of the use of SnO2 films for producing conductive
glass may be found in [5]. Here also is a study of the resistivity of a number of these
films from 1.5 to 500 K.
Tin oxide films can be removed by “nascent hydrogen” produced near the film
surface by dipping a cotton swab first in HCl(aq) and then in powdered zinc dust.
This reactive mixture, when rubbed on the SnO2 film, will efficiently remove it by a
reduction process. Thorough washing in H2O should then be carried out.
References 125
References
1. Aquadag colloidal suspensions may be obtained from Acheson Colloids Co., 1607 Washington
Avenue, P.O. Drawer 611747, Port Huron, MI 48061-1747
2. Organometallic paints containing either Pt or Au may be obtained from the Hanovia Division of
Engelhard Industries, 1 West Central Avenue, East Newark, NJ 07029. These paints contain the
metal chemically bound with organic ligands which are thermally unstable
3. R. Gomer, Field Emission and Field Ionization (AIP, American Institute of Physics, New York,
1993), pp. 169–170
4. J. Burns, E. Yelke, Rev. Sci. Instrum. 40, 1236 (1969)
5. R. Gomer, Rev. Sci. Instrum. 24, 993 (1953)
Chapter 13
Phosphor Screens
The drained screen should be air dried in the sedimentation chamber for several
hours, keeping it covered to avoid dust. After drying, inspect for homogeneity.
A “ring” on the screen indicates a discontinuity owing to a change in the drip rate,
possibly due to a particle being trapped in the stopcock. Such discontinuities may
not be serious unless photometry is to be employed, where phosphor thickness must
be constant. Even without visible evidence of inhomogeneities, the phosphor
thickness will vary over the screen surface owing to the variable angle of inter-
ception of the settling particles by the curved screen. Following drying, the screen
should be baked in an oven at 80 °C for 30 min to remove additional water. Screens
treated in this manner are satisfactory for use in ultrahigh vacuum and show little or
no influence on the base pressure after a bakeout at 150 °C for 24 h [5].
For deposition of phosphor screens in glass vacuum envelopes, minor variations
of the procedure described above may be used. The sedimentation chamber may be
replaced by the envelope itself if the depth of the liquid suspension is sufficient.
Alternatively, the glass vacuum envelope can be immersed in a large sedimentation
vessel, and in either case the draining can be done by very, very slow tipping of the
envelope using a slowly driven motor that turns once in 24 h.
Screens prepared in this manner are approximately 10 μm thick and well suited
for electron beam imaging. Thicker screens may be subject to electrostatic charging.
An unusual electrostatic deposition and adhesion method has been described by
Cetronio and Cross [7].
Phosphor coating of screens for imaging electron beams may be achieved using a
technique developed years ago at AT & T Bell Laboratories [8] and described
briefly elsewhere [9]. This method has the advantage over the sedimentation
method of producing more uniform phosphor coatings, since it does not sense a
curved surface during the deposition procedure.
As shown in Fig. 13.2, a glass or metal substrate, cleaned with a strong hot
alkaline detergent, has a stopper made of cork or plastic (but not rubber) that seals
the hole for the electron gun or other source. A sticky uniform layer of phosphoric
acid, H3PO4, is deposited by the following procedure:
Clean about 100 glass balls (3-mm diameter) and the screen with ethanol and
acetone to degrease. Then from a solution consisting of 3 drops of H3PO4 in 25 ml
of acetone, place 25 drops onto the screen substrate with the stopper in place. Place
the glass balls onto the screen substrate and roll them around as the acetone
evaporates. After 2–3 min, the acetone will have evaporated, and the phosphoric
acid will uniformly wet the substrate. If streaks are observed, repeat the process.
Then shake 1–2 cm3 of phosphor powder onto the wet substrate and tap the screen
until the phosphor coats the entire surface. Invert the screen and tap to remove the
loose powder. Remove the stopper and using a gentle jet of nitrogen, blow off any
130 13 Phosphor Screens
remaining loose phosphor. The screen should be baked in air at 200 °C for 2 h and
it is then ready for use in ultrahigh vacuum.
Caution should be exercised to prevent breathing of phosphor or contact with the
skin.
A unique method for electrostatically depositing and adhering a phosphor screen
is described in [10].
Another unique method for depositing a phosphor powder on a flat, tin
oxide-coated plate is described in [11]. Here, a glass plate which has been chilled in
air below the dew point is coated with phosphor powder from an atomizer. The
phosphor uniformly adheres to the moist plate. Drying the plate in a dessicator
removes the water, and the phosphor adheres to the plate without any adhesive. The
adhesion is sufficiently strong to permit the use of the phosphor screen in any
position in a vacuum system, yet the coating can easily be removed by wiping
under water.
References 131
References
Thermionic emitters are very widely used in surface science experiments as sources
of electrons. The emission of electrons in the absence of applied electric fields
follows the Richardson equation,
I=S ¼ AT 2 ee/=kT
where I/S is the electron current density, ϕ = the work function of the emitter, e is
the electronic charge, T is the temperature in K, and k is the Boltzmann constant.
For metals, A = 120 A/cm2 K2.
Various coatings on thermionic emitters are used to reduce the work function,
permitting the thermionic emitter to operate at lower temperatures while delivering
the same current density as an unmodified emitter at higher temperatures. Some
aspects of the various coatings and their advantages and disadvantages in ultrahigh
vacuum systems are described by Redhead et al. [1]. For various reasons, ThO2
coatings on W or Ir surfaces exhibit the most favorable properties for use in
ultrahigh vacuum. While long-term operation in an oxygen ambient is not possible
with either metal, Ir will withstand short accidental exposure to oxygen without
burnout.
Figure 14.1a shows the simple experimental procedure that may be used for
coating a thermionic emitter with a layer of ThO2 by cataphoretic deposition [2].
A suspension of 200 mesh ThO2 powder (5 g) in a solution containing 0.075 g of
thorium nitrate in 100 ml of 95 % ethanol is used. This suspension is allowed to
stand for a few minutes so that the large ThO2 particles settle to the bottom and are
removed from further considerations. The suspension is electrically deposited on
the filament at a negative potential of a few volts relative to the Pt anode. The
required current density depends on the shape of the surface being coated, and is
found by trial and error. It is best to electropolish tungsten substrates in NaOH
solution (tungsten positive) and to rinse in water before beginning the ThO2
deposition. The white ThO2 powder will form a very uniform deposit on the emitter
surface, and when fired in vacuum, this coating will be rather strongly adherent.
Coatings between 0.025 and 0.040 mm in thickness are satisfactory [2]. Firing in
vacuum to about 1800 K brightness temperature activates the thoria fully [2].
Reference [2] contains many interesting measured properties of ThO2.
The maximum current density that may be drawn from both clean W and ThO2
surfaces is shown in Fig. 14.1b. Since the average work function of W (about
4.5 eV) is reduced to about 3.0 eV by thoriation, a tremendous enhancement of
electron emission is achieved, as shown in Fig. 14.1b [1].
Fig. 14.1 Thoriated tungsten thermionic emitters. a Cataphoretic ThO2 deposition on a thermionic
emitter. b Maximum current density for thermionic emitters. c W Evaporation rate for thermionic
emitters
14.1 Thoriated Thermionic Emitters 135
One of the reasons for using thoriated thermionic emitters is to reduce the
operating temperature for electron emission to levels that involve negligible
evaporation of the metal. Figure 14.1c [1] shows the evaporation rate of W atoms
versus electron emission current density for both W and ThO2/W emitters. At the
higher emission current densities for pure W, significant evaporation of W will
occur over long experimental times, whereas ThO2/W emitters involve negligible
W evaporation rates.
It is important not to bombard the ThO2 coating with electrons, which will
cause damage. Even using AC accelerating voltages is detrimental in degassing
ionization gauge grids, when the hot grid can be a thermionic emitter on the reverse
half-cycle [1].
It has been reported that ThO2-coated W filaments often emit unknown impu-
rities that can be contaminants in surface science experiments [3]. This was studied
for ThO2/W filaments contaminated by F [4], and significant transfer of F or
F-containing species to a clean single crystal of W was observed. Other halogen
impurities such as Cl are also likely to be present in ThO2 coatings exposed to new
vacuum systems, since both HCl and HF are common pickling agents in cleaning
stainless steel [5]. These effects can be eliminated eventually by outgassing new
ThO2-coated emitters for long periods of time [4]. Such halogen contamination of
the grid surfaces of a Bayard-Alpert gauge renders the gauge useless for low
pressure measurements, since electron impact desorption from the halogen-covered
grid surface will supply spurious ion currents [6].
In addition to W wire thoriated by surface coatings of ThO2, it is also possible to
purchase W wire that contains alloyed Th at the l–2 % level [7, 8]. This small
amount of thoria was originally added to decrease crystallite slippage in the
polycrystalline wire as it was formed by drawing through a die. These wires also
exhibit greatly enhanced thermionic emission, compared to pure W. Despite high
vacuum conditions, these emitters are readily poisoned [9]. This poisoning process
may be diminished by carburizing the wire in a hydrocarbon gas, such as acetylene
at temperatures near 1700 °C [8]. An outer layer of W2C, about one-tenth the wire
diameter in thickness is produced [8].
A thoriated Ir thermionic emitter has been reported to have high resistance
to exposure to air and to exhibit low Th evaporation rates at operating
temperatures [10].
Thorium carbide thermionic emitters exhibiting an effective work function of
about 3.3 eV have been produced and will emit about 10 A/cm2 at 2150 K [11].
Solid LaB6 thermionic emitters have been described on p. 137, but these are only
useful for well-defined geometries and are costly. Thus for small, disposable LaB6
emitters, cataphoretic deposition on metal ribbons or filaments provides an excel-
lent route to useful thermionic emitters.
Five factors have been identified as being of importance in LaB6 deposition.
They are:
1. The particle size and the relationship between particle weight and surface area.
2. The concentration of cations in the suspension holding the LaB6 particles.
3. The ability of cations to adsorb onto the LaB6 surface, providing surface charge
to permit particle migration under electrolytic conditions.
4. The electrostatic potential gradient causing the particles to migrate.
5. The adherence of the particles to the metal cathode surface.
The most efficient deposition occurs with particles having a high surface charge and
migrating in a strong electrostatic field.
The present study [12] has been built on earlier methods [13, 14]. Re ribbons,
0.0025 × 0.76 × 12.7 mm3 were used. These filaments were heated white hot in
vacuum to clean them before external cataphoretic deposition. A Ta anode was
used, and the current (0.2–1.0 mA) was adjusted to achieve optimum deposition of
LaB6 in 30–300 s. For best results, 10 g of very finely divided LaB6 was suspended
in 100 ml of methanol and about 0.05 g of NH4NO3 was added as the electrolyte,
dissolving this first in a small volume of methanol. These suspensions have a short
lifetime of the order of several weeks owing to hydrolysis of the LaB6, which
produces a white coating on the powder that interferes with adhesion to the cathode.
A Ta ribbon was used as an anode, and it was noted that the deposition occurs at the
Re ribbon edges owing to higher electric field strength there. In order to avoid this,
the geometry shown in Fig. 14.2 was used, with the Ta anode parallel to the
meniscus of the suspension, and the ribbon pulled almost out of the suspension to
reduce the tendency for deposition on the edges. A narrow gap between cathode
and anode is preferred to increase the electrostatic field. In addition, deposition
against the gravitational field leads to the selection of smaller LaB6 particles which
adhere better than large particles. The LaB6 thickness could be adjusted up to
0.5 mm, but little difference in performance was noted for various thicknesses.
Deposition on filaments can be done with a large anode and with the electrodes in a
vertical position, since the electric field at the cylindrical filament surface is uni-
form [12, 14].
Failure in the adhesion of LaB6 to Re supports has been observed following
cataphoretic deposition. It was found that either boron or boric oxide impurity in the
LaB6 was responsible for the adhesion loss, and a purification procedure was
described which produced LaB6 with excellent adhesion properties on Re surfaces
[13].
LaB6 may also be deposited on Ta foil or filament surfaces if the surface is
carburized first to produce a barrier layer of TaC that protects the underlying Ta
from attack by the LaB6 at high temperatures [15].
14.3 Directly Heated Lanthanum Hexaboride Thermionic Emitters 137
copious emission at 1400 °C, it cracked upon cooling as a result of its high coef-
ficient of thermal expansion. To avoid this problem, rectangular and circular fila-
ments were constructed as shown in Fig. 14.3b. These structures survived
temperature cycling. The structure in Fig. 14.3b is mounted on graphite holders,
while that in Fig. 14.3c is mounted on Mo holders.
Figure 14.3d gives the temperature versus current for a LaB6 strip, 1.6 × 3 mm2
in cross-sectional dimension. The points are experimental points, and the line is a
theoretical line calculated on the basis of the ohmic heating and the radiative losses
[17]. Figure 14.3e shows the thermionic electron emission of LaB6 compared to Ta
and W emitters. Approximately four to five orders of magnitude difference in
current density is observed compared to the refractory metals.
As mentioned before, one is concerned with La evaporation from LaB6 when it
is heated to sufficient temperature. However, one is also concerned about metal
evaporation effects when refractory metals are used as thermionic emitters, since
they must work at very high temperatures (see p. 133). Figure 14.3f shows a
comparison of the evaporation rate from LaB6 compared to W and Ta, obtained
from published data, where the ordinate is in g cm−2s−1. On this basis, a LaB6
cathode can last for several thousands of hours at current densities of 10–
20 A cm−2. Above 50 A cm−2, the useful life may still be greater than 200 h [18].
A method for welding small LaB6 shapes to Re wire for making compact
high-efficiency thermionic emitters has been described [20]. In addition, a method
for protecting a W filament from LaB6 attack by interposing a layer of ThO2 has
been described [21].
A method for mounting conical tipped LaB6 cathodes for electron beam
instruments has been commercially developed. The cathode is mounted between
pyrolytic graphite blocks that are clamped between Mo/Re alloy support posts that
have a high modulus of elastically at high temperatures. The method is also used for
CeB6 cathodes [22].
between the emitter and the conductor wires. This method avoids welding and
allows the emitter coil to be disassembled for replacement with ease.
Connection to the Mo conductor wires is conveniently made with Cu braid, and
a soft Cu connector tube (annealed in a flame) that is crimped tight several times
around the braid and the Mo wire using pliers. The assembly should be arranged
such that the braid does not strain the Mo conductor wires by its weight or motion.
Figure 14.4b shows a similar device that utilizes a Mo electron repeller shield.
The alumina tube is flattened on two sides and is placed into a fitted channel at the
back of the Mo shield. The Mo shield will charge negatively during operation, but
for heavy electron bombardment heating, the crystal must be biased positively.
Macor ceramic supports are inappropriate for these purposes for two reasons.
The material degrades at high temperatures, and fluorine compounds are evolved at
high working temperatures [24].
14.5 Indirectly Heated Cathodes for High-Temperature Operation 141
Cathodes for electron guns and other devices are often coated with low-work-
function materials that permit higher current densities to be drawn than would be
possible with pure refractory metal cathodes such as W or Mo (see p. 133). Coated
cathodes are often heated to temperatures in excess of 2000 K for activation and
operation. Directly heated cathodes may be employed, but the requirement for an
emitter to be at equipotential over its surface rules out direct ohmic heating.
Thus, indirectly heated cathodes are often used. The design shown in Fig. 14.5a
uses an ohmically heated W filament to heat a Mo disk containing a work-function-
lowering material on its surface [25].
A self-supporting W filament (0.38-mm diameter), wound in a double helix on a
mandrel into a coil 9.5 mm in length, is used to heat a Mo disk cathode mounted on
the helix axis. The Mo disk is at the end of a cathode cylinder, 3.8 mm in diameter
and 10 mm long. Mounted on the outside of the Mo cathode cylinder are 3 (0.1-mm
thick) polished Ta radiation shields. These shields reduce the radiation transmitted
to the outer stainless steel support cylinder such that a measurable color temperature
for the stainless steel sleeve is not reached.
Figure 14.5b shows the W heater life behavior versus the heater current for
operation in ultrahigh vacuum. The temperatures associated with each point were
measured pyrometrically on the Mo cathode surface.
In the example shown here, only radiation heating of the Mo cathode was used.
However, this may be assisted by applying a positive potential to the cathode
relative to the filament, and drawing electron bombardment current to the cathode.
This method of heating is well suited for fast flashing of the cathode which is often
necessary during activation of the cathode coating (about 2500 K for ThO2).
Electron bombardment also involves lower W heater temperatures, and will not
shorten the heater lifetime.
Fig. 14.5 Indirectly heated cathode for high temperature operation. a Cathode design, b Cathode
lifetime versus heater current
Many workers have noted that external electron sources in UHV systems cause
erratic readings in ionization gauges. In the case of ionization gauges in the vicinity
of synchrotron or other radiation sources, there are also effects due to external
radiation causing inaccurate gauge readings. In early work by Redhead [27], the
effects of soft-X-ray production leading to photoelectron generation inside the
14.7 Bayard-Alpert Ionization Gauge with Corrections … 143
gauge itself were eliminated in various ways, resulting in new designs for gauges
with for instance two collector electrodes (the modulated ionization gauge), or to
gauges in which the ion collector was electrically and spatially isolated from the
ionization region (suppressor and extractor gauges). These modifications also led to
methods to compensate for positive errors caused by electron stimulated desorption
from adsorbed gases on the ionization gauge grid.
A normal ion gauge operating in the vicinity of an external thermionic emitter or
in a region containing high energy radiation will suffer major errors, caused by
things like external electron collection by the grid or the collector. Also
externally-produced photons interacting with the electrodes can cause X-ray pro-
duction, leading to photo-emission within the gauge, and giving spurious ion
current measurements. In the new gauge design shown in Fig. 14.7, measurements
of the current due to externally-produced electrons are made at the correcting
electrode in an auxilliary gauge which does not contain a thermionic emitter fila-
ment, and these measurements, applied to the current measurement from the normal
gauge, can be used to make corrections to the measured ion current and calculated
pressure. The details of this are dealt with in a recent paper [28], and background
information will be found in earlier papers [29–35].
Figure 14.7 shows the side view of the gauge design and its logic. ICE is the
collector current from the ion collector in the correcting electrode (CE) attachment,
and IC is the current from the normal gauge. The measurement of ICEe, the external
electron current received by the grid (and hence available within the gauge structure
for ionization of gases), is also employed to make corrections to the measured ion
current in the normal gauge. Reference [28] gives a formula to measure a com-
pensated pressure in this gauge. Figure 14.7 shows the measurements of the
uncorrected gauge, BA-S4, compared to measurements of the corrected gauge,
BA-S4. Both gauges are subjected to a flux of externally-produced electrons which
enter the two gauges from the outside. The pressure measurements made with
NIG-2F are made for a gauge which does not see externally produced electrons.
The abscissa is a plot of the filament current causing thermionic emission in an
144 14 Thermionic Emitters
external electron source and large emitted electron currents are produced as one
moves from 2 to 3 A. The indicated emission current in the ion gauge being tested
remains essentially constant at 3.97 mA. It is seen that without compensation, the
ion gauge suffers considerable negative errors when externally-generated electrons
are present. The compensation circuit also includes a method for measuring the
effect of external radiation on currents generated in coaxial cable, ICErr and ICrr.
Fig. 14.7 Bayard alpert ionization gauge with corrections for external radiation and electron flux
effects
14.8 Replacing Filaments in Glass Bayard-Alpert Gauges 145
after thorough cleaning of the gauge with Alconox detergent. The gauge should be
annealed in the upright position before use.
Replacement filaments can be made of many materials, such as iridium, thoriated
iridium, thoriated tungsten, etc. (see p. 133). The use of 0.010-in. diameter W is
easiest. The bent filament is formed into a tight V-shape, and a hook is fabricated
from the filament wire by first twisting the apex region of the wire and then making
the hook as shown in Fig. 14.8. The hook is then inserted into a wire loop at the top
of the filament support rod. When softer filament materials such as iridium are used,
a tungsten hook made of 0.010-in. wire may be used for the upper support.
Hemostats are reported to be useful in the filament insertion procedure.
The lower attachment of the two ends of the filament to the filament attachment
posts can be made in three different ways:
1. Welding, using a tweezer spot welding assembly, and making small 0.002-in.
thick Ta tabs that will weld nicely to W and Ir filaments.
2. The filament post is filed round, and a l/8-in.-diameter stainless steel barrel is
fabricated as shown in Fig. 14.8. The small tungsten locking coil fits onto the
filament post, and when the barrel is twisted in one direction, binding occurs.
Twisting oppositely results in breaking the junction. A 0.020-in. clearance is
needed when the tungsten locking coil is made of 0.010-in. W. The emitter
filament is held under an 0–80 screw head placed in a tapped hole on the
opposite end of the barrel, using a jam nut as shown.
3. The W filament may just be wrapped several times around the two support
posts.
The gauge calibration seems to be well maintained after refabrication [36].
References
1. P.A. Redhead, J.P. Hobson, E.V. Kornelsen, The Physical Basis of Ultrahigh Vacuum
(American Institute of Physics, New York, 1993), Chapter 7, section 7.6
2. T.E. Hanley, J. Appl. Phys. 19, 583 (1948)
3. P.A. Redhead, J.P. Hobson, Br. J. Appl. Phys. 16, 1555 (1965)
4. J.T. Yates Jr, D.A. King, J. Vac. Sci. Technol. 9, 1256 (1972)
5. Stainless Steel Fabrication (Allegheny Ludlum Steel Corp., 1959), p. 309
6. H.E. Winters, J.W. Coburn, Surf. Sci. Rep. 14, 161 (1992)
7. I. Langmuir, Phys. Rev. 22, 357 (1923)
8. R.G. Murray, R.J. Collier, Rev. Sci. Instrum. 48, 870 (1977)
9. R.O. Jenkins, W.G. Trodden, J. Electron. Control 12, 1 (1963)
10. H.F. Gray, E.H. Harris, T. Pankey Jr, G.A. Haas, Rev. Sci. Instrum. 43, 1113 (1972)
11. H.H. Glascock Jr, Rev. Sci. Instrum. 43, 698 (1972)
12. J.E. Delmore, Rev. Sci. Instrum. 54, 158 (1983)
13. L.J. Favreau, D.F. Koenig, Rev. Sci. Instrum. 38, 841 (1967)
14. L.J. Favreau, Rev. Sci. Instrum. 36, 856 (1965)
15. M. Nasini and G. Redaelli, Rev. Sci. Instrum. 42
16. J.M. Lafferty, J. Appl. Phys. 22, 299 (1951)
17. K.N. Leung, R.A. Pincosy, K.W. Ehlers, Rev. Sci. Instrum. 55, 1064 (1984)
References 147
It is common to employ magnetic shields inside ultrahigh vacuum systems for the
elimination in electron spectrometers of stray magnetic fields caused by the Earth or
by magnets associated with ion pumps or other devices near the vacuum system.
A number of formulas have been presented for the calculation of the efficiency of a
magnetic shield [1–3]. These formulas do not consider deviations from pure
cylindrical geometry caused by lips, caps, and pumpout ports. The calculation
method studied here is able to approximate the nonideal conditions caused by
deviations from perfect cylindrical geometry [4].
A computer code, JASON [5], has been used. This code solves the linear Poisson
equation. The Earth’s magnetic field is of the order of 500 mG. It is desired in this
particular example to reduce the magnetic field to less than 1.0 mG inside the shield
shown in Fig. 15.1. The following constraints exist in this design as well as in
designs to be made in other ultrahigh vacuum systems: (1) A small number of
shield layers should be employed to reduce the additional surface area and out-
gassing which occurs from these surfaces; (2) Holes must be cut into the shield for
pumping by the ultrahigh vacuum system. The pumping speed through a hole for
atmospheric gases is about 11 L/cm2 s; (3) The sample must be moved into the
magnetically shielded spectrometer along the Z axis of the cylindrical shield.
The optimum shield material thickness, t, is
t 3a=2l ð15:1Þ
where a is the inner radius of the cylinder and μ is the permeability of the
magnetic shielding material. If the value of t needed, based on (15.1), is larger than
the thickness of available material, two concentric shells may be used in shielding,
at the expense of introducing more surface area.
References
6. Ad-Vance Magnetics, Inc., 625 Monroe Street, P.O. Box 69, Rochester, IN 46975
7. D.L. Martin, R.L. Snowdon, Rev. Sci. Instrum. 46, 523 (1975)
8. J.C. Jones, Rev. Sci. Instrum. 46, 489 (1975)
9. Obtained from: Dupont, 1007 Market Street, N-2426-A, Wilmington, DE 19898; see Dupont
high performance films, Bulletin H-38492 (1983)
10. Obtained from: Dow Corning, P.O. Box 0994, Midland, MI 48640–0994
Chapter 16
Single Crystal Fabrication/Orientation
The production of small spherical single crystals from Au wire provides a conve-
nient way of doing STM experiments on the various single crystal plane facets
which are exposed on the surface [1]. A procedure outlined below has been
employed in a number of STM studies on Au(111) and other crystal planes [2–10].
Low purity Au wire of 0.3 mm diameter was employed. The wire is first cleaned
in piranha solution (3:1 mixture of concentrated H2SO4 with 30 % H2O2) to remove
organic impurities and then in boiling concentrated HNO3 to remove transition
metals for about 20 min. The impurities from the bulk of the wire are removed by a
“zone refining” procedure like that used in the semiconductor industry. The wire is
clamped in a ceramic tweezer with the tweezer face horizontal and the wire hanging
down. Use a 3 cm portion of the wire. Then bring a tiny hydrogen flame slowly
from the lowest end of the wire and start melting it. Due to surface tension effects,
the Au melts into a bead. Do this slowly, and as soon as the bead size is about the
size of the wire diameter, remove the flame and allow the Au to crystallize. Then
reheat, remelt and move the melting front (readily visible) upwards away from the
bead. The impurities get stuck at the melting front and you then swipe them
upwards from the bead to the junction between the bead and the wire. The impu-
rities also segregate to the Au surface in this process.
After two or three cycles of this melting and sweeping procedure, repeat the
cleaning procedure (boiling in HNO3 and aqua regia—a mixture of concentrated
HNO3 and HCl which will dissolve Au) to dissolve a portion of the bead. The
surface of the Au bead and the interface region between the bead and wire, con-
taining zone-migrated impurities, will dissolve removing the impurities segregated
there and the etched bead will appear non-specular. Then repeat the whole proce-
dure several times and finally grow the bead to the desired size less than 1 mm in
diameter. Facets will appear on the bead. If you inspect a facet and see a single
visible particle, then the surface is not going to be STM clean. The single crystal
bead is conveniently mounted by its Au wire tail, spot welding to a Ta foil support.
The <111> facets will be several hundred µm in size. These facets feature zero
miscut angles since they are natural facets. The (111) terrace width can be as large
as several µm in the center of a facet. The edges of the facet approach the curved
surface of the bead providing a variety of vicinal Au(111) surfaces, where
single-atom steps form a superlattice.
This preparation should be done in a dust-free environment. Normally after
installation in the STM the Au surface will need to be ion bombarded with Ar+ and
then annealed in vacuum to present a clean surface (Fig. 16.1).
between the Bond Barrel Holder and the hardened steel jig can also result in a
curved crystal [22].
Grinding and polishing to make highly flat surfaces can be carried out most
effectively using rigid soft metal laps that are very flat and contain embedded
abrasive particles [13, 15, 17]. Such an arrangement is shown in Fig. 16.2d. A set of
laps, containing different particle size abrasive grains, is required, and
cross-contamination with other abrasive grain sizes is to be avoided. Brass and tin
are favorite lap materials. Grinding and polishing on laps has produced crystal
surfaces flat to one-twentieth the wavelength of sodium light or about 250 Å/cm,
which corresponds to only one atomic step for about 4 × 105 Pt atoms, measured
horizontally [15]. At this level of step density resulting from departures from pla-
narity, the step density owing to misalignment will be the dominant factor, being
about 2 orders of magnitude greater for a misalignment of 0.05°. A polishing jig
giving an accuracy of 0.02° in the orientation of a silicon single crystal has been
reported [23]. A method, involving the observation of a single Bragg reflection with
a diffractometer, and employing a crystal holder that can be rotated about its axis,
has been described, and the reported accuracy is better than 0.003° [24].
Combined mechanical and electrochemical polishing techniques are often used
and a description of one method for polishing copper single crystals and alloys is
given in [25] and [26].
Fig. 16.2 Single crystal orientation, grinding and polishing. a Typical crystal grinding. b Typical
crystal polishing. c Measured crystal curvature. d Crystal lapping for optimal flatness
16.2 Single Crystal Orientation, Grinding, and Polishing 159
A special holder for the grinding polishing jig, shown here, has been developed
which allows the device to be used in an automatic grinding-polishing wheel device
as is often employed to prepare metallurgical samples [27]. The holder uses ball
bearings inside a ring which make contact with the outside of the hardened steel shell
which makes contact with the polishing cloth and wheel. This ring-ball bearing
apparatus allows the jig to rotate, while applying downward and lateral force which
causes the polishing jig to translate and oscillate across the rotating wheel. Using this
jig, a spherical surface with 3.4 m radius was generated near the center of the crystal,
with surfaces fitting larger radii being produced near the crystal edge.
A clever method for cutting single crystal boules along any plane involves a
goniometer of very simple design, as shown in Fig. 16.3 [28]. Here, a single crystal
boule is fastened into a holder with a crystal bond compound, using a plastic screw
that will not deform the crystal when tightened against its outer surface. The holder
is placed into a special wedge-shaped device with shoulders cut at angle α with
respect to the upper surface. This wedge-shaped device can rotate inside a tilted
cylinder with an upper surface inclined at angle a with respect to the base of the
device. When the two wedges are rotated relative to each other, the tilt angle of the
crystal, ω, will change.
The crystal is aligned by Laue back-reflection X-ray diffraction by manipulating
the wedged device. A rotation of 180° will result in swinging the crystal axis by a,
as shown in Fig. 16.3. Thus, if the saw cuts in a plane parallel to the baseplate of the
goniometer, the crystal may be cut at any angle over the range α. An engraved
angular scale is provided in two places as shown. The rotations are not simply
related to the angles measured on the Greninger chart, used for X-ray diffraction
measurements, but the original paper gives a method to achieve the conversion
using a small programmable calculator [28].
The alignment of highly reflective single crystals by using simple optical methods is
commonly employed for the setting of vernier adjustments. In addition to the
optical methods, electron beams can also be used for crystal alignment, as shown on
pp. 253 and 255.
Figure 16.5a shows a very simple crystal alignment method [32], using only a
flashlight in a darkened room. An image of the pupil of the eye is reflected off of
the crystal and back into the eye. In addition, the image is also reflected from the
window to the eye. When the crystal surface is parallel to the window surface, the
two reflected images of the pupil will coincide.
Figure 16.5b shows the same method used outside the vacuum system to align a
crystal in its tilt angle so that the normal of the crystal is parallel to the manipulator
mounting flange face [32]. The manipulator is held in a carefully made holder that
is square. The image of the eye, illuminated by a flashlight filament, is reflected
from the crystal and passes back by a pointer. If the center of the image of the pupil
is coincident to the pointer when the pointer is translated along the table, then the
crystal normal is parallel to the tabletop and therefore to the manipulator reference
flange face.
Fig. 16.5 Crystal optical alignment. a Crystal alignment by eye. b Crystal alignment by eye on
table. c Crystal alignment by laser
162 16 Single Crystal Fabrication/Orientation
Figure 16.5c shows the use of a He/Ne laser for crystal alignment. The laser is
moved until its reflection from the window is aligned so that the laser beam returns
to the laser as an aligned beam coincident with the incident beam. The crystal is
then manipulated in the vacuum system until the reflection of the laser beam from
the crystal is also coincident. In this case, then, the normal to the crystal and the
normal to the window will be parallel. The accuracy of the laser alignment pro-
cedure will increase as the optical path is increased, and the accuracy of coalign-
ment can easily be made to be 0.1°, or better. A partially reflecting glass slide may
be used to deflect a fraction of the two reflected laser beams to a card located off the
laser axis. The slight divergence of the laser ultimately limits the accuracy of the
alignment procedures.
It should be pointed out that the establishment of laboratory coordinates inside
an ultrahigh vacuum system, where reference is made to flange alignments, maybe
in error because of machining errors and welding errors made in the fabrication of
the chamber, as well as to distortion of the chamber which occurs both on welding
and due to the atmospheric pressure.
differentiating the collected current with angle, plots that are symmetrical about the
normal will be obtained, provided the plane of the angular scan is perpendicular to a
mirror plane of the crystal. This behavior is a consequence of the band structure of
the crystal, and the more free-electron-like a band is, the higher the electron cou-
pling probability is to this band. Inside the band gap region the reflectivity of
electrons with low energies can reach values close to unity.
The apparatus shown in Fig. 16.7a shows the electron gun and the electrical
circuit used for differentiating the collected current at the crystal. The measurements
Fig. 16.7 Measuring angle of incidence of electrons. a Electron gun + circuit. b First derivative—
collected current
16.7 Measurement of Angle of Incidence of an Electron Beam on a Single Crystal 165
were taken in the ΓXWK mirror plane of the Cu(100) crystal, perpendicular to
another ΓXWK mirror plane. The angle of incidence was varied by rotating the
electron gun around the crystal. The data shown in Fig. 16.7b clearly show
the symmetry with respect to the normal incidence angle for two electron energies.
The symmetry indicates the absence of external magnetic fields and the absence of
stray electrical fields. It is estimated that the position of the normal can be estimated to
±0.5°, although the features in the wings differ in their maxima in the two directions
by several degrees. The procedure can be used to determine the azimuthal orientation
of the crystal by scanning the azimuthal angle. For off-normal angles of incidence,
extrema should be observed when crossing a mirror plane of the crystal [35].
References
30. M. Holzwarth, M. Wissing, D.S. Simeonova, S. Tzanev, K.J. Snowdon, and O.I. Yordanov,
Surf. Sci. 331-333, 1093 (1995). Here polycrystalline Cr films were smoothed with a reduction
in the rms roughness by a factor of 5 to 8 from 9.2 nm, as measured by the STM
31. M. Wissing, M. Batzill and K. Snowdon, Nanotechnology 8, 40 (1997). Here, CaF2 single
crystal surfaces were smoothed from an rms roughness of 0.6 nm to better than 0.1 nm (the
AFM noise level)
32. Dr. Oliver Schaff, Fritz Haber Institut der Max-Planck Gesellschaft, Faradayweg 4-6, D-14195
Berlin (private communication)
33. S.L. Cunningham, W.H. Weinberg, Rev. Sci. Instrum. 49, 752 (1978)
34. J.M. Burkstrand, Rev. Sci. Instrum. 44, 774 (1973)
35. R. Schneider, H. Durr, Th Fauster, V. Dose, J. Vac. Sci. Technol. A8, 3363 (1990)
Chapter 17
Tip Fabrication
Metal tips are required in field electron and field ion microscopy, as well as in
scanning tunneling microscopy. One method for preparing such tips is to use zone
electropolishing while viewing the process under an optical microscope [1–3].
Figure 17.1 shows the setup employed for both wire samples and for irregular
specimens such as thin films stripped from their substrates or slivers of metal cut
from a sheet. The electrolyte is held on a loop by surface tension, and its volume
can be adjusted upward by transfer from a wet glass rod or downward by absorption
by a piece of tissue paper. This loop can be moved laterally using an x, y,
z micromanipulator as shown, and the specimen to be electropolished is attached
mechanically and conductively to a specimen holder. For a wire specimen, selected
regions of the tip may be electropolished by lateral motion of the loop to the desired
region as shown in Fig. 17.1, inset A. Very sharp tips may be produced by first
electropolishing a neck behind the tip apex and then reducing the necked-down
region to a sharp point in an electrolyte film of about 0.2-mm thickness. Additional
sharpening can be achieved by quick thrusts of the electrolyte film onto and away
from the tip.
Irregular specimens can be shaped as shown in Fig. 17.1, inset B. Here, working
on the sides of the sample, etching is achieved to bring the sample to a tip shape
prior to sharpening the tip as shown in Fig. 17.1, inset A.
Either AC or DC electropolishing can be used. In addition, instead of using a
micromanipulator, the lateral stage motion of the specimen holder or the stage of
the microscope can be used with the sample being held rigidly in place.
The microscope magnification is either ×l50 or ×600 with a long working dis-
tance objective. Note the protective coverglass for the objective to prevent etching
of the glass lens by droplets of electrolyte produced during electropolishing.
Special considerations in tip construction apply for tips to be used under elec-
trolytes, where Faradaic currents may damage the tip. This damage problem may be
eliminated by the use of Apiezon wax to shield the tip shank [4].
An excellent review on methods for sharpening STM tips is given in [5].
The use of AC waveform control in shaping tips for field ion microscopy has
been described [6, 7]. Electrochemical etching of Ag tips has recently been
described, using electronic control of the tip movement out of the electrolyte [8, 9].
A review comparing the instrumental characteristics of various STM [10].
A simple apparatus [11] for making a number of STM tips while automatically
replenishing the electrolyte is shown in Fig. 17.2. Section (a) shows a three-hole
electrolyte support in which thin films of the electrolyte are held by surface tension
effects. The central hole of 3-mm diameter is connected to smaller 2-mm diameter
supply and withdrawal reservoirs in a 1-mm-thick Cu plate. Section (b) shows the
beveled end of a (1-mm outside diameter, 0.45-mm inside diameter) supply tube for
fresh electrolyte being fed into the supply reservoir. A similar tube is used to
withdraw electrolyte from the other reservoir. The wire to be etched enters the
17.2 Etching STM Tips Reproducibly 169
central hole, and projects below the bottom of the electrolyte film. This permits the
etched tip to fall away under gravity when the etching is complete, and it is believed
that as the lower tip falls away the local strain is likely to produce an atomic
asperity [12] for STM imaging. Section (c) shows the application of AC current to
the etching apparatus, and the profile of a tip just being started. The thin neck below
the desired tip location is useful for making a detachable handle with which to hold
the tip, and which can be broken off later.
The flow rate of electrolyte and the configuration of the meniscus can be
adjusted by changing the level of electrolyte storage beakers on labjack platforms
170 17 Tip Fabrication
An Omicron STM tip carrier has been converted to a thermal tip flasher. In the
design shown in Fig. 17.3, a Ta sheet, 0.125 mm thick, is fabricated with a key hole
slot. This sheet, mounted by three holes which correspond in position to the holes in
the tip carrier body, is electrically isolated from the supporting bolts by use of
concentric ceramic tubes. The sheet is slightly bent as shown to be able to make
contact with the tip at some point when the tip is magnetically attracted to a
permanent magnet. Current is passed between the Ta sheet and the bottom of the
carrier. Nominal flashing conditions are: 7 A × 3.5 V for 10 s, and outgassing
causing <5 × 10−10 mbar pressure rise is reported [18]. A similar device was
recently described [19].
Many methods have been devised for making sharp tungsten tips for STM and other
applications, and some of these are described in Sects. 17.1–17.3. It has been
discovered that tip blunting can occur during the last stages of tip etching [20]. The
method described here eliminates this blunting effect by control of the electro-
chemical potential during the moments just after tip separation in the etching
solution.
Figure 17.4a shows a typical etching apparatus for making tungsten tips. Using
2 M KOH (aq) as electrolyte, suspended in a lens shaped droplet in an Ir loop, the
17.4 Reverse Bias Method for Sharp Tip Etching 171
tip is roughly etched to a point where the thinning of the neck is readily visible. The
etching is then continued more slowly in a fine etching process with 0.1 M KOH
(aq) solution until gravity separates the parts. At this point the etching potential is
changed from V1 to V2 as shown in Fig. 17.4b. If the power supply is a normal dc
supply, the applied potential can be dropped to zero at the moment of separation,
but it has been observed that there is still a natural electrical potential which exists
between the surface and the tip and the Ir ring, of order 0.5 V, and that charge will
still flow at currents in the 10−7–10−8 A range. This natural current causes blunting
of the tip. Using a power supply designed to apply the reverse potential from that
used for V1, the blunting effect can be eliminated. The reverse bias is applied within
500 ns of the separation time by electronic detection of the moment of separation
which triggers the application of the reverse bias. Figure 17.4c shows the measured
tip radii versus post etching time at the natural potential compared to experiments
where V2 is applied as a reverse bias, eliminating etching. Figure 17.4d shows an
electron micrograph of a tip etched with the reverse bias method, where a 4 nm
radius is achieved. The blunting is thought to be due to the recession of the tip,
172 17 Tip Fabrication
Fig. 17.4 Reverse bias method for sharp tip etching. a STM tip machine. b Current versus time
during etching c. Tip radius versus post-etching time. d STM tip etching to r ˂ 4 nm by Reverse
Bias Technique. e Tip configuration if reverse bias is not used
during the post-separation etching, and the radius will be dependent on the cone
angle achieved. A schematic of the blunting is shown for times increasing from 1 to
5 in Fig. 17.4e after the separation.
17.5 Sharpening Single Crystal Metal Tips by Ion Bombardment 173
Both tungsten and molybdenum single crystal wires can be sharpened into tips by
using axial sputtering. The data shown here are for Mo(110) tips [21], but similar
results are reported for W(111)-oriented wires [22], except in that work the ion
beam direction was from behind the tip, producing very sharp pyramidal shapes.
The Mo(l10) crystal wire is mounted such that the axis of the wire is parallel to
the ion beam, as shown in Fig. 17.5a. In addition, because of variation in ion current
density across the ion beam, the tip should be centered in the beam to achieve
symmetrical sputter etching. The ion beam in this case is 350 μm in diameter. It is
found that Xe+ ions are more effective in making less roughened tips; Ar+ at low
kinetic energies can produce very sharp asperities also. High-energy Ar+ sputtering
produces increased surface roughness compared to Xe+.
Fig. 17.5 Sharpening tips by ion bombardment. a Apparatus. b Tip profiles by TEM
174 17 Tip Fabrication
The profiles of tips produced both by 3 kV Xe+ and 0.5 kV Ar+ sputtering are
shown in Fig. 17.5b. It is estimated that the central asperity on the Ar+ sputtered tip
has a base diameter of about 10 Å.
Ion sputtering for sharpening field electron and field ion tips has been described
[22, 23].
Polycrystalline wires cannot be shaped so well by this technique because of the
presence of grain boundaries from which preferential sputter etching occurs [23].
However, single crystal tungsten tips have been ion milled to a point that is sharp
on the atomic scale [24, 25].
A recent example of the use of a focused ion beam of Ga+ ions for ion milling W
and Pt/Ir tips is given in [26], which also summarizes much other work in this area.
Copper wires of 30–50 μm diameter have been ion etched into sharp tips, and the
dynamics of the process have been studied [27].
References
1. A.J. Melmed and J.J. Carroll, J. Vac. Sci. Technol. A2, 1388 (1984)
2. A.J. Melmed, J. Vac. Sci. Technol. B9, 601 (1991)
3. H. Lemke, T. Goddenhenrich, H.P. Bochem, U. Hartmann, C. Heiden, Rev. Sci. Instrum. 61,
2538 (1990)
4. L.A. Nagahara, T. Thundat, S.M. Lindsay, Rev. Sci. Instrum. 60, 3128 (1989)
5. M. Fotino, Rev. Sci. Instrum. 65, 159 (1993)
6. H. Morikawa, K. Goto, Rev. Sci. Instrum. 59, 2195 (1988)
7. H. Bourque, R.M. Leblanc, Rev. Sci. Instrum. 66, 2695 (1995)
8. K. Dickmann, F. Demming, J. Jersch, Rev. Sci. Instrum. 67, 845 (1996)
9. A.A. Gorbunov, B. Wolf, Rev. Sci. Instrum. 64, 2393 (1993)
10. Y. Kuk, P.J. Silverman, Rev. Sci. Instrum. 60, 165 (1989)
11. J.O. Fiering, F.M. Ellis, Rev. Sci. Instrum. 61, 3911 (1990)
12. P.J. Bryant, H.S. Kim, Y.C. Zheng, R. Yang, Rev. Sci. Instrum. 58, 1115 (1987)
13. L.A. Nagahara, T. Thundat, S.M. Lindsay, Rev. Sci. Instrum. 60, 3128 (1989)
14. Y. Kuk, P.J. Silverman, Rev. Sci. Instrum. 60, 165 (1989)
15. E.W. Muller, T.T. Tsong, Field Ion Microscopy (Elsevier, New York, 1969)
16. A. Cricenti, E. Paparazzo, M.A. Scarselli, L. Moretto, G. Selci, Rev. Sci. Instrum. 65, 1558
(1994)
17. L.A. Hockett, S.E. Creager, Rev. Sci. Instrum. 64, 263 (1993)
18. B. Unal, A. Belianinov and P.A. Thiel, Private Communication
19. F. Bastiman, A.G. Cullis, M. Hopkinson, K.J. Briston, J. Vac. Sci. Technol. B28, 371 (2010)
20. O. Guise, J.W. Ahner, M.-C. Jung, P.C. Goughnour, J.T. Yates Jr, Nano Lett. 2, 191 (2002)
21. S. Morishita, F. Okuyama, J. Vac. Sci. Technol. A9, 167 (1991)
22. J.M. Walls, H.N. Southworth, G.J. Rushton, Vacuum 24, 475 (1974)
23. J.A. Kubby, B.M. Siegel, J. Vac. Sci. Technol. B4, 120 (1986)
24. D.K. Biegelsen, F.A. Ponce, and J.C. Tramontane Appl. Phys. Lett. 54, 1223 (1989)
25. D.K. Biegelsen, F.A. Ponce, J.C. Tramontna, S.M. Koch, Appl. Phys. Lett. 50, 696 (1987)
26. M.J. Vasile, D.A. Grigg, J.E. Griffith, EA. Fitzgerald, and RE. Russell, Rev. Sci. Instrum. 62,
2167 (1992)
27. H. Hasuyama, Y. Kanda, T. Soeda, K. Niiya, M. Kimura, Matls. Sci. and Engr. 90, 33 (1987)
Chapter 18
Spot-Welding
Fig. 18.1 Spot-welding difficult junctions. a Point contact weld. b Welding to single crystal.
c Preventing electrode sticking. d Welding under methanol. e Sandwich weld between refractory
metals. f In situ electropolishing+welding
that changing the welding tip polarity sometimes helps in making a difficult weld,
although the reasons for this are not understood [2].
It is very important to only try to weld metals when the surfaces have been
freshly cleaned. Thus, W and Mo are cleaned by electropolishing in NaOH(aq)
solution, about 1 M in concentration. The surface to be welded is positive during
electropolishing, or in some cases AC electropolishing may be used.
Figure 18.1f shows a clever way to electropolish the metal and then to use the
elec-tropolish solution as a protective boundary from air oxidation. Here, a Mo
single crystal is to be welded to a Mo wire at the crystal edge. The NaOH(aq)
solution in a droplet is placed on the crystal, and electropolishing of the edge is
carried out using a Mo wire electrode in series with a light bulb and a transformer.
The light bulb supplies a protective load if a short circuit to the crystal occurs. Once
the electropolishing is complete, the droplet of solution is left in place, and the
spot-weld of a We/Re thermocouple in this case is made under the electrolyte [3].
After spot-welding, the droplet of electrolyte is pipetted away, and the assembly is
washed well with distilled water. It was noted that lower power was needed to make
effective spot-welds in this manner [3].
18.1 Spot-Welding Difficult Junctions 177
References
In some cases, the provision of wrinkle-free metal foil for windows may be needed.
A very simple method, related to the differential thermal expansion of metal and
Pyrex glass, may be employed to achieve this objective [1].
Figure 19.1a shows a short cylinder of fused quartz or Pyrex glass ground flat on
its end, mounted on top of the wavy foil. The thin foil is supported on its bottom by
a horizontal substrate such as a lapped flat plate of aluminum, coated with a
fluorocarbon dry lubricant [2]. The foil is attached to the plate at the edges with an
adhesive tape to prevent curling. An epoxy adhesive (Torr seal [3]) is applied to the
lower edge of the glass tube and the tube is placed on the wrinkled foil. The
assembly is heated in an oven for several hours at about 60 °C, and then allowed to
slowly cool in the oven overnight. When the assembly is heated, the areal expan-
sion of the metal foil exceeds that of the glass cylinder due to the thermal expansion
coefficient differences. The light weight glass cylinder does not impede the foil
expansion, and wrinkling does not occur in the center of the glass form. Upon
cooling, the foil contracts more than the glass, resulting in a uniformly stretched foil
as shown in Fig. 19.1b. This technique can be used for stretching foil over an
aperture that cannot be heated; the glass form is larger than the aperture, and the
aperture is attached by cement to the foil, previously stretched as shown. A simple
consideration of the strain effects in the circumferential and the radial directions
indicates that for the circular geometry of the glass cylinder, the strain in both
directions is equal, meaning that no shear stress takes place on cooling. This leads
to a uniform tension and a flattening of the foil [1]. HAVAR foil (0.005 mm thick)
[4], aluminum foil (0.025 mm thick) and Mylar foil (0.025 mm thick) have all been
stretched flat in this manner.
Fig. 19.1 Stretching metal foils. a Before assembly. b After cure and cooling
Two methods are outlined for producing micron-sized holes in the ends of tubes to
be used as supersonic jet sources in vacuum [5].
A tube of the appropriate length and inner and outer diameter is made with only one
end open. An inner rod, which fits into the tube, is made with its length just slightly
longer than the inside dimension of the tube. This rod is sharpened on one end, and
19.2 Making Micron-Size Holes for Supersonic Nozzles 181
Fig. 19.2 Making micron-size holes for supersonic nozzles. a Method I. b Method II
a cap is machined on its upper end. By varying the cone angle of the sharpened tip,
the hole size can be controlled. The sharpened rod is then punched with a hammer
until the cap is tightly in contact with the upper tube edge. This should produce a
cone-shaped indentation in the end of the tube as shown. The outer surface of the
tube is then sanded down carefully until a hole just begins to appear. A microscope
is used to judge the hole size using both reflected and transmitted light [6].
182 19 Mechanical Procedures
A tube of the appropriate length and inner and outer diameter is made with only one
end open. The closed end is cut down with a conical surface on a lathe until a tiny
hole appears on the end. A punch die with the same cone angle as the tube end is
made. The punch die is then hammered on top of the tube end to close down the
hole until the desired size is obtained. A microscope is used to monitor the changes
in the hole size using both reflected and transmitted light [7].
Fig. 19.3 Straightening and braiding wires. a Stretching a wire to straighten. b Making uniform
braided wire
19.3 Straightening and Braiding Wires 183
It is often advisable to work with very straight segments of wire, for example, in
placing wires along the outside of a manipulator shaft (see p. 67), where bends
might interfere with the linear construction or where a tight fit through tubulations
is needed. Figure 19.3a shows a method which works for nonrefractory metal wires
such as-Cu, Ni, Ag, and stainless steel [8]. The end of the wire is rigidly clamped in
a vise, and with a pair of pliers, the wire is stretched only a fraction of a percent.
The elastic limit should not be exceeded, and one can feel the give in the wire as
stretching occurs below the elastic limit. It is wise to have prepared a place to store
the straightened wire and a storage tube is recommended for this. After straight-
ening, many wires become work hardened so that further straightening by this
method will not work unless the wire is annealed.
Thin refractory wire will not straighten by this method, and another method must
be used [8]. Lay the wire on a paper tablet and stroke the wire with a round tool at
right angles to the wire axis, keeping the arch upward by rotating the wire during
straightening. A pair of smooth jaw pliers will hold the wire in the desired position.
Work on only a few inches at a time.
Braided wire may be conveniently produced in a uniform bundle by placing the
wires together and holding one end in a vise. A hand-cranked drill is chucked to the
other end of the wire bundle and turned slowly to make a uniform braid, as shown
in Fig. 19.3b.
References
1. W.H. Holt, C.M. Blackmon, Rev. Sci. Instrum. 51, 671 (1980)
2. Obtained from Miller-Stephenson Chemical Co., Inc., George Washington Highway, Danbury,
CT 06810
3. Obtained from Varian Vacuum Products, 121 Hartwell Avenue, Lexington, MA 02173
4. Obtained from Hamilton Technology, Inc., 101 N. Queen Street, P.O. Box 4787, Lancaster, PA
17604
5. Professor Thomas Engel, Department of Chemistry, Box 351700, University of Washington,
Seattle, WA 98195-1700 (private communication). The ideas are not original from the Engel
group and are believed to originate from Dr. Adi Scheidemann, from Peter Toennies’ group,
Göttingen University, Göttingen, Germany
6. From Steve Moskovits, T. Engel group, University of Washington
7. From B. Holm, machinist, University of Washington
8. G.L. Fowler, 6816 Kelly Ave., N.E., Albuquerque, NM 87109 (private communication)
Part III
Measurement Methods
Chapter 20
Electrons
Many descriptions of the design and properties of electron guns are presented in the
literature [1–5] and it is not the purpose of this treatment to summarize this prac-
tically important field. Instead, an ordinary electron gun [6] is described to give the
reader a beginning picture of this technology [7].
Figure 20.1a shows a drawing of the arrangement of elements in an electron gun
like those used in LEED and other surface science probes. It consists of a therm-
ionic emitter filament or ribbon, usually coated with a material that lowers the work
function, yielding higher operating currents at lower filament temperatures. For this
directly heated tungsten cathode, either thoriated tungsten, or ThO2, or LaB6
coatings can be used, as discussed on pp. 206 and 210. Some electron guns, such as
the widely used Varian Model 981-2125, have the thermionic emitter mounted off
axis to avoid line-of-sight exposure to the crystal reducing the possibility of con-
tamination as well as reducing light output. The thermionic cathode is housed in a
Wehnelt cylinder that surrounds it except for the beam extraction aperture. The
Wehnelt cylinder can be biased either positively or negatively, and is usually kept
near the potential of the cathode. This may be done by placing a resistor between
the cathode and the Wehnelt, which results in self-biasing of the Wehnelt; a typical
value of 570 kΩ will produce a small negative bias of the cylinder relative to the
cathode and will maintain beam focus over a several hundred volt range. The
electron beam is extracted by the first lens, G1, which then focuses the initially
divergent beam. This lens forms an image of either the emitter or the aperture of the
Wehnelt cylinder, depending on geometrical factors and the potentials applied. The
focal length is of the order of 10–30 cm. The drift tube, the lens G3, and the exit
aperture form an Einzel lens that focuses the beam. The beam diameter is typically
about 1 mm, and this can be reduced by decreasing the first two aperture sizes
(0.4-mm apertures in guns working at 100 eV energy typically deliver 1 μA cur-
rent). For 0.1-μA beam currents, a minimum beam size of 0.2 mm and a divergence
of 0.2° are possible. Changing the bias on the Wehnelt cylinder with respect to the
cathode drastically affects the beam current and its stability.
Figure 20.1b shows an experimental plot of the electron beam width for the
electron gun in Fig. 20.1a, c shows the beam divergence versus electron energy.
These properties were measured with the multiple Faraday cup assembly and
procedure shown on p. 198.
Modeling of the behavior of an electron gun with cylindrical symmetry such as
this one can be carried out using an electrostatic computer program that is readily
available [8].
High-quality electron guns for high-resolution LEED measurements have been
described and often involve field emission electron sources [9, 10].
A high-resolution low-energy electron source involving a pair of 127° cylin-
drical selectors and delivering 3 × 10−8 A with 30-meV resolution has been
described [11].
Fig. 20.1 Electron gun properties. a Ordinary electron gun. b Beam width. c Beam divergence
20.1 Electron Gun Design and Behavior 189
Electron sources for irradiation of cm2 area surfaces often employ stray electrons
that originate from thermionic emitters in mass spectrometers and ionization gau-
ges, or from bare filaments some distance from the target. While this is often
satisfactory, without adequate modeling of the distribution of electrons originating
from these sources, one may be dealing with non-uniform current densities on the
surface being irradiated. Extended electron beams are employed in cases where
surface analysis over a broad target area is to be done following damage to the
surface by electron bombardment.
A simple electron gun [14] designed to give a uniform current density across an
extended surface is shown in Fig. 20.2a. The gun contains a thoriated tungsten
filament (0.019-cm diameter) (see p. 206) which is hidden from line-of-sight to the
target by an L-shaped shield. Electrons are accelerated through a mesh grid (80 %
transparency) and then decelerated on the way to the target. The SIMION elec-
trostatic lens and analysis program [15] was used to model the behavior of the
electron gun as a function of various potentials, and Fig. 20.2b shows some electron
trajectories calculated for various electron energies. A fairly uniform flux in the
vertical plane is observed for electron energies above 2 eV.
Because of end effects on the thoriated emitter [16], to achieve a uniform current
density laterally, it would be wise to employ a gun width that is perhaps two times
larger than the target width, as was done here.
The current collected by the mesh grid and the target is shown in Fig. 20.2c.
Currents at the crystal in the 2–20 μA range are measured for electron energies in
the 2–31 eV range. Even at the lowest electron energies, about 1 in 5 electrons
emitted are collected by the target for mesh potentials above 10 V.
190 20 Electrons
Fig. 20.2 Low-energy/broad-beam electron gun. a Electron gun design. b Electron trajectory
analysis. c Current/voltage performance
The electron gun described here provides a broad, cylindrically symmetric electron
beam, as opposed to a simpler gun described on p. 290, which has only planar
symmetry. Low-energy electron beams produced by guns of these designs are used
to cause surface chemical reactions by controlled electron beam damage, where the
entire surface of a macroscopic sample is to be irradiated with uniform current
density. Since electron-induced surface reaction is expected to happen at low
electron energies, special design provisions must be made for delivery of high
current densities at low energies.
20.3 Low-Energy Electron Gun for Broad-Beam Irradiation—Cylindrical Symmetry 191
Figure 20.3a shows a cylindrical gun design [17]. A BaO-coated disk cathode
[18] is mounted inside a cylindrical repeller electrode. An anode plate with a
2-mm-diameter, 4-mm-long cylindrical hole is located close to the emitter disk.
Four cylindrical lenses are used to focus the electron beam, and these are stacked
between ceramic spacers [19]. A single positioning ring at the end of the stack holds
the assembly together, and the assembly is housed inside a metal tube housing. To
minimize work function differences along the gun, the inside of the lens is gold
plated.
Figure 20.3b shows calculated electron trajectories for different electrode
potentials. At the top, an anode at high positive potential withdraws electrons from
the cathode, and the trajectories are shaped by the four lenses set for potentials from
+30 V to 0 V as shown. This focuses the beam on the crystal at ground potential.
Figure 20.3c shows the electrostatic potential in the cathode region in a
three-dimensional plot for electrons. The geometry and potential of the repeller and
anode create a steep potential drop for electrons. The shape of the potential is
designed to increase the lateral momentum of the electrons in order to achieve a
broad electron beam. The broad beam exhibits very uniform current density as may
been seen from the trajectory spacings in Fig. 20.3b.
In the simulations, it was found that the emission is very sensitive to the exact
position of the cathode disk relative to the repeller cap and the anode opening.
Figure 20.3d shows the performance of the gun for various electron beam
energies. At an electron energy of 10 eV, a current of 7 μA is obtained at the crystal,
comparable to the performance of the gun shown on p. 293. The optimized anode
and repeller potential are shown for various electron energies in Fig. 20.3e.
It is not the purpose of this book to discuss the details of the various forms of
electron energy analyzers that are used for the analysis of surface phenomena.
Instead, a practical guide to the selection, construction, and use of electron energy
analyzers is available [20]. In addition, an extensive bibliography dealing with the
various design analyzers is also presented in [20].
Six types of electron energy analyzers are described in [20], and these
constitute the most widely used instruments. The analyzers are briefly categorized
below.
Figure 20.4a—Time-of-Flight Analyzers
These analyzers are used primarily for ion detection owing to the extremely high
velocity of electrons. However, time-of-flight methods have also been used for
rapid pulsed sources of slow electrons. Electrons are allowed to drift in a field-free
region, and their arrival at a microchannel plate detector is recorded by a suitable
waveform digitizer. MicroChannel plate systems can provide gains of 106–107 and
192 20 Electrons
Fig. 20.3 Low-energy electron gun—broad beam a Design of the low-energy electron gun.
b Beam profile for different electron energies. c Shape of the potential near the cathode.
d Emission current from low-energy electron gun. e Potential settings for broad beam operation
and constant electrode potentials
20.4 Electron Energy Analyzers 193
Fig. 20.4 Electron energy analyzers. a Time-of-flight analyzer. b Retarding grid analyzer.
c Tandem reflection—transmission analyzer. d Plane mirror deflection analyzer. e Cylindrical
mirror deflection analyzer. f Cylindrical deflection analyzer. g Spherical deflection analyzer
20.4 Electron Energy Analyzers 195
pulse widths of 1 ns. Drift-tube lengths of a few meters are employed, and in such
long tubes magnetic field effects may be significant.
Figure 20.4b—Retarding Grid Analyzers
Here, the source is located at the center of two hemispherical grids and a spherical
collector. The source and the inner grid are grounded to make a field-free inter-
vening region. The outer grid, biased at—VR, retards the electrons, and the collector
is biased positive relative to ground to collect all electrons that pass the retarding
grid. The energy resolution is estimated from the assumption that the source is
spherical, of radius r = a, and that the radius of the collector is r = b. A schematic of
the transmission of the device versus kinetic energy is shown when the retarding
grid is set at VR. The energy resolution, ΔE, at VR is given by the equation in
Fig. 20.4b.
Figure 20.4c—Tandem Reflection-Transmission Analyzer
This type of analyzer allows bandpass filtering of the energy distribution of elec-
trons emitted from the source. The first grid system is a low-pass filter. Electrons
having energies greater than eVREF can overcome the retarding field and are col-
lected on the ellipsoidal collector. Lower-energy electrons are reflected and focused
through a grounded aperture, and thence to a retarding grid analyzer something like
that shown in Fig. 20.4b. The two-dimensional image detector makes possible the
simultaneous detection of the energy and the emission angle of the electrons
received in the band pass of the two analyzers, as shown in the transmission
196 20 Electrons
diagram for the tandem analyzers. Analyzers of 100-meV resolution and large
angular acceptance have been constructed.
Figure 20.4d—Plane Mirror Deflection Analyzer
This represents the most elementary type of the deflection analyzers. Electrons,
constrained to a certain angular range by the apertures, enter from the left into a
region of width = h in which field-free conditions exist. From there, they enter a
second region of width = d, where a constant electric field exists, which field is
produced by making the upper plate negative. Here, the electrons travel in parabolic
paths in the uniform field. Electrons having a specific energy arrive at the exit slits
on the right and are measured below the lower grounded plate.
Figure 20.4e—Cylindrical Mirror Deflection Analyzer
Here principles similar to those used in the plane mirror analyzer apply. However,
because of the cylindrical symmetry, a wide solid angle of acceptance exists,
leading to high throughput and sensitivity. The cylindrical geometry is also con-
venient for the placement of the electron gun on axis, and has made this geometry
popular for Auger electron spectroscopy (AES).
Figure 20.4f—Cylindrical Deflector Analyzer
This type of analyzer is widely used in high-resolution electron energy loss spec-
troscopy (HREELS). The optimum angle between the entrance and exit slits is
θ = 127°. Analyzers of this type are particularly useful for measuring the angular
distribution of electrons emitted from a surface.
Figure 20.4g—Spherical Deflection Analyzer
This type of analyzer is widely used in photoelectron spectroscopy. The optimum
angle between the entrance and exit slits of the analyzer is θ = 180°.
Improvements in the resolution and throughput of these analyzers can be
achieved by preretarding the incident electrons prior to entry into the analyzer [20].
The accurate location of the Fermi level (EF, sample) is often an important
requirement for the calibration of electron spectrometers. Often in ultraviolet
photoelec-tron spectroscopy (UPS) the Fermi level can be determined by measuring
the Fermi edge feature of a clean metal surface. However, if UPS is not available,
the precise calibration of the Fermi edge position often cannot be carried out.
A method to do this, using the secondary electron emission from electron impact
excitation, X-ray photons, or UV photons on a graphite reference sample is
described below [21].
Figure 20.5 shows the secondary electron emission region of a sample of
graphite (highly oriented pyrolytic graphite, or HOPG) and a sample of Aquadag
20.5 Electron Spectrometer Calibration Using Graphite 197
(colloidal graphite) [22] that have been excited by He I (21.22 eV) radiation. In
addition, the upper energy photoelectrons in the vicinity of the Fermi edge are also
shown from a cleaned Ta sample in electrical contact with the graphite samples.
The fact that the upper energy measured is at 21.22 eV indicates that the spec-
trometer energy scale is properly calibrated in this measurement. It can be seen in
the secondary electron emission region that there is a distinct emission feature
centered above the Fermi level at 7.57 ± 0.03 eV and superimposed on the broad
secondary emission tail. This feature is related to electronic excitation of the a state
present in the conduction band of the graphite. This state arises from the a electrons
198 20 Electrons
in the C–C bonds in the graphite lattice and has been well studied previously
[23, 24]. The energy of the σ peak from graphite is independent of the energy of the
exciting radiation or electrons and hence can be used to locate the Fermi level
accurately when it is otherwise not accessible.
The behavior of the a state is independent of the annealing temperature of the
Aquadag over the range 230–750 °C and is independent of air exposure. Also, the
energy is invariant with the angle of emission. From the low-energy cutoff in the
secondary electron spectrum, the sample or electron analyzer work function can be
determined [21].
Fig. 20.7 Measuring anisotropy of electrical conductivity of a single crystal—four point probe.
a Four point probe. b Anisotropy in conductivity of TiO2(110). The radius observed is
proportional to the electrical conductivity of the single crystal in the measured direction
20.7 Measuring Anisotropy of Electrical Conductivity … 201
TiO2. The force on each tip was equivalent to 100 g. The SiC contact points are
located at the corners of a 1 mm square and are 0.1 mm in diameter. Current is
passed through two non-diagonal tips and the potential drop is measured through
the other two tips. From this information, the electrical conductivity is measured at
the angle of the probe with respect to the <001> crystal axis. The calculation
method of Yamashita [31] was employed. This method takes into account the finite
width and thickness of the crystal. The four-point probe is rotated in a systematic
manner and a new conductivity is measured at each azimuth. Figure 20.7b shows a
directional conductivity measurement for TiO2 measured on the well studied <110>
face [30].
light. A lens focuses the light from a LEED beam on a photodiode that is mounted
behind an aperture placed in a thin reflective metal screen. The LEED pattern can be
observed visually through a port using the reflective screen and the spot photometer
is moved laterally for measurement of individual beams. A photodiode mounted
behind the aperture integrates the light received from a selected beam plus the
surrounding background. The diffraction pattern is modulated at 100 Hz by
applying a chopping voltage to the suppressor grid of the order of Ve + δ, which is
sufficient to suppress the LEED pattern [32].
At a spacing of 30 cm from the LEED screen, the photodiode sees a circular area
of about 5.4-mm diameter, about five times the average diameter of a diffraction
spot in the apparatus. The photodiode is a low-noise planar-diffused silicon sensor
photodiode [33]. The reader may see examples of the use of this apparatus in
[34, 35].
Newer methods involving vidicon detectors are widely employed for the study
of LEED patterns and intensities on phosphor screens. It is possible to employ a
bandpass optical filter to remove stray light from high temperature filaments and
emission from a hot sample. For a P11 phosphor, with a narrow band of emission at
459 nm, an interference filter with a central wavelength of 460 nm and a bandwidth
of 50 nm is excellent for removal of the stray light, permitting more accurate
measurement of LEED intensities on a background that is closely similar to the dark
current of the vidicon [36].
Faraday cups are employed for the collection of electrons in collimated beams. One
problem is that specularly reflected electrons from the bottom of the Faraday cup
may be reemitted, causing the measured collected current to be diminished. Various
methods for the suppression of specularly reflected and secondary electron emission
are employed, and one principle employed is to make the Faraday cup depth a
number of times the aperture width. But even this design criterion does not prevent
the reflection of specular electrons from the bottom of the cup. Surfaces are known
to have high reflectivities for low-energy electrons of a few eV energy.
To avoid this problem, a stainless steel cone was inserted in the cup [37], as
shown in Fig. 20.9a. The specularly reflected electrons from this conical surface
maintain their forward component of velocity along the optical axis of the
cup. These electrons have no velocity component along the optical axis toward
the aperture. Thus this design is the electron optical equivalent of a light trap. The
design invalidates the principle that the depth of the cup should be large compared
to the aperture.
20.9 Modified Faraday Cup for Electron Current Measurement 203
Fig. 20.10 HREELS-discrimination between single and multiple electron scattering events.
a EELS and AES measurements of Al2O3/Al(111) at three different levels of oxidation.
b Correlation between the EELS intensities and the Al2O3 thickness
20.10 HREELS-Discrimination Between Single … 205
The deposition of thin films onto electron multipler surfaces occurs after long use,
particularly in vacuum systems having a background of oil vapors or in systems that
have been used for reactive gases, such as fluorine [39]. A simple rejuvention
method has been reported for Cu/Be dynodes exposed to fluorine to the extent that
the multiplier gain had dropped to 1/300 of its original gain. Complete rejuvenation
back to the original gain was observed [39].
For the particular case here, it was decided to expose the dynodes to a plasma
discharge in hydrogen to remove the fluorine, and then to expose them to a plasma
206 20 Electrons
discharge in oxygen to reoxidize the dynode surfaces. This was done, of course, by
removing the multiplier and treating it in a separate metal vacuum system. In
Fig. 20.11, a sequence of rejuvenation steps is carried out, in which neighboring
dynodes are used as anode/cathode combinations, working up the multiplier chain.
For hydrogen plasma generation, a pressure of about 1 Torr of H2 was used, under
flow conditions, with a 350 V DC power supply connected through 5000 Ω
between adjacent dynodes. A Tesla coil was used to initiate the plasma, and the
pressure was increased in the range 1–5 Torr until about 30 mA of current passed.
Hydrogen plasma treatment was carried out for 2 min. After this, for O2 plasma
generation, about 10 mA of current was used for 1 min. The multiplier anode was
not treated.
A cleaning and oxidation treatment, designed to avoid copper oxide formation
on the dynode structures and to place beryllium oxide films on the dynodes, has
been described [40]. Here, oxidation in water vapor at pressures up to 5 × 10−2 Torr
was carried out in cycles in an RF-heated furnace at 625 °C. Beryllium oxide
exhibits a higher secondary electron yield than copper oxide [40].
Figure 20.12 shows the counter, which consists of two parts, the housing unit
and the insertion unit. The housing unit is mounted on a 2 3/4-in. Conflat flange.
A conical ring presses either CaF2 or SrF2 windows [43] onto the housing con-
taining a Viton O-ring. The inner surface of the ring is beveled at 45° to minimize
photon emission from this surface by stray electrons. The window has a diameter of
20 mm and a thickness of 2 mm.
The insertion unit is mounted inside the housing unit, and can be removed for
bakeout. It too is sealed with a Viton O-ring at the 2 3/4-in. flange. Both the inner
wall of the insertion unit and the counting wire (1.5-mm diameter) are polished. The
counting wire is electrically isolated from the apparatus by a Macor insulator, which
is glued to the outer flange and to the counting wire with vacuum cement [44]. The
active length of the counting wire is 40 mm and its end is located 10 mm from the
window. The glue joint is made with the counter wire accurately centered in the
insertion unit. The port serves to evacuate the counter as well as being an inlet for
Ar background gas.
The counter is assembled with the help of a He leak detector to assure tight
sealing between the housing unit and the entrance window with minimum force on
the front-end bolts. After installation, the assembled unit is pumped overnight, then
opened and a few crystals of iodine are dropped in. After pumping again for 30 min,
Ar is introduced to a pressure of 100–200 mbar, and the counter is valved off.
Typical operating voltages are 400–500 volts. All pulses should be of identical
height (about 0.2 V). Occasionally the counter is pumped again and refilled with Ar.
Repeated fillings with Ar may also help if the counter does not work initially.
Deterioration of the counter performance (requiring a steady increase in voltage)
usually indicates insufficient vacuum conditions.
208 20 Electrons
References
37. W.R. Miller Jr, N.N. Axelrod, Rev. Sci. Instrum. 37, 1096 (1966)
38. J.G. Chen, J.E. Crowell, J.T. Yates Jr, Rapid communications. Phys. Rev. B. 35, 5299 (1987)
39. G.E. Salser, Rev. Sci. Instrum. 37, 674 (1966)
40. M.B. Pongratz, Rev. Sci. Instrum. 43, 1714 (1972)
41. V. Dose, Rev. Sci. Instrum. 39, 1055 (1968)
42. V. Dose, Th Fauster, R. Schneider, Appl. Phys. A. 40, 203 (1986)
43. CaF2 and SrF2 windows are available from Harshaw Optical/Bicron, 6801 Cochran Road,
Solon, OH 44139
44. The vacuum cement used is Araldit ATI, Ciba-Geigy GmbH, Oeflinger Str. D-79664 Wehr
(Baden), Germany
Chapter 21
Ions
Normal ion guns operate in an ambient pressure of the gas to be used for sputtering.
This pressure needed to generate a satisfactory ion bombardment current is nor-
mally in the 10−6–10−5 Torr range [1]. Improved ion gun designs involve a direct
feed of the inert gas into the ionization region, and these designs normally involve
an ambient pressure in the 10−8 Torr range. However, any rise in the pressure in the
main chamber during the cleaning process is detrimental, since it is known that gas
release from the chamber walls, from the ionization pump (if left on) and from the
sublimation pump is caused by an increase in the pressure of even an inert gas. This
undesirable gas evolution depends upon the previous history of the vacuum system
and influences the time required to pump down to the base pressure after ion
sputtering.
The design shown in Fig. 21.1a involves direct injection of the sputtering gas
into the ion gun and direct pumping from the ion gun into a turbopump. The gas
beam inside the ion gun is created by a multicapillary array collimator (see p. 421
for gas doser designs employing this collimation principle). The angular distribu-
tion of sputter gas molecules exiting from the array may be calculated [2, 3]. The
sputtering gas is injected into the ion gun through transparent Ta grids that are
welded over windows cut into the walls of the outer shield of the ionization gun.
A multicapillary array of 13-mm diameter and 3-mm thickness is used, with 10-μm
diameter capillaries. Control of the gas pressure is achieved with a leak valve, as
shown in Fig. 21.1b. Two gate valves are used to isolate the main chamber from the
turbopump after sputtering when an ion pump is used for maintenance of UHV.
Normally the 360-L/s turbopump is used to pump the main chamber in parallel to
an ion pump and a Ti sublimation pump (not shown) [4].
While an IQ 10/35 ion gun is shown in the example here, many ion gun designs
will work with this modification.
Fig. 21.1 Ion sputter gun with low gas emission. a Modified ion gun. b Schematic of vacuum
system
21.2 Alternate Ion Bombardment Sources 213
Ion bombardment of elemental crystal surfaces is widely used for removing both
surface and bulk impurities. In the case of bulk impurities, cleaning by sputtering is
often accomplished at elevated temperatures, permitting bulk impurities to diffuse
to the surface for removal. Normally an ion gun is employed for this and guns
giving both focused beams and more diffuse beams are available. Admission of Ar
gas directly to the ionization region of the ion gun will reduce the gas load in the
vacuum system and hence reduce contamination effects during and after sputtering
(see pp. 87–91).
Alternately, very simple ion bombardment sources have been reported, and two
examples of these are illustrated here. In Fig. 21.2a, an open ion gauge is used as a
source of Ar+ ions [5]. The vacuum system is filled with Ar gas to about 10−3 Torr,
permitting ample gas density within the ionization region of the gauge for
Fig. 21.2 Two alternate ion bombardment sources. a Ion gauge sputter source. b Reflector sputter
source
214 21 Ions
A more complex ion gun than that described on p. 211 and 213 is presented here
[7]. A Bayard-Alpert gauge is modified in two ways: first, by removal of the top
part of the cylindrical grid cage for addition of the ion extraction electrodes; and
second, by the addition of a metal shield around the gauge body and the thermionic
emitter.
21.3 Ion Gun Based on Bayard-Alpert Gauge 215
Figure 21.3a shows a schematic of the gun design and the electrical components.
A tantalum cylindrical shield to repel electrons is mounted around the gauge body
and its electrical potential is maintained at the same potential as the positive end of
the thermionic emitter filament. An extractor lens shaped with a waist withdraws
ions from the region enclosed by the grid and these pass into an accelerator region,
and thence into a cylindrical lens and a quadrupole lens, where the ions are ejected
into vacuum and to the target. The flange holding the ion gun is equipped with
radial feedthroughs about its outer edge for supplying potentials to the ion gun
elements.
The extractor lens accelerates mainly off-axis ions to an on-axis focal point, and
its potential can be as high as −200 V. The divergence of the ion beam is controlled
by the bore of the accelerator electrode, and the final focusing is done by the lens.
Since all potentials in the gun, except for the lenses, are referenced to the primary
energy, the extraction of ions is independent of the final beam energy. At low
energies, the ions are of high energy when they arrive at the lens, and are decel-
erated there, while being focused. This leads to high currents at low ion energies.
The quadrupole lens consists of four plates at 90° to each other symmetrically
placed around the gun axis, but its use is not described in the original article [7].
Fig. 21.3 Ion gun based on Bayard-Alpert gauge. a Ion gun and circuit. b Ion spatial distribution.
c Current and current density. d Ion beam energy distribution
216 21 Ions
The spatial distribution of the ions may be adjusted as shown in Fig. 21.3b.
Either a sharply peaked ion beam with high current density at the center may be
produced, or alternatively, a flat-topped beam profile may be produced by adjusting
the acceleration and lens voltages.
In Fig. 21.3c, the current and current density are plotted for a pressure of
1 × 10−4 Torr Ar pressure. Current densities between 0.1 and 120 μA/cm2 are
obtained at various ion energies. The current collected near 1 keV energy is limited
by space charge effects in the extractor region. The collector in these cases was
located 4 cm from the end of the gun.
Figure 21.3d shows the energy distribution measured with a retarding field
analyzer and showing that at 1.2 keV, an ion beam halfwidth of 1.6 eV may be
obtained using an extractor voltage of about −30 V. The ion beam was also mass
analyzed, and at an electron energy of 70 eV, a 1:10 ratio of Ar++ to Ar+ was
measured.
An earlier version of this design for an ion gun has been published [8].
A cold cathode ion gun, useful for sputter deposition, can be made, as shown in
Fig. 21.4 [9]. The gun is useful for sputter deposition from any conducting target
material. The ion gun was first described in [10], and this is a modification of that
design. The gun is capable of producing an ion beam of 1-mA current at a kinetic
energy of 2–6 keV, impinging on an area 5–10 mm in diameter. There is no
filament, and the gun can be operated for hundreds of hours without maintenance.
Currents in excess of 1 mA cause tip heating which shortens the tip life. Sputter
deposition rates of 1–4 Å/s can be obtained.
The beam gas is admitted from a pumped gas handling system through two
ceramic insulator nipples and through a metal frit particle filter into a Conflat
flange-mounted electrical feedthrough passing a 1/4-in. tube that is connected via
Swagelok connectors to a hollow tube made of l/16-in.-diameter tungsten or tung-
sten carbide alloy. This needle is 0.625-in. long and has an orifice of 0.05–0.075-mm
diameter. It is called a wire bonding tool [11]. The high electrical field at the tip
ionizes the gas as it passes through the tip, resulting in a plasma cone that extends to
the target. The diameter of the cone depends on the orifice diameter, supply gas
pressure, and target distance. A 2.5 Torr beam of gas in the gun, combined with a
tip-target distance of 6 mm and a 0.050-mm-diameter orifice produces a beam crater
diameter of 7–10 mm.
A principal difficulty is keeping particles out of the tip. The tip may be
unplugged by passing a 0.003-in.-diameter W wire from the back end. This wire
may be made from larger stock by electrolytic etching in NaOH (aq) solution. Final
cleaning of the tip is done in an ultrasonic bath with acetone.
The beam current and stability depend on the pressure of gas entering the gun.
Stable operation is obtained when the chamber pressure is in the 5 × 10−4 to
21.4 Sputter Ion Gun—Cold Cathode 217
3 × 10−3 Torr range. If the supply pressure is too low, arcing will occur in the
high-voltage nipples on the gas supply line. If the pressure is too high, or the orifice
diameter is too large, a glow discharge will occur in the chamber with little apparent
localization. The beam current increases with supply gas pressure up to the point of
glow discharge initiation. The high-voltage power supply should be current limited
to prevent more than 5–10 mA of current from flowing during such an event.
Another design of a cold cathode sputtering source for metal deposition maybe
found in [12].
Broad-beam ion sources are employed for many purposes including sputter depo-
sition, etching, etc. Figure 21.5 schematically shows a typical conventional source,
which illustrates the basic principles of more complex sources [13, 14]. The
cylindrical chamber contains a cylindrical anode and an axially positioned cathode.
Gas for ionization is admitted from the rear of the source. A magnetic field, sup-
plied by either an electromagnetic winding or a permanent magnet, is aligned
axially, and the magnetic field diverges slightly as one approaches the front of the
source. The purpose of the magnetic field is to cause the electrons emitted from the
cathode to have a spiral path length sufficiently long such that each will cause a gas
phase ionization event before reaching the anode. The source has two grids on its
front side—a screen grid and an accelerator grid. The discharge chamber is nor-
mally operated under low-pressure conditions where the mean free path of an
electron for ionization is much longer than the apparatus dimensions. The plasma
potential is usually within a few volts of the anode potential, and the primary
electrons emitted from the cathode therefore have energies close to the potential
difference between cathode and anode [13, 14].
When Kaufman sources [13, 14] are used with inert gases or reactive gases at
low power levels, they operate for long times without failure. However, at high
pressures with reactive gases like O2, the ion source lifetime is shortened. This is
due to failure of the refractory cathode (such as W or Ta), or to buildup of an
insulating oxide layer on the anode. Such an insulating layer on the anode reduces
the discharge current until it is no longer possible to support a discharge [15]. This
lifetime problem can be reduced significantly by using thorium oxide coated Ir
cathodes (see p. 133) instead of the more standard bare refractory filaments. For
electron emission current densities of 0.2 A/cm2, the thoriated filament lifetime is
1000 h or longer [16]. As the emission current density is increased, the lifetime
decreases as the square of the current density. Multiple filaments, used in series, can
extend the lifetime by reducing the current density at each filament [15].
To avoid the buildup of an insulating layer on the anode, it may be
sputter-coated with a 1-μm-thick film of IrO2, which is a conductive metallic oxide.
Because this oxide cannot be oxidized further, operation in an intense plasma will
not form an insulating layer, as with many anode metal materials. The source,
containing both a thoria-coated Ir filament and an IrO2-coated anode has been
operated at 100-W discharge power for about 50 h without degradation of
source performance. With an uncoated stainless steel anode, failure occurred in
30 min [15].
Other possibilities exist for using this idea. The anode may be coated with other
con-ductive metal oxides such as those from Ru or Re [17, 18]. Alternatively, the
anode could be made from the oxide by chemical vapor deposition. Finally, the
anode could be made from solid Ir, since an insulating oxide layer will result from
its oxidation by the atomic O and other active oxygen species produced in the
source [15].
Steinbruchel [19, 20] has studied the emission of both neutral and ionized
species produced from a Kaufman-type ion source operating with fluoro- and
chlorocarbon species, as well as Ar, Ne, O2, N2, and BCl3. This work involved a
line-of-sight quadrupole mass spectrometer as the detector.
Ensinger [21] has written an excellent review of a broad range of ion source
designs for many purposes.
220 21 Ions
There is some interest in producing surface chemical reactions with low-energy ions
impinging on the surface. In order to do this effectively, ion current densities at very
low energies must be maintained at sufficient levels to carry out surface reactions in
a reasonable time. A method is described in which the efficiency of collection of a
beam of ions with a diameter of about 7 mm is markedly enhanced with an axial
magnetic field [22].
The ions are generated by a Colutron GD-2 ion beam source, and ions are
produced with an energy spread of as low as ±0.1 eV. The ions are extracted,
focused by an Einzel lens, and mass-filtered with a Wien filter. These ions are
directed at an aperture plate of 12-mm thickness containing a 7-mm-diameter hole,
as shown in Fig. 21.6a. A tungsten grid (90 % transparency) is placed across the
aperture plate. The gap between the grid and the sample to be bombarded is kept
between 2 and 5 cm and the ions are decelerated here. The difference between the
beam potential and the positively biased sample defines the ion energy at the sample
surface. A 2000 G Co-Sm permanent magnet (4-cm diameter) is placed behind the
sample at a distance of about 0.5 cm. This magnet can be electrically biased, and
sometimes a magnet potential of +200 V above that of the sample has been found to
increase the measured ion current collected by the sample. This effect may be due to
the collection of secondary electrons from the grid. The 0.5-cm gap is varied during
ion collection to maximize the collected ion current at the sample. The magnet can
also be removed when surface analytical methods influenced by external magnetic
fields are employed. Obviously the sample must be nonmagnetic.
Figure 21.6b shows the effect of magnetic focusing on an Ar+ ion beam incident
on a Pt sample. A dramatic increase in the collected current below an ion energy of
10 eV is noted, where the initial ion current at 1000 eV is 400 nA. The ion beam
energy spread at the sample is about ±0.5 eV. The paper [22] describing this work
contains an analysis of the effect of the cyclotron motion of the ions in the magnetic
field on their divergence. This method is simple and cost effective compared to
electrostatic focusing methods when a small ion beam spot diameter is not a critical
concern.
The conversion of a conventional high energy ion gun into a low energy gun can be
carried out, preserving the ion current capability as well as the spatial and the
energy resolution capability [23]. Several low energy ion guns have been described
in Sects. 21.1–21.6 of this book, but all requizre the construction of new ion optics.
This design uses existing ion optics and electronic control capabilities of a com-
mercial ion gun. It can be used where it is desired to use low energy ions as reactive
21.7 Low Energy Ion Gun—Construction and Performance 221
Fig. 21.6 Focusing of low-energy ions by magnetic field. a Ion deceleration in magnetic field.
b Ion collection efficiency
species without significant sputter damage at the target. The chemistry measured
will be that of the neutral, ion because of efficient ion neutralization at surfaces.
Figure 21.7a shows a schematic of the original and modified ion guns.
Essentially, the conversion to the low ion energy regime is achieved by biasing the
entire ion gun to about −5 kV. This is achieved by the use of ceramic standoffs to
222 21 Ions
Fig. 21.7 Low energy ion gun. a Modification to the ion gun for low energy application. b Ion
potential energy. c Energy spread. d Ion beam profile
isolate the outer case of the ion gun at its 2¾ in. Conflat flange, and then to float the
gun at high negative potential. Great care should be used to prevent laboratory
workers from making accidental contact with the lethal retarding voltages on the
electrically isolated ion gun. This can be done by covering everything at high
potential with a plexiglass shield.
Figure 21.7b shows the potential energy surface experienced by a positive ion
under conditions when it is desired to deliver the ion to a surface with 10 eV of
kinetic energy. Most of the loss of ion energy occurs near the exit of the ion gun,
but small additional retardation occurs near the front of the sample as shown. In this
particular simulation [24], the sample and the walls of the vacuum are at ground.
The measured beam energy spread as a function of ion energy is shown in
Fig. 21.7c. About 2 eV change in halfwidth of the ion beam is seen over the energy
range 10–50 eV.
21.7 Low Energy Ion Gun—Construction and Performance 223
By moving the sample across the ion beam, an approximate estimate of the beam
profile could be made. Using the % FI potential shown, the F2 % potential is varied.
For F2 from 77.8 to 91.1 %, total beam widths from 3–4 mm are observed. As F2 is
raised to 96.3 % the beam width degrades to 8–9 mm. In addition the maximum ion
current received by the sample decreases at high values of F2. Ion beam profiles are
shown in Fig. 21.7d.
The size of ion beams used in research can be measured easily [25] by using the fact
that ion-irradiated Gd2(MoO4)3, called GMO, becomes passivated toward attack by
dilute hydrofluoric acid, HF(aq). This passivation occurs for the c-surfaces (001
plane) of GMO crystals, when irradiated with 1–2 keV He+, Ne+, and Ar+.
The GMO crystal is polished down to a surface finish achieved with 1-μm
alumina abrasive and then may be irradiated with the ion beam as shown in
Fig. 21.8. Following irradiation, the crystal is removed and etched with 1 % HF(aq).
The area exposed to the ion beam is almost unaffected by the HF, whereas the
surrounding areas are etched away. The imaging of the ion beam is reported to have
l-μm lateral resolution, and to be much more sensitive than the competitive tech-
nique in which a Ta2O5 anodic film on Ta is bombarded until metallic Ta is
exposed. The authors warn that the technique is an integrating technique, and the
pattern will therefore reflect any wandering of the ion beam during the irradiation
period. The technique is quite useful for grazing incidence ion beams.
Three more accurate methods for measuring ion beam sizes and the divergence of
an ion beam are given in [26]. Here, a HgCdTe substrate covered with a photoresist
is used in one method, or movable collectors are used in the latter two methods.
Varying the value of the resistance, R, over a wide range has little effect on the
meter deflection, but R should be kept as small as possible to avoid charge buildup
on the segments that could cause beam aberrations.
There are many advantages in pulse counting during the measurement of either
electron diffraction or ion angular distribution patterns produced by electron beam
excitation of surfaces (ESDIAD). Pulse counting affords the maximum sensitivity to
particle measurement, assuming)each pulse is counted. The digitized information
arising from pulse counting allows easy handling in computer analysis and error
analysis. But most importantly, the high sensitivity of pulse counting techniques
allows one to work with electron beams having low intensities, thereby minimizing
damage to the surface layer during measurement [28, 29].
Figure 21.10a shows a LEED/ESDIAD apparatus [30] that uses four grids in
front of a pair of microchannel plate (MCP) amplifiers [31]. Charged particles (and,
in some cases, excited neutral particles) are transmitted through the grid system to
the microchannel plate amplifiers, where each particle exciting the microchannel
plate produces a cascade of electrons that exit the back of the second microchannel
plate and are accelerated into a two-dimensional resistive anode detector [32]. This
pulse of about 105 or 106 electrons expands radially from the point of impact and is
collected at the four corners of the resistive anode. Since the local intensity of
expanding charge decreases as 1/r, the pulse height arriving at the four corners can
be used to calculate the coordinates of the point of impact (the centroid). The pulses
at the four corners are capacitively coupled into four preamplifiers, A-D. These
amplified pulses then travel into the position computer [33]. The position computer
then generates a pair of pulses which are proportional to the x- and y-coordinates of
the impact point on the resistive anode. The x- and y-pulses are then sent to a pair of
analog-to-digital converters that form the input to either a multichannel analyzer [34]
or to a computer. A three-dimensional display of the collected angular distribution of
the electron or ion or excited neutral pattern is then displayed. It is important to point
out that the excitation of the crystal surface by the electron beam will produce a
background due to soft X-rays, and this must be subtracted from the measured
patterns [35, 36]. The ultimate spatial resolution of the apparatus is about 0.3°.
Typical electron beam currents of the order of 1 × 10−9 A are employed over time
periods of the order of 100 s. Damage to even the most electron-beam-sensitive
surface species is generally less than 1 %. The positive ion patterns viewed in this
apparatus may be compressed by applying a positive potential, Vc, to the crystal.
Examples of both typical LEED patterns and ESDIAD patterns are shown in
Fig. 21.10b, c for the case Cu(110) and CO chemisorbed on Cu(110) [37]. An
earlier example of this technology applied to LEED measurements alone can be
found in [38].
226 21 Ions
Fig. 21.10 Pulse counting LEED/ESDIAD. a Pulse counting LEED/ESDIAD analyzer. b Typical
LEED pattern—Cu(110). c Typical ESDIAD pattern—CO/Cu(110)
The counting efficiency for ions on a MCP detector has been carefully measured
using a field evaporation apparatus [51]. It was found that the application of a
positive bias on the front surface of the channel plate did not increase the
capture/counting efficiency much above that expected from the fractional area of
open channels present on the plate. It was, however, argued by others that a small
positive potential (20 V) facilitated capture of secondary electrons and hence
enhanced the counting efficiency [52], and this effect was verified in a careful paper
in which the absolute detection efficiencies for He+, H+, and O+ ions was directly
measured using a Faraday cup to compare with the count rate. Ion detection effi-
ciencies ranged from 18 to 63 % in the energy range 250–5000 eV [53]. These
workers also measured the detection efficiency as a function of incident angle,
finding that an incident angle of about 10° yielded maximum efficiency, with the
efficiency falling 24 % in the range 20–40° off the channel axis direction [53].
The basic physical properties of straight through MCP arrays, chevron MCP arrays,
and curved channel MCP arrays are discussed in some detail in [54].
Field penetration effects from one region of the LEED apparatus to another can
be prevented in most cases by using neighboring grids at the same bias potential to
screen elements at high potential (such as the front of the MCP) from the interior of
the grid system [55].
The hemispherical grid analyzer may be used as a retarding field energy analyzer
for either electrons or ions. For these measurements, the retarding potential ramp is
applied to a central grid along with a small sinusoidal modulating voltage, and
lock-in detection is applied to the transmitted current. This has been discussed in
detail in [56].
A planar grid ESDIAD apparatus has been used to monitor the angular distribution
of ions emitted from an insulator surface as a result of excimer laser pulses [57].
Section 21.10 describes the design and use of an ESDIAD apparatus in which
positive ions, produced in electron stimulated desorption (ESD), are imaged spa-
tially as they are ejected from surface species. These measurements are used to
determine the chemical bond directions in adsorbed species on single crystal sur-
faces [29] and the dynamical behavior of the molecules as they undergo low fre-
quency motions on their binding sites [58].
Detailed information about desorbing ions can be obtained if their momentum is
also measured, using a pulsed electron beam, and measuring the flight time between
the crystal and the detector. The flight time will directly yield the ion velocity,
except that it is safest to measure changes in flight times with various accelerating
potentials to derive the ion energy and mass (i.e. the momentum). An apparatus [59]
to measure the time-of-flight (TOF) of desorbing ions is shown in Fig. 21.11a. The
apparatus is capable of measuring the desorbing direction and the flight time for
228 21 Ions
Fig. 21.11 Time of flight measurements in ESDIAD. a Schematic circuit for TOF measurement of
positive ions in ESDIAD. b Time sequence of pulses and signals. c ESDIAD analyzer and pulsed
electron beam. d Time-of-flight separation of H+ and D+. e Benzoate on Cu(110). f ESDIAD
patterns of H+ and D+ from isotopic benzoate molecules
21.11 Time-of-Flight Measurements for Ions Generated in ESDIAD 229
each ion comprising the ESDIAD pattern. Time resolution is in the picosecond
range and data acquisition up to 100 kHz is achievable. This enables the decon-
volution of the spatial distribution and the momentum of each ion. The reader is
referred to [59] for details of the electronics shown in Fig. 21.11a.
Figure 21.11b shows the time sequences for various pulses. Initiated by a trigger
signal, an electron gun ejects an electron pulse of variable time width down to 5 ns.
Positive ions A and B arrive at the MCPs at different times. The flight time of an ion
is recorded along with the x and y coordinates of its arrival at the MCP. A drawing
of the hemispherical analyzer and the pulsed electron and ion beams is shown in
Fig. 21.11c. The ions, after being amplified and converted to electrons in the MCPs,
are counted by the resistive anode (RA) which yields the coordinates of ion arrival
there. For the adsorption of the benzoate ion, oriented normally on Cu(110) by
adsorption through the COO group, three hydrogenic ion beams are shown in
Fig. 21.11e corresponding to the 3-, 4-, and 5- positions on the phenyl ring. By
using the TOF method, the two H+ ions corresponding to the 3- and 5- C–H bonds
are measured at 60° to the normal. Using a deuterium label at the 4-position, a D+
beam is observed at the 4-position which produces a normally-oriented beam
[60, 61]. The D+ ion has a longer TOF due to its heavier mass as shown in
Fig. 21.11d. The separate ion angles are plotted in Fig. 21.11f as contour diagrams.
The difference in yield of H+ and D+ is due to several factors related to the departing
speed of the ion which strongly controls its neutralization and to smaller differences
in the rate of the neutralization of the departing ions originating from different
positions on the phenyl ring [61].
In photon or electron stimulated desorption one may require both mass analysis of
the ions as well as energy resolution. This problem was first addressed by Feulner
[62, 63], using a simple retarding field analyzer coupled to a microchannel plate
(MCP) assembly and readout by a multichannel analyzer (MCA), as shown in
Fig. 21.12a. The macrostructure of the ion peaks yields the masses of the ions in
each peak; the microstructure (peak shape) yields the energy distribution as sche-
matically indicated in the spectrum shown.
A more elaborate analyzer can be made from a hemispherical electron energy
analyzer as shown in Fig. 21.12b. If “single punch (pulsed) light” is not available,
an electrostatic gate is placed in front of the entrance aperture to the hemispherical
analyzer. It is a pair of deflection plates placed inside a screening box. A symmetric
square wave is applied to these deflection plates in order to modulate the trans-
mitted ion beam. The signals from the 5 channeltron multipliers—used for
230 21 Ions
Fig. 21.12 Ion detector—kinetic energy and mass resolution. a Retarding field TOF [58].
b Hemispherical-TOF analyzer
21.12 Ion Detection with Kinetic Energy and Mass Resolution … 231
increased sensitivity—are fed into multi-page, multi-channel scalars that are syn-
chronized with the modulation of the ions. In this manner, 2D fields are obtained
with the axes “kinetic energy” and “flight time”. From this information the kinetic
energy distributions of the individual ions can be extracted [62].
The detection of neutral atoms and molecules that have been generated from sur-
faces using scattering, electron- and photon-stimulated desorption, and ion-induced
desorption is facilitated by tunable nanosecond pulsed lasers. Laser excitation
permits particle detection with quantum state specificity using laser-induced fluo-
rescence (LIF) or resonance-enhanced multiphoton ionization (REMPI) methods.
The REMPI technique has proven to be more popular because of the high sensi-
tivity of the ion detection technique and the ability to image ion angular distribu-
tions analogous to the ESDIAD technique (see p. 225).
The apparatus shown in Fig. 21.13 was designed to detect laser-ionized neutral
atoms and molecules that have undergone electron stimulated desorption from a
single crystal surface [64, 65]. It is based on the design developed by N. Winograd
et al. [66] for the detection of neutral atoms desorbed by ion bombardment. The
pulsed electron gun shown in the Figure therefore can be replaced by a pulsed ion
gun or a pulsed laser, and the detection scheme is the same. The principle is as
follows. Neutral products that leave the surface as a result of excitation by the
pulsed source must travel a specified distance before they enter the ionizing laser
beam, which is directed parallel to the crystal surface. The laser generated ions are
then accelerated into the chevron electron multichannel plate array (CEMA)
detector [67]. The output of the detector is coupled to a phosphor screen. Provided
that the laser is focused in a sheet-like fashion, the visual display on the phosphor
screen provides angular resolved patterns of the desorbed neutrals, as in the
ESDIAD method, but now with full internal quantum state resolution [65, 66]. The
integrated signal can be detected with a gated integrator or the two-dimensional
image can be digitized using a charge-coupled device (CCD) camera.
By pulsing the electron beam rapidly (1 μs), the velocities of the neutrals can be
measured by the time delay between the pulses of the electron gun and the laser.
Desorbed ions can be a problem, but they can be swept away prior to the laser pulse
by an applied potential to the first grid and/or the crystal. It is also possible to
discriminate against unwanted signals by pulsing the applied voltage on one of the
multichannel plates. Recently several groups have oriented the detection assembly
90° to the surface normal, so that the spread in velocities is obtained simultaneously
with the angular distribution [68].
References
66. R.H. Kobrin, G.A. Schick, J.P. Baxter, N. Winograd, Rev. Sci. Instrum. 57, 1354 (1986)
67. CEMA assemblies are available from Galileo Electro-Optics Corp., Galileo Park, P.O. Box
550, Sturbridge, MA 01566
68. M. Menges, B. Baumeister, K. Al-Shamery, B. Adam, T. Mull, H.-J. Freund et al.,
in Desorption Induced by Electronic Transitions, DIET V, ed. by A.R. Burns, E.B. Stechel,
D.R. Jennison (Springer, New York, 1993), p. 275
Chapter 22
Photons (UV)
The use of vacuum ultraviolet sources for valence band photoemission studies of the
surfaces of solids is widespread, and many such studies are made on synchrotron
sources. Small laboratory vacuum UV sources predated synchrotron sources, and are
still widely used. A high-intensity source delivering both He(I) and He(II) radiation,
depending upon the discharge conditions, is shown schematically in Fig. 22.1a [1].
He gas is admitted to a differentially pumped region containing a hollow cathode,
and is pumped away continuously by a trapped forepump. A second stage of dif-
ferential pumping occurs closer to the ultrahigh vacuum chamber using a diffusion
pump that is backed by the same forepump. The forepump (rated at 500 L/min)
speed was measured to be 100–200 L/min at the lamp and 30–70 L/min at the back
of the diffusion pump for typical operation [2]. The forepump was trapped with a
water-cooled Cu trap and a zeolite trap. An 8-cm-diameter port at the lamp leading to
the liquid nitrogen trap before the diffusion pump was used [1], as an aid to improve
pumping speed over that originally used [2]. This large port was made from a
straight-through valve. This permits the lamp to be isolated from the UHV system.
Figure 22.1b shows the details of the He lamp. A high-voltage UHV feed through
holds a hollow cathode cylinder made of 0.5 mm Ta sheet. This cathode, 2 cm in
diameter and 4–6 cm in length, is held inside a glass tube in a location in which a
stainless steel capillary (2.3-mm internal diameter) picks up radiation at a distance of
1–1.5 cm from the end of the cathode. The glass discharge region can be removed for
cleaning. A needle valve admits He into the side of a Conflat flange that mates to the
flange holding the first stainless capillary tube. Below the first capillary is the region
pumped by the forepump, region A-B. A second capillary, mounted between B and
C, directs the radiation to a third capillary which enters the UHV chamber. The He
pressure in the system with the lamp operating is approximately 1 × 10−8 Torr.
Contaminant gases are measured to be below 10−10 Torr, although such gases will be
Fig. 22.1 Hollow Cathode resonance lamp—He(II) source. a Pumping system. b VUV lamp.
c Au UPS spectrum
highly directed as a beam to the sample by the capillaries, and one should be alert to
the possible build up of surface contamination.
The performance of the lamp for both He(I) (21.2 eV) and He(II) (40.8 eV), and
for 48.4 eV radiation, is shown in Fig. 22.1c, where an evaporated Au film was
studied, using a double-pass CMA electron spectrometer for measurement of the
photoelectron spectrum at a constant resolution of 0.2 eV. The pressure of He in the
discharge was about 1 Torr for optimum He(I) intensity and 0.1–0.2 Torr for
40.8 eV and 48.4 eV photons. This lamp has improved He(II) intensity compared to
earlier lamps utilizing a microwave discharge [3].
Fig. 22.2 He(I) and He(II) UV source. a Cold cathode discharge UV source. b Differential
pumping system
through a 0.1-in.-diameter exit aperture between the anodes at the front of the
lamp. An aperture with a diameter less than one-third the inner diameter of the
cathode allows sampling of the radiation without allowing electrons to escape [9].
The gas used here was 99.9995 % He, which flows through an activated charcoal
trap at 77 K and a variable leak valve. A 0 to +5 kV DC power supply (0.500 mA)
is connected through a variable resistance ballast set at 16 kΩ. This prevents current
surges during initiation of the discharge whose resistance varies from infinity
(before activation) to several kΩ during operation.
The pumping system in the lamp consists of two stages of differential pumping,
followed by a third stage in the UHV chamber. The first stage, across a 1.5-mm-i.d.
Pyrex capillary, 3 cm long, is pumped by a chilled water-trapped 325-L/s oil
diffusion pump. This stage can be eliminated by opening a bypass valve between
entrance region and the first stage of diffusion pumping. The second stage is also
across a similar capillary and is pumped by a liquid-nitrogen-trapped 325-L/s oil
diffusion pump. The third stage of pumping across the exit capillary (2.5 mm i.d,
22.86 cm long) is pumped by the UHV system pumps, which consist of a 450-L/s
turbopump and a 500-L/s ion pump. A gate valve is used to isolate the source from
the UHV system. The original paper shows plots of photoelectron intensity from a
Au surface versus pressure [4].
Under optimal operating conditions, the He(I)/He(II) intensity ratio produces a
10/1 ratio of photoelectrons from the two spectral lines from the Au surface.
Although ultraviolet induced photochemistry research has been carried out for more
than 60 years in homogeneous phase, detailed studies of photochemical processes
on well-defined single crystal surfaces are much more recent. Because ultraviolet
lamps emit more photons over an extended exposure time than most laboratory
lasers, these lamps are often used for inducing photochemical processes on sur-
faces. Both intensity measurements and photon energy range selection using filters
are conveniently carried out, permitting good measurements of cross sections
versus photon energy for surface process induced photolytically.
Figure 22.3a shows an ultrahigh vacuum chamber [10] set up so that a single
crystal may be irradiated with ultraviolet light from a high-pressure Hg arc source
[11]. The single crystal is located in front of an aperture that leads to a quadrupole
mass spectrometer in a differentially pumped section, permitting the direct mea-
surement of photodesorbed species. In addition, the mass spectrometer maybe used
also for thermal desorption studies after ultraviolet irradiation to access the changes
that have occurred due to irradiation of an adsorbed layer. The ultraviolet light
passes through a 10-cm distilled water filter that filters out infrared radiation.
A 200-mm f/1 quartz condensing lens focuses the light onto the crystal through a
UV-grade sapphire window. Interference filters (Oriel Corporation) centered on the
22.3 Photochemistry on Surfaces Using an Ultraviolet Lamp 239
Fig. 22.3 Photochemistry using ultraviolet lamp. a Ultraviolet lamp for photochemistry using
filters and polarizer. b Typical cross section measurements from TPD analysis
240 22 Photons (UV)
intense lines of the mercury arc are often used to produce narrowband radiation and
have a transmission of from 15 to 50 % at the center of the band with a full width at
half maximum (FWHM) of 10 nm, except for the 237-nm filter, which has a
FWHM of 20 nm. One must be careful in using these filters for long irradiation
times, since the transmission band center can shift up to 2 nm when extensive
irradiation occurs.
The apparatus contains a quartz flat partial reflector used to measure continu-
ously the light intensity. The quartz flat reflects roughly 10 % of the light, and this is
directed to a neutral density filter (1 % transmission) and then into a blackened
housing which holds a calibrated photodiode [12]. The calibration of the photo-
diode relative to an absolute thermopile detector can be done externally on an
optical bench, with the thermopile being placed in the same position as would be
occupied by the crystal. Such measurements have found that the intensity of UV
radiation from the arc source is not uniform, and it is interesting to map out the
intensity pattern which gives a view of the arc location in the source in the most
intense region of emission.
A commercially available UV quartz prism polarizer may be placed into the
beam if either s- or p-polarized radiation is desired (see p. 240).
By combining the photoirradiation experiments at various photon energies with
analysis of the surface before and after irradiation (using thermal desorption in this
case) it was possible to observe two major separate photochemical processes for
molecular oxygen chemisorbed on Pd(111)-desorption of O2 and dissociation to
produce chemisorbed O. The cross sections for these two processes are plotted
versus photon energy in Fig. 22.3b, giving an idea of the random errors inherent in
this type of measurement [10]. These cross sections were also measured in another
laboratory using laser-induced photochemistry and the results were in agreement to
within a factor of 2–3 [13].
A photoreactor for studies of powders and single crystals has been described
in [14]. This device can operate at high or low gas pressures.
absorbed energy will vary in accordance with Fresnel’s laws and the optical con-
stants of the crystal absorber. Figure 22.4b shows a calorimetric experiment in
which polarized broad band radiation having 0.23 W/cm2 power density is allowed
to warm the crystal from 79 K. For an incidence angle of 60°, one can see that the
p-polarized radiation is more efficiently absorbed than the s-polarized radiation, as
indicated by the relative increases in temperature. The measured difference in
absorptivity is in fair agreement with the optical constants considering the errors
inherent in this type of experiment [17].
Figure 22.4c shows the comparative effect of s- and p-polarized radiation on a
molecular O2 layer on Pd(111) as studied by thermal desorption following various
periods of irradiation. The α1- and α2-molecular oxygen states desorb by phot-
ostimulation, and also photochemically dissociate to produce a thermal desorption
feature labeled β2. It may be seen that p-polarized radiation is more efficient in
causing these desorption and dissociation processes than is s-polarized radiation
[17]. This is consistent with a process in which electronic excitation of the metal by
ultraviolet light is responsible for the surface photochemical processes, but more
detailed studies would be needed to prove this.
While many chemical methods may be used to witness photoactivity and its
enhancement on semiconductors, the production of clusters of Ag by photoreduction
of Ag+ ions in solution to produce Ag0 clusters is novel in its ability to statistically
witness the reaction, using AFM to measure the cluster density and size [18–21].
As shown in Fig. 22.5a, slices of a TiO2(110) crystal, glued to a microscope
slide, are immersed in a 0.1 M AgNO3(aq) solution. Exposure to about 1017
photons/cm2 causes photo-produced electrons in the TiO2 to reduce Ag+ ions to
Ag0 atoms which cluster together to produce AFM-visible deposits. Figure 22.5b
shows an AFM image of the undoped TiO2 compared to the doped TiO2 and the
enhanced density of silver clusters clearly indicates that doping was effective in
enhancing the frequency of Ag+ reduction events. The volume of the Ag clusters
produced allows an estimate of the quantum yield for the process, and Fig. 22.5c
shows a plot, versus photon energy, of the quantum yield which falls in the 1 %
range. In the case shown, doping was done by treating the TiO2(110) crystal in
NH3(g) at 1 atm and 870 K for 5 min. Initial heating and cooling of the crystal to
870 K was achieved in Ar(g) at 1 atm.
Figure 22.5d shows the N(1s) XPS spectrum achieved after doping. Spectrum a
corresponds to the removal of about 5 Å of the surface by sputtering in order to
achieve a clean surface after handling in air and NH3. After the removal of 60 Å of
material, two N(1s) features are observed. It was shown that the 399.6 eV feature,
due probably to N-H type species, is active for doping, giving a photon threshold
energy at about 2.4 eV which is 0.6 eV below the bandgap of rutile TiO2 [21].
22.6 Reducing Radiation Damage in Raman Spectroscopy of Solids 243
has been widely used in IR spectroscopy as discussed elsewhere in this book. The
grid is held in a square frame with a 1.4 × 1.4 cm2 opening and consists (for
strength) of a front frame made of stainless steel (1.5 mm thick) and a rear Cu
(OFHC) frame for maximum thermal conductivity, which is brazed to Ni leads. Six
bolts and nuts clamp the frame to the grid. The sample is mounted in a Pyrex
Raman cell with provisions for cooling with liquid nitrogen.
The arrangement was compared to a conventional rotating Raman sample holder,
using an OMARS 89 triple-monochromator spectrometer (Dilor, Lille, France)
equipped with a thermoelectrically cooled CCD camera (Princeton Instruments).
A test was carried out with a WOx/ZrO2 catalyst. The Raman spectrum was recorded
with a rotating holder as shown in Fig. 22.6b, using three laser power levels from a
488-nm Ar+ ion laser (Spectra Physics, Type 2020). Since the spectrum does not
change during irradiation at any level, the rotating sample holder is able to prevent
radiation damage. In Fig. 22.6c, radiation damage is not seen for the mesh-supported
sample until the power is increased above 90 mW, where a new band at 1015 nm
appears. An Os3(CO)12/SiO2 was studied also, and the cryogenically-cooled sample
was stable to 24 mW power. These studies show that non-rotating cryogenically
cooled powdered samples held in a grid may be stabilized against laser damage, but
that the rotating sample method is superior for reducing damage [24].
Fig. 22.6 Reducing radiation damage in Raman spectroscopy of solids. a Cryogenically cooled
non-rotating Raman cell. b, c Raman spectrum—WOx/ZrO2
22.7 Lyman-α Lamp Calibration Using a Chemical Actinometer 245
The measurement of the light flux from vacuum UV lamps is important in a number
of areas of photochemistry such as astrochemistry and atmospheric chemistry.
Different methods are used to calibrate such lamps including photoelectron gen-
eration, photoionization of gases, and photodecomposition of gases [26–31]. There
are only a few reports of the use of chemical actinometry for the calibration of
Lyman-α lamps [29–36] (which operate at 10.2 eV). The results of a variety of
calibrations of these lamps are tabulated in [37] and show typical light fluxes
measured over a 47 year history. It can be seen that the various Lyman-α lamps
used in the past have been shown by various methods to deliver approximately
1015 photons s−1 cm−2 over a range of approximately 10-fold.
A chemical actinometer successfully used recently for the calibration involves
the photodecomposition of N2O and the use of transmission IR spectroscopy to
measure the N2O destruction as well as the production NO and NO2 [37]. N2O
decomposes when illuminated by Lyman-α radiation according to the reactions
(22.1)–(22.4):
N2 O þ hm ! N2 þ O ð22:1Þ
N2 O þ hm ! NO þ N ð22:2Þ
1=2O þ 1=2N2 O ! 1=2N2 þ 1=2O2 ð22:3Þ
1=2O þ 1=2N2 O ! NO ð22:4Þ
The quartz Lyman-α lamp was mounted with a Viton O-ring onto a stainless
steel Conflat flange containing a MgF2 window. The lamp was supplied with a
continuous flow of a 10 % H2/90 % Ar mixture at 1 Torr pressure and was excited
by a 60 W microwave supply. The MgF2 effectively cuts off radiation at wave-
lengths above 1100 Å, thereby removing Ar emission [38]. The vacuum UV
radiation passes into a bakeable high vacuum stainless steel transmission IR cell
(Volume = 1.9 L) equipped with turbopumping, Baratron capacitance manometer
pressure gauges, and a mass spectrometer, and having a base pressure of about
1 × 10−8 Torr as shown in Fig. 22.7a. For the lamp calibration, between 1 and
10 Torr of N2O was admitted to the cell and the lamp was turned on and stabilized.
The IR spectrum for N2O and the products during photolysis is shown in Fig. 22.7b.
The rate of consumption of N2O was measured by following the integrated
absorbance (at 2 cm−1 resolution) at two strong fundamental frequencies, ν1 and ν3,
where the IR beam travelled transversely through the cell. Separate pressure versus
integrated absorbance curves were measured for N2O at these frequencies, and the
consumption of N2O, calculated in Torr s−1, was determined. Figure 22.7c shows a
sample of the N2O depletion curves where about 10 % consumption of N2O was
measured, yielding linear plots. It may be seen that the values of dP/dt, measured
246 22 Photons (UV)
Fig. 22.7 Calibration apparatus for N2O photodecomposition by Lyman-α radiation. a FTIR cell
and Lyman-α source. b Spectral evolution of N2O, NO and NO2 features during the
photodecomposition of N2O. c Determination of photon flux
22.7 Lyman-α Lamp Calibration Using a Chemical Actinometer 247
from ν1 and ν3, are in excellent agreement, with less than 1 % difference. The flux
of the Lyman-α light was estimated using a measured efficiency from the literature
[39] of 0.69 ± 0.02 photons/N2O. One should not employ pressure changes in
estimating the light flux as heating effects were found to confuse such measure-
ments. It is interesting to recognize that in the pressure region from 1 to 10 Torr, all
Lyman-α radiation is absorbed in the first few cm from the MgF2 window and the
assumption of fast mixing of the gas was found to hold. In addition, a slight
pressure dependence of the photodecomposition kinetics was found and attributed
to wall effects which become more significant at low pressure [37].
Photochemistry studies are being extended to higher and higher photon energies
because of scientific interest in processes involving deep ionization in molecules, in
solids and on surfaces. A source of radiation of this energy range can be UV lamps,
such as the Lyman-α lamp (10.2 eV), and various other gas lamps excited by
radiofrequency radiation (see Sect. 22.7). Another source of such radiation is syn-
chrotron radiation which extends far into the X-ray region, and which is commonly
monochromatized for use in photoelectron spectroscopy. A laser plasma light source
is another source of Vacuum UV (VUV) or Extreme UV (EUV) light. In the example
shown here, a Ta plasma is generated by a 1064 nm pulsed laser and light is
monochromatized by toroidal gratings in the vacuum spectrometer for the range 4–
108 nm (310–11.5 eV). The apparatus is more than 3 m in length and is pumped by
turbomolecular pumps (TMP) as well as Ti sublimation pumps (TSP) as shown.
As shown in the data below the spectra in Fig. 22.8c, 106 to 108 photons per
pulse are generated which over the duty cycle yields 5 × 107 to 5 × 109 photons/s
averaged over a long time period [40]. This is a far smaller number of photons than
can be generated by a gas discharge lamp such as a Lyman-α lamp, where over the
whole spectrum in the region of 10.2 eV, about 1015 photons/s are generated [37].
GIXPS was first described by Henke [41] and has seen limited use. The technique is
based upon the angle-dependent X-ray penetration depth at small incidence angles
(measured from grazing angle). X-rays that impinge on a flat surface at *1 keV at
sufficiently small angles undergo total external reflections and their penetration
depth is strongly attenuated. This means that the photoelectron yield contains an
increased contribution from the surface of the solid. As the angle of the X-rays
increases, increased contributions from deeper layers are observed. Therefore, depth
dependent XPS analysis is possible. The theory behind GIXPS is given in [42].
248 22 Photons (UV)
Fig. 22.8 Pulsed vacuum UV light source. a Top view. b Side view. c Intensity (photon/pulse)
Figure 22.9a shows a laboratory design for making GIXPS measurements, using
a small X-ray source. The X-ray source may be rotated in angle of incidence and the
center of rotation is the analysis position on the sample. The microchannel plate
amplifier is used to measure the reflected X-ray radiation. A double pass cylindrical
mirror electron analyzer is used. Normalized sample spectra for oxidized and
unoxidized GaAs(100) are shown in Fig. 22.9b, and it may be seen that at higher
angles (66 mR) the relative contribution from the bulk—derived As0(3d) feature
(binding energy = 41 eV) is enhanced compared to As5+ and As3+ which come from
the surface oxide film and overlap near a binding energy = 45 eV [42]. A patent has
been issued [43].
Figure 22.9c shows a schematic for the performance of GIXPS using synchro-
tron excitation. Here, the sample is rotated in the X-ray beam, maintaining the
22.9 Grazing Incidence X-Ray Photoemission … 249
References
1. J.E. Rowe, S.B. Christman, E.E. Chaban, Rev. Sci. Instrum. 56, 759 (1985)
2. J.E. Rowe, S.B. Christman, E.E. Chaban, Rev. Sci. Instrum. 44, 1675 (1973)
3. T.V. Vorburger, B.J. Waclawski, D.R. Sandstrom, Rev. Sci. Instrum. 47, 501 (1976)
4. G.M. Lancaster, J.A. Taylor, A. Ignatiev, J.W. Rabalais, J. Electron Spectrosc. Relat. Phenom.
14, 143 (1978)
5. A.H. Mcllraith, Nature 212, 1422 (1966)
6. A.H. Mcllraith, J. Vac. Sci. Technol. 9, 209 (1972)
7. R.K. Fitch, T. Mulvey, W.J. Thatcher, A.H. Mcllraith, J. Phys. D3, 1399 (1970)
8. R.K. Fitch, T. Mulvey, W.J. Thatcher, A.H. Mcllraith, J. Phys. E4, 533 (1971)
9. A.H. Mcllraith, private communication to J.W. Rabalais
10. L. Hanley, X.-C. Guo, J.T. Yates Jr., J. Chem. Phys. 91, 7220 (1989)
11. Oriel, 250 Long Beach Boulevard, P.O. Box 872, Stratford, CT 06497-0872 (Model 66014,
350 watt)
12. The l-cm2-area circular photodiode-preamplifier package is sensitive from 185 to 1150 nm,
and is calibrated by the manufacturer with standards traceable back to the National Institute of
Science and Technology (EG and G HUV-4000B)
13. F. Weik, A. de Meijere, E. Hasselbrink, J. Chem. Phys. 99, 682 (1993)
14. A.Z. Moshfegh, A. Ignatiev, Rev. Sci. Instrum. 59, 2202 (1988)
15. S.A. Buntin, L.J. Richter, D.S. King, R.R. Cavanagh, J. Chem. Phys. 91, 6429 (1989)
16. E. Hasselbrink, S. Nettesheim, M. Wolf, A. Cassuto, G. Ertl, in Springer Series in Surface
Science, vol. 19, DIET IV (1990), p. 75
17. J. Yoshinobu, X.-C. Guo, J.T. Yates Jr., Chem. Phys. Lett. 169, 209 (1990)
18. P.D. Fleischauer, H.K.A. Kahn, J.R. Shepherd, J. Am. Chem. Soc. 94, 283 (1972)
19. P.A.M. Hotsenpiller, J.D. Bolt, W.E. Farneth, J.B. Lowekamp, G.S. Rohrer, J. Phys. Chem.
B 102, 3216 (1998)
20. J.B. Lowekamp, G.S. Rohrer, P.A.M. Hotsenpiller, J.D. Bolt, W.E. Farneth, J. Phys. Chem.
B 102, 7323 (1998)
21. O. Diwald, T.L. Thompson, T. Zubkov, E.G. Goralski, S.D. Walck, J.T. Yates, Jr., J. Phys.
Chem. B 108, 6004 (2004)
22. J.M. Stencel, Raman Spectroscopy for Catalysis (Van Nostrand Reinhold, New York, 1990),
p. 32
23. S. Chan, A.T. Bell, J. Catal. 89, 433 (1984)
24. E.A. Wovchko, J.T. Yates Jr., M. Scheithauer, H. Knozinger, Langmuir 14, 552 (1998)
25. T.H. Ballinger, J.C.S. Wong, J.T. Yates Jr., Langmuir 8, 1676 (1992)
26. T.A. Chubb, H. Friedman, Rev. Sci. Instrum. 26, 493 (1955)
27. J.J. Bollinger, F.M. Pipkin, Rev. Sci. Instrum. 52, 938 (1981)
28. K. Watanabe, F.F. Marmo, E.C.Y. Inn, Phys. Rev. 91, 1155 (1953)
29. H. Cottin, M.H. Moore, Y.B. Nilan, J. Astrophys. 590, 874 (2003)
30. M.S. Westley, R.A. Baragiola, R.E. Johnson, G.A. Baratta, Nature 373, 405 (1995)
31. P. Warneck, Appl. Opt. 1, 721 (1962)
32. H. Okabe, J. Opt. Soc. Am. 54, 478 (1964)
33. D. Davis, W. Braun, Appl. Opt. 7, 2071 (1968)
34. I. Yamashita, Jpn. J. Appl. Phys. 14, 70 (1975)
35. P. Weber, J.M. Greenburg, Nature 316, 403 (1985)
36. G.A. Baratta, G. Leto, M.E. Palumbo, Astron. Astrophys. 384, 343 (2002)
References 251
37. M. Rajappan, M. Buttner, C. Cox, J.T. Yates Jr., J. Phys. Chem. A 114, 3443 (2010)
38. A. Duncanson, R.W.H. Stevenson, Proc. Phys. Soc., London 72, 1001 (1958)
39. W.E. Groth, W.E. Schierholz, Planet. Space Sci. 1, 333 (1959)
40. I. Arakawa, private communication
41. B.L. Henke, Phys. Rev. A 6, 94 (1972)
42. M.A. Chester, T. Jach, J. Vac. Sci. Technol., B 11, 1609 (1993)
43. Jach et al., United States Patent 5,280,176 (1994)
44. T. Jach, E. Landree, J. Surf. Anal. 9, 339 (2002)
Chapter 23
Kinetics of Adsorption/Adsorbate
Coverages
The measurement of the kinetics of adsorption and the absolute saturation coverage
of an adsorbate on a single crystal is basic to our understanding of surface chem-
istry. While relative surface coverages, compared to the saturation coverage, are
fairly easy to measure using various surface spectroscopies, the measurement of the
absolute coverage and the absolute sticking coefficient are more difficult. One of the
absolute methods employed in some cases involves the use of nuclear activation of
the adsorbed layer, where absolute counting of nuclear disintegration events after
activation may be used with high accuracy [1]. Unfortunately, not all adsorbates are
amenable to this method.
Figure 23.1a shows an apparatus that is capable of measuring both the absolute
sticking coefficient and the absolute surface coverage during an adsorption experi-
ment. The method is a version of the King and Wells method, developed in 1972 [2].
A multicapillary doser, with an absolutely known flux of effusing molecules in a
known angular distribution (see p. 421) is incident on a single crystal of known size,
located at a known distance from the doser. A quadrupole mass spectrometer,
shielded from direct line of sight from the crystal and doser, measures the random
flux of gas in the vacuum system as a function of the exposure of the crystal to the
incident beam. A shutter can be rotated to intercept the beam, shielding the crystal
from the direct flux, and converting all of the incident beam into a random flux by
reflection from its surface. In order to make an uptake measurement, the crystal is
cleaned, brought to adsorption temperature, and positioned behind the shutter. Gas is
admitted, as shown in Fig. 23.1b, and the system pressure rises quickly to a sharp
point where the rate of rise decreases significantly. At time t1, the shutter is opened,
and gas begins to be adsorbed on the crystal, causing the random flux of gas to
decrease. Region A of the uptake curve represents the first phase of adsorption, and
the horizontal trace indicates that the sticking coefficient is constant (and probably
Fig. 23.1 Absolute adsorption uptake measurement from calibrated beam. a Apparatus for
absolute adsorption uptake measurement. b Typical uptake curve C2H2/Si(100). c Increase in C2H2
coverage beyond tb owing to continued slow adsorption
23.1 Absolute Adsorption Uptake Measurements … 255
unity) in this region. Following adsorption to point ta, it is observed that the random
flux begins to rise in region B as the sticking coefficient decreases below its value in
region A. At the end of the experiment, the random flux has asymptotically risen to
the dashed line, which indicates that, at the sensitivity level of this experiment, the
uptake is complete. Placing the shutter into the beam causes little effect at point tb,
since the complete reflection of incident molecules by the inert shutter is identical in
magnitude to the reflection of incident molecules by the saturated crystal. The area of
region A and region B may be used to compute the final absolute coverage at tb [3].
There are three issues that contribute to the uncertainty of this measurement:
1. The nonhorizontal character of the dashed line against which the integrals are
measured. This error is probably due to slow pumping by the vacuum system
walls, which decreases as the exposure is increased. The empirical solution to
this problem is to treat the dashed limiting line as a baseline; it is designated Peq.
2. The adsorption of gas from the random flux. A small correction can be made for
this effect, which typically amounts to 1 % in the region from t0 to t1. This initial
coverage is designated N(t)initial.
3. The low accuracy of determining the saturation point, tb. Here, Auger spec-
troscopy was employed to show that for long exposures beyond tb, the rise in
adsorbate coverage was 6 % as shown in Fig. 23.1c for the example chosen here.
The equation that governs the adsorption is
Z
NðtÞtotal ¼ NðtÞinitial þ ðF þ Fr Þ SðtÞdt; ð23:1Þ
where N(t)total = the total coverage at saturation; F = the beam flux; Fr = the random
flux; and S(t) = the sticking coefficient. S(t) = [Peq − P(t)]/fPeq and f = the fraction of
the beam intercepted by the crystal. The value N(t)total is then increased to account
for the small correction mentioned above in issue 3.
A variation on this technique uses a diffuse source of adsorbate from an
uncalibrated leak valve and a collimated detector used to measure the rate of
adsorbate reflection from the surface during adsorption [4].
sabs ¼ ra =z ð23:3Þ
The absolute sticking coefficient is equal to the net sticking coefficient only
when the rate of desorption is zero, so in order to achieve this condition, con-
ventional sticking coefficient measurements are made at temperatures below the
desorption temperature.
Substituting (23.2) into (23.3) gives (23.4).
Normally, a plot of snet versus 1/z at constant temperature will give an intercept of
sabs and slope rd. Figure 23.2 shows a three-dimensional plot with the linear
extrapolation of the family of curves (black points) to the limit of 1/z = 0, yielding
the white points which form a trace of the absolute sticking probability in the limit
of very high impingement rate, z. Here the symbol ML = monolayer.
The data shown in the Figure are taken at 400 K, whereas the desorption tem-
perature for CO in this case is in the range 350–550 K [5]. The method works well
when rd < z, or correspondingly, when 1/rd > 1/z.
An alternate method, particularly useful for irreversible dissociative chemi-
sorption, has been described by Xia et al. [6].
References
1. Z.H. Lu, K. Griffiths, P.R. Norton, T.K. Sham, Phys. Rev. Lett. 68, 1343 (1992)
2. D.A. King, M.G. Wells, Surf. Sci. 29, 454 (1972)
3. C.C. Cheng, R.M. Wallace, P.A. Taylor, W.J. Choyke, J.T. Yates Jr, J. Appl. Phys. 67, 3693
(1990)
4. A.M. Glines, M.A. DeAngelis, A.B. Anton, J. Vac. Sci. Technol. A10, 3179 (1992)
5. X.-C. Guo, D.A. King, Surf. Sci. 302, L251 (1994)
6. L.-Q. Xia, M.E. Jones, N. Maity, J.R. Engstrom, J. Vac. Sci. Technol. A13, 2651 (1995)
Chapter 24
Mass Spectrometry
One of the most widely used and versatile techniques for the study of adsorbed
species on single crystal surfaces is temperature-programmed desorption (TPD),
often called thermal desorption spectroscopy (TDS). The method extends back to
the early work of Taylor and Langmuir, 60 years ago, who studied the desorption of
Cs from polycrystalline tungsten filaments [1]. The method was extended to other
adsorbates on polycrystalline surfaces by Becker and Hartman [2], Ehrlich [3],
and by Redhead [4]. A review of the early work can be found in [5].
Temperature-programmed desorption is now used extensively for studies on single
crystal surfaces of metals, semiconductors, and insulators. A comprehensive review
of ten methods for analyzing TPD data, and an evaluation of each was recently
written by de Jong and Niemantsverdriet [6].
The collection of TPD data has progressed from the early use of ionization
gauges to the almost universal employment of quadrupole mass spectrometers
(QMS) for the detection of desorbing gases. Using the mass spectrometer in the
multiplex mode, one can simultaneously measure the evolution of various gas
products as the temperature of the crystal is raised. This can also be extended to the
study of isotopic mixing between labeled adsorbates, as was first demonstrated for
adsorbed CO desorbing from polycrystalline W [7]. Large differences in the
desorption kinetics from different single crystal faces was first demonstrated for H2
desorbing from W(100), W(111), and W(110) by the combined works of various
authors [8–12].
The early work with single crystals employed unshielded mass spectrometers
that sampled the random flux of desorbing gases originating from the front of the
crystal, from the electrical support leads to the crystal, and from the back of the
crystal disk. These effects could be eliminated to various degrees by using a dif-
ferentially pumped and shielded mass spectrometer that preferentially collected
desorbing gas only from the front of the crystal. The enhanced sensitivity to
desorbing gas as a result of the shield is related to the fact that the shield retains gas
molecules for an average time period of τ = V/S, where V = volume and
S = pumping speed, both relating to the shield region itself. Values of τ = 50
−100 ms are common, permitting each trapped molecule to pass through the ion
source many times before being pumped away. The apertured shield is selective for
desorption from the front of the crystal, but it is not measuring significant
desorption for molecules making a single pass, compared to the contribution from
trapped molecules within the shield which make multiple passes through the ion-
ization source. The value of τ must be kept below the characteristic time for the
desorption process measured, which is usually of the order of seconds for tem-
perature scan rates of a few K/s.
Figure 24.1a shows a glass shield that was just slipped over the ionization source
of a mass spectrometer. Pumping occurs from the annular space at the rear of the
shield [13]. Variations on this design (Fig. 24.1b) involve a metal shield with
pumping holes present on the side of the shield behind the ionization source. For
N2(g) at 300 K, a 1-cm2 aperture will provide a pumping speed, S = 11 L/s, where
the total pumping aperture area considered should also include the area of the
entrance aperture. Figure 24.1c shows a quadrupole mass spectrometer mounted in
an apertured shield and differentially pumped by a dedicated pump [14], which may
be either an ion pump, a titanium sublimation pump (TSP) [15], or a turbopump.
The reader is referred to p. 72 for a discussion of the relative merits of an ion pump
and a TSP in this application.
The ionization source of the quadrupole mass spectrometer contains a thermionic
emitter producing electron emission at mA currents. A fraction of these electrons
escape from the axial aperture and may be collected by the crystal, causing electron
beam damage of the adsorbate. Stray currents of the order of 10 μA are often
observed. These effects may be eliminated by either biasing the crystal at a negative
potential to retard the collection of electrons, or by using a separately negatively
biased aperture, as shown in Fig. 24.1d [16]. The effect of electron beam damage
from stray electrons in the absence of these precautions can be profound in alter-
ating the chemistry on the crystal surface.
Figure 24.1d also shows a helpful device useful for reproducibly locating the
crystal at a fixed distance from the aperture [16]. A Ta contact with a gently
rounded end is connected to the electrically isolated aperture, and the connection to
the bias potential is opened. An ohmmeter, working on the MΩ scale, is connected
between the Ta contact and ground. The crystal is slowly advanced to the Ta
contact until MQ resistance is measured as a result of very light contact. This allows
one to calibrate the horizontal micrometer drive for the crystal at a fixed point in the
apparatus. The micrometer drive is then retracted by a measured amount, and
knowing the position of the Ta contact allows one to calculate the position of the
crystal relative to the aperture.
The fractional interception of desorbing gas from the crystal will depend on the
distance between the crystal and the aperture and also on the angular distribution of
the desorbing gas, expressed as cosnθ. For a crystal of 3.5 mm radius, and an
24.1 Shielded Quadrupole Mass Spectrometers—Temperature-Programmed … 261
aperture of 0.8 mm radius, the relative collected signal for various values of n is
shown in Fig. 24.1e. It is noted that only small losses are observed in the first 1-mm
crystal-aperture distance, and also that an error in distance will make a smaller
effect in the first 1-mm range compared to errors at larger ranges [17].
A critical review of the advantages and disadvantages of using a shielded mass
spectrometer for temperature programmed desorption studies has been given by
Schlichting and Menzel [18]. A shielded mass spectrometer such as that shown in
Fig. 24.1a will have a constant pumping speed determined by the geometry of the
aperture and the opening at the rear, whereas the pumping speed of a vacuum
system used with an unshielded mass spectrometer will be variable. Also effects due
262 24 Mass Spectrometry
The quadrupole mass spectrometer is widely used in surface science for residual gas
analysis as well as for the study of desorbing species from surfaces. Thermal
desorption, and electron- and photon-induced desorption (including laser-induced
desorption) have been widely studied. For most quadrupole mass spectrometers, a
channeltron multiplier is employed for detection of the mass-separated ions.
A method is described here for improvement of the signal-to-noise ratio of a
channeltron detector; and in addition, the behavior of a channeltron detector is also
described [22].
An electron multiplier has a low inherent noise, leading to a dark current. In the
case of a channeltron multiplier mounted behind quadrupole rods with line of sight
to the ionization source, there are additional background effects that increase as the
electron emission current increases and also as the pressure in the mass spec-
trometer is increased. While these effects are not fully understood [22], they are
thought to be due to soft X-rays and UV photons produced in the ion source, and
also to excited neutral species, both of which pass down through the quadrupole
rods without deflection. The use of an off-axis ion multiplier enables one to
24.2 Enhanced Signal-to-Noise in Quadrupole Mass Spectrometers 263
eliminate these effects to a large degree, and the modification described here
accomplishes this while not causing unacceptable reduction of the ion collection
efficiency.
Figure 24.2a shows the off-axis channeltron detector located beneath the
quadrupole rods. Between the exit region of the quadrupole rods and the chan-
neltron entrance are two ion deflection plates, designed to steer ions into the
channeltron. The channeltron is located 10 mm off axis. A multiplier shield plate,
8 mm in diameter, is mounted 2 mm in front of the channeltron entrance cone.
A deflector plate, with a 45° angle to the shield plate, is mounted on the axis of the
quadrupole, and a second deflector plate is mounted in a direction perpendicular to
the shield plate, and 15 mm from the quadrupole axis. The distance between these
deflector plates and the exit aperture of the quadrupole lens is 2 mm. The 45°
deflector plate may be biased by an isolated lead, which is separately shielded. The
entire structure is housed in a shield can to prevent both photons and charged
particles from entering the channeltron.
Figure 24.2b shows the performance of the channeltron for various applied
multiplier voltages, and also for various positive deflector plate voltages. The large
enhancement of output current for a slight increase of multiplier voltage is evident
from the three curves; in addition, a large effect of the positive deflector plate is also
seen, with maximum transmission of ions being observed at +10 V plate bias.
Figure 24.2c shows the behavior of the dark current versus electron emission
current in the ionization source of the mass spectrometer. A slope of 1.3 in this log–
log plot indicates that slight nonlinearities exist as the filament temperature is raised
in the ionization source to produce a larger electron emission current. A 100-fold
improvement in the dark current is observed for the off-axis detector.
Figure 24.2d shows the behavior of the dark current as a function of Ar pressure
in the mass spectrometer, with data taken at 1 mA emission current. The dark
current rises as the 1.1 power of the pressure in both of these measurements. The off
axis detector reduced the dark current here by a factor of 140.
The performance of the mass spectrometer with the two configurations of the
chan-neltron detector can be summarized in (24.1) and (24.2) below:
Fig. 24.2 Enhanced signal-to-noise—QMS. a Off axis channeltron detector, b detector perfor-
mance, c Dark current behavior, d Dark current behavior
24.3 Elimination of Spurious Electron Emission from a Quadrupole … 265
Electrons emitted from the ion source of a mass spectrometer can be a source of error
in measurements when the electrons strike and damage a sample in front of the mass
spectrometer. For an enclosed mass spectrometer, with an aperture, placing a neg-
ative bias on the aperture will repel electrons and prevent this effect (see p. 259). The
effect of spurious electron emission may also be eliminated by biasing the sample
negative to repel electrons. A third method involves the use of a floating grid
surrounding the mass spectrometer ion source, which is automatically brought to the
potential of the thermionic emitter (usually −70 V in a mass spectrometer). This
occurs as a result of the charging of the isolated grid by the emitted electrons [25].
Figures 24.3a–c show the steps in rebuilding a mass spectrometer ionizer to
provide the floating grid. In Fig. 24.3a, an inner grid cap is placed over the existing
inner grid cage. In Fig. 24.3b, the existing electron repeller grid cage is placed on
the ion source. Inside the repeller grid is the thermionic emitter assembly, not
shown in the drawing. Finally, in Fig. 24.3c, the floating grid, consisting of a
cylindrical grid cage capped with a planar grid, is installed. The floating grid is
electrically isolated by means of a ceramic ring that can be bolted to the mass
spectrometer housing by means of three standoff bolts as shown.
The floating grid at negative potential is capable of extracting positive ions from
the ionizing region inside the inner grid. This is the reason for the addition of the
inner grid cap on the ionizing region, which acts as a shield to prevent positive ion
extraction.
The floating grid collects electron current so that the electron emission current to
the inner grid cage is somewhat reduced. Thus, in the UTI mass spectrometer, the
filament protect circuit must be deactivated, since it will trip when it senses too little
current at the inner grid. With the floating grid design, spurious emitted electron
currents are reduced from about 10 μA to less than 1 nA [25].
A large O+ signal and a large F+ signal may be observed in the mass spec-
trometer operating with the floating grid (the latter especially if fluorinated mole-
cules are introduced to the chamber deliberately). These signals probably originate
from electron stimulated desorption effects from the ceramic ring and from the grid
material itself [25]. Fluorine, a well-known surface contaminant in UHV systems,
with a very high cross section for electron-stimulated desorption as a positive ion, is
often seen when contaminated surfaces are electron bombarded [26].
Fig. 24.3 Elimination of spurious electron emission from a Quadrupole Mass Spectrometer.
a Placing inner grid cap, b Placing repeller grid + emitter (not shown), c Placing floating grid
Ta is much more ductile than W at room temperature and can be wound and
welded easily compared to W. Ta wire (0.006-in. diameter) is wound around a rod
(typically 0.040-in. diameter) to form a coil before being installed into the mass
spectrometer as shown in Fig. 24.4.
A Ta emitter has a built-in property that is sometimes useful. Ta has a suffi-
ciently high vapor pressure that at filament operating temperature there is a
detectable signal from 181Ta, the most abundant isotope. This signal for the 181Ta+2
24.4 Ta Mass Spectrometer Filament 267
ion, observed at M/q = 90.5, may be used as a mass spectrometer calibration point,
and interference from other ions in this region will be absent. The lifetime of the Ta
filament will be years despite its slow evaporation during use.
Fig. 24.5 Modulated QMS ion source for enhanced sensitivity. a Modulation for circuit,
b Enhanced TPD spectrum
wave output at 50 % of the timing chip frequency. The 100 kΩ variable resistor is
used to vary the frequency. When the disable switch is manually closed the ion
source is not modulated and the QMS operates normally.
The advantage of measuring the modulated QMS signal is illustrated in
Fig. 24.5b. Here a fraction of a ML of iodine from Pt(111) is being monitored by
thermal desorption at a temperature sweep rate of 0.1 K/s. In (a), the unmodulated
signal is shown and is very noisy. A smoothing routine is applied to this signal as
shown in (b), and two desorption peaks are barely resolved. In (c), the modulated
signal is shown with good resolution of two desorption processes. By enhancing the
speed of the temperature ramp, further improvements in signal intensity will be
found, but the resolution of overlapping peaks will begin to degrade [28]. Of
course, the method may also be used to enhance the mass spectrometer sensitivity
in other types of measurements.
24.6 Line-of-Sight Mass Spectrometry for Thermal Desorption 269
A mass spectrometer may be used for line-of-sight detection if all arriving mole-
cules from the background are removed. By placing the mass spectrometer behind
apertures which connect to baffle surfaces at 77 K, many gases may be sampled in
this manner. Any molecule with an approximate desorption enthalpy of
≥ 18 kJ mol−1 will be pumped at 77 K, and this includes molecules such as water,
ethene and almost all higher molecular weight organic and inorganic compounds.
Molecules such as H2, O2, N2, CO, Ar and CO2 will not be pumped at 77 K.
Figure 24.6 shows a schematic of the cryo-shield [29–31]. The mass spec-
trometer should have its shield removed to prevent internal reflection from warm
walls. Liquid N2 fills the tank from a reentrant tube which leads to the outside. Two
cylindrical shields, heat stationed on the cryo-reservoir are cooled by conduction.
An entrance aperture, N, admits a beam of molecules originating from the crystal
surface during a desorption experiment. All molecules originating from outside the
focal area will hit a cold surface and be retained. For a gas being admitted from the
focal area at a pressure of 10−8 mbar, the pressure in the mass spectrometer ioni-
zation volume will be about 10−12 mbar at a sample-to-ionization volume distance
of *30 cm. A mass spectrometer with a multiplier operating at 106 amplification
factor typically has a sensitivity of better than 10−14 mbar. The exit aperture, M,
does not have to be maintained at 77 K, since molecules reflected from its surface
will reach the cooled inner cylindrical shield surrounding the mass spectrometer and
be captured. To eliminate the influence of background gas produced in the mass
spectrometer, chopping of the desorbing gas may be carried out.
This method of beam measurement eliminates desorption from the crystal sup-
port assembly and may be used to measure accurate thermal desorption spectra
[31–34]. It may also be used to measure the sticking probability [29–32, 34];
surface reactions can be measured either isothermally or during heating [30, 31, 34,
35]; and radical desorption may be directly measured [30].
The upper pressure limit for the apparatus is about 10−4 mbar, where mean free path
considerations will disrupt the trajectories of molecules coming from the crystal.
Many factors make it difficult to control the partial pressures of two or more gases
in a pumped vacuum system and to know the absolute ratios of partial pressures,
much less the absolute partial pressures themselves. The factors that prevent such
calibrations have to do with fundamental issues such as the pumping speed, for
individual gases, the relative differences in sticking coefficients on the chamber
walls for different molecules, and the absence of absolute knowledge of gas fluxes
from calibrated dosers, etc. One way to overcome some of these difficulties is to
utilize the thermal decomposition of molecules, containing known stoichiometric
ratios of the elements, and to measure the relative signals from the gaseous
decomposition products under dynamic pumping conditions [36].
Figure 24.7 shows a general arrangement that maybe utilized for these mea-
surements. A molecular beam consisting of the pure compound is directed at a
heated surface that is known to decompose the compound, producing only gaseous
products. A mass spectrometer samples the desorbing gas products and yields
signals of the mass spectrometer cracking products and parent ions of the product
molecules from the decomposition.
The absolute mass spectrometer sensitivity, βx, is defined as
may contain contributions from the cracking of more than one type of molecule. It
is also important to remember that this method assumes that the angular distribution
of all of the product molecules from the molecular decomposition is the same, and
probably of cosine form if surface equi-librium is achieved before release of the
product molecules.
The method has been used for two molecules to determine the relative sensitivies
for four gases as diagrammed below:
H2 CO ! H2 þ CO ð24:4Þ
For the vacuum system used in [36], in which a UTI-100C mass spectrometer
was employed, the following relative sensitivities were determined:
References
Figure 25.1b shows typical desorption spectra for H2 from Ni(111) that has been
saturated with H2. The angular distributions of both H2 and CO from Ni(111) are
shown in Fig. 25.1c. It can be seen that CO desorbs in the two azimuths measured
with a cosΘ distribution, whereas H2 has a sharper distribution, described as cosnΘ,
where n = 4.6 or 4.8.
The thermal desorption of species from surfaces is widely used throughout the field
of surface science for studies of adsorbates on metal foils and single crystal surfaces
(see p. 273). Often an unshielded mass spectrometer is used for detection of the
desorbing particles and it will generally measure the density of gas produced in a
random flux within the vacuum chamber. More modern methods involve the use of
apertured mass spectrometers that are differentially pumped within the surrounding
shield (see p. 72). The aperture is often close-coupled to the crystal surface along
the surface normal in order to collect the majority of the desorbing gas evolved
from the region of the crystal viewed by the aperture. Under these conditions of
efficient collection, considerations of the angular distribution of the desorbing gas
are of second-order importance.
However, when the apertured mass spectrometer is located far from the crystal
surface, considerations of the angular distribution of the desorbing gases become
very important, especially if measurements are made of the relative signal from
different gases desorbing simultaneously (and sampled by multiplex methods).
Thus, gases are known to desorb with angular distributions approximated by cosnΦ
where Φ is the angle from the normal. Here n can have a value of unity, or values
running to about 20 in extreme cases. If the value of n differs significantly for two
species desorbing simultaneously, then errors in estimating the relative signal
intensities will occur, and these errors can approach 400 % for collection along the
normal direction when the mass spectrometer is distant from the crystal. Pauls and
Campbell [2] have considered this problem and have found that there is a magic
angle for the mass spectrometer location that will minimize the error in comparing
desorption signals with n varying over the range 1–9. Figure 25.2a shows the
geometry of the mass spectrometer collection optics, where r is the crystal radius,
and d is the distance to the aperture. The polar angle Φ is shown. Figure 25.2b
shows the magic angle, which minimizes the error in the relative magnitude of the
desorption signals having varying angular distribution, where an integration over
the emitting surface of the crystal has been carried out. The function shown varies
as the value of the ratio d/r varies. For typical values of d/r from 2 to 5, the magic
angle varies from 42° to 34°, and at large values of d/r, the optimal angle of
Φm = 34° is found. The variations in the signal for the range of n considered is only
about 26 % at Φm = 42°.
It should be mentioned that the mass spectrometer detection behavior may vary
between two limits. For very fast pumping within the shield, only a single pass of
276 25 Temperature-Programmed Desorption/Reaction
the desorbing species occurs through the ionizing optics, and considerations of the
molecular velocities of the gas molecules would be necessary in measuring relative
fluxes [3]. For very slow pumping in the shield, which is the normal method of
operating the shielded mass spectrometer at high sensitivity (see p. 259), the
desorbing particles will probably acquire the temperature of the ionizing source and
then these considerations of molecular velocity will be unneeded. The detector
signal is then truly proportional to the flux entering the aperture [2].
These considerations also apply to the measurement of any particle desorption
technique, for example sputtering of neutral species from surfaces [4].
25.3 High-Temperature Thermal Desorption Spectroscopy 277
control system without position feedback to the computer. Figure 25.4b shows the
main function blocks for control. Through the DAQ board, the computer controls
the power supply that feeds the heating current to the crystal. It also reads the
amplified output of the thermocouple attached to the crystal. The stepper motors are
driven by two motor control boards (Motion Group, model MS2.0). The motors
provide an angular resolution of 1.8º translating into an angular resolution of 0.045º
inside the vacuum system. The resolution of the micrometer giving linear motion to
the crystal is in the µm range.
The timing of separate steps is shown in Fig. 25.4c. Dosing of the sample to
different doses of gas is accomplished in the constant-gas flow by adjusting the time
of exposure. QMS calibration is then performed with the crystal in a null position.
280 25 Temperature-Programmed Desorption/Reaction
TPD is next carried out with the crystal at position 2 in front of the shielded QMS.
After the upward temperature ramp, the crystal is cooled before continuing the
cycle. Figure 25.4d shows a stack of typical TPD measurements made, where the
exposure is automatically increased by a factor of 21/2 in each step. Figure 25.4e
shows that the coverage increases linearly with exposure, to great accuracy. This
means that the adsorption process is governed by the production of a mobile
precursor species.
The automatic method is highly accurate since it does not involve human
intervention.
Fig. 25.5 Scanning kinetic spectroscopy (SKS). a SKS apparatus. b Typical SKS data.
c Elementary reaction steps as inferred from B
a volatile product is not liberated from this process, but O–H bond scission must
begin below 290 K since the β-recombination process begins at 290 K. The process
γ2 and γ3 involve the consumption of the reactant molecule since the reflected flux
of reactant falls below the reference line in this temperature region. These con-
sumption processes are closely correlated with the D2 and CO desorption processes,
shown near the bottom of Fig. 25.5b, which open Ni sites for adsorption and
reaction processes.
The SKS method is a fast survey method, giving one a partial view of the
characteristic temperatures for various adsorption, desorption, and surface disso-
ciation processes. It has been effectively used with intentional poisoning of the
surface by added species, and with intentional addition of known reactants to the
surface [9]. It has been successfully used to observe deuterium kinetic isotope
effects that give insight into the rate-controlling steps in surface reactions involving
the hydrogen moiety [9, 11]. The original concept of temperature scanning while
making reaction kinetic measurements on single crystal surfaces was first presented
in 1981 by Goschnick and Grunze [12]. The SKS method also resembles the
familiar temperature-programmed reaction spectroscopy (TPRS) [13, 14] (see
p. 282), but SKS has added advantages of working like a flow reactor where fresh
reactant is continuously supplied during the measurements.
The ability to rapidly scan many mass peaks and to record the time behavior of
selected mass peaks has been a familiar subject for a number of years. Recently, a
fast multimass measurement method has been introduced that permits up to 15 mass
spectra of 50 amu width to be recorded per second, using a quadrupole mass
spectrometer [15]. Another variation on this method scans up to 100 complete mass
spectra per thermal desorption experiment. Enhanced signal-to-noise is achieved by
integrating each of the mass peaks rather than by measurement of the peak height
only [16].
There are many other mass-spectrometer based methods used for the study of
surface reactions in which transient surface species may be detected. Among these
is laser induced desorption, in which a Fourier transform mass spectrometer is often
used [17]. Here, the laser may be used to desorb surface species without extensive
thermally induced fragmentation.
An example of the use of TPRS is given here [18]. The interaction of ethylene
glycol, (CH16 18
2 OH)2, with fractional monolayers of chemisorbed O on Ag(110) has
been studied. First, up to 200 mass peaks are scanned to survey the products. Then,
using methods which scan 10 mass peaks in a multiplex mode in 20 ms, the
temperature dependence of evolution of a variety of products was measured as
shown in Fig. 25.6a. The two cross-hatched desorption processes occur in surface
reaction limited processes and contain no 18O, and are the end result of the oxi-
dative dehydrogenation of the ethylene glycol. The five remaining product peaks
are indicative of reactions involving the chemisorbed 18O.
The desorption of H182 O at 255 K occurs as a result of the direct reaction of O–H
bonds of the two O–H groups of the ethylene glycol with adsorbed 18O, as illus-
trated in Fig. 25.6b. The desorption together of H18 16
2 O and CH2 O with a peak
temperature of 300 K is indicative of a surface reaction controlled by C–H bond
scission in the CH2 groups of the 1,2 ethanedioxy species as shown in Fig. 25.6c.
The simultaneous evolution of H2 and C16O18O at 415 K is indicative of the
decomposition of adsorbed formate, HC18OO, on the surface. Similar reactions
were observed in separate experiments with adsorbed formate, and are illustrated in
Fig. 25.6d.
There are other possible mechanisms that might participate in the surface
reactions, and these were eliminated by using D substitution for H in the CH2
groups of the ethylene diol and looking to see whether a primary kinetic isotope
effect was seen, in which the scission of the C–D bond occurs at a significantly
higher temperature than the C–H bond in the unlabeled molecule. This isotope
effect was observed, pointing to C–H bond scission as being a key elementary step
in the surface chemistry.
Two reviews on TPRS have been written [14, 19].
References
1. H.R Steinruck, A. Winkler, K.D. Rendulic, Surf. Sci. 152/153, 323 (1985)
2. S.W. Pauls, C.T. Campbell, Surf. Sci. 226, 250 (1990)
3. G. Comsa, R. David, Surf. Sci. Rep. 5, 145 (1985)
4. A. Wucher, F. Novak, W. Reuter, J. Vac. Sci. Technol. A6, 2265 (1988)
5. K. Pelhos, T.E. Madey, Department of Physics and Astronomy, Laboratory for Surface
Modification, Rutgers, The State University, Piscataway, NJ 08855 (private communication)
6. Model AD210AN. Analog Devices, One Technology Way, Norwood, MA 02062-9106
7. C.M. Furnaces, Inc., 103 Dewey Street, Bloomfield, NJ 07003
8. P. Kondratyuk, J.T. Yates Jr, J. Vac. Sci. Technol. A23, 215 (2005)
9. S.M. Gates, J.N. Russell Jr, J.T. Yates Jr, Surf. Sci. 159, 233 (1985)
10. I. Chorkendorff, J.N. Russell Jr, J.T. Yates Jr, J. Chem. Phys. 86, 4692 (1987)
11. P.L. Hagans, I. Chorkendorff, J.T. Yates Jr, J. Phys. Chem. 92, 471 (1988)
12. J. Goschnick, M. Grunze, J. Vac. Sci. Technol. 18, 561 (1981)
13. R.J. Madix, J. Benziger, Ann. Rev. Phys. Chem. 29, 285 (1978)
14. R.J. Madix, in Chemistry and Physics of Solid Surfaces, vol. II, ed by R. Vanselow (CRC
Press, Boca Raton, FL, 1979)
References 285
15. G. Zagatta, H. Muller, N. Bowering, U. Heinzmann, Rev. Sci. Instrum. 65, 359 (1994)
16. L.H. Dubois, Rev. Sci. Instrum. 60, 410 (1989)
17. D.P. Land, C.L. Pettiette-Hall, D. Sander, R.T. Mclver Jr, J.C. Hemminger, Rev. Sci. Instrum.
61, 1674 (1990)
18. A.J. Capote, R.J. Madix, Surf. Sci. 214, 276 (1989)
19. R.J. Madix, J. Benziger, Annu. Rev. Phys. Chem. 29, 285 (1978)
Chapter 26
Gas Chromatography—Enhanced
Sensitivity
The sampling of gases for gas chromatography analysis suffers from the fact that
gas expansion in the sampling lines invariably causes loss of analytical sensitivity.
This behavior may be reversed by using a high-pressure pulse of the carrier gas to
compress the low-pressure gas sample into the sample loop, where it is then swept
into the gas chromatograph. A factor of 20 or more sensitivity enhancement may be
achieved. The application shown here involves sampling a high-pressure reactor
being used to study the kinetics of a catalytic reaction over a single crystal.1
As shown in Fig. 26.1a, the reactor volume to be sampled is connected by valve
V1 to a length of capillary tubing containing an aliquot volume. The tube is then
connected via a tee to a gas chromatography sampling valve. Gas from the reactor is
expanded into the aliquot volume and the thin tubing (V5 open) and its pressure is
read with a low-volume pressure gauge. The valve V2 to the pump is closed and
valve V3 is opened to the high pressure of the carrier gas regulator (about 800 Torr).
Valve V4 is then opened, and the carrier gas acts as a piston, compressing the
sample gas into the sample loop located on the sampling valve. The contents of the
sample loop are then injected into the gas chromatograph, as shown in Fig. 26.1b.
In the case shown here, the sampling volume is the aliquot volume plus the volume
of the thin tubing, valve, and sampling loop. By keeping valve V5 closed during
sampling, the sampling volume is just the aliquot volume. It was shown that two
consecutive analyses using this method differed by less than 4 %.
1
Personally contributed by Professor C.T. Campbell, Department of Chemistry, University of
Washington, Seattle, WA 98195. See also [1].
Reference
The use of a retarding diode method for measuring work function changes upon
adsorption is one of the oldest and simplest methods for measurement in surface
science. The literature related to this method is mostly from the 1960s or 1970s
[1–6]. Other methods for measuring work function changes are described elsewhere
in Sects. 27.2–27.6.
In Fig. 27.1a is shown a thermionic emitter housed in an ultrahigh vacuum
system and facing a sample whose work function change, Δϕ, is to be measured as a
result of adsorption or desorption. The emission current collected by the sample as a
function of the applied voltage, U, is given by
Fig. 27.1 Retarding diode for measuring work-function changes. a Schematic of circuit,
b Collected current versus retarding potential
temperature will also be constant. The temperature of the cathode emitter will be
sensitive to the heat loss through the leads to the emitter, and thermostating these at
constant temperature with a water cooling jacket will remove this variable.
There are several factors that can affect the measurements using this simple
device. One possible problem is the emission of impurity gases from the hot emitter
27.1 Work Function Changes Using Retarding Diode Method 291
filament, which will influence the work function of the sample. Second is the effect
of the hot filament on the residual gases in the vacuum system (for example, H2
dissociation) making them reactive with the sample. This effect can be eliminated
by turning off the emitter during gas admission. The method will not work well at
pressures above about 10−4 Torr, since otherwise the cathode might be cooled by
heat conduction through the gas. Finally, the effect of electron bombardment and
radiation on the surface layer on the sample must be understood when using this
method.
The retarding diode method has been employed in a conventional LEED
apparatus with continuous derivative measurement of the work function change
during adsorption. A sensitivity of about 5 meV was obtained, using the lock-in
amplifier method for making continuous measurements of the changing work
function [7]. Reference [7] also summarizes background papers in this area.
One of the simplest methods to measure the work function change during
adsorption on a metal single crystal is to employ a retarding potential method, using
an ordinary electron gun and an interposed grid or aperture. Figure 27.2a shows a
simple device first used for the study of hydrogen, nitrogen and xenon adsorption
on tungsten surfaces [8–10]. An electron gun of ordinary design supplies current to
the crystal through an aperture made of coiled tungsten in this case. The coiled
tungsten is also used as an electron emitter for thermally cleaning the tungsten
crystal at 2500 K. The crystal is welded to a thin wire support which may be cooled
with liquid nitrogen. By using 25 eV electrons, and retarding between the aperture
and the crystal surface, I–V plots are obtained, and from the shift in electrical
potential of the knee region, the work function change may be deduced as gases are
adsorbed on the crystal. The reproducibility of duplicate measurements is
about ±20 meV.
Figure 27.2b shows the linear work function change versus coverage for H2
adsorption on W(100). The linearity versus measured coverage (using thermal
desorption spectroscopy) suggests that depolarization effects between neighbor H
atoms are small in this case [8]. Figure 27.2c shows measurements of the adsorption
of N2 on W(110) at 123 K. Here, the work function change is nonlinear with
coverage, and this indicates that depolarization effects exist between the adsorbed
molecular N2 molecules [9]. The plot shown in Fig. 27.2c is a fit to the immobile
Topping model, and the N2 polarizability may be obtained from this fit [9, 11].
The use of an electron gun in a display LEED apparatus for retarding potential
measurements of work function changes has been described [12].
292 27 Work Function, Tunneling Spectroscopy and Ellipsometry
The use of the Shelton diode [13, 14] to measure the work-function difference
between two surfaces, an emitter and a collector, has been employed to determine
the difference in work function between two crystal planes of tungsten. In addition,
the work-function change during adsorption of a gas on the collector, and the
energy-dependent reflection coefficient for electrons incident normally on the col-
lector surface, can be measured with this device [15].
Figure 27.3a shows the schematic arrangement used [15]. Two single crystals of
different crystallographic orientation are mounted by wires and a tungsten spring in
support rings. These crystals face a movable septum containing a small aperture.
Either crystal may be heated by electron bombardment to a temperature where
thermionic emission occurs, and a feedback stabilized controller is used to maintain
emitter and crystal temperature constant, using the current collected from the crystal
emitter by the septum for regulation of the temperature of the spiral tungsten emitter
located behind the emitter crystal. The entire assembly is placed in an axial mag-
netic field of 3600 G, causing the emitted electrons to travel in tight spiral paths
from the emitter to the collector crystal. The movable septum permits sampling of
different regions of the emitter and collector. The collector crystal current is plotted
against the retarding potential applied between emitter and collector, and a typical
retarding potential plot, as shown in Fig. 27.3b, may be measured. The collected
current follows the equation
where I0 is the emitted current from the emitter, VR is the retarding potential
applied, and Δϕ is the difference in work function between the collector and the
emitter, and e is the electronic charge. Thus as the work function of the collector
changes, the retarding potential plot (Fig. 27.3b) will shift. The shift of the “knee”
in the plot is a measure of the work-function change of the collector, and may be
used as a measure of the adsorption of a gas on the collector, the emitter remaining
clean at its high operating temperature. The device shown in Fig. 27.3a measures
collected currents of the order of 0.01 μA and to avoid problems with leakage
currents, the terminals to each crystal are shielded from evaporating metal films by
the hat-shaped shields shown on each feedthrough. The device is symmetric, so that
it may be operated with either crystal acting as the emitter, but such a symmetrical
arrangement is of course unnecessary if one is interested in only one crystal. In
contrast to similar devices [13, 14] that used ohmic heating of the emitter, this
device does not have a potential drop across the emitter region sampled by
the aperture and should therefore perform better by having less uncertainty of the
electrical potential distribution in the sampled emitter region.
By varying the electron energy and measuring the collected current, curves like
those shown in Fig. 27.3c may be generated. These curves are related to the
294 27 Work Function, Tunneling Spectroscopy and Ellipsometry
Fig. 27.3 Work function measurements—shelton thermionic diode. a Shelton diode, b Retarding
potential plot, c Reflection coefficient plot
effective electron reflection coefficient in the plateau region and have to do with
band structure and other possible effects at the collector.
The accuracy of work-function measurements made with this device is of the
order of ±20 meV over many months of operation with tungsten crystals.
27.4 Improved Piezoelectric Drive Kelvin Probe 295
The Kelvin-Zisman vibrating capacitor method for measuring changes in the work
function of a surface is widely used in surface science studies. The principles of this
method are well described in the literature [16, 17] and will not be discussed in
detail here. Basically, a reference electrode is caused to oscillate at audio frequency
and the electrical potential of the reference electrode is adjusted to yield a null
signal in the capacitance circuit. As the work function of the sample changes due to
adsorption or to changes in temperature, the potential applied to the reference
electrode is adjusted to maintain a null signal in the AC circuit. The change in the
applied potential is equal to the change in the work function or surface potential of
the sample. It is of course important to know that the reference electrode is not
changing its work function during the measurements, and inert electrodes such as
Au or SnO2 films are often used. The vibrational motion of the reference electrode
is commonly produced by mechanical excitation of the resonant frequency of the
device, by an electromagnetic driving mechanism, or by the use of a piezocrystal
oscillator.
The design shown in Fig. 27.4a involves the use of a piezoceramic foil actuator
[18]. This avoids the interaction of stray magnetic fields from an electromagnet
driver with the sample, which could cause problems if the sample is magnetic. In
addition, the measured signal is taken from the reference electrode, not the sample,
avoiding complications owing to the connection of the sample to both the heating
circuit and the thermocouple circuit. The device described in Fig. 27.4a is a
modification of the piezoelectric Kelvin probe built by Besocke and Berger
[19, 20]. The device was used to study magnetic materials and yielded an accuracy
of ±1 meV [21]. This compares to a reported sensitivity of better than 0.1 meV in
the original device built by Besocke and Berger [19]. The reader is referred to the
original paper [18] for dimensional details of the Kelvin probe shown in Fig. 27.4a.
The reference electrode is an L-shaped piece of gold, connected by a tungsten wire
to a quartz rod and thence to a shield plate. This assembly is actuated by a piez-
oceramic foil [22], containing vapor deposited Ag contacts on both surfaces, and
glued using UHV leak sealer to a tungsten reed, 20 mm × 4 mm × 0.1 mm thick.
The signal from the Au reference electrode is measured through a very thin wire
connected to a support ceramic before exiting the apparatus. The stable resonance
frequency of the reed is about 130 Hz, the amplitude of vibration is about 0.1 mm,
and the AC driving voltage is about 1.0 Vrms. The spacing between the reference
electrode and the sample is about 0.5 mm.
Before operating, the piezoceramic foil is polarized by a DC voltage pulse above
300 V. The polarization remains as long as the piezoceramic is not heated
significantly.
Figure 27.4b shows a block diagram of the self-compensating feedback circuit
which allows a continuous measurement of the work function change. It is esti-
mated that the drift is about 1 meV/h and the noise is about ±0.5 meV. The response
time constant of the system is less than 10 s. The variable DC power supply
296 27 Work Function, Tunneling Spectroscopy and Ellipsometry
Fig. 27.4 Improved kelvin probe—piezodrive. a Piezodriven kelvin probe, b Circuit for Δφ
measurements, c Work function change on cooling Ni (poly.)
supplies a variable compensating voltage, Vb. The authors note that electron
emission from thermionic sources can disturb the electrical isolation of the sample
and the reference electrode [18].
27.4 Improved Piezoelectric Drive Kelvin Probe 297
Figure 27.4c shows a record of the increase in the work function of a poly-
crystalline Ni sample as it cools through the Curie temperature, TC.
Excellent reviews and analysis of the effects of stray capacitance [23, 24] and
noise [25] on the performance of the Kelvin probe have been published. Kelvin
probes, driven by an electromagnetic voice coil and possessing superior noise
levels, have been described [26, 27].
The Kelvin probe has been used in dynamic measurements of the desorption of
one desorbing surface species from a single crystal surface [28]. The method was
shown to give superior resolution of overlapping desorption processes compared to
temperature programmed desorption into a large chamber with gas monitoring by a
quadrupole mass spectrometer [28]. Also, the dynamic work function change
method can be used over a wider range of dT/dt than the mass spectrometer method
[28].
For more complicated surface phenomena in which surface decomposition of a
molecule may occur without product desorption, the use of the Kelvin probe for
observing work-function changes associated with decomposition has proven helpful
[29]. The Kelvin probe may also be used to measure work-function changes on
semiconductor surfaces when adsorption occurs [30].
should agree with that obtained from the derivative of the work function. It was
found that an improvement in the resolution of the thermal desorption spectroscopy
could be achieved by using the Kelvin probe, compared to the mass spectrometer
detector of desorbing gas. However, the two kinetic curves do not agree because of
the nonlinear behavior of work function with coverage. Figure 27.5c shows data for
the thermal desorption of CO from Ru(001) measured both ways [31].
Reference [31] gives an excellent review of the development of the vibrating
capacitor method for work-function measurements.
An excellent analysis of the effect of stray capacitance on Kelvin probe mea-
surement is given in [32], and the effect of noise on Kelvin probe measurements is
given in [33]. Reference [34] gives a new design for a Kelvin probe involving a
voice coil driver. Reference [35] gives design and performance features for a Kelvin
probe with a reproducibility of 1 meV over long periods of time. In addition,
methods for measuring the work-function change during ohmic heating of a metal
crystal are described [35].
The vibrating Kelvin probe has been described on pp. 295–299. Both mechanical
and electromechanical activation of the motion of the reference electrode have been
employed. For electromechanical activation, a piezoceramic actuator is often used,
but because the upper temperature limitation is near 415 K for these actuators, they
are not generally useful for high-temperature studies.
Figure 27.6 shows a method for using a piezoceramic actuator [36] to drive a
Kelvin probe at some distance from a high-temperature sample [37]. The device,
with a balanced quartz tube lever of 180-mm length, has the actuator on one end
and the Au Kelvin probe electrode on the other. The quartz tube (2-mm diameter)
lever arm is balanced on sapphire rods, as shown in the drawing. The actuator is
coupled to the quartz tube lever with a light weight stainless steel spring. This
refinement avoids the addition of a large mass to the actuator, and the resonant
frequency of the device is close to that of the free actuator (in this case, 325 Hz).
The apparatus is used to study chemical reactions on a single crystal sample
placed under the Kelvin probe. This sample may be heated in flowing gas while the
work function is being measured. The large distance between the hot sample and
the actuator avoids heating of the actuator. In addition, the large separation distance
prevents electromagnetic pickup in the work function measurement from the pi-
ezodrive unit. A sampling tube permits the gas composition to be measured in the
vicinity of the sample, and this provides a measurement of the local rather than the
incident gas composition.
The resolution of the probe was estimated to be better than 5 meV, using known
applied potentials on the sample for calibration. It is believed that improvements in
the electrical measurement methods will lead to substantially lower noise and
higher resolution [37].
300 27 Work Function, Tunneling Spectroscopy and Ellipsometry
Many examples if the use of I/V measurements for probing the local density of
electronic states (LDOS) of adsorbed atoms and molecules are now present in the
literature. Similarly, for alloy metal surfaces, STM-STS measurements are often
used for studying the individual metal atoms, either in the surface or the subsurface
layers of the metal. A common difficulty with these measurements is that the I–V
and dI/dV measurements versus V contain a contribution from the LDOS of the tip
atoms which is mixed in. In the method to be described here, a subtraction of STS
spectra is made in order to cancel out the LDOS contribution from the tip.
Figure 27.7a shows a schematic alloy of composition AxBy. The STM, through a
number of averaged scans across an identical region, provides first a dI/dV scan of
atom A, which contains, also the tip contribution. A similar number of scans across
atom B provides a second dI/dV scan, also containing the tip contribution. By
subtracting A−B, the tip contribution is cancelled out, and the difference spectrum
represents the change in dI/dV when one moves from atom A to atom B. These
differences are able to be handled more easily by theory than the convoluted spectra
from which the difference spectrum is derived [38].
27.7 Scanning Tunneling Spectroscopy (STS)—Compensation … 301
One example shown in Fig. 27.7b concerns the Cu-Pd alloy, where STM can
distinguish surface Cu atoms, surface Pd atoms, and subsurface Pd atoms. The data,
when subtracted give a difference spectrum for the two types of Pd. Here the Cu
spectrum has been used as a subtraction spectrum, removing the tip contribution.
The difference spectrum therefore represents the difference between Pd and Cu. It is
seen that at about −0.4 V, both types of Pd atoms exhibit a slightly lower LDOS
than Cu atoms. Consequently, charge transfer takes place from Pd to Cu atoms, just
the opposite from a prediction based on the higher work function of Pd (5.12 eV)
versus Cu (4.65 eV) [39].
302 27 Work Function, Tunneling Spectroscopy and Ellipsometry
Ellipsometry has been used for years to study adsorption on surfaces, especially in
the study of oxide films grown on metals. There are few comparisons of adsorption
measurements with other surface measurements, especially under ultrahigh vacuum
conditions for model systems with well-understood behavior.
The ellipsometer is an automatic null system with a polarizer-compensator-
sample-analyzer (PCSA) configuration as shown in Fig. 27.8a. Here, two ellips-
ometry parameters are measured: “phase shift” Δ, and “amplitude reflectance ratio”
ѱ, which are components of the optical constant of the surface. The relative change
in Δ, δΔ, measured versus the clean substrate is used as a measure of the amount of
adsorption. This is due to the fact that δΔ is much more sensitive than δѱ for a
physisorbed film thinner than a few 10s of nm [40]. The apparatus shown also
contains XLEED, which is a sensitive low energy electron diffraction method using
a low current electron beam with microchannel plate amplification of the diffraction
pattern. XLEED eliminates the problem of electron stimulated desorption (ESD).
Measurement of the coverage behavior of Xe is achieved by measuring the intensity
of the diffracted Ag(10) beam [41, 43].
Figure 27.8b shows a comparison of Ellipsometry and XLEED for the mea-
surement to the adsorption of Xe on both Ag(111) and Ag(100). At a constant
pressure of Xe of 1.1 × 10−5 Pa, the temperature of the crystal was slowly lowered
from 82 to 56 K. The broken line shows the behavior of the ellipsometry mea-
surement of δΔ, versus T, where equilibrium is maintained at all temperatures. The
initial increase is due to the adsorption of 1 ML of Xe, followed by a second
increase due to the formation of a second layer. The solid line shows the XLEED
behavior of the [10] beam. The two experimental methods compare two crystals,
but even in this case, it may be seen that the condensation temperature for the
second layer is identical to 0.1 K. The slight difference in condensation temperature
for the monolayer, measured by the two techniques, is not due to differences
between the techniques, but is due to slight differences in the polarization of Xe on
the two Ag surfaces, leading to a difference of 0.2−0.4 K in the position of the step
corresponding to monolayer completion.
References 303
References
7. J.L. Taylor, W.H. Weinberg, J. Vac. Sci. Technol. 15, 1811 (1978)
8. T.E. Madey, J.T. Yates, Jr., Colloques Internationaux du Centre National de la Recherche
Scientifique, No. 187, 155 (1970)
9. J.T. Yates Jr, R. Klein, T.E. Madey, Surf. Sci. 58, 469 (1976)
10. M.J. Dresser, T.E. Madey, J.T. Yates Jr, Surf. Sci. 42, 533 (1974)
11. J.R. MacDonald, C.A. Barlow Jr, J. Chem. Phys. 39, 412 (1963)
12. J.H. Fritz, C.A. Haque, Rev. Sci. Instrum. 44, 394 (1973)
13. H. Shelton, Phys. Rev. 107, 1553 (1957)
14. P. Kisliuk, Phys. Rev. 122, 405 (1961)
15. T.E. Madey and J.T. Yates, Jr., Supplemento al Nuovo Cimento, No. 2, Series I, 5, 483, (1967)
16. G. Ertl and J. Küppers, Low Energy Electrons and Surface Chemistry (Verlag Chemie, GmbH,
D-694 Weinheim, 1974), p. 120
17. J. Holzl, F.K. Schulte, in Solid Surface Physics, ed. by G. Hohler, vol 85 (Springer, Berlin,
1979), p. 1
18. S. Saito, T. Soumura, T. Maeda, J. Vac. Sci. Technol. A2, 1389 (1984)
19. K. Besocke, S. Berger, Rev. Sci. Instrum. 47, 840 (1976)
20. G. Wedler, R. Ruhmann, Appl. Surf. Sci. 14, 137 (1982–1983)
21. S. Saito, T. Maeda, J. Vac. Soc. Jpn. 25, 190 (1981)
22. Piezofoil can be obtained from Kyoto Ceramic Ltd., Kyoto, Japan
23. I.D. Baikie, E. Venderbosch, J.A. Meyer, P.J.Z. Estrup, Rev. Sci. Instrum. 62, 725 (1991)
24. see also J.S.W. de Boer, H.J. Krusemeyer, N.C. Burhoven Jaspers, Rev. Sci. Instrum. 44, 1003
(1973)
25. I.D. Baikie, S. Mackenzie, P.J.Z. Estrup, J.A. Meyer, Rev. Sci. Instrum. 62, 1326 (1991)
26. I.D. Baikie, K.O. van der Werf, H. Oerbekke, J. Broeze, A. van Silfhout, Rev. Sci. Instrum.
60, 930 (1989)
27. C.S. Kumar, A. Subrahmanyam, J. Majhi, Rev. Sci. Instrum. 67, 805 (1996)
28. P. Feulner, D. Menzel, J. Vac. Sci. Technol. 17, 662 (1980)
29. L. Hanley, X. Guo, J.T. Yates Jr, J. Phys. Chem. 93, 6754 (1989)
30. J. Abelson, G. de Rosny, J. Physique 44, 993 (1983)
31. H.A. Engelhardt, P. Feulner, H. Pfnur, D. Menzel, J. Phys. E10, 1133 (1977)
32. I.D. Baikie, E. Venderbosch, J.A. Meyer, P.J.Z. Estrup, Rev. Sci. Instrum. 62, 725 (1991)
33. I.D. Baikie, S. Mackenzie, P.J.Z. Estrup, J.A. Meyer, Rev. Sci. Instrum. 62, 1326 (1991)
34. I.D. Baikie, K.O. van der Werf, H. Oerbekke, J. Broeze, A. van Silfhout, Rev. Sci. Instrum.
60, 930 (1989)
35. M. Chelvayohan, R. Gomer, Surf. Sci. 172, 337 (1986)
36. A Phillips PXE 5 piezoactuator with length of 27 mm was used
37. S. Lundgren, B. Kasemo, Rev. Sci. Instrum. 66, 3976 (1995)
38. H.L. Tierney, A.E. Baber, E.C.H. Sykes, J. Phys. Chem. C 113, 7246 (2009)
39. A.E. Baber, H.L. Tierney, E.C.H. Sykes, ACS Nano 4, 1637 (2010)
40. A. Itakura, I. Arakawa, J. Vac. Sci. Technol., A 9, 1779 (1991)
41. S. Igarsshi, Y. Abe, Y. Irie, T. Hirayama, I. Arakawa, J. Vac. Sci. Technol., A 16, 974 (1998)
42. S. Igarashi, A. Tosaka, T. Hirayama, I. Arakawa, Langmuir 19, 4627 (2003)
43. A. Tosaka, I. Arakawa, Surf. Sci. 600, 1071 (2006)
Chapter 28
Radioactive Adsorbates
The high sensitivity of radiochemical methods suggests that these methods would
have utility in studies of chemisorption and reaction on single crystal surfaces. This
description of the use of a silicon surface barrier detector for the measurement of
14
C in chemisorbed layers on Pt(111) represents one of the few attempts to make
measurements of this type [1]. Earlier studies using a Geiger-Müller counter were
also successful [2, 3].
The surface barrier detector [4] is a large area diode consisting of a depleted slice
of n-type Si mounted in an insulator ring as shown in Fig. 28.1a. The front and back
surfaces are metallized. The thickness of the Au front surface coating is optimized
for low energy β counting (158 keV) from 14C. Also, the depth of the depletion
region (300 μm) is chosen to optimize the detection of the β-electrons from 14C,
which have a range of about 150 μm in Si. Free charge carriers produced by the β-
emission are separated by an applied electric bias (50 V). Integration of the current
induced between the front and back electrodes yields an output pulse that is pro-
portional to the ionizing energy. Either β-spectra may be collected or a ratemeter
may be used to count the level of radioactive material, using the appropriate dis-
criminator settings to eliminate noise.
The detector, *20-mm in diameter, is mounted on a rotary feedthrough that
allows it to be positioned in front of the Pt crystal at a distance of 5–10 mm, as
shown in Fig. 28.1b. In the opposite orientation, background counting may be done.
When a bias is applied to the detector, the temperature must be maintained between
−30 °C and +30 °C. This is done by using a flexible Cu braid (see p. 392) that is
anchored on a copper block maintained at the desired temperature. To avoid con-
tributions from detector noise, the lower discriminator was set at 35 keV, and this
resulted in rejection of about 40 % of the 14C β-signal as determined in separate
calibrations. The counting efficiency was determined as 2.9 ± 0.3 % for the
geometry and discriminator settings used here.
Fig. 28.1 Radiotracer techniques for surface science. a Si surface barrier detector. b. Detector
geometry. c C Auger signal versus radioactive 14C coverage absolute surface coverages
28.1 Radiotracer Techniques for Surface Studies 307
Figure 28.1c shows the correlation between the measured count rate and the
Auger intensity from carbon deposited as 14C–C6H6 on the Pt (111) crystal. The
linear correlation between the two measurements indicates that both are measuring
the carbon coverage as expected.
The sensitivity of the 14C counting method depends upon many factors such as
geometry, detector parameters, and the specific activity of the carbon-containing
compounds used. While many hydrocarbons and other molecules may be obtained
with activities up to 64 mCi/mmol, a nominal value of 10 mCi/mmol is more
common. Assuming a detection efficiency of 3 %, and a background signal of 3
counts/min, the detection sensitivities for 20-min counting, at nominal labeling
levels, are 1 × 1013 (23 % standard deviation), 5 × 1013 (10 % standard deviation),
and 1 × 1014 (4 % standard deviation) molecules/cm2. Thus, coverages of 1 % of a
monolayer may be determined accurately. It is found that repeated adsorption
experiments lead to increasing background radioactivity during measurements over
a day; pumping overnight restores the background to a low level.
A second example of the use of this method for the study of hydrocarbon
chemistry on Pt(111) can be found in [5].
The use of radioactive adsorbates is not limited to those labeled with 14C as may
be seen in Table 28.1.
References
1. S.M. Davis, B.E. Gordon, M. Press, G.A. Somorjai, J. Vac. Sci. Technol. A19, 231 (1981)
2. K. Klier, Rev. Sci. Instrum. 40, 372 (1969)
3. K. Klier, A.C. Zettlemoyer, H. Leidheiser Jr, J. Chem. Phys. 52, 589 (1970)
4. Ortec TA-23-25-300. Obtained from Ortec, P.O. Box 1397, Easley, SC 29640
5. S.M. Davis, F. Zaera, B.E. Gordon, G.A. Somorjai, J. Catal. 92, 240 (1985)
Chapter 29
Thin Film Deposition
The ability to evaporate clean thin films and to achieve desired surface coverages
that are absolutely known is of fundamental importance in many areas of surface
science research. Fortunately, quartz thickness monitors (QTM) are commercially
available that have sensitivities of the order of a nominal thickness of 1 Å of
deposited material. With appropriate use, these monitors can be used to monitor,
absolutely, the deposition of thin films on substrates of interest.
Figure 29.1a shows a successful arrangement for producing absolute coverages of
metal on a single crystal [1]. In the case of this work, the metal deposited was Pt and
the substrate was atomically clean Mo(110). The metal deposition source consists of
a tungsten hairpin filament, mounted on degassing loops (see pp. 407 and 437),
which is shielded by a metal shield containing a 15-mm-diameter hole near the
source. A thermostated QTM is able to be moved quickly between three positions—
1, 2, and 3. When the QTM is in position 1, the Pt source is completely shielded, so
that metal does not escape into the vacuum chamber. Position 1 is used for through
outgassing of the source. When the QTM is moved to position 2, the QTM collects
evaporated metal and radiation from the hot source. Thermostatting with circulating
water from a temperature-regulated source maintains the QTM temperature constant
to ±0.5 K, and this condition is necessary for absolute accuracy of the QTM. When
the QTM is rotated to position 3, the crystal receives the metal vapor and the QTM is
out of line-of-sight with the metal source. Stable operation of the metal source can be
achieved by monitoring the current and voltage drop across the source and making
the product of these constant. In practice, the source is stable over hours of operation
without sophisticated wattage regulation. The relative flux at the crystal and the
QTM is determined from the distances by the ratio, d2QTM/d2crystal.
Figure 29.1b shows the duty cycle described above as monitored by the QTM. In
position 1, no Pt is received either by the QTM or the crystal. In position 2, the
mass of the deposited platinum on the QTM rises linearly with time. In position 3,
Fig. 29.1 Thin Film Deposition to Absolute Surface Coverage. a Metal evaporation/quartz
thickness monitor. b Duty cycle for evaporation/measurement. c Layer-by-layer growth kinetics Pt
deposition measurement
29.1 Thin Film Deposition to Absolute Surface Coverages 311
Pt is deposited on the Mo(110) substrate and the QTM again does not show any
change in mass. Finally, in position 2 for the second time, the QTM again receives
Pt, to check that the rate of evaporation has not changed significantly during
deposition. The two slopes in position 2 agree to within 0.4 %.
Following each deposition cycle, the Mo(110) crystal was rotated to the CMA
Auger spectrometer position for measurement of the Pt(NW) intensity, which is a
secondary measure of the extent of Pt deposition. It can be seen in Fig. 29.1c that
the Pt Auger signal grows linearly in at least three branches of different slope. The
break points, located analytically, correspond to the completion of a monolayer of
Pt and the start of the formation of a second layer of Pt as growth proceeds by a
layer-by-layer mechanism. It is interesting to note that the three labeled break points
correspond well, in absolute Pt coverage as measured with the QTM, with the
312 29 Thin Film Deposition
Fig. 29.2 Au Evaporation Source. a Au evaporation source. b Evaporation versus time. c Power
consumption (I × V) versus time
droplet in the notch region. In Fig. 29.2a, the evaporation source is shown. It
consists of a W wire [18] with the notch, welded between two Ta degassing loops.
These loops can be independently degassed at higher temperatures than employed
for the Au evaporation and eliminate outgassing effects, as described in Sect. 38.1.
Current and voltage measurements (Fig. 29.2b) are made as shown, allowing one to
measure the watts expended and to keep this quantity constant throughout the
evaporation. Measuring the voltage drop across the active W evaporation filament is
superior to measurements made across the entire device, since one wishes to control
the resistance and temperature of the active evaporation element. Figure 29.2c
shows a series of Au deposition rate measurements made over several months. The
reproducibly is better than ±20 %, and allows Au to be deposited far down into the
submonolayer range at a slow rate. The thickness measurements for the Au film are
made by measuring the Auger attenuation of the substrate, which in this case is a
TiO2(110) single crystal [19].
314 29 Thin Film Deposition
A convenient and inexpensive secondary standard for the measurement of the rate
of metal evaporation may be established using a pair of ionization gauges with
matched geometry. The flux of metal atoms through one gauge is measured by an
increase in ion current owing to the metal atom steady state gas density during
evaporation. A second gauge is used to make a small correction for the presence of
residual gases produced during evaporation. The calibration of the ion gauge is
made against a quartz thickness monitor (QTM) [20] (see p. 309).
Figure 29.3a shows the design of the apparatus, where a translatable QTM is
located line-of-sight to a metal evaporation source that contains two separate
sources housed in a shield. It is desired to evaporate known coverages of metal onto
the crystal, and the QTM is located very near the crystal location. After outgassing
of the source, the crystal is moved to a line-of-sight position and measurements of
the mass received by the QTM are made. Simultaneously, measurements of the
difference between the two ionization gauge readings are integrated versus time,
and this difference is related to the metal vapor steady state gas density inside the
ionization gauge assembly, which is located line-of-sight between the source and
the crystal and the QTM.
Figure 29.3b shows a cross calibration obtained for Sn evaporation. The linear
correlation over a wide Sn thickness range indicates that both measurement
methods are behaving as expected. However, the sensitivity of the ion gauge is high
compared to the commercial QTM, so that small film thicknesses can be measured
with the ionization gauge conveniently.
The dual-ionization gauge arrangement could also be calibrated against other
thickness standards such as RHEED intensity measurements made during
layer-by-layer growth [21].
Quartz crystals used in thin film measurement are able to hold only so much
deposited material before becoming unusable. Typical thickness limitations are
approximately 105 Å of metal deposit. The thin films can sometimes be etched
away to recover the clean quartz oscillator, but strong chemical etchants will often
destroy the metallic electrodes that activate the crystal, requiring redeposition of the
electrodes [22].
A simple solution has been found to allow thin film overlayers on quartz crystals
to be removed without harming the electrodes [23]. A photoresist coating is
29.4 Reusable Quartz Crystals for Film Thickness Measurement 315
Fig. 29.3 Ionization Gauge Measurements of film Evaporation Rate. a Ionization gauge geometry.
b Correlation of QTM and BAG
deposited on the clean quartz crystal before use. Later, using acetone as a solvent
for the photoresist, the thin film to be removed is stripped away by dissolving the
underlying photoresist layer as shown in Fig. 29.4. It was found that a photoresist
layer of at least 3000 Å thickness is needed to permit stripping of all overlayer
316 29 Thin Film Deposition
deposits cleanly. This loads the quartz crystal only to about 1 % of its mass
capacity. The photoresist [24] is deposited using a spin-coating method, baked at
120 °C for 20 min, and then patterned to cover the area on the crystal used for
deposition. The photoresist coatings made in this manner can be stored in air but
should not be baked at temperatures approaching 200 °C because this makes the
photoresist insoluble in acetone.
29.5 Hybrid Thin Film Deposition Processes—Stable and Metastable Films 317
The deposition of thin films with desirable chemical and physical properties is
widely used for the production of electronic materials, tribological coatings, and
optical coatings. The processes used may be divided into those that produce ther-
modynamically stable compound films and those that produce metastable materials.
It is beyond the scope of this book to describe the wide range of film deposition
methods and materials that have been deposited in thin film form. Books and
reviews are widely available. Below, a short description of the some of the hybrid
film deposition methods is given [25], and figures schematically describing each
method are shown. Figure 29.5a–d illustrate ion bombardment from an external
plasma. Figure 29.5e illustrates the use of ions extracted from a plasma near the
substrate, and this method is now used widely in plasma assisted vapor deposition
processes.
Figure 29.5a—Ion Beam Deposition: Ions are extracted from the source where
they are produced by plasma excitation of the appropriate gas. The method has the
advantage that the energy of deposition of the ions can be controlled rather than
working with a mixture of ions, neutrals, and excited neutrals.
Figure 29.5b—Ion Beam Sputter Deposition: Ions are produced in an ion source
by a plasma. These are used as a means for sputttering a target, and collection
occurs on the substrate.
Figure 29.5c—Dual Beam Sputtering: Two ion beams are used. One sputters a
target; the second bombards the growing film on the substrate.
Figure 29.5d—Ion Bombardment Evaporation: Material is evaporated thermally
using electron beam or ohmic heating. The growing film is bombarded by ions
extracted from an ion source.
Figure 29.5e—Reactive Sputtering with Dissociable Reactive Gas: Metals are
sputter-deposited in a reactive gas such as silane to produce metal silicide films.
Many variations on the schematic methods shown in Fig. 29.5a–e are employed
and a wide variety of preparation methods are described in the literature [26, 27].
The need to coat the inside of a metal or insulator tube with a metal film can be met
by both electrochemical and chemical vapor deposition methods. It is also possible
to deposit pure metal films inside a tube using a special sputtering procedure
described here [28].
As shown in Fig. 29.6, a conical metal target, placed on the axis of a polished
steel tube to be coated, is exposed to a beam of ions that are directed down the tube.
The conical metal target is moved along the tube toward the ion source by an
electrical drive mechanism. In the particular case studied here, the inner diameter of
318 29 Thin Film Deposition
Fig. 29.5 Hybrid thin film deposition processes. a Ion beam deposition. b Ion beam sputtering.
c Dual beam sputtering. d Ion bombardment—evaporative deposition. e Reactive sputtering—
dissociable reactive gases
the tube was 6 mm, and a conical Ag target of 5 mm diameter was employed.
The target was pushed through the tube at a rate of 0.05 mm/s, while a 2 keV Ar+
beam was incident. A sputter yield of about 7 Ag/Ar+ was measured for a 60°
incidence angle.
29.6 Sputter-Coating Inner Walls of Tubes 319
The conical target was chosen so that the maximum sputter yield could be
achieved using a 60° angle of incidence for the ions on the target surface. This
optimum Ar+ incidence angle for Ag sputter deposition was determined experi-
mentally using a separate apparatus that permitted the collection of the Ag metal
sputter deposit and its measurement by chemical analysis.
To determine that the Ag film deposited was uniform in thickness, the tube was
cut along its axis in four sections. Visual inspection showed that a complete Ag film
was deposited along the tube length, and sputter-depth profile measurements
showed that depth of the film was uniform to within about 10 %.
Experiments with Pt deposition on Ta tubes were also carried out and subsequent
mechanical embrittlement tests showed that the Pt deposit was very effective in
preventing hydrogen-induced embrittlement of the Ta tubes.
References
1. A. Linsebigler, G. Lu, J.T. Yates Jr, Surf. Sci. 294, 284 (1993)
2. S. Tanuma, C.J. Powell, D.R. Penn, W.F. Egelhoff Jr, J. Vac. Sci. Technol A8, 2213 (1990)
3. G. Sauerbrey, Zeit. Phys. 155, 206 (1959)
4. L.J. Slutsky, W.H. Wade, J. Chem. Phys. 36, 2688 (1962)
5. B. Kasemo, E. Tornqvist, Surf. Sci. 77, 209 (1978)
6. C. Lu and A. W. Czanderna, in Applications of Piezoelectric Quartz Crystal Microbalances
(Elsevier, New York, 1983), Chapter 2
320 29 Thin Film Deposition
the crystal in the optical cell with a high pressure of gas also present, the spectral
contribution from the gas phase can be canceled almost completely.
As shown in Fig. 30.1b, the single crystal is moved by a long throw manipulator
from the surface analytical positions in the ultrahigh vacuum chamber (Level 2 and
Level 3) up into the IRAS cell (Level 1). The crystal can be rotated through 360° by
means of a doubly differentially pumped Teflon rotary seal (see p. 6) for precise and
reproducible angular orientation to the IR beam, which typically has an angle of
incidence of about 86°. The windows of the IR cell can be made of single crystal
CaF2 mounted in Conflat flanges [4] or can be made of KBr single crystals mounted
by means of a differentially pumped double Viton seal assembly (see p. 521). The
small optical chamber is sealed by a metal-seal gate valve that does not have the
problem of Viton O-ring outgassing under compression nor the problem of per-
meation. For a sealed-off static cell, this is important. For high pressure experi-
ments, the composition and pressure of the gas phase may be determined by
spectroscopic measurements through the reference cell for infrared active gases.
Figure 30.1c shows a comparison of operation of the IRAS apparatus in single
beam mode and in double-beam mode. The optical path is purged with dry nitrogen.
After 6 h, a severe intensity drift in atmospheric H2O (about 0.4 % T) and atmo-
spheric CO2 (about 0.2 % T) can be seen in the spectrum obtained by the single
beam method. In the spectrum obtained by the double-beam method, this effect is
eliminated. The comparison on Fig. 30.1c shows the ability of the double-beam
method to eliminate background effects owing to slow drifts during long periods.
For 900 scans for both S and R infrared measurements, a noise level of about
1 × 10−4 absorbance unit may be routinely obtained. The amount of light delivered
to the detector is significant in determining the noise level. Therefore, large single
crystals of the order of 15 mm diameter are favored in this work.
A FTIR spectrometer mounted on a precision optical mount has been employed
for IRAS measurements in which the path length between the spectrometer and the
external detector has been minimized by using only KBr lenses for focusing the
radiation [14]. Both the spectrometer and the detector are coupled to the UHV
chamber by small bellows, permitting each unit to be independently adjusted with
precision XYZ optical mountings.
Recently, an instrument designed to measure the mass of adsorbate on a thin film
in addition to the infrared spectrum has been described [15]. The working tem-
perature range of the quartz crystal oscillator (see p. 309) used for adsorbate mass
measurements is 80–350 K [15].
The instrument shown in Fig. 30.1 is described in detail in [16].
The IRAS apparatus described on p. 321 has the special capability if working at
elevated pressures. At these pressures, electron spectroscopies fail because of the
reduction of the mean free path of electrons to lengths comparable to or less than
324 30 Infrared Spectroscopy
the electron traveling distances inside the apparatus. Fortunately, no such limita-
tions exist for optical techniques, assuming the gas phase at high density is optically
transparent in the required spectral regions.
An example is shown of the study of N2 adsorption on the Ni(111) surface.
Figure 30.2a shows that as the N2 pressure is raised over 4 orders of magnitude from
1 × 10−8 to 2.9 × 10−4 mbar, the equilibrium coverage of chemisorbed N2 at a crystal
temperature of 89 K increases as indicated by the increase in the absorbance of the
chemisorbed N2. Data of this sort were collected as a function of crystal temperature
in the range 89–115 K, and the results are shown in Fig. 30.2b, where the vertical
scale is the integrated IRAS absorbance. It is clear that as the temperature is
increased, higher pressures of N2 must be employed to produce a given integrated
absorbance for the chemisorbed N2. If one assumes that the integrated absorbance is
proportional to surface coverage (an approximation of unknown accuracy) then the
“isoabsorbance” plots versus reciprocal absolute temperature can be made, as shown
in Fig. 30.2c, and from the slope of these plots the enthalpy of adsorption may be
deduced at particular integrated absorbances. The inset to Fig. 30.2c shows this plot,
indicating that the enthalpy of adsorption is almost constant, or rises very slightly
with increasing absorbance (or coverage within the approximation used above).
Work of this type, in which an initially clean single crystal is exposed to high
pressures of a gas, must be carefully done with highly purified gas in order to avoid
contamination. For the example of about 10−4 mbar of gas, impurity levels of only
10 ppm will give sufficient impurity partial pressure that in about 1000 s of
exposure, one monolayer of impurity could be adsorbed if the impurity has a
sticking coefficient of unity over its coverage range in the presence of the dominant
gas. Effects of this type can often be discovered by examining the reversibility of
the spectral changes, or by desorption of the dominant adsorbate, followed by
readsorption. The purification of gases is discussed on pp. 505–514.
The slight changes in the vibrational spectrum of the chemisorbed N2 are
thought to be due to various island structures that develop at various equilibrium
coverages. The fact that so little shift occurs in the N–N stretching mode (a 24 cm−1
range) suggests that the electronic character of the chemisorbed N2 is not strongly
affected by changes in coverage, and suggests that the approximation that integrated
absorbance is proportional to coverage may be valid [17].
Fig. 30.2 IRAS measurements at high pressure. a Development of IRAS Spectra of N2/Ni(111).
b N2 Adsorption Isotherms on Ni(111) by IRAS. c Isosteric heat of adsorption N2/Ni(111) Infrared
reflection-absorption spectroscopy (IRAS)
326 30 Infrared Spectroscopy
Figure 30.3a shows the geometrical arrangement used for holding a silicon
single crystal on a manipulator and for making connections for electrical heating
[18]. Here, Ta clips are attached to the top of the U-shaped crystal and the crystal
may be heated by the passage of current. Conductive cooling to the clips causes a
temperature gradient across the crystal (100 K at 1000 K), but in the portion probed
by the IR beam, only a small gradient is observed. The edges of the crystal are
beveled and polished for admittance of the focused IR beam. The vector diagram
for p- and s-polarization is also shown in Fig. 30.3a. Oxidation of the top surface
deactivates this surface for adsorption.
30.3 Internal Reflection IR Spectroscopy—Silicon 327
Figure 30.3b shows the UHV system and the associated optical system for use in
the FTIR internal reflection measurements [19]. The U-shaped silicon crystal is
mounted on a manipulator and may be rotated to an Auger spectrometer or to a
LEED apparatus. In its position for internal reflection spectroscopy, the focused IR
beam from the FTIR spectrometer passes through a CsI IR window (see p. 521) and
enters the silicon sample at the proper incident angle. After many internal reflec-
tions, the exiting radiation passes through another CsI window and is focused by a
mirror through a polarizer and onto the detector.
For silicon, the absorption of infrared light below about 1450 cm−1 prevents
transmission with typical sample lengths of 3–5 cm.
Excellent reviews of this field, giving many other optical arrangements for doing
IR surface spectroscopy on semiconductors, are available [20–25].
The use of internal reflection IR spectroscopy for silicon single crystals has been
discussed on p. 324. Different problems exist when the same measurement tech-
nique is applied to gallium arsenide. In order to permit IR transmission through
GaAs, the carrier concentration needs to be kept below about 1016/cm3 [26, 27].
GaAs with this level of carriers is commercially obtainable from one supplier,
which produces 1 Ω-cm material [27]. Normally, wafer GaAs is sold with con-
ductivities of about 106 Ω cm (semi-insulating), or about 10−2 Ω cm. While the
semi-insulating material would be satisfactory for IR measurements, it cannot be
conveniently heated by resistive heating; alternatively, the highly doped material
can be heated electrically, but its resistance is so low that large currents are needed
and temperature nonuniformities are produced [26]. This material is also opaque in
the infrared [26].
The mounting of 1 Ω-cm material for internal reflection infrared spectroscopy is
shown in Fig. 30.4. The infrared beam enters and exits through polished beveled
edges. Heating current flows perpendicular to the IR beam direction, using Ta
mounting clips on the top and bottom of the crystal. A thermocouple is conve-
niently glued to the crystal using a refractory cement, used also for thermocouple
attachment to other semiconductor crystals such as SiC [28]. Welding the ther-
mocouple first to a Ta foil square (1.2 mm × 1.2 mm) and then gluing this to the
GaAs provides good agreement between the thermocouple and pyrometry [26]. The
advantages of this mounting arrangement are that either surface can be studied, or if
both are studied together a two-fold improvement in the signal strength in the FTIR
measurement is obtained. The two sides of the crystal, if terminated by the same
element, exhibit orthogonal orientation of the dangling bonds. Thus, using s- and p-
polarized IR radiation, one can probe the adsorbate orientation [26]. A disadvantage
of working with the 1 Ω-cm material is that the resistance becomes too great for
328 30 Infrared Spectroscopy
electrical heating when the sample is cooled to 120 K. This problem can be alle-
viated by warming the sample slightly with a projector lamp (see p. 371) [26].
InSb and HgCdTe infrared detectors are used in the mid-IR range (4000–500 cm−1)
for the study of adsorbed species on surfaces. The detector element is mounted on
the end of a small metal Dewar which is kept filled with liquid nitrogen during
operation, as shown in Fig. 30.5. The Dewar is often pumped on a separate vacuum
system and then sealed with a Viton-sealed valve. In addition, the detector window
is often sealed with a Viton O-ring. Thus, over time, slow permeation, outgassing,
and leakage can be expected (see p. 108), and the condensation of water and other
materials may occur on the detector surface, causing interference with infrared
measurements. These problems can be prevented by using a small appendage ion
pump to continuously pump the detector [29].
The IR detector is connected to a metal bellows-sealed valve and thence to a tee
as shown in Fig. 30.5. One branch of the tee is connected to the ion pump via a
1.5-cm-diameter bellows that extends up to about a 30-cm length as it makes the
transition from the detector to the ion pump. The second branch is temporarily
connected to an oil diffusion pump (liquid nitrogen trapped) for preliminary
pumping of the Dewar and the appendage pump.
30.5 Infrared Spectroscopy—Continuous Pumping of Detector Dewar 329
Initially the entire system is pumped out by the diffusion pump. The valve on the
detector is then closed and the ion pump (not operating) and the metal tubing are
baked out with heating tape. When the pressure is low enough that the ion pump
can be operated, the valve to the diffusion pump is closed and the diffusion pump is
removed from the vacuum system. The heating is discontinued and when the line is
cooled, the valve to the detector is opened. Under these conditions, the IR detector
will be continuously pumped and condensation effects will be eliminated on the
detector. It is wise to place a conical shield just below the top of the Dewar entrance
to prevent liquid nitrogen from accidentally pouring over the detector window and
the vacuum seal for the amplifier wires, since these vacuum seals may be broken by
such accidental cooling [30].
Infrared spectrometers require that the optical path length be free of absorbing
atmospheric gases. Both CO2 and H2O vapors must be removed by using either
N2(g) or specially-dried and CO2-scrubbed compressed air. Some IR spectrometers
have a sample compartment which is separately purged from the spectrometer
330 30 Infrared Spectroscopy
source and detector compartments. The sample compartment is usually rather large
to admit large IR cells. It usually contains a sample compartment lid which is
opened and closed as the cell is transferred in and out. The cell and its connecting
tubing and wiring will cause problems with compartment sealing, requiring cutting
of the lid and unsatisfactory sealing with tape, etc. In Fig. 30.6a, an arrangement for
a portable cell is shown in which custom-cut rubber purge gas sleeves are arranged
to slide into the sample compartment under slight compression, thereby making a
gas tight seal with the inner walls of the sample compartment and the inserted IR
cell. The seals are held on the cell window flanges with band clamps. With this
arrangement, the IR spectrometer remains purged always, and the small volume
within the rubber sleeves is separately purged with great rapidity. Figure 30.6b
shows a comparison of the purge rate for the sleeves, compared to the entire
spectrometer, for CO2 and H2O. It is seen that the small-volume rubber sleeves are
purged about 25 times more rapidly than the entire spectrometer sample compart-
ment. In addition, complex gas-tight seals involving the compartment lid are
avoided, and the lid may be removed permanently. Soft rubber sleeves may be
obtained in several sizes from the plumbing department of a local hardware store,
where they are sold as flexible couplings for sewer systems. To cut the length to fit,
a mandrel is used on a lathe, and a smooth cut suitable for purge gas sealing at the
wall is produced with a sharp blade. This apparatus was produced in [31].
References
19. Y.J. Chabal, Lucent, AT&T (personal communication). See Y.J. Chabal, E.E. Chaban, S.B.
Christman, J. Electron Spect. Rel. Phenom. 29, 35 (1983b)
20. N.J. Harrick, Internal Reflection Spectroscopy (Wiley, New York, 1967)
21. R.M. Mirabella, N.J. Harrick, Internal Reflection Spectroscopy: Review and Supplement
(Harrick Scientific Corporation, Ossining, NY, 1985)
22. Y.J. Chabal, in Semiconductor Interfaces: Formation and Properties, eds. by G. LeLay,
J. Derien, N. Boccara, Springer Proc. Phys. 22, 301 (1987)
23. Y.J. Chabal, Surf. Sci. Rep. 8, 211 (1988)
24. Y.J. Chabal, in Internal Reflection Spectroscopy: Theory and Applications, ed. by R.M.
Mirabella, Jr. (Marcel Dekker, New York, 1993), printed in Practical Spectroscopy, 15, 191
(1993)
25. M. Katiyar, J.R. Abelson, J. Vac. Sci. Technol. A13, 2005 (1995)
26. Dr. J.R. Creighton, Dept. 1126, MS 0601, P.O. Box 5800, Sandia National Laboratories,
Albuquerque, NM 87185-0601 (personal communication). J.R. Creighton, Appl. Surf. Sci.
82/83, 171 (1994)
27. GaAs with no intentional dopants (HB material, about 1 Ω-cm) is available from Sumitomo
Electric, 551 Madison Avenue, New York, NY 10022
28. F. Bozso, L. Muehlhoff, M. Trenary, W.J. Choyke, and J.T. Yates, Jr., J. Vac. Sci. Technol.
A2, 1271 (1984). Aremco 516 ceramic cement is used. Obtainable from Aremco Products,
Inc., P.O. Box 429, 23 Snowden Avenue, Ossining, NY 10562-0429; the cement should be
preserved in the refrigerator and has a limited shelf life
29. Contributed by Dr. Victor M. Bermudez, Naval Research Laboratory, Code 6863,
Washington, DC 20375-5347
30. Developed in author’s laboratory
31. A. Stevanovic, M. Büttner, Z. Zhang, J.T. Yares Jr, J. Am. Chem. Soc. 134, 324 (2012)
Chapter 31
Calorimetric Heats of Adsorption—Single
Crystals
There are many methods for measuring the energy required for desorption of a
molecule from a single crystal surface using either isosteric heat measurements
under equilibrium conditions, or using desorption kinetic methods. The interpre-
tation of the isosteric measurements requires perfect reversibility in adsorption and
desorption, a condition not met when compound formation or dissolution of the
adsorbate in the crystal occurs. The interpretation of the kinetic measurements are
beset with difficulties having to do with the inability to write the desorption rate
laws that apply with accuracy. The calorimetric method avoids these difficulties by
directly measuring the heat evolved upon adsorption and by using only the first law
of thermodynamics.
In order to employ a calorimetric device, the heat capacity of the calorimeter
must be made very small in order for a measurable temperature rise to be produced.
This has been cleverly done, using single crystal films of metals which are only
2000 Å in thickness [1], using apparatus as shown in Fig. 31.1. These films are
grown epitaxially on an alkali halide crystal surface, and the continuous metal film
is then lifted off of the salt crystal as it dissolves in water. The metal crystal film is
transferred and attached to a ring, made of the same metal, and may be placed at the
focus of a pulsed molecular beam of the adsorbate. The thin crystal may be cleaned
in UHV by ion bombardment and by heating in vacuum or in reactive gases used
for cleaning. Standard surface analysis and LEED studies are routinely performed
on the thin crystal after cleaning. As the gas pulses arrive and are adsorbed on a
2-mm-diameter spot, the temperature of the thin crystal rises, and the pulsed
emission of infrared radiation is measured from the rear of the crystal by focusing
the radiation on a mercury cadmium telluride detector held at liquid nitrogen
temperature. Temperature rises on the crystal per pulse are below 0.1 K for
chemisorbing gases, where each pulse delivers about 0.01 mL of chemisorbed
species to the surface. The emissivity of the backside of the crystal is significantly
enhanced by initially coating it with amorphous carbon from a candle flame. For
calibration, a pulsed laser beam of known energy per pulse is used, and its
reflectivity at near normal incidence is measured. The temporal character of the
laser beam is matched to that of the pulsed molecular beam in the calibration
measurements. The energy losses after each pulse are found to be due to radiation
losses only.
In addition to the errors in measuring the small pulses of heat, errors in the
coverage measurement (and the sticking coefficient) currently limit this method [1].
The sticking coefficient and absolute coverage are measured using the method of
King and Wells [2], which measures the intensity of the reflected gas from the
crystal during each pulse. A gold flag is employed in these measurements.
A stagnation detector (see p. 429), which is a Bayard-Alpert gauge with a small
aperture, is employed for measurements of the absolute beam flux, as discussed on
p. 314.
available from Goodfellow. The PVDF detector is bent into an arch and gently
pressed to the back of the crystal which has been thermally cleaned in UHV before
contact. Pulses of adsorbent gas molecules, arriving at 0.5 Hz, cause temperature
bursts due to the evolution of the adsorption energy. The transient rise of crystal
temperature causes part of the PVDF detector to change its degree of polarization,
releasing an electrical charge through an electronic circuit. The ease of using the
PVDF detector compared to others allows the crystal to be cleaned at high tem-
peratures and also allows calorimeter operation below 300 K. This permits a wide
range of heat-of-adsorption experiments to be performed on thicker (*75 µm)
single crystals than previously possible.
The latest advances in the technique are sketched in Fig. 31.2A–D. It has been
found that noise, caused possibly by electrical leakage at the PVDF edges, can be
reduced by etching the Al film coating away from the ribbon edge. Figure 31.2 A
shows this: (a) indicates the aluminized PVDF film; (b) shows masking of the Al
Fig. 31.2 Improved adsorption calorimeter for single crystals at 100–350 K. A Electrodes on
PDVF front and back surfaces. B Bent PDVF sensor. C contact of sensor with thin sample.
D Experimental adsorption measurement
336 31 Calorimetric Heats of Adsorption—Single Crystals
film on front and back faces of the ribbon with nail polish; (c) shows the
electrochemically-etched ribbon with preservation of Al only under the nail polish
covering; (d) shows the final assembly with the removal of the nail polish with
methanol.
Figure 31.2B shows a schematic of the assembly for holding the bent PVDF
detector and for making electrical contact to both faces. A copper bar of thickness t,
of the order of a few mm, is used to define the width of the arch and the projection
of the arch. For contact with the crystal, the arch is compressed reproducibly by
0.5–1 mm.
Figure 31.2C shows a schematic of the design to compress the arch of the PVDF
detector against the back of the sample. Self-aligning conical holders are driven into
contact, and care in design prevents appreciable bending of the thin crystal sample.
Figure 31.2D shows a series of three experimental measurements of the heat
evolution during the adsorption of cyclohexene on Pt(111). Pulses of gas, involving
1.1 % of a monolayer are incident. As the coverage increases for the chemisorbed
cyclohexene, the heat of adsorption per pulse drops due to site blocking as well as
other effects. At 0.25 mL, the monolayer is complete and second layer adsorption,
with a constant heat of adsorption, takes place. The raw calorimetry data are shown
in the inset, and repeated experiments show a pulse-to-pulse standard deviation of
about 1.3 kJ/mol. The thermal energy released per pulse is 0.24 µJ which corre-
sponds to a temperature rise in the adsorption spot on the crystal of *7–8 mK. The
detector temperature rises in each pulse by 0.3 mK. The use of a pulsed He–Ne
laser on the crystal allows a calibration of the PVDF detector to give energy
liberated at the crystal surface. The improvements shown here have yielded a * 40
fold improvement in the pulse-pulse standard deviation over earlier PVDF detec-
tors. The paper [3] describes future developments to further enhance detection
sensitivity.
References
1. C.E. Borroni-Bird, D.A. King, Rev. Sci. Instrum. 62, 2177 (1991)
2. D.A. King, M.G. Wells, Surf. Sci. 29, 454 (1972)
3. W. Lew, O. Lytken, J.A. Farmer, M.C. Crowe, C.T. Campbell, Rev. Sci. Instrum. 81, 024102
(2010)
Chapter 32
Surface Debye Temperature
The Debye temperature of a bulk solid is a measure of the stiffness of the bonds
within the crystal and is related to the Debye characteristic frequency, νD. This
frequency corresponds to the highest allowed vibrational mode in the crystal in the
distribution of frequencies following a mode frequency distribution function with a
magnitude which is proportional to the square of the mode frequency. The Debye
temperature is HD ¼ hmD =k, where k is Boltzmann’s constant.
The Debye-Waller factor, 2M, is connected to the Debye temperature through
(32.1), and determines the intensity of scattering of an electron beam incident on the
crystal through (32.2).
2M ¼ 24me E cos2 a þ V0 T=ma kðHD Þ2 ð32:1Þ
where ma = mass of the atoms in the crystal, me = mass of the electron, α = the angle
of incidence, V = the inner potential, and E is electron energy.
For an electron beam incident on a crystal, in the kinematic approximation,
I ¼ I0 e2M ð32:2Þ
involved in the diffraction process, and the measured Debye temperature decreases
as a result of the looser binding of the crystal atoms near to the surface.
Figure 32.1b shows a Debye-Waller plot of lnðI Ibk Þ=I0 versus temperature for
various values of ΔK, the perpendicular component of the momentum transfer
between the electron and the metal atoms in the crystal. Here, Ibk is a background
scattering contribution which can either be measured in the vicinity of the 0,0 beam,
or treated parametrically as a variable. In this case, for low angles of incidence, one
can only measure an effective perpendicular Debye temperature as ΔK is varied [2].
Figure 32.1c shows typical behavior of the Debye temperature measurements for
three angles of incidence on a Rh(111) crystal as a function of ΔK. It may be seen
that all of the measurements show that as ΔK increases, and the depth of sampling
of the crystal by the electron beam increases, the measured value of ΘD increases to
a plateau at ΘD = 370 K, which is characteristic of the bulk of the crystal. The lower
limit of ΘD is near 200 K, and is characteristic of the surface metal atoms.
These measurements were made with a LEED apparatus mounted between two
pairs of Helmholtz coils to cancel the earth’s magnetic field. The values of the
diffracted beam intensity were measured as the crystal slowly cooled to avoid the
serious influence of the magnetic field associated with the electrical heating current.
The Faraday cup aperture subtended an angle of 4.6 × 10−3 steradians, and was
large enough to collect the entire beam, I.
References
1. C.M. Chan, P.A. Thiel, J.T. Yates Jr., W.H. Weinberg, Surf. Sci. 76, 296 (1978)
2. G.A. Somorjai, H.H. Farrell, Adv. Chem. Phys. 20, 215 (1971)
Chapter 33
Friction and Fracture
Fig. 33.1 UHV tribometer for measuring the coefficient of friction. a Strain gauge arrangement.
b Forces at slider/sample interface and displacement components
Fig. 33.2 Pin-on-disk measurements—UHV. a Exploded view. b Forces and electrical contacts
Fig. 33.3 UHV fracture stage for surface analysis after fracture. a UHV fracture device. b Load
cell for accurate operation in UHV. c Force/displacement plots for polyester resin
References
scaling and isolation to permit the process controller to control the power output.
An input of 1–5-V DC produces a 0–50-A DC output.
The process controller [3] (Fig. 34.1a), was selected as the primary control
device for the following reasons: (1) the ability to program the readouts in arbitrary
units; (2) accuracy (±0.05 %, full scale); (3) speed (updates 3 times/s); (4) digital
input of tuning parameters.
An illustration of the performance of the programmer is shown in Fig. 34.1c.
The programmer tuning constants were adjusted to produce the linear temperature
versus time performance from about 95–520 K as shown. The temperature stability
at 272 K is shown, where fluctuations of only ±0.35 K were observed. Scan rates
from about 2–8 K/s may be programmed into the controller.
In order to achieve optimum performance, appropriate values of the process
controller constants (gain, rate, and integral time) need to be established. These
three control constants are dependent on sample properties such as mass and
resistivity and on the sample-mounting configuration. Initially the rate and integral
time can be set to zero while an optimum gain is being established. An optimum
gain value is one which minimizes oscillations in the sample temperature, both
during the heating ramp and while the crystal is being maintained at constant
temperature, and the optimum gain may differ slightly for the two operating con-
ditions. It is also necessary to optimize the gain such that the difference between the
ramp voltage and the measured thermocouple voltage is minimized during the
heating ramp. The other control constants can then be adjusted to fine tune the
operation of the controller. Tuning of the integral time is essential, particularly
when trying to avoid an overshoot at the end of temperature ramp.
Another electronic controller, designed to correspond to the system’s thermal
response and capable of programming at 20 K/s without overshoot, has been
described [4].
Almost all of the studies of adsorption, desorption, and reactivity at surfaces involve
control of the temperature of the surface as well as control of temperature pro-
gramming. A traditional solution involves specialized controlling units, one of which
is mentioned on p. 349. These controllers provide a technically simple but rigid
solution for the temperature control. An alternative solution employs an acquisition
board in a personal computer controlled by a program written in any programming
language. This method provides great flexibility and can be used for any type of
temperature sensor and any desired temperature program. A graphical programming
language, LabVIEW [5], is especially appropriate for these tasks. It has a powerful
graphical interface, a large library of subroutines for data acquisition, and a variety of
mathematical operations. An example of a virtual instrument created by LabVIEW
for temperature measurement and control of a silicon crystal is described below.
352 34 Heating Samples
Figure 34.2a shows a block diagram of the measurement setup for a silicon
crystal having a room temperature resistivity up to 100 Ω cm. The crystal tem-
perature is measured by a thermocouple that has a known relationship to the true
temperature (measured by pyrometer) as a function of temperature. A cold junction
compensator [6] is used to establish, electronically, a reference voltage corre-
sponding to the ice temperature, 273.2 K. The signal is fed to a separating amplifier
and amplified by 50–100 times, which minimizes the noise level received at the
acquisition board. A second purpose of the separating amplifier is the rejection of
the DC offset caused by the potential applied to the crystal for heating. The
acquisition board can only handle 10–20 V offsets, which would be normal for
metal crystals because of their higher conductivity compared to a semiconductor
crystal. The signal from the separating amplifier is fed to the acquisition board [7],
which has an analog input/output with a maximum input and output voltage of
10 V. Two DC power supplies are needed, one to power the separating amplifier
(±5 V) and a heating supply (60 W, 120 V, and a maximum current of 50 A).
Figure 34.2b shows a block diagram of the LabVIEW controlling program that
controls the heating power supply according to a predetermined program. A sample
of a complex heating/cooling program and the result is shown in Fig. 34.2c. This
program provides the desired heating ramp from −200 up to 850 °C, and then
cooling by various ramp functions, as shown on the front panel readout in
Fig. 34.2c. The lower right-hand side of the front panel readout shows the actual
program achieved compared to that desired.
Figure 34.2d shows a detailed view of the linearity of the heating ramp and the
error that was measured. Here, P = 3 K/s, the linear ramp function. The lower
difference plot shows the accuracy of this particular ramp and all of the data points
fell in a 1.5 K noise range up to about 1000 K, where the mechanical contact (see
p. 356) between the thermocouple and the Si crystal began to fail.
The ability to control the isothermal temperature of the crystal for long periods
of time is also excellent, with better than 1 K temperature constancy and error
spread. There is often an initial overshoot of the order of several K as one ramps up
to a desired temperature, and this can be minimized by adjusting the proportional
integrating differentiating (PID) parameters in the program, Fig. 34.2b.
The advantages of the LabVIEW approach to temperature programming are
listed below:
1. Flexibility—the temperature profile can be changed as desired and nonlinear
programs are possible. Any type of temperature sensor with a non-linear output
can be used (see p. 405). Any type of correction for errors, such as the error in a
thermocouple because of inefficient thermal contact (see p. 356), can be made.
2. High speed.
3. Universal interface—the acquisition board with multiple inputs and outputs cou-
pled to the LabVIEW program, can be used for computer control of other instru-
ments (e.g., a mass spectrometer). However, the interface cannot be used at the
same time with another acquisition board driven by a program working in DOS.
4. Convenient graphical programming with a large library of subroutines.
34.2 Temperature Programming of Metal and Semiconductor Crystals 353
Fig. 34.2 Temperature programming of metal and semiconductor crystals. a Block diagram—
resistive heating control. b Block diagram—LabVIEW controller logic. c Control parameters for
temperature program. d Linearity and error in heating ramp
354 34 Heating Samples
Crystals such as Ag, Al, Au, and Cu possess very high thermal conductivity and
cannot easily be spot-welded to other metals because of the lack of melting at the
low temperatures achieved at the junction during welding. The mounting of such
crystals to heating/cooling wires requires a special method that will provide good
mechanical, thermal, and electrical contact over many cycles of heating and cool-
ing, and a design for this is shown in Fig. 34.3.
The crystal holder is fabricated out of OFHC copper, sapphire, and tungsten.
These materials all possess optimum thermal conductivity properties below 100 K.
Fig. 34.3 Mounting nonweldable crystals for resistive heating and cooling
34.3 Mounting Nonweldable Crystals for Resistive Heating and Cooling 355
Figure 34.3 shows a scheme to mount such crystals. A slot is carefully cut into
two opposite edges of the crystal. The slot makes a tight sliding fit for the tungsten
wire heating lead, and the crystal should not be distorted to make this fit.
The tungsten heating/cooling leads are fabricated with a pair of U-loops that
remove the effects of linear expansion that otherwise would push the wire from the
slot. Thus, the crystal, incorrectly mounted as shown in the right-hand insert, is
subject to variation in the cooling and electrical contact as the wire is heated and
cooled owing to bowing of the wire in the slot. The looped wires on the other hand
undergo thermal expansion effects which are almost counterbalanced on either arm
of the U-loops.
Make the assembly as follows using gloved fingers:
1. Mount the preshaped W wire (0.38-mm diameter) on the lower set of Ta screws
and tighten.
2. Fix the preshaped upper W wire on the right hand screw and do not tighten the
screw.
3. Fix the crystal with its thermocouple (hole-mounted—details not shown) on the
lower wire.
4. Rotate the upper wire counter clockwise until it fits into the slot and then tighten
both screws carefully.
This mounting procedure produces a stable junction over many cycles that
permits cooling of the crystal to about 32 K when liquid He is used in the reservoir
intended for liquid nitrogen [8].
One of the most difficult tasks in the laboratory is uniformly heating silicon single
crystals by resistive heating and measuring the temperature of the crystal. A variety
of methods of making electrical contact to silicon crystals have been devised that
34.5 Heating Silicon Crystals and Temperature Measurements 357
involve attachment to metal springs [10, 11] or elastic metal sheets [12, 13]. These
contact methods often result in the formation of hot spots at certain regions of contact
and can lead to nonuniform heating. Ta and Mo are used in these applications, since
other metals can lead to metal contamination of the silicon crystal, either through
metal diffusion through the bulk (Ni; Fe) or by transport by vaporization.
The attachment of thermocouples is also a difficult task owing to metal diffusion
or metal vaporization. Here, too, Ta-protective envelopes must be used, as in the
design of Bozack et al. [13], which employed Ta-protected W26 %Re/W5 %Re
thermocouples inserted into slots cut into the crystal edge [13]. We have found that
chromel-alumel (type K) thermocouples are unsatisfactory owing to the high vapor
pressure of Ni (10−5–10−4 Torr at 1450–1500 K) [14]. A few atomic percent of Ni
produces, on Si(100), the well-known split-off dimer defects [15]. Thus, W26 %
Re/W5 %Re thermocouples, shielded by Ta, have been used successfully in our
own work.
Adhesive thermocouple contacts, such as zirconia-based Ultratem (Aremco),
have been found to diffuse through Si above 1100 K. A Ta thermocouple tab
contact, joined by a layer of Ge that alloys with both Ta and Si was proposed [11],
but adhesion was lost on heating to high temperatures [16].
Figure 34.5a shows a design [17] that addresses the clamping of Si crystals to
provide rather uniform electrical contact and temperature distribution. A Ta sheet
(20 × 9 × 0.125 mm3) is bent as shown in the diagram to form a bracket shape,
which is then welded to a W rod (2-mm dia.), using a copper plate as an alignment
jig, producing the structure shown in Fig. 34.5a(d). Four slips of Si wafer material
are prepared (4.5 × 8.5 × 0.4 mm3) and used to sandwich a Si wafer slice
(23 × 8 × 0.4 mm3) as shown in Fig. 34.5a(e). The entire sandwich structure is
compressed by a flat U-shaped Ta spring that is wedged into the slot between the
top Si slip and the top of the bracket, Fig. 34.5a(f). The fully assembled mounting is
shown in Fig. 34.5a(g).
Figure 34.5b shows a method for attaching a W26 %Re/W5 %Re thermocouple
(0.08-mm dia.) to the crystal by a U-shaped clip design [17]. To prevent direct
contact of the ther-mocouple material to the crystal, it is shielded with a Ta tube
(0.19-mm inner dia., 0.06-mm wall thickness, 5-mm length). The prewelded ther-
mocouple is inserted into the Ta tube at the end, which has been previously cut off
at an angle [Fig. 34.5b(a, b)], and the tube is flattened [Fig. 34.5b(b)], and formed
into a bracket shape that tightly clips over the crystal edge [Fig. 34.5b(c)]. The
thermocouple exits from the angled end and is bent upward as shown in Fig. 34.5b
(c). The thermocouple voltage is converted to temperature using a calibration
method similar to that of Smentkowski et al. [18], as discussed on p. 405. A cold
junction compensator is used for the reference junction at 273 K.
Two designs for thermocouple attachment to Si crystals have been compared
using a calibrated optical pyrometer. Figure 34.5c shows the two designs where the
slotted crystal design from Bozack et al. [13] and the design shown here [17] are
compared. The deviation for the slotted design is from 50 to 150 K, whereas the
deviation of the U-shaped clip design is from 50 to 90 K over the same temperature
range.
358 34 Heating Samples
The uniformity of heating of the Si crystal with this mounting design is found to
be ±15 K at 1400 K, as measured by an optical pyrometer. At temperatures above
about 1100 K, temperature oscillations are sometimes observed with the U-shaped
clip thermocouple.
Great care must be exercised in making the mount and the thermocouple clip. All
Ta parts must be boiled in concentrated HCl solution to remove Ni contaminants
[15]. All tools should be made of plastic to avoid metal contamination, and fabri-
cation should be done under clean conditions without contact to metals other than
cleaned Ta. In addition, Mo parts, which are more springy, may be substituted for Ta
parts, although the HCl cleaning procedure followed by STM studies has not been
done for crystals with Mo/Si contacts, as has been done for W/Ta contacts [15].
The use of interferometry for measuring the temperature changes in semicon-
ductor crystals has been reported [19]. Here, both temperature dependent changes in
the refractive index and in the thickness of the crystal wafer cause the optical path
in the wafer to change in comparison to λ/2n, where λ = the laser radiation
wavelength, and n = the refractive index. Temperature changes as small as 0.2 K
have been detected, but one must start at some known temperature in order to make
temperature measurements at other temperatures.
Single crystals of metals may be heated in several ways for studies at constant
temperatures. Unfortunately the presence of large AC or DC currents used for
heating and the concomitant presence of stray electric and magnetic fields will be a
severe disturbance to measurements involving electron currents to or from the
crystal. A general method used to avoid this problem involves pulsing the heating
current and making current measurements from the crystal during the time when the
heating pulse is off.
Figure 34.6a shows a typical mounting for a metal single crystal on a manipu-
lator using ohmic heating for the crystal. Large-diameter (1.0–1.5 mm) tungsten
support leads are connected to current and cooling leads on the manipulator and
conduct current to a pair of tungsten wires (0.3-mm diameter) welded to the back of
the crystal. Tungsten is preferred to tantalum because of its higher coefficient of
thermal conductivity at low temperatures. In this configuration the crystal is heated
by thermal conduction from the hot wires rather than by ohmic heating of the
crystal itself, and at incandescence, it will be seen that the parallel wires are
comparatively bright in the region between the crystal and the support leads. The
uniformity of welding to the crystal and the support leads can be judged by the
relative brightness of the four intermediate regions. Poor welds will yield higher
brightness and will possibly cause premature burnout. The thermocouple welded to
the crystal edge is used to measure the crystal temperature.
360 34 Heating Samples
Fig. 34.6 Avoiding extraneous electrical effects owing to heating crystals. a. Interrupted ohmic
heating. b Interrupted electron bombardment heating
heating depend upon the magnitude of the heat leakage to the support wires through
the thinner wires welded to the crystal.
Three circuits for interrupted heating of single crystals have been reported in the
literature [20–22]. For ohmic heating, at 1000 K, the temperature controller [20]
required a duty cycle of 2/3. The switching rate was approximately 0.3 Hz, and the
temperature regulation was ±15 K. Better control would be possible at higher
switching rates [20]. For electron bombardment heating, a second circuit has been
published in which a mechanical relay is used [21]. Here, bounce within the relay
limits the time period for data acquisition to about 40 ms at 10 Hz. It is necessary to
switch both the filament heater current and the bias voltage and since the power
supplies are slow to recover from the switching, dummy loads are used in the off
cycle. A difficulty reported for switching large currents and high voltages is that
transients are produced that disturb electronic measuring equipment being used.
Proper shielding of the elements of the circuit and the use of shielded cables
eliminates these effects. A third circuit, which has shown favorable utilization in
conjunction with LEED, Auger, and HREELS measurements is reported in [22].
The circuits shown in [20, 22] employ older electronic components, and the user
should consult with an electronic specialist for designing similar circuits using
modern components.
Electron bombardment heating of a metal crystal can often be used to bring the
crystal to higher temperature than can be done conveniently by ohmic heating.
Thus, for the highly refractory metals such as W, Ta, Mo, and Re, all of which
require very high temperatures to remove impurities, electron bombardment is often
used (see also p. 363.)
A typical design for electron bombardment heating, adapted from [23], is shown
in Fig. 34.7. The crystal is supported on a manipulator by one or more welded wires
that connect to a stub on the end of the manipulator. This stub is electrically isolated
from the manipulator shaft by a high-voltage ceramic spacer that will work at
positive bias potentials of several kilovolts. The high-voltage connection to the stub
is provided by a well-insulated wire that passes through a ceramic sheath, and this
sheathed wire is firmly attached to the manipulator shaft by means of a ceramic
support designed to hold it tightly, and in isolation from the rotary manipulator
shaft or any other grounded parts.
The very simple electron bombardment gun is made of a stainless steel tube with
an aperture cap placed over the open end, and a custom made ceramic base to hold
the heavy filament leads. Filament current is supplied through copper braid that is
connected to the ceramic support, and thence to flexible power leads that connect to
feedthroughs on the manipulator flange. The electron gun is mechanically con-
nected to the ceramic support by a rod, as shown.
362 34 Heating Samples
The filament for the electron gun can be made of pure W wire, or thoriated W
wire (see p. 133). Emission currents of 10–50 mA with a positive bias of 1 kV on
the crystal will supply 10–50 W, usually enough for heating to the desired tem-
perature, depending on the thermal conductivity of the welded leads to the crystal
and the radiation loss at the desired crystal temperature. For gentle heating, only
radiation heating from the filament can be used, and in this case a non-thoriated W
filament would be preferable. While the winding of W filaments is easily done,
34.7 Electron Bombardment Crystal Heating 363
The use of indirect heating of samples is especially important for materials that are
nonconductive and therefore cannot be heated by ohmic heating. However, indirect
heating of conductor materials is also a favored method of heating, since excellent
control of temperature can be achieved by indirect heating methods. Both heating
by radiation and heating by electron bombardment are termed indirect heating,
since neither method depends on the electrical resistance of the sample, as does
direct ohmic heating.
Figure 34.8a shows a typical method for indirectly heating a sample using a
button heater [25] that is supplied commercially [26]. The sample is held onto the
top surface of the Mo cylindrical shell that surrounds the W heater wire. The W
heater is immersed (potted) in an aluminum oxide powder. In the case shown here,
the sample is held in place by means of a Mo sleeve that fits snugly over the Mo
cylinder and has a lip that grips the front surface of the sample. This arrangement
may be used up to about 1500 K without damage to the heater assembly. Since the
heater wire is noninductively wound, electron spectroscopic measurements may be
made with heating power on [25].
364 34 Heating Samples
Fig. 34.8 Indirect sample heating. a Conduction heating (T ≤ 1500 K). b Radiation and
e-bombardment heating (T ≥ 1500 K). c Static heating—config. B d Programmed heating and
cooling—config. B. e Ohmic support heating
Figure 34.8b shows another arrangement that maybe used at temperatures above
1500 K, but which also is useful at lower temperatures [27]. A Mo cylindrical can is
polished flat on its top surface, and the sample is placed in contact with the upper
surface. In the case shown here, a single crystal of diamond is held by two tungsten
or tungsten alloy wires by means of parallel slots in two opposing edges. The wires
are spot-welded to the edges of the can and tightly bind the diamond to the polished
top of the can. Inside the can is a flat W pancake spiral made by winding a W wire
onto a mandrel into a watch spring configuration as a tight coil. No aluminum oxide
potting powder is present, so that the Mo can is heated only by radiation from the W
spiral, or by radiation plus electron bombardment if more power is needed.
A reflective radiation shield behind the Mo can assists in increasing the efficiency of
heating the Mo support by the incandescent filament. A more complex electron
bombardment heating device involving a shielded thermionic emitter has been
reported [28].
The performance of the heating assembly in Fig. 34.8b has been measured for a
diamond single crystal [27], and the results are shown in Fig. 34.8c, d. In Fig. 34.8c
the temperatures of two thermocouples, buried in laser-drilled holes in the edge of
34.8 Indirect Sample Heating 365
the diamond crystal, are shown for static heating experiments, where the assembly
is brought to steady temperature in ultrahigh vacuum. The two thermocouples differ
by only 8 K in these experiments, indicating that the true temperature of the
diamond is being measured. However, at a diamond temperature of 1400 K, the
difference between the Mo support temperature and the diamond temperature
amounts to 140 K. This difference decreases at lower temperatures. Since diamond
is an excellent thermal conductor, it is likely that heat transfer from the Mo surface
to the diamond back surface is the limiting factor in these experiments. Similar
temperature differences between the temperature of a heated supporting surface and
a contacting sample have been measured by others at about 10−3 Torr pressure. The
sample was contacted to the supporting surface using a silver paint that worked up
to 1073 K, where a 22 K temperature difference was measured [29].
Figure 34.8d shows the difference between the heating behavior of the diamond
crystal and the Mo support can under conditions of temperature programming at a
rate of 1.3 K/s. At a diamond temperature of 1150 K, a difference of temperature
between the Mo support and the diamond of 225 K was noted (compared to only
90 K under static conditions as shown in Fig. 34.8c). The Mo can cools faster than
the diamond when the heating is interrupted, as would be expected for cooling
controlled by thermal conduction from the assembly.
The maximum filament heating power used to heat the diamond may be reduced
by employing a mixture of radiation heating combined with electron bombardment
heating. Thus, for example, a positive bias of 250 V produces thermionic heating of
12 W at a diamond temperature of 1273 K, whereas 47 W of filament heating power
was needed under conditions of zero bias to achieve the same temperature.
A successful variation of this method, involving ohmic heating of the support
assembly [29] is shown in Fig. 34.8e. Here, a polished Ta block of the same size as
a TiO2 single crystal is heated by means of W wires which pass through the body of
the Ta. The temperature of the crystal is measured with an embedded thermocouple
cemented [30–32] into a diagonal slot in the crystal corner, and programming of the
heating power is controlled by this thermocouple which acts to supply a feedback
signal. Somewhat similar methods for heating and cooling alkali halide crystals,
and using an In metal film as a heat transfer medium have been reported [33].
used for resistive heating. A thermocouple is attached to the Ta film for the mea-
surement of the temperature.
The indium film is applied conservatively so as to not cause a short circuit
between the Ta clip leads that support the sapphire plate on 0.75-mm Ta support
leads. The indium film application may be easily done by using the Ta film resistor
to raise the sapphire temperature above the melting point of In (157 °C). The Ta
resistor film requires 1.5–2 A at 8–10 V to heat the unit to about 900 K. The GaAs
sample is grounded by using a Ta spring contact that is welded to a Ta clip.
The disadvantages of this technique are that the In reacts slowly with GaAs,
causing failure after long heating. This effect can be minimized by using only small
amount of In. In addition, at high temperatures, the In has a significant vapor
pressure (10−7 Torr at 833 K; 10−6 Torr at 895 K).
A number of methods for mounting semiconductor crystals in ways where they can
be electrically heated and cooled are discussed throughout this chapter beginning on
p. 349. In the case of crystals which cannot be heated by electrical conduction,
either because of unfavorable conductivity or because of unfavorable shape, it is
possible to use a refractory cement that holds the crystal to a tungsten disk that is
heated through welded contacts to refractory metal heating wires [38, 39].
In Fig. 34.11, an irregularly shaped SiC crystal has been cemented to a W disk
using Aremco Ultratem 516 cement, which has a ZrO2 base [40]. The W disk has
previously been welded to a pair of Ta heating wires that pass current from the
heavy W (1.5-mm-diameter) support leads. The cement is made electrically con-
ductive by mixing with the appropriate amount of powdered graphite. This permits
the electron spectroscopies to be used without sample charging.
A chromel-alumel thermocouple is connected to the back side of the SiC crystal
by means of an alloy solder (90 % Au/10 % Ta) that anchors a tiny Ta tab to the
crystal. The connection to the SiC crystal must be done in a preparative vacuum
chamber to avoid gross oxidation of the crystal during induction heating.
Measurements with AES and XPS indicate that no migration of the ceramic
cement occurs to the front of the SiC crystal up to 1400 K. The cement is adequate
for mounting light crystals, but it is by no means robust.
At 1000 K, the temperature indicated by the thermocouple may be about 100 K
too high compared to the true SiC temperature. The temperature range from 85 to
1400 K was employed, and minimal outgassing occurs at 1400 K [38].
34.12 Low-Stress Mount for Fragile Semiconductor Crystals 369
Single crystals may be heated by radiation from hot filaments mounted internally in
the ultrahigh vacuum system, and this method possesses the best efficiency of
coupling the radiant source (usually a hot W wire) to the crystal. For special cases,
however, it may be preferable to employ radiation heating from external tungsten
halogen lamp sources, which typically have maximum power limits in the 250–
1000-W range. Usually the maximum temperature that can be reached with an
external lamp is insufficient for cleaning the crystal and electron bombardment or
ohmic heating methods must be employed for initial cleaning.
A very early use of external radiation heating for programming the temperature
of semiconductor crystals can be found in the work of Palmberg and Peria [42]. In
Fig. 34.13a, their glass ultrahigh vacuum apparatus is shown. A track allows the
translation of the crystal, using a magnetic device, from location to location in the
linear vacuum system. Shown is a 200-W halogen lamp inside a spherical reflector
(silvered glass flask section). Figure 34.13b shows a variation on the design of
Fig. 34.13a, where a commercial 750-W lamp (operating at reduced power) is
housed in an ellipsoidal reflector, which has reflective aluminum focusing surfaces
behind and in front of the lamp [43]. This arrangement permitted crystal temper-
atures near 1000 K to be achieved using the lamp alone, and cooling of the
hemispherical glass surface was necessary to prevent damage to it by the
lamp. A nonlinear heating rate was observed. The crystal was mounted on a
0.25-mm-diameter W support wire, which minimized cooling by conduction to the
liquid-nitrogen-cooled support lead. A tungsten thermionic emitter is shown behind
the crystal and this is used for electron bombardment cleaning at 2500 K, assisted
by magnetic collimation. Figure 34.13c shows a lens arrangement for focusing light
from a 900-W lamp [44]. The single-glass double-convex lens, f/1.32, has a
magnification of unity. Lenses of this type are used as condenser lenses in pho-
tographic enlargers. The window is Pyrex glass with 95 % transmission in the near
infrared, where the spectral radiancy is greatest for tungsten. A silicon crystal could
372 34 Heating Samples
Fig. 34.13 Radiation heating of crystals through glass windows. a Hemispherical reflector.
b Ellipsoidal reflector. c Lens + hemispherical reflector. d Temperature program for C
be raised to 900 K with this arrangement. The temperature program using the lamp
in Fig. 34.13c is shown in Fig. 34.13d, and the reproducibility over a period of
several days is within ±5 K.
Fig. 34.14 Pyrolytic graphite heating elements. a Heater design—current flow in C direction.
b Design for internal heating element—metal. c Resistivity in C direction
Resistive sections are separated from each other using a band saw or preferably a
diamond saw wheel. Care must be taken not to fracture the graphite during sawing.
The heater is mounted on threaded Mo connection rods, as shown in Fig. 34.14a.
These rods are not heated too much since the thermal conductivity along the
C-direction is small compared to the other directions. When the temperature in the
central region is in the 2000 K range ±30 K, typical power dissipation is about
2.0 kW for 155 V applied with 13.5-A current.
Figure 34.14b shows the heater used as in insertion device inside of a block to be
heated to about 1200 K. The pyrolytic graphite heater was mounted in a rectangular
374 34 Heating Samples
hole and was insulated from the metal by an alumina spacer. Electrical contact was
made with a tungsten foil spring attached to the bottom of the heater as shown.
Although the heater is mechanically fragile, once in operation it may be
repeatedly cycled in temperature without damage. The authors report that this
design is compatible with ultrahigh vacuum.
The resistivity of the pyrolytic graphite heater was measured versus temperature,
as shown in Fig. 34.14c. Relatively poor agreement was obtained compared to the
manufacturer’s specifications. This level of variation is typical for CVD pyrolytic
graphite measured in the C-direction.
A simple resistance heater which can be painted on an insulating surface [47] may
be useful in indirectly heating components in an ultrahigh vacuum system (as
shown in Fig. 34.15). A thin layer of Aquadag [48], which is a colloidal graphite
suspension, is painted onto a roughened quartz surface and outgassed for 0.5 h by
baking in an oven at 300 °C, which is reported to be satisfactory for complete
outgassing [48]. The graphite layer can then be smoothed with fine abrasive paper.
Contact to the graphite layer is made using Ni or Pt strips, 0.1 mm thick, which are
sandwiched between the graphite layer and a second quartz plate that is clamped to
the first to assure good contact.
For a 27-mm-diameter plate, typical room temperature resistances are about
15 Ω. With a variable transformer, up to 20-V maximum, the heater could be used
continuously for long periods at temperatures of 500 °C, and operation at higher
temperatures may be possible. The application of these ideas to heating irregularly
shaped surfaces is possible.
A cylindrical single crystal exposes different single crystal planes at different angles
around its perimeter. Thus experiments which sample small areas may be carried
out at different azimuthal angles, sampling different single crystal facets as well as
at different atomic step geometries and densities on vicinal planes. The crystal
geometry shown in Fig. 34.17a features a crystal shell with inside diameter of
14 mm and outside diameter of 20 mm, and involves a crystal length of 14 mm. The
crystal is clamped to the bottom of a cryogenic reservoir containing liquid N2,
which with long cooling times can bring the crystal down to 89 K. The crystal may
be heated for cleaning to above 900 K by radiation from the encased tungsten
filament or by electron bombardment. Cooling occurs at an initial rate of
*45 K min−1, with the lowest temperature being achieved in about 60 min cooling
376 34 Heating Samples
Fig. 34.17 Rotatable single crystal. a Structure of cylindrical single crystal. b H2 dissociation
probability versus atomic step density
34.17 Rotatable Single Crystal 377
Tungsten lamps are excellent clean radiation sources for the interior of UHV sys-
tems. They may be of the quartz-halogen or quartz-Xenon type where the added gas
prevents W evaporation on the inner surface of the quartz envelope. UHV com-
patibility is related to having only quartz and metal exposed to the vacuum system.
Such bulbs are used in home lighting systems, auto headlamps, and projector lamps.
They have the advantage of being able to operate at high power without filament
burnout or W evaporation into the vacuum system. Also, there is no thermionic
electron emission. Gases in the bulb will not escape into the vacuum system. These
bulbs have been used also for passing focused radiation through windows for
heating of samples in UHV (see Sect. 34.13).
A particular use of a lamp for heating an evaporant is shown in Fig. 34.18. An Al
or Cu body, containing a glass vial with the evaporant, is heated from behind with a
lamp encased in the body as shown. The lamp is carefully shielded by aluminum
metal shields to prevent radiation heating of an insulating PTFE support or other
parts of the support assembly. Thermocouple wires may be attached to the evap-
orator, but electrical pin contacts to thermocouple wires should be shielded from
radiation to prevent heating of junctions which will supply spurious voltages. The
design shown has been used up to 300 °C [51].
References
1. R.J. Muha, S.M. Gates, P. Basu, J.T. Yates Jr., Rev. Sci. Instrum. 56, 613 (1985)
2. Model 2B50B, obtainable from Analog Devices, One Technology Way, Norwood, MA
02062-9106
3. Model UDC 500, obtainable from Honeywell, Micro Switch Division, 11 W. Spring Street,
Freeport, IL 61032-4353. More advanced controllers are now available from Honeywell, such
as the UDC 5000 ultrapro
4. M.A. DeAngelis, A.B. Anton, J. Vac. Sci. Technol. A10, 3507 (1992)
5. LabVIEW is available from National Instruments, 6504 Bridge Point Parkway, Austin, TX
78730-5039, USA
6. Type MJC-C, obtainable from Omega Engineering, Inc., P.O. Box 4047, Stamford, CT 06970,
USA
7. Type Lab-PC+, obtainable from National Instruments, see Ref. [5]
8. J. Ahner, D. Mocuta, R.D. Ramsier, J.T. Yates Jr., J. Chem. Phys. 105, 6553 (1996)
9. K.L. Rostov, Institute of General and Inorganic Chemistry, Bulgarian Academy of Sciences,
1113 Sofia, Bulgaria (private communication). This design is used in the laboratory of
Professor Peter Toennies, Max Planck Institut, University of Göttingen, Stroemungsf,
Bunsenstr. 20, D-37037 Göttingen, Germany
10. R.C. Henderson, R.F. Helm, Surf. Sci. 30, 310 (1972)
11. J.A. Stroscio, W. Ho, Rev. Sci. Instrum. 55, 1672 (1984)
12. K. Jacobi, W. Ranke, J. Electron. Spectrosc. Relat. Phenom. 8, 225 (1976)
13. M.J. Bozack, L. Muehlhoff, J.N. Russell Jr., W.J. Choyke, J.T. Yates Jr., J. Vac. Sci. Technol.
A5, 1 (1987)
References 379
14. Metal vapor pressures are listed and graphed, in Handbook of Electron Tube and Vacuum
Techniques, ed. by F. Rosebury (American Institute of Physics, New York, NY, 1993),
pp. 142–148
15. V.A. Ukraintsev, J.T. Yates Jr., Surf. Sci. 346, 31 (1996)
16. W. Ho (private communication)
17. H. Nishino, W. Yang, Z. Dohnalek, V. Ukraintsev, J.T. Yates Jr., J. Vac. Sci. Technol. A15,
182 (1997)
18. V. Smentkowski, J.T. Yates Jr., J. Vac. Sci. Technol. A14, 260 (1996)
19. V.M. Donnelly, J.A. McCaulley, J. Vac. Sci. Technol. A8, 84 (1990)
20. G.E. Thomas, W.H. Weinberg, J. Vac. Sci. Technol. 16, 87 (1979)
21. G.B. Hoflund, R.E. Gilbert, J. Vac. Sci. Technol. 18, 56 (1981)
22. J.A. Stroscio, L.J. Richter, W. Ho, Rev. Sci. Instrum. 55, 732 (1984)
23. J. Lecante, R. Riwan, C. Guillot, Surf. Sci. 35, 271 (1973)
24. Standard and custom wound W filaments may be obtained from C.M. Furnaces, Inc., 103
Dewey Street, Bloomfield, NJ 07003
25. J. Klebanoff, V.H. Ritz, R.E. Thomas, J. Vac. Sci. Technol. A2, 1396 (1984)
26. Spectra-Mat Inc., 100 Westgate Drive, Watsonville, CA 95076
27. V.S. Smentkowski, J.T. Yates Jr., J. Vac. Sci. Technol. A11, 3002 (1993)
28. P.H. Holloway, J.B. Hudson, Rev. Sci. Instrum. 43, 1045 (1972)
29. T.E. Jones, W.C. McGinnis, J.S. Briggs, Rev. Sci. Instrum. 65, 977 (1994)
30. J.C.S. Wong, A. Linsebigler, G. Lu, J.-F. Fan, J.T. Yates Jr., J. Phys. Chem. 99, 335 (1995)
31. F. Bozso, L. Muehlhoff, M. Trenary, W.J. Choyke, J.T. Yates Jr., J. Vac. Sci. Technol. A2,
1271 (1984). (Here the cement used was mixed with graphite in order to permit electrical
conductivity for electron spectroscopy of a SiC single crystal)
32. A zirconia-based refractory cement was employed which matched the thermal expansion
coefficient of TiO2. This particular cement is Aremco 571. The address is Aremco, P.O. Box
429, Ossining, NY 10562-0429
33. E.P. Marsh, F.L. Tabares, G.A. Bach, J.P. Cowin, Rev. Sci. Instrum. 57, 3134 (1986)
34. J.R. Creighton, Dept 1126, MS 0601, P.O. Box 5800, Sandia National Laboratories,
Albuquerque, NM 87185-0601 (personal communication)
35. J.C. Burns, P.C. Stair, R. Haidle, J. Vac. Sci. Technol. A8, 3886 (1990)
36. Vacseal is made by Space Environment Laboratories, RO. Box 1061, Boulder, CO 80306
37. Metal Finishing Guidebook and Directory (Metals and Plastics, Hackensack, NJ, 1989),
pp. 244–265
38. F. Bozso, L. Muehlhoff, M. Trenary, W.J. Choyke, J.T. Yates Jr., J. Vac. Sci. Technol. A2,
1271 (1984)
39. M.J. Bozack, L. Muehlhoff, J.N. Russell Jr., W.J. Choyke, J.T. Yates Jr., J. Vac. Sci. Technol.
A5, 1 (1987)
40. Obtain from Aremco Products, Inc., P.O. Box 429, 23 Snowden Avenue, Ossining, NY
10562-0429; the cement should be preserved in the refrigerator and has a limited shelf life
41. K.A. Schultz, E.G. Seebauer, Rev. Sci. Instrum. 63, 218 (1992)
42. R.W. Palmberg, W.T. Peria, Surf. Sci. 6, 57 (1967)
43. T.E. Madey, J.T. Yates Jr., in Structure et Proprietes des Surfaces des Solides,
No. 187 (Editions de Centre National de la Recherche Scientifique, Paris, 7–11 July 1969),
p. 155
44. M.J. Bozack, L. Muehlhoff, J.N. Russell Jr., W.J. Choyke, J.T. Yates Jr., J. Vac. Sci. Technol.
A5, 1 (1987)
45. C.S. Pitcher, P.C. Stangeby, G.E. Austin, G.M. McCracken, J. Vac. Sci. Technol. A5, 383
(1987)
46. Pfizer, Inc., 235 E. 42nd Street, New York, NY 10017
47. P. Winsemius, H.P. Lengkeek, Rev. Sci. Instrum. 44, 229 (1973)
380 34 Heating Samples
48. Obtain “Dag 660” from Acheson Colloids, 1607 Washington Avenue, P.O. Drawer 611747,
Port Huron, MI 48061-1747
49. J.R. Conrad, S.F. Home, D.R. Pirkle, R.P. Torti, Rev. Sci. Instrum. 51, 144 (1980)
50. C.L. Juurlink, C. Hahn (personal communication)
51. R.G. Jones (private communication)
Chapter 35
Cooling Samples
Cooling single crystal samples in ultrahigh vacuum below 77 K was at one time
considered to be technically difficult because of the perceived need for cooled
radiation shields for the structure holding the crystal. Recently, excellent cooling
properties have been achieved without a radiation shield using a number of
refinements that have permitted rapid cooling to 6 K, control of temperature to
30–70 mK at 10 K, and measurements with absolute temperature errors of about
0.7 K at 10 K and 1.5 K at 100 K [1].
Figure 35.1 shows the design for efficient cooling [1]. A liquid He cryostat [2]
containing concentric ducts for liquid admission and gas transport away is mounted
on a standard UHV manipulator. While the publication [1] shows the use of OFHC
Cu cooling members, recent refinements have substituted highly polished Ag
members for Cu primarily because the surface of Cu oxidizes slowly in use, giving
a nonreflective surface that is not as efficient as the shiny Ag surface in reducing
warming from the radiation fromthe surroundings [3]. All heat-conducting inter-
faces are made by pressing smooth surfaces together and not by welding, which
produces poor thermal contact. A 10 × 10 × 55 mm3 block of Ag is bolted to the
bottom of the cryostat and holds two 1-mm-thick sapphire plates that are separated
from each other by a gap. Three 2-mm-diameter screws, with ceramic isolation
washers, apply force to hold two Ag electrical conduction leads in good contact
with the sapphire plates. The screws pass through the lower Ag bar in large
clearance holes that assure that no short circuits can occur. Each Ag bar is covered
with a stainless steel force plate to avoid distortion of the Ag owing to the force of
the screws. The two Ag conduction leads have 4-mm-diameter stainless steel power
leads exiting from their back and passing up the cooled cryostat surface. These
power leads are heat-stationed twice each, in series, to the cooled cryostat wall to
reduce heat flow downward to the crystal assembly. The heat stations involve
sapphire electrical isolation junctions, mounted 50 mm apart on the cryostat shield.
The thermocouple (in this case, type K) is welded to the back of the crystal and
the leads pass upward after also being heat-stationed on the cryostat with isolation
by sapphire. Heating rates from 0.01 to 100 K/s are possible. A heating current of
70 A DC at about 8 V is sufficient to bring the crystal to 1600 K. The cooling time
between 650 and 6 K is 3 min.; from 1600 to 6.5 K the time is 5 min. At steady
state, the consumption of liquid He is 4 L/hr. With the cryostat cool, the base
pressure is below 10−11 mbar.
Reference [1] is highly instructive about the pitfalls of using cryogenic methods
in ultrahigh vacuum applications.
Others have employed closed-loop He cryostats for cooling of crystals in UHV
that do not achieve the very low temperatures described here [4–6].
Fig. 35.2 Spring-loaded cooling by contact. a Side view of manipulator + thermal contact,
b cooling curve
Figure 35.3a shows the experimental apparatus used in this work. A Cu reservoir
with polished planar faces holds two Cu support bars that are clamped with equal
force to the reservoir. The total contact area is 3.13 cm2 for both support bars. The
temperature of each bar (T1 and T2) can be measured during cooling, using type K
35.3 Cooling Samples on Manipulators … 385
thermocouples that are wedged into small holes in the support bars. In addition, the
crystal temperature can also be measured as cooling occurs to determine the
influence of the 0.025-mm-diameter tungsten wire (and the associated spot welds)
on the crystal cooling rate, compared to the support bar cooling rate.
Figure 35.3b shows cooling experiments comparing three different insulator
materials. The slowest cooling rate of the support bars is achieved using Kapton
films, 0.037 mm in thickness. Faster cooling of the support bars is achieved using
0.500-mm sapphire single crystal insulators. The fastest cooling is achieved when
an indium film is used to join the sapphire to the cooling reservoir and the support
bars. For cooling times up to 90 s, cooling of the central reservoir to 77 K is taking
place, and the data in this time period are disregarded in the measurements.
Figure 35.3c shows semilogarithmic plots of the cooling rate after the reservoir
has reached constant temperature. The slope of these plots is a measure of the rate
of cooling through each insulator junction. It may be seen that the Kapton exhibits
cooling at 81 % of the rate of the sapphire, and that the indium coating on the
sapphire enhances its cooling rate by about 6 %.
386 35 Cooling Samples
Long manipulator rods filled with liquid nitrogen can act as cryogenic surfaces that
can physisorb or condense gases used in surface science studies. As the liquid
nitrogen level drops, reevolution of these gases can cause problems in experiments
conducted under ultrahigh vacuum. The design shown in the Fig. 35.4a eliminates
this problem by using a vacuum jacket to isolate the outer wall of the manipulator
rod from the inner tube, which is cooled with liquid nitrogen [11].
A copper reservoir is supplied with liquid nitrogen through a 1/8-in. diameter
stainless steel tube. The excess liquid and vapor exits through a 5/16-in.-diameter
outer tube, which is concentric with a l/2-in.-diameter stainless tube welded to a
Conflat flange. The outer tube is pumped and contains an adsorbent material for
maintenance of the vacuum, as is sometimes used in metal Dewar flask vacuum
jackets. The liquid nitrogen inlet and outlet tubes are able to rotate inside a com-
pression fitting that contains a spring-loaded Teflon seal that allows the sample to
be rotated while the nitrogen delivery tube and the outlet tube remain fixed. Liquid
nitrogen is drawn through a latex rubber tube into the assembly from an unpres-
surized storage Dewar by means of a small carbon vane vacuum pump [12] attached
to the outlet tube. A valve at the inlet of the pump is used to adjust the flow of liquid
nitrogen. This pump is isolated from the liquid nitrogen reservoir by a length of
copper tubing that permits the liquid nitrogen to convert to gaseous nitrogen before
entering the pump. The length of the manipulator is 1.3 m below the Conflat flange.
Owing to thermal contraction, the inner tube will decrease in length upon
cooling. Since this is fixed at only one end, it may contract and expand freely. The
differential expansion of the outer two tubes is taken up by a formed bellows that is
1
Tabulated values for thermal conductivity are sapphire, 41.90 J/s km at 273 K; 100 HN Kapton,
0.10-0.35 J/s km at 276 K. Goodfellow Catalog (Goodfellow Metals, 1995), p. 433; p. 501. In
addition reference [10] gives information and references on thermal conductivities as a function of
temperature. Sapphire exhibits a greatly enhanced thermal conductivity at 80 K of 960 J/sKm.
35.4 Vacuum-Jacketed Cryogenic … 387
Fig. 35.4 Vacuum-jacketed cryogenic manipulator rod. a Vacuum-jacketed sample support tube.
b Sample holder cryogenically coupled to copper reservoir
welded to the compression fitting. Some contraction and expansion of the outer tube
still occurs as the inner tube changes temperature.
The sample holder shown in Fig. 35.4b is made of OFHC copper and is held on
the base of the copper reservoir by four screws that pass through a 0.5-mm-thick
sapphire plate that permits efficient cooling (see p. 383) as well as electrical iso-
lation. A Ta support structure consisting of two arms for the sample is bolted to the
copper support arms. Heating leads and thermocouple leads to the sample pass on
the outside of the vacuum-jacketed manipulator rod and are heat-stationed on the
lower cooled sections of the assembly.
In operation, the sample is cooled to 90 K when liquid nitrogen is the refrigerant.
Measurements indicate that the outer tube slowly cools to about 200 K owing to
cooling from its lower end, where it contacts the copper reservoir. This could be
prevented by the installation of a small electrical heater on the outer tube.
Because of the length of the manipulator rod, a lower bearing is supplied in
which the manipulator rod rotates. This prevents vibration of the sample inside the
apparatus. Rotation and translation of the sample are supplied by a standard dif-
ferentially pumped rotary seal (see p. 6) and a standard x-y-manipulator stage.
388 35 Cooling Samples
A simple design [13], involving indirect heating of a sample that can be cooled to
cryogenic temperatures and undergo limited azimuthal rotation, is shown in
Fig. 35.5. The sample is mounted on a metal cap fixed to a sapphire cylinder that
makes thermal contact with a Cu-Be cup through a novel ring of Cu-Be finger stock
[14]. The Cu-Be cup is held by a hollow shaft that rotates in the Cu-Be block through
stainless steel ball bearings that permit azimuthal rotation of the cup and the asso-
ciated parts connected to the sample. The sapphire is a good thermal conductor but is
an electrical insulator. In order to provide more cooling of the sample, four Cu braids
are connected between the sample mount and the Cu-Be block, which contains
internal pathways for flowing liquid nitrogen refrigerant. Azimuthal rotation is
achieved using a rack and pinion connection allowing linear translation of the rack to
produce rotation of the Cu-Be cup and the sample. An internal heater may be used
for radiation heating or electron bombardment of the sample and its mount. It is
electrically isolated and consists of a two-hole ceramic tube fixed in position with a
ceramic collar so it does not rotate as the Cu-Be cup is rotated. The azimuthal
freedom is ±45°, limited by the length of the copper cooling braids. The temperature
range of the sample is from 130 to 1000 K, with cooling times of about 45 min.
The use of cooling braids to supply flexibility in cooling of samples also causes a
limitation in cooling rate owing to the low thermal conductivity associated with
braids having a desired flexibility. Therefore, methods to cool movable samples by
direct contact with a cold reservoir have been developed.
The first of the methods [15] to be described is shown in Fig. 35.6. Here a 200-g
Cu tank is connected directly to a closed-cycle He refrigerator. The cold second
stage is shielded from external radiation by a first stage shield as shown. The Cu
tank contains a slot for the insertion of the sample holder containing a freshly
cleaved semiconductor sample in this case. This sample holder makes contact with
a droplet of molten Ga for the production of a metallic joint of high thermal
conductivity. The Ga can be melted by radiation heating from an attached heater as
shown in Fig. 35.6. The Cu tank can then be cooled to 20 K over a 140 min cooling
period after starting the refrigerator. The long cooling time is probably a result of
the large heat capacity of the massive Cu tank in this design. Temperature
adjustment can be achieved by using the radiative heater in opposition to the He
refrigeration unit. During heating, outgassing of the assembly with bursts of
pressure is minimal since cooling is being maintained and slow heating occurs.
The closed cycle refrigerator is mounted on a bottom-entry manipulator per-
mitting rotation and x, y, z-translation. Translation in the z-direction is especially
important because of thermal contraction of the device on cooling.
A thermal switch arrangement involving a high area Cu-Cu contact between a
liquid-He cooled reservoir and a crystal support assembly has produced a fast cooling
performance and a crystal temperature of 20 K after long equilibration times [16].
A design of a highly efficient attachment of a removable closed-cycle refrigerator
to a crystal mount has been described, with a limiting low temperature of 11 K [17].
Other arrangements for direct contact cooling are shown on p. 383 and 392.
390 35 Cooling Samples
The design shown in Fig. 35.7 uses two copper disks as heat transfer devices,
and at all rotational angles the disks are immersed in liquid nitrogen which is
confined by gravity in the expanded end of the manipulator [18]. The two current
carrying leads are brazed into holes in alternate copper disks so that efficient
thermal conduction will occur to each lead at all rotational angles. The use of two
Cu disks assures electrical isolation between the current carrying leads, which are
insulated up to the braze points with Teflon sleeves. The thermocouple leads pass
through an elongated hole in each disk and are insulated from the disks and each
other by Teflon sleeves.
The design shown in Fig. 35.7 is coupled to the manipulator by VCR fittings,
and the reader is referred to the original article [18] for details of this arrangement
and the alignment procedures for assembly. Tests with the design shown have
reached sample temperatures in the range 103–123 K.
392 35 Cooling Samples
The design of efficient heat exchange methods that also permit motion of a cooled
sample is made difficult by two factors: (1) the provision of sufficient thermal
conductivity in the materials making the thermal connection, and (2) the provision
of excellent thermal contacts between interfaces along the conduction pathway.
The best method for making a thermal contact between interfaces along the
conduction path is by welding [19, 20]. Brazing, while better than most pressed
mechanical connections, provides an alloy that is inferior to its metal components in
thermal conductivity. Wherever possible, welded thermal contacts between the
same materials (such as Cu or Ag) should be made.
Figure 35.8 shows a braid assembly made of OFHC Cu that is connected to a
cryostat and to a Cu sample holder by means of welds at two of the three contact
points. Because the Cu flexible braid work hardens upon extended usage, it should
be considered an expendable component. The Cu heat conductivity can be
improved by hydrogen firing at 1 bar pressure at a temperature of 600–800 °C for
3–5 min. Longer hydrogen firing may cause hydrogen embrittlement of the braid.
The effect of limited hydrogen treatment is probably to remove copper oxide films
that impede heat transfer at interfaces in the braid and on its connectors. Hydrogen
firing should be done only by experienced persons!
Many experiments require sample cooling below room temperature and below the
temperature of liquid nitrogen. The use of copper thermal braid for this purpose
provides flexibility in sample motion, and with proper design copper braid can be
employed for efficient cooling.
In Fig. 35.9a, a copper braid is employed with a detachable thermal switch to
provide intermittent cooling of a sample mounted on a movable manipulator [21].
Fig. 35.9 Cooling using thermal braid. a Detachable cooling assembly. b Continuously Coupled
cooling assembly
The braid is firmly connected to the sample mount by means of clamps made of Cu.
The other end is connected to a solid Cu cylinder, weighing about 100 g which
makes contact with a Cu cup. The Cu cup is cooled by flowing liquid nitrogen
through a coil wrapped and brazed in place. The use of brazing is highly recom-
mended, since the thermal conduction through a brazed joint is far superior to any
other mechanical contact except for a weld between identical metals. A reversible
metal contact between the Cu cylinder and the Cu cup is made using a liquid
gallium droplet (mp = 32 °C at a vapor pressure of 10−14 Torr). Thus, to melt the
Ga, hot air is passed through the coil on the cup; the weight is lowered into the cup
by moving the sample holder to which it is attached, and subsequent cooling freezes
the Ga, making a good thermal and mechanical contact. Sample cooling to
93 ± 10 K then occurs within about 200 s.
In Fig. 35.9b, a thermal braid is permanently attached to the sample holder [22].
The manipulator swings about a radius of 15 cm, and various degrees of freedom
are present. The sample holder is thermally isolated from the manipulator shaft and
most of the arm. The thermal braid is attached to a closed-cycle He refrigerator [23],
which has two stages of cooling to 80 K (15 W) and to about 10 K (1 W). The 80-K
stage provides cooling for a radiation shield that surrounds the low-temperature
stage, as shown schematically in the Figure. Modeling of the energy transfer was
done in order to compute the optimal dimensions for the sample holder and the
braid. The minimization of thermal conduction to the sample holder makes thermal
radiation the primary source of unwanted heat. Since the ratio of the surface area of
the sample holder to that of the thermal braid is about 1:12, a cooled radiation shield
around the sample is unnecessary in this design. It was calculated that the
cross-sectional area of the Cu braid was 0.075 cm2 on the basis of thermal con-
duction and thermal radiation considerations. It is of interest to note that this design
achieved sample temperatures of 14–15 K routinely with only 1 W of cooling in the
second stage of the closed-cycle refrigerator.
394 35 Cooling Samples
The use of copper braid for cooling samples on a manipulator is widespread. This
method of cooling avoids complex crystal motion devices that must also conduct
heat efficiently, and mountings with six degrees of motional freedom are often built
with braid cooling.
As the crystal and the manipulator shaft move, there will be the need for relief of
the strain in the copper cooling braid. While a long braid may be used to give a
sufficient degree of motion, this decreases the cooling efficiency of the system. It is
also possible, however, to mount the cooling reservoir itself on a flexible mount as
shown in Fig. 35.10 [25], thereby permitting a shorter braid to be used.
A mini-Conflat flange, mounted on the manipulator flange, is welded to a
standard Swagelock Tee. A stainless tube enters the vacuum system through this
flange and is connected to a very flexible bellows and then to a double kovar/glass
seal, which acts as an insulating break, electrically isolating the cooled reservoir
from the vacuum system and ground. An OFHC Cu reservoir is brazed to the end of
the tube assembly, and contains a polished flat to which the copper braid and its
copper lug are mounted (see p. 392). An indium foil gasket makes good thermal
contact when compressed by a bolt that passes through the flat and is drawn down
by a nut on the back side of the flat. The alternate use of threads in the soft copper
flat is not recommended because of the possibility of stripping threads.
Liquid nitrogen is supplied to the copper reservoir by suction from a styrofoam
cup or small Dewar through 1/8-in.-diameter Teflon tubing that can be sealed inside
the Swagelock fittings. Sample temperatures below 100 K can easily be achieved.
A simple method [26] has been devised and tested for cooling below the temper-
ature limit established by liquid nitrogen (77.3 K). The method involves bubbling
He gas through the liquid nitrogen (transpiration); evaporation of liquid nitrogen
35.11 Enhanced Cooling Using Liquid Nitrogen 395
requires heat absorption from the fluid itself, and the temperature falls as a
consequence of the evaporation until the triple point (63.1 K) is reached. Thus, for
work requiring temperatures slightly below that achieved using liquid nitrogen, this
method is convenient. Compared to pumping liquid nitrogen to reduce its equi-
librium vapor pressure and to achieve lower temperatures, this method offers the
convenience of not requiring high-capacity pumps connected to manipulators by
large flexible tubing, a procedure that is often mechanically complex and
inconvenient.
A He bubble in equilibrium with N2(l) will contain both He and N2, so that the
partial pressure of N2 will be less than the total pressure within the bubble. Thus as
N2 vaporizes within the bubbles, cooling will occur to reduce the temperature below
the normal boiling point of N2(l). Figure 35.11a shows a demonstration apparatus
where this experiment was carried out. A fritted disk bubbler was used in the
right-hand Dewar flask, while the thermocouple reference junction was maintained
at 77.3 K in the left-hand Dewar. After about 80 min of transpiration, solid nitrogen
crystals were observed, and the temperature reached that of the triple point. When
the He flow was stopped, the temperature remained stable for some minutes as the
solid nitrogen melted at the triple point; heat leakage into the Dewar was respon-
sible for this as well as for the gradual warming beyond 120 min.
Figure 35.11b shows a similar experiment in a 1.42-m column of liquid nitrogen
that comprises the cooling reservoir of a reentrant stainless steel manipulator on an
ultrahigh vacuum system. The temperature of a metal single crystal was measured
as He bubbled into the bottom of the reservoir from the end of a 0.097-cm-diameter
tube. The crystal temperature fell from 86.5 to 74 K during He flow. Stability of
±0.5 K was achieved, and excellent cooling control maybe maintained in this
temperature range by adjustment of the He flow rate. Because of the limited amount
of N2(s) produced at the bottom of the reservoir, interruption of the He flow
produced an almost immediate rise in the crystal temperature as shown. Because of
the length of the column of liquid N2, additional liquid N2 may be added without
disturbance, as shown at the time indicated by the dots in Fig. 35.11b. This is a
distinct advantage of using a long column of refrigerant where transpiration cooling
occurs primarily at the bottom.
There are many designs for level alarms and controllers for liquid-nitrogen Dewars
and Fig. 35.12 shows one of the simplest alarm devices for keeping a small supply
Dewar filled [27]. A cork float senses the level of liquid nitrogen in the Dewar, and
when the level drops to its lower limit, contact is made between a metal plate and an
inverted metal cup at the top of the sealing cork lid. This completes the circuit,
causing both an audio and a light signal to be emitted, alerting the operator to refill
the Dewar through the refill funnel. The audio oscillator normally used to drive an
Auger spectrometer employing a lock-in amplifier can be used.
35.12 Level Alarm-Liquid Nitrogen 397
Fig. 35.11 Cryogenic trick for enhanced cooling with liquid N2, a Attainment of N2 triple point
by He transpiration, b Cooling of metal crystal by he transpiration
The Dewar is sealed to the top of the entrance tube using cotton plus ice to make
the seal. Simply wet the cotton from a plastic squeeze bottle, admit a little liquid
nitrogen to freeze the water, and repair any leaks with more water on the cotton.
This type of seal does not put any stress on the glass Dewar during warm-up in
contrast to compression seals, which will allow the expanding metal to strain the
glass Dewar bottom, sometimes resulting in breakage.
398 35 Cooling Samples
the central tube and above the platform. The conical shape permits the Dewar to be
tightly sealed to the central tube by forcing it to slightly compress the many layers
of slippery Teflon tape. This makes a good seal for liquid-N2 which normally has a
low viscosity and tends to leak. Rubber seals become brittle in this application and
will fail. The Dewar is fitted with a loose-fitting styrofoam cover to prevent
excessive ice from forming. There may be condensation near the Teflon seal and
this can be eliminated by using a small fan for drying.
For studies of adsorbed He, it is necessary to cool below the boiling point of He.
This is especially difficult because of the need to reduce radiation heating at the
crystal to minimum values. The manipulator shown here achieves this with minimal
use of liquid-He. It is designed to also raise the crystal temperature to more than
1570 K, and this has been demonstrated with Ru single crystal samples. The
manipulator is immersed in a radiation shield in a liquid-N2 bath which itself is
cooled to below 77 K by pumping with a Roots blower. After heating the crystal to
high temperature in UHV to clean it, liquid He is admitted to the central Dewar
assembly and about 0.5 L accumulates. The inner He cryostat is then pumped with a
Roots blower until the final temperature of *1 K is reached, in an amazingly short
time of about 20 min after the high temperature cleaning procedure. About 0.3 L of
liquid He is consumed leaving 0.2 L for operation at the base temperature for about
400 35 Cooling Samples
The ability to turn on and turn off cooling to a sample and its support may be useful
in ultrahigh vacuum operations where condensation on the sample/support system
must be avoided. This could be the case when the sample and its support are to be
immersed in a high-pressure gas in a high-pressure vessel inside the UHV system.
The design of a gas thermal switch to vary the cooling rate to a sample and its
support is shown here [31].
The principle of the gas thermal switch is to use a heat transfer gas (He(g) is
ideal) to conduct heat when cooling is needed, and then to evacuate the He(g) when
cooling is not required. As shown in the exploded view of Fig. 35.16a, the OFHC
copper cooling stage is maintained at the cryogenic temperature desired by means
of contact with a reservoir of the cryogenic fluid (l-N2 in this case). Brazed to this
heat station is a concentric assembly of copper tubes. The cooling tubes nest into a
set of concentric tubes connected to the OFHC copper sample holder that is ther-
mally connected to the sample. Cooling is achieved by introducing about 0.5 Torr
of He(g) into the 10-cm3 volume containing the thermal switch. The cooling is
interrupted by pumping the He(g) away. Given the geometry of the thermal switch
(not exactly specified in the paper), the thermal conductivity varies between about
402 35 Cooling Samples
0.8 and 1.4 W/K with He(g) as a heat transfer medium to about 0.03 W/K with a
vacuum of <5 × 10−3 Torr, which is achieved in 10-s pumping time.
Figure 35.16b shows the performance of the gas thermal switch. He(g) is
admitted with the sample at 300 K and the cooling station at 100 K. Cooling starts
immediately with a time constant of about 2 min; after some cooling, the He(g) is
35.16 Gas Thermal Switch 403
Fig. 35.16 Gas thermal switch. a Thermal switch—exploded view. b Cooling performance using
vacuum and He(g)
evacuated, and cooling slows. Then He(g) is reintroduced, and cooling occurs with
a time constant of about 2.6 min. The decrease in cooling rate in the second stage of
cooling may be due to the decrease in the thermal conductivity of He as its tem-
perature falls. Cooling to about 102 K is achieved in less than 10 min.
404 35 Cooling Samples
References
temperature region. The maximum difference above 800 K is near 2 K. For tem-
peratures below 32 K, it is advisable to use the calibration in [3], which is exclu-
sively concerned with this temperature range. The fitting curve of [3] cannot be
extrapolated with accuracy above 32 K.
The power series for the range 4–2588 K is given below, and tabular solutions to
this equation are given in [4] at intervals of 0.05 or 0.10 mV.
36.1 Tungsten-Rhenium Thermocouples—Calibration Over a Wide Temperature Range 407
Table 36.1 Power series for the range 4–2588 K and solutions
T = ΣCnBn
C0 = 276.5327901 C6 = −4.147285358 × 10−3
C1 = 83.31475475 C7 = 1.529502993 × 10−4
C2 = −15.19381797 C8 = −3.422732361 × 10−6
C3 = 4.431847913 C9 = 4.253729476 × 10−8
C4 = −0.7258516658 C10 = −2.351967817 × 10−10
C5 = 0.06999557967
One problem that can arise in the use of high-temperature filaments is the
evaporation of the filament material, causing the diameter of the filament to
decrease, upsetting the temperature calibrations. In addition, filaments containing
attached pieces of evaporant material can change resistance over time due to this
effect as well as to alloying effects.
A useful nomograph for estimating the power requirements needed to heat
tungsten filaments of various diameters is shown in Fig. 36.2c. To determine
heating parameters, a straight edge is used across the chart. The example shown in
the center of Fig. 36.2c is a guide to tungsten wire heater design, including cor-
rections for tungsten wire cooling at the ends welded to the feedthrough leads.
The measurement of the brightness temperature of a hot object can be carried out
with an optical pyrometer working at a particular wavelength, λ. Often optical filters
in the pyrometer are used to define λ. Planck’s law for a perfect blackbody emitting
radiation energy, Wb(λ, T ), is
Wb ðk; T Þ ¼ 2phc2 =k5 ½expðhc=kkT Þ 1 ð36:1Þ
When an object is not a blackbody, the surface of the object is a partial inward
reflector of the radiant energy being emitted from the object, and the energy
emitted, Wn(λ, T ), is
where R is the reflectivity of each surface of the window material through which the
measurement is being made. R is given by
Both the emissivity, ε, and the window surface reflectivity, R, may be deter-
mined as correction factors for Tsample, as shown below.
For the emissivity, an experimentally determined number that is a specific
property of the material being measured must be used. Tabulations of the emis-
sivities for various metals are given in [6], as well as convenient correction plots to
directly convert the brightness temperature to the true temperature for various
emissivities.
For the particular case of Si at λ = 0.65 μm, the emissivity was measured
separately by Allen [7] and by Sato [8]. Their data are fit by the following equation
[9] in the temperature range 540–1700 K.
eð0:65 lm; T Þ ¼ 0:4826 2:09 105 T þ 0:1583 f1 þ exp½ðT 1410Þ=112:9g
ð36:7Þ
Figure 36.3b shows the fit of the empirical (36.7) to the data. The relationship
between Tpyro and Tsilicon can be calculated using (36.6) with values of R and
ε(0.65 μm, T) calculated from (36.5) and (36.7), respectively. In the case of glass
windows, the refractive index for 0.65 μm light is 1.46, and R = 0.035.
Figure 36.3c shows the deviation between the measured brightness temperature
with the pyrometer at 0.65 μm, and the true temperature for Si, viewing through a
glass window.
Similar considerations apply for pyrometric measurements made by electronic
methods in the infrared region, as is common for studies of semiconductor mate-
rials, except that a specific emissivity will apply for the particular material and the
wavelength of light employed.
36.4 Pyrometer Measurements Through Windows Coated … 411
The deposition of a thin film on an optical window can cause a negative error in
optical pyrometer readings, which will increase with the thickness of the deposited
film. In order to avoid this effect, an empirical method has been developed and
tested against thermocouple readings on a hot sample [1].
Figure 36.4a shows the basic principle of the method. A movable ionization
gauge having an emission regulated thermionic emitter is used as a standard. The
temperature of the emitter should remain constant in good vacuum for a standard
constant emission current that is set into the controller. When the window is first
installed, and when it is free of any thin film deposit, the brightness temperature of
the thermionic emitter is measured as a function of the emissivity setting on the
412 36 Temperature Measurements
Fig. 36.4 Use of standard filament to calibrate pyrometer through film-coated windows. a Ion
Gauge pyrometry standard. b Typical temperature versus emissivity setting for various window
film deposits
pyrometer, and a curve like that of Fig. 36.4b (clear) is constructed. As the window
becomes coated with an inadvertent thin film deposit, the brightness temperature of
the thermionic emitter filament versus the emissivity setting will decrease, as shown
for various time periods in Fig. 36.4b after the thin film has grown to various
thicknesses. For any temperature, an emissivity setting returning the filament
brightness temperature to the same value can be determined, as shown for example
36.4 Pyrometer Measurements Through Windows Coated … 413
for a brightness temperature of 973 K (dotted line), where emissivity settings of 0.6,
0.5, and 0.3 are needed to bring the pyrometer back to the standard temperature of
973 K. This same emissivity setting is then used when the sample temperature is
being studied in the position of the thermionic emitter.
The pyrometer must, of course, be calibrated accurately in the absence of any
thin film deposit. Both thermocouple and phase transformation information may be
used for this purpose, or in the case of materials with known emissivities, tables of
emissivity versus temperature may be used to make the correction from brightness
temperature to true temperature. If the sample temperature is considerably lower
than the ion gauge filament temperature, it will be necessary to employ an inter-
posed neutral density filter for the thermionic emitter measurements, as shown in
Fig. 36.4a. Because the thermionic emitter operates at about 1500 K, a neutral
density filter is needed if measurements of sample temperatures considerably below
this temperature are needed. In the example given, a neutral filter of optical density
2.0 (10−2 transmission) is needed to reduce the apparent filament temperature to
about 1000 K for an emissivity setting of 0.6.
Separate calibrations using a clean window that did not become coated with the
thin film verified that this procedure was accurate and that the thermionic emitter
temperature was constant over a long time [1]. In the case shown here, an IRCON
Model 2000 pyrometer, working in the range 0.7 mm–1.05 μm, was employed [10].
The provision of a calibration point for silicon at high temperatures for optical
pyrometers may be useful in certain situations. A method developed for doing this
in rapid wafer thermal processors [11] can be extended to other situations in the
laboratory.
In Fig. 36.5, a silicon wafer, 0.3 mm thick, and covered by a 5000-Å-thick SiO2
film, is coated with a 4000-Å Ge film, and then overcoated with a protective 1000-
Å Si3N4 film. This sandwich structure serves as a useful calibration standard. The
optical reflectivity of Ge at 0.6328 μm changes from 49 % just before semicon-
ducting Ge melts to 72 % for melted Ge, which behaves with metallic properties.
The melting point of Ge is 938 °C, and melting can be viewed through the Si3N4
overlayer. As shown in Fig. 36.5a, the sandwich test structure contains a K type
thermocouple embedded in the backside of the silicon wafer, and an optical
pyrometer to be calibrated at the melting point of Ge. A chopped laser beam is
reflected off of the Ge sandwich face, and the reflectivity of the surface is measured
as the temperature is raised, using external radiation for heating the wafer.
Figure 36.5b shows the result. At a thermocouple temperature of 938 °C, a sharp
414 36 Temperature Measurements
Fig. 36.5 Pyrometer calibration at fixed point for Silicon. a Ge/Si test structure. b Reflectivity
changes at 0.6328 μm
36.5 Calibration Point for Optical Pyrometer Used for Silicon 415
Both thermocouples and pyrometers are used to measure and control the temper-
ature of silicon crystals during electrical heating. As discussed elsewhere (p. 548),
the use of thermocouples for this purpose often produces metal contamination of the
silicon crystal. A method is described here involving the measurement of the
electrical resistance of the crystal during heating and cooling in order to determine
and control the temperature. For purposes of evaluation of the resistance method,
the crystal used in these measurements contains a W/26 % Re:W/5 % Re ther-
mocouple which has a ball end buried inside a laser-drilled hole in the crystal and
held in place with a MgO cement (Aremco 571P) [12].
The crystal is mounted on a Ta holder as shown in Fig. 34.5a, p. 358. The crystal
dimensions are 9.5 × 3 × 0.3 mm3. The silicon is P type, 100 Ω cm, boron doped.
The resistance of the crystal was calculated using voltage and current measurements
made during heating. Comparisons to the thermocouple and to an optical and an IR
pyrometer are described in the original paper [12]. Figure 36.6a shows the
dependence of the logarithm of the measured resistance versus 1/T. Above about
600 K, the expected linear relationship is observed, where the dominant intrinsic
conductivity of the silicon exists. Below 600 K, the dominating conductivity is due
to the dopants for this particular crystal. The slope in the linear region corresponds
to the quantity EG/2kBT, and the best value of EG (the silicon band gap) for this
crystal was measured to be in the range 1.22 to 1.25 ± 0.05 eV, in good agreement
with literature values of 1.170–1.205 eV [13–15].
Figure 36.6b shows the type of temperature control that may be achieved using
the resistance and comparing to the temperature measured directly with the ther-
mocouple. A programmed temperature ramp of 3 K/s, interrupted by a constant
temperature period, and then a cooling ramp of −3 K/s are shown. The solid line
represents the temperature measured with the thermocouple, and the dotted line
with the silicon resistance. Good agreement is found with a maximum deviation of
20 at 1150 K. Control was carried out using LabVIEW [16], which is described on
p. 351.
416 36 Temperature Measurements
The method must be calibrated against a standard at one or more points. This
may be done best using a pyrometer and comparing the resistance with the cor-
rected pyrometer reading (see p. 409). It would be better to do the calibration
against the pyrometer at several points in the intrinsic region in order to make the
best calibration, although the knowledge of the bandgap permits single point cal-
ibration [12].
References 417
References
Beam dosers for quantitative exposure of surfaces to gases have become widely
used in surface science research. This is because of the disadvantages in filling an
ultrahigh vacuum system with gas in order to study adsorption processes. These
disadvantages relate to the large gas loads needed for system dosing as compared to
beam dosing. Such large gas loads result in adsorption on the walls of the vacuum
chamber, which then contributes to slow achievement of base pressure after
adsorption by the sample surface is complete. In addition, large gas loads in a
vacuum system can cause displacement effects from the system walls leading to
impurity adsorption, as well as regurgitation effects from ion pumps in which
impurity gases are produced (see p. 72). Also, the use of quantitative and absolute
calibration procedures for the gas flux coming from a collimated doser is much
superior to dosing with a system dose using an ionization gauge or a mass spec-
trometer, both of which have serious problems with absolute calibration.
Figure 37.1a shows a design for a microcapillary array beam doser [1].
A pinhole aperture made of stainless steel serves as a deformable gasket inside a
Cajon vacuum fitting [2]. Typical aperture diameters are about 5 μm for convenient
dosing fluxes for adsorption studies, as shown on p. 425, where calibration methods
are described. The flux of gas issuing from the pinhole aperture travels into the
doser body, where it strikes a baffle made of metal shim stock, held into place by
springy arms that bend against the inside wall of the doser body. This baffle
randomizes the gas molecules’ motion inside the doser body and behind the mi-
crocapillary collimator plate. Collimation of the gas occurs in the high-density
capillary array where parallel capillaries of 10 μm diameter and 500–1000 μm
length are arranged in a hexagonal close packing arrangement [3]. The microcap-
illary array disk has a radius, RS and should be purchased with the capillary axes
perpendicular to the array surface, rather than inclined as in the chevron design
Fig. 37.1 Design and performance of microcapillary array doser, a Design of doser (R = radius),
b Fractional interception by target
commonly manufactured [3]. The capillary array is held in place by a metal cap
that, with a clearance of several micrometer, confines the edge of the array in a
close-fitting groove on the end of the doser body. Care should be taken not to
compress mechanically the very fragile glass capillary array during installation. The
capillary array has a relatively high conductance for gases and acts mainly as a
collimator in the gas doser.
Figure 37.1b shows the intercepted flux fraction from the array by the target of
radius RT for various geometries [4, 23]. The most uniform dosing over the target
will take place with larger values of RS /RT but of course the fraction of the beam
intercepted will decrease as shown in Fig. 37.1b. The assumptions used in deriving
the functions shown in Fig. 37.1b assume that diffuse scattering occurs on the inside
walls of the capillaries under molecular flow conditions where the mean free path is
considerably longer than the diameter of the capillaries.
If the capillary array is too close to the target, reflected molecules from the target
will have a finite probability of incidence on the crystal for a second encounter
when they are reflected back by the capillary array surface.
37.1 Design and Performance of Microcapillary … 423
The effective use of the microcapillary array beam doser shown on p. 421 requires a
gas-handling system of appropriate design, as shown in Fig. 37.2 [17]. A stainless
steel system, pumped by a turbopump, is designed to attach to several glass storage
gas bulbs, using Cajon VCR fittings [17].
The storage of gases in glass bulbs at pressures near 1 atm prevents accidental
over-pressurizing of the gas-handling system by compressed gases in cylinders.
Also, during gas transfer to prebaked storage bulbs (for removal of adsorbed H2O
from the glass walls), one has the opportunity to use purification steps for gases
such as the freeze-pump-thaw method commonly used to remove noncondensable
impurities.
For calibration of the gas doser system, a test gas, such as N2, is admitted to the
gas-handling system to a few Torr pressure. The ballast bulb, of exactly known
volume, is also filled with this gas. As will be shown on p. 425, the conductance of
the pinhole aperture may be measured by measuring the time rate of change of
pressure in the gas-handling system as the N2 is slowly pumped through the pinhole
aperture. To interrupt the flow from the array doser, Valve-1 is closed, and the
region behind the doser is rapidly evacuated through Valve-2, keeping the gate
valve to the UHV system closed.
424 37 Gases
Stainless steel systems such as this can be baked with heating tape to 150–200 °C
while pumping with the turbopump. Because of the low pumping speed through the
narrow tubing and the relatively high internal surface area compared to the volume,
limiting pressures in the 10−8 Torr range are usually measured. For gases added to
the Torr level, this background gas represents much less than 1 ppm of added
impurity, which is usually superior to the initial purity of the gas being studied.
The internal surface of the stainless steel gas-handling system has a significant
capacity to react with active gases such as Cl2. Gases of this sort may often be used
themselves to passify the stainless walls. This is done by storing the active gas at
pressures of several Torr pressure overnight, and then pumping the gas away before
adding fresh gas to the system [18]. It may be necessary to do this passivation treatment
in several cycles. In other cases, treatment of the gas-handling system with oxygen gas
to passivate the inner walls by oxidation has been found to be effective [19].
37.3 Calibration and Use of Array Beam Doser 425
Figure 37.3a shows a typical design for an ultrahigh vacuum system equipped to
measure the kinetics of adsorption of a gas on an atomically clean crystal surface.
A quadrupole mass spectrometer (QMS detector), shielded to receive only a random
flux of background gases, and the microcapillary array doser are the principal
components of the apparatus for this measurement.
The calibration of the doser was performed with N2 over a period of about 105 s,
as shown in Fig. 37.3b. The data, plotted logarithmically, against time, t, provide a
linear curve whose slope is related to the conductance of the pinhole by (37.1).
Fig. 37.3 Use of beam doser for kinetics and coverage measurements. a Typical apparatus—
absolute coverage and sticking probability measurement, b Typical calibration curve—pinhole
conductance. c Adsorption uptake curve. d Sticking probability versus absolute coverage
426 37 Gases
[20]. For molecular flow conditions, where the mean free path of the gas on the
high-pressure side of the pinhole is larger than the diameter of the pinhole, the rate
of effusion through the pinhole will be proportional to M−1/2, where M = the
molecular mass of the gas. This condition continues to be approximately true as one
enters the transition region in gas density, at higher pressures, where viscous flow is
beginning to occur in the pinhole aperture which has finite wall length.
Figure 37.3c shows measurements of the adsorption of O2 on a Fe(110) surface at
90 K [21]. For this measurement, the shutter is placed between the doser and the
crystal, and a steady base pressure of the order of 1 × 10−10 Torr or less is estab-
lished, using the mass spectrometer tuned in this case to 32 amu. After steady
conditions are achieved, the doser is turned on with the shutter still in place and the
pressure measured by the mass spectrometer rises. Ideally the pressure should level
out prior to opening the shutter to allow adsorption from the beam to occur, but since
a small flux of random gas scattered from the back of the shutter is incident on the
crystal, causing a slow increase in coverage, the waiting time is limited and a
compromise must sometimes be reached in which “almost steady state” conditions
for Pback are established, as shown here. At this point, the shutter is opened, and
adsorption occurs until asymptotic behavior of the scattered flux of gas is observed.
Then, when the sticking probability has fallen to a low value, placing the shutter into
the beam causes no discontinuity to be observed in the mass spectrometer signal.
The data shown in Fig. 37.3c can be converted into sticking coefficient vs.
absolute coverage, as shown in Fig. 37.3d, if the fractional interception of the doser
beam is calculated (see p. 421) (or measured [20]). It is of interest to note that the
entire measurement of sticking probability and absolute coverage are determined
solely by the accuracy of the flux calibration and the accuracy of the fractional
interception of the collimated gas beam by the crystal [22, 23]. The mass spec-
trometer does not have to be calibrated in any absolute sense.
This method is based upon earlier work with variations on the features of the
apparatus and the measuring conditions [24–27].
Fig. 37.4 Sticking coefficient and surface reactivity. a Sticking coefficient and surface reaction
apparatus. b Typical uptake kinetics data
Very simple measures can sometimes be used to interrupt the flow of gases from a
gas doser. As indicated on p. 430, the use of a small deadspace valve of special
design is helpful in defining reproducible and absolute doses of gases into a colli-
mated gas doser system. In some cases, where gases condensable in liquid nitrogen
are being used with a gas doser, it is possible to interrupt the flow rapidly by using a
cold trap as a pump. In Fig. 37.5, a cold trap made of Pyrex glass is immersed in a
Dewar flask containing liquid nitrogen [30]. This cold trap is a very rapid pump for
all gases that are condensable at 77 K. Flowing gas through the doser (see p. 423)
may be turned off rapidly by closing the admission valve and then quickly opening
the dump valve. This lowers the gas pressure very rapidly behind the conductance
limiting aperture, and turns the doser off. Many doses of gas, and different con-
densable gases, may be trapped without cleaning the collector tube. For cleaning,
the tube is disconnected from the doser line using the VCR fitting, cleaned with
solvents, baked briefly in an oven, reconnected, and evacuated using the pump in
the gas-handling system. This method is actually preferable to pumping corrosive
gases from the trap, after each dose, through a turbopump because of possible
failure of the turbopump after exposure over time to corrosive conditions.
Ps ¼ Qo =Corf ð37:2Þ
where Qo is the net outgassing rate and Corf is the conductance of the orifice.
When the valve to the calibrated leak is opened, the pressure increase will be
where QL is the calibrated leak rate. The conductance of the orifice, which should
be sharp-edged is, under molecular flow conditions
where D is the orifice diameter in cm, T is the absolute temperature, and M is the
molecular mass of the gas.
When the molecular beam passes through the calibrated orifice it will cause a
pressure rise in the stagnation chamber given by
It is important that the beam diameter be larger than the orifice and that the baffle
inside the stagnation detector be arranged to prevent direct entry of the gas from the
calibrated leak into the ionization gauge. For reactive gases such as oxygen, the
operating temperature of the ionization gauge filament and the emission current
should be reduced as much as possible to prevent pumping effects.
The intensity at a sample position that differs from the orifice position is related
to the intensity at the orifice by the inverse square law for the distances involved.
A good article detailing the many experimental aspects of the use of supersonic
molecular beams in surface chemistry studies may be found in [32].
The use of gas dosers for adsorption experiments is highly recommended in order to
avoid impurity effects from wall displacement and ion pump regurgitation processes
as well as to have quantitative gas dosing capabilities, as discussed on pp. 421–428.
37.7 Low Deadspace Aperture Valve for Repetitive Gas Dosing 431
In the conventional gas dosing design, a rather large volume of gas is held in a
storage section of the gas line, and to interrupt dosing, one must pump out this gas
or some fraction of it held in a section of the gas storage line. For repetitive gas
dosing, this is inconvenient, and in addition for expensive gases it is wasteful. The
design shown here [33] avoids this effect by minimizing the deadspace volume so
that only a few seconds of gas dosing occurs following closure of the valve. In
addition, when a conductance-limiting aperture at the outlet to the small deadspace
volume is used, quantitative gas dosing can be done when the valve is opened.
Figure 37.7a shows a view of the aperture valve assembly. The valve body of a
stainless steel diaphragm ultrahigh vacuum valve (Nupro DL Series) is drilled and
Fig. 37.7 Low deadspace aperture dosing valve. a Aperture valve assembly. b Pumping behavior—
40 μm aperture. c Duty cycle—open and close
432 37 Gases
tapped with 1/16 NPT pipe threads to receive a stainless steel rod also threaded on the
upper end. This is done at the correct angle inside the inlet side of the valve. This rod
has a central hole, 2 mm in width, bored its length, except for an 0.003-in.-thick
membrane at the inner end. The membrane is then laser-drilled with a nominal
60-μm-diameter conical hole [34]. The upper end of the insert contains a screwdriver
slot for convenient insertion into the valve; Teflon tape is used to make a gas tight seal.
The inlet to the valved assembly (originally the outlet) is connected to a conventional
gas-handling system, containing pumping capabilities, a capacitance manometer, and
a ballast volume, V, used for conductance calibration of the laser-drilled aperture. The
outlet of the valve assembly is connected to a small flange that connects to a colli-
mating gas doser (see p. 421) inside the ultrahigh vacuum system. With this
arrangement, the deadspace when the valve is closed is about 0.2 cm3. Thus, when the
dosing is completed, and the valve is closed, evacuation of the deadspace into the
ultrahigh vacuum system by way of the doser does not involve the flow of large
amounts of gas, even when the backing pressure is near 0.1 Torr. Under these con-
ditions, the number of molecules in the deadspace is about 4 × 1014. If 20 % of these
molecules strike the crystal and the sticking coefficient is 0.1, about 2 % of a
monolayer will adsorb from the exhaust of the deadspace after the valve is closed.
Figure 37.7b shows a comparison of the pumping speed through the opened
valve for both Ar and H2, and may be used to measure the conductance of the
laser-drilled aperture. It is noted that the effusion rate follows a semilogarithmic
behavior from backing pressures of 10 Torr and lower, indicating that slight
departures from molecular flow conditions do not sensibly disturb the conductance
of the aperture from that predicted by the kinetic theory of gases for molecular flow.
The superposition of points for the two gases on the curve gives further confidence
in the notion that the behavior can be predicted from the molecular flow equations.
Figure 37.7c shows the response of a mass spectrometer in the vacuum system as
the valve is opened and closed for one cycle. A total rise time of about 3 s is seen as
the valve is opened. Constant Ar flux occurs beyond this time until the valve is
closed, where the pumping out of the deadspace volume exhibits a time constant of
1.3 s. Even though these on-and-off periods give relatively slow response times, the
doser may still be calibrated to give reproducible doses containing the on-off
periods as well as the constant flux period. The use of a smaller laser-drilled
aperture is possible, but the time constant for pumping will increase to unacceptable
levels. A major feature of the apparatus is that the unit can deliver rather large
fluxes at fairly low backing pressures and that it can be quickly cut off without
wasting gas. Multiple exposures that are reproducible can be achieved.
A variation on this idea, involving a trapping volume of about 0.4 cm3, has been
employed to dump known quantities of gas onto a crystal by way of an effusive
doser [35–37]. The trapping volume contains about 1.4 × 1016 molecules/Torr; and
the repeatability of the dose, using a capacitance manometer for pressure mea-
surements in the gas line feeding the doser volume, is estimated to be 0.5 %
standard deviation.
37.8 Gas Flow Regulation—Squeeze Valve 433
A very simple method for regulating the flow of gases into an ultrahigh vacuum
systern with a minimum of contaminants involves the use of a brass squeezing
device to restrict the flow through a thick-walled copper tube, as shown in
Fig. 37.8a [38]. The copper tubing (3 mm OD × 1 mm ID) is annealed in a flame,
then cleaned internally with nitric acid and rinsed with water and alcohol. The brass
squeeze device is clamped to the tube, and the flow rate may be adjusted over a
Fig. 37.8 Gas flow regulation—squeeze valve. a Squeeze valve. b Gas-handling system
434 37 Gases
wide range using the tightening screws. This action is reversible tens of times. The
valve cannot be closed completely. To adjust the flow rate after the valve is set, the
pressure in the high-pressure region is varied using the pressure gauge for mea-
surement. To shut down the flow, the upper UHV valve is closed and the region
between this valve and the squeeze valve is evacuated to the pump, as may be seen
in Fig. 37.8b.
A convenient molecular beam doser which turns itself on or off upon retraction is
shown in Fig. 37.9. A leak valve or a calibrated leak can be used to control gas
flow. It is convenient to turn these dosers on or off by quickly pumping out the gas
supply to the doser. In this case, a translational manipulator is used to move the
doser in and out. When in the in position, the 4 holes in the doser tube between the
two O-ring seals are isolated from differential pumping and the doser delivers gas at
the established flow rate. When the doser is in the out position, the four pumping
holes are exposed to differential pumping, turning the doser off. The O-rings are
cemented into grooves machined in the on-off assembly. They slide into a polished
cylindrical region with a slight taper on the left side for easy admission. The doser
is guided by a Cu guide with a loose-fitting alignment hole. The doser contains a
microchannel plate for gas collimation and a baffle to prevent beaming of the gas to
the plate from the delivery tube, spoiling the collimation [39].
References
1. M. Bozack, L. Muehlhoff, J.N. Russell Jr, W.J. Choyke, J.T. Yates Jr, J. Vac. Sci. Technol.
A5, 1 (1987)
2. P.L. Hagans, B.M. DeKoven, J.L. Womack, J. Vac. Sci. Technol. A7, 3375 (1989).
Laser-drilled micron-size pinholes are produced in disks fitting VCR fittings by Precision
Laser Services, 314 E. Wallace St., Ft. Wayne, IN 46803
3. Capillary array blanks, made without being coated with active surfaces for use as electron
multipliers, are available from Galileo Electro-Optics Corporation, Galileo Park, P.O. Box
550, Sturbridge, MA 01518. Sizes up to 25-mm diameter are available
4. A. Winkler, J.T. Yates, Jr., J. Vac. Sci. and Technol. A6, 2929 (1988)
5. S.M. Gates, J.N. Russell Jr, J.T. Yates Jr, Surf. Sci. 146, 199 (1984)
6. P.K. Leavitt, P.A. Thiel, J. Vac. Sci. Technol. A8, 148 (1990)
7. M. Henderson, R.D. Ramsier, J.T. Yates Jr, J. Vac. Sci. Technol. A9, 2785 (1991)
8. A.M. Glines, R.N. Carter, A.B. Anton, Rev. Sci. Instrum. 63, 1826 (1992)
9. E.C. Henn, M.E. Bussell, C.T. Campbell, J. Vac. Sci. Technol. A9, 10 (1991)
10. D.W. Goodman, T.E. Madey, M. Ono, J.T. Yates Jr, J. Catalysis 50, 279 (1977)
11. D.W. Goodman, J.T. Yates Jr, T.E. Madey, Chem. Phys. Lett. 53, 479 (1978)
12. S.M. Gates, J.N. Russell Jr, J.T. Yates Jr, Surf. Sci. 159, 233 (1985)
13. D.E. Ibbotson, T.S. Wittrig, W.H. Weinberg, Surf. Sci. 110, 294 (1981)
14. J.E. Deffeyes, A.H. Smith, P.C. Stair, Surf. Sci. 163, 79 (1985)
15. D.M. Murphy, J. Vac. Sci. Technol. A7, 3075 (1989)
16. D.E. Kuhl, R.G. Tobin, Rev. Sci. Instrum. 66, 3016 (1995)
17. M.J. Bozack, L. Muehlhoff, J.N. Russell Jr, W.J. Choyke, J.T. Yates Jr, J. Vac. Sci. Technol.
A5, 1 (1987)
18. A.L. Linsebigler, V.S. Smentkowski, M.D. Ellison, J.T. Yates Jr, J. Am. Chem. Soc. 114, 465
(1992)
19. M.A. Henderson, R.D. Ramsier, J.T. Yates Jr, J. Vac. Sci. Technol. A9, 2785 (1991)
20. C.C. Cheng, R.M. Wallace, P.A. Taylor, W.J. Choyke, J.T. Yates Jr, J. Appl. Phys. 67, 3693
(1990)
21. V.S. Smentkowski, J.T. Yates, Jr., J. Vac. Sci. Technol. A7, 3325 (1989); Surf. Sci. 232, 113
(1990)
22. A. Winkler, J.T. Yates Jr, J. Vac. Sci. Technol. A6, 2929 (1988)
23. C.T. Campbell, S.M. Valone, J. Vac. Sci. Technol. 13, 408 (1985)
24. D.A. King, M.B. Wells, Surf. Sci. 29, 454 (1972)
25. T.E. Madey, Surf. Sci. 33, 355 (1972)
26. A.A. Bell, R. Gomer, J. Chem. Phys. 44, 1065 (1966)
27. C. Wang, R. Gomer, Surf. Sci. 84, 329 (1979)
28. D.A. King, M.B. Wells, Surf. Sci. 33, 355 (1972)
29. F. Zaera, Int. Rev. Phys. Chem. 21, 433 (2002)
436 37 Gases
The deposition of thin films by evaporation constitutes one of the main experi-
mental procedures in surface and materials science. Indeed, the study of chemi-
sorption on clean surfaces has as one of its origins the use of thin metal films as
substrates in early studies done in the 1940s.
A major problem in making clean thin films originates from the outgassing of
heated surfaces during evaporation, leading to contamination of the thin film by
chemisorption from the background vacuum. This may be avoided by careful
design in which prior out-gassing is achieved for components that will be heated
during evaporation. Figure 38.1 illustrates how this is done, using two principles.
The first principle involves the use of degassing loops to support all filaments
that will be heated [1]. In the center of the evaporation shield, the evaporation loop,
containing welded pieces of the evaporant at its apex, is mounted on two degassing
loops. Knowing the current (I1) required to bring the evaporation loop to its desired
temperature for evaporation of the evaporant, one passes current I2 through the
degassing loops prior to the evaporation step. The current I2 should be somewhat
larger than I1/2 so as to bring the degassing loops to a higher temperature than they
will ever attain during evaporation, and the two loops should be outgassed sepa-
rately so as to avoid passing current through the evaporation loop during
outgassing.
The second principle illustrated in Fig. 38.1 is the use of a shield-outgassing
filament for heating the shield to temperatures higher than will be achieved while
evaporation occurs from the evaporation loop [2]. This is achieved by radiation
from the outer shield outgassing filament, where the shield thermocouple may be
used to monitor the temperature achieved during outgassing compared to that
during evaporation. Note that this shield outgassing filament is also mounted on
degassing loops.
Schematic performance curves are shown in Fig. 38.1 to illustrate that extremely
rapid employment of a properly outgassed evaporation source is possible compared
to a poorly designed source that may outgas over days prior to being useful for
evaporation under good vacuum conditions. The shutter should not be opened until
an appropriate limiting pressure is achieved during evaporation.
Some evaporation sources will be subject to continual outgassing because of
design limitations preventing the use of methods shown here. In that case, reduction
of the unwanted gas fluence at the substrate may be achieved by differentially
pumping the source and the substrate chambers that are connected by an aperture [3].
38.1 Degassing Evaporation Sources 439
Two oven designs for the downward evaporation of metals using radiative heating
have been described that operate at very low outgassing rates [4]. The problem with
most downward evaporation sources that involve melting of the evaporating metal
is that gravity may cause molten metal to fall from the evaporator to the substrate.
The cylindrical oven body is made of tantalum foil, 0.05 mm thick, spot-welded
into shape and containing end caps made of the same size material. The oven is then
wrapped with several layers of 0.01-mm-thick Ta foil to provide substantial thermal
shielding. This heat shielding is essential to permit good thermal efficiency and
therefore low outgassing rates. Two 3-mm-diameter Mo rods are used to bring in
power to a W coil filament, 1 mm in diameter and 250 mm long. The longer of the
two rods is spot-welded to the inside of the Mo oven body to provide mechanical
support for the assembly. The second passes through a hole in the top cap and is
insulated by a thin tube of pyrolytic boron nitride, pBN. When the oven is operating
the resistance of the W coil is about 0.5 Ω. A W25 %Re/W3 %Re thermocouple
monitors the temperature of the top end cap, but does not necessarily read the
correct temperature of the metal source.
Figure 38.2a shows the design for a nonmelting metal (3-mm slab) that is sup-
ported with thin W wires on the bottom of the oven. A W shield, placed on the top of
the source material reduces upward evaporation, extending the life of the oven. This
source has been used for Fe film deposition rates of between 5 and 30 Å/min at
power levels of 90 W and above without melting. The oven-to-substrate distance is
10 cm in this case. Mn has also been evaporated with this design.
Figure 38.2b shows a modified design that permits evaporation of metals that
must be melted to achieve the desired rate of film growth. A set of Ta rings
440 38 Evaporation Sources
(0.01 mm thick) alternate with rings of the evaporating metal (0.1 mm thick). The
volume of the metal charge may approach 0.4 cm3. A thicker Ta shield disk
(0.05 mm thick) is placed above the stacked rings to direct the metal deposition
downward and to shield the substrate from the radiation of the hot W filament. For
Cu deposition, a power level of 95 W will give a deposition rate of 10 Å/min, and
145 W gives a rate of about 45 Å/min. The melted metal is held between the thin Ta
rings by surface tension. Silver has also been evaporated successfully with this oven
design. Metals that alloy with Ta may self-destruct the oven.
The ovens are surrounded by tightly coiled water-cooled Cu tubing that does not
touch the oven and that extracts much radiant heat. It is reported that with this
design, the total pressure rise is <5 × 10−11 Torr [4].
An excellent review of metal evaporation sources and materials compatibility at
high temperatures is given in [5].
The use of a quartz thickness monitor (QTM) for absolute measurement of the
thickness of deposited films has been discussed elsewhere (p. 309). The extension
of this technique to very low coverages of deposited material has been made by
Chin et al. [6]. Essentially they arrange for the rate of evaporation to be determined
by a QTM placed rather close to the evaporation source (7.6 cm), while collecting
38.3 Ultralow Coverage Metal Evaporation 441
Fig. 38.3 Ultralow coverage metal evaporator, a multiple evaporator source, b sources mounted
at long distance
the deposit on the crystal substrate at a much longer distance (dcrystal = 112 cm).
The apparatus shown in Fig. 38.3a consists of a set of multiple evaporation sources
of similar geometry mounted at a distance, dQTM, from the QTM, which is centrally
located. The crystal substrate is mounted in an ultrahigh vacuum chamber facing the
evaporation source, which is located far from the substrate in a long flight tube as
shown in Fig. 38.3b. Considering both the distances and the angles between the
normal to the QTM and the crystal, αQTM and αcryst., the equation for the relative
rates of deposition is
In the design and testing of this long-path evaporator, the effect of gravity on the
evaporating beam was shown to be negligible as was the effect of beam scattering
by residual gases at ultrahigh vacuum. For the geometry shown, it was calculated
that because of the lower pumping speed at the evaporation source compared to that
at the crystal, the pressure in the evaporator region would be about 40 % greater
than in the ultrahigh vacuum chamber.
In order to obtain a calibration factor for the apparatus, a second QTM was
placed at the crystal position and the ratio of the signals received during evaporation
from each of the individual sources was measured. It was found that each of the
evaporators gave the designed ratio of about 200 within a 20 % error range, whether
the evaporator source material is in the form of a bead or a coil of wire. It was also
shown that shadows cast by the QTM and its support structure were unimportant at
the flight distances employed. Minimum deposition levels of the order of 10−4 ML
are possible with this apparatus. An added advantage of the geometry used is that
heating of the crystal substrate by radiation from the sources is negligible.
An excellent review of metal atom evaporation sources and materials compati-
bility at high temperatures is given in [7].
The evaporation of alloys and mixtures will almost always result in composition
changes during evaporation because of the differences in vapor pressures of the
different components. Thus, flash evaporation is often employed where the entire
alloy or mixture is rapidly evaporated to yield films of more uniform composition,
equivalent to that of the starting material. Often, the starting material is dispensed as
a powder onto a heated surface where evaporation occurs rapidly. This procedure
has certain deficiencies, such as the need to powder the material, which might
introduce impurities, and the problem with blockage of tubes, etc., with powdered
material. Often, too, the powdered material is mechanically ejected, causing film
inhomogeneities. The device shown in Fig. 38.4 overcomes these problems by use
of a liquid material for the dispenser device [8]. Films of InSb and As/Sb/Se have
been made by a predecessor of the method to be described, and films of doped Se
have been made by the new method [8].
A quartz capillary tube is partially immersed in a molten alloy of the desired
composition, and the liquid is kept slightly above the melting temperature. The
molten alloy is forced through the capillary tube by means of an inert gas with a
pressure of the order of 10 Torr, and this pressure is dynamically regulated by
means of pumping and a gas emission needle valve. The molten source material
streams out of the capillary tube into an evaporator heated to a temperature that
produces rapid evaporation from its inner surfaces, and is above the melting point
of any congruent compound that might form. The evaporator is baffled so that direct
38.4 Congruent Evaporation of Multicomponent Materials 443
line of sight between the capillary end and the substrate is not present. In this
arrangement, the molten source material can be thoroughly outgassed prior to film
deposition. It is advantageous to heat the top of the capillary and the evaporator to
temperatures where the source material already starts to evaporate before leaving
the capillary.
444 38 Evaporation Sources
Often, eutectic-forming metals are combined to form evaporation sources for the
higher vapor pressure metal. However, at the temperature of vaporization, a
lower-melting eutectic alloy may form, causing the mechanical failure of the
evaporator device by melting.
A method to avoid this problem has been developed that prevents the formation
of the eutectic in sufficient quantities to cause mechanical failure. In the case of a
zirconium evaporator, this method has proved satisfactory [9]. Figure 38.5a shows
the basic principle of the method. A tungsten rod serves as a mandrel for the
evaporator coil, which consists of smaller-diameter W wire wound together with Zr
wire. Ohmic heating of the assembly by passing current through the large Ta
connector and support plates raises the Zr and W to the correct evaporation tem-
perature (about 2300 K), using 38–40 A, and permits Zr decomposition at a rate of
0.25–0.35 ML/min on the sample located 3 cm from the evaporation source.
A special outgassing procedure was employed for this source [9], which is gassy
given the high operating temperature and the method of construction.
The principle of operation may be seen from considerations of Fig. 38.5b. Here,
a lower melting eutectic mixture of Zr and W is produced at intermediate atomic
fractions of W. In the case of the source described above, one avoids arriving at the
eutectic region for a substantial part of the assembly by limiting the amount of Zr
available compared to the amount of W. Thus, even if the eutectic forms in the
immediate region of a Zr/W interface, diffusion of Zr from this region will of
necessity bring one to the left-hand region of the phase diagram and into the
composition range where extreme depression of the W melting point does not
occur. This principle is often applied in designs for W/Ni and W/Al [10] evapo-
rators where the same kind eutectic formation occurs, causing difficulties.
An excellent article describing metal evaporation methods and materials com-
patibility at high temperatures is available [11].
The use of alkali metals as adsorbed species and as modifiers of surfaces is well
documented in the literature [12], and several alkali metal evaporation sources are
used, as indicated on p. 483. A particularly nice source for Li has been described
and is presented here [13, 14].
A binary alloy of Al and Li, containing 6.5 % Li, was used in the form of a disk,
10 mm in diameter and 1.5 mm thick [15]. The material was rinsed in acetone
before use and is known to be covered with a thick oxide layer. Two
0.25-mm-diameter Ta wires were looped through two pairs of holes, and a fifth hole
was used for attachment of a type K thermocouple, as shown in Fig. 38.6a.
Although not used by the developers of this Li source [13, 14], the Ta heating wires
38.6 Simple Lithium Metal Evaporation Source 445
should be mounted on degassing loops (see p. 407 and 437) for maximum effi-
ciency in outgassing the support. The evaporation source is surrounded by a
stainless steel shield.
Li evolution from the source was monitored with a quadrupole mass spec-
trometer, and the temperature dependence of Li evolution is shown in Fig. 38.6b.
For actual film deposition, a temperature of 695 K was employed, and Li was
446 38 Evaporation Sources
Fig. 38.6 Li thin film evaporation source, a Li evaporation source, b Li+ mass spectrometer signal—
T scan rate = 0.5 K/s
that retards alloy effects [17]. A variety of the most recent methods for evaporating
Al have been reviewed in a paper which also presents a different solution to the
lifetime problem [18].
The evaporation source is made of zero-porosity alumina tubing (DIN 40 685
type AL23-710) [19], 1.1 mm inner diameter and 1.7 mm outer diameter, 6 mm
long. Into this was inserted a 6-mm piece of high-purity (99.999 %) Al wire, with a
1-mm diameter. A tungsten coil, made of 0.25-mm wire and wrapped around a
1.37-mm-diameter mandrel prior to use, was fastened tightly around the alumina
tube by virtue of the size difference between the coil and the tube. This assembly
was then spot-welded to two Mo rods of 2-mm diameter. The assembly was rigidly
supported between the Mo rods using ceramic locating blocks, not shown in
Fig. 38.7. The evaporation source was surrounded by a shield that was supported on
the mounting flange, and that contained an 8-mm-diameter hole on the axis of the
device. A rotatable shutter was used in front of the aperture to control the Al
deposition time and to permit outgassing of the source without deposition.
The source was degassed for hours at a 60 % power level compared to the
operating conditions used for evaporation. Typical operating conditions were
achieved at 2.5 V, 1.4 A, and 3.5 W, under which conditions the alumina tube
assembly glowed with a bright orange-yellow color. Final outgassing, with the
shutter closed, was carried out for 30 s at the operating temperature. Operation
could occur for 5–10 min before the pressure rose to 5 × 10−10 Torr. Under these
conditions, deposition occurs at a rate of about 0.5 ML/min. Auger spectroscopy
and X-ray photoelectron spectroscopy both indicated that impurities on the evap-
orated Al film were below the detection limit. The source could be operated
reproducibly for several months before failure. Since Al melts at the operating
temperature, the source should be mounted horizontally.
An improved design for this type of source has recently been reported [20]. Two
improvements are incorporated to outgas the source and its supports more thor-
oughly. First, the heating leads wrapped around the Al2O3 are mounted on
degassing loops as described on p. 407 and 437; secondly, the entire source is
surrounded by a Ta radiation shield that can be separately degassed by means of
electron bombardment from a W filament mounted inside the shield, somewhat as
shown on p. 407. After degassing by these methods, the source could be operated
for hours in the 10−11 mbar range [20]. In addition, the outside of the Al2O3 tube
was modified by grinding a helical channel along its length. This reduced the mass
of the tube and increased the thermal contact between the heater filament and the
tube. Finally a Chromel-Alumel thermocouple was connected to a Pt sphere that
itself was attached to the tube by melting a Pt wire with a hydrogen flame and
observing that a portion of the molten Pt drops away while another portion wets the
alumina [20]. The thermocouple reads a temperature hundreds of degrees Kelvin
below the Al temperature but this may still be used to control the Al evaporation
rate even when the thermal contact between the heating wire and the alumina tube
varies [20].
Often the methods used for metal film evaporation require large, high-current
evaporation devices that are incompatible with experimental apparatus designed to
study the films themselves. Both heating and large outgassing effects are associated
with many of the high-power evaporation devices. Therefore, very simple designs
will often give better results in research experiments that do not require massive
film deposition capabilities. The Cr evaporator discussed here is one example of
ultimate simplicity which meets these requirements [21].
Two sizes of W wire loops, 0.25 and 0.5 mm in diameter, are used. These wires
are cleaned with 600 grit sandpaper followed by ultrasonic cleaning in acetone and
methanol, and then are bent into a hairpin shape as shown in Fig. 38.8. The hairpin
filaments are suspended in an electroplating bath with about 2-mm submersion. The
electroplating bath is of the following composition [22]:
100 ml H2O, deionized
30 g CrO3
5 g NaOH (add slowly)
0.6 g H2SO4
0.1 ml ethanol
Electroplating is carried out with a Pb anode. Currents of about 45 and 90 mA
are used for the small and large filaments, respectively. The plating voltage is about
3 V. Electroplating for 18 h typically deposited 75 and 250 mg of Cr on the small
and large loops, respectively. Before use, the filaments are rinsed in deionized water
38.8 Cr Evaporator Sources of Compact Design 449
and methanol. After installation, degassing is done before bakeout at 5–6 A and
15–16 A for the small and large filaments, respectively. The filaments are outgassed
a second time during chamber cooling after bakeout, and may then be used in
ultrahigh vacuum without appreciable gas evolution in the 6 × 10−10 Torr range.
The small evaporators yielded Cr film growth rates of 4 Å/min at a current of 7 A
for a distance of 5 cm. The larger evaporator yielded 4 Å/min at a 10 cm distance at
20-A heating current. These evaporation rates could be increased by a factor of
10–20 by working at higher filament currents. Both filaments produced about
4000 Å of total Cr film thickness before depletion. Auger spectroscopy showed that
impurities were not present at the limit of Auger detection if the sources were
thoroughly outgassed before use. For early stages of evaporation from a poorly
outgassed filament, oxygen impurities were seen. The films produced are of suffi-
cient purity to be used in studies of the magnetic properties of multilayer structures
[23].
beam evaporation methods is the emission of both high- and low-energy stray
electrons, which cause effects such as:
1. Electron stimulated desorption from walls and the cold shroud which may lead
to film contamination.
2. Film damage.
3. Interference with current measuring instruments, such as ionization gauges,
mass spectrometers and electron diffraction apparatus.
Of the two types of electron gun evaporators (magnetically and electrostatically
focused instruments), the magnetically focused instruments are more susceptible to
the difficulties mentioned above [24]. Figure 38.9a shows a schematic diagram of
an electrostatically focused electron gun. A thoriated W filament of circular
geometry is located between a shield and a cage, and both are biased at between −5
and −10 kV (−HV). The filament is heated by a 12-V, 30-A power supply.
Electrons are emitted and follow trajectories to the grounded hearth containing the
sample to be evaporated. A focusing mechanism (not shown) can change the rel-
ative position between the hearth and the rest of the components of the gun,
permitting an exact focusing of the electron beam on the top of the molten metal in
the hearth. A second degas mode involves keeping the filament at −3 kV and
grounding the remaining components of the electron gun evaporator. Out-gassing
was reduced by making all components out of refractory materials, such as W, Mo,
and Ta. The shield and hearth were made of Mo; the cage was constructed of Ta
mesh on a Mo frame with a W top cover containing a central hole. Using a Ta mesh
cage prevents heat buildup, and radiation-induced heating of large surfaces leading
to outgassing.
The secondary electrons produced in the gun cannot escape from the cage, which
is biased at −HV. The high-energy electrons have a potential V in the range
−HV < V < −HV + Vf, where Vf is the potential drop across the emitter filament due
to its resistance. In addition, there are high-energy backscattered electrons, which
mostly are at a potential of V or a slightly lower potential owing to energy loss upon
scattering. Since the potential at the hole in the cage can be positive relative to −HV
by a few hundred volts, backscattered electrons will escape from it. To eliminate
this problem, a tungsten mesh grid with 2-mm spacings was placed over the hole to
retard electrons there. In addition, by biasing the filament at −9.9 kV, excellent
focusing is achieved and stray electron currents are minimized, as depicted in
Fig. 38.9b, where calculated trajectories for cylindrical symmetry are shown.
This evaporation source can operate for refractory metal evaporation at pressures
<2 × 10−9 Torr [24]. A cryogenically shrouded source of somewhat similar design
has been described [25]. A highly stable electron beam evaporation source has been
described in [26]. A method involving the measurement of the ion current generated
by ionization of evaporating metal atoms and relating this measurement to the rate
of metal evaporation is described in [27].
An excellent review of metal evaporation methods and materials compatibility at
high temperatures is available in [28].
38.10 Small Electron Beam Evaporator for Refractory Metals 451
The device is simple to build, does not require cooling water for its components,
operates in the 10−9 Torr range, and produces clean films of refractory metals.
The evaporation device is mounted on a 2.75-in.-diameter flange containing
high-voltage feedthroughs. A tungsten emitter filament, about 1 cm in length and
0.012 cm in diameter, is spot-welded on 0.1-cm-diameter Ni rods as shown. It is
positioned 1–2 cm from the evaporation source. The source is a 6-cm wire
(0.025-cm diameter) made of the refractory metal to be evaporated that is fashioned
38.10 Small Electron Beam Evaporator for Refractory Metals 453
into a loop on the outer end; the base of this wire is welded to a 0.1-cm-diameter Ni
rod connected to a feedthrough pin by a set-screw contact for easy replacement. The
source wire is placed on the axis of a Mo wire helix with a diameter of about
1.5 cm, made of 0.025-cm-diameter Mo wire.
For operation, the source wire is set at about +6 kV = Va, and the grid is set at
about −3 kV = Vg. The filament voltage, Vf, is slowly raised using a current supply
until the collected current at the evaporator source is in the 5 to 20-mA range. Next
the negative voltage on the grid is empirically adjusted until electron bombardment
heating occurs only on the loop of the source wire. Overheating the loop can result
in melting which is undesirable. Films with greater than 1000-Å thickness have
been evaporated from W, Ta, and Ir at rates in the range 1–10 Å/min.
A variation on this source in which a small Mo crucible replaces the pendant
droplet source described here has been designed and tested [30]. The mass of the
crucible is kept to a minimum, and the emitter coil is wound so that it encircles the
crucible edge and is slightly below the edge [30].
An excellent review of metal evaporation sources and materials compatibility at
high temperatures is given in [31].
The cover of this second edition features this simple metal evaporator source
because of its wide application.
A design [32] differing somewhat from those shown elsewhere in this section, which
focuses a high-energy electron beam at a point on the end of a source wire for the
purpose of evaporating a metal is shown in Fig. 38.11. The metal source wire is
1–2 mm in diameter, mounted on a high-voltage Cu feedthrough that can be
translated in the vertical direction. In normal operation, a molten droplet of metal is
produced on the end of the wire by electron bombardment. The circular electron
emitter filament, made of 0.25-mm-diameter W wire, is looped over itself in a
15-mm diameter circle, and welded to two Ta wire leads (1-mm diameter), which are
connected to vacuum feedthroughs as shown. The distance between the plane of the
loop and the source tip is typically 6–9 mm. Nominal operating conditions are shown
in Fig. 38.11. As the source wire is evaporated, the beam current decreases and may
be restored by advancing the source to a position where the current returns to its
original value. With a source-to-sample distance of 15 cm, several hundred Å of film
may be deposited before source motion is necessary. The filament is shielded from
the substrate receiving the evaporated film by the aperture plate, and this reduces
radiative warming of the substrate. The source is readily outgassed behind a closed
shutter by running with the cooling water removed and at a reduced voltage (0.5–
1 kV). During normal operation, with water cooling, the pressure in the UHV system
remains in the 1–4 × 10−10 Torr range. Impurity transport to the film has been
measured for Fe and Co source wires using both Auger spectroscopy and X-ray
454 38 Evaporation Sources
photoelectron spectroscopy, and impurity levels below the detection limit were
found.
A design similar to this, but using a very small Mo crucible in the position of the
end of the source wire, has been described [33] for the evaporation of low-melting
metals under conditions where the low outgassing qualities of this evaporator are
retained.
A review of metal evaporation sources is contained in [34]. In addition, an
excellent review of metal evaporation sources and materials compatibility at high
temperatures is given in [35].
Often low-vapor pressure compounds are used as precursors for the growth of
compound semiconductor films on a substrate. Many of these compounds react with
stainless steel and therefore cannot be transferred through heated tubing. This
design places a glass ampoule of the compound close to the substrate so that the
vapors do not come into contact with metal prior to deposition [36].
A glass ampoule is filled with the compound in a dry box under an inert
atmosphere and the ampoule is sealed. The ampoule is rigidly attached to the end of
a transfer rod inside a copper carrier assembly that fits the ampoule (Fig. 38.12).
The copper assembly is wound with W wire, sheathed in fiberglass or quartz fiber
sleeving, and a thermocouple is placed in contact with the copper carrier for
temperature measurements during vaporization from the ampoule. The use of a
bellows transfer device is superior to a magnetically driven device, since the heater
and thermocouple wires are fixed along the transfer rod during motion and do not
have to flex each time the ampoule is moved.
A linear motion device with a stainless steel blade is used to break off the neck of
the ampoule when it is desired to start the vaporization process. Once the neck is
broken, the ampoule can be moved into the UHV chamber through a gate valve,
and the temperature of the doser can be adjusted to produce the desired rate of
precursor vaporization.
A cryopump is chosen for the first chamber because it pumps water vapor
efficiently and it is not impeded by glass fragments.
Thin films of CdS may be thermally evaporated from CdS powder. The apparatus
shown in Fig. 38.13 is made entirely of tantalum, tungsten, and boron nitride [37].
A large area baffled aperture produces a uniform CdS deposit. In addition the baffle
shields the substrate from radiation from the indirect heater. The baffle has a
3-mm-wide gap between its outer edge and the Ta shield.
The sublimation source is made of a Ta cup welded to a Ta tube. This slips over
a cylindrical shield made of BN, which contains a rigid coil of fairly heavy wound
W wire in three strands. The entire assembly may be unloaded by pulling upward
on the handle, leaving the W wire exposed. In operation, the CdS nearest to the
central heater sublimes first. The source temperature is measured with a Pt/PtRh 13
thermocouple welded to the baffle. At a temperature of 550 °C, CdS is deposited at
a rate of 7 Å s−1 on the substrate at 120 °C and 24 cm distant. The operation time at
this rate of evaporation is about 4 h before refilling is needed.
By choosing the correct evaporation and source temperatures, the resistivity of
the deposited CdS film may be controlled.
A CdS evaporation source using elemental evaporation has been described [38].
The use of elements from Groups II, III, IV, V, and VI in the production of semi-
conductor devices is widespread. Many of the elements sublime as atoms, but in
Groups V and VI, sublimation as dimers and tetramers is commonly observed. It is
advantageous—for example, in the growth of GaAs surfaces—to employ atomic As,
which forms surfaces with fewer defect sites than surfaces grown from As2 or As4
species. The device shown in Fig. 38.14 permits the substantial atomization of As2
and As4 by use of a final heated atomizer stage [39]. The principles embodied here
can also be used with other elements where atomization is desired. It should be
remembered that the equilibrium between atomic species and higher molecular
38.14 Arsenic Atom Source 457
weight species is favored toward the atomic species by high temperatures and by low
pressures.
Figure 38.14 shows, schematically, the device used to make atomic beams of As
containing small fractions of As2 and As4. A sublimation source, loaded with As(s),
is connected to a first-stage cracking furnace through an adjustable aperture.
Adjustment of this aperture allows the beam pressure of As species to be controlled
throughout the apparatus. The subliming species are conducted through the first
stage cracking furnace and then exit to a heated plate that serves as the second-stage
atomizer. The products from the second stage are significantly enriched in atomic
As, and are emitted from a second aperture toward the substrate. The atomizer can
be operated in the range above about 1200 K when the local pressure is below about
10−3 Torr; it can also be operated in the range above 1400 K when the local
pressure is about 10−2 Torr. Many materials can be used for the atomizer plate with
various elements to be atomized, including refractory metals such as W, La, Nb, Re,
Mo, and Ta, and stainless steel, Ni, graphite, Pt, B, Si, ceramics, and BN.
Compound formation between the material of the atomizer plate and the beam is to
be avoided.
The preferred range of operation for various materials is given in Table 38.1, and
substantial but not complete atomization will be produced under the listed
conditions.
38.15 Portable Microevaporator for Depositing Low Coverages of Single Atoms 459
Table 38.1 Preferred range Substance Pressure at atomizer Temperature range (K)
of operation—atom sources (Torr)
As <10−2 >1400
<10−3 >1200
P <10−3 >2000
<10−5 >1800
Sb <10−2 >1400
<10−4 >1200
S <10−2 >2000
<10−4 >1700
Se <10−2 >1600
<10−4 >1300
Te <10−2 >1300
<10−4 >1100
Fig. 38.15 Portable microevaporator for depositing low coverages of single atoms
The field of surface science is increasing concerned with the study of larger mol-
ecules with low vapor pressures. While a number of types of heated dosers are
described in this book none have the capability to deposit an accurately known flux
38.17 Quantitative Molecular Beam Dosing of Slightly Volatile Organics 461
Fig. 38.17 Quantitative molecular beam dosing of slightly volatile organics, a heated gas doser,
b vacuum system—heated, c reproducibility of vapor pressure measurements of naphthalene,
d enthalpy of sublimation for naphthalene, e mass spectrometer for response in UHV chamber
during dosing
38.17 Quantitative Molecular Beam Dosing of Slightly Volatile Organics 463
region is connected to the doser in the UHV system by a gas line which is heated
about 20 K above the source temperature in the oven. Flow to the heated doser can
be interrupted by pumping the source through the turbopump.
Figure 38.17c shows the measured vapor pressure of naphthalene in the tem-
perature range 294–346 K. Three separate measurements agree well. Figure 38.17d
shows a Clausius-Clapeyron plot of the vapor pressure; the slope yields the sub-
limation enthalpy of 71.0 ± 0.2 kJ/mole, which agrees well with the literature.
Figure 38.17e shows the behavior of the mass spectrometer in the vacuum
system as naphthalene vapor is admitted at various vapor pressures through the
heated doser system. The walls of the UHV system initially adsorb naphthalene
molecules, and the mass spectrometer observes this wall-adsorption during the
approach to steady state system pressure, which is achieved after about 400 s
dosing. However, a crystal placed in the molecular beam will see an immediate
steady flux of naphthalene and it was shown that the coverage increased in linear
proportion to the pressure of naphthalene as measured in the oven source multiplied
by the time of exposure [43].
A critical feature in the apparatus is a heated adjustable leak valve, 1, which is a
Granville-Phillips (series 203) variable leak valve. In the different experiments, the
leak valve is not adjusted to open or close it, since these valves are not reproducible
464 38 Evaporation Sources
upon cycling open and closed. It serves therefore only to adjust the fraction of the
incident vapor which passes. It can be adjusted initially to select the time required to
achieve a desired coverage on the single crystal but is not adjusted further in
quantitative dosing measurements. Its leak rate during our measurements was about
2 × 1014 molecules naphthalene/Torr s.
This doser method has been shown to work well using other low vapor pressure
organics [43, 44].
A simple doser for subliming volatile organic materials is made from a 1 in. ceramic
standoff with vented screws mounted in both ends. The top screw covers a small
reservoir containing the powdered material to be sublimed. The bottom reservoir is
empty and also contains a vented screw through which a thermocouple is forced
and spot welded. Heating is accomplished by passing current through 0.010″ W
(3 %Re) wire, although ordinary W wire would certainly serve too. A shutter may
be used to interrupt evaporation [45] (Fig. 38.18).
Many evaporators for organic materials are made of ceramic crucibles (e.g. boron
nitride) with ohmic heaters and thermocouples for achieving and measuring the
evaporation temperature. At least 3 electric feedthroughs are necessary for each
crucible if one leg of the ohmic heater is connected to ground. In this design, only a
single feedthrough is needed as the ohmic heater is also used as a temperature
466 38 Evaporation Sources
the Pt wire and the voltage drop (V) is measured. For the I/V ratio (R), the resistance
is obtained and used as a measure of the temperature by using the well-documented
polynomial of the Pt resistance versus T. For pulsed evaporation, the current is
increased using simple electronics, keeping the crucible at preset temperatures
before and after evaporation. The accuracy of the measurement of R and T can be
increased by using larger diameter Pt lead wires welded to the end of the resistance
wire, minimizing effects on the total circuit resistance by heating of the Pt lead
wires [47] (Fig. 38.20).
468 38 Evaporation Sources
References
1. The use of degassing loops originates from the thin film community but the historical origin of
this useful method is not known
2. Professor R.J. Behm (personal communication)
3. W.F. Leonard, Rev. Sci. Instrum. 40, 187 (1969)
4. D.A. Steigerwald, W.F. Egelhoff Jr, J. Vac. Sci. Technol. A7, 3123 (1989)
5. K.J. Ross, B. Sonntag, Rev. Sci. Instrum. 66, 4409 (1995)
6. K.K. Chin, P. McKernan, I. Lindau, J. Vac. Sci. Technol. A4, 1949 (1986). This paper
contains references to tests of this device as well as to related devices
7. K.J. Ross, B. Sonntag, Rev. Sci. Instrum. 66, 4409 (1995)
8. F. Jansen, J. Vac. Sci. Technol. 21, 106 (1982)
9. P.R. Davis, H.R. Poppa, J. Vac. Sci. Technol. 15, 1771 (1978)
10. P.J. Chen, M.L. Colaianni, J.T. Yates Jr, Phys. Rev. B 41, 8025 (1990)
11. K.J. Ross, B. Sonntag, Rev. Sci. Instrum. 66, 4409 (1995)
12. H.P. Bonzel, A.M. Bradshaw, G. Ertl (ed.), Physics and Chemistry of Alkali Adsorption
(Amsterdam, Elsevier, 1989)
13. F.J. Esposto, K. Griffiths, P.R. Norton, R.S. Timsit, J. Vac. Sci. Technol. A12, 3245 (1994);
for details of the surface behavior of Li/Al alloys
14. F.J. Esposto, C.-S. Zhang, P.R. Norton, R.S. Timsit, Surf. Sci. 290, 93 (1993)
15. Supplied by Alcan, International, Ltd. It is rumored that Canadian beer cans are made of this
alloy (personal communication, P.R. Norton)
16. R.P. Schorn, E. Hintz, S. Musso, B. Schweer, Rev. Sci. Instrum. 60, 3275 (1989)
17. G. Polanski, J.P. Toennies, Surf. Sci. 260, 250 (1992)
18. W.J. Wytenburg, R.M. Lambert, J. Vac. Sci. Technol. A10, 3597 (1992)
19. Degussa Limited, Wilmslow, Cheshire, England
20. H.-G. Keck, E. Kopatzki, and R.J. Behm (private communication)
21. J.J. McClelland, J. Unguris, R.E. Scholten, D.T. Pierce, J. Vac. Sci. Technol. A11, 2863
(1993)
22. H.F. Mar et al. (eds.), Encyclopedia of Chemical Technology (Wiley, New York, 1979), vol. 6,
p. 104
23. J. Unguris, R.J. Celotta, D.T. Pierce, Phys. Rev. Lett. 67, 140 (1991)
24. M. Heilblum, J. Bloch, J.J. O’Sullivan, J. Vac. Sci. Technol. A3, 1885 (1985)
25. L. Ramberg, E. Flemming, T.G. Andersson, J. Vac. Sci. Technol. A4, 141 (1986)
26. Y.W. Beag, K. Egawa, R. Shimizu, Rev. Sci. Instrum. 64, 3647 (1993)
27. T. Jones, J. Sawler, D. Venus, Rev. Sci. Instrum. 64, 2008 (1993)
28. K.J. Ross, B. Sonntag, Rev. Sci. Instrum. 66, 4409 (1995)
29. J.R. Waldrop, R.W. Grant, J. Vac. Sci. Technol. A1, 1553 (1983)
30. J. Araya-Pochet, G.A. Mulhollan, J.L. Erskine, Rev. Sci. Instrum. 62, 2288 (1991)
31. K.J. Ross, B. Sonntag, Rev. Sci. Instrum. 66, 4409 (1995)
32. B.T. Jonker, J. Vac. Sci. Technol. A8, 3883 (1990)
33. J. Araya-Pochet, G.A. Mulhollan, J.L. Erskine, Rev. Sci. Instrum. 62, 2288 (1991)
34. L.I. Maissel, R. Glang, Handbook of Thin Film Technology (McGraw Hill, NY, 1970), pp. 1–51
35. K.J. Ross, B. Sonntag, Rev. Sci. Instrum. 66, 4409 (1995)
36. Dr. M.S. Jackson, J.G. Ekerdt, Department of Chemical Engineering (The University of Texas
at Austin, Austin, TX 78712-1062) (private communication)
37. R.T. Galla, Rev. Sci. Instrum. 36, 403 (1965)
38. D. Beecham, Rev. Sci. Instrum. 41, 1654 (1970)
39. Professor Stephen R. Leone, University of Colorado at Boulder, Joint Institute for Laboratory
Astrophysics, Boulder, CO 80309-0440 (private communication). U.S. Patent No. 5,541,407
Arsenic Atom Source, S.R. Leone, A.L. Alstrin, P.G. Strupp, A.K. Kunz, S.M. Casey, and
R.V. Smilgys
References 469
40. H.-P. Rust, T. Konig, G.H. Simon, M. Nowicki, V. Simic-Milosevic, G. Thielsch, M. Heyde,
H.J. Freund, Rev. Sci. Instrum. 80, 113705 (2009)
41. K. Lammle, A. Schwarz, R. Wiesendanger, Rev. Sci. Instrum. 81, 053902 (2010)
42. M. Feng, J. Lee, J. Zhao, J.T. Yates Jr, H. Petek, J. Am. Chem. Soc. 129, 12394 (2007)
43. L. Thompson, J.-G. Lee, P. Maksymovych, J. Ahner, J.T. Yates Jr, J. Vac. Sci. Technol. A21,
491 (2003)
44. J. Lee, O. Kuzmych, J.T. Yates Jr, Surf. Sci. 582, 117 (2005)
45. K. Vijalakshmi, R.A. Rosenberg, private communication
46. N. Sato, K. Seki, H. Inokuchi, Rev. Sci. Instrum. 58, 1112 (1987)
47. P. Feulner, private communication
Chapter 39
Ions
Fig. 39.1 Solid state Cs+ ion gun + other alkalis. a Ion gun design. b Performance. c Zeolite
source Performance for various alkalis
aration and evaluation of the various sources. Figure 39.1c shows the saturation
current density for the four zeolite emitters coated on Pt surfaces as a function of
emitter temperature, measured in a simple diode arrangement in which the ion
emission extrapolated to zero bias was evaluated. An approximately 0.5-cm2 area of
Pt mesh emitter was used. A Rb+ ion source, operating with a mixture of Rb+-
substituted aluminosilicate zeolite and powdered Mo has been described. Here too a
Pierce-type electrostatic focusing structure was used [15].
The investigation of the Cs+ purity from various composition aluminosilicates
has been reported [16].
The design of a high-quality ion gun for delivering Li+ ions in the energy range
1–100 eV with an energy spread of only 0.22 eV has been published [17].
Although high-temperature sources for production of the alkalis are well known, the
use of similar alkaline earth sources is less well established. Recently, a study of the
oxides MgO, CaO, SrO, and BaO has been made, and the purity of the emitted
Mg+, Ca+, Sr+, and Ba+ ions have been investigated [18].
Filaments (0.25-mm diameter) of either W or Ta were used to support the oxides
in front of a quadrupole mass spectrometer as shown in Fig. 39.2a. This was done
by making a slurry of the powdered oxide in distilled water and transferring powder
to the metal coil. Tungsten wire was used for CaO and MgO, and Ta wire was used
for the other oxides, including MgO in later studies, since it was found that MgO
reacted with W. Measurements were made of the total ion emission as well as the
purity of the emitted ions as a function of the temperature. Temperature measure-
ments were made with a pyrometer.
474 39 Ions
Fig. 39.2 Alkaline earth ion sources. a Ion detection by mass spectrometer. b Ion current versus
T. c BaO/Ta. d SrO/Ta. e CaO/W
Figure 39.2b shows the ion current measured for three of the filaments as the
temperature is raised. Currents as high as 1 μA are observed at elevated tempera-
tures, but failure of the filaments occurs rather rapidly at extreme temperatures
owing to the reaction of the metal oxide with the refractory metal filament.
Figure 39.2c–e show the impurity behavior for BaO, SrO, and CaO as a function of
the temperature. In all cases, alkali metal ion impurities are observed, even though
the oxides were of purity of 99.9 % or better. This is due to the highly preferred
39.2 Alkaline Earth Metal Ion Sources 475
surface ionization of the alkalis as a result of their low first ionization energy. As the
filaments are heated to higher temperatures, the alkali ion impurities decrease in
yield and rather pure Ba+, Sr+, and Ca+ ions are liberated. The purity of Mg+ from
MgO coatings was less than 50 % at all temperatures, and this would therefore be
an unsuitable source for Mg+.
References
Atomic hydrogen is employed for research in the surface science field as a reagent
for rapidly producing chemisorbed hydrogen on semiconductor surfaces and also on
metal surfaces that do not readily dissociate molecular H2. The surface chemistry of
atomic hydrogen has been investigated as early as the work of Langmuir [1, 2] who
first produced it on a hot tungsten filament. The production of atomic hydrogen has
been investigated by a number of workers [3–5]. It has been found that the effi-
ciency of H atom production is proportional to (PH2)1/2 at H2 pressures above about
10−6 Torr where pseudoequilibrium conditions exist. At lower H2 pressures, the rate
is proportional to PH2. Under these conditions, the reported efficiency of H2 dis-
sociation is about 0.3 per H2 collision with the tungsten surface at temperatures
above 2500 K [3]. It has been observed that the efficiency of dissociation is
decreased by impurities often present on a tungsten filament [4].
Two sources that place a relatively high H2 gas density near a heated tungsten
surface have recently been described [6, 7]. Both of these sources produce large
fluxes of atomic hydrogen compared to a bare tungsten filament operating in the
vacuum system, if comparisons are made at the same gas throughput to the system’s
pumps. In Fig. 40.1a a tungsten capillary (0.6-mm ID, 1.6-mm OD) is electron
bombarded from a tungsten filament so that the end of the capillary is heated to
1800–1815 K over a 4-mm length. An acceleration voltage of +560 V on the
capillary (HV) and a 30 mA emission current achieve this condition. A Cu shield,
with a conical hole drilled on axis, serves as a radiation shield. It is cooled by a
water circulation tube that is butt-brazed to the solid Cu shield. The sample is
placed 10 mm in front of the shield, or 20 mm from the end of the W capillary. The
back end of the W capillary is snugly fitted into a ceramic sleeve, which also
connects to a 4-mm-diameter stainless tube that delivers H2 gas to the atomizer.
Since the gas density is low during operation, no need exists for vacuum tight seals,
and the adapter uses only friction fits to the capillary and the tube. The working H2
pressure at the entrance to the W capillary is about 10−2 Torr. In order to change the
position of the H atom doser, it is mounted on a translational device. It is reported
that a total atomic H flow of about 3 × 1014/s may be achieved under these
conditions. The degree of dissociation approaches 50 %. Therefore the total gas
influx is small, minimizing the background H2 pressure. It is estimated that about
20 % of the H atoms are emitted in a central portion of the beam with a cone angle
of ±15° [8, 9].
In Fig. 40.1b, a spiral W filament (1.5-cm diameter, 0.025-cm-diameter wire,
14.5-cm length) [10] is located within a cup-shaped stainless steel doser having a
perforated array in its central portion [7]. Hydrogen gas flows into this doser
through a tube that is connected to a calibrated conductance aperture (see pp. 89–
91; 421–428), and the flow rate is adjusted using a capacitance manometer on the
high-pressure side of the aperture. A typical flow rate of about 2 × 1016 H2/s is
employed, at a system pressure of 2 × 10−7 Torr. The incoming H2 encounters a
deflector baffle that produces a random flux on the back of the perforated array. The
sample is shielded from radiation by the wall of the doser body. In order to reflect
the H atoms to the sample (avoiding radiation heating of the sample as well as
possible impurities from the hot filament) a cooled Pyrex plate at 120 K is located
close to the outlet of the atomic hydrogen source. This reflects atomic H with high
efficiency [11, 12]. The Pyrex is held onto a stainless steel cooling reservoir with
metal clips and is thermally connected to the l–N2 reservoir with a film of In metal.
In studies of the efficiency of the doser [7], it was found that a reduction in
efficiency of a factor of 4.8 occurred using the reflector source compared to a
line-of-sight geometry. However, experiments working at the same system gas
throughput indicate that the doser produces about 50 times more atomic hydrogen
than a bare W filament, operating without a doser geometry. Thus, even the factor
of 4.8 reduction in efficiency owing to the reflector is compensated by the higher
atomic H production rate achieved at the high gas densities in the doser. This source
is therefore about a factor of 10 more efficient than a bare W wire in delivering
atomic H to the sample for identical H2 gas loads. A monolayer of H atoms (about
4 × 1014/cm2) was adsorbed by a Si(100) surface at 400 K when 4 × 1018 H2
molecules had passed through the doser with the W filament operating at 2000 K.
Figure 40.1c shows the relative efficiency of atomic H production plotted versus
the reciprocal temperature (K−1). The source produces atomic H at a rate which
reaches a maximum near 2000 K. Similar results were reported by George et al.
[13].
An interesting study, comparing the pumping speed of atomic H with that of H2
during adsorption on engineering surfaces in the range 2.5–4.3 K, has been pub-
lished [14]. Interestingly, the pumping speed for H2 is between a factor of 3 and 40
above that of atomic H. This difference in either condensation or adsorption is
attributed to the higher internal energy of the atomic H compared to H2.
A hydrogen atom source involving a design similar to that of [6] has been
carefully evaluated to determine the efficiency of atomic H production [15].
A second similar source has also been evaluated by mass spectrometric methods
40.1 High Flux Atomic Hydrogen Sources—Thermal 479
Fig. 40.1 High flux atomic H sources. a H atom source—hot tube design. b H atom source—
spiral filament design. c Efficiency of design B for H-atom formation
480 40 Active Gases and Species
[16]. In addition, atomic H beams, cooled by collision with surfaces near 6 K, have
been produced [17].
A number of detectors for atomic H have been devised and are reviewed briefly
in [18]. One of these [18] involves the use of a thermistor (primarily NiO) for
measurement of an atomic H flux. Electrical conductivity changes, measured with a
Wheatstone bridge and phase-sensitive detection, were used for the evaluation of
the flux of atomic H made in a microwave discharge. Control experiments elimi-
nated the possibility of effects owing to the influence of heating and radiation on the
detector. The mode of operation of the detector is unknown and could be due to
local heating as a result of H atom recombination or to electronic surface effects.
The lower limit of sensitivity is estimated to be 1011 H/cm2 s.
An atomic H source involving the use of a W/26 % Re alloy tube ohmically
heated to 2400 K has been described. The W/26 % Re alloy does not have the brittle
properties of pure W [19].
Many interesting uses for atomic H have been described in the literature and are
beyond the scope of this book.
The use of thermal sources for the production of atomic H has been described on
p. 477. It is also possible to produce beams of atomic H using microwave discharge
sources. These sources are convenient to utilize but one must be cautious that
impurities and radiation from the discharge region do not affect the experiments.
There are many different RF discharge sources available, as described in [20].
A typical source not unlike others that have been made [20] is described below.
Figure 40.2 shows a diagram of a Pyrex glass source of simple design. It is
capable of producing atomic H with up to 95 % dissociation of H2(g). The density
of atomic H ranges from 5 × 1011 to 2 × 1013 H/cm3. The tube is 18-mm inside
diameter and 240 mm long. It is terminated at each end by a 1-mm-ID capillary to
prevent the discharge from spreading outside the source, and it is cooled by a water
jacket to reduce the temperature and to reduce H atom recombination on the inner
walls [21]. This may be responsible for the long life and stability of this design.
The discharge is excited by about 25 W of RF power into a coaxial cavity
resonant at 35 MHz. The cavity consists of a copper cylinder, surrounding the
discharge tube, and a 12-turn helix of constant pitch. Power is coupled via a
single-turn link coupling. Details of cavity design and cavity dimensions have been
published [22]. The coaxial cavity is properly impedance-matched to the generator
to insure 90 % power absorption into the cavity. The RF generator is properly
designed to prevent the propagation of unwanted RF signals. The RF oscillator and
driver stages are housed within individual screened compartments. The amplifying
stages are narrow band with sharply tuned filters in each stage, eliminating har-
monics. Notch filters are used in the power input to prevent RF propagation into the
laboratory AC supply. Low noise was achieved in auxiliary instruments when the
RF supply was operating.
A hydrogen pressure of about 0.15 Torr was used in the source. Great care was
taken to clean the discharge tube before use, successively washing with hot chromic
acid (in sulfuric acid), water, acetone, HF(aq), and distilled water. A palladium leak
(see p. 511) was used to purify the H2 before use and back diffusion of contami-
nants was avoided. The tube operated for about 5000 h without trouble.
The delivery of reactive atomic species in high purity to surfaces poses one of the
difficult experimental problems in adsorption and surface reaction studies. Species
such at atomic F are particularly difficult to prepare and transfer without secondary
reactions on the walls of the source. An excellent design [23], probably applicable
in general to producing a number of reactive species, is shown in Fig. 40.3.
A low pressure 10 %F2/90 % He mixture is flowed through an alumina tube
which passes through an Evanson-type microwave discharge cavity, and the
products are then pumped through the apparatus using a rotary pump which is
corrosion resistant. The microwave discharge products are admitted to the prepa-
ration chamber through a 25-μm-diameter orifice mounted on the end of the outer
stainless steel tube. The orifice was laser-drilled (see p. 421) in a stainless steel
sheet of 25 μm thickness. The alumina tubing inside the source is thought to
become passivated with the formation of a nonvolatile layer of AlF3, and is superior
to a tube made of silica [23, 24]. The microwave discharge source [25] is operated
at 30–60 W of power in the total pressure range of 0.05-10 Torr.
The species emitted from the orifice were studied with a modulated quadrupole
mass spectrometer, which measured the consumption of F2 (38 amu). The results
show that essentially complete dissociation occurs below a total pressure of 2 Torr,
as shown in Fig. 40.3b. When working in the molecular flow regime below a
pressure of 5.7 Torr, the angular distribution is described as having a cos2 distri-
bution, and for sample sizes of the order of 0.5 cm at source-to-sample distances of
482 40 Active Gases and Species
about 1 cm, the fluence is rather uniform. It is calculated that at a total pressure of
0.7 Torr, the source delivers F atoms at a flux of about 1 × 1014/cm2 s in the beam
with a random flux from the source and the chamber being at about 5 % of this
beam flux.
Microwave sources are widely used in molecular beam studies [24, 26], but this
source is one of the first such sources to be successfully employed in surface studies
[27–29]. It is also useful for generating chlorine atoms. For chlorine, a quartz tube
may be used instead of an alumina tube [30, 31].
It should be noted that large gas loads accompany the use of this source, and that
a rapidly pumped preparation chamber, isolated from the ultrahigh vacuum surface
analysis chamber, is one way to reduce the gas load in the analysis chamber when
the source is being used.
40.4 Alkali Metal Sources 483
The alkali metals form one of the most interesting class of adsorbates on the
surfaces of both metals and semiconductors [30]. As a result they have been widely
studied since the pioneering work of Langmuir in the 1930s [32]. The alkali metals
are also used in the production of low-work-function photocathode surfaces, and in
other applications in the field of physical electronics. As a result, excellent com-
mercial sources of alkalis for research are available from SAES getters [33].
Figure 40.4a shows a view of a typical SAES alkali getter source; these are
available to deliver Li, Na, K, Rb, and Cs. A powdered mixture of the alkali metal
chromate, M2CrO4, and a reducing agent (St 101 alloy—84 % Zr and 16 % Al) is
held in a nichrome boat with a very narrow slit running along its length, as shown.
A nichrome retaining wire is squeezed along the backside of the slit to prevent
powdered material from being ejected from the slit. Butting pieces of two designs
are described, which block the ends of the boat and which serve as connectors for
electrical heating. As current is passed, the temperature rises into the 550—850 °C
region, as indicated in Fig. 40.4b, and the solid state reduction reaction liberates the
alkali metal which escapes from the slit. The Zr also acts as a getter for impurity
gases such as H2 and CO. The typical operating temperature for deposition must be
adjusted on the basis of experience, but it is usually preferable to use lower rates for
controlled deposition experiments. The typical amount of H2 emitted from one of
the sources is about 1 Torr-cm3 per cm of getter length, and degassing is achieved
by initially working at 500 °C. Various geometrical shapes and sizes are available.
The solid state oxidation-reduction reaction between the reducing agents, Zr and
Al in the metal alloy reactant can be written as shown in Fig. 40.4c, but the exact
details of the reaction are probably not well worked out. An excess of the Zr-Al St
101 alloy is employed so as to have material for the gettering of active gases during
the reaction.
In addition to alkali metal getters, SAES also makes alkaline-earth Ba getter
sources of a different configuration from the alkali metal sources (Type St
15A/20HY/400 and St 15/20KLS/200). The reactant consists of 45 % BaAl4 alloy
mixed with 55 % Ni powder. This composition, on being heated to about 850 °C,
reacts exothermically instantaneously releasing about 25–30 % of its Ba. Continued
heating by RF methods releases the remainder of the Ba. These sources are used to
process vacuum tubes by throwing a heavy getter film of Ba, which is active for CO
[34], H2 and other gas adsorption. An SAES source of Ba geometrically similar to
the alkali sources is the type St 14/Fr, which can be ohmically heated as is done
with the alkali sources.
A description of an alternate design for a Cs ion source, involving diffusion of
Cs through a heated porous W plug, is given in [35].
Alkali metal sources may also be based on the direct evaporation of the metal,
and many such sources have been described. A novel source, which delivers large
quantities of Na as an atomic beam, has been described; it employs a metal mesh
wick to return excess Na to the evaporator section. This source also works under
484 40 Active Gases and Species
Fig. 40.4 Alkali metal sources, a Structure of SAES alkali metal dispenser. b Typical
temperature-current performance. c Solid state reaction—using Zr or Al
40.4 Alkali Metal Sources 485
ultrahigh vacuum conditions [36]. In addition, zeolite sources containing alkalis are
often employed, and one of these is discussed on p. 471.
is a fairly clean reaction, and if gas phase reactions of the methyl radicals are
avoided, then only surface reactions in the source can contribute impurities.
Figure 40.5 shows a schematic design of the pyrolysis apparatus.
Dimethyldiazine is allowed to pass through a quartz tube that has its central section
heated by means of a closely wound Ta wire, 0.25-mm diameter, which is cemented
to the outside of the pyrolysis tube using a ceramic cement (Aremco Type 576). The
quartz tube is connected to the source of dimethyldiazine using a graphite ferrule
[43] in a swagelock reducing union. The thermocouple embedded into the heating
section allows the temperature to be adjusted to the optimum one for complete
conversion of dimethyldiazine to pyrolysis products, which occurs near 1200 K.
A cooling section beyond the pyrolysis region allows the CH3 radicals to become
accommodated at a lower temperature than their production temperature. Various
lengths of the three zones shown in Fig. 40.5b were studied [42]. A typical mass
spectrum for dimethyldiazine and for the completely pyrolyzed product is shown in
Fig. 40.5c. In addition to CH3, measurable fractions of CH4, C2H6, and H2 are
produced along with N2 at 1173 K and a chamber pressure of 5 × 10−7 Torr.
The dimethyldiazine was synthesized using the method of Renaud and Leitch
[44] and was stored in a stainless steel container over a period of several months
without decomposition. Ar was mixed with the dimethyldiazine as an internal
standard for quantitative flow measurements [42]. A quadrupole mass spectrometer
was used to acquire the mass spectra shown, working inside a stainless steel
chamber pumped by a turbomolecular pump [42].
486 40 Active Gases and Species
Fig. 40.5 Methyl radical sources. a Methyl radical source. b Pyrolysis tube. c Mass spectrum of
gas from source
The ozone molecule, O3, is a powerful oxidizer, useful in situations where oxida-
tion of surfaces is needed under mild conditions. It has been used, for example, in
the growth of Ba2YCu3O7-δ thin films [45, 46]. However, when stored as the liquid
at cryogenic temperatures, O3 is dangerous and may explode. The apparatus
described here avoids this problem by storing O3 by adsorption in the pores of silica
gel at temperatures slightly above its boiling point [47, 48]. The rate of decom-
position under these storage conditions is about 5 % per week.
40.6 Production, Storage, and Use of Ozone 487
The system is equipped with O3 destruct tubes containing copper sponge that
catalytically decomposes O3. These are also available commercially. The rotary
pump oil should be non-hydrocarbon-based, using either Fomblin [51] or Krytox
[52]. Teflon O-rings should be used in all locations, and not rubber O-rings, since
ozone is well known to destroy rubber by attacking carbon-carbon double bonds.
Extreme care should be taken to keep the dry ice + acetone trap cold at all times
when O3 is trapped on the silica gel.
An alternate method [53] employs a cryopump controlled by a thermocouple
feedback to boil off O3 from liquid O3 at a precisely controlled temperature. The O3
is condensed in a vacuum-jacketed reservoir whose temperature is precisely
adjustable through the operation of a cryopump. These authors provide a rupture
disk for this O3 storage system in case of an explosion, which is a possibility for
liquid O3 stored in this manner.
A simple electrical discharge apparatus for producing 100 % pure O3 in a
nonflow system, and for continued repurification, has been described. Less than
100 cm3 of pure oxygen at atmospheric pressure is converted in a batch operation,
and purification occurs by pumping traces of residual O2 away at 77 K [54].
The calibration of a mass spectrometer for O3 and the measurement of its
cracking pattern in a magnetic mass spectrometer has been carried out carefully,
using a 5.3-L storage volume (Teflon coated) to receive the O3 from a cold trap after
pumping off impurity O2, etc. The storage volume was the source for a molecular
beam of O3 that entered the ionizer of the mass spectrometer. It was found that the
half-life of O3 at a pressure of 0.1 Torr in the storage volume was about 12 h at
room temperature [55].
Ozone is widely used in surface chemistry and materials chemistry as a high power
oxidizing agent. In Sect 40.6, we show how to make relatively large quantities of
very pure laboratory ozone and how to quantitatively dose it into an ultrahigh
vacuum system. For occasions where small quantities of ozone are needed in an
ultrahigh vacuum system, a simpler doser working on a different principle [56] may
suffice as shown in Fig. 40.7.
Oxygen gas at a pressure less than about 100 Torr is admitted into a glass tube
specially constructed for the preparation of O3 and for the removal of excess O2.
The glass reservoir is immersed deeply in a liquid nitrogen bath and a Tesla coil is
used to convert O2 to O3. A dark blue liquid O3 will be seen to form on the inner
walls and to flow to the bottom of the reservoir. Several doses of O2 may be used to
build up the quantity of collected O3. The conversion process is aided by the
presence of a small amount of Cu wool in the neck of the preparation/storage
reservoir and the discharge from the Tesla discharge is directed into the gas by the
Cu. The discharge should be done through the liquid nitrogen in the bath.
Following partial conversion of O2 to O3, the excess of O2 is distilled away at liquid
40.7 Ozone Generator and Molecular Beam Doser Source 489
nitrogen temperature being aware that liquid O3 has an appreciable vapor pressure
of a few Torr at 77 K and will also be removed slowly. The pump used for
distillation should not contain any oxidizable oil.
Admission of the O3 to the doser is accomplished by using a stainless steel
metering valve which allows the stored O3 to slowly leak into the glass beam doser
assembly and into the UHV system. The use of glass between the storage reservoir
and the Pyrex transfer tube assures minimal wall decomposition of the O3 by
collision with metal surfaces. Passivation by exposure to O3 will deactivate the
small amount of metal exposed. It will be found that even the Pyrex walls act
initially to decompose the O3 to some degree (see Sect. 40.6), and only by pas-
sivation in O3 will the rate of wall decomposition approach zero. To shut off the
doser beam at the UHV system, a small gate valve is placed into the middle of the
two section-glass doser assembly. Since it will be useful to have a way to monitor
the rate of flow of O3 into the ultrahigh vacuum system it is suggested that one uses
the response of the ion gauge in the UHV system to the flow of O3 to be monitored.
The O3 which is not adsorbed on the sample becomes randomized by collision with
the walls of the stainless steel system and will all decompose to O2 and the ion
gauge therefore measures an increase in O2 pressure which is proportional to the
beam flux of O3.
490 40 Active Gases and Species
Silver reacts only slowly with molecular oxygen, while its reactivity with O3 is very
rapid. In the apparatus shown in Fig. 40.8a, O3, originating from liquid O3
vaporization in an O3 cryogenic storage system, is incident on a Ag film (500 Å
thickness) evaporated onto a quartz crystal microbalance [57] (see p. 486 and 309).
The rate of increase of mass is proportional to the O3 flux in the initial stages of
oxidation, and then falls off to a lower rate as bulk diffusion effects begin to control
the growth kinetics as shown in Fig. 40.8b. The O3 angular distribution from the
delivery tube (stainless steel) was also measured by moving the detector off axis,
maintaining a constant tube-to-detector distance of 6.3 cm, and a constant O3 flow
rate of 1.9 standard cm3 s−1. The profile of O3 is Gaussian (Fig. 40.8c), and when
Ar diluent was used, the flux of O3 decreased and the angular profile was broadened
somewhat. It was estimated that the detection efficiency of the Ag film was 26 % for
O3, assuming that one O atom from O3 oxidizes the Ag, and that molecular O2 is
evolved. This efficiency assumes that the O3 is 100 % pure when it exits the
delivery tube, and if this is not so, then the detection efficiency is higher. The near
100 % purity of the distilled O3 was chemically checked using reaction with KI
solution [58], and also by its vapor pressure. O3 fluxes in the range 1014 to
1016 cm−2 s−1 were measured.
Fig. 40.8 Detection of ozone or atomic oxygen. a Geometry for O3 detection. b Mass change
during O3 + Ag reaction. c Angular distribution of O3
40.8 Detection of Ozone or Atomic Oxygen by Oxidation of Silver Films 491
The fate of the O3 in the film growth system was studied by placing O3 detectors
in various positions outside the beam. It was found that almost no O3 was detected,
suggesting that efficient wall decomposition reactions occur, converting O3 to O2.
A similar study, involving atomic O made in an RF discharge, was made using
Ag films on a quartz crystal microbalance [59].
A number of reports have been made of the use of hot Ir and Pt filaments, operating
in an oxygen atmosphere, for the production of atomic oxygen, . Usually the atomic
oxygen is delivered directly to a surface by line-of-sight transmission from the
filament to the surface. The work to be described here has used mass spectrometry
of the gas phase above hot Ir and Pt filaments and Auger spectroscopy measure-
ments of the composition of an atomically clean Au surface to investigate exactly
what is evolved from the Ir and Pt surfaces [56, 60]. Unfortunately, it is found that
atomic oxygen is not cleanly produced and that the transport of Ir and Pt oxides also
occurs making suspect any use of this method to produce pure atomic O.
Figure 40.9a shows line-of-sight mass spectroscopy of an Ir filament operating in
1.8 × 10−5 Torr of O2. The detection of the Ir oxides is shown as the Ir filament
temperature is raised. It was shown that the Ir signal corresponds mainly to Ir(g)
production which occurs also in the oxygen atmosphere. Figure 40.9b shows the
Auger spectrum of a Au surface after long exposure to a hot Ir filament at 2096 K,
operating in ultrahigh vacuum. It is seen that even at this low temperature (compare
to Fig. 40.9a), over long times, the Au Auger spectrum is modified by the depo-
sition of Ir. An Auger ratio, A/B, may be used to deduce the magnitude of the Ir
coverage. In Fig. 40.9c, this ratio maximizes after about 100 h of exposure to Ir at
2096 K, meaning that the Au Auger signal has been obscured by several mono-
layers of Ir which have been deposited. In Fig. 40.9d, the same experiment, done in
O2(g), shows that Ir is detected by Auger spectroscopy on the Au substrate even at
*1600 K, whereas atomic O (or Ir oxides) are not detected until 1760 K. Thus
atomic oxygen, transport to Au does not occur in the absence of Ir transport
meaning that Ir should not be used as a source of pure atomic O [60].
A similar set of experiments were performed with Pt. As shown in Fig. 40.9e, the
low energy Auger spectrum of the Au is altered by the deposition of Pt at 1683 K in
ultrahigh vacuum. In 137 min of deposition the, Au 69 eV feature is extinguished,
meaning that several ML of Pt have deposited. In Fig. 40.9f, the development of the
O-Auger intensity and the Pt coverage, as measured by an Auger ratio, are com-
pared. It may be seen by a careful comparison of the two panels that Pt transport is
observed prior to atomic O (or Pt oxides) transport to the Au surface [61].
492 40 Active Gases and Species
Fig. 40.9 Purity of Ir and Pt hot filament sources for atomic O production. a Line-of-sight mass
spectrometry—Ir + O2, b Auger ratios—Ir deposition in vacuum on Au (2096 K = TIr),
c Ir/Vacuum—2096 K—on Au, d Deposition of O and Ir in O2 on Au versus T, e Auger ratios—
Pt in vacuum on Au (1683 K = TPt), f Deposition of O and Pt in O2 on Au versus T
40.9 Purity of Ir and Pt Hot Filament Sources for Atomic Oxygen Production 493
Both of these experiments indicate that neither Ir nor Pt hot filaments, operating
in an oxygen atmosphere, can be trusted to deliver pure atomic O to an exposed
surface.
Hydrogen peroxide, H2O2, is a strong oxidizer, often used in aqueous solution for
oxidizing inorganic or organic substances. It works by a chemical reduction process
involving the oxygen, where the oxidation number of the peroxide ion (O1−)
changes to O2− in water product. A balanced equation for its use as an oxidizer is
2 þ 2e ! 2O
O2 2
reduction of peroxide
2A ! 2A 1þ
þ 2e oxidation of A
References
13. B.E. Koehler, C.H. Mak, D.A. Arthur, P.A. Coon, S.M. George, J. Chem. Phys. 89, 1709
(1988)
14. V.G. Luppov, M. Mertig, T. Roser, B.S. van Guilder, B. Vuaridel, YuM Melnik, A.F.
Prudkoglyad, Rev. Sci. Instrum. 62, 2738 (1991)
15. B. Van Zyl, M.W. Gealy, Rev. Sci. Instrum. 57, 359 (1986). It is interesting to note that in
these studies, the maximum in H2 dissociation probability is not reached until tungsten
temperatures above 2400 K are employed. This may be due to the presence of gas phase
collisions here, where collisionless conditions generally apply in sources used for surface
studies in ultrahigh vacuum
16. H. Koschmieder, V. Raible, Rev. Sci. Instrum. 46, 536 (1975)
17. A. Hershcovitch, A. Kponou, and TO. Niinikoski, Rev. Sci. Instrum. 58, 547 (1987)
18. K.C. Harvey, C. Fehrenbach Jr, Rev. Sci. Instrum. 54, 1117 (1983)
19. M. Eisenstadt, Rev. Sci. Instrum. 36, 1878 (1965)
20. J. Slevin, W. Stirling, Rev. Sci. Instrum. 52, 1780 (1981)
21. B.J. Wood, H. Wise, J. Phys. Chem. 66, 1049 (1962)
22. W.W. MacAlpine, R.O. Schildknecht, in Proceedings of IRE, 2099 (1959)
23. C.D. Stinespring, A. Freedman, C.E. Colb, J. Vac. Sci. Technol. A4, 1946 (1986)
24. C.E. Kolb, M. Kaufman, J. Phys. Chem. 76, 947 (1972)
25. Microwave sources are available from Opthos Instruments, Inc., 17805 Caddy Drive,
Rockville, MD 20855
26. S. Sibener, R.J. Buss, C.Y. Ng, Y.T. Lee, Rev. Sci. Instrum. 51, 167 (1980)
27. C.D. Stinespring, A. Freedman, Appl. Phys. Lett. 48, 718 (1986)
28. A. Freedman, J. Appl. Phys. 75, 3112 (1994)
29. G.N. Robinson, A. Freedman, R.L. Graham, Langmuir 11, 2600 (1995)
30. A. Freedman, C.D. Stinespring, J. Phys. Chem. 96, 2253 (1992)
31. H. Bonzel, A.M. Bradshaw, G. Ertl, Physics and Chemistry of Alkali Metal Adsorption
(Elsevier, Amsterdam, 1989)
32. J.B. Taylor, I. Langmuir, Phys. Rev. 44, 423 (1933)
33. Getter sources may be obtained from SAES Getters S.p.A., Head Office, Via Gallarate, 215,
20151 Milano, Italy. USA Offices: SAES Getters/U.S.A. Inc., 1122 E. Cheyenne Mtn. Blvd.,
Colorado Springs, CO 80906
34. F. Ricca, P. Delia, Porta. Vacuum 10, 215 (1960)
35. D.L. Schaefer, Rev. Sci. Instrum. 41, 274 (1970)
36. L.V. Hau, J.A. Golovchenko, M.M. Burns, Rev. Sci. Instrum. 65, 3746 (1994)
37. F. Kalos, A.E. Grosser, Rev. Sci. Instrum. 40, 804 (1969)
38. K.N. Walzl, C.F. Koerting, I.M. Xavier, A. Kuppermann, J. Chem. Phys. 86, 88 (1987)
39. F.A. Elder, C. Giese, B. Steiner, M. Inghram, J. Chem. Phys. 36, 3292 (1962)
40. P. Chen, S.D. Colson, W.A. Chupta, J.A. Berson, J. Phys. Chem. 90, 2319 (1986)
41. J. Dyke, N. Jonathan, E. Lee, A. Morris, J. Chem. Soc., Faraday II 72,1385 (1976)
42. X.-D. Peng, R. Viswanathan, G.H. Smudde Jr, P.C. Stair, Rev. Sci. Instrum. 63, 3930 (1992)
43. Alltech Assoc., Inc. 2051 Waukegan Rd., Deerfield, IL 60015
44. R. Renaud, L.C. Leitch, Can. J. Chem. 32, 545 (1954). Proper caution should be employed
working with dimethyldiazine as it is a carcinogenic agent
45. D.D. Berkley, A.M. Goldman, I. Maps, B.R. Johnson, J. Morton, T. Wang, Rev. Sci. Instrum.
60, 3769 (1989)
46. D.D. Berkley, B.R. Johnson, N. Anand, K.M. Beauchamp, L.E. Conroy, A.M. Goldman, K.
Mauersberger, M.L. Mecartney, J. Morton, M. Tuominen, Y.-J. Zhang, Appl. Phys. Lett. 53,
1973 (1988)
47. E. Coleman, T. Siegrist, D.A. Mixon, P.L. Trevor, D.J. Trevor, J. Vac. Sci. Technol. A9, 2408
(1991)
48. G.A. Cook, A.D. Kiffer, C.V. Klumpp, A.H. Malik, L.A. Spence, Advances in Chemistry
Series 21 (American Chemical Society, Washington, DC, 1959), p. 44
49. Matheson Gas Data Book, 6th ed. (Matheson Gas Products, Secaucus, NJ, 1980), pp. 574–577
496 40 Active Gases and Species
50. Aldrich #24,982-3; CAS (7631-86-9). Sold by Aldrich Chemical Co., P.O. Box 2060,
Milwaukee, WI 53201
51. Fomblin is a registered trademark of Montedison Specialty Chemicals, Milano, Italy
52. Krytox is a registered trademark of E.I. dupont De Nemours and Co., Inc., Wilmington, DE
53. S. Hosokawa, S. Ichimura, Rev. Sci. Instrum. 62, 1614 (1991)
54. M. Griggs, S. Kaye, Rev. Sci. Instrum. 39, 1685 (1968)
55. S. Anderson, K. Mauersberger, Rev. Sci. Instrum. 52, 1025 (1981)
56. L. Andrews, R.C. Spiker Jr, J. Phys. Chem. 76, 3208 (1972)
57. J.T. Kucera, J.D. Perkins, K. Uwai, J.M. Graybeal, T.P. Orlando, Rev. Sci. Instrum. 62, 1630
(1991)
58. D.H. Byers, B.E. Saltzman, Advances in Chemistry Series, vol. 21 (American Chemical
Society, Washington, DC, 1959), p. 93
59. V. Matijasevic, E.L. Garwin, R.H. Hammond, Rev. Sci. Instrum. 61, 1747 (1990)
60. V.S. Smentkowski, J.T. Yates Jr, J. Vac. Sci. Technol., A 12, 219 (1994)
61. V.S. Smentkowski, J.T. Yates Jr, J. Vac. Sci. Technol., A 12, 224 (1994)
62. W.T. Foley, P.A. Giguere, Hydrogen Peroxide and its analogues II. Phase equilibrium in the
system hydrogen peroxide-water. Can. J. Chem. 29, 123 (1951)
63. M.J. Loeffler, B.D. Teolis, R.A. Baragiola, J. Phys. Chem. B 110, 6911 (2006)
64. CRC Handbook of Chemistry and Physics, 84th edn. (CRC Press, Boca Raton, 2003–2004),
Section 6-71
65. N. Izyumskaya, V. Avrutin, W. Schoch, A. El-Shaer, F. Reuβ, Th. Gruber, A. Waag, J. Crystal
Growth, 269, 356 (2004)
Chapter 41
Electrochemical Sources and Devices
Halogens are important surface species in surface chemistry and their delivery to
surfaces in a controlled manner forms an important research technology. This
discussion concerns the delivery of iodine gas by means of electrochemical
41.2 Electrochemical Halogen Sources 499
generation [8], but others have used similar methods for both chlorine and bromine
[7, 9, 10].
The source shown in Fig. 41.2a is a silver iodide pellet. At moderate tempera-
tures, an electrochemical reaction, 2AgI(s) → 2Ag(s) + I2(g), occurs in the pellet as
a result of a potential applied between two electrodes, relying on the ionic mobility
in the heated pellet. The mobile defect concentration may be increased by mixing
about 5 % CdI2 with the AgI before casting the pellet.
The current flow through the heated pellet can be regulated using a constant
current DC source, and a voltage of about 25 V will cause a current of about 25 μA
to flow without external heating. The cells display extremely non-ohmic behavior,
especially during start-up. In order to increase the efficiency of the source, it is
housed in a Pyrex heater tube coated on the outside with a semiconducting SnO2
thin film (see p. 123) having a resistance of about 10–100 Ω/cm of tube.
Alternatively, the Pyrex tube can be heated using a spiral nichrome wire wrapped
around the tube [11]. The temperature of the tube and the pellet is adjusted by
passing current through the SnO2 film, which has a band of Aquadag colloidal
graphite on each end to act as a contact between the conductive film and two
clamps. At a temperature of 390 K, an applied voltage of 20 V produces a stable
cell current of about 100 μA. The cell walls may adsorb I2 during initial operation,
but then stable operating conditions are achieved. Faraday’s laws may be used to
roughly estimate the flux of iodine from the source.
Figure 41.2b shows the output of the electrochemical cell as a function of the
current as measured with a quadrupole mass spectrometer in line of sight, and on
axis with the source. The system pressure was simultaneously measured with an
ionization gauge that did not have line of sight to the source, and it is believed that
it measures mainly I2 gas. The mass spectrum indicates that I+ at 127 amu and I2+ at
254 amu are the main ions generated in the mass spectrometer from I2(g), although
some H2 and Cl2 are also produced. If the pellet is heated to higher temperatures,
near 430 K, the conductivity may be increased to give higher I2 fluxes, but
deposition of CdI2 also occurs slowly. It is also reported that a baffle at room
temperature may be used to avoid deposition of undesirable substances, and that
gaseous iodine can still be deposited on the sample with this arrangement.
To make the AgI/CdI2 button, 7 g of the mixed salt is melted and poured onto
the Pt/Rh mesh that is held on a glass plate under a cylindrical mold made of Pyrex.
The Ag coil is inserted into the top of the melt before solidification. If the Pt/Rh
mesh is heated to redness in a Bunsen burner flame prior to being covered with the
liquid melt, and if it is held flat on the plate, the melt will cover only the exposed
wire and will not submerge the electrode. Alternatively, red-hot Pt/Rh can be fused
into the cold pellet [7]. The pellet can easily be removed from the Pyrex mold after
cooling for insertion into the long support-heater tubing, which acts as a beam
collimator in use.
Similar devices have been made for producing all of the halogens [7], and their
performance is similar to that reported here. It is reported that after many hours of
operation, a dead short often occurs through the pellet. This indicates that a metallic
Ag dendrite has grown inside the pellet; brief passage of a high current through the
500 41 Electrochemical Sources and Devices
Fig. 41.2 Typical electrochemical halogen source. a Cell. b Mass spectrum at 150 °C
41.2 Electrochemical Halogen Sources 501
heated pellet may break the dendrite. In addition, for the Cl2 cell used at excessive
temperatures (>420 K) significant AgCl(g) was evolved. The halogen sources may
be voltage modulated in the frequency range of 0.5–22.5 Hz, and the modulated
halogen beam produced can then be used in kinetic studies on single crystal sur-
faces [9], and useful waveforms may be employed.
The production of pure F2 by electrochemical means has been partially suc-
cessful [10], although the high reactivity of fluorine with materials has caused
problems. A CaF2 pellet, held by springy contacts between a Pt plate and a Pt
gauze, was employed for electrochemical generation at temperatures of 420–450 K.
The gas evolved at the anode was focused onto the sample with a short tube, coated
with CaF2. Small amounts of SiF4 were detected if glass tubing was used. The
source delivers small amounts of platinum fluorides and HF. Studies with this
source of the adsorption of F2 on Pt showed that PtF2 is evolved above about 500 K
from Pt at low coverages, and that PtF4 is evolved at high coverages, in addition to
F(g) at 890 K [12]. The HF yield from the electrochemical cell may be reduced to a
few percent of the F2 yield by outgassing and electrolysis. Currents between 2 and
50 μA were used. During F2 dosing, the total pressure of the vacuum system was
below 1 × 10−9 mbar.
Halogen atom sources have been described and one example is a pulsed source
involving Cl2 dissociation in a microwave cavity. One interesting feature of this
work is the study of 2P3/2- and 2P1/2-Cl atom losses on Teflon walls [13].
A source of halogen ions produced by surface ionization on a LaB6 surface of
low work function has been described [14].
Halogen dosers, based on electrochemical generation are popular and have been
treated in Sect. 41.2 of this book. Recently an I2 doser working at low temperature
has been developed which adds to the technology described previously (Fig. 41.3).
Previous dosers required high temperatures to achieve electrical conductivity in the
solid state electrolyte. By using the compound Ag4RbI5, high ionic conductivity is
achieved near room temperature. Ag4RbI5 is prepared and about 0.5 g is com-
pressed into a disk of 2 mm thickness and 7.5 mm width using a hydraulic press at
600 MPa, for 5 min. The disk is made with Pt mesh on the anode side. The disk is
placed between two stainless electrodes and the application of a voltage causes
current to flow with the production of I2(g). The cell had an operating resistance of
about 2.5 kΩ and operates at about 200 µA when dosing. The collimator directs the
I2 flow to the sample. The Ag4RbI5 is stable up to 500 K allowing mild vacuum
system bakeout to be used. The design cannot be extended to similar compounds of
Cl and Br because these mixed salts have insufficient conductivity. Outgassing at
100 µA is recommended prior to opening the shutter for dosing [15].
502 41 Electrochemical Sources and Devices
The application of UHV surface analysis methods to working solid state electro-
chemical devices is of widespread interest. To study working devices, it is necessary
to carry out electrochemical redox processes involving gases, and to control elec-
trical potentials in the device. In addition, since ion mobilities govern device per-
formance, one must be able to control the device temperature during operation.
Figure 41.4a shows a top view of a working device, with a Ni-film deposited
working electrode. The device is mounted on a base plate which is transferrable to
one or more manipulators and which connects to an external potentiostat/galvanostat
used for impedance spectroscopy. Three springy electrodes mounted around the
device exterior make contact at various points, as shown in Fig. 41.4b. The small
contact area of the probes minimizes heat transfer from the heated device. Device
41.4 Electrochemical Device Working in UHV 503
Fig. 41.4 Electrochemical device working in UHV. a Top view. b Side view. c Thermal behavior
References
The use of high-pressure gases for experiments involving single crystal surfaces
always involves the problem of the preferential adsorption on the crystal of trace
impurities present in the gas. At high pressures, even 1 ppm of adsorbable impurity
can result in the selective adsorption of a monolayer or more of impurity. This
problem maybe eliminated by several methods, depending upon the gases and the
impurities involved.
For gases with low condensation temperatures, storage over previously baked
(573 K) zeolite pellets [1] at liquid nitrogen temperature is a favored method for the
adsorption of trace impurities with high condensation temperatures [2]. This is a
favored method for the additional purification of 99.9999 % pure methane, for
example [2]. In addition, carbon monoxide almost always contains traces of Ni
(CO)4 and Fe(CO)5 as impurities produced within the storage cylinder or in the
stainless steel gas handling system, and these metal carbonyls can be removed by
storage of the CO over a wide pore zeolite at liquid nitrogen temperature. See
p. 509 for a discussion of the use of cryogenic storage and distillation at liquid
hydrogen temperature. Alternatively, metal carbonyls can be removed by passing
CO through a copper coil heated to 100 °C, which is sufficient to decompose the
carbonyls, and then through a trap cooled with liquid nitrogen. It is important to
avoid the use of stainless steel tubing following the purification of the CO [3].
Adsorption of active chemisorbing gases in the presence of inactive gas is often
effectively carried out by means of a Ti getter film which is freshly deposited inside
a storage vessel under vacuum [3–5]. See p. 507 for the use of a Ti getter film for
the final stages of purification of He(g) [5]. An apparatus involving the use of
zeolite trapping followed by storage over a Ti film is shown in Fig. 42.1a.
A commercial titanium sublimation pump cartridge is employed in a stainless steel
storage vessel, and gas storage takes place at 77 K.
Fig. 42.1 Gas purification for experiments at high pressure. a Gas purification by adsorption.
b Gas purification over catalyst (high area)
42.1 Gas Purification for Experiments at High Pressures 507
Catalytic chemists have for years used a method involving gas exposure to a
high area catalyst prior to admission to the reactor chamber. This method has also
been applied to Ni single crystals being studied under high pressures of methane [6,
7]. Figure 42.1b shows the configuration of the gas purification apparatus used in
these studies. A U-tube is filled with a Ni catalyst (in this case), which consists of
Ni particles supported on a high surface area oxide support. The particle size ranges
from 0.3 to 0.8 mm diameter. Both ends of the U-tube are plugged with quartz wool
to prevent transfer of dust away from the catalyst. The catalyst is activated by
reduction in flowing H2 at 800 K, and then is evacuated at 800 K to remove H2.
Either methane or nitrogen may be cleaned of impurities such as CO, O2, CO2, and
hydrocarbons (excluding CH4) using this apparatus.
The cleanup of H2 maybe achieved using a supported Cu catalyst in the same
manner. The chemisorption of H2 by Cu is suppressed by working at liquid nitrogen
temperature. The activation of the Cu catalyst is first carried out using H2, and one
must be careful to initially use a low pressure of H2, since the exothermic reduction
of copper oxides may cause a runaway condition. Reduction temperatures of 650 K
are recommended [8, 9].
The production of He with extremely high purity for use in ultrahigh vacuum
systems is a challenge requiring special gas handling techniques. It must be
remembered that hydrogen is a major impurity in stainless steel ultrahigh vacuum
systems, and that to avoid hydrogen impurities, glass vacuum systems must
therefore be employed. The bakeout treatment of a glass vacuum system differs
from a metal system, and bakeout temperatures of the order of 600 K must be used
[10].
A number of methods for producing pure He have been reported in the literature.
The use of superfluid He leaks [11] and diffusion through Vycor glass [12, 13] have
been used by others, but the diffusion rate for H2 through hot Vycor is about 0.1 of
that for He [11], so that other methods must be used to achieve large factor
reductions in the hydrogen impurity level.
The apparatus [14] shown in Fig. 42.2 was mainly constructed of Pyrex glass
with bakeable metal valves, and it was processed by bakeout to achieve a base
pressure of 1 × 10−10 Torr. The pumping speed in this apparatus is only about 2 L/s.
Operation of the purification system begins with the deposition of Ti getter films on
the inner walls of the two bulbs, G1 and G2, by heating W loops to which Ti wire
segments have been spot welded. The surface area of these films is each about
300 cm2. The Vycor leak is outgassed at 1025 K until the outgassing cannot be
detected using the Bayard-Alpert gauge at 1 × 10−10 Torr pressure. Valve V4 is then
closed and He is effused through the Vycor leak at 975 K. The He used on the high
508 42 Gas Purification
pressure side of the Vycor leak is specified to be 99.999 % pure, and it is admitted
to approximately 1 atm pressure, into a gas handling system initially at 10−7 or
10−8 Torr. Typical He effusion rates are approximately 6 × 1014 He atoms/s.
Storage of He takes place overnight at several Torr pressure in the closed system
over the freshly evaporated Ti getter film, G1.
The Rh(111) single crystal was used as an integrating detector of H2 in the
purified He, using temperature-programmed desorption (TPD) with the mass
spectrometer (QMS) as the detector. Purified He was admitted to less than 1 Torr
pressure, with Valve V2 open to the system, and with a freshly deposited Ti film in
getter bulb, G2, in order to continue getter purification of the He during exposure to
the clean Rh(111) single crystal. An exposure of 1000 s was used in the He purity
test, leading to exposures of from 100 to 900 Torr s to the He. Following pumping
with the Hg diffusion pump, followed by use of the ion pump, TPD of both H2 and
CO was carried out, and compared to blank experiments without He. A least
squares fit to the data gave 5.4 × 1010 H/cm2-Torr s (He) for the contamination rate
with the CO contamination rate about an order of magnitude smaller. It was shown
that the rate of H2 adsorption was not limited by gas diffusion kinetics at the He
pressures used. Assuming a sticking coefficient of unity for the H2, these mea-
surements indicate that P(H2)/P(He) = 1.9 × 10−11, which is a factor of about 106
improvement over the original high purity He [14]. Flowing He has been similarly
purified using hot Ti sponge in the flow lines [15].
Superleaks for the control of He flow into a vacuum system have been described
[16]. They are made of porous Vycor glass [17] with approximately 40-Å-diameter
42.2 Production and Purity Measurement for Extremely Pure He 509
For some applications the supply of extremely pure gases into a vacuum system is
needed, and conventional cylinder sources are inadequately pure for this purpose.
Thus, for years, chemists have used permeation through high-temperature mem-
branes for the removal of impurities from the gas. In addition to supplying pure gas,
the permeation technique also affords easy control of flow rate by the adjustment of
the temperature.
In Fig. 42.4a, either hydrogen or oxygen permeation sources are shown [20]. For
hydrogen, pure Ni permeation tubing is used [21]; for oxygen, pure silver perme-
ation tubing is used [21]. The permeation tubes are 10–13 cm in length. The thin
permeation tubes are silver brazed (in hydrogen) to the ends of 0.25-in.-diameter
copper tubes mounted on ceramic insulator feedthroughs welded into a flange.
Passing current through the device provides the heating necessary to allow gas
permeation to occur. The purity of gases produced in this manner has been studied
by Whetten and Young [22], Young [23], and comparisons are made in the book by
Redhead [24]. The worker should remember that while high gas purity can be
achieved by permeation methods, it is also likely that the gas in a UHV system will
become impure owing to displacement from the walls of the chamber or by
interaction with hot filaments [24]. Studies of Ag permeation sources for the pro-
duction of pure O2 have shown that long periods of operation of the source must
occur before gases like H2S, SO2, CO, and CO2 are eliminated to the percent level
[25, 26]. It may be best to condition a Ag source on a separate vacuum system to
avoid the evolution of the gases into the main system [25, 26].
In Fig. 42.4b, a glass device involving a permeation thimble is shown. For
hydrogen, both pure Ni thimbles and pure Pd thimbles have been used. For oxygen,
pure silver thimbles are employed. The thimble, brazed to a Kovar seal, is centrally
mounted in a glass envelope containing a helical refractory metal heating filament
that heats the thimble by radiation. Permeation of purified gas from the outer
512 42 Gas Purification
envelope to the inside of the thimble takes place. In both experimental arrange-
ments, shown in Fig. 42.4a, b, it is appropriate to flow the source gas slowly
through the apparatus to replenish gas at the source.
A solid state source of oxygen, followed by permeation through a Ag tube wall,
is described on p. 513.
The use of a pure iron permeation source for purification of hydrogen and
deuterium has been reported in [27–29].
42.5 O2 Source—Solid State 513
The permeation of hydrogen gas through pure metals and alloys is of technological
and scientific interest. One of the methods used to make permeation measurements
is to electron-beam-weld the metal sample diaphragm to a cylindrical tube end [31].
Another method is to use a sample with machined knife edges that contact a
deformable gasket [32]. Elastomer gasket or O-ring sealing is only useful at room
temperature. The method to be described here differs from the earlier methods, is
simple, and the apparatus is reusable [33].
Two standard Conflat flanges with a standard Cu gasket are used. The sample, in
the form of a circular disk, is mounted inside the Cu gasket, and two Al gaskets are
placed on either side of the sample. The Cu gasket, when bolted in place, protects
the sample from the outside environment. The Al gaskets are plastically deformed
between the sample and the inner region of the flanges, and their thickness should
be properly adjusted to a dimension leading to compression and sealing. In cases
where an incomplete seal at the Al gasket was found, this was attributed to work
hardening of the gasket during manufacture. This could be corrected by annealing
the assembled sample holder for 1–3 h at 573 K, then retightening the bolts. The
sample holder was used up to 623 K, and is shown in Fig. 42.6.
An entirely different method for measuring hydrogen diffusivity in metals,
involving electrical measurements as hydrogen diffuses along the length of a foil, is
described in [33].
References
1. One may use Varian Vacsorb molecular sieve, catalog No. 944-0000, obtainable from Varian
Vacuum Products, 121 Hartwell Avenue, Lexington, MA 02173
2. T.P. Beebe, D.W. Goodman, B.D. Kay, J.T. Yates Jr, J. Chem. Phys. 87, 2305 (1987)
3. Dr. J. Wambach, Danish Technical University, Department of Physics, Building 307,
DK-2800 Lyngby, Denmark (private communication)
4. L. Hanley, Z. Xu, J.T. Yates Jr, Surf. Sci. 248, 265 (1991)
5. J.T. Yates Jr, J.J. Zinck, W.H. Weinberg, Rev. Sci. Inst. 50, 132 (1979)
6. I. Chorkendorff, Technical University of Denmark, Building 307, DK-2800, Lyngby,
Denmark (private communication)
7. I. Chorkendorff, I. Alstrup, S. Ullmann, Surf. Sci. 227, 291 (1990)
8. P.A. Taylor, P.B. Rasmussen, C.V. Ovesen, P. Stoltze, I. Chorkendorff, Surf. Sci. 261, 191
(1992)
References 515
As surface chemistry deals with more and more complex systems it has become
necessary to work at high gas pressures as described elsewhere in this book. In
addition, examples are available where work under liquids has been done while
maintaining clean surfaces. The simple addition of a drop of liquid water to an
atomically clean TiO2 single crystal was found to be very difficult when pure water
injection with a syringe through a rubber septum was attempted. Large quantities of
silicon were found on the crystal following injection and evaporation of the water
droplet, and this was thought to be due to silicones present in the rubber seal.
Therefore, a syringe-injection procedure using only a tiny stainless guide tube for
the syringe needle was employed as shown in Fig. 43.1a. The wetting chamber was
baked to achieve ultrahigh vacuum with the syringe seal closed. Upon establishing
an atmospheric pressure of very pure flowing O2 gas (<0.05 ppm hydrocarbons)
(plus a few extra Torr to cause outward atmospheric flow), it was found that
ultrapure water could be injected into the stainless guide tube while O2 gas con-
stantly flowed from the inside to the outside along the loose fit between the
hypodermic syringe and the guide tube wall. Following droplet evaporation on
the atomically clean TiO2 crystal inside the chamber (achieved by reducing the
relative humidity of the flowing gas), and evacuation, the crystal was withdrawn
into the analytical chamber and an Auger analysis was performed. Figure 43.1b
shows the Auger spectrum after a wetting/evaporation cycle. The C Auger signal,
compared to the TiO2 Auger features, showed that less than 0.05 ML of carbon
remained after evaporation. The lower limit of carbon for the cleaned crystal was
about 0.005 ML. The syringe needle and the syringe were treated in boiling H2O2
(aq) for several hours to remove hydrocarbons prior to the water deposition
experiments [1].
Fig. 43.1 Adding liquid to atomically clean surfaces. a Apparatus for wetting studies of
ultrahigh-vacuum-prepared TiO2 surfaces. b Proof of surface cleanliness by auger spectroscopy
Reference
1. T. Zubkov, D. Stahl, T.L. Thompson, D. Panayotov, O. Diwald, J.T. Yates Jr, J. Phys. Chem.
B 109, 15454 (2005)
Part VI
UHV Windows
Chapter 44
Spectroscopic Windows
For the transmission of infrared radiation to and from an ultrahigh vacuum system
for the purpose of carrying out infrared spectroscopic measurements, it is essential
to employ windows that are highly transparent in the desired spectral region. Often,
alkali halide crystals are used for this purpose. Alkali halides have a small linear
thermal expansion coefficient relative to metals (about 0.1 of the metal) and thus
undergo high stresses under elevated temperature conditions when rigid contact is
made between the window material and the metal. Several window-mounting
designs that place elastomer seals between the window and the metal have been
reported [1–5]. In each of these designs the elastomer yields by means of shear
strains during bakeout relieving the strain in the alkali halide window and thereby
avoiding fracture. The common elastomer in all of the designs is Viton (see p. 108
for studies of gas diffusion through Viton), and this material will stand repeated
bakeout at 200 °C. Two of the designs, shown in Fig. 44.1a, b, involve differential
pumping to avoid leakage and/or diffusion of atmospheric gases through the Viton
seals. The designs shown in Fig. 44.1c, d do not employ differential pumping. The
design of Fig. 44.1c places an Al gasket between two Viton O-rings, eliminating
leakage or diffusion through the outer O-ring. The design of Fig. 44.1d eliminates
differential pumping or seals between the Viton gaskets.
Figure 44.1a is the original differentially pumped design by Hollins and
Pritchard [1], and has proven effective in many different laboratories [2]. The NaCl
or KBr window is retained in a stainless steel flange by an outer clamping flange.
The inner space between the Viton O-rings is differentially pumped to pressures of
the order of 0.1 Torr by a rotary pump, and it is reported that this pumping is
essential for the best system pressure. On the ultrahigh vacuum side of the window,
the O-ring is compressed outside of a loosely fitting retainer ring that keeps it from
moving inward and also avoids local trapping of gas introduced during assembly. If
the differentially pumped space is let up to air before bakeout, the system pressure
responds immediately; however, after bakeout, the O-ring seals mate more effec-
tively to the metal and window, and only a small degradation of vacuum occurs if
air is introduced into the differential volume. The indicated base pressure after
bakeout is about 5 × 10−11 Torr.
Figure 44.1b shows a modification of the original design in which no extension
beyond the system outer flange occurs (zero profile) [3]. Here, a modified Conflat
flange, equipped with differential pumping, receives an O-ring plus a metal ring
retainer to hold the O-ring in a circular geometry. An IR window is seated on this
O-ring, and a double O-ring plus retainer ring is compressed on the outside of the
window by a threaded thrust ring that screws into the flange opening. The inner
O-ring (1/16-in.) of the pair seals against the window, while a slightly larger
diameter 1/16-in. O-ring seals against a ledge in the flange. The atmospheric
pressure forces the inner O-ring against the outer rim of its groove in the retainer
ring. This device has been employed with KRS-5 and Csl window material, and is
routinely baked at 100–110 °C. Once the seal is made, no effect of pumping out the
differential space is observed, and pressures of the order of 5 × 10−11 Torr are
reported.
Figure 44.1c shows a third method for sealing IR windows to an ultrahigh
vacuum system [4]. The components are constructed of KF 50 flanges, which are
commercially available. The outer section has its groove widened in order to hold
the O-ring. The inner section has been modified by welding tubulation containing
an O-ring groove into a hole cut into the KF 50 flange. A specially designed Al
gasket is placed between the inner and outer flanges before assembly. An IR
window is then pressed between the inner and outer O-ring seals and compression
by three separate clamps results in gasket deformation, leading to a vacuum-tight
seal between the two O-rings, and removing the need for differential pumping.
Bakeout of the window was limited to 150 °C and the O-rings were observed not to
stick to the sealing surfaces. The leak rate was estimated to be smaller than
1 × 10−12 Pa m3 s−1.
Figure 44.1d shows a double seal made of two flat Viton gaskets that seal against
the IR window [5]. A special thrust ring is bolted onto a modified Conflat flange
using six bolts and a small volume at the edge of the IR window is captured by both
gaskets. Bake-out temperatures up to 150 °C have been used, and the base pressure
is reported to be 2 × 10−10 mbar.
All four of these designs must be carefully made to avoid metal contact with the
brittle window material upon compression of the O-ring seals.
A design similar to that of Fig. 44.1a, b has been reported [6] for sealing IR
windows to a glass system.
524 44 Spectroscopic Windows
For some applications, CaF2 single crystal windows1 [7–10] and MgO windows2
[11, 12] may be attached to flanges for IR studies, and these seals require neither
differential pumping nor O-ring-sealing technology. A review of this method may
be found in [13], and an RF heating method for making these seals is given in [14].
Irtran 4 windows (ZnSe, polycrystalline) have been sealed to Voscovit 501 alloy
using Pyroceram 7575 sealant [15].
The use of a lead gasket in conjunction with sealing a NaCl window to a
stainless steel flange has been reported to be successful for ultrahigh vacuum
applications. A specially shaped flange sealing surface and a series of conical
washers acting as compression springs are involved to compensate for differential
thermal expansion. These windows assemblies are bakable to 275 °C, and exhibit a
leak plus outgas rate of <3 × 10−10 atm cm3/s [16].
A simple method involving the sealing of Irtran 2 and Irtran 6 [17] into rotatable
Conflat flanges has been given in [18]. Here, a silicone cement (Gevac [19]) is used
to seal the window to the rotatable insert, and a ring of Al is placed between the
outer flange and the insert in order to apply pressure directly over the underlying
knife edge to prevent flexure of the flange during tightening. This design is superior
to that of [20], which involves flexure of the nonrotatable flange employed.
The Viton O-ring technology for sealing windows such as KBr, CaF2, and MgO to
ultrahigh vacuum systems involve severe limitations on the bakeout temperatures
that may be employed (see p. 521). Other sealing methods involving, for example,
fused AgCl cements, are susceptible to leak formation upon exposure to ultraviolet
light, and also are incompatible with many metals because of corrosion effects.
A new method, which works for both CaF2 and MgF2 windows, is described
below [21].
Figure 44.2 shows a stainless steel window flange welded to a tubulation that
connects to the ultrahigh vacuum system. A 0.65-cm-wide step is provided to
support the window, which is 1 mm smaller in diameter than the outer diameter of
1
CaF2 windows, sealed directly to Conflat flanges through a thin silver/silver chloride intermediate
sealing assembly, may be obtained commercially from Harshaw Crystal and Electronic Products,
Solon Technologies, 6801 Cochran Road, Solon, OH 44139. These windows are unsatisfactory for
use with ultraviolet light because of the sensitivity of the AgCl cement to photodecomposition.
A description of the sealing process is given in [7–10].
2
MgO windows have been sealed to Pyrex glass sleeves using gold/silver chloride sealing with the
Pyrex being sealed by a Kovar graded seal to stainless steel flanges. For a diagram of this method,
see [11, 12].
44.2 Windows for Vacuum UV Transmission into Ultrahigh Vacuum Systems 525
the step. An outer window retaining ring presses the window against the seal, and
the pressure at the seal can be adjusted using bolts which compress springs around
the perimeter. A soft aluminum foil cushion is used between the retaining ring and
the outer part of the window.
The vacuum seal is made by building up a Vacseal [22] bead on both the
window and the step surface of the flange as shown in Fig. 44.2. The Vacseal bead
is not considered to be a leak sealer between the window and the flange, but a
deformable gasket material. The following recipe is recommended for the pro-
duction of a high temperature seal which is He leak tight:
1. Clean and degrease the window and the flange.
2. Apply Vacseal as shown to the step and window surfaces as shown, keeping
both surfaces horizontal. A 1.5-2.5-mm-thick bead is formed on each surface.
3. Place flange and window into an oven at room temperature and heat to 125 °C
for 5 h. Then, following this precuring procedure, allow to cool to room tem-
perature. There should be no bubbles in the beads.
4. Invert window and carefully place on the flange. The two beads should coalesce
into a single seal about 2–2.5 mm in thickness without voids or bubbles.
5. Gently tighten the retainer ring but do not displace the Vacseal sealing material.
6. Cure at 150 °C for 15 h. Cool to room temperature and inspect for cracks or
voids. The sealant should be amber in color. The spring-loaded bolts may be
tightened slightly in this step.
7. Place the seal into the oven at 200 °C for 8 h. Cool and leak check.
526 44 Spectroscopic Windows
If voids or cracks appear during the above procedure, start over. Methyl ethyl
ketone (MEK) may be used to dissolve the cured Vacseal. Ethyl alcohol will not
dissolve the seal and may be used to clean the window without causing a leak to
form. One must be careful to not apply mechanical stresses to the cured window
assembly. The seals are bakeable to 330–400 °C and are leak-tight for periods in
excess of 6 months [21]. This sealing technique has been used on windows of
7.5-cm diameter [21]. The optical distortion of the widows is small, and thermal
cycling may be carried out.
Another method, useful for mounting quartz windows on Al flanges, has been
reported [23]. Here, an indium seal is employed and the window may be baked to
120 °C, which is the bakeout temperature needed for Al ultrahigh vacuum systems.
Thin Be windows are often employed for transmission of X-rays into UHV sys-
tems. Although electron-beam-welded windows are commercially available [24],
brazing onto stainless steel Conflat flanges is sometimes used [25, 26]. The pro-
cedure for doing this is reported here [26].
Figure 44.3a schematically shows the mechanical procedure employed for
making an ultrahigh vacuum brazed seal to type 304 stainless steel. Beryllium foil,
0.25–0.50 mm thick, must be specified to be of vacuum quality [27]. Such foils may
be used for spans up to 20 cm in length. The stainless steel surface is abraded with
250 grit sandpaper, and all sharp edges on the vacuum side are rounded. A brazed
region 6–9 mm wide is employed. The Be foil is shaped in exterior dimensions by
etching, using common fingernail polish to protect the regions to be preserved. The
protective layer can later be removed using acetone solvent. The acid etching
mixture is specified in Table 44.1. The braze material is 99.99 % Ag foil of
0.075-mm thickness, cut into a ring. The silver should be cleaned with silver polish
or by means of a chemical bath specified in Table 44.2, in which metallic Al is used
to reduce oxidized and sulfided Ag layers. The cleaned components are loaded into
a stainless steel cylinder and the lower flange is tack welded to the cylinder. The
stacked assembly is covered with quartz cloth and loaded with a weight as shown in
Fig. 44.3a. The assembly is fired in a vacuum furnace according to the procedure
outlined in Table 44.3.
He leak detection indicates no leaks at the sensitivity of the detector,
10−11 mbar L s−1.
The transmission of X-rays through Be foils of different thicknesses is shown in
Fig. 44.3b.
44.3 Construction of Be Windows for Ultrahigh Vacuum Use 527
Fig. 44.3 Beryllium Window Fabrication. a Brazing Be windows to stainless steel. b X-ray
transmission through Be
Low atomic number elements are often used for windows to transmit X-rays or
high-energy particle beams. A very simple bakeable aluminum foil window
mounting design is shown in Fig. 44.4 [28]. The Al foil window, cut into a circle, is
mounted between two Cu gaskets that are used to make vacuum seals to two
Conflat flanges. The window material is Alcoa 3003 aluminum with an H19 temper.
The only special machining required is to put a radius, R, onto the inside edge of the
Cu gasket placed on the vacuum side of the window. The radius R should equal the
thickness of the window material, T.
Both 0.05- and 0.13-mm Al foil of 3.81-cm diameter and 10-cm diameter exhibit
a leak rate for He of <5 × 10−10 std. cm3 s−1. Foils of 0.13-mm thickness have been
baked at 260 °C overnight with one side at atmospheric pressure without devel-
oping leaks; foils of 0.05-mm thickness have been baked at 180 °C overnight
without developing leaks. This sealing technique has been shown to be unsuc-
cessful with Be windows.
Very thin metal foils may be welded to a stainless steel flange using electron beam
welding in vacuum. However, there is danger of damaging the foil because of the
lack of heat conduction away from the fusion region. A method [29] for successful
welding is shown in Fig. 44.5.
The foil, which may be of Cu, Ni, or stainless steel, is several microns in
thickness. It is placed on the flange and a backup washer of 304 stainless steel
(0.05–0.13 mm thick, made by photoetching from sheet material) is placed on top
of the foil, as shown. A hold-down fixture and clamps are used. The electron beam
(120 kV, 250 μA, 0.13-mm diameter) is directed as shown in Fig. 44.5, and the foil
is fused to the flange, after which the washer may be removed. A rotating table,
with a velocity of 76 cm/min at the weld point is used.
If the foil initially contains wrinkles, it will flatten as a result of welding at the
circumference. This is thought to be due to radial forces exerted at the circumfer-
ence during cooling of the weld puddle.
mounting bolts that can be transmitted to the window. A re-entrant design such as
that shown in Fig. 44.7 can eliminate this type of strain [35].
To construct this mounting, the glassblower will start with a Kovar-to-glass seal
that has been welded into the flange. This seal first has glass matched to the thermal
expansion coefficient of Kovar alloy, and then a graded seal leading to Pyrex on the
outer end of the seal. The Pyrex is fashioned into an S-shaped reentrant seal, with
the window mounting flared and polished flat to receive the window and its sealant.
The length of glass in the S-section is sufficient to eliminate the transmission of
stress to the window when the flange is tightened.
Glass-to-metal seals other than Kovar are sometimes used in these applications3
[36, 37].
References
1. P. Hollins, J. Pritchard, J. Vac. Sci. Technol. 17, 665 (1980) (this report also contains
references to earlier window sealing methods)
2. Z. Xu, J.T. Yates Jr, J. Vac. Sci. Technol. A8, 3666 (1990)
3. R.G. Tobin, C. Chung, J.S. Luo, J. Vac. Sci. Technol. A12, 264 (1994)
4. C. Damerow, W. Erley, J. Vac. Sci. Technol. A5, 2974 (1987)
5. A. Bradshaw, private communication. The gaskets may be obtained from Busak and
Sham-ban GmbH & Co., Handwerkstr. 5–7, 70565 Stuttgart, Germany
6. R. Raval, S. Munro, A. Coats, G. Beange, J. Vac. Sci. Technol. A10, 3593 (1992)
7. D.D. Berger, Rev. Sci. Instrum. 53, 1619 (1982)
3
Glass-to-metal sealing methods are described in [36, 37].
534 44 Spectroscopic Windows
8. R.W. Roberts, J.F. Harrod, H.A. Poran, Rev. Sci. Instrum. 38, 1105 (1967)
9. B. Caras, Rev. Sci. Instrum. 39, 1441 (1968)
10. Y.-C. Chiu, Rev. Sci. Instrum. 41, 639 (1970)
11. J.T. Yates Jr, D.A. King, Surf. Sci. 30, 601 (1972)
12. R.W. Roberts, J.F. Harrod, H.A. Poran, Rev. Sci. Instrum. 38, 1105 (1967)
13. Y.-C. Chiu, Rev. Sci. Instrum. 41, 639 (1970)
14. M. Aslam, Rev. Sci. Instrum. 42, 1892 (1971)
15. P.E. Wierenga, G.J. Mollenhorst, A.T.B. Wikkerink, Rev. Sci. Instrum. 49, 408 (1978)
16. T.J. Manuccia, J.R. Peele, C.E. Geosling, Rev. Sci. Instrum. 52, 1857 (1981)
17. Irtran optics are available from Eastman Kodak Co., 5/81 RC MC 02010, Rochester, NY
14650-2010
18. M. Kottke, R.G. Greenler, Rev. Sci. Instrum. 42, 1235 (1971)
19. A silicon resin, type KL5, like that used in Ref. [14] can be obtained from Kurt J. Lesker
Company, 1515 Worthington Avenue, Clairton, PA 15025-9911
20. L.J. Schkolnick, Rev. Sci. Instrum. 39, 122 (1968)
21. C.H. Muller III, M.W. Barrett, D.D. Lowenthal, Rev. Sci. Instrum. 59, 1425 (1988)
22. Vacseal is sold as a viscous liquid in 3 oz. bottles. Space Environment Laboratories, Box
1061, Boulder, CO 80306
23. H. Saeki, J. Ikeda, H. Ishimaru, Vacuum 39, 563 (1989)
24. Electrofusion Corp., 34325 Ardenwood Blvd., Fremont, CA 94536
25. P.L. Hartman, Rev. Sci. Instrum. 57, 2564 (1986)
26. G. Navrotski, J. Vac. Sci. Technol A12, 262 (1994)
27. Brush-Wellman, Elmore, OH 43416
28. J. Edgecumbe, Rev. Sci. Instrum. 37, 1419 (1966)
29. P. Fowler, J.V. Minges, Rev. Sci. Instrum. 40, 841 (1969)
30. N.W. Cheung, Rev. Sci. Instrum. 51, 1212 (1980)
31. Obtained from Connecticut Hard Rubber Co., P.O. Box 2560, Meriden, CT 06450
32. C.L. Huang, T. van Duzer, IEEE Trans. Electron Devices ED-23, 579 (1976)
33. N.W. Cheung, L.C. Feldman, P.J. Silverman, I. Stensgaard, Appl. Phys. Lett. 35, 859 (1979)
34. N.W. Cheung, M.-A. Nicolet, J.W. Mayer, in Proceedings of the Symposium on Thin Film
Phenomena- Interfaces and Interactions, vol. 80.2, ed. by J.E.E. Baglin, J.M. Poare (The
Electrochemical Society, Princeton, 1980)
35. Professor David Waldeck, Department of Chemistry, University of Pittsburgh, Pittsburgh, PA
15260 (private communication). The design originated with Professor C.B. Harris’ group at
the University of California, Berkeley, CA
36. E.L. Wheeler, Scientific Glassblowing (Wiley Interscience, NY, 1958), p. 157
37. W.E. Barr, V.J. Anhorn, Scientific and Industrial Glass Blowing and Laboratory Techniques
(Instrument Publishing Co., Pittsburgh, 1947), p. 116
Chapter 45
Observation Windows
points opposite each other, is about 2 Ω at room temperature and about 10 Ω at the
operating temperature. Electrical contact to the Ti film is made using W clips made
from 0.25-mm-diameter wire and protective Ta foil pads prevent abrasion of the Ti
film by the W clips. Electrical heating at a power level of about 20 W maintains the
temperature around the edge of the quartz window at 550–600 °C, and the tem-
perature in the center is estimated to be 350–400 °C, based on a calculation for
radial heat transfer [4]. It was reported that evaporation of over 1000 g of As over
1 year was carried out using one protective quartz window, and it was found that
the optical transmission of the internal window only decreased by 5 %, corre-
sponding to a temperature error of only 2° for pyrometry measurements at 600 °C.
For GaAs deposition, it was found that the heated window did not introduce
impurities into the films produced by molecular beam epitaxy.
Alternatively, the use of disposable internal glass windows has been employed
successfully for this purpose [5].
Glass seals in Conflat flanges are generally made by a brazing technique that
distorts the window, limiting its usage for certain optical studies. In addition, when
optical coatings are applied to flange-sealed glass windows, the high temperature of
538 45 Observation Windows
the coating process may further distort the window. A number of alternative win-
dow seals have been described [6–10] but often require considerable modification
of the standard flanges.
In Fig. 45.3, a modified Conflat flange and gasket are employed to seal BK 7
[11] glass into an ultrahigh vacuum chamber [12]. The window material can range
from high precision optical flats to window glass. These glass blanks can be
optically coated if desired. A commercial OFHC Cu gasket is machined to have its
own knife edge. The window seals the gasket by deforming the Cu knife edge. The
dimensions shown are somewhat arbitrary. Pressure is applied to the window seal
using a rotatable bolt ring which may be modified to increase the area which applies
force to the window. Twenty sheets of 0.001-in.-thick Al foil are used as a cushion
between the retaining ring and the glass. Alternatively either Cu or Pb wire may be
used for the cushion.
The bolts are tightened uniformly and slowly until no leak is observed with a He
leak detector. For a 4.5-in.-diameter flange, torques of about 3 ft. lb/bolt are
employed for sealing. The technique forms seals with less than 10−10 atm cm3 s−1
He leak rates. Baking to 250 °C may open leaks, and these can be eliminated by
retightening the flange bolts. A test for stress-induced birefringence showed that
only 0.01 % of 851-nm light passed through two crossed linear polarizers with a
window mounted between them.
It is important to keep the Cu knife edge at least 0.1 in. from the edge of the glass
to prevent breakage, and this may necessitate offsetting the Cu knife edge from the
underlying stainless steel knife edge, as shown in Fig. 45.3 [12].
Several methods for mounting windows on ultrahigh vacuum systems are available,
but the method shown here is the most simple, requiring only standard components.
As shown in Fig. 45.4, a window is retained between two Viton O-ring seals, using
a rotatable outer flange without its normal insert [13]. The assembly is mounted
onto a normal Conflat flange attached to the vacuum system. The inner seal may be
either a O-ring or a special Viton gasket manufactured by Varian for use in place of
copper gaskets. The upper bakeout temperature for the Viton is 150 °C. The general
design is also applicable to larger flanges.
References
Fig. 46.1 Surface segregation and cleaning of metal surfaces chemically. a Typical phase diagram
for alloys. b Si segregation on Pt. c Si segregation in vacuum and in O2 on Pt. d Silicon
segregation on Pd. e B-O vibrational modes on Rh(111) after reaction at 970 K in O2
46.1 Surface Segregation and the Cleaning of Metal Single Crystals 545
effective in purifying the bulk (or at least the near surface region) so that bulk
impurities are not available at the surface during an experiment.
A description of the surface cleaning methods employed for a total of 74 ele-
ments has been given in a very useful review by Musket et al. [2].
Several examples of the use of chemical methods for cleaning surfaces in this
manner are presented below to illustrate the general principles involved and the
types of measurements used to follow the cleaning process.
Platinum
Elements such as Si, Ca, Al, S, C, and P are often found in zone-refined Pt crystals,
and an excellent review of and insight into this problem has been provided in the
work of Bonzel et al. [3–5]1. Si segregates strongly to the surface of Pt, but this was
not appreciated until the early 1980s because the prominent Si (92 eV) feature is
almost coincident with a Pt(94 eV) feature. This may be seen in Auger spectra in
Fig. 46.1b, where heavy Si segregation is observed under equilibrium conditions at
873 K, and less heavy segregation is observed at higher temperatures. Spectrum
(a) is of the freshly sputtered surface (Si free) and the small Pt Auger feature at
about 94 eV is evident. Studies of the effect of O2 on the rate of Si segregation to
the surface show that at 10−7 mbar and at Pt temperatures of 873 K, no enhance-
ment of the Si segregation occurs in the presence of O2, as shown in Fig. 46.1c.
Studies at higher temperature [3–5] indicate that bulk SiO2 will form for
T > 1000 K. It is pointed out that the Auger intensity of the O(KLL) feature,
associated with Si atoms, is the best monitor of the amount of Si present after
oxidation. Platinum oxides are thermodynamically unstable in vacuum at the
temperatures discussed here. The properties of the oxide on silicon-contaminated Pt
first led investigators to postulate the presence of a “platinum oxide” with unex-
pected thermodynamic stability on Pt surfaces [3–5].
Palladium
Similar results to those found for Pt are reported for Pd [6]. Here, too, Si-stabilized
oxides are formed under conditions where PdO would be thermodynamically
unstable, as shown in Fig. 46.1d. The oxidation occurred at 1073–1273 K in
10−6 mbar O2 for spectra (a) and (b). Argon ion sputtering was used prior to
acquiring spectrum (c).
Rhodium
A common impurity in single crystal Rh is boron, which is difficult to remove by
sputtering [7, 8]. It has been found that heavy oxidation is effective in removing
boron by the formation of a boron oxide species that is best detected by
high-resolution electron spectroscopy (HREELS) [9]. The interference between Rh
and B Auger spectral features makes Auger spectroscopy impractical for seeing
small amounts of B on Rh. Figure 46.1e shows an HREELS spectrum of the B-O
vibrational modes obtained when Auger spectroscopy was ineffective in detecting B.
1
Earlier important studies of this problem include [4].
546 46 Cleaning Metal and Semiconductor Crystals—Examples
Fig. 46.2 Chemical method to measure and remove C from Pd. a Pd auger spectra for increasing
C coverage. b Auger ratio for increasing C coverage. c 2C + 2 15NO(a) → 15N2 + 2CO
548 46 Cleaning Metal and Semiconductor Crystals—Examples
in deionized water. This procedure has been shown to reduce the Ni concentration
in the Si surface to below the Auger sensitivity (signal to noise = 0.0003) level,
giving Si crystals that are long-lived and almost Ni-free as measured by STM [17].
The cleaning procedure for Si(100) wafer strips is then to heat to 900–950 K
overnight to outgas and then to anneal to 1420–1450 K for 30–60 s, cool quickly to
1000–1100 K, hold at temperature to 120 s, and finally cool down to 300 K at a
slow rate of 2 K/s. Alternating current heating is used.
550 46 Cleaning Metal and Semiconductor Crystals—Examples
Fig. 46.4 Preparation of clean tungsten single crystals. a Sample holder and flashing geometry.
b Thermal desorption spectra of W(110) crystal behavior at 6 × 10−8 mbar O2 pressure. Both CO
and O2 partial pressures are monitored
46.4 Preparation of Atomically Clean Tungsten Single Crystals 551
Tungsten and some of the other refractory transition metals are easily contaminated
in ultrahigh vacuum as a result of hydrocarbon adsorption and decomposition to
make a metal carbide film. This film can be oxidized in vacuum to make a very
stable molecule, CO, which desorbs, leaving pure clean metal behind. The surface
cleaning procedure occurs in two steps: (1). Cycles of low power flashes
(T * 1200 K) in P(O2) = 6 × 10−8 mbar; and (2). a single high desorption pressure
flash (T = 2200 K) in vacuum to remove the oxide layer produced in the first stage
of treatment. The loss of oxide at high pressure leaves a clean surface. In the case of
tungsten wire, the loss of tungsten oxides has been carefully studied [19–21].
Figure 46.4a shows a typical W single crystal held in a molybdenum holder and a
spiral W heater filament. Figure 46.4b shows the typical oxygen, carbon monoxide
and power in the cleaning of the carbon from the crystal surface. It may be seen in
Fig. 46.4b that CO desorption reaches a constant behavior during heating more than
once. The first CO desorption differs from later stages, indicating that most CO is
lost in the first cycle under oxygen [22].
References
1. J.J. Burton, E.S. Machlin, Phys. Rev. Lett. 37, 1433 (1976)
2. R.G. Musket, W. McLean, C.A. Colmenares, D.M. Makowiecki, W.J. Siekhaus, Appl. Surf.
Sci. 10, 143 (1982)
3. H.P. Bonzel, A.M. Franken, G. Pirug, Surf. Sci. 104, 625 (1981)
4. M.J. Cardillo, G.E. Becker, Surf. Sci. 99, 269 (1980)
5. H. Niehus, G. Comsa, Surf. Sci. 102, L14 (1981)
6. S.D. Bader, L. Richter, T.W. Orent, Surf. Sci. 115, 501 (1982)
7. D.G. Castner, G.A. Somorjai, Appl. Surf. Sci. 6, 29 (1980)
8. P.A. Thiel, J.T. Yates Jr, W.H. Weinberg, Surf. Sci. 82, 22 (1979)
9. S. Semancik, G.L. Haller, J.T. Yates Jr, Appl. Surf. Sci. 10, 546 (1982)
10. R.D. Ramsier, K.-W. Lee, J.T. Yates Jr, J. Vac. Sci. Technol. A13, 188 (1995)
11. H. Conrad, G. Ertl, J. Küppers, E.E. Latta, Surf. Sci. 65, 235 (1977); ibid 245 (1977)
12. K. Kato, T. Ide, S. Miura, A. Tamura, T. Ichinokawa, Surf. Sci. 194, L87 (1988)
13. H. Niehus, U.K. Kohler, M. Copel, J.E. Demuth, J. Microscopy 152, 735 (1988)
14. X.F. Lin, K.J. Wan, J. Nogami, Phys. Rev. B 47, 13491 (1993)
15. A.E. Dolbak, B.Z. Olshanetsky, S.I. Stenin, S.A. Teys, T.A. Gavrilova, Surf. Sci. 218, 37
(1989)
16. R.M. Wallace, C.C. Cheng, P.A. Taylor, W.J. Choyke, J.T. Yates Jr, Appl. Surf. Sci. 45, 201
(1990)
17. V. Ukrainsev, J.T. Yates Jr, Surf. Sci. 346, 31 (1996)
18. E.R. Weber, N. Wiehl, Mat. Res. Soc. Symp. Proc. 14, 19 (1983)
19. D.A. King, T.E. Madey, J.T. Yates Jr, J. Chem. Phys. 55, 3236 (1971)
20. D.A. King, T.E. Madey, J.T. Yates Jr, J. Chem. Phys. 55, 3247 (1971)
21. D.A. King, T.E. Madey, J.T. Yates Jr, J. Chem. Soc. Faraday Trans. 1(68), 1347 (1972)
22. Kh Zakeri, T.R.F. Peixoto, Y. Zhang, J. Prokop, J. Kirschner, Surf. Sci. 604, L1 (2010)
Part VIII
High-Area Solids
Chapter 47
Infrared Cells
Fig. 47.1 IR cell for adsorption studies on supported catalysts. a IR cell. b. Transmission IR
spectra
single crystal disk held on a hot plate at about 70 °C. Here a 9:1 acetone:water
solution is the solvent for the metal salt and a carrier for the oxide support. Flash
evaporation of the solvent occurs during the spraying, and the oxide sticks to the
disk and retains the ions from the salt. In some cases it will be necessary to lightly
abrade the CaF2 disk to permit better adhesion. The disk is masked so that one-half
is sprayed with the catalyst/support mixture while the other half is sprayed with the
47.1 Infrared Cell for Adsorption Studies on Supported Catalysts 557
Over 50 cell designs for use in the study of high area surfaces by infrared spec-
troscopy have been reported since 1967, as reviewed in [1]. The cell shown in
Fig. 47.2 possesses unique features, which, in combination, are not found in other
designs reported. These features include ultrahigh vacuum operation, wide tem-
perature range extending from about 100 K to above 1500 K without use of an
external furnace, accurate temperature control to 1 K, linear temperature pro-
grammability up to 8 K/s, and very rapid sample cooling rates [1].
The cell design in Fig. 47.2a comprises a six-way stainless cube to which
windows and gas delivery and pumping tubes may be attached by means of 2.75-in.
Conflat flanges. The sample itself is held on a tungsten grid that is clamped between
high thermal conductivity Ni clamps [5], mounted on an electrical feedthrough that
exits from the bottom of a reentrant Dewar made of stainless steel. The feedthrough
also contains matched thermocouple leads that are attached to a thermocouple
welded to the top center of the grid. The entire assembly is oriented using a
558 47 Infrared Cells
Fig. 47.2 Wide-temperature-range IR cell for high area solids. a Cell + sample holder. b Spraying
technique for sample preparation. c Hydraulic pressing technique. d Grid temperature distribution
at steady state. e Temperature programming grid
rotatable flange and is inserted into the six-way cube, along with other desired
attachments.
The grid material is made of thin tungsten foil that has precise square holes
photo-etched in 0.0125–0.05-mm-thick material and exhibiting 58 % optical
transparency in the IR [1]. Each of the square holes is 0.22 mm in width, sur-
rounded by 0.05-mm boundaries. The uniform geometry is advantageous in con-
trolling the uniformity of powder deposition, as will be shown below. Gold grids of
the same geometry have also been employed in this cell.
Two methods may be used for the deposition of high-area solids on the W grid.
As shown in Fig. 47.2b, Step 1, a spraying technique [1] like that described on
p. 123 may be used in which the powdered solid is suspended in a volatile solvent
and sprayed onto the heated grid. Heating to about 70 °C is achieved using tem-
porary power leads to the grid and its Ni clamp support, causing flash evaporation
of the solvent. The grid is rigidly held in a T-shaped Macor jig-mask that is
designed to maintain correct spacing between the Ni clamps so that they may be
slid over the power feedthrough leads on the bottom of the reentrant Dewar without
47.2 Wide-Temperature-Range IR Cell for High-Area Solids 559
flexure of the grid containing the powder deposit as shown in Step 2. The Macor jig
is then removed, as shown in Step 3. The powder first sticks to the boundaries of the
0.22-mm-wide openings and then as the deposition continues, the powder uni-
formly fills the holes. Oxide powder densities up to about 0.01 g/cm2 may be
achieved in the grid openings.
A second method for placing powder into the grid openings is shown in
Fig. 47.2c [6]. Here, the unmounted grid is held between two sheets of glossy
weighing paper and an excess of powder is sprinkled over the grid. Two stainless
steel dies are used to sandwich the grid and its protective paper, and hydraulic
pressure of about 12,000 psi is applied. This causes the powder to enter the grid
openings and to be rigidly held there. Upon removal of the glossy paper, a razor
blade is used to scrape off the excess powder from the top of the grid, and the grid is
then mounted in the Ni clamps for loading into the IR cell. The powder material in
the grid openings will withstand gentle bending of the grid. Powder densities of the
order of 0.004 g/cm2 may be achieved. The spectroscopic, gas transport, and
thermal behavior of the pressed powders is very similar to that of the sprayed
deposits [6].
The thermal properties of the grid during electrical heating to a stable temper-
ature by a feedback controller have been characterized as shown in Fig. 47.2d, e [1].
In Fig. 47.2d, four thermocouples were attached as shown in the diagram. The open
circle and black triangle points were used to measure temperature gradients along a
vertical line in the center of the grid. The three open point measurements were made
to determine the temperature gradient along a horizontal line through the center of
the grid. In the central 4-mm-wide region of the grid that is sampled by the IR
beam, no vertical gradient in temperature is observed. A small dropoff in temper-
ature is seen in the central region as one moves horizontally owing to heat con-
duction away from the grid by the Ni clamps. The ability to temperature-program
the grid and the supported powder to elevated temperatures is illustrated in
Fig. 47.2e. Here, almost linear heating rates between 1.9 and 8.4 K/s are demon-
strated, with the controller being programmed to stop the heating ramp at about
750 K in this case. Rapid cooling in 10–20 s from 900 K to the base temperature of
325 or 150 K maybe achieved [1].
Other cells for heating catalysts to high temperatures and cooling to low tem-
peratures have been described. One of these is used for EXAFS measurements and
works in a range from about 77 to 700 K [7]. Another, involving the use of a
pressed sample in a copper grid, has been described and characterized for thermal
behavior [8].
Cells in which the sample can be heated provide an ideal situation for the study
of infrared emission spectroscopy. While many IR emission studies have been done
on planar surfaces containing coatings, a few studies of high area surfaces as IR
emission sources have been carried out too [9, 10]. One must be careful to rule out
artifacts in this type of measurement [10].
560 47 Infrared Cells
Core-shell structures are sometimes able to transport adsorbed molecules from the
outside to the inside. Since the transporting of adsorbed molecules undergoes dif-
ferent interactions with the shell surfaces and the core surface, its frequency is an
excellent indicator of its progress through the shell to the core. Figure 47.4a shows
a core-shell structure which has been investigated in this manner. A *10 nm
diameter CoO shell has been prepared around a *2 nm diameter Pt particle [19].
Carbon monoxide has been added and the temperature has been raised while
recording transmission IR spectra. CO is postulated not to transport by a gas phase
route through pores in the CoO shell; however CO does transport as an adsorbed
species across CoO surfaces which surround the pores in the shell structure.
Eventually the diffusing CO reaches the Pt core and chemisorbs, giving a charac-
teristic vibrational spectrum for CO/Pt. Figure 47.4b shows the intensity of the
CO/CoO and CO/Pt spectral features as a function of temperature when the sample
is placed under 2 Torr of CO at 110 K and then warmed. Initially, no spectral
features are resolved, as the CO has not entered the shell structure. At about 150 K,
CO/CoO appears and increases in coverage as CO enters into the CoO pores. At
about 200 K, CO transport through the shell structure carries it to Pt sites on the
core, and CO/Pt spectral features begin to develop. Figure 47.4c shows the spectral
measurements as a function of temperature. This is a modification of the method
discussed in Sect. 47.3 where adsorbates are observed to transport through mac-
roscopic distances into the depth of a packed bed of a high surface area adsorbent—
but here the transport distances involved are in the nanometer regime.
monitor the free hydroxyl groups as they slowly become complexed with the
adsorbing molecules. Usually isolated M–OH groups will shift downward in fre-
quency when hydrogen bonded to adsorbed molecules, producing M–OH···X
species. This includes the Xe molecule [20] as well as many types of organic
molecules which sense the strong dipole of the M–OH group. Figure 47.5a shows
the cross section of a pressed powder involved in this inward-diffusion experiment.
It is important in such an experiment to be certain that the source of the diffusing
molecules, the ice, is not depleted as this would change the boundary condition for
the diffusion process. Figure 47.5b shows the drop in absorbance of isolated Al–OH
species and the production of complexed species as diffusion inward occurs. The
drop in normalized integrated Al–OH absorbance is shown versus time as diffusion
occurs and using Fick’s second law of diffusion, the diffusion coefficient can be
extracted from these data [12].
In the case of outward diffusion, shown schematically in Fig. 47.5c, molecules
become mobile inside the pore structure of the powdered adsorbent, which has been
saturated with adsorbate, and eventually desorb, causing the integrated absorbance
to fall. Figure 47.5d shows a fitted kinetic plot for the diffusion where two activated
diffusion processes were found, leading to slower and faster diffusion, as shown by
the Arrhenius plot [13]. Defect sites on the oxide surface are thought to be
responsible for the slow rate of diffusion over these sites. The value of the diffusion
coefficient is derivable from these data where the boundary condition of zero
coverage applies to the outer geometrical surface [13]. The outward diffusion is
sometimes accompanied by small changes in the IR spectrum as the adsorbed
molecules leave, since one preferentially samples the more strongly-bound species
as the diffusion/desorption process proceeds [13–15]. The outward diffusion
method more readily lends itself to studies of temperature dependence since in the
inward process one must remain below the melting temperature of the ice. Recently,
diffusion studies through a metal catalyst supported on an oxide were able to detect
the separate transport of an adsorbed molecule from metal- to- metal site as well as
transport across the oxide sites [21].
The adsorption of gases on high surface area solid surfaces is of interest techno-
logically. Transmission IR spectroscopy can be applied to such materials (see
Sect. 47.2) if they possess sufficient optical transparency. If the materials are
opaque, they may be studied by DRIFTS (Diffuse Reflection Infrared Fourier
Transform Spectroscopy) [22–25]. However, even opaque materials may be studied
by transmission IR spectroscopy, where transmission occurs by light scattering
through the powdered material and the adsorbed species are located along scattering
47.6 Transmission IR Spectroscopy Through Opaque Materials 565
pathways [26]. Much of the history of the use of transmission IR spectroscopy for
studying adsorption on powdered materials is contained in [27–29].
For the study of opaque materials it is well known that suspending the opaque
particles in an optically transparent matrix (the famous pressed disk method) is
useful. Figure 47.6a shows a hydraulic pressing die which may be used for making
6 different samples on a single tungsten grid support as described in Sect. 47.2.
Figure 47.6b shows a schematic of carbon particles suspended in pressed KBr,
566 47 Infrared Cells
Fig. 47.6 Transmission IR spectroscopy through opaque materials. a Hydraulic pressing die—6
samples. b Cross section through grid + KBr/C sample. c Transmission IR—CO/C/KBr.
d Transmission IR—CO/KBr
where the carbon particles are stored at KBr grain boundaries in the pressed sample.
The individual square openings in the grid are 0.22 mm2 in area, separated by W
grid elements. The pressed samples are 0.051 mm in thickness. Light scattered
through the KBr matrix interacts with the adsorbed species on the suspended carbon
particles.
Figure 47.6c shows the IR spectrum for a high area C/KBr sample when
adsorbed CO is in equilibrium with the gas at the cryogenic temperatures shown in
the figure legend. Two regions of IR absorption are seen. At 2145 and 2131 cm−1,
CO on the carbon surface is detected; at lower CO frequencies, adsorption on the
KBr surfaces is detected. For a KBr matrix alone, Fig. 47.6d shows only the low
frequency CO species as strong bands, accompanied by very low coverage species
with a frequency of 2145 cm−1, close to that of gas phase CO. It is noted that for
many molecules adsorbed on surfaces, there is a matrix shift of a few cm−1 due to
the dual interaction of the molecule with its adsorbent, coupled with interactions
with the surrounding KBr matrix [30].
References 567
References
1. P. Basu, T. Ballinger, J.T. Yates Jr, Rev. Sci. Instrum. 59, 1321 (1988)
2. H.P. Wang, J.T. Yates Jr, J. Phys. Chem. 88, 852 (1984)
3. T.P. Beebe, P. Gelin, J.T. Yates Jr, Surface Sci. 148, 526 (1984)
4. Windows of CaF2 Single Crystal Mounted in Conflat Flanges may be Obtained from Harshaw
Crystal and Electronic Products. Solon Technologies, 6801 Cochran Rd., Solon, OH, 44139
5. Custom Fabricated Grids made of W or Other Metals are Available from Buckbee Mears, 245
E. Sixth Street, St. Paul, MN 55101
6. T.H. Ballinger, J.C.S. Wong, J.T. Yates Jr, Langmuir 8, 1676 (1992)
7. F.W.H. Kampers, T.M.J. Maas, J. van Grondelle, P. Brinkgreve, D.C. Konigsberger, Rev. Sci.
Instrum. 60, 2635 (1989)
8. M. Hughes, L. Bourget, R.G. Greenler, J. Cat. 139, 691 (1933)
9. D. Kember, D.H. Chenery, N. Sheppard, J. Fell, Spectrochim. Acta 35A, 455 (1979)
10. D. Kember, N. Sheppard, J. Chem. Soc., Faraday Trans. 2(77), 1321 (1981)
11. D.B. Mawhinney, J.A. Rossin, K. Gerhart, J.T. Yates Jr, Langmuir 16, 2237 (2000)
12. S. Kim, O. Byl, J.-C. Liu, J.K. Johnson, J.T. Yates Jr, J. Phys. Chem. B 110, 9204 (2006)
13. S. Kim, X. Wang, C. Buda, M. Neurock, O.B. Koper, J.T. Yates Jr, J. Phys. Chem. C 113,
2219 (2009)
14. X. Wang, S. Kim, C. Buda, M. Neurock, O.B. Koper, J.T. Yates Jr, J. Phys. Chem. C 113,
2228 (2009)
15. I.X. Green, C. Buda, Z. Zhang, M. Neurock, J.T. Yates Jr, J. Phys. Chem. C 114, 16649
(2010)
16. W.H. Weinberg, in Kinetics of Interface Reactions, eds. by M. Grunze, H.-J. Kreuzer
(Springer, New York, London, 1986)
17. J.H. de Boer, The Dynamical Character of Adsorption (Oxford University Press, London,
1968)
18. D.D. Do, Adsorption Analysis Equilibria and Kinetics (Imperial College Press, London, 1998)
19. S. Kim, Y. Yin, A.P. Alivisatos, G.A. Somorjai, J.T. Yates Jr, J. Am. Chem. Soc. 129, 9510
(2007)
20. T.H. Ballinger, P. Basu, J.T. Yates Jr, J. Phys. Chem. 93, 6758 (1989)
21. I. Green, W. Tang, M. Neurock, J.T. Yates Jr, Science 333, 736 (2011)
22. J.T. Yates Jr, T.E. Madey (eds.), Vibrational Spectroscopy of Molecules on Surfaces (Plenum,
NY, 1987)
23. P.E. Fanning, M.A. Vannice, Carbon 31, 721 (1993)
24. B.J. Meldrum, C.H. Rochester, J. Chem. Soc. Faraday Trans. 86, 861 (1990)
25. M.B. Mitchell, V.N. Sheinker, E.A. Mintz, J. Phys. Chem. B 101, 11192 (1997)
26. A.T. Bell, M.L. Hair, Vibrational Spectroscopies for Adsorbed Species (American Chemical
Society, Washington, DC, 1980)
27. M.L. Hair, Infrared Spectroscopy in Surface Chemistry (Dekker, New York, 1967)
28. L.H. Little, Infrared Spectra of Adsorbed Species (Academic Press, New York, 1966)
29. A.V. Kiszelev, V.I. Lygin, Infrared Spectra of Surface Compounds (Wiley, New York, 1975)
30. D.B. Mawhinney, J.A. Rossin, K. Gerhart, J.T. Yates Jr, Langmuir 15, 4617 (1999)
Chapter 48
Adsorption/Desorption—Thermal
The study of the surface behavior of high area solids can be carried out using
extensions of the fundamental measurement methods developed for studies of low
area single crystal surfaces. Among the most widely used methods in single crystal
studies is temperature-programmed desorption (TPD), often called thermal
desorption spectroscopy (TDS) [1]. The TPD method, with appropriate calibration,
can be used to measure absolute surface coverages as well as the kinetics of
desorption as a function of surface coverage on both single crystals and high-area
solids.
Figure 48.1a shows a thermal desorption spectrometer that was used for the
study of ethylene adsorption on a high-area (423 m2/g) NaX zeolite [2]. The
sample, of mass 10 μg, was deposited on a Ta disk by use of a micropipet that
deposited known volumes of a slurry containing the suspended zeolite in suspen-
sion. Upon evaporation of the solvent, it was found that the zeolite particles were
separated from each other as they adhered to the Ta surface. Thus, interparticle
diffusion effects during adsorption and desorption were eliminated by the high
particle dispersion, although pore diffusion in the zeolite particles was clearly
evident in the measurements [2].
The high-area sample is first cleaned by heating in vacuum using a temperature
programmer that operates in a feedback circuit involving the thermocouple. Gases
are adsorbed in the pumped system, or for high exposures, with the gate valve
closed.
The sample is held in front of a 2-μm sampling aperture that may be opened and
closed using an O-ring seal on the end of a linear motion device. With the gate
valve closed, this permits the sampling of gases near the sample surface at high
pressures as would be necessary for the study of catalytic reactions. However, with
the gate valve open after adsorption, sampling of desorbing gas from the powdered
sample is possible in the pumped system through direct transmission from the
sample to the mass spectrometer. The mass spectrometer and pumping speed of the
system may be separately calibrated using an absolute flow of gas into the pumped
system through a calibrated leak (see p. 425). Knowing the mass of the sample and
its surface area per unit mass, the absolute coverage of adsorbate may be deter-
mined by thermal desorption measurements, integrating the pressure signal over the
time duration of the desorption peak.
Figure 48.1b shows an example of the uptake of ethylene versus exposure on
NaX zeolite at 114 K. At low exposures, adsorption is postulated to occur on the
outer surfaces of the isolated zeolite particles [2, 3]; at exposures above about 200 L
(1 L = 1 × 10−6 Torr s), adsorption occurs into the pores of the zeolite, and the
efficiency of adsorption increases significantly as indicated by the lower curve of
Fig. 48.1b. Over the entire upper range of exposures, a typical sigmoid uptake curve
is found, as shown in the upper curve of Fig. 48.1b. The inset shows an analysis that
may be used to predict the saturation coverage of the adsorbate, which occurs in the
limit of 1/ε = 0, where ε is the gas exposure.
Extensive work on the analysis of TPD spectra from high-area solids has been
reported and appropriate references are listed in [2]. Noteworthy is a study by
Demmin and Gorte [4]. A major consideration is the effect of interparticle diffusion
on the observed desorption kinetics, and the method shown here, involving very
high particle dispersion, avoids this effect, while also offering sensitivity at the
10−4 ML level for powders with very high specific surface areas. Small corrections
for the adsorption on the Ta support may be easily applied.
Similar methods have been applied to larger samples (0.1 g) of powdered
material in the form of compressed pellets, but the influence of interparticle
diffusion is not eliminated in these studies [5].
A recent review of temperature-programmed desorption from high area solids
maybe found in [6].
Fig. 48.2 Thermal desorption from high area materials. a Thermal desorption apparatus.
b. n-Heptane thermal desorption from SWNT. c Experimental detection of different SWNT
binding sites for several n-Alkanes. d Graphite versus single walled nanotubes—n-Heptane
desorption
48.2 Thermal Desorption from High Area Materials 573
Chemical vapor synthesis (CVS) is well suited for the production of alkaline earth oxide
nanoparticles (MgO, CaO, SrO, and BaO) and mixed oxides such as CaxMg1−xO.
Due to the absence of solvents, the vapor synthesis can yield clean material and controlled
particle size. The reactor is shown in Fig. 48.4a, allowing controlled oxidation of metal
vapors under reduced pressure [25]. Concentric quartz tubes are placed in a furnace with
good temperature control. The inner tube contains two ceramic boats containing several
Fig. 48.4 Chemical vapor synthesis of oxide nanoparticles. a CVS reactor. b TEM image of MgO
nanocubes
48.4 Chemical Vapor Synthesis of Oxide Nanoparticles 577
grams of metal and the temperature is adjusted under Ar flow to produce sufficient metal
vapor pressure. The inert Ar carries the metal vapor to the end of the inner tube where it
meets flowing oxygen. The exothermic oxidation reaction produces a bright stable flame
at the end of the inner tube and alkaline earth oxide nanoparticles are produced as a result
of homogeneous nucleation in the gas phase. Due to continuous pumping, the residence
time of the oxide nuclei within the flame remains short enough to prevent substantial
coarsening and coalescence. A by-pass system (Valve V1) allows one to avoid particle
collection during uncontrolled process conditions during initial heating. Following par-
ticle synthesis and collection in the particle collection net, the furnace is rapidly cooled by
flowing air through the region surrounding the outer quartz tube. The oxide nanoparticles
may be transferred in air to other instruments for study, and cleaning procedures in
vacuum should be used to remove contaminants adsorbed on the nanoparticles during
transfer. An electron micrograph of nanometer sized MgO nanocubes, exposing MgO
(100) faces is shown in Fig. 48.4b.
References
Power interruptions can play havoc with UHV systems operating with a turbopump,
and particularly when these systems are being baked out. A power failure has the
effect of venting a turbopump, using the commercial vent valve designed to prevent
oil back-streaming from the forepump into the turbopump and the vacuum chamber
during a power interruption. A partial solution to avoid massive air entry into the
UHV system is to employ a pneumatic gate valve between the turbopump and the
vacuum system. However, there are two defects in this protection system: (1) the
time of closure of the gate valve is too long to prevent at least partial loss of vacuum
in the chamber; and, (2) when the power returns, the gate valve will quickly open,
exposing the chamber to a turbopump which is at or near atmospheric pressure.
A circuit to prevent these effects is described below [1]. There is a commercial
device that will do almost the same thing electrically [2], and a pneumatic device
also for the same purpose [3].
The first problem mentioned above is solved by venting the turbopump only
after the gate valve is completely closed. This is done by supplying short-term
emergency power to the turbopump vent valve for the necessary period of time
(approximately 8 s is used). The second problem is solved by controlling the
pneumatic gate valve with a relay [4] that latches in the off position upon power
failure. This prevents the valve from reopening when the power returns until the
relay is overridden.
Figure 49.1 shows the circuit, costing less than $100, which accomplishes both
objectives. The nondelayed socket supplies power to the gate valve relay. A line
switch between the nondelayed socket and the gate valve supplies manual control
of the gate valve. The nondelayed socket power is designed to latch in the off
position in the case of a power interruption, and pushing the reset button reactivates
power at this socket when the main power is restored.
Fig. 49.1 Delay circuit for turbopumping protection against vacuum loss from power interruption
The delayed socket supplies emergency power to the turbopump vent valve. It is
powered by a battery-driven backup AC power supply [5]. When main power is
interrupted, emergency power is supplied and then shuts off, venting the turbo-
pump. For momentary main power interruptions lasting less than 8 s, the turbo-
pump vent valve will not open because power is supplied from the backup supply.
The control system does not assume that when power returns the turbopump
restarts and can be automatically opened to the vacuum system, unlike other sys-
tems [2, 3]. It is deemed best to leave the system shut down after power inter-
ruptions, awaiting manual restarting.
49.2 Fast-Closing Beam Valve for UHV Chamber Protection 583
not degrade the vacuum. The housing is tightened into place by using a nut that
engages a threaded rod at the top of the assembly.
When the current is on, the wedged-shaped Cu seal is drawn up into the sole-
noid; interruption of vacuum results in turning the solenoid off, and the Cu seal is
dropped into the sealing position, where it is held in place by a spring and by
gravity. The seal is made against a Viton O-ring in the lower portion of the valve.
Copper cooling coils are used to dissipate heat from the solenoid. Electrical con-
nections are brought out from the solenoid housing through a Conflat flange, not
shown.
The leak rate through the closed valve is 4 × 10−5 Torr L/s, and is too large to be
used to protect an ultrahigh vacuum system without a pumping section between the
valve and the system. In the case of the use of this valve, a differentially pumped
chopper region is present and serves this purpose [7].
A similar principle has been used to insert a large gate valve into an ultrahigh
vacuum system [8].
Modern oil diffusion pumps are now used to achieve ultrahigh vacuum, and these
pumps exhibit some advantages over more expensive ion pumps or turbomolecular
pumps. Among the advantages are lower cost, the ability to pump all gases, the
absence of magnetic fields, and the absence of vibrations. Among the disadvantages
are the possibilities of system contamination if failure of various components occurs
[9]. The failure items are loss of vacuum, loss of power, loss of cooling water, and
loss of liquid nitrogen cooling in the ultrahigh vacuum trap. A pneumatically
actuated valve between the pumps and the ultrahigh vacuum system is the key
element that closes to protect the system if some failure should occur.
A schematic drawing of an oil diffusion pumped ultrahigh vacuum system is
shown in Fig. 49.3a. It consists of a standard bakeable chamber with an ionization
gauge and a titanium sublimation pump (TSP). The pneumatically actuated gate
valve separates the system from the external diffusion pump and liquid nitrogen
trap. The oil diffusion pump contains special perfluoro polyether pump oil (Fomblin
type) or polyphenyl ether (Santovac-5) pump oil used for achieving the best
pressure. Silicone diffusion pump oils should be avoided in all UHV apparatus. The
oil diffusion pump is protected from the oils of the mechanical pump by a bakeable
molecular sieve trap, and the foreline pressure is monitored by a thermocouple
gauge or other pressure gauge.
The simple relay circuit shown in Fig. 49.3b is designed to protect the system
from the various failure possibilities [9]. Digital electronic components were
avoided to reduce the circuit sensitivity to electrical noise. The control items in the
vacuum system are: (1) foreline thermocouple gauge maximum pressure set point—
49.3 Safety System for Oil Diffusion-Pumped UHV Systems 585
Fig. 49.3 Safety system for oil-diffusion-pumped UHV systems. a Schematic oil diffusion
pumped UHV system. b Safety system schematic
to J1; (2) water flow sensor—to J2; (3) ionization gauge maximum pressure setpoint
—to J3; (4) liquid nitrogen filler with low level sensor—to J4; (5) pneumatically
actuated system valve—to J5. The relay, Rl, is a double-pole, single-throw relay
and is turned on by closing the power switch, SW1, and then depressing the
momentary contact switch, SW2. This sets R1 if the fore-line pressure is below the
586 49 Protecting the Vacuum System
thermocouple gauge set point and if there is sufficient water flowing through the
diffusion pump, closing J2. The switch SW3, which controls the relay, R2, may
then be used to switch on the diffusion pump. The system valve, SW4, can then be
actuated if continuity exists through the right-hand side of the circuit, meaning that
the liquid nitrogen trap and the ion gauge are operating in the proper range. If one is
in the initial pumpdown period, the pressure will be too high at the ionization
gauge, and this gauge and the thermocouple gauge can be bypassed temporarily by
closing switch SW5.
Table 49.1 shows the various failure conditions and the components’ condition
when these failures occur.
Rather long linear and rotary magnetic translators are often used in ultrahigh
vacuum systems, and these are always in danger of being hit by laboratory per-
sonnel as they pass by. The design shown in Fig. 49.4 prevents this by folding the
device housing containing the retracted magnetic translator/rotor.
Figure 49.4a shows a side view of the device. The welded bellows is folded
upward through a 90° angle by a hinge and locked into place. In the horizontal
position the bellows is held in a slightly compressed configuration by the gimbal
subassembly. The outer end of the outer cylinder of the translator has a threaded rod
attached on axis and this is retained in a “scissors” subassembly that permits
accurate vertical positioning of the end of the translator. The bench holding the
device is attached to the system lower flange by a rigid yoke assembly.
Figure 49.4b shows the gimbal subassembly and the two rotational adjustments
that are possible as well as a lateral translation for alignment. A double-sided flange
is used in order to be able to disassemble either the bellows or the translator section
independently. Washers, used as spacers under the gimbal, permit vertical
adjustment.
A critical design consideration is the appropriate size of the welded bellows.
Figure 49.4c shows the dimensional parameters associated with a welded
bellows-namely, the relaxed or free length (FL) at one atmosphere, the extended
49.4 Folding Linear Magnetic Translator 587
Fig. 49.4 Folding linear magnetic translator. a Side view. b Gimball subassembly. c Welded
bellows—dimensional definitions. d Bellows in bent configuration
length (EL), and the compressed length (CL). These parameters are specified by the
bellows manufacturer. In the bent condition, the inner radius of the bellows is
defined as the working length (WL) in Fig. 49.4d. The outer length of the bent
bellows is therefore WL + θr, and the condition in (49.1) must be met:
WL þ hr\EL ð49:1Þ
References
1. J.A. Polta, P.A. Thiel, J. Vac. Sci. Technol. A5, 386 (1987)
2. Balzers Part No. TCV 102. Obtained from Balzers, 8 Sagamore Park, Hudson, NH 03051
3. J.P. Saint-Germain, G. Abel, B.L. Stansfield, J. Vac. Sci. Technol. A4, 2391 (1986)
4. Time delay CNS-35-76 from Newark Electronics, 4801 N. Ravenswood Avenue, Chicago, IL
60640-4496
5. Available from Tripp Lite, 500 N. Orleans Street, Chicago, IL 60610
588 49 Protecting the Vacuum System
The quantitative effects of electrical current flowing through the human body are
listed in the Table 50.1 [1]. It is the current flow through the body that is the
fundamental parameter involved in electrical shock, and the level of danger with a
high-voltage power supply is related to the current rating of the supply. Safety with
electrical devices involves minimizing accidental current flow through the body,
especially through the heart. One may have to work with one hand in a pocket when
there is danger of shock from high-voltage wires, but of course this is not an
adequate protection against shock. Electrical shielding of leads and terminals is an
obvious precaution that must be taken.
The avoidance of electrical shocks in the laboratory is carried out with an
understanding of the meaning of the electrical ground and how to connect electrical
equipment to ground. Figure 50.1a shows a three-wire wall outlet and a polarized
male plug. The green ground wire is a non-current-carrying voltage reference line
that is connected to the earth somewhere near the laboratory building. The white
wire, which is neutral, is connected to the ground wire in the vicinity of the wall
outlet. At the point of connection, the electrical potential of the white wire will also
be at ground potential. However, as one moves from this connection point, there
will be a change in the potential of the white wire owing to the IR drop caused by
the current flow through the white wire. One should avoid connecting the white
wire to the ground wire in the laboratory. The black wire is the hot wire, and it
remains isolated from both the ground and the neutral wire. The wall outlet contains
three contacts—the large slot for the neutral wire, the small slot for the hot wire,
and the larger hole for the ground connection.
Fig. 50.1 Electrical shocks in the laboratory. a 110 V AC wall outlet + polarized plug. b Coupled
chassis ground
potential in the kilovolt range is applied to the grid for normal electron bombard-
ment cleaning. An electrically isolated feedthrough inside the vacuum system is
immersed in the plasma and comes to the plasma potential which is near the
potential of the grid. Contact with this feedthrough electrode external to the vacuum
system can result in exposure to high electrical potentials, and one example of this
592 50 Protecting Personnel
effect leading to death has been reported. A safe operating procedure to avoid this
type of effect is never to attempt to outgas an ionization gauge when the pressure is
high, which would be a pointless thing to do anyway.
Reference
French, T.E. (Revised by Vierck, C.J.), A Manual of Engineering Drawing, 7th Ed.
(McGraw Hill, New York, NY, 1947).
Gomer, R., Field Emission and Field Ionization (American Institute of Physics,
New York, NY 10017-3483, 1993).
Gordon, A. J. and Ford, R.A., The Chemist’s Companion—A Handbook of Data,
Techniques, and References (Wiley, New York, NY, 1972).
Hablanian, M.H., Diffusion Pumps, AVS Monograph Series, M-5 (American
Vacuum Society, New York, NY 10017, 1983).
Kennard, E.H., Kinetic Theory of Gases (McGraw-Hill, New York, NY, 1938).
Kohl, W.H., Handbook of Materials and Techniques for Vacuum Devices
(American Institute of Physics, New York, NY 10017-3483, 1995).
Leco Corporation, Metal Etching and Polishing (St. Joseph, MI 49085, 1977).
Leybold-Heraeus, Vacuum Technology Its Foundations, Formulae, and Tables
(Leybold-Heraeus, Pittsburgh, PA 15632).
Loeb, L.B., The Kinetic Theory of Gases, 3rd Ed., (Dover, NY, 1961).
Lunn, G. and Sansone, E.B., Destruction of Hazardous Materials in the
Laboratory (Wiley, New York, NY, 1990).
Madey, T.E., Ed., Vacuum Science and Technology, History of Vacuum Science
and Technology, Vol. 1 (American Vacuum Society, AIP Press, New York, NY,
1984).
Maissei, L.I. and Glong, R., Handbook of Thin Film Technology (McGraw-Hill,
New York, NY, 1970).
Mark, H.F., Ed., et al., Encyclopedia of Chemical Technology (Wiley, New York,
NY, 1979).
Moore, J.H., Davis, C.C. and Coplan, M.A., Building Scientific Apparatus
(Addison-Wesley Reading, MA, 1983).
Müller, E.W. and Tsong, T.T., Field Ion Microscopy (Elsevier, New York, NY,
1969).
O'Hanlon, J.F., A User’s Guide to Vacuum Technology (Wiley, New York, NY,
1980).
The Omega Complete Temperature Measurement Handbook and Encyclopedia,
Omega Engineering, Inc., P.O. Box 4047, One Omega Drive, Stamford, CT 06907.
Pierce, J.R., Theory and Design of Electron Beams (D. van Nostrand Co., Inc.,
New York, NY, 1949); 2nd Ed. (1954).
Redhead, P.A., Hobson, J.P. and Kornelsen, E.V., The Physical Basis of Ultrahigh
Vacuum (American Institute of Physics, New York, NY 10017-3483, 1993).
Redhead, P.A., Ed., Vacuum Science and Technology, History of Vacuum Science
and Technology, Vol. 2 (American Vacuum Society, AIP Press, New York, NY,
1994).
Roberts, R.W. and Vanderslice, T.A., Ultrahigh Vacuum and Its Applications
(Prentice-Hall, Engelwood Cliffs, NJ, 1963).
Robinson, N.W., The Physical Principles of Ultrahigh Vacuum Systems and
Equipment (Chapman and Hall, London, 1968).
Rosebury, F., Handbook of Electron Tube and Vacuum Techniques (AIP,
American Institute of Physics, New York, NY, 1993).
Appendix A 595
Manufacturers
Note: Page numbers followed by ‘f’ denotes Figure and ‘n’ denotes Footnote