New Developments in Transmission Electron Microscopy For Nanotechnology.

REVIEW
New Developments in Transmission Electron

Microscopy for Nanotechnology**
By Zhong Lin Wang*
High-resolution transmission electron microscopy (HRTEM) is one of the most

powerful tools used for characterizing nanomaterials, and it is indispensable for na-
notechnology. This paper reviews some of the most recent developments in electron microscopy techniques for
characterizing nanomaterials. The review covers the following areas: in-situ microscopy for studying dynamic
shape transformation of nanocrystals; in-situ nanoscale property measurements on the mechanical, electrical and
field emission properties of nanotubes/nanowires; environmental microscopy for direct observation of surface
reactions; aberration-free angstrom-resolution imaging of light elements (such as oxygen and lithium); high-angle
annular-dark-field scanning transmission electron microscopy (STEM); imaging of atom clusters with atomic
resolution chemical information; electron holography of magnetic materials; and high-spatial resolution electron
energy-loss spectroscopy (EELS) for nanoscale electronic and chemical analysis. It is demonstrated that the
picometer-scale science provided by HRTEM is the foundation of nanometer-scale technology.
1. Electron Microscopy and Nanotechnology designing, modifying, and constructing nanoscale structures.
Third, the newly discovered nanostructures, such as carbon
Nanotechnology, as an international initiative for science nanotubes, quantum dots, semiconducting oxide nanobelts, etc.
and technology in this century, is a focused area of current re- display the diversity and richness of the nano-scale world.[2]
search. Advanced nanomaterials and advanced manufacturing The unique, novel, and largely improved properties demon-
are the foundation of nanotechnology. Tracking the historical strated by these structures have illustrated a blue print for the
background of why nanotechnology and why now, it is be- next technological revolution in human civilization. Finally, the
lieved that the following facts may contribute to this global in- powerful modeling techniques and supercomputers can predict
itiative.[1] First, the road map for microelectronics and micro- possible phenomena that could be realized experimentally,
systems will be reaching its limit in about 10 years. As the size providing guidance in materials design and system analysis.
of interconnects is thinner than 100 nm and one electron can High-resolution transmission electron microscopy
ignite the switching of a device (so called single electron de- (HRTEM) is one of the most powerful tools used for character-
vice), quantum mechanical phenomena are becoming appreci- izing nanomaterials, and it is indispensable for nanotechnol-
able and dominant. Nanoscale technology is not simply a min- ogy.[2] In fact, decades before the national nanotechnology in-
iaturization in sizes, but an in-depth revolution in physical itiative, scientists had started examining ªsmall particlesº
concepts, system design, and materials manufacturing. Philo- (nowadays these are called ªnanoparticlesº) by HRTEM. It
sophically, a change in quantity results in a change in quality. was not until the early 1990s that inventions of various types of
Secondly, the development of transmission electron microsco- scanning probe microscopy allowed scientists to manipulate at
py and scanning probe microscopy allows direct imaging of the nanoscale. Traditionally, HRTEM has been mainly applied
atomic structures in solids and on surfaces, and these powerful for imaging, diffraction, and chemical analysis of solid materi-
tools provide the ªeyesº and ªhandsº for imaging and manip- als.[3] Carbon nanotubes, for example, were first identified by
ulating the nanoscale world, fulfilling the dream of moving HRTEM.[4,5] Analysis of such tubular structures requires ex-
atoms ªone-by-oneº. This provides a unique opportunity for tensive development of electron microscopy, which has been
covered very comprehensively in the book edited by Wang.[6]
± Conventional imaging and diffraction are the two most power-
[*] Prof. Z. L. Wang ful methods in characterizing the phase structure and phase
School of Materials Science and Engineering
Georgia Institute of Technology transformation of inorganic materials. With the assistance of
Atlanta, GA 30332-0245 (USA) energy dispersive X-ray spectroscopy (EDS) and electron en-
E-mail: [email protected]
ergy-loss spectroscopy (EELS), the transmission electron mi-
[**] The author is grateful to his collaborators: Drs. Xuedong Bai, Ruiping
Gao, Enge Wang, Yolande Berta, Christopher Ma, and Mostafa El-Sayed. croscope is a versatile and comprehensive analysis tool for
Thanks the financial support from NSF and NASA (URETI) and NSFC. characterizing the chemical and electronic structure at nano-
Adv. Mater. 2003, 15, No. 18, September 16 DOI: 10.1002/adma.200300384 Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1497
Z.L.Wang/New Developments in Transmission Electron Microscopy
REVIEW
scale.[7] HRTEM has been extensively applied to determine The large percentage of surface atoms in nanocrystals is the
the shape of nanoparticles[8±10] and structures of nanotubes.[6] origin of their unique properties. The melting point of a nano-
In recent years, a number of new and novel developments crystal is much lower than the bulk melting temperature.[14]
have been made in HRTEM for nanotechnology, such as in- Here, the melting of Pt particles is taken as an example.[15]
situ microscopy for observing dynamic processes at nanoscale, Platinum nanoparticles with a high percentage of cubic-, tetra-
nanomeasurement techniques that directly correlate physical hedral-, and octahedral-like shapes, respectively, have been
properties with structures, holographic imaging of electric and synthesized by changing the ratio of the concentration of
magnetic fields, quantitative chemical mapping at sub-nano- polymer capping material (polyacrylate) to that of Pt2+ ions
meter resolution, and ultra-high resolution imaging techniques. being reduced by H2 from K2PtCl4 at room temperature.[16]
In consideration of the large diversity and amount of literature Cubic and tetrahedral are the two most typical shapes for the
available in electron microscopy, instead of giving a vast review Pt nanocrystals. We now apply in-situ TEM to determine the
about HRTEM, this article reviews only the novel develop- stability of particle shapes and the melting behavior.
ments that have been made in the last few years for solving the Figure 1 shows a series of TEM images recorded from the
problems in nanotechnology. Readers who are interested in same region when the specimen temperature was increased
fundamentals of TEM are referred to existing text books.[11±13] from 25 to 610 C. These images were selected from a group
of images to present the most significant changes in particle
shape. For easy notation, particles are labeled as groups to
2. In-Situ Microscopy track their shape transformation behavior. Most of the parti-
cle shapes showed no significant change when the specimen
In-situ microscopy refers to the techniques that allow a di- temperature was below 350 C (Figs. 1a±c). Truncated cubic
rect observation of the dynamic properties at the nanoscale and tetrahedral particles were formed when the temperature
through imaging and diffraction. The most traditional in-situ reached 410 C (Fig. 1d). The corners and edges of the parti-
technique is thermal induced structural transformation and cles were disappearing because these local atoms have a high-
transition. More recent developments include in-situ property er energy. The tetrahedral particles could still be identified
measurements of nanotubes/nanowires, electric transport even when the temperature reached 500 C (Fig. 1e), while
through a nanotube, and environmental microscopy. the cubic particles became spherical when the temperature
was above 500 C. This indicates that the tetrahedral particles
are more stable than cubic ones, possibly because the {111}
2.1. Thermally Induced Surface Dynamic Processes of surfaces have lower surface energy than the {100}. A feature
Nanocrystals observed in Figure 1 is that the tetrahedral shape of a particle
labeled as T1 was preserved when the specimen temperature
In-situ study of the temperature induced phase transforma- was as high as 610 C. This is possibly due to the contact of the
tion and the structural and chemical evolution of nanocrystals apexes of the particle with the adjacent particles, so that the
is important for understanding the structure and structural interparticle diffusion can still sustain the shape of the apexes.
stability of nanomaterials. TEM is an ideal approach for con- These results indicate that the surface capping polymer is
ducting this type of experiments, in which a specimen can be removed by annealing the specimen to a temperature of 180±
cooled down to the temperatures of liquid nitrogen or liquid 250 C, while the particle shape shows no change up to
helium or heated to 1200 C. The in-situ process can be re- ~ 350 C. In a temperature range of 350 to 450 C, a small trun-
corded at TV rate for exhibiting the time and temperature de- cation occurs in the particle shape but there is no major shape
pendent phenomena. transformation. The particle shape experiences a dramatic
Zhong Lin Wang received his Ph.D. in Physics from Arizona State University in 1987. He is cur-
rently a Professor, the Director of the Center for Nanoscience and Nanotechnology, and the
Director of the Electron Microscopy Center at the Georgia Institute of Technology. Prof. Wang
has authored and co-authored four scientific reference and textbooks and 320 journal articles,
edited and co-edited nine volumes of books on nanotechnology, and is amongst the top 20 most
cited authors in nanotechnology from 1992 to 2002. He was elected to the European Academy of
Science in 2002, and received the 2001 S.T. Li prize for Outstanding Contribution in Nano-
science and Nanotechnology, the 2000 Georgia Tech Faculty Research Award, the 1999 Burton
Medal from the Microscopy Society of America, 1998 US NSF CAREER grant award, and 1998
China-NSF Oversea Outstanding Young Scientists Award. He is an associated editor for The
European Physical JournalÐApplied Physics, member of the editorial board of Micron,
Advanced Functional Materials, Journal of Nanoscience and Nanotechnology, and Progress in
Natural Science. Prof. Wang's most recent research focuses on oxide nanobelts and nanowires,
in-situ techniques for nano-scale measurements, self-assembly nanostructures, fabrication of
nanodevices, and properties of magnetic nanostructures.
1498 Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim http://www.advmat.de Adv. Mater. 2003, 15, No. 18, September 16
REVIEW
Fig. 1. A series of TEM images recorded in-situ
from Pt nanocrystals dispersed on a carbon sub-
strate and heated to different temperatures, exhib-
iting the shape transformation as well as surface
melting phenomenon of the particles at tempera-
tures much lower than the melting point of the
bulk. Models for cubic and tetragonal polyhedra
are also shown.
transformation into spherical-like shape when the tempera- cess.[24] The nanobelts have a distinct structural morphology,
ture is higher than ~ 500 C (the melting point of bulk Pt is characterized by a rectangular cross-section and a uniform
1773 C). structure, which could be directly used as nanocantilevers
Using the confined and unique geometry provided by nano- andnanoresonators in nanoelectromechanical systems
materials, other interesting experiments have been carried out (NEMS).[30,31] A key phenomenon for applying nanobelts in
using in-situ TEM. Gao and Bando[17] have demonstrated a NEMS technology is their electromechanical resonance
nanothermometer in TEM. By trapping liquid gallium in a behavior, and an important physical quantity for cantilever
closed carbon nanotube during the nanotube growth, the vol- applications is their bending modulus.
ume/height of the gallium liquid increases as the temperature Measurement of the bending modulus of a nanowire/nano-
increases, forming a nanothermometer. tube/nanobelt object with a length of a few micrometers is
rather challenging because its small size prevents the applica-
tion of the established techniques for mechanical testing. Re-
2.2. In-Situ Measurement of Mechanical Properties of cently, a new experimental approach, based on the electric-
Nanotubes/Nanobelts/Nanowires[18] field-induced resonant excitation, has been developed for di-
rect measurement of the mechanical properties of individual
Cantilever-based sensors present wide range applications in nanowire-like structures by in-situ TEM.[32,33] Using this
the field of physical, chemical, and biochemical sciences,[19±21] method, mechanical properties of carbon nanotubes[32,34] and
which are based on the bending induced by the surface ad- silicon carbide-silica composite nanowires[35] have been inves-
sorbed foreign molecules as a result of changes in surface en- tigated. Here the application of this technique is demonstrat-
ergy. The most conventional cantilevers are based on Si, SiC, ed for studying the resonance behavior of a nanobelt.[36]
or SiN, with a typical length of 50 lm, width of 5 lm, and To carry out property measurements of a nanobelt, a key
thickness of 0.5 lm. Miniaturization of the cantilever dimen- step is to design and build a specimen holder that can apply
sions toward the nanometer scale will enable an improvement an electric field across nanostructures. The specimen holder
in sensitivity, spatial resolution, energy efficiency, and time of requires the translation of the nanobelt via either mechanical
response.[22,23] Semiconducting oxide nanobelts[24] and nano- movement by a micrometer or axial directional piezo. The
wires,[25] as a new class of nanostructures, are showing novel static and dynamic properties of the nanobelt can be obtained
electronic, chemical, and optical properties.[26,27] Nanowires by applying a controllable static and alternating electric field.
are the fundamental building blocks for fabricating nanoscale Figure 2a shows a typical TEM image of the ZnO nanobelt,
sensors and logic circuits.[28,29] Recently, nanobelts of semicon- displaying high structural uniformity. The selected area elec-
ducting oxides have been synthesized using a solid±vapor pro- tron diffraction pattern indicates that the nanobelt grows
Adv. Mater. 2003, 15, No. 18, September 16 http://www.advmat.de Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1499
REVIEW
Fig. 2. a) A typical TEM image of a ZnO nanobelt. b) Schematic geometrical

shape of a nanobelt. Fig. 3 Electromechanical dual-mode resonance of a ZnO nanobelt: a) closely
perpendicular to the viewing direction (mx = 1.25 MHz), b) nearly parallel to the
viewing direction (mx = 1.38 MHz).
along [0001] and is enclosed by ± (21Å1Å0) and ± (011Å0) facets.[24]
Figure 2b represents the geometry of a nanobelt with length
L, width W, and thickness T. In our experiment, the nanobelts the fundamental resonance frequency (m1) and the dimensional
were glued at one end onto a copper wire, while the other end sizes (L, W, and T) of the investigated ZnO nanobelts. Based
near the counter electrode was free. This is a simple one-end on the experimentally measured data, the bending modulus of
clamped cantilever with a rectangular cross section. An oscil- the ZnO nanobelts is calculated using Equation 1 and Equa-
lating voltage with tunable frequency was applied to the nano- tion 2. The experimental results are summarized in Table 1.
belt. Mechanical resonance of the nanobelt can be induced if The bending modulus of the ZnO nanobelts was ~ 52 GPa.
the applied frequency matches the natural vibration fre- This value represents the modulus that includes the scaling ef-
quency (Fig. 2). From the classical elasticity theory for a rec- fect and geometrical shape, and it is in excellent agreement
tangular beam,[37] the fundamental resonance frequency cor- with the elastic modulus measured by a nanoindenter.[21] Al-
responding to the thickness direction (x-axis) (Fig. 2c) is though nanobelts of different sizes may have slight difference
2 r in bending modulus, there is no obvious difference if the calcu-
b T Ex
mxi i 2 (1) lation was done using either Equation 1 or Equation 2. The ra-
4pL 3r
tio of two fundamental frequencies my1/mx1 is consistent with
where bi is a constant for the ith harmonic: b = 1.875 and the aspect ratio W/T, as expected from Equation 1 and 2 (my1/
b2 = 4.694, Ex is the bending modulus for the vibration along mx1 = W/T if Ex = Ey). The nanobelts could be used as a new
the x-axis, L is length of the nanobelt, r is the mass density; type single crystalline nanoresonator and nanocantilever with
and the corresponding resonance frequency in the width di- dual operation modes, which could be useful in nanoelectro-
rection (y-axis) (Fig. 2c) is given by mechanical systems and highly functional nanodevices.
2
s Other in-situ mechanical testing, such as stretching and
bi W Ey
myi (2) nano-indentation have also been observed by in-situ TEM.[38]
4pL2 3r
The failure modes of carbon nanotubes under load have been
A selected ZnO stationary nanobelt is given in Figure 3a. By observed by in-situ TEM. Nanoindentation to measure the
changing the frequency of the applied voltage, we have found hardness of a thin film, while the creation and movement of
two fundamental frequencies in two orthogonal transverse vi- dislocations was observed simultaneously, has also been illus-
bration directions. Figure 3b shows a harmonic resonance with trated.[39] But quantitative calibration of the force applied re-
its vibration plane nearly parallel to the viewing direction, and mains a key challenge. Recently, atomic force microscopy has
Figure 3c shows the harmonic resonance with the vibration also been used in conjunction with TEM to study gold±gold
plane closely perpendicular to the viewing direction. For calcu- interaction, showing the possibility of integrating AFM with
lating the bending modulus, it is critical to accurately measure TEM.[40]
Table 1. Bending modulus of the ZnO nanobelts. Ex and Ey represents the bending modulus corresponding the
resonance along the thickness and width directions, respectively.
Length Width Thickness Fundamental Frequencies Bending Moduli
L [lm] W [nm] T [nm] [Gpa] [Gpa]
Nanobelt (± 0.05) (± 1) (± 1) W/T mx1 my1 my1/mx1 Ex Ey
1 8.25 55 33 1.7 232 373 1.6 46.6 ± 0.6 50.1 ± 0.6
2 4.73 28 19 1.5 396 576 1.4 44.3 ± 1.3 45.5 ± 2.9
3 4.07 31 20 1.6 662 958 1.4 56.3 ± 0.9 64.6 ± 2.3
4 8.90 44 39 1.1 210 231 1.1 37.9 ± 0.6 39.9 ± 1.2
REVIEW
2.3. In-Situ Measurement of the Electric Transport in a ing cross-section leads to the disappearance of the quantum
Carbon Nanotube[41,42] conductance effect.
My co-workers and I have previously shown that a defect-
Measurements of electric transport in nanotubes usually free nanotube can be a room-temperature ballistic conductor
use the two-point or four-point contact methods, in which the without heat dissipation.[45] However, at the junction of two
contacting electrodes are fabricated using lithography and the nanotubes, defects must be introduced, such as the case shown
nanotube is laid down onto the electrodes.[43] An alternative in Figure 5a, thus, heat dissipation at the defects is possible.
method has been developed to measure the nanotube electric After applying a higher voltage, the burning point between
transport in TEM.[44] Using the same type of specimen holder the two nanotubes is just the junction point, as shown in
as used for mechanical property measurement, by changing Figure 5b. TEM provides powerful evidence for this transport
the count electrode into a soft contacting metal, such as a tiny phenomenon.
mercury droplet, a carbon nanotube can be directly put in
contact with the electrode, and thus its transport properties
can be measured (Fig. 4). The advantage offered by such a set
up is that the nanotube can be directly imaged under the elec-
tron beam, while its electrical properties are characterized.
Fig. 4. In-situ observation of electric transport through a single carbon nano-

tubes. a) Applied voltage V = 0.1 V, current I = 7.3 lA, the resistance of the
carbon R = 12.7 kX, which corresponds to a conductance G = (12.7 kX)±1
= 1.02 G0. b) After applying a 4 V voltage, the nanotube was broken and the Fig. 5. Examples of nanotube failure modes under applied current. Multi-walled
break occurred at the contact of the nanotube and the Hg. c) The conductance carbon nanotubes a) before and b) after passing a large current, showing the
of this carbon nanotube contaminated with graphitic particles is 0.25 G0. break points at the defect sites. c,d) A nanotube before and after passing a large
current through it. In (d) the tip of the nanotube had been contacted to a Au
surface and current was passed trough it. Only the outer layer, starting at the
The contact area between the tube and the mercury can be contact point has been damaged, suggesting that the current flows over the sur-
significantly increased by inserting the nanotube into the mer- face.
cury, thus, the contact resistance is drastically reduced and be-
comes negligible. This technique was first accomplished using The room-temperature ballistic transport of a carbon nano-
the tip of an AFM and it demonstrated room-temperature tube was attributed to the single layer conductivity of the
ballistic conductance of a multi-walled carbon nanotube.[45] nanotube. Graphite has a unique layer structure, in which the
The in-situ TEM provides a direct image of the nanotube conductivity parallel to the graphitic plane (a±b axis plane) is
while the electric properties are being measured. For a multi- about 50 times higher than that along the c-axis. If the applied
walled carbon nanotube, if the applied voltage is a fraction of voltage is small, so that the interlayer tunneling effect can be
a volt, the conductance of the nanotube is measured to be ignored and the end of the nanotube is closed, the electric cur-
close to (12.7 kX)±1, which is very close to the single chance rent mainly flows along the top surface layer, provided the
conductance G0 = 2e2/h = (12.9 kX)±1 predicted by quantum tube is perfect and there are no defects. Since the layer thick-
mechanics. For a carbon nanotube without defects, the link ness of ~ 0.34 nm is comparable to the wavelength of the con-
between the graphitic sheets is weak, and there is little tunnel- duction electron in the graphitic (~ 0.5 nm), quantum conduc-
ing effect between the layers if the applied voltage is small. tance is possible. Figure 5d shows a multi-walled nanotube
The current is effectively carried by the outmost layer al- after passing a large current, which was brought into contact
though the nanotube is multi-walled.[45] If the applied voltage with an Au surface during the field emission experiments.
is increased to 4 V, the nanotube burned out at the contacting Due to a flow of a shock short-circuit current, the surface of
point with the Hg electrode (Fig. 4b). This is due to the con- the nanotube is disrupted starting at the contact point while
tact between the nanotube and the mercury. At such a high the interior layers are not visibly affected. This observation
voltage, tunneling between graphitic layers is strongly en- provides further evidence that the electronic current passes
hanced, thus the entire thickness of the wall would be effec- over the surface layers of the nanotube, e.g., the single layer
tive for conducting current. The increased size of the conduct- conductance.
REVIEW
2.4. In-Situ Measurement the Work Function at the Tip of a The force acting on the NT is proportional to the square of
Carbon Nanotube[46] the total charge on the nanotube
An important physical quantity in electron field emission is F = b [Q0 + ae(VDC + VAC cos 2pft)]2
the surface work function, which is well documented for ele-
mental materials. For emitters such as carbon nanotubes = a2 b {[(WAu ± WNTT + eVDC)2 + e2 VAC
2
/2]
(NTs), most of the electrons are emitted from the tips of the + 2eVAC(WAu ± WNTT + eVDC) cos 2pft
carbon NTs, and it is the local work function that matters to + e2 VAC
2
/2 cos 4pft} (4)
the properties of the NT field emission. The work function
measured from the ln(J/E2) versus 1/E characteristics curve,
where E is the macroscopic applied electric field that is an where b is a proportional constant. In Equation 4, the first
average over all of the aligned carbon NTs that are structural- term is constant and it causes a static deflection of the carbon
ly divers in diameters, lengths, and helical angles, and J is the NT; the second term is a linear term, and the resonance occurs
measured electric current density. We have developed a tech- if the applied frequency, f, approaches the intrinsic mechanical
nique for the measurement the work function at the tip of a resonance frequency, f0, of the carbon NT (Fig. 7a). The last
single carbon nanotube.[47] term in Equation 4 is the second harmonics. The most impor-
Our measurement is based on the electric field induced me- tant result of Equation 4 is that, for the linear term, the reso-
chanical resonance of carbon nanotubes, but with a slight nance amplitude A of the NT is proportional to VAC(WAu ±
modification. The principle for work function measurement is WNTT + eVDC). By fixing the VDC and measuring the vibration
schematically shown in Figure 6a. We consider a simple case amplitude as a function of VAC, a linear curve is obtained
in which a carbon nanotube, partially soaked in a carbon fiber (Fig. 7c).
produced by arc-discharge, is electrically connected to a gold Experimentally, we first set VDC = 0 and tuned the fre-
quency, f, to find the mechanical resonance induced by the
applied field. Secondly, under the resonance condition of
Fig. 6. a) Schematic diagram showing the static charge at the tip of a carbon
nanotube as a result of difference in work functions between the nanotube and
the gold electrode. b) Schematic experimental approach for measuring the
work function at the tip of a carbon nanotube.
ball. Due to the difference in the surface work functions be-

tween the NT and the counter Au electrode, a static charge Q0
exists at the tip of the NT to balance this potential difference
even at zero applied voltage.[48] The magnitude of Q0 is pro-
portional to the difference between the work functions of the
Au electrode and the NT tip (NTT), Q0 = a(WAu ± WNTT),
where a is related to the geometry and distance between the
NT and the electrode.
The measurement relies on the mechanical resonance of the
carbon NT induced by an externally applied oscillating volt-
age with tunable frequency. In this case, a constant voltage,
VDC, and an oscillating voltage, VAC cos 2pft, are applied onto Fig. 7. a) Mechanical resonance of a carbon nanotube induced by an oscillating
the NT, as shown in Figure 6b, where f is the frequency and electric field; b) halting the resonance by meeting the condition of WAu ±
VAC is the amplitude. The total induced charge on the NT is WNTT + eVDC0 = 0. c) A plot of vibration amplitude of a carbon nanotube as a
function of the amplitude of the applied alternating voltage VAC. d) A plot of
vibration amplitude of a carbon nanotube as a function of the applied direct
Q = Q0 + ae(VDC + VAC cos 2pft) (3) current voltage VDC, while the applied frequency is 0.493 MHz and VAC = 5 V.
REVIEW
keeping f = f0 and VAC constant, slowly change the magnitude
of VDC from zero to a value that satisfies WAu ±
WNTT + eVDC 0 = 0 (Fig. 7b); the resonance amplitude A
should be zero although the oscillating voltage is still in effect.
VDC 0 is the x-axis interception in the A ~VDC plot (Fig. 7).
Thus, the tip work function of the NT is WNTT = WAu + eVDC0.
Our data show that most of the nanotube tips have a work
function of 4.6±4.8 eV, lower than the work function of carbon
(WC = 5.0 eV).[47]
The technique demonstrated above, in principle, can be ap-
plied to measure the work function at tips of any nanowires,
conducting, semiconducting, or insulating. We have applied
the technique to measure the work function at the tips of
ZnO semiconducting oxide nanobelts.[49] We found that the
work function is very sensitive to the surface structure at the
tip. A small metallic layer at the tip could significantly reduce
the tip work function.
For multi-walled carbon nanotubes, the bonding between
the shells is van der Waals, thus the inter-wall linkage is rather
weak. Therefore, sliding between the walls is possible. By
opening one end of a nanotube, carbon nanotubes can be used Fig. 8. A) Schematic diagram showing the mechanical stretching of atom-wire
as ultralow-friction nanoscale linear bearings and constant- by a scanning tip in TEM. B) High-resolution TEM images showing the gold
nanowires with widths of a) five, b) four, c) three, d) two, e) one, and f) zero
force nanosprings.[50] The inner walls can be repeatedly, re- (broken) atom columns, created by mechanical stretching. C) Corresponding
versibly pulled out by the electrostatic force and self-sucked measurement of the electrical conductance across the wire as shown in (B) as a
in. Repeated extension and retraction of telescoping nano- function of the stretching time, showing quantum conductance (Images courtesy
of Prof. K. Takayanagi, Tokyo Institute of Technology).
tube segments revealed no wear or fatigue on the atomic scale
as revealed by in-situ TEM.
thickness of one atomic column. The corresponding measure-

2.5. Transport Through a Molecular Nanowire ment of the electrical conductance across the atom-wire while
being stretched is given in Figure 8C. The conductance pla-
Introducing STM into TEM was first demonstrated by teaus measured as the wire thickness is reduced are marked in
Spence and co-workers,[51] aiming at improving the resolution correspondence with the wire thickness indicated in Fig-
for imaging surfaces using reflected electron microscopy.[52] ure 8B, which are given in the unit of Q0 = 2e2/h, the single
This approach has been extensively developed in recent years channel conductance predicted by quantum mechanics. In the
for nanotechnology. Takayanagi and co-workers have applied cases of (a±c), the conductance Q is measured to be 25, 16, 9
the technique to study molecular nanowires in an ultra-high Q0, corresponding to atom columns of 5 5 = 25, 4 4 = 16,
vacuum TEM. They used a gold wire as the tip of the STM, and 3 3 = 9, respectively. For the case of (e), the measured
which is directly in contact with a gold counter electrode. A conductance changed from 4Q0 to 2Q0, which might have re-
slow retraction of the tip results in the formation of Au molec- sulted from the shrinkage of the atomic columns from two to
ular nanowires of a few atoms in thickness. The electric cur- one if they were aligned in the direction parallel to the inci-
rent passing through the nanowire is monitored simultaneous- dent electron beam direction.
ly while the nanowire is stretched, providing one-to-one Ugarte and co-workers[55] have also carried out similar ex-
correspondence between the structure and the conductance of periments using a conventional TEM. They have thoroughly
the nanowire.[53,54] They have observed the structural transfor- analyzed the dynamical structural evolution of Au, Ag, and Pt
mation from a six-prism to a helical multi-shell tubular struc- nanowires due to mechanical stretching. Atomically resolved
ture if the diameter of the gold nanowire is less than 2 nm. images have revealed that just before rupture, nanowires are
The conjunction application of high-resolution TEM and crystalline and free of defects and have only a reduced num-
STM was successfully demonstrated for correlating atomic ber of structures whose formation is determined by surface
structure with nanoscale properties. energy (faceting pattern).[56] The structural information de-
Figure 8A shows a schematic diagram of creating an atom- rived from HRTEM observations was essential to understand
wire by mechanical stretching in TEM. The gold atom-wires the variable conductance behavior of nanowires generated by
with widths of a) five, b) four, c) three, d) two, e) one, and mechanical stretching. These results emphasize the need of
f) zero (broken) atom columns are shown in Figure 8B. These considering an atomic description when analyzing the trans-
images show the first direct observation of atom-wires to port properties of nanoscopic systems.
REVIEW
2.6. In-Situ Environmental Microscopy for Observing Gas±

Liquid-Nanoparticle Interaction
Structural analysis in TEM is always carried out in vacuum,

typically 10±8 torr. It is possible that surface processes occur-
ring in high vacuum are very different from those in gaseous
or wet environments, thus, it is very possible that the dynamic
surface process revealed by TEM may not be the realistic pic-
ture of a practical reaction. It is known that catalytic activities
on heterogeneous catalysts occur entirely at the surface under
ambient conditions. It has been a long time dream to develop
a microscopic technique that can image surface processes and
catalytic behavior on nanoparticles in-situ in a more realistic
gaseous environment. But the technical difficulty is that the
gas molecules near the sample would collide with the incident
electrons, producing a diffuse scattering pattern, and thus the
image resolution would be greatly reduced.
Great progress has been made in recent years in developing
environmental transmission electron microscopy. An environ-
mental cell is built around the objective lens that contains the
specimen in a lower pressure gas environment. The electron
beam enters the cell through a tiny hole, outside of which a
powerful pumping system is designed to maintain the required
vacuum for operating the electron source. Instrumental design
allows imaging, diffraction, spectroscopy, and chemical map-
ping to be carried out in-situ simultaneously using the envi-
ronmental TEM.[57] Gas-catalysis reactions can be directly ob-
served in-situ at sub-nanometer resolution.[58] This technique Fig. 9. In-situ TEM imaging of liquid-catalyst reaction processes and the pro-
duction and growth of the linear polyamide nylon-6,6. The liquids are amidic
has been extended to a wet cell for directly observing liquid- acid and hexamethylene diamine (HMD) and the catalyst is Ru-Co on titania
catalyst reactions at the operating temperature at nanometer (m). The reaction temperature in is 188 C (Image courtesy of Dr. Pratibha
scale.[59] Dynamic liquid hydrogenation and polymerization L. Gai, Du Pont R&D)
reactions in the manufacture of polyamides have been imaged

directly, opening up the possibility of electron microscopy in a 5.0 mbar in a composed gas environment at T = 220 C.[62] The
wet environment. particle shapes can be identified through the lattice images
Figure 9 shows an in-situ liquid-catalyst reaction process observed directly. For face-centered cubic (fcc) structured
and the production and growth of the linear polyamide nylon- metals, the surfaces of the particles can be described by the
6,6.[59] The liquids are amidic acid and hexamethylene dia- Wulff construction with the inclusion of {111}, {100}, and {110}
mine (HMD) and the catalyst is Ru-Co on titania (m). The re- types facets. The proportions of the {100} and {110} are signifi-
action temperature in Figure 9b is 188 C, and the formation cant at 1.5 mbar pressure, while these faces are greatly re-
of polyamide nylon-6,6 is directly observed. This is an out- duced after the pressure increases to 5 mbar. The change in
standing example that shows the true reaction process at oper- the crystal facets surrounding the particles significantly affects
ating temperatures of the catalyst.[60] their catalytic behavior.
Direct observation of the location, state, and function of a
promoter in heterogeneous catalysis is critical in understand-
ing the surface processes in catalysis. Hansen et al.[61] has re- 3. Quantitative Chemical Imaging
cently achieved atomic resolution (0.14 nm) in-situ imaging of at Sub-Nanometer Resolution
nanoparticles exposed to reaction gases at elevated tempera-
tures. In the most active barium-promoted ruthenium catalyst Although high-resolution TEM imaging is powerful for di-
for ammonia synthesis, through in-situ TEM observation at rectly observing atomic structures of nanomaterials, the image
552 C and 5.2 mbar in a gas composition of H2/N2 of 3:1, they contrast is not very sensitive differences in atomic number,
suggest that the increased activity is related to a two-dimen- and in cases of light elements, such as oxygen, the weak scat-
sional barium-oxygen over-layer on the Ru crystals. A dy- tering of the atom results in vanishing contrast. To quantita-
namic shape change of the supported copper nanocrystals has tively determine the structure of nanomaterials, chemical
been revealed at atomic resolution.[62] analysis at spatial resolutions of sub-nanometers or approach-
Figure 10 shows the high-resolution TEM image Cu nano- ing angstrom is critical in solving many material problems.
crystals supported on ZnO substrate at pressures of 1.5 and There are two techniques for chemical analysis in TEM: EDS
REVIEW
Fig. 10. In situ TEM images (A,C,E) of a
Cu/ZnO catalyst in various gas environ-
ments together with the corresponding
Wulff constructions of the Cu nanocrys-
tals (B,D,F). A) The image was recorded
at a pressure of 1.5 mbar of H2 at 220 C.
The electron beam is parallel to the [011]
zone-axis of copper. C) Obtained in a gas
mixture of H2 and H2O, H2/H2O = 3:1 at a
total pressure of 1.5 mbar at 220 C.
E) Obtained in a gas mixture of H2
(95 %) and CO (5 %) at a total pressure
of 5 mbar at 220 C (Image courtesy of
Dr. P. L. Hansen et al., with permission
from AAAS).
and EELS. Due to the small thickness of samples used for

TEM, the spatial resolution of the chemical analysis is ap-
proximately determined by the size of the electron probe,
which can approach 0.1 nm. The thickness-projected elemen-
tal maps can be obtained by collecting the characteristic sig-
nals while the electron probe is scanned across a sample in a
scanning transmission electron microscope (STEM). This is a
well-established technique for chemical mapping.
There are two important developments in microanalysis in
the last few years. Using the newly corrected electron optical Fig. 11. Composite nanotube with the C-BN-C sandwich structure. A) TEM im-
age showing the graphitic-like structure. B) Elemental profiles measured across
system (see Sec. 5) and the high-coherence, high-brightness
a tube similar to that shown in (A); the arrows indicate the anticorrelation be-
field-emission electron source, an electron probe with a diam- tween C and B (or N). C) Simulated profiles for C and B (or N) distributions
eter as small as ~ 0.1 nm has been achieved. Such a fine elec- for the coaxial heterostructure shown in (D) (Courtesy of Dr. K. Suenaga and
Dr. C. Colliex, University of Paris-Sud).
tron probe is ideal for probing the chemical and electronic
structure of a sample, atomic column by atomic column, pro-
viding the most powerful analytical technique for nanotech- pure BN tubes, all tubes were found to have well-separated
nology research. The other key advance is the improvement BN layers and carbon layers. From the concentration profiles
of energy resolution in TEM/STEM, so that electronic struc- taken across such a tube, the presence in the carbon profile of
ture mapping can be performed. We use the following exam- four maxima and two minima coinciding with the two maxima
ples to illustrate the advances in these two major areas. of the B and N profiles clearly indicates an alternating C-rich-
The analysis of multi-walled C-BN nanotubes is an exam- BN-rich-C-rich geometry (Fig. 11a). These profiles were simu-
ple. Tubular structures can be formed by either a rolling gra- lated for various distributions of layers with different chemi-
phitic layer or a BN layer. Arc-discharging a mixture of car- cal compositions (Fig. 11b). The best agreement with experi-
bon and BN produces composite C-BN nanotubes. The ment was obtained in this case for a 14-layer tube, 12 nm in
structures of the C-BN nanotubes cannot be distinguished external diameter, formed with three inner layers of graphitic
from carbon nanotubes using high-resolution TEM because carbon covered by six (BN)xC2y layers (with y/x ~ 0.05 ± 0.05)
there is no significant difference of the scattering from B or N and five outer carbon layers. Chemical mapping using EELS
compared to that of carbon. Chemical mapping using a fine has also been applied to directly imaging the presence of a
electron probe is unique for this type of study. By collecting single Gd atom trapped inside C60 molecules (the peapod
the EELS spectra of the transmitted electron beam, while the structure).[64] The data give unambiguous proof of the pro-
electron beam is scanned across the diameter of the tube, a posed model.
group of thickness-projected chemical compositions of the To illustrate the application of high-spatial resolution
sample is recorded (Fig. 11).[63] The layers of the tubes have EELS, we use the Cu±Al2O3 interface to be presented in Fig-
various chemical compositions. Besides the typical case of ures 20 and 21 as an example. Figure 12 shows the Cu L-edge
REVIEW
EELS spectra acquired from the Cu film and from the region
containing the Cu±Al2O3 interface using an electron probe of
~ 0.5 nm in size.[65] The difference spectrum is received by
subtracting the EELS spectrum acquired from the Cu side
from that from the interface region with a proper scaling fac-
tor. The interfacial Cu L-edge shows a chemical shift of
Fig. 13. EELS spectrum acquired from CoO, showing the O-K and Co-L ioniza-
tion edges.
onstrated to map the valence state distributions of Co (or Mn)

using energy-filtered electron microscopy.[70,71]
An energy filtered image can display solid state effects. Fig-
ure 14 shows a group of energy-filtered TEM images from a
triple point in a CoO±Co3O4 two-phase specimen. Chemical
images from the O K- and Co L-edges directly reflect the
thickness-projected distribution of oxygen and cobalt, respec-
tively (Figs. 14a,b). The Co L3/L2 white line ratio image was
calculated based on the experimental images recorded using
the L3 and L2 edges, after subtracting the contribution from
the continuous energy-loss region that is due to single atom
scattering; the image clearly displays the distribution of cobalt
valence states (Fig. 14c). The region with lower oxidation
state (Co2+) shows stronger contrast, and the ones with high
oxidation states show weaker contrast. The thickness-pro-
jected atomic ratio O/Co image was calculated from the im-
Fig. 12. EELS spectra acquired from the Cu±a-Al2O3 interface as presented in
Figure 20, a) from the Cu film, b) from the interface region, and c) the differ-
ence spectrum. d±f) Standard spectra acquired from bulk Cu2O, CuO, and
CuAl2, respectively (Spectra courtesy of Dr. C. Scheu and Prof. M. Rühle,
Max-Planck Institute für Metallforschung, Stuttgart).
~ 1.8 eV to higher energy-loss and a prominent peak at the

edge onset. The standard spectra from Cu2O, CuO, and CuAl2
are also shown for comparison. The difference spectrum
closely resembles the spectrum of CuAl2, suggesting that the
Cu atoms at the interface have a similar electronic structure
and environment as in CuAl2.
Using EELS, we have illustrated the sensitivity of near edge
fine structures to the valence states of transition metal ele-
ments. Energy-filtered imaging by selecting electrons with an
energy-loss in correspondence to the near edge structure con-
tains the spatial distribution of the valence states across the
region. A typical example is the mapping of sp2 and sp3 bonds
in diamond-like materials.[66] Figure 13 shows an EELS spec-
trum acquired from CoO that clearly indicates the presence of
fine structures in the oxygen K-edge, and the split white lines Fig. 14. A group of images recorded from the same specimen region using sig-
nals from a) the O K-edge, b) the Co L-edges, c) the processed L3/L2 image,
in the Co L-edge. The relative intensities of the Co-L3 and displaying the distribution of the valence states. d) The atomic concentration
Co-L2 white lines strongly depend on the valence state of co- ratio image of O/Co. The continuous background contributed from the single
balt. The ratio L3/L2 image gives the distribution of valence atom scattering has been removed from the displayed Co-L3 and Co-L2 images.
The O/Co image is normalized with reference to the standard composition of
states across the specimen.[67±69] Using the images acquired CoO for the lower portion of the image in order to eliminate the strong influ-
from the white lines, an experimental approach has been dem- ence on the ionization cross-section from the white lines.
REVIEW
ages recorded from the O K-edge and the L3 + L2 white lines in Figure 15a, the crystal appears to be a single piece without
(Fig. 14d). Although the energy-filtered O K-edge image ex- typical characters. Energy-filtered images using the Mn L2-
hibits some diffraction contrast and thickness effect, the O/Co and Mn L3-lines (Figs. 15b,c) show distinct differences in in-
compositional ratio image greatly reduces this effect. The high tensity distribution. After subtracting the image formed by
intensity region in the O/Co image indicates higher local con- the continuous energy-loss past the L2 white line, a ratio im-
centrations of oxygen (e.g., higher Co oxidation states), the age of L3/L2 is attained (Fig. 15d). A line scan across the im-
low intensity region contains comparatively less oxygen (e.g., age with an average of 50 pixels in width is given below the
lower Co valence state), entirely consistent with the informa- L3/L2 image, in which the distributions of Mn4+, Mn2+, and
tion provided by the L3/L2 image. A spatial resolution of Mn3+ are unambiguously displayed. The reduction of oxides
~ 2 nm has been achieved. This is remarkable in comparison first occurs at the surface region because of the easy desorp-
to any existing techniques. tion of the oxygen, while the core of the crystal is still MnO2.
The chemical imaging is sensitive to the transition in va- This type of imaging is powerful in analysis of Co- and Mn-
lence states. In reference to our studies of thermal induced based oxides.
transition metal oxides,[72] a reduction of MnO2 was carried Using the fine electron probe, valence excitation in multi-
out in-situ in vacuum at a temperature of 350 C, and the re- walled and single-walled carbon nanotubes have been stud-
sulting reduced phases were a mixture of oxides of Mn with ied.[73,74] The electron probe can be focused at specific posi-
valences of 2+, 3+, and 4+. Figure 15 shows a series of images tion in reference to the center of the object; the dielectric ex-
acquired by selecting the different energy-loss signals in citation in a small region of a few nanometers can be
EELS. Looking at the conventional bright-field image given investigated. Theoretical calculation based on quasi-classical
electron energy loss theory gives a good interpretation of the
observed data.
4. Quantitative Microscopy and Electron

Diffraction
4.1. Dark-Field Imaging of Small Clusters and Point Defects

in STEM
Imaging in the TEM is mainly dominated by elastic scatter-

ing, and diffraction effects are highly pronounced if the im-
ages are formed by Bragg reflected beams. Bright-field imag-
ing is dominated by diffraction and phase contrast, and is not
very sensitive to differences in atomic number. A monolayer
HgS grown on the surface of a CdS nanocrystal, for instance,
cannot be easily imaged by HRTEM. To reduce the diffrac-
tion effect and enhance chemical sensitivity, a dark-field imag-
ing technique has been developed. We now examine the scat-
tering behavior of a single atom, whose scattering factor is
given by the Mott formula
fe(s) = e[Z ± fx(s)]/16p2e0s2, (5)
where fx is the X-ray scattering factor of the atom, Z its atom-

ic number, and s a variable related to the scattering angle by
s = sinh/k, where k is the electron wavelength. Taking the case
of large angle scattering with large s, fe(s) » eZ/16p2e0s2, is di-
rectly proportional to the atomic number. This means that
atomic number information can be revealed by collecting the
signals scattered to high angles. This is done in scanning trans-
mission electron microscopy (STEM) using a high-angle an-
Fig. 15. A group of energy-filtered images acquired from the same specimen re-
nular dark-field (HAADF) detector to detect the electrons
gion of mixed phases of MnO2 and Mn3O4. a) The conventional bright-field
TEM image, b,c) the energy-filtered TEM images of Mn-L2 and Mn-L3 white scattered to high angles. This technique was first demonstrat-
lines, respectively, d) the calculated Mn L3/L2 ratio image, and e) the normal- ed by Crewe for direct imaging of a single gold atom.[75] Fig-
ized (L3 + L2)/post-line image. A pixel averaged line scan of the L3/L2 ratio
from (d) is also given, from which the distribution of Mn4+, Mn3+, and Mn2+ are
ure 16 shows the scattering geometry of a fine electron probe
identified. across a specimen in STEM. An HAADF detector is placed
REVIEW
In many cases, it is believed that the surfaces of small parti-

cles are the most active sites for catalysis. In practice, besides
the small particles, single atoms may be deposited on the sub-
strate surface, which may also be active sites for catalysis.
However, the classical technique was unable to observe single
atoms due to the strong effect from the substrate. Using
HAADF STEM, individual single atoms can be imaged un-
ambiguously. Shown in Figure 17 is a HAADF STEM image
of single La atoms on a c-alumina support recorded at 300 kV
with a Cs-corrected electron probe. This is an important ad-
vance for studying nanostructured catalysts. A histogram of
the full width at half maximum (FWHM) of intensity profiles
Fig. 16. Scattering geometry in STEM and the corresponding image formation
process.
at the diffraction plane that detects the electrons scattered to

high angles.
For crystalline specimens, phonon scattering (thermal dif-
fuse scattering, or TDS), which was first proposed by Wang
and Cowley[76] as the dominant contributor for HAADF
STEM imaging, is the main contributor to scattering to high
angles. Details about thermal diffuse scattering can be found
Fig. 17. HAADF STEM image (Z-contrast image) recorded at 300 kV from
elsewhere.[12] The intensity contributed by a single atom at single La atoms supported on c-alumina substrate, showing greatly enhanced
high scattering angle is contrast with the improvement in electron probe size (Image courtesy of
Dr. S. J. Pennycook, Oak Ridge National Laboratory).
ITDS(s) = [fe(s)]2 [1 ± exp(± 2Ws2)] (6)

across single La atoms gives the probe FWHM as about 0.7±
where W is the Debye-Waller factor that is sensitive to speci- 0.8 . The smaller probe also gives a greatly improved image
men temperature. ITDS(s) is proportional to Z2 at high scatter- of the alumina itself. Very faint lattice fringes can be identi-
ing angle. Since the vibration of the nuclei is responsible for fied in the image.
TDS, phonon scattering is a very localized scattering process. Imaging of line and planar defects is the most conventional
The resolution of HAADF STEM is approximately deter- application of TEM. Analysis of point defects, however, is
mined by the size of the electron probe. probably the most challenging task. For points distributed
TEM is a powerful technique for imaging the projected par- with short-range ordering, electron diffraction can be applied
ticle shapes and surface atom arrangement through profile im- to determine the correlation among point defects. If the point
aging.[77] Imaging of surface adsorbed atoms on a nanoparti- defects are distributed randomly, TEM can hardly provide
cle, however, can never be an easy task. Due to the rotation of any useful information about point defects because of the dif-
the particle under the scanning tip, direct resolution of the ad- ficulty in isolating the information from the defect site from
sorbed atoms by scanning tunneling microscopy is in practice the huge contribution of the diffraction of the main crystal as
difficult to achieve. Structure analysis of heterocatalysts is a well as the surface atoms. Recently, Voyles et al. have demon-
challenge because of strong effects of the substrate. The strated direct imaging of individual Sb dopant atoms in crys-
strong scattering from the substrate greatly reduces the signal talline Si lattice using HAADF STEM.[79] Doping is a conven-
from the supported catalyst particles. Conventional high-reso- tional processing technique in semiconductor industry. The
lution TEM is dominated by the scattering from the substrate carrier introduced by dopants is usually described based on
and is insensitive to the adsorbed atom, especially lighter ele- statistics because the doping level is typically a few parts per
ments. Using HAADF-STEM, the signal from the catalyst million (ppm). Low doping in nanocrystals could be different
particles can be enhanced by the stronger scattering power of because each particle, on average, may have only one or even
the adsorbed atoms. This study has been demonstrated for less than one doping atom. Thus, the question is whether the
platinum and rhodium atoms dispersed on c-Al2O3 sup- dopants are located inside the particle or on the surface of the
ports.[78] The configuration of the metal atom cluster could be particle? The HAADF-STEM image may provide a possible
constructed from the image and a model for surface adsorp- technique for studying dopants in a single particle if the dop-
tion could be proposed. ants are heavier atoms.
REVIEW
4.2. Holography for Imaging Magnetic Nanostructures
Imaging of magnetic domains and interdomain interactions

is vitally important for understanding the behavior, improving
the properties, and device designing of magnetic materials.
The most frequently used techniques for imaging domain
structures are Fresnel contrast and Lorentz microscopy (see
literature[80] for recent progress) and off-axis electron holo-
graphy.[81] A comprehensive review of the imaging techniques
for magnetic structures can be found elsewhere.[82] This sec-
tion mainly focuses on the introduction of off-axis electron
holography for imaging magnetic structures. The mechanism
of imaging magnetic domains in magnetic materials relies on
the Aharonov±Bohm effect, which describes the relative
phase shift of two electron waves traveling from a point along
two routes enclosing a zone with magnetic field distribution to
the same point,
2pe R
Du B ´ dS (7)
h
where dS is a surface integral over the surface area enclosed

by the two routes. A key step in observation of magnetic flux
Fig. 18. a) Formation of electron holograms in TEM using an electron biprism;
lines is to create a magnetic field-free zone in which the speci- b) reconstructing the amplitude and phase information from the electron holo-
men to be examine is placed. This can be achieved using a gram.
Lorentz objective lens. Electron holography can be applied to
image magnetization in magnetic particles, flux lines in super- aberration effect and the defocus introduced by the objective
conductors, and magnetic domains in thin films.[83] lens, thus, the true phase of the electron wave is retrieved.
The principle of electron holography was first introduced Figure 19 shows a phase image of a Co nanoparticle of size
by Gabor[84] in an attempt to exceed the point-to-point resolu- ~ 200 nm, where the contour lines in the interference micro-
tion of an electron microscope. Holography is based on the in- graph directly indicate the magnetic lines of force in h/2e
terference and diffraction properties of waves, thereby pro-
ducing a true image of an object (including amplitude and
phase) without any distortion of the lenses. The development
of high-brightness, high-coherence electron sources has made
it possible to obtain holograms using electron waves in
TEM.[85±87]
Electron holography is designed as a two-step imaging tech-
nique. The first step is to form an interference hologram as
shown in Figure 18a. The other step is to reconstruct the holo-
gram to get the real object image. A high-brightness, high co-
herence electron beam is generated by a field emission source
with an energy-spread smaller than 0.5 eV. The specimen is Fig. 19. Electron phase map reconstructed from a holographic image of a plate
positioned to cover half of the image plane, leaving the other Co particle (Courtesy of Dr. A. Tonomura, Hitachi).
half for the reference wave. The off-axis holograms are re-
corded by means of an electron biprism inserted between the units.[86] The magnetization direction is clockwise or counter-
back focal plane of the objective lens and the intermediate clockwise corresponding to the wave-front protruding like a
image plane. By applying a positive voltage to the filament of mountain or a valley. Electron holography is a powerful tech-
the biprism, the waves on both sides of the biprism are de- nique for quantitative analysis of magnetic domain structures.
flected towards each other, forming an interference pattern in
the image plane, i.e., the hologram. The two waves passing by
the biprism are analogous to the waves emitted by the double 4.3. Quantitative Electron Diffraction for Refining Atomic
slits in the Young's interference experiments in optics. In the Structure and Charge Density Distribution
second step, as shown in Figure 18b, the goal is to retrieve the
amplitude and phase information contained in the hologram. Analogous to X-ray diffraction, refinement of crystal struc-
This procedure can be done optically using an interferometer, ture using electron diffraction has been a goal for scientists. It
or by numerical calculation. This step removes the spherical is known that HRTEM imaging is limited and distorted by
REVIEW
aberrations of the objective lenses, but electron diffraction is the atomic structure at the interface is quantitatively refined.
aberration free. Quantitative analysis of electron diffraction Figure 21a reveals a simulated HRTEM image of the inter-
data is rather challenging due to strong electron±specimen in- face using an Al-terminated Al2O3 model, and the result is in-
teraction that results in dynamic diffraction effects, which has serted in the figure. It was noticed that the calculated image
to be quantified by solving the Schrödinger equation. On the gives a smaller value of 0.08 ± 0.02 nm for the projected Cu±
other hand, by taking advantage of the strong electron±atom Al distance at the interface, and the visible contrast at the in-
interaction for high-energy electrons, the diffraction intensity terface is not reproduced in the calculated image. Therefore,
is sensitive to charge redistribution in crystalline specimens. the positions of the Al atoms have to be relaxed at the interfa-
Thus, a quantitative fitting of the experimental data with the cial region, and the model that gives the best fit to the experi-
calculated data produces the charge transfer and bonding mental image is given in Figure 21b. This is an excellent ex-
maps in functional oxides.[88,89] This study is feasible for per- ample of interface structure refinement using HRTEM.[91]
fectly crystalline specimens without defects and surface ef-
fects, but its applications for nanomaterials has not been dem- (a)
onstrated as of now.
Surface/interface structure is likely to be the most pro-
nounced structural character of nanomaterials. Studying
bonding in inorganic solids is important in understanding
many of their properties. Our common knowledge is that met-
als have metallic bonds and ceramics have covalent and/or
ionic bonds. At an interface between metal and ceramics, such
as in Nb±Al2O3, what is the bonding type? This type of study
has been systematically carried out by Rühle's group.[90] They
have adopted a first principle calculation of the structure of
the interface, which is then used for image calculations based
on the well established imaging theory. By systematic compar-
ison between the super quality experimental HRTEM images (b)
with the theoretically calculated results, the theoretical model
is refined, and thus the interface structure is revealed.
Figure 20 shows a HRTEM image recorded from the inter-
face between Cu and a-Al2O3.[65] The orientation relationship
Fig. 21. A quantitative comparison of the high-resolution TEM images re-

corded from the interface between Cu and a-Al2O3 and the calculated images
(inserted) based on the model shown on the right-hand side (Image courtesy of
Dr. C. Scheu and Prof. M. Rühle, Max-Planck Institute für Metallforschung,
Stuttgart).
5. Aberration-Corrected High Resolution Imaging

Fig. 20. High-resolution TEM image recorded from the interface between Cu
and a-Al2O3. The incident beam is [112] Cu and the image was recorded using Due to the imperfection of magnetic lenses, the resolution
a 1250 kV atomic resolution TEM with a point-to-point resolution of 0.12 nm of the electron microscope is limited by spherical aberration,
(Image courtesy of Dr. C. Scheu and Prof. M. Rühle, Max-Planck Institute für an effect originating from the dependence of the electron fo-
Metallforschung, Stuttgart).
cal length on its scattering angle. There are two typical defini-
between the film and substrate is: (112Å0)Al2O3 [0001]Al2O3 i tions of resolutions in TEM. The Scherzer resolution or the
(111)Cu [112Å]Cu No strain field indicative of misfit dislocations structural image resolution refers to a resolution at which the
was observed, the interface is atomically abrupt, and there is images can be directly related to the projected structure of
no reaction layer. Quantitative interpretation of the image the object, which is given by
and building the interface atomic structure model depends on
image simulation. Based on the well-established dynamic Rs = (3/4)1/2 k3/4 Cs1/4 (8)
electron scattering theory,[91] the well accepted computer soft-
ware,[92] and iterative matching between the experimental im- where k is the electron wavelength, and Cs is the spherical
age and the calculated image based on the proposed model, aberration coefficient, which is a key quantity in determining
REVIEW
the structural resolution that can be achieved by a TEM. The (see Ch. 4 in Wang and Kang[7]). Direct imaging of light ele-
other resolution is the information limit, which means the fi- ments by electron microscopy has been demonstrated by the
ner details presented in the image that might exceed the Cs-corrected HRTEM very recently.[97] Figure 22A shows an
Scherzer resolution but may not have a direct relationship experimental image recorded from a chemically non-stoichio-
with the structure of the sample. To enhance the structure im- metric SrTiO3±x that has oxygen vacancies. Based on the mod-
age resolution as stated in Equation 8, the classical technique el inserted in Figure 22A, the projected intensity at the oxy-
is to reduce the wavelength of the electrons, which means a gen positions should be equal, but the image, as illustrated by
great increase in the acceleration voltage of the microscope to the line scan across atom columns 1 and 2, indicate lower in-
a range of over one million volts.[93] The consequence of rais-
ing acceleration voltage is not only a gigantic increase in cost
of manufacturing, operation, and maintenance, but also en-
hanced radiation damage due to the high energy electrons.
An alternative approach is to reduce the spherical aberration
coefficient by introducing an aberration corrector around the
objective lens, a so-called Cs-corrector. The advantage of
using a Cs-corrector is that a sub-angstrom resolution can be
achieved while the microscope is operated at lower voltage
such as 200 kV. This offers a great advantage for materials
analysis.
Correcting the aberration and extending the image resolu-
tion have been a constant effort for scientists. The most classi-
cal technique is electron holography, as demonstrated by
Lichte's group.[94] Electron holography gives the possibility of
retrieving electron wave functions at the image plane, which
includes the amplitude and phase images. If the experimental
parameters such as objective lens defocus Df and Cs are pre- Fig. 22. A) Experimental image of SrTiO3 oriented along [011] using a Cs-cor-
cisely known, the aberration introduced by the objective lens rected high-resolution TEM. The spherical aberration coefficient was adjusted
can be eliminated from the electron wave, thus the true wave to Cs = ±40 lm, and the overfocus Df = 8 nm. The atomic columns in the pro-
jected direction shows bright contrast, and the oxygen atoms are clearly re-
function at the exit face of the specimen is received. solved. The intensities in columns 1 and 2 are lower than the neighboring oxy-
In hardware, great and fast progress has been made in the gen columns due to the presence of oxygen deficiency. B) Simulated images
based on a model for introducing 85 % and 80 % occupancies in the columns 3
last few years on Cs-corrected TEM, a resolution of better and 4, respectively, for oxygen occupancy, shows excellent agreement with the
than 0.1 nm has been achieved at 200 kV, while such a resolu- experimental image as indicated by the line trace (Image courtesy of Prof. K.
tion requires 1250 kV without the use of a Cs-corrector. There Urban and Dr. C. L. Jia, Institute of Solid State Research, Jülich).
are several approaches having been developed for correcting
the Cs. Urban and co-workers[95] have used a hexapole correc- tensity in comparison to the neighbor oxygen columns. Quan-
tor system, which improved the point resolution for a 200 kV titative image calculation by introducing oxygen deficiency
TEM from 0.24 nm to 0.14 nm. Krivanek et al.[96] developed a for 85 % and 80 % oxygen occupancy in columns 1 and 2, re-
quadrupole±octupole corrector for improving the probe size spectively, matches the observed image. This is a beautiful ex-
of a scanning transmission electron microscope. ample demonstrating direct imaging of ordered oxygen defi-
An improvement in resolution provides following advan- ciencies in oxides.[97] It is known that the properties of
tages for nanotechnology. Firstly, high quality experimental function oxides are determined by the oxygen deficiency.
data is provided for quantitative structure analysis. Secondly, ZnO, for example, can be tuned from semiconductor to con-
light elements, such as oxygen[97] in functional oxides and ductor simply by introducing oxygen deficiencies into the lat-
even lithium[98] in battery materials, can be directly resolved tice. The entire series of Ce-, Pr-, and Tb-based oxides have a
by TEM. Thirdly, it gives the possibility of increasing the gap wide range of stable homologous phases when ordered oxy-
between the pole pieces of the objective lens, thus scanning gen vacancies are introduced.[7] Quantitative determination of
probe techniques and in-situ techniques can be integrated light elements using electron based techniques is a key prog-
with TEM, and more innovative measurement tools can be ress in nanomaterials analysis.
used in conjunction with TEM. Finally, a fine and bright elec- Among the list for light elements in solid materials, hydro-
tron probe can be formed for atomic resolution spectroscopy gen, lithium, beryllium, boron, carbon, nitrogen, and oxygen
and chemical analysis (see Sec. 6). are the most important members. Recent experiments by
Structure determination for light elements is usually per- Shao-Horn et al.[98] have shown the direct imaging of lithium
formed using neutron diffraction because of the significantly atoms in battery materials. Figure 23 shows the HRTEM im-
larger scattering cross-section. For function oxides that have age recorded from LiCoO2, where the projected positions of
an ordered distribution of oxygen vacancies, electron diffrac- the oxygen, cobalt, and lithium atoms are indicated. It is ap-
tion can be applied to determine the structure of the crystal parent that Li atoms are clearly resolved for the first time by
REVIEW
rector for improving the probe size of a 100 kV STEM to

0.1 nm,[101] which has been applied to image gold nanocrystals
at a resolution of 0.1 nm.[102]
Cowley has proposed a new approach that can achieve a
probe size of 0.05 nm without using the Cs corrector.[103] The
approach is based on the atom lens, in which a heavy atom,
placed in front of an electron probe in STEM, could focus the
electron probe to a size of ~ 0.05 nm. Experiments on carbon
shell have demonstrated the feasibility of this proposal.[104,105]
Spectrum-imaging of single atoms in STEM must overcome
several hurdles. The S/N ratio, which is affected by the ioniza-
tion cross-section of the element to be detected and the thick-
Fig. 23. Sub-angstrom image of LiCoO2 battery material from Shao-Horn et al.
[98] shows all atom species. As indicated by the superimposed model, the strong
ness of the substrate, which is usually carbon. The beam dose
white peaks occur at the positions of oxygen atom columns, the strong fuzzy is also important. Radiation damage, particularly to light ele-
peaks at cobalt sites, and the weak white peaks show lithium positions. The ments and biological specimens, is a main concern, which is
authors confirmed the identities of the peaks with image simulation. (Reprinted
with permission; copyright Nature Publishing Group 2003). likely to induce structural degradation, atom diffusion, and
limited data acquisition time. Specimen drift must be mini-
electron imaging. This exciting advance will no doubt be a mized to extend the data acquisition time. It is now possible
great boost in the research and development of nanoscale en- to digitally correct the drifting effect. As the size of the elec-
ergy storage materials. tron probe is approaching 0.1 nm, a single light atom can be
Aberration correction can also be carried out by image pro- identified. Leapman and Rizzo[106] have successfully imaged
cessing, which has been systematically developed by Van two phosphorus atoms using EELS. Single atom imaging has
Dyck's group.[99] An important variable that can be changed recently been demonstrated by Suenaga et al.[107] who imaged
in recording electron images is the objective lens focus. If a fo- a Gd atom trapped in a C60 molecule. The C60 molecules are
cus series of HRTEM images is recorded from the same speci- aligned as a chain inside a single-walled carbon nanotube, and
men area, the nonlinear information transfer function of the each C60 molecule is the cage for storing one Gd atom.
optical system can be corrected and the complex wave func- Using a fine electron probe, local electronic structure can
tion at the exit face of the specimen can be retrieved by image be examined with atomic resolution. With the improvement
processing. This process requires high stability of the micro- in instrumentation, it is anticipated that EELS can be ac-
scope and image drift correction. A large improvement in im- quired at an energy-resolution of better than 0.1 eV at a spa-
age resolution has been achieved. tial resolution better than 0.1±0.2 nm. Such a dramatic perfor-
mance of the future EELS is likely to significantly impact the
nanoscale chemical and electronic analysis in the following as-
6. Sub-Angstrom Electron Probe and Single Atom pects. Firstly, with an ultra small electron probe, chemical
Detection analysis and local electronic structure can be quantitatively
detected from a single atomic column. This provides a novel
A unique advantage of chemical imaging in STEM is that technique for probing the solid state structure, column-by-col-
the spatial resolution is solely determined by the size of the umn, allowing a direct probe of the local density of states at a
electron probe for very thin specimens. In STEM, the elec- defect site, the atoms at interfaces and at grain boundaries.
tron-specimen interaction occurs after the formation of the This is vitally important for understanding the impurity states
electron probe by the condenser lens, thus, the chromatic introduced by defects and the local bonding transformation at
aberration effect arising from inelastic interaction is elimi- interfaces. Secondly, chemical analysis with atomic resolution
nated, a major advantage over TEM. Therefore, the broaden- provides an effective approach for ªatom-by-atomº nanoengi-
ing of the electron probe as a result of electron energy-loss is neering. Finally, chemical imaging towards atomic resolution
solely determined by angular broadening introduced by the has a vast impact for solving many problems in materials
intrinsic physics, the enlargement from the lens is diminished. science. It could provide a powerful approach for identifying
By considering the high localization of the inner shell ioniza- point defects, grain boundary segregation, interface reaction
tions of atoms, it is possible to achieve atomic resolution and diffusion, and impurities at defect sites, etc.
chemical analysis and imaging in STEM.[100]
For a Cs-limited STEM, the optimum electron probe size is
given by 7. Sub-eV Resolution Spectroscopy
dp = 0.4 k3/4 Cs1/4 (9) The energy resolution in TEM is in the order of 0.5±1 eV.
To be adequate for resolving solid state electron structure, a
A correction in Cs means a great improvement in probe size. resolution of better than 0.1 eV is required. The energy
Krivanek et al.[96] have developed a quadrupole±octupole cor- spread of a field emission source is 0.5±0.7 eV. A major effort
REVIEW
in microscopy now is to develop a monochromator that re- tures can be fabricated by controlling the scan of the electron
fines the energy spread of the electron beam that transmits probe with the presence of the chemical vapor.
the specimen to an order of 0.1 eV. With the proposed energy
resolutions one has the potential to examine local mechanical,
thermal, and acoustic processes. Phonon excitation in the bulk 8. Summary and Perspectives
and at the surface is an outstanding example. If it would be
possible to observe phonon excitations at a spatial resolution This paper reviews some of the most recent progress in elec-
at the atomic level, the structural stability of atoms at inter- tron microscopy techniques for nanotechnology. Based on ex-
faces and defects could be detected. For instance, in materials isting experiments, the review focuses on novel approaches
that undergo certain structural phase transitions, the onset of that have been demonstrated for solving problems in nanoma-
the transition is signaled by a softening of the structure. This terials. The review covered the following areas: in-situ micros-
softening is detectable by observation of phonon modes. copy, in-situ nanoscale property measurements, environmen-
tal microscopy, angstrom-resolution imaging, high-angle
annular-dark-field STEM imaging, electron holography, and
7. Creation of Nanoscale Structures by a Fine high-spatial resolution EELS. The reviewer believes that
Electron Probe these areas will be the future directions of electron microsco-
py for nanotechnology.
Nanoscale patterns can be fabricated by an electron beam. Understanding atomic structure, detecting local chemistry,
Radiation damage by an electron beam is dominated by three measuring local electronic structure, and probing local proper-
processes: breaking bonds by knock-on collision with the ties are essential in ªatom-by-atomº engineering for nano-
atoms, ionization of inner shell electrons, and local heating. It technology. Picometer-scale imaging provided by TEM is
is known that some of the materials are sensitive to electron essential for nanometer-scale technology. It is apparent that
beam illumination, and their structures change under the elec- atom-by-atom engineering is the state of the art for na-
tron beam. Using this characteristic, a fine electron probe in noscience, but it may not be practical for nanotechnology to
STEM can be useful for creating nanoscale patterns. It was re- meet the need of mass production with extremely high repro-
ported by Humphreys et al.[108] that an electron beam in ducibility. We are convinced that self-assembly will be the fun-
STEM can drill a nanometer size hole through a MgO cube. damental of nanotechnology, just like the self-assembly pro-
a-Al2O3 can be reduced into metallic Al by a strong electron cess in biosystems. Self-assembly will be the bridge that links
beam.[109] The mechanism of this radiation damage is attribut- the ªbottom±upº with the ªtop±downº approaches for nano-
ed to an internal Auger decay process. Figure 24 shows an im- technology. Understanding the column-by-column atomic
age of an AlF3 cube prior and after scanning an electron beam structure, properties and manipulation are the first steps for
across its surface, resulting in the formation of a metal wire, as controlling the self-assembly process of future nanomaterials
indicated.[110] The metal nanowire is metallic Al because AlF3 and processing. Electron microscopy will be indispensable and
crucial for bringing the blueprint of nanotechnology to reality.
Received: March 18, 2003
Final version: June 6, 2003
±
[1] Research Horizon (Ed: J. M. Sanders), Georgia Tech, Atlanta, GA 2001,
p. 5.
[2] Z. L. Wang, Y. Liu, Z. Zhang, Handbook of Nanophase and Nanostruc-
tured Materials, Vols. I±IV, co-published by Kluwer Academic Press, New
York, and Tsinghua University Press, 2002.
[3] P. Buseck, J. M. Cowley, L. Eyring, High Resolution Transmission Elec-
tron Microscopy: Theory and Applications, Oxford University Press, New
York 1989.
[4] G. G. Tibbetts, J. Cryst. Growth 1984, 66, 632.
[5] S. Iijima, Nature 1991, 354, 56.
[6] Z. L. Wang, C. Hui, Electron Microscopy of Nanotubes, Kluwer Academic
Fig. 24. Scanning electron microscopy images of AlF3 cube synthesized by a sol-
Publisher, New York 2003.
id-vapor process, a) before and b) after scanning an electron beam slowly along
[7] Z. L. Wang, Z. C. Kang, Functional and Smart MaterialsÐStructural Evo-
the line indicated, showing the formation of wire in the sample.
lution and Structure Analysis, Ch. 6±8, Plenum Press, New York 1998.
[8] Z. L. Wang, J. Phys. Chem. B 2000, 104, 1153.
can be easily reduced by the electron beam. The width of the [9] Z. L. Wang, Adv. Mater. 1998, 10, 13.
[10] Z. L. Wang, Z. R. Dai, S. Sun, Adv. Mater. 2000, 12, 1944.
Al nanowire could be controlled by choosing the size of the [11] D. B. Williams, C. B. Carter, Transmission Electron Microscopy, Plenum
electron beam, the acceleration voltage and the beam radia- Press, New York 1996.
[12] Z. L. Wang, Elastic and Inelastic Scattering in Electron Diffraction and
tion time. Imaging, Plenum Press, New York 1995.
It is known that some precursors can be dissociated by an [13] L. Reimer, Transmission Electron Microscopy, 3rd ed., Springer-Verlag,
electron beam even at low temperatures, thus thin films can New York 1993.
[14] P. Buffat, J.-P. Borel, Phys. Rev. A 1976, 13, 2287.
be made by electron-beam assisted chemical vapor deposi- [15] Z. L. Wang, J. Petroski, T. Green, M. A. El-Sayed, J. Phys. Chem. B 1998,
tion. Using a fine electron probe in STEM, patterned struc- 102, 6145.
REVIEW
[16] T. S. Ahmadi, Z. L. Wang, T. C. Green, A. Henglein, M. A. El-Sayed, [58] P. L. Gai, Top. Catal. 1999, 8, 97.
Science 1996, 28, 1924. [59] P. L. Gai, Microsc. Microanal. 2002, 8, 21.
[17] Y. H. Gao, Y. Bando, Nature 2002, 415, 599. [60] P. L. Gai, Curr. Opin. Solid State Mater. Sci. 2001, 5, 371.
[18] For a review see Z. L. Wang in Electron Microscopy of Nanotubes (Eds: [61] T. W. Hansen, J. B. Wagner, P. L. Hansen, S. Dahl, H. Topsoe, C. J. H. Ja-
Z. L. Wang, C. Hui), Kluwer Academic Publisher, New York 2003, Ch. 7. cobsen, Science 2001, 294, 1508.
[19] R. Berger, C. Gerber, H. P. Lang, J. K. Gimzewshi, Microelectron. Eng. [62] P. L. Hansen, J. B. Wagner, S. Helveg, J. R. Rostrup-Nielsen, B. S. Clau-
1997, 35, 373. sen, H. Topsùe, Science 2002, 295, 2053.
[20] G. H. Wu, R. H. Datar, K. M. Hansen, T. Thundat, R. J. Cote, A. Majum- [63] K. Suenaga, C. Colliex, N. Demoncy, A. Loiseau, H. Pascard, F. Willaime,
dar, Nat. Biotechnol. 2001, 19, 856. Science 1997, 278, 653.
[21] K. A. Stevenson, A. Mehta, P. Sachenko, K. M. Hansen, T. Thundat, [64] K. Suenaga, M. TencØ, C. Mory, C. Colliex, H. Kato, T. Okazaki, H. Shino-
Langmuir 2002, 18, 8732. hara, K. Hirahara, S. Bandow, S. Iijima, Science 2000, 290, 2280.
[22] G. Abadal, Z. J. Davis, B. Helbo, X. Borrise, R. Ruiz, A. Boisen, F. Cam- [65] J. M. Cowley, A. F. Moodie, Acta Crystallogr. 1957, 10, 609.
pabadal, J. Esteve, E. Figueras, F. Perez-Murano, N. Barniol, Nanotechnol- [66] J. Mayer, J. M. Plitzko, J. Microsc. 1996, 183, 2.
ogy 2001, 12, 100. [67] D. H. Pearson, C. C. Ahn, B. Fultz, Phys. Rev. B, 1993, 47, 8471.
[23] Z. J. Davis, G. Abadal, O. Khn, O. Hansen, F. Grey, A. Boisen, J. Vac. Sci. [68] H. Kurata, C. Colliex, Phys. Rev. B 1993, 48, 2102.
Technol. B 2000, 18, 612. [69] Z. L. Wang, J. S. Yin, Y. D. Jiang, Micron 2000, 31, 571.
[24] Z. W. Pan, Z. R. Dai, Z. L. Wang, Science 2001, 291, 1947. [70] Z. L. Wang, J. Bentley, N. D. Evans, J. Phys. Chem. B 1999, 103, 751.
[25] M. H. Huang, M. H., S. Mao, H. Feick, H., Yan, Y. Wu, H. Kind, E. We- [71] Z. L. Wang, J. Bentley, N. D. Evans, Micron 2000, 31, 355.
ber, R. Russo, P. Yang, Science 2001, 292, 1897. [72] Z. L. Wang, J. S. Yin, W. D. Mo, Z. J. Zhang, J. Phys. Chem. B. 1997, 101,
[26] M. S. Arnold, P. Avouris, Z. W. Pan, Z. L. Wang, J. Phys. Chem. B 2003, 6793.
107, 659. [73] T. Stöckli, Z. L. Wang, J.-M. Bonard, P. Stadelmann, A. Châtelain, Phys.
[27] E. Comini, G. Faglia, G. Sberveglieri, Z. W. Pan, Z. L. Wang, Appl. Phys. Rev. B 2000, 61, 5751.
Lett. 2002, 81, 1869. [74] T. Stöckli, J.-M. Bonard, A. Châtelain, Z. L. Wang, P. Stadelmann, Appl.
[28] Y. Cui, Q. Wei, H. Park, C. M. Lieber, Science 2001, 3, 1289. Phys. Lett. 2002, 80, 2982.
[29] M. K. Gudiksen, L. J. Lauhon, J. Wang, D. C. Smith, C. M. Lieber, Nature [75] A. V. Crewe, Chem. Scr. 1978±1979, 14, 17.
2002, 415, 617. [76] Z. L. Wang, J. M. Cowley, Ultramicroscopy 1989, 31, 437.
[30] H. G. Craighead, Science 2000, 290, 1532. [77] L. D. Marks, D. J. Smith, Nature 1983, 303, 316.
[31] W. Hughes, Z. L. Wang, Appl. Phys. Lett. 2003, 82, 2886. [78] P. D. Nellist, S. J. Pennycook, Science 1996, 274, 413.
[32] P. Poncharal, Z. L. Wang, D. Ugarte, W. A. de Heer, Science 1999, 283, [79] P. M. Voyles, D. A. Müller, J. L. Grazul, P. H. Citrin, H.-J. L. Gossmann,
1513. Nature 2002, 416, 826.
[33] Z. L. Wang, P. Poncharal, W. A. de Heer, Pure Appl. Chem. 2000, 72, 209. [80] Y. Zhu, V. V. Volkov, M. De Graef, J. Electron Microsc. 2001, 50, 447.
[34] R. P. Gao, Z. L. Wang, Z. G. Bai, W. de Heer, L. Dai, M. Gao, Phys. Rev. [81] A. Tonomura, Electron Holography, Springer-Verlag, New York 1993.
Lett. 2000, 85, 622. [82] M. De Graef, Y. Zhu, Magnetic Imaging and its Applications to Materials,
[35] Z. L. Wang, Z. R. Dai, Z. G. Bai, R. P. Gao, J. Gole, Appl. Phys. Lett. Academic Press, New York 2000.
2000, 77, 3349. [83] T. Matsuda, S. Hasegawa, M. Igarashi, T. Kobayashi, M. Naito, H. Kayiya-
[36] X. D. Bai, E. G. Wang, P. X. Gao, Z. L. Wang, Appl. Phys. Lett. 2003, 82, ma, J. Endo, N. Osakabe, A. Tonomura, Phys. Rev. Lett. 1989, 62, 2519.
4806. [84] D. Gabor, Proc. R. Soc. London, Ser. A 1949, 197, 454.
[37] L. Meirovich, Elements of Vibration Analysis, 2nd ed., McGraw-Hill, New [85] G. Möllenstedt, Adv. Opt. Electron Microsc. 1991, 12, 1.
York 1986. [86] A. Tonomura, Microsc. Soc. Am. Bull. 1994, 24, 501.
[38] D. Qian, E. C. Dickey, J. Microsc. (Oxford) 2001, 204, 39. [87] H. Lichte, Adv. Opt. Electron Microsc. 1991, 12, 25.
[39] E. A. Stach, T. Freeman, A. M. Minor, D. K. Owen, J. Cumings, M. A. [88] J. M. Zuo, M. Kim, M. O'Keeffe, J. C. H. Spence, Nature 1999, 401, 49.
Wall, T. Chraska, R. Hull, J. W. Morris, A. Zettl, U. Dahmen, Microsc. Mi- [89] Y. Zhu, L. Wu, J. Tafto, Phys. Rev. Lett. 2000, 85, 5126.
croanal. 2001, 7, 507. [90] M. Rühle, J. Eur. Ceram. Soc. 1996, 16, 363.
[40] D. Erts, A. Lohmus, R. Lohmus, H. Olin, A. V. Pokropivny, L. Ryen, [91] C. Scheu, W. Stein, M. Rühle, Phys. Status Solidi B 2000, 222, 199.
K. Svensson, Appl. Surf. Sci. 2002, 7699, 1. [92] P. Staddelmann, Ultramicroscopy 1987, 21, 131.
[41] For a review see Z. L. Wang in Electron Microscopy of Nanotubes (Eds: [93] S. B. Lee, W. Sigle, M. Rühle, Acta Mater. 2002, 50, 2151.
Z. L. Wang, C. Hui), Kluwer Academic Publisher, New York 2003, Ch. 8. [94] H. Lichte, Adv. Opt. Electron Microsc. 1991, 12, 25.
[42] P. Poncharal, C. Berger, Y. Yi, Z. L. Wang, W. de Heer, J. Phys. Chem. B [95] M. Haider, H. Rose, S. Uhlemann, E. Schwan, B. Kabius, K. Urban, Ultra-
2002, 106, 12 104. microscopy 1998, 75, 53.
[43] S. J. Tans, M. H. Devoret, H. J. Dai, A. Thess, R. E. Smalley, L. J. Geer- [96] O. L. Krivanek, N. Dellby, A. R. Lupini, Ultramicroscopy 1999, 78, 1.
ligs, C. Dekker, Nature 1997, 386, 474. [97] C. L. Jian, M. Lentzen, K. Urban, Science 2003, 299, 870.
[44] Z. L. Wang, P. Poncharal, W. A. De Heer, Microsc. Microanal. 2000, 6, 224. [98] Y. Shao-Horn, L. Croguennec, C. Delmas, E. C. Nelson, M. A. O'Keefe,
[45] S. Frank, P. Poncharal, Z. L. Wang, W. A. De Heer, Science 1998, 280, Nat. Mater. 2003, 2, 464.
1744. [99] H. W. Zandbergen, D. van Dyck, Microsc. Res. Tech. 2000, 49, 301.
[46] For a review see Z. L. Wang, in Electron Microscopy of Nanotubes (Eds: [100] P. E. Batson, Ultramicroscopy 1992, 47, 133.
Z. L. Wang, C. Hui), Kluwer Academic Publisher, New York 2003, Ch. 9. [101] N. Dellby, O. L. Krivanek, P. D. Nellist, P. E. Batson, A. R. Lupini,
[47] R. P. Gao, Z. W. Pan, Z. L. Wang, Appl. Phys. Lett. 2001, 78, 1757. J. Electron Microsc. 2001, 50, 177.
[48] Z. L. Wang, Adv. Mater. 2000, 12, 1295. [102] P. E. Batson, N. Dellby, O. L. Krivanek, Nature 2002, 418, 617.
[49] X. D. Bai, E. G. Wang, Z. L. Wang, unpublished. [103] J. M. Cowley, J. C. H. Spence, V. V. Smirnov, Ultramicroscopy 1997, 68,
[50] J. Cumings, A. Zettl, Science 2000, 289, 602. 135.
[51] J. C. H. Spence, W. Lo, M. Kuwabara, Ultramicroscopy 1990, 33, 69. [104] J. M. Cowley, Phys. Rev. Lett. 2001, 87, 016 101.
[52] Z. L. Wang, Reflection Electron Microscopy and Spectroscopy for Surface [105] J. M. Cowley, Ultramicroscopy 2000, 81, 47.
Studies, Cambridge University Press, Cambridge 1996. [106] R. D. Leapman, N. W. Rizzo, Ultramicroscopy 1999, 78, 251.
[53] Y. Kondo, K. Takayanagi, Science 2000, 289, 606. [107] K. Suenaga, T. Tence, C. Mory, C. Colliex, H. Kato, T. Okazaki, H. Shino-
[54] H. Ohnishi, Y. Kondo, K. Takayanagi, Nature 1998, 395, 780. hara, K. Hirahara, S. Bandow, S. Iijima, Science 2000, 290, 2280.
[55] V. Rodrigues, J. Bettini, A. R. Rocha, L. G. C. Rego, D. Ugarte, Phys. [108] C. J. Humphreys, I. G. Salisbury, S. D. Berger, R. S. Timsit, M. E. Mochel,
Rev. B 2002, 65, 153 402. EMAG'85, Inst. Phys. Conf. Ser. 1985, 78, 1.
[56] V. Rodrigues, D. Ugarte, Nanotechnology 2002, 13, 404. [109] Z. L. Wang, J. Microsc. 1991, 163, 261.
[57] R. Sharma, Microsc. Microanal. 2001, 7, 494. [110] C. Ma, Y. Berta, Z. L. Wang, Solid State Commun., in press.
______________________

New Developments in Transmission Electron Microscopy For Nanotechnology.

Uploaded by

Copyright:

Available Formats

New Developments in Transmission Electron Microscopy For Nanotechnology.

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

New Developments in Transmission Electron Microscopy For Nanotechnology.

Uploaded by

Copyright:

Available Formats

REVIEW

New Developments in Transmission Electron

High-resolution transmission electron microscopy (HRTEM) is one of the most

Fig. 2. a) A typical TEM image of a ZnO nanobelt. b) Schematic geometrical

Fig. 4. In-situ observation of electric transport through a single carbon nano-

ball. Due to the difference in the surface work functions be-

thickness of one atomic column. The corresponding measure-

2.6. In-Situ Environmental Microscopy for Observing Gas±

Structural analysis in TEM is always carried out in vacuum,

reactions in the manufacture of polyamides have been imaged

and EELS. Due to the small thickness of samples used for

onstrated to map the valence state distributions of Co (or Mn)

~ 1.8 eV to higher energy-loss and a prominent peak at the

4. Quantitative Microscopy and Electron

4.1. Dark-Field Imaging of Small Clusters and Point Defects

Imaging in the TEM is mainly dominated by elastic scatter-

fe(s) = e[Z ± fx(s)]/16p2e0s2, (5)

where fx is the X-ray scattering factor of the atom, Z its atom-

In many cases, it is believed that the surfaces of small parti-

at the diffraction plane that detects the electrons scattered to

ITDS(s) = [fe(s)]2 [1 ± exp(± 2Ws2)] (6)

Imaging of magnetic domains and interdomain interactions

where dS is a surface integral over the surface area enclosed

Fig. 21. A quantitative comparison of the high-resolution TEM images re-

5. Aberration-Corrected High Resolution Imaging

rector for improving the probe size of a 100 kV STEM to

You might also like