Detekcija Ostecenja

J. Phys. D: Appl. Phys. 29 (1996) 1389–1417.
Printed in the UK
REVIEW ARTICLE
Industrial ultrasonic imaging and

microscopy
R S Gilmore
General Electric Company, Building KWD, Room 251, PO Box 8, Schenectady,
NY 12301, USA
Received 12 March 1996
Abstract. Ultrasonic imaging and scanned acoustic microscopy are terms used to
describe similar imaging processes at different magnifications and frequencies.
Both processes form images by acquiring spatially correlated measurements of the
interaction of high-frequency sound waves with materials. With the exception of the
interference measurement, called V (z ), and the gigahertz frequencies used by the
higher frequency scanning acoustic microscopes, it is difficult to establish
operational differences between them. This is especially true since almost all
commercial ultrasonic imaging systems use transducers producing focused beams
and can display magnified high-resolution images.
Ultrasonic C-scan imaging was developed largely by the ultrasonic
nondestructive testing industry. The development was gradual and evolutionary.
Over a 50-year period, better and better broadband transducers, electronics and
scanners were developed for operation at progressively higher frequencies, now
ranging from 1.0 to 100 MHz. Conversely, scanning acoustic microscopes made a
relatively sudden appearance 20 years ago on the campus of Stanford University.
The first scanning acoustic microscopes operated at gigahertz frequencies and
used microwave electronics that produced acoustic tone bursts with many
wavelengths per pulse.
Three factors control resolution in an acoustic image:
• diameter of the acoustic beam or its point spread function (PSF);

• size and spacing of the pixels making up the image;
• signal-to-noise ratio (contrast) of the feature being resolved.
The beam diameter, or PSF, is controlled by the frequency of the ultrasonic

pulse and the focal convergence of the beam (or focal length to diameter ratio
Z /d ). In the coupling fluid, the Z /d ratio is determined by the transducer diameter
and lens, but in the material, Z /d is established by the materials ultrasonic
velocities. Pixels are the squares of colour or greyscale that make up computer
displays of scanned images. Following Nyquist’s criterion, the resolution of those
images is twice the size and spacing of the pixels. It follows, therefore, that in
order to support the resolution of an ultrasonic beam, the pixels must be no larger
than half that beam diameter. Finally, the contrast of the feature being studied
must be (at least) a clear shade of grey above the background produced by the
image noise. The noise can be due to the material or the electronics.
Written to support industrial ultrasonic inspection of materials, this discussion
will emphasise the similarities between imaging and microscopy rather than the
differences. The roles of the focusing lens, the pulse frequency, and the material
being imaged, with respect to the final resolution of an acoustic image, will be
considered in detail. It will be shown that additional improvements in resolution can
be achieved with image processing. Finally, applications studies in metals,
ceramics, composites, attachment methods, coatings, and electronic assemblies
will be used to demonstrate specific roles for imaging/microscopy in nondestructive
testing.
1. Introduction and historical review acquire images of sound waves interacting with materials,
considerable effort has gone into defining the differences
Because both ultrasonic C-scan imaging and scanning between them. A typical ultrasonic imaging/microscopy
acoustic microscopy (SAM) describe processes used to system is shown schematically in figure 1. The early
0022-3727/96/061389+29$19.50
c 1996 IOP Publishing Ltd 1389
R S Gilmore
z entire reflected waveforms from a region in the sample.

All three types of image data can be acquired with many
y FOCUSED ULTRASONIC TRANSDUCER
commercially available ultrasonic imaging systems, and the
WATER images can be displayed as C-scans, B-scans, or by solid
ULTRASONIC
SURFACE TRANSMITTTER/ SCAN
CONTROLLER
model imaging formats.
RECEIVER
ELECTRONICS
In industrial quality control, almost all ultrasonic
imaging systems use broadband electronics, detect time-
x DATA
resolved signals, and acquire images with mechanical
θ ACQUISITION/DISPLAY scanning. One exception would be the use of the
COMPUTER
MULTI-AXIS SCANNER phase interference method V (z). This method requires
narrowband systems and, when mechanically scanned,
produces phase interference images.
MAGNETIC/
OPTICAL
At present, and after years of development, ultrasonic
CRT
DISPLAY
STORAGE C-scan immersion inspection continues to provide signif-
icant assistance in materials evaluation. It provides hard
copy and/or computer records of test results. Producing
Figure 1. Schematic of a typical ultrasonic
imaging/microscopy system. greyscale or colour images, it simplifies detection by dis-
playing both high- and low-amplitude signals and discrim-
inates sound material from regions containing flaws on the
acoustic microscopes operated at gigahertz frequencies, basis of signal amplitude. In addition, C-scan-type imag-
and ultrasonic imaging was done at megahertz frequencies ing spatially correlates multiple signals to indicate a single
but, recently, work with even that distinction has become flaw when successive scan lines and pulses provide multiple
blurred, and there are many more similarities than signals.
differences between them. The terms scanned ultrasonic Table 1 shows that the use of high-frequency acoustic
microscopy and scanned acoustic microscopy can be used waves for visualizing the interiors of opaque solid
interchangeably, as can ultrasonic C-scan imaging and materials was first recognized in Russia by Sokolov and
scanned acoustic imaging. All are processes which in Germany by Muhlhauser in 1929. Muhlhauser filed
form images of soundwave interactions with materials for a German patent for the concept of ultrasonically
and involve scanning an acoustic source by electronic or inspecting and visualizing material volumes in 1931. Until
mechanical means. 1940, only continuous-wave transmission techniques were
If ultrasonic/acoustic microscopy refers to processes used in all ultrasonic research and development. When
that create ‘magnified’ images of objects, then it follows Bergmann published Ultrachall in 1937, he included
that ultrasonic or acoustic imaging refers to processes that approximately 600 papers on various continuous-wave
create images of objects without magnification. Because ultrasonic investigations. Several of these papers mentioned
magnified images involve acquiring acoustic data from imaging concepts. An additional 90 papers were added
many small areas of the object or region, either the when the English translation of Bergmann’s book was
interrogating acoustic beam must be focused, or the energy published in 1938 [1].
used to create the acoustic energy must be focused, in Sokolov was the first to produce a device to make
order to insonify each of these small areas independently. acoustic images and also to realize that 3 GHz acoustic
However, it is not unusual for ultrasonic beams at images would rival optical images in resolution and detail.
essentially any frequency to be focused in order to improve The Sokolov Tube, first described in 1936, was an image
the resolution and contrast of the acoustic image acquired converter that produced a television display by scanning the
with that beam. Acoustic microscopy has been used to back surface of a (quartz) piezoelectric receiver element
describe images acquired at 1 to 3 MHz and ultrasonic with an electron beam. The acoustic amplitude pattern
imaging for images at 75 to 100 MHz. To further the or shadowgraph on the receiver was produced by passing
confusion, product names have been introduced by various an acoustic beam through the object being imaged. The
companies with acronyms that are very similar to C-scan early images produced by Sokolov Tubes were at best
and SAM. uninspiring. This was because of poor resolution and
Historically, industrial applications for ultrasonic contrast, and very low signal strength. Pohlmann (1939)
imaging have evolved toward broadband rather than also recognized the ‘possibility of an acoustic image in
narrowband (tone burst) signals. When broadband systems analogy to an optical image’. Pohlmann Cells depended
are used, the short time duration of the pulses obscures on the intensity of the ultrasonic waves developing patterns
the near entry surface material for only a few round- in liquid suspensions of aluminium powder. The resulting
trip wavelengths; therefore, thinner material sections can images showed little improvement in comparison to the
be evaluated. Most of the ultrasonic imaging/microscopy Sokolov Tube, and required a longer development time to
applications described here were done with broadband form the particles into patterns. The work of both has been
pulses and systems. These data can be acquired during described by Bergmann [1] and others [2, 3].
uninterrupted mechanical scanning by peak detecting a Firestone’s 1940 invention and development of
single maximum amplitude, detecting both the amplitude devices to produce pulsed acoustic waves at ultrasonic
and the phase of the largest pulse, or by digitizing the frequencies [3], when combined with the development of
1390
Industrial ultrasonic imaging and microscopy
Table 1. Acoustic imaging and materials evaluation: a timeline.
Year Event
1929 Muhlhauser and Sokolov independently propose ultrasonic waves for materials evaluation.
Scanned imaging is suggested outright in manuscripts by both [1–3].
1931 Muhlhauser obtains German patent for ultrasonic testing of materials using continuous wave transmission [1–3].
1936 Sokolov Tube patented and published; Sokolov observes 3 GHz acoustic images equal optical resolution [1–4].
1937 Bergmann writes Ultraschall (in German) with ∼ 600 references (English translation in 1938) [1].
1940 Firestone invents pulse–echo ultrasonic testing (patent awarded 1942). SNT (later ASNT)
chartered to provide a professional forum for nondestructive testing [2, 3].
1945 Sperry acquires Firestone patent (ultrasonic reflectoscope). Erdman, Krautkramer, Pringle, and
to Smack develop ultrasonic C-scan equipment. Hastings, using an Erdman system, makes
1958 greyscale C-scan images: film, paper [3].
1959 First ASNT handbook by McMaster. C-scans, focused probes, scanned images, CRT
greyscale, etc [3]. Dunn proposes scanned ultrasonic testing absorbtion microscope [5].
1963 Jacobs adds electron multiplier to Sokolov Tube [6].
1966 Korpel et al at Zenith Corp. invent scanning laser acoustic microscope [7].
1967 First international symposium on acoustical holography (later changed to ‘imaging’) [8].
1969 Batalle founded Holotron Inc., later Holosonics Inc., to market acoustic holography systems.
1971 Fowler at Panametrics Inc. introduces and markets a quartz buffer–rod–lens focused 50 MHz transducer†.
1973 Lemons and Quate invent and introduce 1 GHz SAM [9]. Stanford group includes G Kino, P Khuri-Yakub,
and B Auld.
1974 Sonoscan Inc. founded by L Kessler to market SAM‡. E Ash builds SAM group at University
College, London, UK‡, that includes C Tsai and H Wickramasinghe.
1977 Tsai builds second SAM group in US at Carnegie Mellon University [10].
1980 EI Leitz Ltd and Olympus Ltd introduce scanning acoustic microscopes to international
market. Imaging is now seeing rapid growth.
1990s More than 30 firms manufacture industrial acoustic imaging/microscopy systems for an international market.
† Panametrics Inc., 221 Crescent Street, Waltham, MA 01754, USA.

‡ Sonoscan Inc., 530 E. Green Street, Bensen, IL 60106, USA.
mechanical scanning systems initiated rapid development in In 1963 Jacobs et al [6] added an electron multiplier
ultrasonic pulse–echo testing and C-scan immersion testing. to the Sokolov Tube for a factor of ten increase in signal
The scanned ultrasonic systems developed by Erdman, strength. Even then Sokolov Tube images showed poor
Krautkramer, Pringle, and Smack during the 1950s were resolution and contrast in comparison to the ultrasonic C-
summarized and described in ASNT’s first Nondestructive scan images that were being developed during the same
Testing Handbook [3]. This work also referenced greyscale period, but they had the advantage of providing a real-
images that were originally published in 1955 by Buchanan time image. In presentations Jacobs et al usually showed
and Hastings [11]. Most of these scanning systems moving pictures because object motion gave the observer
recorded ultrasonic amplitude data on electrostatic paper. better apparent contrast than still images. Korpel et al [7]
The greyscale images were recorded on photographic film developed the scanning laser acoustic microscope during
through Z-axis modulation of the electron beam of an this period.
X–Y oscilloscope. Polaroid camera backs became the C-scan testing continued to develop. Because
most popular method for recording the brightness of the electrosensitive paper was manufactured in widths up to
oscilloscope displays. However, because the Polaroid 75 cm (30 in), large parts could be scanned and imaged with
images were limited to a 4 × 5 inch format and had to mechanically linked plotters. This made paper the most
be scanned and then viewed, they were more useful for popular C-scan recording medium for industrial testing.
research and development studies than for nondestructive The pulse–echo or pulse–transmission images were usually
evaluation. of flaw echos that exceeded an amplitude threshold. For
1391
R S Gilmore
many systems, these images were binary (either black or SAPPHIRE

white), typically at 1 × magnification, and consisted of TRANSMITTER
LENS
MYLAR
SUPPORT SAMPLE
shaded areas with flaws and white areas without flaws.
RECEIVING
Thresholds used to detect the flaw signals are usually OSC TRANSDUCER
set to some fraction of the amplitude from a reference AMP
reflector. The reference was made the same size as the
flaw to be detected. Reference reflectors are still the SENDING
TRANSDUCER Y SAPPHIRE CRT DISPLAY
most commonly used sensitivity references for ultrasonic WATER
RECEIVER
LENS
nondestructive evaluation (NDE). They are typically slots, X
flat-bottomed holes (FBHs), or side-drilled holes and are SCAN CONTROL
MECHANICAL
machined into the same material used to manufacture the SCAN
AND CRT DRIVE
CIRCUITRY
part under evaluation. This ensures that the attenuation in
the calibration sample is the same as in the part, and that (a )
the reference flaw is placed in an equivalent microstructure. CIRCULATOR PULSED OSCILLATOR
Acoustic noise from the material is excluded by the
threshold.
In the later 1960s, greyscale displays using electrosen- SENDING
AND RECEIVING RECEIVER
sitive paper or thermal paper began to replace the binary TRANSDUCER
displays of the preceding decade. But it was the introduc- CRT
SYNC
SAPPHIRE
tion of small computers in the 1970s that provided both LENS WATER
DISPLAY
acoustic imaging and acoustic microscopy with vastly im- Y

SAMPLE
proved capabilities for industrial NDE, materials science X SCAN CONTROL
AND CRT DRIVE
MECHANICAL
and medicine. SCANNER
CIRCUITRY
In 1973, work at Stanford University under the

(b )
direction of Quate [9] introduced a new dimension to
scanned acoustic imaging. The Stanford scanning acoustic Figure 2. Schematic diagrams of scanning acoustic
microscopes: (a ) pulse–transmission and (b ) pulse–echo
microscopes operated at gigahertz frequencies, produced modes (C Quate and L Lam, Stanford University).
images with optical microscope resolution, and gave a new
excitement and focus to the research effort in acoustic
imaging and microscopy worldwide. Four decades after complete introductory survey, it is necessary to list,
Sokolov’s prediction, images to rival optical microscopy describe, and reference these four methods.
became a fact. Unlike the lower frequency C-scan systems,
these gigahertz-frequency acoustic microscopes used tone-
burst systems having only a few per cent bandwidth. 1.1. Scanning acoustic microscopy (SAM)
During this period, broadband commercial transducers
In scanning acoustic microscopy, images are formed of the
and electronics appeared with centre frequencies up to
surface or interior regions of materials by mechanically
100 MHz. Possibly the most popular transducer design
scanning piezoelectric transducers that produce focused
was the broadband 50 to 100 MHz ultrasonic transducer
acoustic beams. These image data are acquired scanned line
with a single-surface quartz lens, introduced by Fowler in
by scanned line. The resolution of the image is controlled
1971 [19].
by the diameter of the focused acoustic beam and by the size
Tone burst excitation was also used in SAM studies
and spacing of the pixels that display the assembled data.
at frequencies down to 100 MHz. However, with
the exception of almost all ultrasonic industrial NDE A single focused transducer may be used to transmit and
[11, 12] and the pulse compression acoustic microscopy receive reflected signals, or two transducers may be used
by Nikoonahad et al [13] most SAM studies until the to transmit and receive signals passing completely through
mid 1980s utilized acoustic pulses containing multiple the object [22–26] (figure 2).
wavelengths and narrow-band systems. Calculations
for material properties require amplitude and phase 1.2. Photoacoustic microscopy (PAM)
measurements at several heights of the acoustic transducer
above the sample. Algorithims for the use of these V (z) Short pulses of light, typically from a laser, are focused and
data have become highly sophisticated as reported in the scanned over the surface of the object. The light pulses are
literature [14–26]. formed by chopping or pulsing the light source. Acoustic
Between 1973 and 1980, other methods for generating pulses result from the rapid localized heating of the object
and scanning highly focused acoustic waves were surface by each of the light pulses. The acoustic pulses are
developed and used to make acoustic imaging/microscopy in turn monitored by a piezoelectric receiver or by laser
systems. Most acoustic microscopy may be summarized sensing the time-delayed surface displacements due to the
under four methods that are used to generate, focus, and reflected acoustic pulse. Although the transmitted acoustic
scan the acoustic energy: scanned acoustic microscopy, pulses may be used to monitor the interior of an object, the
photoacoustic microscopy, electron acoustic microscopy, best resolution is obtained at or very near the surface of the
and scanning laser acoustic microscopy. To make a material [22, 23, 27] (figure 3).
1392
FOCUSED MODIFIED
SAMPLE ACOUSTIC SCANNING
ELECTRON MICROSCOPE
RECEIVER
SAPPHIRE
LENS
LASER LENS
AMP
FOCUSED
LIGHT SPOT RECEIVING
Y TRANSDUCER CRT
SOUND DISPLAY SCAN CONTROL
X AND CRT
WAVES DRIVE CIRCUITRY
ELECTRON
SCAN CONTROL BEAM
MECHANICAL SCANNING
AND CRT DRIVE
SCAN COILS
CIRCUITRY
SYNC X
(a ) Y
SYNC
SAMPLE CRT
LASER SIGNAL OUT
PROBE SQUARE LAW
(RECEIVER) DETECTOR STAGE
SOUND WAVES
BEAM LASER
SPLITTER TRANSMITTER
RECEIVER
LENS SCAN CONTROL PIEZOELECTRIC AMPLIFIER
AND LASER TRANSDUCER
SAMPLE DRIVE CIRCUITRY RECEIVER
Y
X
Figure 4. Schematic diagram of a scanning electron
MECHANICAL
SCANNER acoustic microscope (S Cargill [28]).
COMPUTER
CRT DISPLAY
SCAN CONTROL
(b ) AND CRT DRIVE
CIRCUITRY
Figure 3. Schematic diagrams of photoacoustic
LASER SCANNER
microscope systems: (a ) C Williams, Stanford University PHOTODETECTOR
and (b ) Y Martins and E Ash, University College, London.
LUCITE
COVER-SLIP
CRT KNIFE-EDGE
DYNAMIC RIPPLE
1.3. Scanning electron acoustic microscopy (SEAM) FLAW
SAMPLE
Electron acoustic microscopy is similar to photoacoustic
microscopy in that a pulsed or chopped electron beam
ACOUSTIC MICROSCOPE
is used to produce the acoustic waves. The apparatus is FREQUENCY STAGE
typically made by incorporating a piezoelectric receiver GENERATOR
PIEZOELECTRIC INCIDENT SOUND
on the stage of a scanning electron microscope (SEM). TRANSDUCER
Scanning electron acoustic microscopy is also most
effective on very near surface conditions, and the samples Figure 5. Schematic diagram of a scanning laser acoustic
must be small enough to fit on a SEM stage [28, 29] (figure microscope (L Kessler, Sonoscan Inc.).
4).
1.5. Ultrasonic testing versus ultrasonic imaging
Manual observations of ultrasonic waveforms are still the
1.4. Scanning laser acoustic microscopy (SLAM) most widely used industrial ultrasonic test method. This has
been primarily due to the low cost of the instrumentation
The scanning laser acoustic microscope was the first
in comparison to the cost of an imaging system. Scanned
acoustic imaging system that used the term acoustic imaging systems have steadily expanded in industrial use,
microscope in the description of the apparatus. This method especially where the highest probability of detection for
introduces a broad and uniform acoustic beam through flaws is required. This is because the spatial correlation
one surface of the object being imaged and then scans provided by C-scan images gives a higher probability of
the displacement patterns of the opposite surface with a detection than is possible with manual observations, even
focused laser. While the resulting displays are often clear with the use of automatic flaw alarms and mechanical
and informative, they are transmission shadowgraphs, like scanners. Because scanned acoustic microscopy utilizes the
those formed by the Sokolov Tube and Pohlmann Cells, same technology as ultrasonic C-scan imaging, acquiring
rather than images. As for all shadow-graphs, the resolution magnified images of detected flaws is becoming a natural
of a feature is dependent on its depth from the scanned extension of ultrasonic immersion testing.
surface [7, 23, 24] (figure 5). Until the development of greyscale imaging, little
1393
R S Gilmore
ultrasonic surface wave scanning was done because of the PAM Photoacoustic microscopy—a pulsed light source
sensitivity of sharply inclined beams to finish roughness, scanned to image a material surface/ volume
finish direction, and to flaw orientation. This surface finish SEAM Scanning electron acoustic microscopy—an
noise made ultrasonic surface wave testing one of the electron beam scanned to image a material
least dependable methods of interrogating surfaces. The surface/volume
surface waves generated by scanning acoustic microscope SNR Signal to noise ratio
techniques are also sensitive to surface roughness but, X Lateral coordinate direction, X = 0, specified
because it images the material structure, surface finish, and at the intersection of the material surface with
any detected flaws, it is less sensitive to the direction of the axis of symmetry of the acoustic beam
the finish or the orientation of a flaw. Therefore, SAM Z Vertical coordinate direction, Z = 0, also
methods have greatly contributed to surface inspection at at the intersection of the material surface with
all ultrasonic frequencies. In addition to the detection of the beam axis
flaws, acoustic images also contain substantial information
for materials characterization. Grain size can be measured 3. Description and theory of acoustic
with standard observational techniques applied to acoustic imaging/microscopy
greyscale images of surfaces. Weglein [15], Liang et
al [17], Briggs [26], Kushibiki and Chubachi [19] and To describe the process of acoustic imaging/microscopy,
others [16, 23, 25] have developed analytic methods for consider the steps required to acquire and display an
tone burst microscopy, and similar calculations have been acoustic image.
and are being developed for broadband systems. These
developments will be surveyed and summarized for NDE (1) A sample is selected for imaging and mounted on
and materials characterization. a scanning stage, and the type of imaging is selected, i.e.
surface or subsurface.
(2) A transducer (size, frequency, and focus) is selected
2. List of symbols and abbreviations on the basis of
type of imaging to be done;
CL Velocity of longitudinal wave (mm µs−1 )
acoustic velocities in the sample;
CS Velocity of shear or transverse wave
resolution required for the image.
(mm µs−1 )
(3) Based on the beam diameter (resolution) produced
CR Velocity of surface wave (Rayleigh wave)
by the transducer, a pixel size and spacing (of no more
(mm µs−1 )
than half the beam diameter) is selected, and a scan plan is
C1 Longitudinal velocity in acoustic lens (mm µs−1 )
loaded into the scan controller.
C2 Longitudinal velocity in coupling (focusing)
(4) Data for the image are acquired in the form of
medium (mm µs−1 )
signal amplitude, amplitude and phase, time of flight from
C3 Velocity in test material, this could be Cl , CS ,
a reference, or fully sampled waveforms which can be
or CR , depending on application or equation
processed to give all of the preceding information.
(mm µs−1 )
(5) Colour or greyscale is selected to display the image
DR Diameter of entry circle for mode-converted
data, and the dynamic range of the acquired/processed data
surface wave (mm)
is gain adjusted to coincide with the dynamic range of the
d Diameter of lens or transducer generating
display.
acoustic wave (mm)
EX Lateral acoustic beam diameter (mm) To describe the process fully, it is necessary to treat
ER Diameter of surface wave focus (mm) each of these steps in detail, especially the transducers, the
EZ Depth of focus for acoustic beam (mm) materials, and the procedures for surface and subsurface
Z Focal length of lens or curved element imaging.
transducer (mm)
Z2 Focal length in couplant medium with 3.1. Focusing acoustic beams
velocity C2 (mm)
Z3 Focal length/depth in test material with Most commercial focused transducers are constructed with
velocity C3 (mm) a flat piezoelectric element producing a quasi-plane wave
f Frequency of acoustic wave (Hz) pulse that is focused by a spherical lens. More recently,
λ Wavelength of acoustic wave (mm) transducers with spherically curved piezoelectric elements
λ2 Wavelength in velocity C2 (mm) have become available. Both methods produce convergent
NDE Nondestructive evaluation—a term becoming spherical waves and well defined focal zones in the coupling
more commonly used than nondestructive testing fluid. However, when highly focused beams are projected
2L First critical angle (longitudinal waves) into solid substrates with high acoustic velocities, the beams
2S Second critical angle (shear waves) become badly aberrant.
2R Rayleigh critical angle (surface waves) Figure 6 summarizes the focusing of acoustic beams.
SAM Scanning acoustic microscopy—a focused The point focus produced by a spherically curved
piezoelectric transducer scanned to image a piezoelectric element or a very high index lens (figure
material surface/volume 6(a)) would be subject to at least some spherical aberration
1394
HIGH VELOCITY
HIGH VELOCITY LOW VELOCITY SPHERICAL
SPHERICAL PARAXIAL SPHERICAL PARAXIAL LENS
LENS RAY LENS RAY
C1
C1
C1
C2 C2 C2
C3
MARGINAL MARGINAL HIGH
DIFFRACTION ABERRATED RAY
LIMITED
RAY VELOCITY
FOCUS
FOCUS SUBSTRATE
(a ) (b ) (c )
ASPHERIC LENS
DESIGNED FOR
ASPHERIC LENS
SUBSTRATE DESIGNED FOR
C1 SUBSTRATE
C1
C2
C2
PARAXIAL
C3 FOCUS, WATER
HIGH (FPW) WITH SUBSTRATE
REMOVED
VELOCITY MARGINAL ASPHERIC
SUBSTRATE FOCUS, WATER WILL NOT FOCUS
(d ) (e )
Figure 6. Five general conditions for focusing acoustic beams: (a ) point focus from a spherically curved element; (b )
aberrant focus produced by a low index spherical lens; (c ) aberrant subsurface focus produced by a high-velocity material;
(d ) point focus produced by a Fermat lens or element; (e ) distributed focus produced in water by a Fermat lens or element.
when focused by a low-index spherical lens as in figure distance to the point where the ray crosses the axis is Z,
6(b). Regardless of the focus of the beam in the coupling then X = RC sin 21 , Z 0 = RC (1 − cos 21 ) and
fluid, it becomes badly aberrant if a high-velocity substrate
X RC sin 21
is inserted between the lens and the focal point (figure Z − Z0 = = (2)
6(c)). An aspheric lens or piezoelectric element (figure tan(21 − 22 ) tan(21 − 22 )
6(d )) can be designed to produce a well behaved focus and finally
in the solid of a specific velocity and at a specific depth;
however, it will not produce a focused beam in the coupling sin 21
Z = RC + 1 − cos 21 . (3)
fluid (figure 6(e)). Transducers with spherical piezoelectric tan(21 − 22 )
elements do produce converging spherical waves and
diffraction limited beams in the coupling fluid, where For small values of 21 , 22 = (C1 /C2 )21 , both Z 0
spherical lenses produce aberrant beams, hence the term and (21 − 22 ) are small, cos 21 = 1.0, and sin(21 −
spherical aberration. However, for those spherical lenses 22 ), tan(21 − 22 ) and (21 − 22 ) can be taken as equal.
with very high indices of refraction to the coupling fluid, Equation (3) now becomes the ‘lens makers equation’.
the aberration in the fluid is very small. Consider a family
1
of acoustic rays parallel to the axis of symmetry (figure Z0 = RC . (4)
(1 − C2 /C1 )
7(a)), where each ray is incident to a spherical surface
of radius RC at some angle 21 . If the spherical surface Equation (4) is in fact precise for small-angle low-index
bounds two media characterized by acoustic velocities C1 lenses and for large-angle high-index lenses. Comparing
and C2 , then the refractive relationships between the rays the focal lengths determined by equations (3) and (4) shows
in medium 1 and 2 are how large angles and low-index lenses can and do produce
C1 spherical aberration. Figure 7(b) shows the aberration
sin 21 = sin 22 (1) curves (Z/Z0 ) for acoustic lenses manufactured from five
C2
commonly used lens materials. Note that sapphire and
where equation (1) is Snell’s law and C1 /C2 is the index [111] silicon produce very little aberration and form a point
of refraction for the fluid-to-solid interface. Note that high- focus even for lenses with significant curvature.
velocity lens materials such as quartz, silicon, and sapphire Regardless of how perfectly the beam is focused in the
produce large values of C1 /C2 , the refracted angle 22 is coupling fluid, figure 6(c) shows that significant aberration
small, and the refracted ray falls close to the radius of will occur upon entry into a high-velocity substrate. In
curvature of the lens. If the lateral distance from the axis the case of fluid-to-solid or low velocity to high velocity,
of symmetry to the ray–lens intersection is X, and the the high indices of refraction produce greater aberration
1395
R S Gilmore
θ1
x
Z'
C1 X
C2
θ1
θ2
C
R
Z
Figure 8. Schematic showing the point spread function for

θ2
θ1 –
a circular lens.
sin 23 = 1.0 and

C2
sin 22 = . (6)
C3
Because Snell’s law is not linear, to produce a well
(a ) behaved focus in a high-velocity material, either 23 must
1.0
be small (actually less than 30◦ works well) or an aspheric
C1/C2 = 7.6 focusing lens must be used (figure 6(d )).
C1/C2 = 6.0 The term point focused is misleading. If all is managed
well, the point focused transducer produces a diffraction-
C1/C2 = 2.0 C1/C2 = 4.0
limited spot size described by its point spread function
(PSF). For a circular transducer of diameter d transmitting
0.9
a wavelength λ2 , the theoretical beam amplitude profile is
[12, 17, 30–33]:

2J1 (x) 2
P(X) = (7)
C1/C2 = 1.5 (x)
0.8
where X = 0.5kd(r/Z), k is the wave number (2π/λ), Z is
INDEX MATERIAL the lens-to-focus distance, and r/Z is the angular distance
from the axis of symmetry of the beam. If this function is
2.0 PLEXIGLASS
1.5 POLYSTYRENE evaluated at −1, −3, and −6 dB with respect to its central
0.7 4.0 FUSED QUARTZ maximum, the beam diameters (figure 8) are:
6.0 SILICON (111 CUT)
7.6 SAPPHIRE (001) Z
Ex = Kλ2 (8)
d
0 10 20 30 40 50 60 70 80 90
ANGLE OF INCIDENCE θ1 for K = 0.43, 0.74, and 1.03 respectively. All three
numbers are important for acoustic imaging. The −1 dB
(b )
diameter is a good value for the pixel dimension, the −3 dB
Figure 7. (a ) Ray in medium 1 incident on a spherically beam is the value producing a single shade of grey in the
curved surface of radius RC bounding media 1 and 2 image, and the −6 dB diameter is the 0.5 amplitude beam
where C2 < C1 . (b ) Aberration curves plotted for the
marginal focus F divided by paraxial focus FP . used industry wide to specify the transducer resolution.
In addition to having a diffraction-limited diameter, the
focal zone also has a diffraction-limited depth of focus
[12, 17, 30, 31, 34]. Using the same notation the −3 dB
rather than reduce it. Consider the spherically convergent
depth of focus is.
beam in the fluid (C2 ) incident on a third higher velocity
2
medium with velocity C3 . Snell’s law again establishes the Z C2
relationships: EZ = 3.6λ2 . (9)
d C3
C3
sin 23 = sin 22 . (5) Figure 9 shows the range in beam diameter and depth of
C2
focus for the typical industrial ultrasonic range from 2.25–
When refracted angle 23 approaches 90 degrees of arc, 100 MHz.
1396
4.5˚
0.7 θL = 1 CL = 5.97
17.5 8˚ LENS θL
˚ 26.
9.9 =
= 2 θS θ
θ
θR S
R
0.6 5 MHz 15.0

WATER
0.5 Cw = 1.48
12.5
εz (INCHES)
10 MHz
0.4
εz (mm)
10.0
STEEL
C
0.3 7.5
Z (Cw)L
CL = 5.90
25 PARAXIAL CS = 3.28
Z (CCw) PARAXIAL
LONGITUDINAL
FOCUS CR = 2.97
0.2 5.0
S
SHEAR
FOCUS
Z
0.1 50
2.5
100
θR
0 5 10
(a )
Figure 10. The angular spectrum for a water–steel
.100 2.5 interface showing the critical angles for longitudinal and
shear waves, and the Rayleigh critical angle.
2.0
HZ
.08
5M
εX (INCHES)
0˚
2.2
5.5˚6.8˚ POLYCRYSTALLINE
εX (mm)
.06 5 1.5 8.6˚

DIAMOND
COMPACT
.04 10 1.0 θL = 5.5˚
θS = 6.8˚
θR = 8.6˚
.02 25 0.5
50
100 STEEL
0 5 10 θL = 14.5˚
θS = 26.8˚
(b ) 0˚ 14.5˚ θR = 29.9˚
Figure 9. Beam diameters and depths of foci for a range of 26.8˚
frequencies and F /d ratios. 29.9˚
4. Role of imaged material: permitted resolution
Most scanned acoustic imaging systems, including acoustic

microscopes, use some form of fluid coupling to introduce PLASTIC
0˚ CHIP CARRIER
their focused acoustic beam into the test material. It 28.7˚
θL = 28.7˚
is apparent from equations (5) and (6) that a practical θS = 55.3˚
55.3˚
understanding is required of the interaction of focused θR = 66.0˚
66.0˚
acoustic beams with fluid–solid interfaces if one is to select
the correct transducer for acoustic imaging.
Consider a planar water–steel interface (figure 10). The
ambient (22 ◦ C) longitudinal velocity of deionized water
is CL = 1.48 mm ms−1 . Shear waves in low-viscosity
fluids are neglected because their propagation distances
are usually less than one wavelength. For this case,
the steel is specified to be fine grained, homogeneous, Figure 11. Schematic indication of the angular spectra for
and isotropic. The longitudinal, shear, and surface wave polycrystalline diamond, steel, and Al2 03 reinforced plastic
velocities respectively are listed in table 2. Restricting (the plastic is used in electronic chip carriers) where water
is the incident medium.
this discussion to geometrical acoustics and neglecting
nonpropagating waves (sometimes called evanescent or
head waves), Snell’s law (equation (5)) describes the a spherically focused acoustic beam with a water–steel
direction of the incident and refracted rays and the three interface is complicated even though the elastic symmetry
critical angles at the water–steel interface as shown. The of the isotropic material is as simple as can be specified for
critical angles for the water–steel interface are tabulated a solid. The behaviour shown in figure 10 is often referred
in both figure 10 and table 2. Clearly the interaction of to as the angular spectrum for steel.
1397
R S Gilmore
DL
(1) (6) L
C1
lens (5)
(2)
θR (I) θR θR
C2 Z2
RC
(3) (4) liquid
CR solid
θL
θR ZA
ZB
DR
3 4
Flaws
D = DR/2 60°
3 4
(60°)
3 4
Figure 12. Schematic showing 50 MHz surface wave arrival from a superalloy steel at defoci of ZA and ZB or water paths of
Z2 − ZA and Z2 − ZB . The entry circle ray geometry producing these signals is shown with two different-sized flaws, L and S.
The angular spectrum changes with the material. Figure point metals (such as gallium and mercury), mineral or
11 shows three schematics indicating the critical angle fluorocarbon oils, gases under high pressure, and liquid
behaviour for the water–material interface for each of the gases, depending on the frequency of the acoustic beam,
materials. Five general conclusions can be drawn from a the material velocities, chemical reactivity, and temperature
consideration of Snell’s law, table 2, and figure 10 and respectively. Regardless of how the beam is focused, the
figure 11 with respect to the interaction of focused beams indices of refraction of the coupling fluid to test material
with fluid–solid interfaces. control the imaging process.
The highest acoustic velocity measured to date is the
(1) Transducers with foci that include longitudinal,
extentional velocity in a highly crystalline carbon whisker
shear, or surface wave critical angles will produce aberrated
(19.3 mm ms−1 ). However, for solids not in whisker
foci.
form, at temperatures near ambient (22 ◦ C) the longitudinal
(2) Beyond the first critical angle 2L , longitudinal
acoustic velocities are bounded above by the [111]
waves do not penetrate into the test material.
crystallographic direction in diamond (18.6 mm ms−1 ) and
(3) Beyond the second critical angle 2S , shear waves
below by the [100] direction in caesium (approximately
do not penetrate into the test material.
1.0 mm ms−1 ). Shear or transverse velocities are also
(4) Beyond the Rayleigh critical angle 2R (surface
bounded by the same materials, the maximum being in
waves), no acoustic waves penetrate the test medium
the [100] direction in diamond (12.7 mm µs−1 ) and
or interact with the water–material interface, i.e. total
the minimum being in the [111] direction in caesium
reflection occurs for angles of incidence greater than 2R .
(approximately 0.4 mm ms−1 ). Depending on Poisson’s
(5) Surface waves cannot be generated by mode
ratio for the material, the surface wave propagates at 0.87
conversion when the surface wave velocity is slower than
to 0.95 times the velocity of the transverse wave. These
the longitudinal velocity in the fluid coupling medium.
data and the acoustic velocities of other materials have been
(This can be avoided by the use of a lower velocity fluid).
compiled by several authors [35–37].
The coupling fluid typically used in imaging is Eleven fine-grained homogeneous isotropic polycrys-
water, but studies have been done using low melting talline compacts are tabulated in table 2. Isotropic com-
1398
Table 2. Acoustic properties of example materials.
2L 2S 2R
Density CL CS CR (degrees (degrees (degrees
Materials (gm cm−3 ) (mm ms−1 ) (mm ms−1 ) (mm ms−1 ) of arc) of arc) of arc)
Solids
Aluminum ‘2.71 ‘6.37 ‘3.11 ‘2.89 13.4 28.4 31.0
Diamond compact ‘3.51 17.05 11.51 ‘9.90 ‘5.5 ‘6.8 ‘8.6
Diamond [100]* ‘3.51 17.63 12.81 11.10 ‘4.8 ‘6.6 ‘7.7
Diamond [111]* ‘3.51 18.65 12.00 10.80 ‘4.6 ‘7.1 ‘7.9
Beryllium ‘1.85 13.05 ‘8.97 ‘7.89 ‘6.5 ‘9.5 10.8
Caesium† ‘1.87 ‘1.10 ‘0.59 ‘0.55 ** ** **
Copper ‘8.90 ‘4.76 ‘2.40 ‘2.20 18.1 38.1 42.3
SiO2 (fused) ‘2.20 ‘5.97 ‘3.75 ‘3.40 14.4 23.3 25.8
Si [100]* ‘2.33 ‘8.90 ‘5.84 ‘4.91 ‘9.6 14.7 17.5
[111]* ‘2.33 ‘9.37 ‘5.12 ‘4.70 ‘9.1 16.6 18.3
SiN ‘3.25 11.15 ‘6.16 ‘5.54 ‘7.6 13.9 15.5
Steel ‘7.89 ‘5.90 ‘3.28 ‘2.96 14.5 26.8 30.0
Al2 0 (sintered) ‘3.96 10.8 ‘6.35 ‘5.82 ‘7.9 13.5 14.7
Sapphire [001]* ‘3.99 11.22 ‘6.10 ‘5.50 ‘7.6 14.0 15.6
Silver (Ag) 10.50 ‘3.70 ‘1.69 ‘1.57 23.6 61.1 70.5
AgCl ‘5.59 ‘3.13 ‘1.20 ‘1.14 28.2 ** **
Polystyrene ‘1.10 ‘2.35 ‘1.09 ‘1.05 39.0 ** **
Plexiglass ‘1.23 ‘2.67 ‘1.12 ‘0.95 33.7 ** **
Lead (Pb) 11.32 ‘2.23 ‘0.86 ‘0.70 41.6 ** **
W 19.2 ‘5.20 ‘2.90 ‘2.70 16.5 30.7 33.2
WC (4% Co) 16.5 ‘7.02 ‘4.10 ‘3.90 12.2 21.2 22.3
Liquids (22 ◦ C)
Gallium (Ga) 5.9 2.90
Water 1.0 1.48
SAE 20 Oil 0.9 1.45
Freon-4 1.6 0.52
* Longitudinal shear and surface waves are dependent on both direction and polarization in single crystal and
other anisotropic materials.
† Caesium (like sodium and lithium) explodes in water.
** Propagation modes in materials with velocities slower than water cannot be produced by mode conversion,
and do not produce critical angles.
pacts have been chosen as example materials because they Systems that record only the maximum amplitude
are the easiest to describe and understand. These materi- pulse in the signal gate usually use gated peak detectors.
als essentially cover the entire range of acoustic velocity Additional circuitry can detect the phase and arrival time of
behavior for isotropic solids. Having discussed diamond, the maximum signal in the gate. Full waveform acquisition
beryllium (Be) is the highest velocity metal, and lead (Pb) requires digitizers with sampling frequencies of at least
the lowest velocity material that is encountered in practical twice the highest carrier frequency component of the pulses.
situations such as materials joining (soldering). Once full waveforms are acquired, they can be processed
Acoustic velocities in liquids have also been compiled to extract the information in any other acquisition method.
by several investigators [16, 38]. The highest velocity is The price one pays is in the storage media required to
liquid gallium at about 30 ◦ C and the lowest is freon-4, one acquire and store the additional information on the signal.
of the heavier fluorocarbon fluids. Water and SAE 20 motor Amplitude alone is typically acquired at 8 bits per pixel,
oil are similar in velocity. and the position at which it is stored on the computer disk
corresponds to its spatial position in the image. Amplitude
4.1. Type of data to be acquired and phase, or amplitude and time of flight, would require
at least two 8-bit words per ultrasonic pulse. Multiple gate
Ultrasonic imaging systems usually fall into one of the acquisitions in turn require one or more 8-bit words for each
following four categories, with respect to data acquisition: gate, each ultrasonic pulse. If full waveforms are acquired,
(1) single or multiple gates are used to acquire peak then at least one 8-bit word is acquired for each sample,
detected amplitudes; and often hundreds of samples are acquired for each pixel
(2) single or multiple gates are used to acquire phase or image position.
and/or amplitude and phase; For example, a 1024 × 1024 amplitude image can be
(3) a single signal gate may be used to acquire contained in 1 Mb of computer memory; 2 Mb are required
amplitude and time of flight; for amplitude and phase or amplitude and time of flight
(4) fully sampled waveforms can be acquired. images. Finally, 256 Mb are needed if 256 samples are
1399
R S Gilmore
Figure 14. 50 MHz surface wave image of a Cu98 –Ag2

Figure 13. 20 MHz surface wave image of a alloy sample 30 mm across. Note the intragranular
100 mm × 100 mm titanium sample showing a fully structure.
annealed grain structure.
will be developed first and then the same principles applied
acquired for each waveform/pixel. to V (z). Liang et al [17] and Briggs [26] have published
V (z) images in their most fundamental form are simply extensive discussions on tone burst surface wave imaging or
scanned amplitude measurements of the phase interference V (z). This discussion will therefore concentrate on time-
between the direct surface reflection and the surface wave. resolved surface wave imaging, because that is the type
They may be scanned at constant height to form an image most used in industrial NDE.
or scanned at various heights (z). Either produces multiple Consider the steps required to acquire and display an
interferences between the surface wave and the direct acoustic image of the surface and near the surface of a
reflection. sample with an acoustic imaging system such as shown in
figures 1 and 2.
4.2. Imaging surfaces (1) The sample is selected, its surface checked for finish
and polished if necessary, and it is mounted on a scanning
Since the discovery of V (z) in the mid 1970s, imaging stage.
surfaces and near surface conditions have become (2) A transducer (size, frequency, and F /d) is selected
synonymous with V (z) imaging. V (z) can only occur on the basis of the surface wave velocity in the sample and
when multiwavelength pulses are used, and the direct the resolution required for the image. For normal incidence,
surface reflection overlaps the surface wave arrival in surface wave imaging the cone of focus must include
time. Figure 12 shows these arrivals time-resolved with the surface wave critical angle. For a high-resolution
broadband signals. Because the two signals are necessarily image, a pixel size/spacing is chosen of 1/6 of the surface
of the same frequency, their phase interference is directly wavelength, and the appropriate scan plan is loaded into
controlled by the path length and velocity of the surface the scan controller.
wave over the surface of the sample. As will be shown, this (3) Depending on system capabilities, data for the
is in turn controlled by the height of the transducer above image are acquired as signal amplitude, amplitude and
the sample surface and the critical angle for the surface phase, time of flight from a reference, or fully sampled
wave mode conversion. Because both time-resolved surface waveforms, which can be processed to give all of the
wave imaging and V (z) imaging are controlled by the same preceeding information.
wave–material interaction physics, the broadband methods (4) A colour or greyscale display is selected and
1400
DT Lens
(1) (2)
C1
DIRECT
REFLECTION
DE Water l2
θR
Z2 C2
SURFACE
WAVE
C3
ZA Z3
FOCAL
GEOMETRY
εZ IN SUBSTRATE (3)
3
(1) Z2
εX
2 and 3
(2)
SIGNALS
GATE FOCAL
GEOMETRY
SURFACE WAVE IN WATER (2)
AMPLITUDE X ε2
2
BROAD BAND
SIGNALS (1)
NARROW BAND (2)

SIGNALS
Figure 16. Schematic showing the geometry for a
diffraction limited beam, Z /d > C3 /C2 . The figure
demonstrates that the beam in the solid refracts to the
same diameter as in the fluid although the depth of the
focus is shortened.
V(z ) (1) + (2)
SIGNALS
GATE
INTERFERENCE AMPLITUDE
Figure 15. Fundamentals of V (z ) (surface wave) imaging.
the dynamic range of the acquired/processed data is gain

adjusted to coincide with that of the display and the image
is displayed.
Figure 10 shows that when the full angular spectrum
of the material is subtended by the cone of focus, then
subsurface information from the longitudinal and shear
components may be included in a surface wave image.
This, however, seldom presents a serious problem. Figure 17. The nickel-based superalloy sample shown has
Often the relative amplitude of the pixels displaying a a slant layer of 250 µm (0.010 in) alumina spheres sintered
into it. The layer increases in depth from the surface, upper
subsurface feature will indicate the nature of the energy left, to 12.7 mm (0.5 in) deep lower right. The sample entry
forming the image. Because of the pulse transmission surface is 47 mm × 46 mm (1.875 in × 1.750 in), the
nature of surface wave imaging, a region in the image sample depth is 25.4 mm (1.0 in).
1401
R S Gilmore
Figure 19. Superalloy sample imaged with a 15 MHz,

(a )
F /D = 6 transducer focused at 8.0 mm depth. Equations
(8) and (9) give EX = 0.6 mm and EZ = 8.0 mm. Although
the ability to separate and image the spheres is lower, they
still produce easily detected signals.
containing lower amplitudes than normally produced by

sound material often indicates that a superficial flaw is
blocking the surface path for the wave. Conversely,
regions containing higher amplitudes often indicate that
longitudinal or shear wave echos from subsurface features
are positively interfering with the surface wave signal.
Features with low-amplitude edges around a high amplitude
centre can indicate a near surface feature initially blocking
surface wave paths within the entry circle but then reflecting
both longitudinal and shear energy as those much smaller
focal zones begin to interact with the feature.
(b ) The time domain photographs in figure 12 show two
surface wave pulses at two different path lengths (ZA
and ZB ) in the isotropic material tungsten carbide. In
forming the surface wave arrival, each ray proceeds from
the piezoelectric element (1) to the lens (2). Following
specifically those rays which are refracted at the Rayleigh
critical angle 2R , they proceed in a converging truncated
cone of rays to intersect the specimen surface (3) where
they intersect it to form the entry circle shown in the
lower part of the figure. Those rays at and around 2R
then mode convert to surface waves and propagates on a
circle diameter to the far side (4), where they reconvert and
proceed back to the lens (5) and the piezoelectric element
(6). For a buffer rod of length L, a lens of radius RC ,
focusing at an axial distance Z2 , where it is defocused to
axial distances ZA and ZB ; the time t1 required for the
directly reflected pulse following the ray path on the lens
(c ) axis of symmetry to travel from the piezoelectric element
and return for Z is
Figure 18. The three image segments of nickel-based
superalloy sample in figure 17 were made with the three 50 2L 2(Z2 − ZA )
MHz transducers indicated. Equations (8) and (9) were t1 = + (10)
used to determine diameters EX and depths EZ of foci.(a ) C1 C2
Z /d = 2.0, focused at 1.5 mm, EX = 0.075 mm
EZ = 0.5 mm, (b ) Z /d = 3.0, focused at 8.0 mm depth, where C1 and C2 are the velocity in the lens and coupling
EX = 0.115 mm EZ = 1.0 mm, (c ) Z /d = 4.0, focused at liquid respectively. The round-trip travel time for the pulse
12.5 mm depth, EX = 0.150 mm EZ = 2.0 mm travelling along the ray incident at the Rayleigh critical
1402
Figure 20. Table of sizes and image format for US Air Force resolving power test target 1951 as provided by
Teledyne-Gurley Inc., Troy, NY.
Figure 21. Acoustic image of a block with multiple

flat-bottomed holes (FBHs). There are four 3 × 3 arrays of
holes spaced at the hole diameter, and a single hole at the
block centre. Each array has nine holes with diameters of
400, 800, 1200, and 1600 microns respectively. The single
hole is 400 microns. Note that the 400 microns holes are
detected but not resolved (separated) by the 20 MHz,
F /7.0 acoustic beam which is 500 microns in diameter.
Figure 22. Schematic illustration of the procedure used to
angle (figure 12) is produce the subsurface target pattern in single crystal
sapphire blanks 1.5 mm and 3.0 mm thick. (a ) coated
2L 2(1 + cos 2L )RC
t2 = + substrate; (b ) selective exposure through a USAF 1951
C1 C1 mask; (c ) ion beam etching of substrate and patterned
2(Z2 − ZA ) − 2RC (1 − cos 2L ) 2ZA tan 2R photoresist; (d ) etched photoresist and sapphire; and
+ + (e ) bonding of etched and unetched sapphire
C2 cos 2R CR
(11) blanks.(Courtesy of Rodel and Glaeser [40].)
where 2L , 2R , and CR are the lens angle, the Rayleigh

critical angle, and the Rayleigh velocity respectively. From direct reflection 1t1 and the surface wave 1t2 at the two
figure 12 and Snell’s law, the diameter DR of the entry defocus distances ZA and ZB and combining equations
circle, where the incident pulse in the liquid is mode (10)–(12) and solving for CR gives
converted into a leaky Rayleigh wave and 2R are s
C2 C22
DR = 2ZA tan 2R and 2R = sin−1 . (12) CR = . (13)
CR 1 − (1t2 /1t1 )2
Writing the differences in round-trip travel time for the Equation (12) permits the Rayleigh group velocity
1403
R S Gilmore
(a )
(a )
(b )
(b )
Figure 23. (a ) Magnified image of groups 2, 3, and 4 as
described in figure 20. The 25 µm pixels are now visible as
is the loss of resolution as the line pairs approach
32 lines/mm. (b ) Image of the same area as in figure 23(a )
but undersampled in two directions by a factor of two. The
image now displays group 3 patterns with the same relative (c )
resolution as figure 23(a ) shows for group 4. Figure 24. (a ) 50 MHz, F /3.0 image of the sapphire
resolution target taken from the 1.5 mm surface to pattern
depth. The flaws evident in the image were actually very
to be determined from time measurements at two water useful. One of the smallest was used to define the point
path settings where both the velocity in the water and spread beam function used to provide the image
the material are determined. However, for temperature- enhancement shown in figure 24(c ). (b ) Magnified image of
the central portion of figure 24(a ). (c ) Wiener filter
stabilized water baths, C2 is known and invariant. enhancement of figure 24(b ) showing approximately a
Therefore, for a constant water velocity, a flat sample, and factor of two improvement in resolution.
an invariant waterpath during scanning, the time delay of
the surface wave arrival behind the direct reflection t1 − t2
can give the surface wave velocity directly. the curvature of the entry surface circle. Entry circles
The entry circle schematic (figure 12) shows that any of three to four surface wavelengths are required to time
discontinuities on the surface that interrupt or change resolve the leaky Rayleigh wave from the direct reflection.
direction of any of the converging bundle of surface wave But, unlike subsurface scanning, this does not obscure any
rays changes the amplitude received by the piezoelectric portion of the entry surface material. In isotropic materials
element. The mechanism by which this occurs can be the surface wave originates at an entry circle and provides a
clarified by considering the two opposing 60◦ segments 360◦ cylindrically convergent, pulse transmission, surface
of that entry circle, and again to consider only those rays wave inspection method that can detect surface features
propagating from left to right. The 60◦ segment depicts a with equal probability regardless of the direction of their
converging ray bundle, point focused at the centre of the surface strike. Utilizing broadband pulses with center
entry circle, and then diverging to the opposite side of the frequencies at 1.0 MHz extends the depth of interrogation
circle perimeter. What is in fact displayed is a surface of the broadband systems to three surface wavelengths,
wave transmission acoustic microscope that is focused by or 9.0 mm in steels and equivalent velocity media. This
1404
surprising depth is due to the low-frequency components specifies the energy lost back into the coupling fluid for the
of the broadband pulse. range of frequencies that make up the ultrasonic pulse.
Two parameters describe the resolution of a time- Equations (11) and (12) show that DR is specified by
resolved surface wave imaging system. These are the the ratio of the surface wave velocity to the longitudinal
diameter of the entry circle DR and the spot size of the velocity in the coupling medium (usually water) and
surface wave focal zone or crossing zone of focus at the defocus distance Z. Written in terms of C2 , CR , and Z,
centre of the circle. The spot diameter ER can be calculated DR becomes
from the amplitude point spread function of the double C2
cylindrical lens system placed on the entry surface by the DR = 2Z q . (15)
entry circle. The point spread function produced by a CR − C22
2
cylindrical lens has been shown by Born and Wolf [6] and
Kino [7] to be of the form (sin X/X)2 , where X has the At constant Z, high values for CR result in smaller
same definition as equation (7) and figure 8, for a pulse– entry–exit paths DR ; lower CR , on the other hand, requires
echo system. Gilmore et al [12] used these results to larger Rayleigh angles to generate the surface wave and
calculate the −3 dB and −1 dB diameters for the point result in longer entry–exit paths.
spread function for the crossing zone at the entry circle When the surface wave velocity varies in the direction
centre: of travel in the plane of the surface, such as for images of
ER = KR λR (14) anisotropic grains, then the travel time along each entry–
where KR equals 0.32 and 0.16 respectively. The exit path around the entry circle is different, i.e. the entry
surface wave entry circle taken as 180◦ transmitter/receiver figure is no longer a circle. In this case the received
segments may be treated as F /0.5 lenses, which produce amplitude results from the sum of all of the arrivals back
very sharply focused spot sizes. The surprisingly small at the transducer in both amplitude and phase. In the case
−3 dB diameter (0.32) of the centre focus explains the of an image of a polycrystalline material, if the maximum,
consistently high resolution of mode converted surface minimum, and average velocities for each of the grains on
wave images that involve cylindrically convergent surface the surface can be uniquely related to the orientation of
waves. the surface with respect to the crystal axes of the grain,
The general rules for broadband surface wave imaging then the received amplitude (and therefore the greyscale
may be summarized from inspection of figure 12. The at which each grain is displayed) can be related to the
entry circle defined by DR must be at least three surface crystallographic orientation. The interaction between 1717
wavelengths in diameter in order to time resolve the surface anisotropic grains and single-frequency signals was first
wave arrival from the direct surface reflection. Therefore, considered by Somekh et al [21].
DR is typically nine times the −3 dB diameter of the Figures 13 and 14 show images of fully annealed
focus at its centre, which is approximately three times the titanium and copper alloys, specifically Ti83 –Al5 –Sn2 –Zr2 –
−20 dB size (1.0λR ). Cracks, seams, and other linear Mo4 –Cr4 and Cu99.85 –Ag0.15 . Each titanium grain shows
discontinuities having little width but with lengths that a consistent shade of grey and therefore a consistent
exceed DR are imaged as having a width that approaches amplitude. From grain to grain, however, the titanium
that circular diameter. Features with dimensions smaller signals vary over a range of 1–64 (six bits). The Cu–
than DR produce detectable changes in amplitude only Ag sample shows less grain-to-grain contrast (five bits) but
when they interact with the crossing zone of focus at the more intragranular variation. If the surface wave arrivals
centre. They will, however, produce some decrease in shown in figure 12 were produced by a tone burst generator
amplitude as soon as they become included in the entry instead of a half wavelength impulse, then, depending on
circle. the length of the resulting pulse, the surface wave arrival
In specifying a surface wave scan, half of the −3 dB
would be overlapped by the direct reflection from the
spot diameter, 0.16, should be used for the line-to-line
surface and the two would interfere. When the surface
spacing and the pulse-to-pulse spacing along the line. This
wave is delayed an even number of half-wavelengths, the
will limit the amplitude ripple in the scanned acoustic field
interference will be positive and the two signals add. When
and support the −3 dB resolution in the image. For 50 MHz
it is delayed an odd number of half wavelengths, the
surface wave images in most materials, this spacing is
0.02 mm or less. interference is destructive and the two signals subtract.
The difference in amplitude of the surface wave signals Raising and lowering the transducer increases and decreases
produced by the two water path distances (Z2 − ZA ) the surface wave pathlength regardless of the shape of the
and (Z2 − ZB ) in figure 12 is inversely proportional to entry figure or the anisotropy of the material being imaged
the distance each pulse travels across the surface of the (figure 15). The best contrast is achieved in a V (z) image
tungsten carbide. A small amount of this amplitude loss when the transducer is placed at a height midway between
is due to material attenuation, but most of the drop in a minimum and a maximum where the slope of the received
amplitude is due to the continuous radiation of elastic interference amplitude is greatest. Adjusting a signal gate to
energy, characteristic of leaky Rayleigh waves, back into a position in time where the arrivals overlap, and scanning
the water during propagation. The received amplitude at the described height, will produce the greatest net change
decreases proportionately with the length of the entry–exit in the contrast due to any velocity changes in the sample
path. Therefore, when path length DR is specified, this also or flaws blocking the surface wave arrival.
1405
R S Gilmore
Figure 25. Applications of ultrasonic imaging/microscopy versus frequency.
4.3. Subsurface imaging of volumes into the scan controller.

(4) Focus the transducer at the depth of interest Z3
Consider the steps required to acquire and display an
where Z3 = (C2 /C3 )(Z2 − `2 ) as shown in figure 16.
acoustic image of a subsurface region in a sample.
Acquire data for the image in the form of signal amplitude,
(1) Select and mount the sample on the scanning stage. amplitude and phase, time of flight from a reference,
(2) Select a transducer (diameter, frequency, and focal or fully sampled waveforms, the latter of which can be
length) on the basis of (a) the range in depth to be imaged, processed to give all of the preceeding information.
(b) the index of refraction between the sample and the fluid (5) Select a colour or greyscale display, gain adjust the
coupling medium, and (c) the resolution required. dynamic range of the acquired/processed data to coincide
(3) Based on the beam diameter (resolution) produced with that of the display, and display the image.
by the transducer, select a pixel size and spacing (of no
more than half that beam diameter), and load a scan plan Volume imaging also requires that an acoustic field
1406
Figure 26. 2.25 and 5.0 MHz images of a 2.0 in × 6.0 in × 3.3 in (50.8 × 152.4 × 83.8 mm) reactor pressure vessel
calibration block. The 5.0 MHz and 2.25 MHz beam diameters are 1.2 mm and 3.1 mm (0.048 in and 0.125 in) respectively.
The 0.62 in (1.5 mm) side-drilled holes are easily detected in both images.
be scanned to a uniform amplitude throughout the length, be evaluated with respect to the detectable flaw size by a
width, and depth of the inspected material. For this to suitable calibration target. In this case fracture mechanics
be accomplished in an economical number of scans, the and probabilistic analysis must be used either to approve
interrogating beam must have a depth of focus (equation or disapprove the test with respect to its noise-limited
(9)) that is, if not equal to, at least an appreciable fraction detection capability.
of the material depth. For many industrial parts this could The first concern of an ultrasonic inspector, however, is
be 1 to 10 cm or considerably greater. Typically the to make sure that the critical flaw is detected. If a reflecting
image is scanned to a uniformity of 1 dB, using the 1 dB flaw is smaller than the focused beam, then the reflected
beam diameter (equation (8)) to establish the pixel spacing. amplitude is in proportion to the ratio of the area of the
However, because it is not unusual for the dynamic range flaw to the area of the beam [34]. To a first approximation,
of the data acquisition to be 40+ dB (eight bits is 48 dB), the shape of the flaw cross section may be ignored if it is
for many industrial inspections, the −3 dB diameter and planar, perpendicular to, and totally contained in the beam
depth of focus may be used. For industrial NDE, the flaw area. Therefore, this analysis will discuss flat flaws that
size to be detected is the key parameter that establishes the are located on the central axis of the interrogating beam at
sensitivity for the nondestructive evaluation of a volume some time during the scan.
of material. This flaw (figure 16) must block a sufficient One method of determining the beam diameter required
fraction of the focused beam that a detectable signal is for an inspection is to determine the flaw diameter required
reflected back to the transducer. The detectable size can to block it totally. A circular reflecting flaw in the shape of
be expressed as a fraction of the beam area, or as a fraction a disk of diameter Df , perpendicular to and centred on the
of the reflector size used to calibrate the system. Once axis of symmetry of an acoustic beam focused at distance
detected, a flaw may be characterized with respect to its size Z in a material of longitudinal velocity C3 producing a
and shape by imaging with a much higher resolution beam. wavelength λ3 , by a lens of diameter d, will totally block
For small flaws there may be no economical solution. A the beam if the back-reflected beam half-angle from the
beam diameter small enough to detect the flaws may require disk is equal to the half-angle of convergence of the focused
a considerable length of time to complete a scan of the beam. Taking the sine as equal to the angle and setting the
material volume. half-beam angle [12, 34] for a flat circular reflecting flaw as
In cases of substantial coherent acoustic noise from the equal to the half-angle of convergence of the focused beam
focal zone, an alternative would be to specify the detectable gives
1.2λ3 dC3 Z
flaw in terms of the ratio of its reflected amplitude to that = or Df = 2.4λ2 . (16)
of the noise. The detectable amplitude is usually given as Df 2zC2 d
two to five times the mean acoustic noise, depending on Note that equation (16) has the same form as equation (8)
the permissible false alarm ratio. This amplitude can then with K = 2.4.
1407
R S Gilmore
Figure 27. Positioning the focus for successful flaw detection. Three scanning depths with respect to an array of sixteen
0.015 65 in (0.4 mm) flat-bottomed holes at 2.0 in (50.8 mm) depth in ring forged Ti–6Al–4V. The 5.0 MHz beam is 0.096 in
(2.4 mm). Ultrasonic amplitude is plotted vertically. Note improved signal-to-noise ratio as the focal zone approaches the
same depth as the holes.Horizontal distances are in mm.
While the insonification by the focused beam across 0.25 mm in diameter. The layer goes from the top surface
the face of disk Df does not have the uniformity that is on the left side of the sample to 12.7 mm depth over its
assumed in the derivations, experimental measurements on 50 mm diameter, giving it an angle of approximately 15
flat-bottomed holes show that equation (12) does give a degrees of arc with respect to the plane of that surface.
reasonable blocking diameter. Similar measurements show The longitudinal velocity in the nickel-based super alloy is
that smaller flaws are usually detectable and that for data 6.1 mm ms−1 .
acquisition systems with at least 32 dB dynamic range, The images shown in figure 18 were made with
flaws with reflecting areas equal to 1/20th of the beam area three 50 MHz transducers focused at Z/d values of two,
can be detected when they reflect echos that are at least three, and four respectively. These transducers produce
twice the acoustic noise in the material. progressively larger lateral (−3 dB) beam diameters of 0.75,
With the critical flaw size assigned, and expressions for 0.115, and 0.150 mm respectively, and hence poorer lateral
the blocking flaw size Df and the diameter and depth of resolution. They also produce greater (−3 dB) depths of
focus for the acoustic beam, a scanning plan can now be field in proportion to the square of their Z/d ratio, giving
established for an inspection volume. Figure 17 shows such 0.5, 1.0, and 2.0 mm respectively. Even the Z/d = 4
a representative volume. This is a hot isostatically pressed transducer, however, can focus over only a fraction of the
and sintered sample with a slant layer of alumina spheres, 12.7 mm depth over which the layer ranges.
1408
Figure 28. Four 10 MHz pulse–echo images of four

100 mm × 100 mm composite samples. Each sample is (a )
constructed with two 0.025 in (0.635 mm) woven ply of
Kevlar, five ply of unidirectional carbon (0.005 in/0.127 mm
each), and an additional two ply of Kevlar. A good
structure is shown in photo (a ); Teflon delaminations are
shown in photos (b ) and (d ); and backing paper is shown
in photo (c ). Beam diameter is 0.75 mm (0.030 in),
512 × 512 displays.
The image shown in figure 19 was acquired with a

15 MHz, Z/d = 7 transducer producing a depth of field
in the sample of 8 mm. This transducer is able to produce
detectable signals for two thirds of the entire depth of the
slant layer. Note that this increase in detection depth has
cost substantially in lateral resolution. The beam diameter
is now 0.6 mm, or approximately 2.4 times the diameter of
the target spheres.
4.4. Measuring resolution in scanned images

The resolution inherent to and obtainable from scanned
acoustic images is basically determined by the beam
diameter and by the scanning increment or pixel size. (b )
The effect of the pixel size on the image resolution is
summarized by Nyquist’s theorem, which states that in Figure 29. Five megahertz pulse–transmission images are
shown of two 100 mm × 100 mm areas in a six-ply
order to support the spatial resolution of the beam it must be carbon–epoxy composite. The good sample shows uniform
spatially sampled at less than half that dimension. In other transmission (a ); the dark region in (b ) is due to porosity.
words, to support the −3 dB beam diameter resolution the Beam diameter 1.2 mm (0.48 in), pixel size 0.2 mm
pixels that make up the image must be less than half that (0.008 in), 512 × 512 display.
size.
As shown in figure 10, subsurface foci in high- imaging system. Targets such as the US Air Force 1951
velocity substrates always contain significant refractive target (figure 20) were first used by Gilmore et al [12]
aberration. In addition, the foci are also subject to micro- to measure the resolution in acoustic images. It is also
aberrations due to grain-to-grain anisotropy. The grain-to- possible to machine flat-bottomed hole targets that provide
grain anisotropy, however, is the mechanism that permits this information (figure 21).
acoustic waves, and particularly surface waves to image the Optically transparent resolution targets make it feasible
microstructure. Surface wave images of microstructure are to verify visually that the resolution target has been
simply micro-aberration images. correctly fabricated. An image of the target with a
The most straightforward method to determine the candidate ultrasonic transducer then permits quantitative
resolution produced in a high-velocity solid is to scan a image resolution estimates to be made even when the
resolution target fabricated from the same or a velocity interrogating acoustic beam contains significant refractive
matched material. A number of resolution targets can aberration. This is important because useful subsurface
provide this type of calibration information for an ultrasonic images can be acquired with the use of acoustic beams
1409
R S Gilmore
(a ) (a )
(b )
Figure 30. Silicate flaw in a 1.2 in (31.2 mm) circular (b )
trepan from a Ni–Cr–Mo–V turbine rotor forging. (a ) The
15 MHz beam diameter is 0.024 in (0.6 mm); (b ) the
50 MHz beam diameter is 0.010 in (0.25 mm). The cross is
a cursor on the CRT display, 512 × 512 images. (Courtesy
of L B Burnet, General Electric Company.)
Figure 31. 50 MHz image of an electrical resistance

heated solid state weld in a titanium alloy. The ultrasonic
beam diameter is 0.0024 in (0.063 mm). The surfaces to
be attached were deliberately contaminated by an artificial
fingerprint (made with lanolin and NaCl) (The weld is
50 mm wide, imaged at 1024 × 256 pixels). (Courtesy of R
Sundell, General Electric Company.) (c )
Figure 32. Four 12.7 mm diameter (0.5 in) BaTiO3
capacitor blanks imaged at 50 MHz. The lower left blank is
that are aberrated to the point that diffraction-limited beam zoomed to show high detail, 20 µm beam and pixel
width calculations are meaningless. diameter. (a ) 1024 × 1024 image; (c ) 256 × 256 image.
Military Standard 150-A describes the basic pattern
which has become known as the ‘USAF-1951’ or simply the
‘Air Force’ target. Each element consists of two patterns change in pattern size is based on the sixth root of two, i.e.
oriented at right angles to each other, each containing three for every six target elements, the number of line spaces per
lines and two spaces. The line and spacing width are equal; millimetre doubles. Each six-target-element set is known
the line length is equal to five times the line width. The as a group, and the group number (0, 1, 2, 3, and 4 for the
1410
Figure 35. Complex metal circuitry in and at the surface of

a ceramic chip carrier. 50 MHz direct reflection
1024 × 1024 image, 20 µm beam diameter, 10 µm pixels.
Figure 33. Surface of a Canadian penny imaged with a
The larger black features at the centre are irregularities in
20 µm diameter 50 MHz beam. The 2048 × 2048 image
the manufacture of the device.
file permits a series of zoomed images to give highly
magnified surface detail, 10 µm pixels.
similar acoustic longitudinal and shear velocities assures
that the refractive aberration in the transparent target will
be similar to that of the opaque materials under study.
Sapphire can also be used to match many structural
ceramics, such as polycrystalline SiN, Al2 O3 , BeO, MgO,
and SiC. Lead borosilicate glass produces targets that match
the zirconium alloys and other lower velocity materials
integrated circuit applications. Fused quartz is a good
velocity match to most structural steels, including those
used for high-temperature turbine engine discs and blades,
and for titanium alloys. It is also a good velocity match
to the oxide-loaded glasses used in computer chip carrier
applications.
The development of standards for establishing the
sensitivity and resolution of all ultrasonic NDE inspection
methods has proven to be an ongoing problem for the
testing industry. The traditional sensitivity standard is
a flat-bottomed hole (FBH), first proposed in 1959 [34].
Drilled into materials that are identical in composition
and in microstructure to those being inspected, the FBH
provides a reflector of known circular (and therefore
mathematically definable) scattering cross section.
Since, as a first-order approximation, pulse–echo
amplitude is linearly proportional to the area of a compact
flat reflecting void perpendicular to the acoustic axis of the
Figure 34. Deformation processes are clearly displayed by interrogating beam, the FBH area establishes the reflecting
first lapping a modern copper-plated zinc coin flat and then area that can be detected, and hence the test sensitivity.
imaging the subsurface microstructure at 50 MHz. Beam However, an ultrasonic image of a single hole can only
diameter and pixel diameter are both 20 µm, image is
1024 × 1024.
show that it was detected; no resolution information is
supplied. Another drawback is that the circular bottom
of the hole is always accompanied by a cylindrical shaft
targets used in this paper) is the power of two, to which the connecting it to the drilled surface, and therefore it is
first element in the group is raised to express the number of surface connected and only approximates the buried circular
lines per millimetre. Therefore, the numbers 0 to 4 for these void/crack it is intended to represent.
targets correspond to 1, 2, 4, 8, and 16 lines per millimetre, In addition, for materials with scattering microstruc-
as shown in figure 20. tures (i.e. materials with grains, or more than one mate-
Fabricating the targets from transparent materials with rial, such as composites) the echos from the bottoms of
1411
R S Gilmore
(a ) (b )
(a )
(c ) (d )
Figure 36. (a ) Silicon wafer, 1.5 in (38 mm) diameter,
0.020 in (0.5 mm) thick. (b ) 50 MHz image of a badly
cracked guard ring under an uncracked silicon entry surface
in (a ), also the 37.5 mm diameter thyristor gate array in
that device. (c ) 50 MHz surface wave image of a 25 mm
[100] silicon wafer fragment containing surface flaws and
(d ) a zoomed image of the surface wave structure around
defect caused by [100] anisotropy in silicon.
small holes add algebraically to the grain boundary reflec-

tions, producing significant echo amplitude variations from
holes of precisely the same size. Therefore, single-reflector
calibration blocks can overcalibrate or undercalibrate, de-
pending on the location of the FBH in the material mi- (b )
crostructure. To meet these problems and to try to solve Figure 37. 100 MHz pulse transmission image acquired
the sensitivity/resolution dilemma, standards have been de- with a scanning laser acoustic microscope, showing
veloped with multiple hole patterns (figure 20). These hole metalized ribbon leads on an Al2 O3 ceramic substrate. The
bonded areas (circled) are 125 µm2 . (Courtesy of
patterns can determine the variance in amplitude and res-
L Kessler, Sonoscan.)
olution produced by scattering materials and are easy to
make for materials that can be drilled, such as titanium,
steel and zirconium. However, ceramics are difficult to
drill and holes only a few microns in diameter, such as are
required to calibrate inspections for small voids in structural of the Nyquist theorem applied to spatial resolution. Stated
ceramics, are especially difficult. simply: in order to resolve dimension d, image data must
Lithography has furnished us with a new tool for be acquired at a sampling interval, less than or approaching
studying and for creating subsurface void arrays with d/2. Applying the Nyquist theorem to the 25 m acoustic
very precise geometries. These arrays may be produced beam, used to acquire the data file for figure 7(a) and figure
in high-temperature structural ceramics, in many glasses, 7(b), indicates that the micron pixels do not support the
and at ceramic–metal interfaces. The combination of 25 m resolution provided by the beam. To support the
photolithographic methods with ion beam etching and hot beam the pixel size/spacing should be less than 12.5 m.
pressing provides the ability first to produce surface features Figure 23(b) shows the data file in figure 23(a)
with highly precise geometries and locations and then to undersampled by two in each direction, increasing the pixel
transform these features into internal features without losing
size to 50 m. Note that it is now not possible to resolve
this detail. The methods summarized in figure 22 are
even the largest pattern in group 4 of the resolution target.
reported in greater detail elsewhere [40, 41]. This target
contains groups 0 through 4. However, looking at the next largest group, it can be
Figure 23(a) is a magnified image of the target shown observed that patterns in group 3 (upper right in the figure
in figure 20. The 0.025 mm beam and pixels resolve the 23(b)) show the same resolution as for group 4 in figure
largest bar–space patterns in group 4 (16 lines/mm), but 23(a). Since the spatial resolution due to pixel size/spacing
the smallest patterns in the group (28.51 lines/mm) are not is decreased by a factor of two, the displayed decrease in
resolved. This observation provides one more verification resolution is expected.
1412
(a )
(b )
Figure 39. 2.7 GHz V (z ) image of the structure of a FET
transistor. Note that this image resolution is very close to
the best optical microscopes. (Courtesy of C Quate,
Stanford.)
limitations imposed by the beam diameter and the pixel

size. The purpose of this section is to demonstrate that
digital image processing methods may be used to provide
improvements of as much as a factor of two in resolution
[43, 44] providing that the spatial resolution at which the
original data was acquired supports the final/ improved
resolution in the processed image. The details for digital
image processing are extensive [45, 46]. This discussion is
limited to Wiener filtering.
Figure 24(a) shows a 1024 × 1024 longitudinal wave
image of the 25.4 mm × 25.4 mm sapphire target described
Figure 38. Three images of a small Inconel sample: (a ) above. The 50 MHz, F /3.0 transducer used to acquire the
and (c ) are an optical and an acoustic image of the image used a 6.37 mm diameter lens focusing the beam at
unetched surface respectively; (b ) is an optical image of 19.05 mm in water. This produces a half-angle of incidence
the surface etched. The 2.7 GHz V (z ) surface wave of 9.5◦ , well beyond the 7.6◦ critical angle of incidence for
acoustic image has a −6 dB beam diameter of 2 µm. (b ) longitudinal waves in water incident on a C-axis sapphire
and (c ) principally show grain structure. (Courtesy of
C Quate, Stanford.) substrate (table 1). The longitudinal focus produced in
sapphire is similar to that shown for steel in figure 10
(between 0 and 14.5◦ ). Clearly aberration, not diffraction,
4.5. Image processing
controls the resolution in this image.
Clearly, like most other imaging techniques, ultrasonic For the sapphire target used to produce the acoustic
images are degraded by blurring due to the practical images in figure 24, vacuum hot pressing (1370 ◦ C, 15 MPa,
1413
R S Gilmore
2.6 mPa) was used to bond a 1.5 mm thick unetched • Evaluation of attachment integrity of welds and
single crystal [001] sapphire slab to a 3.0 mm thick etched composite structures, as a process development and process
slab of the same [001] orientation. This produced the control tool as well as a quality assurance tool.
USAF-1951 target embedded in a volume of single crystal • Inspection of electronic devices and materials.
sapphire. The target contains bar–space patterns 0, 1, 2, 3, Most industrial inspection and medical diagnostic
and 4, as described in figure 20 with an entry-surface-to- imaging is done at ultrasonic frequencies from 1.0 to
target-pattern depth of either 2.0 or 4.0 mm. Referring to 10.0 MHz. In this frequency range, metal and composite
figure 20, these range from 1 to 28.5 lines per millimetre. sections multiple inches in thickness may be penetrated, and
Note that in addition to providing the geometrical precision equivalent depths of penetration are achieved in medical
required for image calibration these void arrays are truly imaging. Almost all in vivo (in living) medical imaging is
buried flaws surrounded by intact solid material. They done in this frequency range.
scatter sound like the subsurface cracks they are. Calibration blocks are the key to a successful and well
A display of the raw data, as shown in figure 24, is designed inspection; therefore, they should be carefully
able to resolve only the first two patterns in group 2 (a evaluated before using them. Figure 26(a) shows a reactor
little better than 4.5 lines/mm or 220 m). The resolution pressure vessel calibration block. The upper surface of the
inherent to the high spatial frequency of the image (25 m block is clad by 0.3 in (8.0 mm) of roll-bonded stainless
pixel size), however, can still be utilized. Figure 24(b) and steel cladding, and four 0.062 in (1.5 mm) side-drilled holes
figure 24(c) zoom in on the outlined portion of figure 24(a) are located at three positions in the clad and just below the
and then enhance the resolution using a Wiener filter image cladding-to-basemetal interface. Figure 26(b) and figure
processing technique [42, 43]. Note that in figure 24(c) 26(c) show 2.25 MHz and 5.0 MHz pulse–echo images
the patterns resolved in group 3 suggest that the processed of the cladding and clad-to-steel interface. The inspection
image shows twice the resolution (is able to resolve line– frequency, 2.25 MHz, clearly shows all four flaws but
space combinations one half those) of the unprocessed shows large edge effects at the block ends. The higher
image. resolution provided at 5.0 MHz gives much more detail
at the cladding interface and is better for characterizing
5. Applications and qualifying the calibration block. Both images can be
acquired by most industrial C-scan inspection systems.
In considering the applications for acoustic microscopy, Positioning the focus for successful flaw detection and
it is important to remember that any industrial ultrasonic determining the effect of the material on flaw signals are
inspection system capable of producing magnified C-scan important for inspection development. Figure 27 shows
images can be used as an acoustic microscope. Therefore, three transducer scanning depths with respect to an array
although a complete applications summary of ultrasonic of sixteen 0.015 65 in (0.4 mm) flat-bottomed holes at 2.0 in
imaging/microscopy is beyond the scope of this work, (50.8 mm) depth in forged Ti–6Al–4V. The titanium block
figure 25 indicates what that range could be. was machined from a large forged ring. The transducer is
Since resolution in an ultrasonic image is dependent 5.0 MHz, focused to a −6 dB beam of 0.096 in (2.4 mm).
on both frequency and focal convergence, it is difficult to For this demonstration the amplitudes reflected by the holes
separate applications as a function of frequency without are plotted vertically rather than as a C-scan image. Note
some overlap. that as the focal zone of the transducer approaches the same
If most industrial ultrasonic systems can be used as depth as the holes, the signal-to-noise ratio of each of the
ultrasonic microscopes, it follows that most ultrasonic hole echos becomes much greater, even though the absolute
microscopes can be used to inspect parts, providing the amplitude is the same range (0 to 200 arbitrary units) for all
parts are small enough to be scanned. three scan depths. When the focal zone is in the same plane
Although few ultrasonic microscopes are constructed as the flaws, the enhanced signal-to-noise is produced by
large enough to inspect large industrial parts, many are large decreasing the noise with respect to the hole echos rather
enough to do inspection development for heavy sections. than enhancing the hole echos. The high microstructure
In these studies the transducer focus and bandwidth are noise in the first two plots clearly shows the curvature of
selected for the inspection, as well as the scan index. In the ring forging.
addition, the calibration standards for the inspection may Selecting the correct transducer, frequency, and focus
be qualified more precisely than would be possible with an for composite inspection requires careful studies of the
industrial inspection system. flaws that would be expected during manufacture. Figure
The broad area applications chosen for this discussion 28 shows four 10 MHz pulse–echo images of four
consist of the following. composite samples, each with dimensions of 4.0 in × 4.0 in
(101.6 mm × 101.6 mm). Three of the samples have
• Evaluation and qualification of calibration blocks intentionally seeded flaws. Each sample consists of two
at higher resolution and frequency than the inspection 0.025 in (0.635 mm) woven plys of Kevlar, five ply of
frequency. unidirectional carbon (0.005 in/0.127 mm each), and then
• Inspection development studies in metals and an additional two plys of Kevlar (same thickness). The
composites. total composite thickness is 0.125 in (3.1 mm). By gating
• High-resolution/frequency characterization of flaws each depth in the sample, the structural segments may be
detected by lower frequency inspections. imaged separately. Figure 28(a) shows the first 0.050 in
1414
(1.27 mm) of Kevlar to be without flaws; (b) shows a the displays in figure 33(b), 33(c) and 33(d ). The modern
delamination between the second and third carbon ply; (c) US penny is made with copper-plated zinc. When such a
shows a backing paper flaw between the fourth and fifth coin is struck, the zinc body sustains many microfractures
carbon ply; and (d ) shows a Teflon-induced delamination such as shown in the 50 MHz image in figure 34(b). This
between the bottom two Kevlar ply. Note that even though deformation pattern was imaged by first lapping the raised
the 10 MHz image has a depth of focus that is greater coin surface flat and then focusing the 20 µm beam just
than the composite thickness, the images do degrade in below the surface.
resolution because of attenuation of the higher frequencies Electronic devices and assemblies have been one of the
in the broadband signal. A number of commercial systems largest benefactors of the development of high-frequency
now marketed permit up to eight images to be acquired ultrasonic imaging. Figure 35 shows the complex metal
simultaneously at increasing depths. circuitry in and at the surface of a ceramic chip carrier. The
Figure 29 shows 5 MHz pulse–transmission images of pulse–echo image was acquired at 50 MHz with a 20 µm
two six-ply carbon-fibre-reinforced epoxy samples. Sample beam diameter. The larger black features at the centre are
(a) good; the sample on the right (b) shows considerable irregularities in the manufacture of the device.
porosity, which attenuates the 5 MHz amplitude to produce Many silicon-controlled power devices require wafers
a darker image. several inches in diameter such as shown in figure 36(a).
As the frequency increases above 10 MHz, resolution Figure 36(b) shows a 50 MHz image of a badly cracked
increases, but attenuation increases also. In the range from guard ring under an uncracked 0.5 mm [100] silicon entry
10 to 50 MHz, medical applications are pretty much limited surface. The image also shows the finger structure of the
to in vitro (excised tissue) studies except for measurements 37.5 mm diameter thyristor gate array.
on the eye and studies of the structure of burns. However, Figure 36(c) shows a surface wave image of a [100]
there are many industrial applications in this frequency silicon wafer fragment that contained several small surface
range. flaws. Note the complex surface wave structure figure
Consider the role of ultrasonic microscopy in evaluating 35(d ) around such a defect caused by the anisotropy of
flaws trepanned from large steam turbine rotor forgings. the silicon wafer.
Figure 30 shows (a) 15 MHz and (b) 50 MHz images of a Most of the images displayed so far have been pulse–
flaw that was trepanned from a Ni–Cr–Mo–V medium alloy echo and acquired with mechanically scanned systems.
turbine rotor. The flaw was detected by a visual pulse–echo Figure 37 shows a pulse transmission image acquired with
inspection conducted at 2.25 MHz. Serial sectioning of the a 100 MHz stage in a scanning laser acoustic microscope.
largest segment of the flaw showed a silicate inclusion. The The image shows metalized ribbon leads on an Al2 O3
composition of the inclusion suggested that it resulted from ceramic substrate. The bonded areas (one is circled) are
the liner of either the furnace or the ladle when the steel was 125 × 125 microns.
poured. Note the 50 MHz image shows an improvement in Both metals and electronic assemblies can be studied
both resolution and detection. in the gigahertz frequency range. Figure 38 shows three
Figure 31 shows a 50 MHz image of a (electrical images of a small Inconel sample. Figure 38(a) and figure
resistance heated) solid state weld in a Ti–6Al–4V alloy. 38(c) show an optical and an acoustic image of an unetched
The ultrasonic beam diameter is 0.0024 in (0.063 mm). The surface respectively. The 2.7 GHz surface wave acoustic
surfaces to be attached were deliberately contaminated by image has a −6 dB beam diameter of 2 µm. By comparing
an artificial fingerprint (made with lanolin and NaCl). Note features in the acoustic image of figure 37(c) and the etched
that in the vicinity of this artificial flaw the microstructure optical image in figure 37(b), it is apparent that both display
shows a much narrower heat-effected zone. This is to the very fine-grain structure in Inconel. Figure 37 was
be expected since the contamination also decreased the acquired with the mechanically scanned V (z) surface wave
electrical conductivity at the weld inter-face. imaging technique.
Porosity in BaTiO3 capacitor blanks is both common The detail at which an integrated circuit can be imaged
and acceptable, as indicated by the four 12.7 mm (0.5 in) at 2.7 GHz is shown by imaging the structure of a FET
blanks displayed in the 50 MHz image in figure 32(a). transistor (figure 39). Note that the resolution of this image
The upper left blank, however, shows some connectivity is indeed very close to that of the best optical microscopes.
between the pores, and that may lead to cracking. Analysis
of the zoomed images shown in (b) and (c) suggests that the 6. Conclusions and future work
pores are all smaller than 100 µm (0.004 in). A number of
commercial image processing programs can pan and zoom Several definitive acoustic microscopy and acoustic
across large images. An example of this capability is shown imaging works have been referenced that have very
in figure 33. Here a 2048 × 2048 image is acquired of complete bibliographies—the works of Lemons and Quate
the reverse face of a 1973 Canadian penny by monitoring [16], Bennett [23], Kessler [24], Khuri-Yakub [25] and
the amplitude of a ultrasonic beam focused directly on the Briggs [26] are five. Briggs has the most recent and
surface. The ultrasonic beam used to acquire these 50 MHz complete bibliography but has written from the aspect
images is approximately 20 µm (0.0008 in) in diameter, of materials characterization rather than nondestructive
and the image is acquired with a pixel size of 10 µm. testing. Lemons and Quate and the papers edited by Bennett
This very high resolution image can then be zoomed for summarize the development of the instrumentation and
an additional magnification of as much as 8×, as shown in techniques as of 1979 and 1985 respectively. All of these
1415
R S Gilmore
reviews assume that the reader is well versed in the details [7] Korpel A, Adler R, Desmares P and Watson W 1966 A
of acquiring and displaying mechanically scanned images. television display using acoustic deflection and
This review has used these details as an outline to organize modulation of coherent light Appl. Opt. 5 1667–75
[8] El-Sum H and Larmore L 1967 Proc. 1st Int. Symp. on
the presentations for surface and subsurface imaging and Acoustic Holography (later Acoustic Imaging) ed A
is written specifically from the aspect of nondestructive Metherell (New York: Plenum)
inspection. The Acoustic Imaging Symposia, currently at [9] Lemons R A and Quate C F 1973 Acoustic microscopy by
volume 24 [46], provide an informative annual review of mechanical scanning Appl. Phys. Lett. 24 165–7
[10] Tsai C S, Wang S K and Lee C C 1977 Visualization of
the current state-of-the-art on ultrasonic/acoustic imaging.
solid material joints using a transmission acoustic
In addition, no discussion of ultrasonics or acoustics would microscope Appl. Phys. Lett. 31 791–3
be well referenced without a careful review of the Physical [11] Buchanan R W and Hastings C H 1955 Ultrasonic flaw
Acoustics series [38]; many of these volumes are edited by plotting equipment: a new concept in industrial
Warren P Mason. inspection Nondestruct. Test. 15 17–25
[12] Gilmore R S, Tam K C, Young J D and Howard D R 1986
Future progress in ultrasonic/acoustic imaging and Acoustic microscopy from 10 to 100 MHz for industrial
microscopy will almost certainly be driven by the dramatic applications Phil. Trans. R. Soc. A 320 215–35
growth in computers, with respect to both availability [13] Nikoonahad M Yue and Ash E 1985 IEEE Trans. Sonics
and power, for image acquisition, processing and analysis. Ultrason. Special Issue on Acoustic Microscopy 32
Image data based on waveform capture is memory 130–375
[14] Weglein R D and Wilson R G 1978 Characteristic materials
intensive, but permits the operators developed for ultrasonic signatures by acoustic microscopy Electron. Lett. 14
spectroscopy to be used in analysis of the data as well 352–4
as the image processing operators such as Wiener filters [15] Weglein R D Acoustic micro-metrology IEEE Trans. on
and fast Fourier transforms. It seems reasonable to Sonics and Ultrasonics 32 225–35
[16] Lemons R A and Quate C F 1979 Acoustic microscopy
suggest that all of the analytic procedures for V (z), as
Physical Acoustics vol 14, ed W P Mason and R N
published by Weglein and Wilson [14, 15], Liang et al Thurston (New York: Academic) pp 2–90
[17], Kushibiki and Chubachi [19] and Somekh et al [21] [17] Liang K K, Kino G S and Kuri-Yakub B T 1986 Material
have corresponding operators which can be developed to characterization by the inversion of V (z) IEEE Trans. on
extract the amplitude–frequency information in broadband Sonics and Ultrasonics, Special Issue on Acoustic
Microscopy 32 213–24
ultrasonic images. Applicable work in ultrasonic spectral [18] Bertoni H L 1984 Ray-optical evaluation of V (z) in the
analysis, such as summarized by Fitting and Adler [39] can reflection acoustic microscope IEEE Trans. on Sonics
also be applied, extracting the multifrequency information and Ultrasonics 31 105–16
from broadband ultrasonic/acoustic images. [19] Kushibiki J and Chubachi N 1985 Material characterization
The lately re-emerging air-coupled acoustic imaging by the line-focus-beam acoustic microscope IEEE Trans.
on Sonics and Ultrasonics, Special Issue on Acoustic
methods such as originated by Wickramasinghe and Petts Microscopy 32 130–375
[47], developed by Fortunko et al [48, 49], and still later [20] Wichramasinghe K 1979 Contrast and imaging
developed by Bond and others [50] promise to provide a performance in the scanning acoustic microscope
considerable body of information in the oncoming decade J. Appl. Phys. 50 664–72
[21] Somekh M, Briggs G A D and Ilett C 1984 The effect of
on plastics and other low-velocity materials. Low-velocity
elastic anisotropy on contrast in the scanning acoustic
fluids such as the freons also promise to provide similar microscope Phil. Mag. 49A 179–204
opportunities. [22] Ash E (ed) 1980 Scanned Image Microscopy (New York:
Academic)
[23] Bennett S (ed) 1985 IEEE Trans. on Sonics and
Acknowledgments Ultrasonics, Special Issue on Acoustic Microscopy 32
130–375
[24] Kessler L 1989 Acoustic microscopy Nondestructive
The author wishes to acknowledge M Gigliotti, P Howard, Evaluation and Quality Control Metals Handbook
R Klaassen, K Mitchell, L Perocchi, R Trzaskos, E Nieters, (Materials Park, OH: ASM International)
J Young, and most of the authors cited in references for [25] Khuri-Yakub B 1992 Scanning acoustic microscopy
many helpful discussions over the years. Ultrasonics 31 361–72
[26] Briggs A 1992 Acoustic Microscopy (New York: Oxford
University Press)
[27] Martins Y and Ash E Photodisplacement microscopy, using
References a semiconductor laser Electron. Lett. 18 763–4
[28] Balk L and Domnik M 1993 Quantitative scanning electron
[1] Bergmann L 1938 Ultrasonics (and Their Scientific and microscopy of silicon Scan. Microsc. 7 37–48
Technical Applications) (New York: Wiley) [29] Cargill G S 1980 Electron acoustic microscopy Scanned
[2] Carlin B 1949 Ultrasonics (New York: McGraw-Hill) Image Microscopy (New York: Academic) pp 319–31
[3] McMaster R C (ed) 1959 Nondestructive Testing Handbook [30] Monk G S 1937 Principles of Light (New York:
vol II (New York: Ronald) McGraw-Hill)
[4] Sokolov S 1936 USSR patent 49 [31] O’Neil H 1949 Theory of focusing radiators J. Acoustic
[5] Dunn F and Fry W 1959 Ultrasonic absorbtion microscope Soc. Am. 21 516
J. Acoustic Soc. Am. 31 632–33 [32] Born M and Wolf E 1980 Principles of Optics 6th edn
[6] Jacobs J, Berger H and Collis W 1963 An investigation of (London: Pergamon)
the limitations to the maximum attainable sensitivity in [33] Kino G S 1980 Fundamentals of scanning systems Scanned
acoustical image converters IEEE Trans. Ultrason. 10 Image Microscopy ed E A Ash (London: Academic)
83–8 [34] Krautkramer J and Krautkramer H 1990 Ultrasonic Testing
1416
of Materials (Berlin: Springer) pp 75–9 [43] Howard P and Gilmore R 1994 Ultrasonic C-scan images
[35] Simmons G and Wang H 1971 Single Crystal Elastic for hard-alpha flaw detection and characterization
Constants and Calculated Aggregate Properties: A Review of Progress In Quantitative NDE vol 13, ed D O
Handbook (Cambridge, MA: University Press) Thompson and D E Chimenti (New York: Plenum)
[36] Chung D and Buessem W 1968 The elastic anisotropy of [44] Jain A K 1989 Fundamentals of Digital Image Processing
crystals Anisotropy in Single-Crystal Refractory (Englewood Cliffs, NJ: Prentice Hall)
Compounds vol 2, ed F Vahldiek and S Mersol (New [45] Lim J S 1990 Two Dimensional Signal and Image
York: Plenum) Processing (Englewood Cliffs, NJ: Prentice Hall)
[37] Anderson O and Liebermann R 1968 Sound velocities in [46] 1967–1994 Acoustic Imaging vol 1–24 Various editors
rocks and minerals: experimental methods and (New York: Plenum)
extrapolation to very high pressures Physical Acoustics [47] Wickramasinghe K and Petts C 1980 Acoustic microscopy
IVb ed W Mason (New York: Academic) in high pressure gases Proc. IEEE Ultrasonics Symp. ed
[38] Mason W (ed) 1990–1994 Physical Acoustics I–XX ed W B McAvoy (New York: IEEE) pp 668–70
Mason (New York: Academic) [48] Fortunko C, Renken M and Murray A 1991 Examination
[39] Fitting D W and Adler L 1981 Ultrasonic Spectral Analysis of objects made of wood Proc. IEEE Ultrasonics Symp.
for Nondestructive Evaluation (New York: Plenum) ed B McAvoy (New York: IEEE) pp 1099–103
[40] Rodel J and Glaeser A M 1989 Photolithography: a new [49] Fortunko C, Dube W and McColskey J 1993 Gas-coupled
tool for ceramic science Mater. Res. Soc. Symp. Proc. acoustic microscopy in the pulse-echo mode Proc. IEEE
155 293–306 Ultrasonics Symp. ed B McAvoy (New York: IEEE)
[41] Rodel J and Glaeser A M 1987 Production of [50] Bond L, Chang C H, Fortunko C and McKolskey J 1992
controlled-morphology pore arrays: implications and Absorption of sound in gases between 10 and 25 MHz:
opportunities J. Am. Ceram. Soc. 70 c172–c176 Argon Proc. IEEE Ultrasonics Symp. ed B McAvoy
[42] Mitchell K W and Gilmore R S 1992 A true Wiener filter (New York: IEEE) pp 1069–73
implementation for improving the signal to noise in
acoustic images Review of Progress in Quantitative NDE
vol 11, ed D O Thompson and D E Chimenti (New
York: Plenum)
1417

Detekcija Ostecenja

Uploaded by

Copyright:

Available Formats

Detekcija Ostecenja

Uploaded by

Document Information

Original Description:

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

Detekcija Ostecenja

Uploaded by

Copyright:

Available Formats

J. Phys. D: Appl. Phys. 29 (1996) 1389–1417.

Industrial ultrasonic imaging and

Received 12 March 1996

• diameter of the acoustic beam or its point spread function (PSF);

The beam diameter, or PSF, is controlled by the frequency of the ultrasonic

z entire reflected waveforms from a region in the sample.

Table 1. Acoustic imaging and materials evaluation: a timeline.

1963 Jacobs adds electron multiplier to Sokolov Tube [6].

† Panametrics Inc., 221 Crescent Street, Waltham, MA 01754, USA.

many systems, these images were binary (either black or SAPPHIRE

acoustic imaging and acoustic microscopy with vastly im- Y

In 1973, work at Stanford University under the

Figure 8. Schematic showing the point spread function for

sin 23 = 1.0 and

0.6 5 MHz 15.0

.06 5 1.5 8.6˚

4. Role of imaged material: permitted resolution

Most scanned acoustic imaging systems, including acoustic

Table 2. Acoustic properties of example materials.

Figure 14. 50 MHz surface wave image of a Cu98 –Ag2

NARROW BAND (2)

Figure 15. Fundamentals of V (z ) (surface wave) imaging.

the dynamic range of the acquired/processed data is gain

Figure 19. Superalloy sample imaged with a 15 MHz,

containing lower amplitudes than normally produced by

Figure 21. Acoustic image of a block with multiple

where 2L , 2R , and CR are the lens angle, the Rayleigh

Figure 25. Applications of ultrasonic imaging/microscopy versus frequency.

4.3. Subsurface imaging of volumes into the scan controller.

Figure 28. Four 10 MHz pulse–echo images of four

The image shown in figure 19 was acquired with a

4.4. Measuring resolution in scanned images

Figure 31. 50 MHz image of an electrical resistance

Figure 35. Complex metal circuitry in and at the surface of

small holes add algebraically to the grain boundary reflec-

limitations imposed by the beam diameter and the pixel

You might also like