Professional Documents
Culture Documents
Abstract The influence that microstructure has on high frequency (3070 MHz)
ultrasonic wave propagation is being investigated. In this work the ultrasound sys-
tem, wave propagation simulation results and experimental results for highly struc-
tured porous aluminium samples are presented. Finite element modeling shows that
a single pseudo S0 lamb wave propagates in the porous aluminium with a velocity of
5.18 km/s. Experimentally a velocity of 5.10 km/s was measured. For angled mea-
surements the porous aluminium samples were observed to normalise the incident
waveforms allowing significant energy to be transmitted beyond angles of incidence
of 40 , well past the solid aluminium P and S critical angles of 14 and 29 .
1 Introduction
A. Dawson
School of Chemical & Physical Sciences, Victoria University of Wellington, P.O. Box 600,
Wellington, New Zealand, e-mail: dawsonand@hotmail.com
P. Harris
Industrial Research Ltd., P.O. Box 31-310, Lower Hutt, New Zealand, e-mail: P.Harris@irl.cri.nz
G. Gouws
School of Chemical & Physical Sciences, Victoria University of Wellington, P.O. Box 600,
Wellington, New Zealand, e-mail: Gideon.Gouws@vuw.ac.nz
S.C. Mukhopadhyay, G.S. Gupta (eds.), Smart Sensors and Sensing Technology, 221
c Springer-Verlag Berlin Heidelberg 2008
222 A. Dawson et al.
is also the property used in anodising, historically in making humidity sensors and
even in the formation of some nanostructures.
In this work broadband high frequency (3070 MHz) longitudinal probing waves
were chosen to examine collinear porous aluminium samples since these are signifi-
cantly influenced by the sample microstructure [2]. In order to carry out the investi-
gation the construction of a suitable ultrasound system for material characterisation
was completed. Finite element modelling of wave propagation in immersed porous
aluminium was performed and confirmed by experiment on two types of porous
aluminium samples.
2 Experimental System
X,Y,Z micrometers
PC
Panametrics V390
25 75MHZ
Needle Hydrophone
The transmitter circuit [3] shown in Fig. 2 is a bridge circuit with an input
matched to the excitation impedance, and a 2:1 output transformer to match to the
low transducer impedance. For an excitation burst amplitude of 2.5Vp and op-amp
gain of 8.2 an output power of 1W is delivered to the transducer. The frequency
response of the circuit is shown in Fig. 3. With a source power of 45 dBm a power
gain of 12 dB is observed with a 3 dB bandwidth of 100 MHz to accommodate the
broad bandwidth of the transducer.
The receiver circuit [3] schematic is shown in Fig. 4. Two AD8337 variable gain
amplifiers (VGAs) are used in series forming the front end of the circuit. The VGAs
possess very low noise, wide bandwidth and a linear-in-dB gain i.e. a ramp with a
slope equal to the attenuation (dB/mm) achieves signal compensation. An adjustable
DC input to an op-amp integrator, with an analog switch for integrator reset, pro-
vides the ramp generator. The function generator sync signal is used to control the
reset function. The VGAs consist of a configurable pre-amp, here set for a gain of 3,
followed by a 24 dB attenuator stage (controlled via the gain pin), and a fixed 18 dB
gain amplifier. The output cable driver used to drive the transmission cable to the
oscilloscope is configured for a gain of 3. This results in a variable gain range for
the circuit from 17 to 65 dB. The summing blocks seen in Fig. 4 control the VGAs
gain. They are configured so that once the first VGA reaches its maximum gain, the
second VGA gain begins to increase. In this way both amplifiers can be dedicated to
DC gain if no time-gain compensation is required, all time-gain compensation, or a
combination of the two [3]. The receiver circuit frequency response from minimum
to maximum gain is shown in Fig. 5.
Each of the traces displays an excellent frequency response satisfying the trans-
ducer requirements up to 100 MHz with a steep roll o at higher frequencies. The
224 A. Dawson et al.
noise characteristic is analysed in Fig. 6. The lower trace shows the noise of the
spectrum analyser, and the higher trace shows the noise of the receiver circuit plus
the spectrum analyzer noise when configured for maximum gain. Based on the dif-
ference between the traces the noise of the receiver at maximum gain is approxi-
mately 36 dBm equating to 2nV/ Hz i.e. the receiver noise is dominated by the
VGA input noise.
Fig. 5 Frequency response of receiver output power for five dierent gains
Two types of porous aluminium samples were used in the experiments. One type was
prepared by Industrial Research Limited (IRL) [5], the second type was obtained
commercially.
Figure 7 shows the SEM pictures of an IRL sample snapped to reveal the highly
collinear (pores from top to bottom) porous structure. The insert picture shows a
plan view of the sample highlighting the regular pore spacing. The 30 nm sized
pores are spaced 60 nm apart. The thickness of the sample used was 160 um and
12 mm in diameter. The IRL samples were found to be relatively robust and able to
be handled comfortably using standard tweezers.
Figure 8 shows the SEM pictures of the commercial sample. The main SEM pic-
ture shows these samples are relatively poorly collinear with many irregularities.
This is easily observed in the insert SEM with many pores not connecting from top
to bottom and with varying pore wall thickness. The pores are not regularly spaced
and large in size at around 200 nm. These samples were only 60 um thick and found
to be very brittle, fracturing easily during handling.
Fig. 7 Cross section and plan view of IRL porous aluminium samples
High Frequency Ultrasonic Wave Propagation 227
Fig. 8 Cross section and plan view of commercial porous aluminium samples
Finite element analysis (using PZFlex) was used to model wave propagation in the
IRL sample microstructure. The simulation was performed in 2D with a material
model comprising of 1) an axially aligned water filled porous layer containing 1000
pores representing the aluminium oxide, and 2) a homogenous block of solid alu-
minium. The dimensions and material properties are shown in Fig. 9. The FEA mesh
size is usually calculated using the minimum velocity in the simulation i.e. the water
velocity 1500 m/s, divided by the maximum possible frequency (100 MHz) result-
ing in a spatial resolution of 15 um. Here, a mesh size of 15 nm (1/1000H2O ) was
used in order to resolve the 30 nm pores. The temporal resolution was calculated us-
ing the spatial resolution of 15 nm divided by the maximum velocity i.e. the P wave
speed in aluminium 6 km/s, resulting in a sampling interval requirement of 2.5 ps;
we used 2 ps.
The excitation is a 50 MHz Blackman pulse applied over the width of a pillar at
the bottom left corner of the aluminium oxide. The bottom boundary in the model is
symmetrical, the left boundary free, and both the top and right boundaries absorbing.
The modeling results can be observed in Fig. 10, which shows three simulation
snapshots (top) 20 ns, (center) 40 ns and (bottom) 60 ns after the excitation pulse was
injected. Large concentrations of energy are represented by darker shades. During
228 A. Dawson et al.
the simulation it is clear from the top and center snapshots that the wave travels as
a single compact wavefront in the porous aluminium being guided by the material
pores. This compact wavefront, a lamb wave (or plate wave), propagates when the
insonifying wavelength is comparable to the thickness of the plates or pore walls
in which it propagates [1, 6]. Lamb wave velocity can vary significantly as various
modes are possible. Here the wave propagates as a pseudo lamb mode with a sim-
ulated velocity of 5.18 km/s. The frequency-thickness product for the aluminium
oxide suggests the wave travels as a fundamental symmetric mode (S0 ) with a theo-
retical velocity given by
1
VS2 2
VS O = 2VS 1 2 (1)
VL
Using (1) a theoretical VS0 of 5.20 km/s is calculated. An S0 lamb mode is predom-
inantly a longitudinal displacement and is therefore little aected by the presence of
the water. Whilst in the porous aluminium the bulk of the energy is observed to be
concentrated to a small region of pores as the material acts as a waveguide. How-
ever, the close proximity of adjoining pillars facilitates the transfer of some energy
which forms an associated wake as can be observed in the top and center snap-
shots. Upon reaching the solid aluminium interface, the compact wavefront ceases
to exist as, no longer guided by the material pores, the P and S components are
free to travel at their respective velocities. This is shown in the bottom snapshot
of Fig. 10, where the individual waves spread spherically away from the material
interface.
The simulation was repeated with pore sizes of 200 and 100 nm spacing, ap-
proximating the dimensions of the commercial samples. The results observed were
High Frequency Ultrasonic Wave Propagation 229
Fig. 10 Simulation (top) 20 ns, (center) 40 ns, and (bottom) 60 ns after beginning
identical to those shown for the IRL sample dimensions. Since the frequency-
thickness product is still very small (much less than 1) this result is expected.
5 Experimental Results
Experimental work has been performed on both porous aluminium samples as well
as solid aluminium samples to confirm the results obtained from the simulation.
With the transducer and hydrophone vertically aligned and immersed in water, sam-
ples of known thicknesses were inserted between them (see Fig. 1). Measurements
230 A. Dawson et al.
6.5
Solid P
6
Soild S
IRL
5.5 Comm
Velocity (km/s)
4.5
3.5
3
0 5 10 15 20 25 30 35 40 45
Angle (degrees)
Fig. 11 Wave velocity versus transducer angle of incidence for the samples
were made for transducer angles of incidence from normal to 40 for each sam-
ple. Using the time dierence between transmitted and reflected waveforms from
the oscilloscope, and measured sample thicknesses, propagation velocities can be
calculated. The resulting velocity measurements are show in Fig. 11.
For the solid aluminium a P wave velocity of 6.3 km/s was measured up to the P
critical angle of 14 , and an S wave velocity of 3.1 km/s up to the S critical angle of
29 . These values are in agreement with standard tables. Beyond the critical angle
the transmitted wave travels along the surface of the water/sample interface and is
not detected in these measurements.
For the IRL sample only a single compact waveform was received up to angles
greater than 40 with an almost constant velocity of 5.10 km/s. This velocity agrees
well with those determined in the simulation and calculated theoretically confirming
the propagation of an S0 lamb wave in the IRL sample.
The commercial sample results are considerably dierent. Only a single wave-
front was received, however the velocity for low angles at 5.7 km/s drops signif-
icantly with angle. Given the poor co-linearity this variation in velocity could be
expected as the velocity will depend on the wave path traveled.
Figure 12 shows the influence angle has on the amplitude measured. In the solid
aluminium the P wave shows a decrease in amplitude out to the P critical angle,
whereas the S wave amplitude stays relatively constant out to its critical angle. The
IRL sample wave amplitude initially decreases and then is constant out to 40 ; the
amplitude seems to follow that of the P and S amplitudes of the solid aluminium.
This is not the case for the commercial sample as its amplitude is constant, seem-
ingly tracking the shear component only.
High Frequency Ultrasonic Wave Propagation 231
350
300 Solid P
Solid S
250 IRL
Comm
Amplitude (mV)
200
100
50
0
0 5 10 15 20 25 30 35 40 45
Angle (degrees)
Fig. 12 Wave amplitude versus transducer angle of incidence for the samples
It should be noted that the significant amplitudes measured for the IRL and
commercial samples beyond the second critical angle potentially convey important
information.
6 Conclusions
References
1. Cheeke D (2002) Fundamentals and Applications of Ultrasonic Waves, CRC Press, Boca
Raton
2. Dawson A, Harris P, Gouws G (2007) High Frequency Ultrasound Wave Propagation in
Anisotropic Materials. In: Proc. 2nd International Conference on Sensor Technology 2007
ICST2007, Palmerston North, New Zealand, pp 552556
3. Dawson A, Harris P, Gouws G (2006) Design and Evaluation of Ultrasonic Transducer Cir-
cuits for Material Characterisation. In: Proc. 13th Electronics New Zealand Conference 2006
ENZCon06, Christchurch, New Zealand, pp 261265
4. Harris P, Dawson A, Young R, Lecarpentier F (2007) High Frequency Propagation in
Structured Solids. In: Proc. 2007 IEEE International Ultrasonics Symposium, N.Y, USA,
pp 690693
5. Kirchner A, MacKenzie K, Brown I, Kemmitt T, Bowden M (2007) Structural characterization
of heat-treated anodic alumina membranes prepared using a simplified fabrication process.
J Membrane Sci 287:264270
6. Rose JL (1999) Ultrasonic Waves in Solid Media, Cambridge University Press, Cambridge