Вы находитесь на странице: 1из 8

ELSEVIER

Ultrasonics 34 (1996) 1-8

Near-field ultrasonic Rayleigh waves from a laser line source


P.A. Doyle, C.M. Scala *
DSTO Aeronautical and Maritime Research Laboratory, Z?O.Box 4331, Melbourne 3001, Australia

Received 28 August 1995

Abstract

The generation of Rayleigh waves by a thermoelastic laser line source was studied to establish a quantitative basis for using this
source in nondestructive testing and in materials characterization. Experiments using interferometric detection were carried out
which showed that, for the near field of the line source, the Rayleigh pulse is a monopolar depression of uniform amplitude. A
theoretical model was developed which led to an approximate expression in closed form for the normal surface displacement of
the Rayleigh wave. Quantitative agreement was found for a parameter-free comparison between theory and experiment for the
Rayleigh wave characteristics in both time and frequency domain.
Keywords:

Laser ultrasonics; Laser line source; Ultrasonic Rayleigh waves; Ultrasonic surface waves; Nondestructive

1. Introduction
The considerable advantages of using lasers rather
than more conventional methods for the generation and
detection of ultrasound are increasingly being demonstrated, both in nondestructive evaluation (NDE) [l]
and in the study of materials properties [ 21. One feature
of laser ultrasonics which has b$en exploited recently is
to use a cylindrical lens to focus the incident laser beam
into a thermoelastic line source of Rayleigh waves propagating on a specimen surface. This technique permits an
improved signal/noise ratio for measured surface waveforms compared with that for a circularly symmetric
source, because more energy can be injected into the
specimen while keeping the energy density low enough
to avoid surface damage [ 11. In addition, the resulting
wavefield is highly concentrated in the direction normal
to the line source [ 31, which is advantageous for surface
crack sizing [4] and for materials characterization
[S--7]. Developments based on the line source include
tone bursts of Rayleigh waves produced by an array of
line sources [S], and continuous surface waves generated
by scanning the laser beam [9].
While the laser line source for Rayleigh-wave generation offers great potential for use in quantitative NDE,
* Corresponding author.
Free of copyright for Commonwealth of Australia purposes.
SSDZ 0041-624X(95)00089-5

evaluation

its application to date has been restricted by limited


knowledge of source characteristics. Preliminary observations on Rayleigh waves generated by a laser line
source in the thermoelastic regime were made by Edwards
et al. [lo] as part of their investigation of laser interference gratings for surface wave generation. They used
a thin line source of length 30 mm and placed a detector
approximately 28 mm along the axis perpendicular to
the line source. A Michelson interferometer was used for
detection, allowing undistorted measurement of normal
surface displacement. The observed Rayleigh waveform
was a sharp, monopolar depression of the specimen
surface; the measured displacement returned quickly to
background noise level after the passage of the Rayleigh
pulse. This behaviour contrasts with the more familiar
Rayleigh wave produced by a circularly symmetric
thermoelastic source, which is bipolar, followed by a
more gradual return to zero [l,ll,lZ].
Edwards et al.
[lo] presented a brief, qualitative explanation of the
monopolar nature of the Rayleigh pulse, in terms of
differentiating the impulse response for forces distributed
along a line [ 133. Recently, Berthelot [14] exploited
half-order derivatives to show that, within some approximations, the Rayleigh wave from a thin, infinitely long
line source is indeed a monopolar depression, which
reproduces the time dependence of the source. However,
for the line-source Rayleigh wave to be exploited fully
in quantitative NDE measurements, a parameter-free
comparison between theory and experiment is required.

P. A. Doyle, C. M. Scala/Ultrasonics 34 (19%)

In the present paper, we first present systematic observations of Rayleigh waves in the near field of a thermoelastic laser line source, using a wide-band interferometer
for ultrasonic detection at a range of source-detector
separations. These observations allow determination of
all the parameters necessary for an absolute comparison
with theory. Next, a new approximate theory is developed for the Rayleigh wave in the near field of the finite
line source. In contrast with the theory of Berthelot
[14], the new model takes account of both the finite
length and the finite width of the line source. Finally, a
comparison which is free of any adjustable parameter is
made between theory and experiment for the laser line
source in both the time and frequency domains.

2. Experiment
2.1. Procedure
Near-field characteristics of the Rayleigh (R) waves
thermoelastically generated by the incidence of a laser
line source on the surface of a specimen were studied in
both the time domain and frequency domain up to
20 MHz. The experimental arrangement used for this
study is shown schematically in Fig. 1. The test specimen
was made from 7079-T6 aircraft aluminium alloy and
had sufficiently large dimensions to be regarded as a halfspace for the present investigation. The R waves were
generated using a Q-switched Nd: YAG laser focussed
to a laser line source on the specimen. R-wave characteristics were measured using an UltraOptec OP-35 MachZender interferometer which is sensitive to out-of plane
displacement of a surface over a wide frequency range.
Detection was carried out at a series of distances, x,
from 2 mm to 12 mm along the axis perpendicular to
the laser line source, where x = 0 corresponds to the
centre of the line source.
The Nd : YAG pulsed laser used to produce the laser
line source was operated in the TEMoo mode with a

helerodyne
interferometer

cylindrical lens

beam &pander

Fig. 1. Schematic of the experimental arrangement used for the nearfield study of Rayleigh waves generated thermoelastically on the
surface of a specimen by a laser line source. The Rayleigh waves were
detected on this surface using an interferometer probe at a distance,
X, along the axis of the line source.

l-8

half-width pulse duration of 25 ns and a Gaussian beam


cross-section with an energy half-width of 1 mm [ 151.
As shown in Fig. 1, the Nd: YAG beam was passed
through a beam expander to create a parallel beam with
the original half-width expanded by a factor of 16. This
expanded beam was passed centrally through a 10 mm
radius circular aperture. The Nd:YAG beam was then
sent through a large cylindrical lens. The lens focussed
the beam through a dichroic filter onto the surface of
the half-space specimen, forming a thermoelastic ultrasonic line source of length 20 mm. The dichroic filter
was used so that both the generation and detection laser
beams could be applied to the same surface for the
R-wave study.
A quantitative comparison between theory and experiment for R waves from the laser line source required
measurement of the energy and width of the focussed
Nd : YAG laser line incident on the specimen. The total
energy in the incident laser line was measured using
a Scientech Astral AD30 energy/power meter. For the
laser conditions used in this study to ensure a thermoelastic line source, the averaged incident energy in the
focussed line was found from 20 separate laser firings
to be 2.9 + 0.3 mJ. The width of the focussed line was
measured using the procedure outlined by Letalick and
Renhorn [16]. Firstly, the beam profile was obtained
by scanning a narrow slit aperture across the focussed
line and measuring the energy transmitted by the slit
as a function of slit position. For each position, energy
measurements from 20 laser firings were averaged.
Separate scans were made across the Nd : YAG focussed
line using slit widths of 50 urn, 40 pm, 30 pm and 20 urn
respectively. The full beamwidth at half-maximum energy
(FWHM) was then obtained from the width of the
measured beam profiles by application of Letalick and
Renhorns correction factors, and was found to be less
than 30 urn. A more accurate estimate was not possible
due to the low energy levels transmitted at narrower
slit widths.
The heterodyne UltraOptec OP-35 interferometer used
for R-wave detection contained a 5 mW HeNe laser
whose beam was split into (i) an internal reference beam
and (ii) an external probing beam [ 173. As shown in
Fig. 1, this external probing beam was reflected at 90
by the dichroic filter described above, and focussed on
the specimen surface using the interferometers internal
microscope objective; the beam then returned to the
interferometer along the reverse path. Therefore, the
focussed probing beam on the specimen surface was
essentially a point detector of R waves for frequencies
of interest in this study, i.e. up to 20 MHz.
The interferometers calibration was verified up to
20 MHz for the present study to allow the parameterfree comparison of experiment and theory. Procedures
used were based on (i) extensive application of manufacturers recommended calibration procedures and (ii) our

P. A. Doyle, C. M. Scala/ Ultrasonics 34 ( 1996) I-8

participation in an international calibration programme


involving round robin testing of various types of interferometers [ 181. These studies showed the interferometer
had a uniform response of 0.054 nm mV_ (-J loo/) over
the frequency range 0.5 MHz to 20 MHz with sharply
decreasing response below 0.5 MHz.
R-wave signals detected by the interferometer were
amplified by a LeCroy 6103 amplifier with uniform gain
between 0.2 and 100 MHz and with filtering beyond
these limits. This preamplifier was connected to a LeCroy
8818 transient recorder. The recorder was electrically
triggered by the laser as it fired and digitized 8K points
of each amplified ultrasonic signal at a sample rate of
10 ns. The resulting time-domain waveforms were stored
in an IBM PC-AT under LeCroy Waveform-Catalyst
software control. Twenty waveforms were stored and
averaged for each set of experimental conditions. Timeaveraged waveforms were imported into Asystant software for FFT analysis. Magnitude frequency spectra
were calculated from 1024 points in each averaged
waveform. Curve fitting procedures were then applied
to the spectra using Asystant software for comparison
with theory.

2mm

6mm

7mm

6mm

9mm

10mm

11 mm

2.2. Results: Rayleigh waves observed by laser


interferometer

Fig. 2 shows near-field averaged waveforms observed


on the surface, at various distances, x, along the axis
perpendicular to the thermoelastic laser line source.
These waveforms are the normal displacement components, measured in nanometres as outlined above. In
each case, the waveforms show a monopolar R pulse,
with a positive displacement indicating a depression of
the surface. Analysis of Fig. 2 showed that (i) the peak
amplitude, (ii) the full width of the pulse at l/e of peak
amplitude, and (iii) the duration of the R pulse, remain
constant, within the axial near-field of the line source;
their values were 0.37 nm (+ lo/), 63 ns (& 10%) and
185 ns (+ 5%) respectively. By comparison, the peak-topeak R-pulse amplitude for a circularly symmetric source
decreases as (distance) -12 for all practical observation
distances [ 121, as could be anticipated from energy
conservation requirements.
Fig. 3 shows typical examples of magnitude frequency
spectra for near-field R waves from the thermoelastic
line source; the spectra shown correspond to the
waveforms in Fig. 2 for the interferometer probe at
(a) x = 3 mm, and (b) x = 9 mm, along the axis perpendicular to the source. Clearly, the near-field laser linesource generates R waves over the full band-width
studied from 0.5 MHz to 20 MHz, with the spectral
magnitude decreasing with increasing frequency. Further
analysis of the experimental data in Fig. 2, including
details of the Lorentzian fitting shown, is presented

Fig. 2. Rayleigh waves observed in the near field of a narrow line


source of length 20 mm, generated thermoelastically on the surface of
an aluminium block by a laser pulse. The observations were made
using an interferometer at distances from x = 2 mm to x = 12 mm
along the axis perpendicular to the line source.

following the development of a theory for the R waves


generated by a laser line source.

3. Theory

In this section, we develop a model to explain the


experimental observations. Initially, we consider a line
source whose width is negligible compared with the
minimum wavelength in the observed Rayleigh-wave
spectrum. Subsequently, a finite line width is included.
Both diffraction by the edge of the circular aperture
(Fig. 1) and line distortion by lens aberrations are
regarded as negligible.
3.1. Thin line source
Consider a coordinate system (x, y, z) with (x, y) as
the surface of a half-space, and z defined as positive
inwards. Suppose that a laser pulse is focussed into a

P. A. Doyle, C.M. Scala/Ultrasonics 34 (1996) I-8

In order to derive w(y), note that the beam crosssection produced by the Nd : YAG laser operating in its
lowest (TEM,) mode is Gaussian. As described above,
this beam is expanded so that the light intensity in
the aperture plane is Gaussian with half-width, a. This
Gaussian beam is truncated by the aperture of radius d,
so that after focussing to the y-axis by the cylindrical
lens, w(y) is of the form
(d2_ ,+1/z
exp { -(x2 + y2)/a2 ) dx
W(Y)a
(2)
s0

(4

Eq. (2) exploits the symmetry whereby the light intensity


in each quadrant of the aperture plane is identical.
Because w(y) describes the redistribution of energy by
the cylindrical lens from a circular to a line geometry, it
should be normalized by equating its integral over the
full line to unity. It then follows that

FREQUENCY (MHz)

(b)

w(y) = W exp( - y2/a2)@ {(d2 - ~)~/a},

0.6

where @ is the error function [ 193 and the constant W


is given by the normalization condition

E
2

(3)

0.6

5
4

0.4

d exp(- y2/u2)@ {(d2 - ~~)//a} dy = 1.

s0
02

0.0;

10

15

20

FREQUENCY (MHz)

(4)

To specify the normal point source displacement


u,,(l, t), we first introduce the source time dependence
q(t) and a convenient constant K For the TEMoo laser
mode, Ready [20] has given

Fig. 3. Relative magnitude spectra (solid curve) for the frequency


range from 0.5 MHz to 20 MHz for near-field Rayleigh waves from
the thermoelastic laser line-source. Results are given for the interferometer probe at distances (a) x = 3 mm, and (b) x = 9 mm, along
the axis of the line source. A Lorentxian fit (dashed curve) to each
spectrum is also shown for comparison with theory. In each case, the
spectrum is scaled so that the Lorentzian fit has a magnitude of one
at zero frequency.

where H(t) is the Heaviside step function and u is a


measure of the laser pulse duration. The constant I
depends on mechanical and thermal parameters of the
isotropic solid:

narrow line of length 2d in the y-direction and centred


at the origin of coordinates. For this source, the superposition principle permits the normal displacement on
the x-axis, t&x, t), to be written as a line integral over
suitably weighted point source contributions:

where 1 and ,u are the Lame constants, tlT is the linear


coefficient of thermal expansion, k and K are the thermal
diffusivity and conductivity, respectively, and Q is the
total energy absorbed from the laser pulse by the solid.
The displacement uo3(1,t) can be written as the time
convolution [ 121

d
l&x,

t)=

s0

w(Y)u,(~ t) dy

(1)

where w(y) is a suitable weighting function and ~~~(1,t)


is the normal surface displacement at time t and distance
I = (x2 + 3)
from a point thermoelastic source. Such
a point source is equivalent to two equal, orthogonal
dipoles in the surface plane, which Rose [ 1l] has called
a surface centre of expansion, or SCOE. The weighting
function w(y) is proportional to the absorbed energy at
(0, y) on the specimen, and is normalized to include the
equal contribution to z&x, t) from the part of the line
source with -d I y I 0.

4(t) = (Uv12)exp(- WW)

(5)

r = (31+ 2&xTkQ/K,

(6)

UO3(
I, t) = rq( t)* aSii,i( I, t)/at.

(7)

The function Sz,i( I, t) is the normal surface displacement


at time t and distance 1 from a point SCOE whose time
dependence is a linear ramp function, tH(t). The usual
Einstein convention is employed for the subscripts of
S, except that the repeated index i is summed only over
the surface coordinates (1,2).
At the R-wave arrival, Sii,i(l, t) is dominated by a
singularity of the form [ 1l]
Ski,i(1,t) Z

(2/7Z&)MH(t

S,l)/(

t2 -

Si12)12.

(8)

Here, sR is the slowness for R waves and A4 is a

P. A. Doyle, C. M. Scala/Ultrasonics 34 (19%) I-8

dimensionless constant whose value for aluminium was


found, from numerical calculations of surface waveforms
generated by a thermoelastic source [ 121, to be 0.11086;
the plus sign is appropriate when, as here, the z-axis is
chosen to point down into the specimen. Also, cL is
the longitudinal wave speed, and the factor 2/7r,uc~
re-normalizes the dimensionless units used in the generalized ray programs of Hsu [ 211 as the basis of deriving
the value for M.
We now write t&(x, t) by inserting Eqs. (3), (7) and
(8) into Eq. (1):
l&x, t) = (2TMW/(7c&)}q(

t) * az( t)/&

(9)

where
for t < sRx, (10a)

Z(t) = 0

d exp(-y2/a2)@ {(d2 - y2)12/u}


t -

s,l)/(

t2 -

s:

12)12

dy

for t 2 sRx. (lob)

Eq. (10a) simply reflects the absence of any normal


displacement before the arrival of the R wave from the
centre of the line source, within the approximation to
Sii,i( 1,t) (Eq. (8)) which neglects the weak surface pressure wave; Scala and Doyle [ 151 have described detailed
observations of this surface pressure wave when generated by a circular thermoelastic source.
It is convenient to change the variable in Eq. (lob) to
8, defined by
sin 8 = y/X,

where X2 = t2/si - x2.

(11)

Then for t 2 sRx, and writing 5 = (d2 - X2 sin2 Q112,


il
Z(t) = (l/srJ

exp(- X2 sin e/a2)q</a)

de,

s0

where
a = 7112
= arcsin(d/X)

= -2t/(d$)
n/2
X

(12)

for sRx I t I sR(x2 + d2)2,


for t > sR(x2 + cI~)/~.

(13)

The two values for the integration limit a distinguish


observation times respectively before or after the arrival
of the R waves from end poini of the line source.
Differentiating Z(t),
iTZ(t)/c%= {7c@(d/u)/(2s*)}6(r) + {az(t)/at}H(z).

(14)

where the retarded time z = t - sRx. Experimentally, we


note that the wave amplitude in all cases has returned
to the noise level by 200 ns after the first arrival of the
Rayleigh wave (Fig. 2). Thus, we need only consider t
within the range sRx I t I sRx + 200 ns, so for the values
of d and x of interest, Eq. (13) gives a = 7r/2. Therefore, using a/& = [(X2 + x2)12/sRX]a/aX in Eq. (12),
the second term in Eq. (14) is

CW-x

+ (u/&/~)

sin e/a2p(gu)

exp(- d2/u2)] sin2 e de. ( 15)

Examination of Eqs. (9), (14) and (15) suggests a


Rayleigh waveform which has an initial positive (i.e.
inward) spike, followed by a negative tail. However,
numerical estimation of Eq. (15) shows that this negative
waveform tail is negligible in the near-field limit x -+ 0,
rising only to less than 2% and 4% at x = 10 mm and
20 mm, respectively, of the peak value of the initial
spike, for the experimental value q = 25 ns. Therefore,
the second term in Eq. (14) can be omitted in modelling
the experimental waveforms of Fig. 2. Eqs. (5), (9) and
(14) then give
uS(x, t) = Pr exp(-r/$H(r)

s0
x H(

(az(t)/atjzqz)

(16)

where
fl= TMW@(d/u)/(psRc~tj2).

(17)

Because M is positive, t&x, t) is positive, and this


implies a depression of the surface. Thus, the near-field
Rayleigh wave on the axis of a sharp thermoelastic
line source produced as in Fig. 2 is a monopolar, linear
depression of constant amplitude propagating away from
the line source at speed l/sR. This was precisely the
qualitative form of the experimental observations summarized in Fig. 2. In addition, Eq. (16) shows that the
waveform reproduces the time dependence of the incident laser pulse energy. These conclusions also agree
with the infinite, narrow line model of Berthelot [ 141.
It is noteworthy that increasing the line length 2d
influences u!(x, t) in Eq. (16) only via the factor @(d/u),
which is related to normalization of the total incident
intensity. That is, increasing the length of the line source,
while keeping all other parameters constant, does not
itself increase the near-field R-wave. This can be understood by noting, from time-of-flight considerations, that
the monopolar axial Rayleigh pulse originates only from
close to the centre of the line. The near-field R-waves
from the remainder of the line source cancel out within
the present approximations.
On the axis of the line source in the far field x + co,
the second term in Eq. (14) becomes significant, so
that the limiting Rayleigh waveform is bipolar. In fact,
R-wave contributions from all points on the finite line
source arrive in phase at a distant axial point, so the
waveform is identical to that from a point source of
equal total strength. Note, however, that the far field
waveform only occurs at large distances compared with
the line length which, for the present case (2d = 20 mm),
is beyond the region of frequent experimental interest.
Also, the far field Rayleigh waveform off the axis will
differ from the point source case, and will be of lower

P.A. Doyle, C.&i. Scala/Ultrasonics 34 (1996) 1-8

amplitude in accord with the line-source


function, which is highly directional [ 31.

directivity

where, for sR(x - b) I t s sR(x + b),


A = exp(r/q - r2/4n2y2) - exp(- bs,/q - d2/a2),
B = @(Y + d/a) - @{Y - ~/(2~)),

3.2. Finite-width line source

(224

while for t > sR(x + b),


We now extend the analysis to derive the vertical
R-wave displacement, u,(x, t), produced on the axis by
a line source of finite width. Suppose that the cylindrical
lens described above focuses the laser pulse not to an
infinitely thin line, but to a narrow ellipse of length
2d and maximum width 2b on the x-axis. Because, as
we have noted from time-of-flight considerations, the
Rayleigh pulse of interest comes from the central part
of the line, the finite-widths source should be weighted
according to the absorbed laser energy on the x-axis.
Since the cylindrical lens compresses the light intensity
by a factor (d/b) in the x-direction, the appropriate
weighting function u(x) is
u(x) cc exp { -(dx/ab)2}

for 1x1I b.

(18)

In order to derive the influence of finite source width


on +(x, t) for the R wave, it is convenient to replace the
independent variable x in Eq. (18) by the corresponding
propagation time t = sRx. In addition, v(t) should be
normalized to unity over the range -s,b I t I s,b, so
that
for ItI > s,b,

v(t) = 0
=exp(-t2/~2)/{~/211@(d/a)}

for ItI <s,b

(19)

where [ = (s,ba)/d. The cut-off of v(t) at ItI = s,b corresponds to the neglect of edge diffraction of the laser light
by the circular aperture used in the experiments; for the
large aperture used (2d = 20 mm), this approximation
has a negligible influence on the final Rayleigh pulse
predicted.
The displacement due to the line source of finite width
can be written as
u,(x, t) = u(t)*&,

t).

A = 2 sinh(bs,/q) exp(-d2/u2),

B = @(y + d/a) - @(y - d/u).

(22d)

The factors A and B differ, depending on whether all


(Eq. (22d)) or only some (Eq. (22~)) of the R waves have
arrived at an observation point at distance x from the
source, which is itself extended over -b I x I b. Clearly,
Eq. (22b) reduces to the infinitely-thin line result (Eq. ( 16))
in the limit b + 0.

4. Parameter-free comparison of theory to measured


Rayleigh waves
In this section, we make an absolute comparison
between theory and experiment for the time- and
frequency-domain characteristics of Rayleigh waves generated by the laser line source. Theoretical prediction of
line-source characteristics, based on Eq. (22), requires
evaluation of the various parameters in Eq. (17) for 8,
for the laser conditions and material used experimentally.
Firstly, the purely geometrical factor W@(d/u) is found,
using Eq. (4) and the experimental values d = 10 mm
and a = 16 mm, to equal 1.3587 x lo2 m-i. The pulse
duration parameter for the Nd:YAG laser was q = 25 ns.
The basic physical parameters for the aluminium alloy
specimen required for B, and the factor r which it
includes, are listed in Table 1. The only factor in /I which
requires further attention is the absorbed energy, Q.
To estimate Q, a model is needed to divide the incident
laser-pulse energy, E, between reflection and absorption. For the infrared wavelength, LI = 1.06 urn of the
Nd: YAG laser, it is a reasonable approximation to
neglect quantum mechanical effects and use classical

(20)

From Eq. (16),


(r+s,b)
n,(x, r) = /J

T exp(- T/q)u(t - T) dT.

s max(0.r

(21)

parameters
[ 20,231

used

to model

the 7079-T6

aluminium

alloy

- s,b)

Eq. (21) can be expressed in closed form by straightforward, though lengthy, analysis to give, in terms of a
dimensionless constant y = sRbu/( 2dq),
u,(x, t) = 0

Table 1
Physical
specimen

for t < sR(x - b),

(224

= B exp(-W@(d/a)
x Cvl~Al~~+ (B/2)(7 - 2vy2)

for t r sR(x - b)

exp(y2)1
(22W

Longitudinal
wave speed, ct.
Rayleigh wave slowness, sa
Lame constants, 1

6.42
0.341
57.69
26.45

p
Linear expansion coefficient,
Thermal diffusivity, k
Thermal conductivity,
K
Relative permeability,
Electrical conductivity,
(measured)

aT

2.3 x 1O-5
9.8 x 1O-5
2.38 x lo*

km s-
s km-r
GPa
GPa
K-
m2 s-r
W m-l

1.00
c
(1.82 f 0.04) x lo7

S m-r

K-

P. A. Doyle, C. M. Scala/Ultrasonics

electromagnetic

theory, which leads to [l]

Q = 4(7r~//4,~oovA)~E

(23)

where ,uis the relative permeability, h = 47~x lo- H m-l


is the free-space permeability and v = 2.998 x lo* m s-l
is the speed of light in vacuum. The electrical conductivity, 0, for the aluminium alloy specimen was
measured directly, using a straightforward eddy current
technique [22], as (1.82 f 0.04) x 10 S m-. (It is noteworthy that, unlike the other specimen parameters
involved, this value differs substantially from that for
pure aluminium, which is 3.77 x 10S m-l.) With these
data, Eq. (23) predicts an absorption of 8.3% of the
incident laser pulse energy which, as described under our
experimental procedure, was measured as 2.9 + 0.3 mJ.
Therefore, Eq. (23) predicts that 0.24 + 0.02 mJ were
absorbed in our experiments.
The normal displacement on the axis of a finite-width
line source can now be calculated using Eq. (22). It is
informative to express the maximum line-width 2b in
terms of a length scale q/sR (=73.3 urn in the present
case), since this length is a measure of how far the R
wave travels in the time the energy is deposited by the
laser pulse. Fig. 4 shows +(x = 10 mm, t) for 2b = nr&,
with n = 0, 1,2,3. As expected, increasing the line width
causes a decreased amplitude and an increased monopolar pulse width for the Rayleigh wave.
Because the predicted R-pulse amplitude remains constant in the near-field, it follows that the experiments
described above, conducted in the range 2 I x.5 12 mm,
can be compared directly with the theoretical results in
Fig. 4. For the experimental line width 2b < 30 urn, Fig. 4
shows that the Rayleigh waveform does not differ signifi-

I-

n=o j

Time (microsec)

Fig. 4. Effect of line-source width on the predicted inward displacement


of the Rayleigh wave at 10 mm along the axis of a line source 20 mm
long. Results are given for line-source widths of n x 73.3 pm, where
n = 0, 1, 2 and 3 are shown (73.3 pm is the distance a Rayleigh wave
travels in 25 ns, which is the experimental pulse duration parameter).
The other parameters used for the calculation are specified in the text
and Table 1.

34 ( 1996) I-8

cantly from that for an infinitely sharp line. Thus, the


peak inward displacement is predicted in a parameterfree manner by the curve n = 0 in Fig. 4 to be 0.31 nm,
in fair agreement with the measured value of 0.37 nm
(2 10%). Approximately half of the discrepancy can be
attributed to the cumulative effect, on the scaling constant /?, of errors in the basic physical specimen parameters; the approximate form for the Rayleigh pulse
(Eq. (8)) may also contribute significantly. Another comparison is the full width of the pulse at l/e of the peak
amplitude, which was 63 ns (&-10%) experimentally, in
reasonable agreement with the theoretical prediction of
75 ns in Fig. 4.
Completion of the comparison between theory and
experiment requires investigation of the form of the frequency spectrum for the near-field R-waves from the laser
line source. Transformation of Eq. (16) gives the magnitude of the frequency spectrum, @(x, sZ)l in the form
[l&x, sz)l = (1+47X2$522)?

(24)

Fits to the experimental frequency spectral data using


the theoretically predicted Lorentzian shape in Eq. (24)
and the known q = 25 ns are shown for x = 3 mm and
9 mm, respectively, in Figs. 3(a) and (b) where, in each
case, the spectrum is scaled so that the Lorentzian fit
has a magnitude of one at zero frequency. Good agreement is evident between the theoretical prediction and
experiment for the form of the frequency spectra of the
R wave.

5. Conclusions
The principal achievement of this work has been to
establish a quantitative basis for the use of the nearfield Rayleigh waves thermoelastically generated by a
laser line source in nondestructive evaluation.
Experiments using interferometric
detection have
shown that for the near field of the line source the
Rayleigh pulse is a monopolar depression and has uniform amplitude; both these properties of the Rayleigh
pulse are useful for NDE applications. A model has been
developed which led to an approximate expression in
closed form for the normal displacement of the Rayleigh
wave. The model took account of both the finite length
and finite width of the line source. Quantitative
agreement was found for a parameter-free comparison
between theory and experiment for the Rayleigh wave
characteristics in the time and frequency domain. This
quantitative agreement should provide the basis for the
more effective use of Rayleigh waves in both defect sizing
and surface characterization.

P. A. Doyle, C. M. Scala/Ultrasonics 34 ( 1996) l-8

Acknowledgements

The authors are indebted to S.J. Bowles for assistance


with the experiments, and to Dr S.K. Burke for valuable
discussions and for a critical reading of the manuscript.

References
Cl1 C.B. Scruby and L.E. Drain, Laser Ultrasonics: Techniques and
Applications (Adam Hilger, Bristol, 1990).
PI C.B. Scruby, F.K. Brocklehurst, B.C. Moss and D.J. Buttle,
Nondestr. Eval. 5 (1990) 97.
c31 A.M. Aindow, R.J. Dewhurst and S.B. Palmer, Optics Commun.
42 (1982) 116.
c41 P.A. Doyle and C.M. Scala, Ultrasonics 28 (1990) 77.
CSI A. Aharoni, M. Tur and K.M. Jassby, Appl. Phys. Lett. 59
(1991) 3530.
C51P.A. Doyle and C.M. Scala, J. Acoust. Sot. Am. 93 (1993) 1385.
c71 C.M. Scala and P.A. Doyle, J. Nondestr. Eval. 14 (1995) 49.
PI J. Huang, S. Krishnaswamy and J.D. Achenbach, J. Acoust. Sot.
Am. 92 (1992) 2527.

[9] K. Yamanaka, Y. Nagata and T. Koda, Appl. Phys. Lett. 58


(1991) 1591.
[lo] C. Edwards, A.C. Bushell, S.B. Palmer and H. Nakano, Nondestr.
Test. Eval. 10 (1992) 15.
[ 111 L.R.F. Rose, J. Acoust. Sot. Am. 75 (1984) 723.
[ 121 P.A. Doyle, J. Nondestr. Evaluation. 8 (1989) 147.
[ 131 W.L. Pilant, Elastic Waves in the Earth (Elsevier, Amsterdam,
1979) p. 103.
[ 141 Y.H. Berthelot, Ultrasonics 32 (1994) 153.
[ 151 C.M. Scala and P.A.Doyle, J. Acoust. Sot. Am. 85 (1989) 1569.
[ 161 D. Letalick and I. Renhom, Rev. Sci. Instrum. 58 (1987) 765.
[17] J.-P. Monchalin, IEEE Trans. Ultrasonics, Ferroelectrics and
Frequency Control UFFC-33 (1986) 485.
[ 181 C.M. Scala, P.A. Doyle and S.J. Bowles Nondestructive Testing
Australia 31 (1994) 90.
Cl93 M. Abramowitz and I.A. Stegun, (Eds.) Handbook
of
Mathematical Functions (Dover, New York, 1965).
[20] J.F. Ready, Effects of High Power Laser Radiation (Academic
Press, New York, 1971).
[21] N.N. Hsu, US National Bureau of Standards Rep. No. NBSIR
85-3234 (Gaithersburg, MD, 1985).
[22] S.K. Burke, DSTO Aeronautical and Maritime Research
Laboratory, Australia, Private communication (1994).
[23] G.W.C. Kaye and T.H. Laby, Tables of Physical and Chemical
Constants (Longmans, London, 1959).

Вам также может понравиться