Laser-Generated Ultrasound [Good for Theory]

J. Phys. D Appl. Phys. 26 (1993) S29-348. Printed in the UK

REVIEW ARTICLE

Laser-generated ultrasound: its properties, mechanisms and multifarious applications

S J Daviest, C Edward&, G S Taylor5 and S B PalmerS t AEA Fusion, Culham laboratoly, Abingdon, Oxon OX14 3DB, UK $ Department d Physics, University d Warwick, Covently CN4 7AL, UK 5 GEC Hirst Research Centre, Wembley, Middlesex HA9 7PC UK

Received 8 September 1992, in final form 17 November 1992

Abstract. The laser generation d ultrasound in solids is reviewed with particular emphasis on the application of this unique ultrasonic source. Three regimes for the generation of ultrasound in solids using lasers in the visible near infrared wavelength region exist: thermoelastic, plasma and constrained surface source regimes. The mechanism for ultrasonic generation in each of these regimes is given. Recent experimental investigations into laser-generated unrasound are also described, including a description d a different mechanism for the generation of ultrasound in solids using a COn laser. Finally, the many applications of lasergenerated ultrasound are reviewed. These applications range from non- destructive testing to the determination of candle flame parameters. Possible future applications are also outlined.

1. Introduction

Since the early 1960s there has been considerable inter- est in the production of ultrasonic waves by irradiation of a solid with a laser pulse. This interest reached its zenith, in terms of the number of papers published, in the early 1980s. Early experiments concentrated on recording the elastic wave produced by the inter- action of a laser pulse with a solid. Subsequently, more detailed investigations were undertaken, in association with mathematical modelling, to characterize more fully the laser source. In addition to obtaining an understand- ing of the laser source, investigations were undertaken to determine areas of applicability. Most of the present research efforts in laser-generated ultrasound concen- trate on the areas of application of this non-contacting, highly repeatable source of broadband ultrasound.

This review paper describes the mechanisms for laser generation of ultrasound. The initial investigations and more derailed descriptions of the generation mech- anisms may be found elsewhere, for example Scruby et a1 (1981), Cooper (1985), Aindow (1986), Hutchins and 'Eim (1986), Aussel et a1 (1988), Scruby (1989) and Scruby and Drain (19%). It is the intention of this paper to concentrate on the areas of application of this unique ultrasonic source.

W22-3727/93/030329+20$07.50 @ 1993 IOP Publishing Ud

2. Characterization of the laser ultrasonic source

Three regimes for the production of ultrasound on metals using lasers with wavelengths in the near infraredhisible wavelength region were identified; the comments about the bulk waves refer to the epicentral amplitude.

(i) The thermoelastic regime with typical power den- sities < io7 W for a metal in the near infra- redhiBible wavelength region. In this generation regime the amplitudes of the compression, shear and surface (Rayleigh) waves increase linearly with increasing power density (Scruby er al 1981).

(E) The plasma regime with typical power densities 2 lo7 W for a metal in the near infraredhrisible wavelength region. The generation of compression and surface (Rayleigh) waves is enhanced in this genera- tion regime, whilst the shear wave amplitude reaches a maximum near the onset of a plasma and then steadily decreases with increasing power density (Scruby er a1 1981).

(i) The constrained surface source regime, which occurs for power densities in both the thermoelastic and plasma regimes. The amplitudes of the compression, shear and surface (Rayleigh) waves are enhanced when a transparent constraining layer is used.

329

S J Davies et a/

Figure 1. Theoretical surface displacement for a thermoelastic source and point detector 50 mm away from the source (alter Cooper (1985)).

2.1. The thermoelastic regime

Generation of ultrasound in the thermoelastic regime occurs through thermal mechanisms. The acoustic source model consists of a thin heated disc extending into the solid to a depth determined by heat diffusion. The diameter of the source is determined by the laser beam diameter and the depth does not exceed that to which significant heating takes place in the dur- ation of the laser pulse. Elastic waves are generated within this zone by strains arising from thermoelastic expansion which, because of the nature of the source, act radially outwards from the centre, in the plane of the surface, with approximately Heaviside time de- pendence. Physically this can be thought of as rapid heating by the laser pulse of the thin disc, with sub- sequent rapid expansion, followed by a much slower rate of contraction.

Using the differential heat equation for a semi- infinite medium, the temperature distribution, and con- sequently lateral strain, developed across the surface can be calculated. This strain is proportional to the laser energy absorbed from the incident beam and is thus inversely proportional to the wavelength as the ab- sorption generally decreases with increasing wavelength in the infraredh.isible wavelength region Studies of the variation of the amplitude of the compression and shear waves as a function of laser wavelength verified this prediction (Aindow el a1 1980).

Using this force distribution it is possible to calcu- late the surface and epicentral displacements. The cal- culation of the surface displacement for a point radial loading of a half space with Heaviside time dependence has been described by, for example, Cooper (1985) and Sinclair (1979). The theoretically predicted surface dis- placement for a point radial force with Heavidde time dependence is shown in figure 1. The compression wave arrival is too small to be seen on this scale. The dis- continuity at the Rayleigh wave arrival time disappears when a detector with a finite frequency response and sensing region is used. Cooper (1985) has shown ex- perimental waveforms which agree with this theoretical prediction.

330

Time < us )

l b l

BO 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

T h e ( us )

Figure 2 (a) Theoretical thermoelastic epicentral displacement in Stainless steel. (b) Thermoelastic epicentral displacement recorded on 10 mm thick Stainless Steel sample.

Scruby ef a1 (1980) and Dewhurst ef a1 (1982) used a Green function formalism to obtain the epicentral dis- placement due to a thermoelastic source. Similarly Rose (1984) considered the p i n t source as a surface centre of expansion (SCOE) and obtained a formal solution again using a Green function formalism. Cooper (1985) used the method outlined above to obtain the epicentral dis- placement for a thermoelastic source and this is shown in figure 2(u). An example of the experimental epicen- tral waveform obtained with a thermoelastie source is shown in figure 2(b). Note the relatively small positive spike before it goes negative at the compression wave arrival time.

The theoretical modelling of the thermoelastie source described above took no aecaunt of thermal diffusion into the solid. Doyle (1986) and McDon- ald (1990) considered the effect of thermal diffusion on the predicted epicentral displacement to determine whether it was a significant contribution. The theoret- ical epicentral waveform obtained by this method had an initial positive spike at the compression wave arrival time. The ratio of spike height to the compression wave step was about 0.7, which agrees with the ratio obtained experimentally by Scruby ef al (1981) and with the ratio obtained in figure 2(b). Thus thermal diffusion was con- sidered to be a major cause of the initial positive spike of the epicentral waveform.

Physically the appearance of the initial positive dis- placement at the epicentre can be understood by con- sidering the contribution arising at a particular time

from a small depth below the irradiated surface. The source can be considered to propagate, in a diffuse manner, into the bulk. This causes a small time de- lay between the direct compression pulse and the pulse that is reflected at the surface boundaly, so that there is incomplete cancellation and pulses appear at the com- pression arrival. As the rise time of the incident laser pulse decreases the amplitude of the positive displace- ment increases.

Other recent papers reporting on the laser- generation of ultrasound in the thermoelastic regime include Rudd (1983), Aindow (1986), Krylov et al (1986), Aharoni d al (1989, Aleshkevich d al (1987), Askar’yan and Yurkin (1987), Bresse and Hutchins (1988), Kolomenskii (1988) and Sullivan ef d (1988). Rudd (1983) calculated the sound power and efficiency of ultrasonic generation for the thermoelastic regime as well as for generation by radiation pressure and in the plasma regime. Sullivan et d (1988) considered the thermoelastic generation of ultrasound using a numeri- cal finite-element technique. Their results compared favourably with the more rigorous analytical treatments outlined above.

Aindow (1986) and Aharoni d a1 (1987) conducted further experimental investigations of the properties of laser-generated Rayleigh waves. They demonstrated that the width of the ultrasonic source affects the Rayleigh wave profile. As the width of the source increased, the Rayleigh wave time profile broadened. This is a result of the thermal distribution generated by the laser pulse. For example, the thermal stress generated by a Gaussian shaped laser pulse profile has a distribution which is the differential of the laser pro- file. The maximum stress occurs at the half-width points and generation occurs throughout the ultrasonic source. Therefore, as the incident laser pulse profile broadens, ultrasonic generation occurs over a larger source size, resulting in a broadened Rayleigh wave profile.

The effect of source size on the generation of com- pression and shear waves was considered by Bresse and Hutchins (1988). They considered the acoustic radia- tion fields from wide or finite-size sources, which is a more realistic approach to the experimentally obtained source. They demonstrated (see figure 3(u)) that, for a uniformly illuminated source, all signals are generated at the edge of the laser beam. An apodized source, figure 3(6), results in a waveform with compressional and shear components being emitted from the whole source.

Off-epicentre it was observed that the amplitude of the compression and shear waves varied as a function of angle measured from the surface normal. For a circular ultrasonic source in an isotropic solid, the di- rectivity would be expected to be axisymmetric. This was found to be the case, as the amplitude of the wave mode recorded at a particular angle from the sur- face normal was the same, within experimental error, at all corresponding locations around the source. Cooper (1985) and Scruby e! 01 (1981), using a method simi- lar to that described by Lord (1966) and Rose (1984),

‘ 1 1

Force U F I r l = b l r - a l

I

i f - L Energy E l r i = l r c a

E l r l =O elsewhere

Laser-generated ultrasound

2%- I I

,//)//),,),,,

normalized time l t / t C l

Figure 3. Sources and waveforms generated by a pulsed wide laser beam using the thermoelastic generation mechanism: (a) a uniform distribution cf optical energy leads to forces in the form of a ring of expansion at the beam edge, resulting in mmpression and shear wave generation at the edge; (b) an apcdization function leads to forces that increase linearly from the centre outwards, leading to mmpression and shear wave components from the whole source (after Bresse and Hutchins (1988)).

calculated the expected directivity of the compression and shear wave generated in the thermoelastic regime. These directivities are illustrated in figure 4 along with the experimental compression wave directivity, obtained by, for example, Hutchins (1981), Cooper (1985) and Aussel ef al (1988). Measurement of the shear wave directivity is hampered by the head wave which arrives at approximately the same time. The directivities of Rayleigh waves generated by a circular and line-shaped laser ultrasonic source in both the thermoclastic and plasma regimes have also been determined (Aindow er a1 1982). The partition of the total ultrasonic energy in the thennoelastic regime into the various acoustic modes as a function of Poisson’s ratio was determined by Rose (1984) and Cooper (1985). Their results are summarized in table 1.

2.2 The plasma regime

At higher power densities, thermoelastic generation is supplemented by the ablation of electrons and ions, forming a plasma, with consequent expansion of the plasma away from the surface of the metal. The resulting momentum pulse is transmitted into the solid, thus enhancing the production of the epicentral com- pression wave pulse. A small ablation pit, about 5 pm deep, is formed by this process. Although the amplitude of the epicentral compression wave generated is in-

331

S J Davies et af

Table 1. Patiiion d wavefront energy for the laser uitrasonic source (after Rose 1984 and Cooper 1985).

"noelastic regime Plasma regime Compression to Rayleigh to Rayleigh to

Poisson's ratio shear ratio shear ratio shear ratio

0.2 1.304 6.93 0.293 4.65 0.267 3.58 0.185

0.4 0.019 3.14 0.968

0. l a 1

e-3 Radial stress - Theory

+Measured

0. I bl

Radial stress

Figure 4. (a) Theoretical and experimental directivity d compression wave for thermoelastic regime source in aluminium (after Cooper (1985)).

creased, this damage may make it impractical for use in some applications.

The source of ultrasound in the plasma regime has a stress distribution which may be simply modelled as the transient response of an elastic half-space due to the application of a normal, point force at the surface with arbitraly time dependence. Its time dependence con- sists of the recoil force of evaporating surface material and a longer time scale decreasing force due to the pres- sure exerted over an increasing circular area by the ex- panding plasma. The response of the surface to a nor- mal driving force with Heaviside time dependence has been derived by Pckeris (1955) and is shown in figure 5(a) for a source-to-detector displacement of 55 mm on an aluminium surface. If the source size, detector size, bandwidth and decreasing driving force are taken into ac- count then the constant negative displacement after the Rayleigh wave arrival is replaced by a displacement tend- ing towards zero over a longer time scale than the ini- tial rise. This is shown in figure 5(b) for a plasma regime laser-generated Rayleigh wave propagating in aluminium

332

6,I I I I I I I I I 0 5 10 15 20 25 30 35 40

Time ( vs )

Ibl I I I I I

L

501 I I I

15.5 18.0 16.5 17.0 17.5 18.0 18.5 Time ( 11s )

Flgure 5. (a) Theoretical surface dlsplacement in aluminium for plasma regime source (after Pekeris (1955)). (b) Experimental Rayleigh wave generated in plasma regime in aluminium.

and detected using a modified Michelson interferome- ter (McKie 1987) situated 50 mm away from the source. Note that the first part of the Rayleigh pulse is now pos- itive (outward) as opposed to negative (inward) for the thermoelastically produced Rayleigh wave.

The epicentral displacement for a normal driving force with Heaviside time dependence has been obtained by Knopoff (1958) and Sinclair (1979). The epicentral displacement for a normal driving force with 6 time de- pendence is obtained by differentiating KnopoFs solu- tion and this 6 shown in figure qa) . Figures 6(b) and (c) show experimental epicentral waveforms for a weak and a strong plasma generated above the surface of a 10 mm thick stainless steel sample. This source produces pos- itive (outward) displacements, whereas the thermoelas- tic source causes negative (inward) displacements. Note that, for a relatively weak plasma, the thermoelastic con- tribution is still significant

The directivity patterns for ultrasound generated in the plasma regime are shown in figure 7 and table 1 sum-

Laser-generated ultrasound

333

S J Davies et al

l o )

2

0

-1 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time ( 0s 1

( b l

140

100 120 :~~ 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

160

150

I 4 O 130 0.5 i 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time (&IS )

Figure 6. (a) Theoretical epicentral displacement for a 6-function normal force in Stainless steel. (6) Epicentral displacement of ultrasound generated in weak plasma regime in 10 mm thick stainless steel sample. (c) Epicentral dlsplacement of ultrasound genereted in strong plasma regime in 10 mm thick stainless steel sample.

marizes the partition of the total ultrasonic energy into the various acoustic modes as a function of Poisson’s ra- tio. As with the thennoelastic regime directivities, the amplitude of the wave mode recorded at a particular an- gle from the surface normal is the same, within experi- mental error, at all corresponding locations around a cir- cular ultrasonic source in an isotropic solid.

Experimental epicentral waveforms for a normal driving force with Heaviside time dependence have recently been obtained using a CO2 laser (Thylor et al 1990). The mechanism for the production of ultrasound in metals at the CO2 wavelength (10.6 fim) is different from those described here and will be discussed in a later section.

2.3. Constrained surface source

Investigations of the effect of various coatings on the

334

t Normal Stress - Theory

m M e o s u r e d Ibl

Normal slross

Figure 7. (a) Theoretical and experimental directivity of compression wave for plasma regime source In aluminium (after Cooper (1985)). (b) Theoretical directivity of shear wave for plasma regime source In aluminium (aner Cooper (1 985)).

irradiated surface upon the amplitude of the different wave modes have also been conducted (Hutchins et a1 1981a). Coatings such as oil, resin and grease as well as a transparent constraining layer were used. For all the coatings used, an enhancement in the amplitude of the various acoustic modes of > 20 dB was observed. The various modifications to the surface effectively result in a buried source at the coating surface boundary. A stress-free boundary at the surface no longer exists and this results in the development of normal stresses. This occurs at all incident power densities. The nature of the source is different and more closely resembles the plasma regime source. Ablation of the coatings was observed at power densities sufficient to form a plasma.

3. Recent experimental investigations wing laser-generated ultrasound

A number of investigations have been concerned with the laser generation of ultrasound on different mate- rials using lasers with different wavelengths and pulse profiles. These investigations are summarized in table 2 and, since the experimental arrangement was similar in each case, it is not considered worthwhile to describe them in more detail here. The interested reader is referred to the references. Other research has concen- trated on using a laser pulse with appropriate optics to achieve a mode of operation to suit a particular appli- cation and these are now described.

Wagner el a1 (1988) repetitively Q-switched the light output from an Nd :YAG laser during a single flashlamp

Laser-generated ukrasound

Figure 8. Measured compression wave directivity patems using an array with five elements and a delay of 0.25 fis between the successive elements: (a) 1.5 mm element spacing, (b) 2.0 mm element spaclng, (c) 2.5 mm element spacing and (d) 3.0 mm element spacing (after Bruinsma and Vogel (1988)).

cycle. This resulted in a series of pulses which could ir- radiate a solid and hence produce a series of ultrasonic pulses. Individual pulses from the @witched Nd :YAG laser had typical time widths of about U ) ns and pulse energies of 5-15 mT per pulse. These were focused onto a steel surface so as to produce a peak power density of

10' Wan-'. The smallest spacing between individual pulses was 18.7 ps, corresponding to a repetition fre- quency of about 50 WZ. Up to ten individual acoustic pulses were detected using a 500 kHz bandwidth acous- tic emission type piezoelectric transducer. The authors concluded that it was possible to generate a narrowband ultrasonic pulse with a modulation frequency the same as the optimum frequency sensitivity of an interfero- meter. McKie et of (1989) used a lenticular lens array to generate up to seven simultaneous line sources from a single Q-switched laser pulse. This had a similar effect to that described by Wagner et a f (1988) above, in that a series of ultrasonic pulses was generated and recorded at angles of 6oa and 80" from the plane of the sample surface.

Another variation on the use of multiple sources is to develop a laser array such that the laser source moves over the surface of a sample. A moving thermo- acoustic array has been considered theoretically by, for example, Westervelt and Larson (1973), Chotiros (1988) and Berthelot (1989). By varying the speed at which the thermoacoustic source moves over the surface, it is pos- sible to generate a plane wave in a particular direction. Also, as the incident laser energy is spread over a larger area, it is possible to remain within the thermoelastic regime whilst using larger laser pulse energies. Addison et al (1987) considered both an array of thermoelastic sources and the use of a Michelson interferometer to form an array of detectors. Bruinsma and Vogel (1988) directed an incident laser pulse into several optical fibres of differing length. They demonstrated (see figure 8) that the directivity of the laser array source could be

changed by wrying the distance between successive fibre ends on the surface of the sample.

Ishikawa et af (1990) set up an interference pattern on a sample surface by splitting an incident laser beam into two and directing the two beams onto the same re- gion of the sample surface as illustrated in figure 9(a). A similar amngement had been reported previously by Nelson et ut (1982). This resulted in the production of ultrasound (coherent acoustic phonons) with frequen- cies ranging from about 80 MHz up to about 1 GHz By changing the crossing angle of the laser beams, it is possible to change the fringe spacing and hence the frequency of ultrasound produced (see figures s(b) and (c)). The technique has also been used to generate periodic Rayleigh and Lamb waves in the 1-20 MHz region (Nakano and Nagai 1991).

Von Gutfeld et uf (19834 b) described a technique for increasing the efficienq of the thennmlastic gen- eration of ultrasound by effectively increasing the area over which the laser energy is distributed. This was aG complihed using an acoustic holographic lens structure which is basically a grated mask with absorbing and non- absorbing strips evaporated onto the sample surface. Cielo et uf (1985) described another method for increas- ing the area over which the laser energy is distributed (see figure 10). Using a positive axicon lens, an annular source was produced on a sample surface (12 mm wide of diameter 15 mm. This annular source produces a convergent Rayleigh wave which has i$ maximum am- plitude at the centre of the annulus. The amplitude of the Rayleigh wave at the centre of the annulus was a factor of 20 larger than the Rayleigh wave produced using the same sample from a line source with the same incident laser energy.

A new and exciting development in the area of laser- generated ultrasound has recently been described (Thy- lor 1990, Bylor et al 1990) whereby ultrasound is gen- erated by a mechanism different from those previously

335

S J Davies et a!

\ . -

(a)

0 Joo 0 590 m e < n m 1 rima < "I_ 1

Figure 9, (a) Experimental set-up for highfrequency phonons (after lshikawa et a' (1990)). (b) (i)? (ii), (iii) Time evolutions d the diffracted light intensity when the resonance mnditlons for n = 1, 2 and 3 respectively are satisfied. Samples are ethanol (i), methanol (ii) and (iii), containing 5 x lo-* mol I-' Na-fluorescein. The maximum diffraction efficiency is about 25% (i), 35% (ii) and 20% (iii). (c) Observed phonon signals, oscillating components of the diffracted light intensity: (i) to (iii) corresponds to (i) to (iii) In (b). Phonon frequencies in (i), (ii) and (iii) are 81.8, 164 and 245 MHz respectively,

described. The source originates from a COa laser, operating at 10.6 j" with a pulse energy of about 1 J. At this wavelength most metals reflect more than 98% of the incident light, such that the thermoelastic and consequently the plasma regimes for ultrasonic genera- tion are not achieved. The CO2 laser pulse interaction with a metal surface does, however, generate a plasma above the irradiated metal. The rapid formation of a plasma, generated within the rise time of the laser pulse, produces an intense shock wave both in the target and in the surrounding air. The sudden application of this shock wave results in a near-normal force at the sample surface and the subsequent generation of compression, shear and surface (Rayleigh or Lamb) waves within the material.

The mechanism for the plasma breakdown of the air, and hence ultrasonic generation, relies upon the presence of either oxides or small defects at the target surface (Walters et a1 1978) which, upon irradiation, produces a dense region of electrons above the sample surface (Weyl et a1 1980). Subsequent absorption, by in- verse Bremsstrahlung, of the incident radiation by these electrons results in a cascade ionization process and the formation of a laser-detonated plasma. Thc resultant

336

OIstancB from centre. mm

Figure 10. Schematic diagram of the experimental apparatus for the generation and detection of a converging SAW, a = axicon, d = dichroic mirror, s = sample. The inset shows a photo d the YAG annulus On the sample plane (after Cieio et a/ (1985)). (b) Amplnude d the detected SAW as a function of the distance d the probed point from the centre a( the YAG annulus.

motion of plasma depends on the initial radiation den- sity and that portion of the laser pulse still interacting with the plasma. Essentially two distinct plasma mo- tions exist: the laser plasmatron (LP) (Barchukov and Prokhorov 1976), which expands hemispherically about the target surface, and the laser-detonated wave (LDW) (Ramsden and Savic 1964), which rapidly propagates away from the target and subsequently expands spheri- cally in the air as the laser pulse finishes. The two regimes depend on the plasma electron density which, in turn, is a function of the initial radiation density. The most efficient ultrasonic generation results from an Lp source producing epicentral displacements com- parable to those produced in the plasma regime by a 200 mJ, about 15 ns wide, Q-switched Nd :YAG laser pulse focused to a 2 mm diameter source size pylar et a1 1990). Unlike the Nd : YAG laser source, the CO2 LF- ultrasonic source produces no surface damage.

The source strength from a CO2 laser thermoelastic interaction with non-metals is enhanced over that of, for example, an Nd :YAG laser thermoelastic source in

Laser-generated ultrasound

(vi) It produces simultaneously compression, shear and surface (Rayleigh) waves on thick (typically > 1 mm) samples. On thin samples (typically < 1 mm thick) it produces symmetric and antisymmetric Lamb WaveS.

(vii) Each acoustic mode has a characteristic direG tivity pattern which may be optically modified to suit a particular application. Using appropriate optics, the laser can be used to produce a train of pulses at a particular frequency.

(viii) Owing to its non-contact nature, the ultra- sonic source does not load the surface of the sample in which ultrasound is generated and avoids the problem of material degradation if a couplant is used.

(ix) It has the ability to monitor changes in ultrasonic travel time as a function of pressure and temperature.

(x) Used in association with interferometric de- tection (termed ‘laser ultrasonics’ by Monchalin et a1 (1988a,b)) a completely remote generation and detection system is available.

(xi) By virtue of its remoteness it allows achievement of faster scanning of a surface.

(xii) Using a laser of high repetition rate, it is possi- ble to generate ultrasound in a series of components on, for example, a production line, to increase the signal- to-noise ratio or to increase the inspection resolution.

I 0 5 10 15

Figure 11. Thsrmoelastic epicentral waveform recorded on Perspex with a COn laser source radius of 3.5 mm.

a metal because the ultrasonic source is buried a small distance, about 50-200 pm, beneath the surface. The normal component of stress into the material is not relieved as it is in the case of an Nd :YAG laser thermo- elastic source. This results in a positive precursor to the compression wave signal as shown in figure 11. Increas- ing the incident COz laser pulse energy at the sample surface results in the ablation of ions and electrons, with subsequent production of a plasma, resulting in an ultrasonic source similar to the plasma regime source described previously.

Time ips1

4. Applications of laser-generated ultrasound

The application of laser-generated ultrasound has been investigated, to varying degrees, since 1963. Most of the examples given are for ultrasound generated using lasers with wavelengths in the near infrared/visible wavelength regioa Before these areas are outlined, the properties of laser-generated ultrasound are summarized.

(i) It is a non-contacting source of ultrasound which permits the generation of pulsed ultrasound on hot or rough samples and in hostile environments.

(ii) It is a reproducible source of ultrasound from shot to shot.

(iii) Accurate time resolution is possible owing to its broadband characteristics.

(N) Ultrasound is generated by three mechanisms at the near infrared/visible wavelengths. thermoelastically, ablatively or beneath a constrained layer. Each of these mechanisms results in signifcantly different source char- acteristics. Thus a particular mechanism for ultrasonic generation may be chosen to suit a particular applica- tion. Plasma regime production of ultrasound results in some surface damage, which may make it inappropriate for use in some applications. At the CO2 wavelength, ultrasound is generated by a different mechanism and no surface damage on metals results.

(v) The laser pulse can be focused to produce a good approximation to a point source. Thus it may be used to generate ultrasound on very small samples or on areas inaccessible to piezoelectric probes.

4.1. Non-destructive testing (NDT)

The immediate and most obvious application of laser-generated ultrasound was in ultrasonic non- destructive testing of materials. Essentially the inspec tion techniques used are the same (see, for example, Krautkramer and Krautkramer (1977), McGonnagle (1982) and Silk er al (1987)) except that the piezo- electric probes are replaced by the laser source and a remote detector. Examples of the type of remote ultra- sonic detectors used include the capacitance transducer (see, for example, %ruby and Wadley (1!378), Aindow et a1 (1987) and Kim et a1 (1989)), the electromagnetic acoustic transducer or EMAT (see, for example, Frost (1979) and Hutchins and Wdkins (1985)) and interfer- ometry (see, for example, Monchalin (1985, 1986) and Paul et a1 (1987)). Thomsen ef a1 (1986) have described a technique for the detection of ultrasound generated by a picosecond laser from measurements of the change in reflectivity of the sample surface. A range of broad- band ultrasonic transducers were compared as detectors of laser-generated ultrasound by Dewhurst et aZ(1987a).

In a major@ of practical applications in indus- try, piezoelectric probes are sufficient for the detec tion and characterization of defects. Laser ultrasonics (laser generation and interferometric detection) can, by virtue of its properties, improve the already estab- lished techniques. It also allows further applications to be exploited which are not possible with piezoelectric probes. One disadvantage of laser ultrasonics for non- destructive testing is the surface damage caused when ultrasound is generated in the plasma regime. This damage may be acceptable on some components. When

337

S J Davies et a/

a COz laser is used to generate comparable amplitude ultrasound in metals, no surface damage results.

The earliest descriptions of the use of laser- generated ultrasound in locating flaws within a sample were given by Bondarenko er a1 (1976) and Von Gutfeld and Melcher (1977). Bondarenko et d (1976), using laser-generated ultrasound and interferometric detec- tion, noted that the technique could be used for thick- ness gauging, high-resolution flaw detection and for the determination of velocity and elastic moduli of materi- als. Von Gutfeld and Melcher (1977) demonstrated the flawdetection capabilities of laser-generated ultrasound by scanning the source over an aluminium cylinder con- taining several drilled holes. With the source and deteG tor directly in line with a defect, a change in the signal shape was observed. The effect of various constraining layers on the amplitude of the ultrasonic signal was also investigated. Other early papers which give examples of the use of laser-generated ultrasound in the location of flaws within a sample include Bar-Cohen (1979), Calder and Wdcox (1980) and Von Gutfeld (1980). In addition to nondestructive testing, Calder and Wdcox (1980) demonstrated that laser ultrasonics could be used to measure the acoustic velocity in a liquid metal column at the point of melting. Such measurements are useful in obtaining a more complete evaluation of the thermo- dynamic state of a liquid metal.

Dewhurst et d (1986), Hutchins and Hauser (1989, Hutchins el d (1987, 1989a) and Hu et a1 (1989) have reported on the use of laser-generated ultrasound and EW detection. Initial investigations by Dewhurst ef d (1986) and Hutchins and Hauser (1987) demonstrated the use of a I ~ ~ C I - E M A T system in defect detection and location. Later papers used ultrasonic tomographic re- construction to produce images of the internal s t r u ~ ture of objects containing defects. The technique was demonstrated using a cylindrical aluminium sample con- taining a entral cylindrical hole and the experimen- tal arrangement is shown in figure 12(u). Experiments were conducted with the hole containing air and various solid inserts to represent different acoustic impedance defects. Resulting tomographic reconstructions for two inserts are illustrated in figures 12(b) and (c).

Monchalin et a1 (1988a, b) have concentrated on developing a laser ultrasonic system for industrial applications. One drawback in the implementation of laser ultrasonics in an industrial environment for non- destructive testing is the relatively poor surface liniih of the majority of the components to be tested. A con- focal Fabry-PCrot interferometer has been developed which has been shown to operate for rough surfaces. However, on totally absorbing surfaces the interfero- meter cannot be used. 'lb overcome this problem, the incident Nd :YAG laser pulse, in addition to generating ultrasound, was used to clean @y ablating some of the absorbing coating) the irradiated area, thus allowing the Fabry-Pkrot interferometer to be used. This cleaning- for-receiving arrangement, illustrated in figure 13, was used in thc thickness gauging of seamless pipes and in locating internal flaws.

338

f

Figure 12. (a) Schematic diagram d apparatus- used for laserctm tomographic investigations (efter Hutchins et d (1987, 1989a)). (b) Laser-EMAT tomographic reconstructions d an aluminium cylinder with a Teflon insett. (c) As (b), cylinder wlth a glass insert.

Other papers reporting on the use of laser gener- ation and interferometric detection for non-destructive testing include Sarrafzadeh et a1 (1986) who directed the generating laser beam to the sample surface using a fibre optic cable. The acoustic transients were detected by a fibre-optic interferometer. Such a system, shown in figure 14, provides the basis for a safe, portable laser ultrasonic system. Generation of ultrasound using a laser pulse directed by an optical fibre has also been described by, for example, Burger er ai (1987), Prause and Hering (1987) and Peters et ai (1989). High p e r densities, about 80 mJ in about 10 ns, can be transmit- ted down large (about 1 mm) core diameter fibres at the Nd:YAG laser wavelength, but at present no viable 10.6 pm optical fibre is available.

Aindow et a1 (1983, 1984) and Cooper et a1 (1985) investigated the use of laser-generated ultrasound in the non-destructive testing of materials. In addition to locating defects using compression and shear waves, extensive investigations into the interaction of laser- generated Rayleigh w a v e with surface-breaking dc- fects were conducted (see, for example, Cooper et a1 1986). It was demonstrated that such defects could be characterized by analysis of the reflected and trans- mitted Rayleigh wave profiles. Similar laser-generated Rayleigh wave investigations have been reported by Maldague et ai (1986), Doyle and Scala (1987, 1990), Davies and Palmer (1990) and Davies (1991).

Maldague et a1 (1986), using the annular source described previously in association with interferometric

Laser-generated unrasound

Generated Ultrasound - Ea" mixing and focussing optlcs

Figure 13. Schematic diagram of the cleaningfor-receiving experimental setup (after Monchaiin et a/ (1988a, 1988b)).

Pulsed excitation

High power laser Multi-mOde

Micraxope Objective

ObjRtlve lens

Scanning directions

Specimen

Objective lens on the back sudace

lRiS - - Photo detector

Figure 14. Tw@fibre system for through-transmission laser ultrasound investigations (after Sarrafradeh et al (1986)).

detection, characterized surface-breaking and near- surface defects. Similarly Kuei Jen ef a1 (1985) and Cielo ef nl (1987), using the same annular source, char- acterized piezoelectric ceramics used in piezoelectric transducers. They illustrated that it was possible to de- termine from the Rayleigh waveform whether or not the piezoelectric sample had a surface electrode on it. Cielo er nl (1987) compared the various techniques available for the characterization of samples such as piezoelectric ceramics.

Doyle and Scala (1987, 1990), developing a tech- nique previously described theoretically by Mendelsohn ef a1 (1980) and Achenbach et a1 (1980), character- ized corner surface-breaking defects using the oscillation which occurs in the reflected and transmitted displace- ment waveforms. They demonstrated that the Iength of such a defect could be determined from a measurement of the period of oscillation in the reflected Rayleigh wave amplitude when dotted as a function of ultra-

Lorenz el uf (1989, 1990), Scruby (1989) and Scruby a a1 (1990) reported on the non-contact inspection of surface coatings. Lorenz ef nl (1989, 1990) (see figure 15) used the dispersive properties of the broadband laser- generated Rayleigh waves to determine the layer thick- ness from phase velocity data. Whilst Scruby (1989) and Scruby er a1 (1990) recorded oscillations in the ultrasonic waveform, which occurred directly after the Rayleigh wave arrival, m determine the coating thickness (see fig- ure 16). Scruby el ul (1990) also investigated the effect of a hardened region at the surface of an iron sample on the shape of the Rayleigh wave, as well as the degree of anisotropy in a variety of composites.

Symmetric and antisymmetric Lamb waves generated by a laser in samples typically < 1 mm thick may be used to locate flaws in pipelines. Harker (1988) and k s n e et a1 (1%)) described the methods used to detect defects using this wave mode. The use of Lamb waves to measure sample thickness will be discussed in a later section.

sonic frequency. Such a method has also been used by Davies (1991) to characterize surface-breaking defects in plane surfaces. Other techniques for the characteri- zation of surface-breaking defects using laser-generated

4.2. Laser ultrasonic velocity and attenuation measurements: determina*ion o~ elastic eOnStantS and grain size

Rayleigh waves have also-been describLd by 6aGes and Palmer (1990) and Davies (1991).

The elastic constants and moduli of a solid can be de- termined from measurements of the compression and

339

S J Davies et al

(a )

(d) ,p$ Highwarder

0 d ..... "* .............I .................

Fundamenla1

nms c arb. Uni~J I F.84YB"Cy

Figure 15. Rayleigh-wave response for two stiffening combinations: nickel on brass (a) and the characteristic dispersion curve for stiffening layers (b), reflecting tape on aluminium (c) and the characteristic dispersion curve for loading layers (d). V, is the shear velocity of the substrate, VR and VR the Rayleigh velocities of the substrate and the layer respectlvely.

shear wave velocities. In an isotropic solid, the veloo ities are the same when measured in any propagation direction, whilst the velocities are direction dependent in an anisotropic solid. If a single crystal is used, various ultrasonic wave modes may be propagated in a direo tion parallel to each of the principal crystal axes and the calculated velocities used to determine the elastic constants of the crystal (Musgrave 1970).

A laser source with interferometric detection en- ables the source and detector to be considered as point- like. This property, in association with the high reso- lution of laser-generated ultrasound, permits accurate measurements of the travel times of the various acous- tic modes. Accurate techniques for the measurement of velocity using piezoelectric probes at temperatures ranging from a few kelvin up to about 500 K are possi- ble (Butler et ai 1979, Carnevale et a1 1964, Eccleston 1990). Above these temperatures it becomes impracti- cal to use piezoelectric probes, and laser ultrasonics is a practicable alternative. It should be noted that laser ultrasonics could be used to measure the velocity within samples ranging in temperature from a few kelvin up to the samples' vaporization temperature.

Urious examples of the use of laser-generated ultrasound to measure the velocity of anisotropic solids at ambient temperatures have been given by Castagnede et a1 (1989). Ultrasonic slowness curves were con- structed for fibreglass/polyester and silicon (100) sam- ples using an iso-angular scanning geometry. Good agreement between the experimental and theoretical values was obtained, as illustrated in figure 17. How- ever, one should be aware that point source and receiver techniques measure the velocity of the wavefront and this can be very different from the velocity surface.

This is illustrated in figure 18, where measurements, performed by the present authors, of the wavefront ve-

340

locity surfaces in singlecrystal aluminium are presented. The experimental data, determined using EhiAT deteo tion of the normal and tangential motions, lie closer to the theoretical wavefront surface than to the veloo ity surface. Aluminium is only slightly anisotropic and the two surfaces only deviated slightly, but in more anisotropic media the surfaces can deviate geatly and the shear wave can exhibit cusps and be multi-valued in particular directions (Musgrave 1954). Iron, stainless steel and silicon have cusps and the wave surface can deviate by as much as 30% from the velocity surface. In uniaxially aligned fibrous composites, such as car- bon fibre reinforced plates, the deviation can be several hundred percent (Musgrave 1954, Rose er a1 1988).

Monchali et a1 (1987), LeBrun and Pons (1989, Dewhurst et ul (1988) and Aussel and Monchalin (198%) described measurements of the ultrasonic veloG ities in various samples using laser ultrasonics at ambient and elevated temperatures. Dewhurst et a1 (1988) mea- sured the velocity of compression waves as a function of temperature in aluminium, iron and steel samples (see figure 19). Significant changes in the velocity- temperature profile were observed in the iron sample both at its Curie temperature (about 768°C) and at the ferritioaustenitic transformation (about 910 "C).

The use of ultrasonic velocity and attenuation mea- surements to locate transition temperatures is an estab- lished technique p e l 1 et ul 1969). Current techniques generally use piezoelectric probes cemented to a crys- tal. It is possible that laser ultrasonics will be used in the future, especially on superconducting crystals, since, using velocity and attenuation measurements, it is pos- sible to determine the bandgap of the superconducting crystal p e l 1 et a1 1969). The present constraint on performing these measuremcnts on high-temperature superconducting crystals is that their coherence length is, at present, too low. Aleksandrov (1989) described how Mandlestamm-Brillouin (MB) light scattering has been used to measure the Rayleigh wave velocities on superconducting single crystals. A cw laser modulated at about 500 M H z using a Bragg cell is used to heat the surface periodically. This produces a Rayleigh wave whose velocity can be inferred from the frequency shift of a probe laser beam observed in a high-finesse Fdbry- Perot cavity.

Aussel and Monchalin (198%) measured the acous- tic velocities of a variety of solids using a cross cor- relation technique. Tbe diffraction correction owing to the finite size of the laser source and detector was estimated and subtracted from the time delay measure- ments. The absolute accuracy of the ultrasonic velocity in piezoelectric ceramics, metal-ceramic composites and single-crystal germanium ws typically < 1%, as indi- cated in table 3. This accuracy in the determination of the ultrasonic velocity resulted in corresponding errors in the elastic moduli of < fl%, which was judged to be as accurate as the more classical piezoelectric pulse- echo and resonance methods. They are compared in tables 4 and 5.

Measurements of laser-generated ultrasonic attenu-

Laser-generated ultrasound

I I I Rayleigh wave

0 10 2m3 Time . 1 2 3 4

Nominal coating IhiCkne%/no. of layers

Figure 16. (a) Surface waveforms from uncoated material and material covered by one to fNe layers of plasma-sprayed aluminium. Note attenuation and delay of Rayleigh wave in coated material and appearance of guided wave in coating (after Scruby (1989)). (b) Period of reverberation obseNed in coating is propoRionai to coating thickness (i.e. number of layers).

C,, = 16057 GPa 2 C,* = 9.224 GPa

C, = 3.026 GPa

- la1

- 2

- - 1 1

i , o ~ v e l m i t y ~ ' (ps/cm)

GI, = 16,521 GPa C,, = a435 GPa

Shear CB6 = 3.543 GPa

C, = 28.i72 GPa C, = 2.671 GPa CIS = 13.489 GPB

- (b) 3

*+ 't

.+ - - 1

2

1.0

1.0 tvelocity~~' (wJcm)

Figure 17. (a) Fibreglasdpolyester: scan in the isotropic plane (1,2) (after Castagnede et a/ (1989)). (b) Fibregiass/polyester: scan in the anisotropic plane (1,3).

ation in various solids, ranging from porous solids to metals, have been described by Rlschow (1988). Hutchins et a2 (1989b), Aussel and Monchalin (1989b), Scruby (1989) and Paul er al (1989). Ultrasonic attenu- ation measurements are well suited for the microstruc-

ture characterization of materials. The wide-bandwidth nature of ultrasound generated by a laser can be used to investigate materials with large and small grain sizes with comparable accuracy to piezoelectric techniques, as shown, for example, in figures U) and 21. Paul et a1 (1989) used the Rayleigh approximation, where the amount of scattering is inversely proportional to the fourth power of the ultrasonic wavelength, to de- scribe the scattering of the laser-generated ultrasound. The total attenuation spectrum obtained from various steel plates was related m the average grain size using this approximation. Other investigations in this field include that by Greenough ef al (1987), who used laser- generated symmetric Lamb waves to assess the s t rw tural change in metallic amorphous ribbons. The results are summarized in figure 22.

4.3. Determination of thin film thickness

The short temporal width of the laser-generated ultra- sonic compression wave pulse using a typical Q-switched laser may be used to determine the thickness of 6lms \< 1 mm thick This is not possible using a piezoelectric probe, as the temporal width of the ultrasound produced in this manner is typically > 1 ps. If the ultrasound is generated so as to reflect back and forth between the thin film surfaces, overlapping of successive echoes 00

curs which cannot be resolved without data processing. Use of a laser to generate non-overlapping compression waves to determine the thin film thickness has been described by Hutchins et al (1980) and %m (1984).

Hutchins et d (1980) used an Nd: YAG laser to gen- erate compression waves in a liquid layer. Timing be-

341

S J Davies er al

Table 3. A summary of the samples used and the accuracy achieved by Aussei and Monchalin (1989a) using laser ultrasonics for precise determination of the ultrasonic velocity and elastic constants.

Identification N203-AcSiC PZT A203-AiSiC A1203-N Ge Ge [No 1) (No 2) (100) (''1)

Shape

Thickness (mm1 Thickness Drecision (%) Time Drecision from laser Aignment Signal-to-noise ratio Time standard deviation Precision of time delay measurement Diffradion c a r e d i 0 n s Dispersion Absolute accuracy of velocity

Planar polished 3.249

0.08% 2.5 pm 0.32%

L 28 dB

0.032/1 (/Is) f in MHz 1.0% at 10 MHz -0.10%

Resolved Y 1% at 10 MHz

Planar polished 6.066

0.04% 2.5 pm 0.04%

L 23 dB

L 0.02%

L 0.05%

-0.03%

Negligible 0.1%

Culved unpolished 2.360

0.4%

0.15%

L 12 dB

L 0.1%

L 0.2%

-0.2%

Not resolved A few percent, iimked by dispersion

10 pm

CUNEd unpolished 3.800

0.3% 10 pm L 0.24% s 0.12% L 20 dB S 10 dB L 0.15% s 0.20% L 0.31% S 0.23% -0.10%

Not resolved A few percent, limited by dispersion

Planar polished 3.167

0.08% 2.5 pm L 0.16% s 0.12% L 40 dB S 20 dB L 0.02% S 0.03% L 0.16% s 0.12% -0.10%

Negligible L 0.24% s 0.20%

Planar polished 3.132

0.08% 2.5 pm L 0.17%

L 40 dB

L 0.02%

L 0.17% -0.10%

Negligible

L 0.25%

tween successive echoes enabled the thickness of the liquid layers to be determined independently of the liq- uid containment vessel. ?am (1984) demonstrated the use of about 1 ns wide compression wave pulse echoes in the measurement of thin 6lm thickness. The film thickness ranged from 14 to 262 pm and good agree- ment, shown in table 6, between calculated and mea- sured thickness was obtained. This technique also has application m the measurement of bond thickness in adhesively bonded samples (Segal and Rose 1980).

Another method for obtaining the thickness of thin films is to record the symmetric and antisymmetric Lamb waves generated by a laser source. Dewhurst et ai (198%) and Hutchins el ai (1989~) have demonstrated the use of Lamb waves in determining the thickness of metal sheets down to about U) pm thickness (see figure 23). Thylor (1990), using a CO2 laser and a mod- ified knife-edge detector, has demonstrated the CO2 laser production of Lamb waves in polymer sheets (see figure 24). This technique employs both the surface flexing of the film, to detect the symmetrical wave, and the elasto-optic interaction between the probe beam and the ultrasound to detect the antisymmetric wave. Such a technique is ideally suited to on-line monitor- ing of film thickness during manufacture. A prototype industrial system has been tested.

4.4. Use as a standard source The highly repeatable nature of the laser source and its ability to change its source characteristics makes it an ideal tool to use as a standard source. Use of the laser source in this manner has been described by Hutchins el ai (1981b), Scruby er a1 (1981) and Hutchins and Tam (1986). In acoustic emission studies, the laser source can be used to model a particular crack mode. Thus it is possible to generate a waveform using the laser source

342

" '"3 /I1101

Figure 18. Measurement d the shear wavefront velocity surfaces In single-crystal aluminium. Also shown are the velocity and wave surface profiles calculated for the given elastic constants and density.

similar to that received in an acoustic emission study. The properties of the irradiated surface and the incident laser pulse define the type of source, which in tum can be used to determine the crack mode stress intensity. For example, the stress field around the thermoelastic source, to a first approximation, appears to be of a si,- ilar form to that surrounding a vertical mode I crack at the surface of a plate, in that the latter is also com- posed principally of horizontal force dipoles (see figure 25). When a laser pulse is incident upon a transfer block with a 4.5' notch it generates a stress field that includes a vertical force dipole which simulates a hori- zontal mode I crack (Scruby er al 1981). It is also used to calibrate acoustic emission transducers (Scruby er al 1981, Scruby 1989).

Another area in which the laser source may be

Laser-generated uhiasound

6.4 - . E 6.3

X 6.2

3 6.1

"2

0 5, I ._ P

6.0

Table 4. Sound velocities in germanium at 25 "c (after Aussel and Monchalin 1989a).

- - - - -

Sound velocity (m s-')

Method

Contact pulse-echo method 5561 492 1 3550 5323 Resonance method 5587 4959 3536 5350 Laser ukrasonics 5577 4937 3522 5323 (measured)

Table 5. Elastic constants of germanium at 25 "c (after Aussel and Monchalin 1989a).

Elastic constant (N m-*)

Method c t 1 c t 2 c44

Contact pulse-echo method 1.289 0.483 0.671 Resonance method 1.316 0.509 0.669 Laser ultrasonics 1.297 0.514 0.660 (calculated using data from previous table)

used as a standard source is in the characterization of other transducers. This involves comparing the ul- trasonic signals produced from the source to be Cali- brated with a standard laser source (Cooper et a1 1984, LeBrun and Pons 1987, Dewhurst ef a1 1987a). Using a wide-bandwidth non-contacting detector such as an interferometer, it is possible to record and compare the ultrasound produced by other transducers with that produced by a laser source. Comparison of the signals allows the determination of the effective stress distri- bution representative of the transducer used. Interfer- ometerS may also be used to characterize the ultrasonic field of piezoelectric transducers (Scruby 1989).

4.5. Acoustic microscopy Acoustic microscopy as a technique for the characteri- zation of a surface has been described by, for example, Lemons and Quate (1979). Discussions on the use of a laser pulse to replace the ZnO transducers and coupling medium (usually water) in generating high-frequency ul- trasound may be found in Wickramasinghe et a2 (1978), Hutchins and %m (1986), Lyamshev (1987), Favro et a1 (1988) and Hauden et al (1989). Use of an interfer- ometer to detect the ultrasound would greatly enhance the technique. To generate the high frequencies either a laser with a short (picoseconds) pulse width or repetitive @switching, at a frequency sufficient to produce ultra- sonic pulses with a fundamental frequency of hundreds of megahertz (Wickramasinghe et a1 1978), is required.

4.6. Miscellaneous applications

In addition to the applications of laser-generated ultrasound described already, there are several other applications which cannot be as conveniently classified. This section gives a summary of the other applications published.

(a)

6.5 I* Dural

+ +

+ +

+ +

+

5.9 I I I I I+ 0 1w 200 3w 4w 5w I

Sample temperature ('C)

(b) 6.0 p 1

I

E m0 5.61 - + +

5,2t , , , ':+i*J 5.0

0 2w 4w 6M) ew 1 m Sample temperature (*C)

Figure 19. Temperature dependence of the compression velocity in (a) Duval and (6) iron. For the iron profile the markers (i) at about 768 "c and (ii) at about 910 C correspond to the samples' Curie temperature and the ferritioaustenfiic transformation respectively (after Dewhurst etal (1988)).

Yang (1974) proposed that the generation of stress waves in thin metallic films by a Q-switched laser pulse could be used in the detonation of high explosives. This made use of the high pressures developed by irradiating a metallic surface through a constraining layer with a laser pulse. Fairand and Clauer (1979) suggested that the high pressures generated could be used to produce significant changes in the in-depth microstructure and mechanical properties of metal alloys. More recently, Bournot ef a1 (1987) have investigated the surface hard- ening of aluminium, titanium and steel after a high- pressure shock wave was generated using an Nd : YAG laser. They suggested that large shocks could be created without affecting the surface quality of metals.

Hutchins and Tam (1986) described a method for the potential monitoring of annealing processes by ob- serving the laser-generated acoustic signal. Of particu-

343

S J Davles et al

la)

loo I t

0.1 1 10

Frequency < MHz )

I

1.0 I! I I I I 10 20 30 40 50

Frequency ( MHz 1

Figure 20. (a) Compressional ultrasonic atlenuation measured by laser ultrasonics on a PZI ceramic sample, illustrating the spherical wave liml. The detecting sensor is a laser- heterodyne interferometer. Also shown is the variation obtained wlh a piezoelectric transducer after diffraction correction and smoothing (after A u s s l and Monchalin (1989b)). (b) Compressional ultrasonic attenuation d a hot-roiled steel plate measured by a conventional piezoelectric transducer and by laser ultrasonics using a Fabry-PBrot velocity interferometric receiver.

Table 6. Summary of results for five samples CT type 302 stainless steel films d different thickness (after Tam 1984).

3

Thickness by Echo round Ultrasonic Fined Effective ultrasonic micrometer trip time velocity thickness attenuation a! ( rm) (ns) (io5 cm s-l) (pm) about 300 MHz (f2 Pm) (neper cm-l)

14 4.24(2) 6.6(9) 12.57(13) 73 26 8.57(4) 6.1 (5) 25.4(3) 31 76 25.66(5) 5.9(2) 76.1(8) 23.0

128 42.7(1) 6.00191 126,61131 23.6 262 m.4i2j 5.93(6j (reference) 19.7

lar interest was the laser annealing of silicon during its manufacture. Another use they described was the deter- mination of charge distribution in dielectrics. The laser pulse was used to provide reproducible shortduration pressure transients within the dielectric. This pressure pulse produced a local electric field (which was deter- mined from measurement of the current in an external circuit) which could be related to the charge density. Deev and F'yatakov (1988) described the generation of acoustic waves by the action of a pulsed laser beam on a space-charge grating formed in a bismuth germanate crystal. It was stated that a special kind of photoacous- tic effect with memory occurs in such a photorefractive crystal, which could have applications to acousto-optical signal processors.

Harata and Sawada (1988) have used laser- generated ultrasound in the determination of candle flame parameters. A laser-induced grating (Harada et a1 1987) was used to generate the ultrasound, which was optically detected using Bragg diffraction of another

344

5 10 30 FleWOWIMW

Figure 21. Comparison of the measured attenuations in three lnconel samples using the laser and piezoelectric source methods (after Telschow (1988)).

laser probe (see figure 26). Measurement of the ultra- sonic velocity enabled the flame temperature to be de- termined by assuming the flame to be an ideal gas. This research has application in the remote monitoring

Tektronix 464

7812 A-D

1w j. .5 -D 5 0 - a e

- - D v) - s o *

-50 2

-100

(b)

1DOvm hotpolythenebubble -

-

(C)

0 4 8 12 16 20 Arrival time ( vs )

25

B - 656K 220 minutes C - 769K 1W minutes 20

15

10

5

n " 0 25 50 75 IW

Figure 22. (a) Experimental arrangement for assessment of the structural change in metallic amorphous ribbons (after Greenough eta/ (1987)). (6) Typical acoustic waveform measured by the magnetomechanical sensing coil. This waveform was produced for a sample when the applied magnetic field was about 32 Oe (about 2.5 kA m-'). (c) Variation of signal amplitude with magnetic field applied to the ribbon. All samples 4 B and C were based on Metgias 2605 SC. Sample A was in an as-received state from the manufacturers, sample B had been heat treated at 660 K for 220 min, and sample C had been heat treated at 773 Kfor 100 min (after Greenough et a/ (1 987)).

of real combustion diagnostics. Von Gutfeld (1980) proposed a scheme for medi-

cal diagnostics using laser absorption at a rubber-water interface in order to perform acoustic tomography. As far as the authors are aware, the technique has not re- ceived any further investigation. Johnson et al (1987), Johnson (1988) and Carlson and Johnson (1988) con- sidered theoretically, using finite-element techniques,

0 4 8 12 16 20 nme(wi

- (c),

Laser-generated ukrasound

(b)

-1.2 I(I(IJIJJII 0 2 4 6 8 1 0

n m e ( p s )

I

0.0 0 100 200 3w

Frequency ( kHz )

Figure 23. Lamb waves generated by a pulsed laser and detected with an interferometer 24 mm away from the acoustic source in aluminium samples of thickness (a) 425 pm and (6) 80 pm (after Dewhurst etal (1987b)). (c) Frequency dependence of the square d the group wave velocity. A least-squares fd through the wigin and using the first four data points gives a thickness of 26.5 pm, compared with a micrometer measurement of 27 pm.

moving plastic bubble anb detected using a He-Ne laset beam deflection technique.

and experimentally the laser generation of ultrasound in a weld pool. The aim of these investigations is to develop a truly automated welding system using in- formation obtained from ultrasonic measurements as a feedback device to the welding process. Bnkovsky and Pelzl (1989) generated resonant elastic modes in bonded materials and in optical fibres by the thermo- elastic mechanism. The acoustic spectra excited were observed to be unique for a particular material. This

345

(b)

(i) Simulated emisdon 80urce (laser) Iksm enww 41mJ. 3mm apenure. depth 25"

g 100

$ 0

5 Timelua - lW

(iii) Model d crack source

.. 0

15 4 rimelus

(iii) Model d crack source

.. 0

15 4 rimelus

C%7WeSiOn Shear way* wave

Flgure 25. (a) Schematic diagram lo compare acoustic generation ty ( i ) absorption ci pulsed laer radiation in a notched specimen, and (ii) a horizontal mode I crack in a half-space (after Scruby et a! (1981)). (b) Acoustic waveforms for (i) ultrasound generated ty pulsed laser in a notched specimen, (ii) ultrasound generated by propagating cleavage crack in steel 'Yobell' specimen and (iii) theoretical waveform for a crack source in a half-space.

could have applications in the control or evaluation of the elastic parameters for optical fibres.

One application for which no research has at present been reported is in the use of laser ultrasound for the determination of residual stress (described by, for ex- ample, Pa0 er a1 (1984). Allen and Sayers (1984) and Pritchard (1987)). Precise measurement of velocity as a function of propagation direction within a particular sample could be achieved using a point source and p i n t detector. The change in velocity with angle in rolled steel owing to its residual stress has a magnitude of about Changes owing to the crystal structure also occur but these may be eliminated if time measurements can be made to an accuracy of or better. Laser ultrasonics has the potential for achieving the requisite accuracy, thus enabling accurate determination of the residual stress within a material.

346

(b)

n 1 a - I II I

Time ( ns I

0

Figure 26. (a) Experimental arrangement for the production d a laser-induced grating (LIO) on a candle flame (after Harata and Sawada (1988)). (b) Typical singleshot response cf UG signal and various background signal traces (after Harata and Sawada (1988)). (c) Acoustic wavenumber versus acoustic frequency in a candle flame (after Harata and Sawada (1988)).

5. Conclusions

brious aspects of the laser-generation of ultrasound have been described with particular emphasis on its areas of application. Descriptions of the different mech- anisms for, and properties of, laser-generated ultra- sound in different generation regimes were given. This was followed by a description of the many applications of laser-generated ultrasound which is the main research activity at present.

Most of the research has been conducted in a lab- oratoly environment, but laser ultrasonics h a technique which seeks application in an industrial environment. Wing to the great expense and safety requirements for a portable laser ultrasound system, it seems unlikely that it will replace those areas of application where the use of piezoelectric transducers is adequate. Thus laser ultrasonics may be developed for applications in which it is not possible to use piezoelectric transducers, such as in hostile environments or on awkwardly shaped

Laser-generated ultrasound

Carnevale E y Lynnworth L C and Larson G S 1964

Castagnede B, Sachse Wand Thompson M 0 1989

Chotiros N P 1988 J. Acoust Soc. Am. 83 2145 Cielo I: Nadeau F and Lamontagne M 1985 L n ~ o n i c s 23

55 Cielo I: Maldague X, Krapez J C and Lewak R 1987

Nondestmcrive Characteizarion of Matm'als .U Roc.

J. Acoust. Soc. Am 36 1678

Lnh.nsonics Int Confi Roc. p 71

n 711

components. Examples of possible future applications have been given by Scruby (1989).

One area of application which may be developed in the near future is the active testing of products, such as hot rolled steel, aluminium, plastics or possibly paper along a production line. The ultrasonic information may be used as an active feedback element in the manufac- turing process. It may also be used to test small com- ponents such as solder joints on printed circuit boards. In the USA, there is a programme under way to test nondestructively, using laser ultrasonics, aircraft wings constructed from composite materials. The possibilities for the use of laser ultrasonics are limited only by the needs of industry and the imagination of researchers in this field.

References

Achenbach J D, Gautesen A K and Mendelsohn D A 1980

Addison R C, Graham L J, Jinebarger R S and Tittmann B

Aharoni A, 'lhr M and Jassby K M 1987 Lnfmonics Inf.

IEEE Pam. Sonics Ultraon 27 124

R 1987 E E E UIimonics Symposium p I109

Conf: Proc. D 623 AindowA Dewhurst R J, Palmer S B and Hutchins D A

1980 SPIE 236 478 Aindow A M, Dewhurst, R J and Palmer S B 1982 Opf.

Aindow A M, Cooper J A, Dewburst R J and Palmer S B

Aindow A M. Dewhurst R J. Palmer S B and Scrubv C B

Commun 42 116

1983 UIirasonics Inf. Confi Roc. p 20

1984 Non-desfr, resting Ink 17 329 Aindow A M 1986 PhD Thesis. Universitv of Hull Aindow A M, Cooper J A, Dewhurst R fan, Palmer S B

1987 3. Phys. E: Sci Inshum 20 204 Aleksandrov V V 1989 Phys. Left. A 142 307 Aleshkevich V A, Gaivoronskii V Ya and Matveev A N 1987

Allen D R and Sayers C M 1984 LntmoniCr 22 179 Askar'yan G A and Yurkin A V 1987 Sov. Phys.-Acoust. 33

Amel J D, Le Brun A and Baboux J C 1988 Urmonics 26

Ausrel J D and Monchalin J P 198% Ufrasonics n 165

Sov. Phys.-Acousr. 33 457

219

245 ..

- 1989h 3. Appl Phys 65 2 1 8 Bar-Cohen Y 1979 f i r . J. Non-desrr Testine 21 76 Barchukov A I and Prokhorw A M 1976 h p . Fir Nauk 119

Bazylev P Y Bondarenko A N and Lugovoi V A 1989 Sox J.

Berthelot Y H 1989 3. Acowt. Soc. A m 85 1173 Bondarenko A N, Drobot Yu B and Kruglov S V 1976

Bourkoff E and Palmer C H 1985 AppL Phys. Lm. 46 143 Bournot Ph, Dufresne D, Autric M, Giovanneschi-Testud P

Bresse L F and Hutchins D A 1988 Rev. Rog. Quam

Bruinsma A J A and Vogel J A 1988 Appl. Opt. 27 46W Burger C P, Dudderar T D, Gilbert J A, Peters B R and

Butler B, Palmer S B and Primavesi G J 1979 Utrasonics 17

Calder C A and Wilcox W W 1980 Marer, Evol. 38 86 Canella G, Pedicelli L and addei M 1989 Non-desfr, Tesfing

Carlson N M and Johnson J A 1988 Rev. F'rog. Quam

425

Non-destr Rsfing 25 253

Defe?doskopiya 6 85

and Caquerelle C 1987 SPIE High Power Lasen- 801 308

Non-destr: EvaL A 7 629

Smith J A 1987 J. Nondest, Eval. 6 57

249

(Pmc. i'ivelfrh World Con[) p 643

Non-destr EvaL 7 1485

r Cooper J A, Aindau A M, Dewhurst R J and Palmer S B

Cooper J A 1985 PhD Thesis, University of Hull Cooper J A, Crosbie R, Dewhurst R J and Palmer S B 1985

1984 LEE Roc. A 131 275

~ a s o n i c s Int. Cont Pmc. D 207 Cooper J A, Dewhurst R J and-hmer S B 1986 PhiL Z-ms.

Davies S J 1991 PhD Thesis, University of wan*ick Davies S J and Palmer S B 1990 Nondesh W. EvaL 5 227 Dew V N and Pyatakov P A 1988 Sov. Phys.-Acousf. 34 358 Dewhurst R J, Hutchins D A, Palmer S B and Scruby C B

Dewhurst R 3, Edwards C and Palmer S B 1986 AppL Phys.

Dewhurst R J, Edwards C, McKie A D W and Palmer S B

- 19871, hol . Phvs. Le& 51 1066

R. Soc. Lond. A 320 319

1982 J. Appl. Phys. 53 4064

Left. 49 374

1987a Lnimonics 25 315

- 1988 J. XppL Pity. 63 1225 Dewhurst R J and Al' Rubai W S 1989 Lntrasonics 27 262 Doyle P A 1986 J. Phys D: AppL Phys 19 1613 Doyle P A and Scala S M 1987 Ln!msonics In& 87 Con5

Eccleston R S 1990 Private "munication Eesley G I Clemens B M and Paddock C A 1987 AppL

Roc. 407 1990 U W O S O ~ ~ C S 7.s n -

Phvs Lelt. 5Ll 717 ~~

Fairani B P and Clauer A H 1979 1. Appl. Phys. 50 1497 Fassbender S, HolTmann B and Arnold W 1989 Mater, Sci

Fmo L D, Shepard S M, Kuo P K and Thomas R L 1988

Frost H M 1979 Physical Acoustics MI 14 (New York

Eng. A 122 37

Rev. Pmg. @ant Non-destc Eva!. A 7 225

Academic) Ghosh A P and Hurst J E 1988 J. Appl. Phys. 64 287 Grahn H 'E Maris H J and BUC J 1989 E E E J. Lhuurt -

Electron 25 ?S62 Greenough RD,bwhurs t R J and Edwards C 1987 J. AppL

Phvs 62 4728 Harada Y, Kanemitsu Y, "aka Y, Nakano N, Kurcda H

and Yamanaka K 1989 J. Phys. D: AppL Phys. 22 569 Harata A, Sawada T and Gohshi Y 1987 lapan. 3. Appl

Phys. Suppl 26 41 Harata A and Sawada T 1988 Japan I. Appl. Phys. SuppL 27

223 Harker A H 1988 Elastic Waves in Soli& (Bristol: Adam

Hilger) Hauden D, Cretin B, Mahmoud A, Lesne J L, Xie Wand

Vermersch S 1989 UIimroniCs Inf. ConJ Roc p 152 Hu J K, Hutchins D A, Ungar J, Zhang Q L and Mak D K

1989 MateK Eval. 47 136 Hutchins D A 1981 J. Acoust. Soc. Am. 70 W62 Hutchins D A, Dewhurst R J and Palmer S B 1980 Acousk

- 1981a Lntrasonics 19 103 Lett. 4 95

Hutchins D A, Dewhurst R J, Palmer S Band SCNW C B

Hutchins D A and Wlkins D E 1985 J. AppL Phys. 58 2469 Hutchins D A and Z?m A C 1986 EEE-UFFC 33 439 Hutchins D A and Hauser F Non-desmctive Esting

Characferizarion of Materiak II Roc D 725

1981b AppL Phys. Len. 38 677

Hutchins D A, Young-R P, Stoner R, Janien D, Hu 5 K and Zhang Q L 1987 EEE Lnfrasonics Symp. Pmc. 1037

347

S J Davies et a/

Hutchins D A, Hu J K, Young R P, Stoner R, Jansen D and B a n g Q L 1989a J. Acoust. Soc A m 85 747

Hutchins D A, Young R P and Ungar J 1989b Ultrasonic Methods in Evaluation of Inhomogeneous Mafenah (MTO Advanced Study Inslintre) Seria E no 126, p 353

Hutchins D A Lunderen K and Palmer S B 1989c J. AcousL Soc. Am. 'ss 1441

Ishikawa K Takahashi Y. 'Emaka U, Fukuda Y and Hashi T 1990 Opt. Commm. .76 81

EvaL 6 147 Johnson J A, Carlson N M and Lolt A 1987 J. Nondesrr.

Johnson J A 1988 Rev. Prog @ant. Non-dese EvaL 7 1477 Kanemitsu Y, Harada Y, Tanaka Y and Kuroda H 1988

J. AppL Phys. 63 4751 Kanemitsu Y, Harada Y, Tanaka Y, Nakano N, K d a H

and Yamanaka K 1989 Japan. J. AppL Phys. SuppL 28 234

Kim K Y, Niu L, Castagnede B and Sachse W 1989 Rex Sci In ." 60 2785

Knopoff L 1958 J. AppL Php. 29 661 Kolomenskii AI A 1988 Sov. Phys.-Acoust. 34 504 Krautkramer J and Krautkramer H 1977 Ulrmsonic Tesling of

Krylov V V, Ponomarev E P and Shtentsel T V 19% Sov.

Kuei Jen C, Cielo P, Maldague X and El-Assal K 1985 J.

LeBrun A and Pons F 1987 Pmc Fourth European Con!

Lemons R A and Quate C F 1979 Physical Amiulics vvI

Lesne J L, LeBNn A and Pons F 1990 Nondese Tesl. EvaL 5

Lord A E 1966 J. Acoust. Soc. Am. 39 650 Lorenz M, Bruinsma A J, Stelwagen U, Vogel J A and

Materia& 2nd edn (Berlin: Springer)

Phyx-Acousi. 41 46

A m C e r m Soc. C 68 146

Non-destructive %ling MI 3, p 1593

XW

171

Berkhout A 1989 Non-Destnrctive i'hling (Proc Tn!e@h World Con/.) p 1449

Nondesfx Tat. EvaL 5 187 LDrenz M, Vogel J A, Bruinsma A J and Berkhout A J 1990

Lyamshev L M 1987 Sm. Phys.-Usp. 30 252 Maldame X Cielo P and Jen C K 1986 Mater: Eval. 44 1120 McGohagle W J 1982 Non-Destructive Tesiing 2nd edn

McDonald F A 1990 Appl. Phys. Len. 56 230 McKie A D W 1987 PhD Thesis, University of Hull McKie A D W Wagner J W, Spicer J B and Penney C M

Mendelsohn D A, Achenbach J D and Keer L M 1980 Wave

(New York Gordon and Breach)

1989 Ultrasonics 27 323

Motion 2 277 Monchalin J P 1985 Appl. Phys. Lett. 47 14 - 1986 LEEE UFFC 33 485 Monchalin J P, Heon R, Bussiere J F and Rrahbakhsh B

1987 Non-destrucfiw Tesling Churacferiurtwn of Materials XI Roc. p 717

Monchalin J P, Aussel J D, Heon R, Bussiere J r; Bouchard P, Guevremont J and Padioleau C 19% IEEE Ultrasonics Symposium Roc. 1041

Monchalin J P, Aussel J D, Bouchard P and Heon R 19886 Rev. Prop. Quant. Non-desk Eval. 7 1607

Monchalin J P, Aussel J D, Heon R, Jen C K, Boudreault A and Bernier R 1989 J. Nondesk EvaL 8 121

Musgrave M J P 1954 Proc R SOC A 226 356 Musgrave M J P 1970 C!ystul Acoustics (San Francisco, CA:

Nakano H and Nagai S 1991 Ultrasonics 29 270 Nelson K A, Miller R J, Lutz D R and Fayer M D 1982

Neubrand A and Hess P 1989 Mater: Sci En& A 122 33 Pa0 Y H, Sachse W and Fukuoka H 1984 Physical Acousfics

Paul M, Betz B and Arnold W 1987 Appl. Phys. Left. 50 1569 Paul M, Peukert H and Arnold W 1989 Non-destrucfive

348

Holden-Day)

J. AppL Phys. U 1144

vol XVII

Tesling (Pmc TwIfIh World Con!) p 261 Pekeris C L 1955 Roc. Nat. Acoust. Soc. 41 469 Peters B R, Gilbert J A and Dudderar T D 1989 Proc. 1989

Prause L and Hering P 1987 Laser und Optoelektronik Pritchard S E 1987 Non-destr: Testing I d . 20 57 Ramsden S A and Savic P 1964 Nature 203 1217 Rose J L, Pilaski A and Tkng A A 1988 New Dreclion on

SEM Spring Con/. p 794

h'DE of Advanced Maierials (ASME) ed J L Rose and A A Beng

Rose L R F 1984 J. Acoust. Soc Am. 75 723 Rudd M J 1983 Rev. Pros @ant. Non-dest. EvaL B 2 1763 Sarrahadeh A, Churchill R J and Niimura M G 1986

Acousto-ultrasonics: %oy and Applicatwns ed J C Duke

Scruby C B and Drain L E 1990 Laser Ultrasonics, Techniques and Applications (sristol: Adam Hilger)

Scruby C B and Wadley H N G 1978 1. Phys. D: AppL Phys, 11 1487

Scruby C B, Dewhurst R J, Hutchins D A and Palmer S B 1980 J. Appl. Phys. 51 6120 - 1981 Research Tchniqties in Non-destmtive i'hling vvl V (New York Academic)

Scruby C B 1989 Ultrasonics 27 195 Scruby C B, Brocklehurst F K, Moss B C and Buttle D J

1990 Nondesir: Test. Eval. 5 97 Segal E and Rose J L 1980 Research Techniques in

Non-deshuctive Testing MI lV (New York Academic) Sessler G M, Gerhard-Multhaupt R, West J E and yon

Seggern H 1985 J. AppL Phys. 97 119 Silk M G, Stoneham A M and Emple J A G 1987 The

Reliability of Non-destructive Inspection (Bristol: Adam Hilger)

Sinclair J E 1979 1. Phys. D: AppL Phys. 12 1309 Sullivan J M, Ludwig R and Stern S 1988 IEEE Utrasonh

Svmu. Roc. LI 481 Tam Tam A C and Coufal H 1983 Auul. Phvs. Lea. 42 33

e 1984 AjpL Phys. Left. 45 510

' bnkmky N and Pelzl J 1989 ililmoiics Int. On! Roc.

Taylor G S 1990 PhD Thesis, University of Warwick Taylor G S, Hutchins D A, Edwards C and Palmer S B 1990

Rlschow K 1988 Rev. Rog. Quant. Non-desk EvaL 7 1211 Thomsen C Strait J. Mrdenv Z Maris H J and Taw J 1984

P 31

Ultrasonics 28 343

, , Phys. Rkv. LetL s? %39

Thomsen C, Grahn H T Maris H J and Tauc J 1986 Phys.

Tittmann B R, Graham L J and Linebarger R 1987 B E E

TNdl R Elbaum C and Chick B B 1969 Ultrasonic Methods

Rev. B 34 4129

Ultrasonics Symp. Roc. 1121

in Solid Sfate Physics (New York Academic)

257 Von Gutfeld R J and Melcher R L 1977 Appl. Phys. Leu. 30

Vbn Gutfeld R J 1'380 Ultrasonics 18 175 Von Gutfeld R J, Mgliotti D R, Ih C S and Scott W R

1983a Appl. Phys. Lett. 42 1018 - 1983b Rex Rog. puutt Non-dest,: EvaL B 2 1797 Wagner J W, Deaton J B and Spicer J B 1988 Appl. Opt. 27

4696 W t e r s C T Barnes R H and Beverly R E 1978 J. Appl.

Phys. 49 2937 Wang M R and Meng H C 1969 1. Appl. Phys. 66 205 Westervelt P J and Larson R S 1973 J. Acoust. Soc. A m 54

Weyl G, Pirri A and Rwt R 1980 J. AlAA 19 460 Wickramasinghe H K, Bray R, Jipson V, Quate C F and

Salcedo J R 1978 Appl. Phys. Lett. 33 923 Yang L C 1974 J. AppL Phys. 45 2601 Zhang S Y, Paul M, Fassbender S and Arnold W 1990 Res.

Non-desk EvaL 2 169

121