Laser ultrasonic monitoring of ceramic sinteringK. L. Telschow, J. B. Walter, and G. V. Garcia Citation: Journal of Applied Physics 68, 6077 (1990); doi: 10.1063/1.346895 View online: http://dx.doi.org/10.1063/1.346895 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/68/12?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Application of Laser AE Technique to Monitoring of Sintering in Structural Ceramics AIP Conf. Proc. 760, 1706 (2005); 10.1063/1.1916876 Ultrasonic monitoring of laser damage in fused silica Appl. Phys. Lett. 78, 3403 (2001); 10.1063/1.1376431 Laser ultrasonic monitoring of steel microstructure at elevated temperatures J. Acoust. Soc. Am. 97, 3372 (1995); 10.1121/1.412658 Laser ultrasonic measurements in highly porous materials: Sintering of ceramics J. Acoust. Soc. Am. 86, S110 (1989); 10.1121/1.2027294 Sintering of new oxide ceramics using a high power cw CO2 laser Appl. Phys. Lett. 44, 1132 (1984); 10.1063/1.94666

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

75.102.71.33 On: Mon, 24 Nov 2014 19:04:21

Laser ultrasonic monitoring of ceramic sintering K. L. Telschow, J. B. Walter, and G. V. Garcia Idaho National Engineering Laboratory. EG&G Idaho. Incorporated. Idaho Falls. Idaho 83415-2209

(Received 15 May 1990; accepted for publication 21 August 1990)

Noncontacting laser-ultrasonic measurements of the sintering of ceramics in real-time are described. Lasers are used for both generation and detection of ultrasonic waves propagating through the material. A pulsed laser generates the ultrasonic wave from thermoelastic absorption at the material surface. Detection is accomplished by a confocal Fabry-Perot interferometer, which is insensitive to the speckle nature of the scattered light from the sample surface and can obtain measurements from samples with rough or diffusely reflecting surfaces. The densification of the ceramic sample is determined by changes in the bulk longitudinal wave velocity and the sample shrinkage.

I. INTRODUCTION

Ceramic components are normally consolidated from powder compacts by sintering at high temperatures. Often the properties of these components are determined by their residual porosity. Improvements in efficiency of production and in the consistency of properties of ceramics could be achieved if the porosity could be monitored, and con­trolled, during the sintering stage. Noncontacting measure­ments of sintering are presented in this paper.

In a material with porosity, the effective compliance is increased due to the presence of voids, which means that elastic modulus and the velocity of ultrasonic wave prop­agation decrease with increasing porosity. The velocity of longitudinal waves in a solid material is given by VI = (M / p) 1/2, where M is the effective elastic modulus and p is the sample density. In a porous ceramic the elastic constants are complicated functions of the host material's Young's modulus and Poisson's ratio, and the number, size, and shape of the voids. I However, for some materials, particularly sintered ceramics and metals, a monotonic in­crease of longitudinal wave velocity with porosity is found. 2

•3 Ultrasonic methods, therefore, have potential for

determining porosity in ceramics in a nondestructive man­ner. However, a noncontacting measurement technique is required for monitoring the evolution of material micro­structure during processing.

Recently, several advances have been made using la­sers for both generation and detection of ultrasonic waves in a totally noncontacting manner for material microstruc­ture evaluation.4-6 The generation of ultrasonic waves in materials by pulsed laser absorption has been studied, and recently reviewed,1-10 for many cases involving thermoelas­tic expansion or ablation. Several detection schemes have been developed that measure ultrasonic motion with laser light reflected from the material surface. These also have been recently reviewed. II •12 Much work now is underway to determine material properties directly with the laser ul-

. 13-16 trasomc measurement process. Due to their noncontacting nature, laser ultrasonic

techniques now open possibilities for real-time monitoring of materials in very hostile environments, such as are en-

. 17 Th' countered during high temperature processing. IS pa-

per describes the successful application of laser ultrasonic techniques to the monitoring of sintering of zinc oxide at temperatures from 850 to 950 0c. Prior to this work, ultra­sonic wave measurements of the sintering of ceramics used buffer rods that made direct contact with the material. 18- 20

Only optical access to the sample being sintered is required with the laser ultrasonic technique described in this paper.

Zinc oxide was chosen as the ceramic material to be studied as it sinters at varying rates at temperatures in the range of 700--1100 °C. Also, it does not require a sintering additive to reach full theoretical density. Laser ultrasonic measurements are presented for samples of zinc oxide with rough and/or optically diffusely reflecting surfaces, includ­ing in situ measurements taken within a tube furnace.

II. EXPERIMENTAL TECHNIQUE

Figure 1 is a schematic of the laser ultrasonic measure­ment setup used for the sintering experiments. A pulsed Nd-Y AG laser with a 10-ns pulse width and pulse energies up to 300 mJ was used as the ultrasound source. The gen­eration mechanism for ultrasound in ceramics generally includes both thermoelastic and ablation. Most ceramics have a significant optical penetration depth, which alters the thermoelastic generation mechanism. This can be used to advantage. Thermoelastic generation at the surface pro­duces an elastic wave radiation pattern with large direc­tionality away from the surface normal, whereas ablation produces its largest wave motion in the direction of the surface normal. 8 Therefore, for through transmission or for pulse echo measurements the ablation mechanism is pre­ferred. Ablation is easily produced in ceramics due to their small thermal diffusivity compared to metals. However, material surface damage results from ablation, which may be undesirable. In ceramics that exhibit a sizeable optical penetration depth, however, the thermoelastic mechanism itself produces a significant component of the ultrasonic wave in the direction of the surface normal because the thermoelastic expansion extends into the material. This subsurface expansion produces a precursor waveform that is normal to the surface and produces a significant normal surface motion. 8

•2

1.22 The excitation laser beam used for this work was focused to a diameter of 3-4 mm and ab­sorbed to a depth of about a millimeter. It produced a

6077 J. Appl. Phys. 68 (12),15 December 1990 0021-8979/90/246077-06$03.00 © 1990 American Institute of Physics 6077

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

75.102.71.33 On: Mon, 24 Nov 2014 19:04:21

Nd-YAG Pulsed Lase~ r---

Sample

Fabry-Perot Confocal I-(T I-- ~'"" 1\

Interferometer ....,. /,- V Tube Furnace

FIG. I. Block diagram of the laser ultrasonic experiment. The laser has a pulse width of 10 ns at 1.06 /lm. The interferometer uses an argon ion laser with a maximum of 1.0 W in the green 0.514-l'm line.

strong longitudinal precursor waveform due to the optical penetration. Some ablation of material from the surface was evident from discoloration and minor pitting; how­ever, the measurements were completed without ablation of the ceramic to a depth greater than about 25-50 f..Lm.

A confocal Fabry-Perot interferometer, modeled after that described by Monchalin and Heon, II was used to de­tect the ultrasonic waves on the surface of the samples. The detector is sensitive to the Doppler shift of light reflected from the sample surface, which is moving due to the ultra­sonic wave. An argon ion continuous laser (maximum out­put of about 1.0 W in the green line) was used to illumi­nate the sample surface as shown in Fig. 2. A smail portion of the beam was diverted for electronic stabilization of the Fabry-Perot cavity length with respect to the laser fre­quency such that the operating point was maintained on the side of the interferometer transmission curve (Le., ap­proximately half of the input light intensity passes through the interferometer). The output of the interferometer is detected by a photodiode yielding a signal proportional to the frequency shift of the light scattered from th<:; moving surface with respect to the s:abihzation point.! 1 The elec­tronic stabilization corrected fDr the laser drift and low frequency instability due to ambient vibrations ;:,f the ap­paratus and allowed the entire apparatus to be mounted on a fixed tabie without additional isolation.

\_ .....

(I)

FIG. 2. Schematic of the confocal Fabry-Perot interferomt:'t .. r used for detection of the sample surface velocity due to the ultrasonic wave. (a) is the argon laser. (b) the Fabry-Perot interferometer. (cl palarized beam splitter cubes. (d) 1/4 wave plates. (el 1/2 wave plate, (f) mirrors. (g) lenses for collecting the scattered light, (h) photodiodes, (i) the sample, and (j) the central tube of the furnace.

6078 J. Appl. Phys .. Vol. 68. No. 12. 15 Decernber 1990

(a)

o 1.5 2 25

Time (~s)

FIG. 3. Comparison of laser generated ultrasonic waveforms recorded with the Fabry-Perot interferometer for two sample, of silicon nitride. The two 6.I-mm-thick samples were identical except for surface finish. The polished sample (a) had a mirrorlike surface and the unpolished sample (b) was coarse ground with 120 grit paper. Detector laser powers were 1.6 mW (polished) and 50 mW (unpolished). respectively. The pulsed laser beam, used as the source of ultrasonic waves for both ,am­pies, impinged on the opposite side from the detection.

This detection scheme is self referencing, i.e., the pho­todiode signal originates from the interference of light scat­tered from the sample surface and delayed portions of that light within the Fabry-Perot interferometer. Since all of the optical wavefronts are nearly identical to that scattered from the sample surface, light from a relatively large sur­face area (about 4 mm2 ) including many speckles is col­lected efficiently. This gives this detection scheme the sen­sitivity needed for detection from rough and/or optical1y diffuse material surfaces, such as are encountered in ce­ramic materials in general and "green" state ceramic ma­terials in particular. Increasing the detection laser power can compensate for much of the lack of specular reflection due to the rough sample surfaces. Figure 3 shows ultra­~onic waveforms detected from polished (a) and unpol­Ished (b) opaque samples of silicon nitride, which were used to assess the capabilities of the Fabry-Perot interfer­ometer for rough surface detection. The figure shows that a very similar waveform was recorded from the unpolished sample as from the polished sample, with essentially the same signal to noise ratio. This was achieved by increasing the argon ion laser power from 1.6 mW for the polished sample to 50 mW for the unpolished sample to account for the. increased scattering of the laser light into a very large solid angle, essentIally hemispherical. No surface damage resulted from using the detection laser at these powers. The differences evident in Fig. 3 between the results for the polished and unpolished samples is under further investi­gation and may originate from different illumination dis­tributions of the confocal Fabry-Perot between the pol­ished and unpolished surfaces. Such a difference would produce different frequency responsivities of the interfer­ometer for the two samples.

The material used was commercial zinc oxide (Fisher ZS2-S00) with a very small particle size, around a micron.

Telschow et al. 6078

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

75.102.71.33 On: Mon, 24 Nov 2014 19:04:21

(b)

(c)

(d)

o 2 4 6 8 10 12 14 16 18 20

Time (jts)

FIG. 4. Real-time laser ultrasonic measurements in the through transmis­sion mode for zinc oxide taken while heating in a tube furnace at 850°C. The firing of the pulsed laser is marked by the optical feedthrough pulse near the origin for each A scan.

and some agglomeration. The green state samples were pressed to approximately 47% of theoretical density in disks 25 mm in diameter and 8-10 mm thick. They were particularly soft with roughly the consistency of chalk. The green state samples were ablated significantly by the source laser at the energy levels required for adequate detection signal to noise ratio. Therefore, the ultrasonic response of the green state samples was recorded with a minimum of signal averaging, usually 10 pulses. Surface ablation was significantly reduced after the sample reached greater than 50% theoretical density. The zinc oxide samples exhibited nearly complete diffuse reflection from their surfaces, even after sintering. The optical collection of scattered light from the sample, the determining factor for the signal to noise ratio, was limited by the opening of the furnace tube.

III. REAL-TIME SINTERING EXPERIMENTS

Figure 4 shows typical waveforms recorded while a sample was being heated to 850·C during the sintering process. (The true temperature of the sample as a function of time was not measured.) The signals represent longitu­dinal waves measured in a through transmission mode. Trace (a) was taken at room temperature before the sam­ple was inserted into the furnace. Several minutes were taken to insert the sample into the tube furnace; trace (b) was recorded upon completion of the insertion. At a fur­nace temperature of 850 ·C, the sample sinters to nearly theoretical density in about 2 h. The waveforms show a rapid decrease in the ultrasonic time of flight after inser­tion. This is due to the bonding between powder particles, the changing sample thickness, and the increasing densifi­cation taking place. Sintering is a complicated dynamic process, and Fig. 4 shows that the laser ultrasonic tech­nique can monitor the overall process as a function of time, but it does not distinguish between shrinkage, densification and possible temperature dependencies in the material properties, such as elastic constants. However, for materi-

6079 J. Appl. Phys., Vol. 68, No. 12, 15 December 1990

als whose elastic constants are not highly dependent on temperature, the density can be directly determined from the ultrasonic data given suitable knowledge about the shrinkage of the sample as it sinters.

At the higher temperatures the sample reflectivity was significantly lower than at room temperature and therefore the detection laser power had to be increased in order to keep the signal to noise ratio constant. Detection laser powers from 100 to 900 m W were used.

IV. RELATIVE DENSITY ALGORITHM

As stated above, the measurements were of the time of flight of longitudinal ultrasonic waves propagating through the sample, as it sinters. In order to determine the sample average density from these measurements, it is necessary to know the sample thickness and the effect of density and temperature on the ultrasonic velocity. In addition, since heating in the sintering process generally takes place from the surface towards the center, the internal temperature gradient causes nonuniform shrinkage. This produces a material which has a nonuniform density as it sinters. The time of flight of the ultrasonic wave is an average over this nonuniformity and cannot resolve the density variations. The extent of this nonuniformity for the zinc oxide sintered here is not known. However, in practice, the sintering pro­cess is usually continued long enough that internal temper­ature and density gradients are eliminated and the final product is uniform in density. It can be assumed that if the starting material is uniform in properties, then the end result from the sintering is as if the sample shrunk uni­formly from the initial to the final state. This is often uti­lized in the ceramics industry to predict the final shape of the material after sintering. This assumption was checked for the zinc oxide samples by measuring the sample initial and final masses, thicknesses and diameterli. From these results the average sample density was calculated both be­fore and after the sintering process. If the sample shrunk uniformly, the ratio of the final to initial diameterwould be equal to the ratio of final to initial thickness. This was found to be the case for the zinc oxide samples as is shown in Fig. 5, where the final mass is plotted as a function of the initial mass for a large number of samples. The data are plotted in the form of density x thickness, 3 which utilizes the fact that the diameters shrink in the same proportion as the thicknesses (i.e., uniform shrinkage). Densities are re­ported in percent of the theoretical density, which for zinc oxide is 5.6 g/cm3

. The slope of the linear fit to the data is 0.96, which indicates a maximum mass loss of about 4%, at most, due to the sintering process, although a small amount of mass was lost on many samples due to chipping of the edges and would lessen the apparent mass loss due to sintering.

During the sintering process, the ultrasonic measure­ments can be used to determine the average density of the sample in real time. With the uniform ~'£n:1.Tl\<...ag<: "':>:>Ulnp­

tion, the sample thickness can be shown to be directly dependent on density during sintering. This allows the cal­culation of both the average density and shrinkage of the sample as it sinters. Uniform shrinkage will be assumed

T elschow et 8/. 6079

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

75.102.71.33 On: Mon, 24 Nov 2014 19:04:21

8

7

'" 6 :i1 :::E

~ 5 :; ()

1il 4 0

~ $ 3 c: iii 1il .S 2 U-

Starting Calculated M9.SS

FIG. 5. Rdatioll>hip between the final sintered mass and the initial mass of tht' line oxide sampks calculated from Eq. (I). The results are pre­",nted in the unit<, of sampk relative density (':'c,) x thickness; [(mill ') , 104]10 show Ihat uniform shrinkage occurs during sintering.

throughout the reminder of this paper, and the results must be understood in terms of an average over the sample dimensions. In the rare instance that uniform shrinkage cannot be assumed. an independent measurement of the sample thickness would be required to completely deter­mine the sample's relative density during sintering.

The temperature dependence of the ultrasonic velocity in the zinc oxide compacts during the sintering process is unknown. Measurements on fully sintered sampk's have shown that the longitudinal wave velocity is essentially temperature independent (to within 5%) over the range of temperatures from ambient to 1100 0c. Therefore, no cor­rection for temperature has been considered for the data presented here.

During sintering, the ultrasonic time of flight and ma­terial dimensions decrease significa.ntly as the material's density and wave velocity increas{;'. A 250/, reduction in thickness was observed between the 47% dense sample and the 97'5',:, dense sample. The attenuation was not measured directly in these experiments; however, Fig. 4 shows that the waveshape becomes narrower as the material sinters. This is consistent with a decrease in attenuation, which is expected as the material densifies and the pores hecome small compared to the ultrasonic wavelength. A well­developed ultrasonic waveform for all samples that have been partially sintered (> 52%) was found. as :".nown in Fig. 4.

To complete the density determination algorithm, the dependence of the ultrasonic wave velocity on density for the material must be known. Several samples of zinc oxide were prepared in the "green" state by compressing the powder into disc shaped preforms about 25 mm in diame­ter by \0 mm thick with relative densities of about 47%. Samples were then sintered at temperatures from 800 to 960°C and for times from 2 min to 2 h in order to produce a set of partially sintered samples with relative densities from 52% to 97%. Longitudinal ultrasonic wave velocities

6080 J. Appl. Phys., Vol. 68, No. 12, 15 December 1990

o -laser

+ - "green"

x - piezoelectric

60 70 80 90 100

Relative Density (%)

FIG. 6. Longitudinal wave velocities of the presintered zinc oxide sample set measured both with the contact piezoelectric and noncnntacting la,er techniques.

were recorded for this sample set to provide basic informa­tion on the effects of sample density. First arrival times were measured with a 5 MHz contact piezoelectric trans­ducer, utilizing a vacuum coupled polymer film 23 with gel couplant and pulse echo overlap, and the laser ultrasonic setup as shown in Fig. 1. Figure 6 shows the longitudinal wave velocities for the sample set, measured with both the piezoelectric and laser techniques. Both measurement methods agreed to within 0.5%. Detection laser powers of about 150 mW were used for the optical measurements.

In the green state the samples have roughly the con­sistency of chalk and are fragile. The graph of Fig. 6 shows that the velocities are low for the green state, around 0.7 mm/ps, and not very dependent on the sample density. However, the velocity rises abruptly after a small amount of sintering, as depicted by the 52% sample~, probably due to bonding between the powder particles in the initial sin­tering. This results in a large change in elastic constant for the sample but very little densification. Similar changes in ultrasonic velocity have been recorded for packed powders under load,24 where the measured velocity was found to increase dramatically with loading pressure.

Further sintering produces an approximately linear change in velocity with relative density up to values ap­proaching that of the bulk material as full densification is achieved. It is not known if the linear dependence of ve­locity on density is valid over the complete densification range for ceramic materials other than zinc oxide; how­ever, similar results have been recorded for silicon car­bide. 3

The assumptions of conservation of mass and uniform shrinkage yield the following relation between density and thickness, which is applicable before, during and after sin­tering:

T 3 L3 L3 P;<--I =P =PI I' ( 1 )

where (i,f) refer to initial and final, with respect to sinter­ing. This, coupled with the longitudinal velocity depen­dence on porosity, C(p), determined previously, yields a

Telschow et a/. 6080

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

75.102.71.33 On: Mon, 24 Nov 2014 19:04:21

l 80

'f 70 0 II>

~ lIi

60 a:

50

40 1 1.5 2 2.5 3 3.5 4

Normalized Time Of Flight (T I T minimum)

FIG. 7. Sample relative density calculated by inversion of the longitudinal wave velocity versus density relationship and the assumption of uniform shrinkage.

relation between density and ultrasonic wave time of flight (T) of the following form:

(2)

Here (m) can refer to either before or after sintering val­ues. This equation can be inverted to obtain the density as a function of time of flight through a given sample. The size of different samples is taken into account by the nor­malization values, determined either before or after the sample is sintered. The range of applicability for this algo­rithm is determined by the range over which the velocity as a function of density is reliably known. For the zinc oxide samples described here, only data from samples of density greater than 50% theoretical density were used. This ex­cludes the "green" state. Figure 7 shows the inverted form of the algorithm plotted as the sample relative density ver­sus the normalized time of flight. For a given material, this curve must be determined by presintering a number of samples to different final densities. The universality of this curve for a given material is not known at present; other parameters may affect its shape, such as starting particle size and "green" state additives and significant tempera­ture dependence in the wave velocity. These effects are under further study.

Figure 8 shows typical results obtained for zinc oxide sample no. 43, which had a starting density around 47% and was presintered by heat treating at 700 ·C for about 1 h. At the time indicated in the graph, the temperature was raised quickly to a final temperature of 900·C and the longitudinal wave time of flight recorded as a function of time. With the aid of the above algorithm, the sample den­sity is calculated from the experimentally measured time of flight and from the density, the sample thickness (shrink­age) is determined using equation (1). These values are also shown in Fig. 8. Several samples with roughly the same starting density and size were monitored and sintered in this way to a variety of final temperatures ranging from 850 to 1050·C. Similar results to those shown in Fig. 8

6081 J. Appl. Phys., Vol. 68, No. 12, 15 December 1990

0.9

0.8

0.6 T (10oo'C) Relative Dens~y LI L Initial

0.40~"""""'0"':.2--:0:"c.4:--:0~.6:---0::'".8::---~--:-1.'::"2--:,'-:.4--:'"':.6--:':'":.8:----1

Time (h)

FIG. 8. Sintering plot for sample no. 43, with initial density of 50.6% and thickness 8.62 mm, showing the relative density. shrinkage. and sintering temperature history with time.

were obtained in all cases. The results from several samples sintered with different final furnace temperatures showed that the speed with which a sample sintered and the final density reached were a monotonic function of the final sample temperature. Comparisons between the shrinkage and relative densities of sample no. 43, sintered at 900 cC, and no. 49, sintered at 975 cC, are shown in Fig. 9. The slope of the densification and shrinkage curves becomes small after times of about 1.5-2.0 h for all the samples tested. The final density of the sample after sintering was predictable from this procedure to a precision of about 2%-3% theoretical density, which is about the precision to which the various parameters used in the algorithm is known. The fact that the final sample density seems to level off after some time, or at least increases very slowly, indi­cates that the furnace temperature is the most important parameter in determining the final density for this zinc oxide material.

V. DISCUSSION AND CONCLUSION

These results show that laser-ultrasonic techniques, which allow measurements on samples with rough and un­polished surfaces, can be used for noncontacting in situ measurements of elastic properties and, with an appropri­ate algorithm, sample density and shrinkage at high tem­peratures. The technique has been successfully applied to the sintering of zinc oxide. A FabrY-Perot interferometer was used for detection, with a sufficient signal to noise ratio to record ultrasonic surface motion from samples with rel­ative density of 47%-97%. It can be concluded that the ultrasonic velocity provides a direct measure of the densi­fication of the material during ~\",\.~ ..... :"'''»''b' <;;''C'V'CH.'' 61'fixrcn.

factors in the sintering process are still unknown, such as the temperature uniformity of the material during heat up and the effects of powder size, compaction pressure, and composition. These factors can now be easily and quickly

Telschow et al. 6081

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

75.102.71.33 On: Mon, 24 Nov 2014 19:04:21

0.95

0.9

O.B

0.75

0.65

#49

T (lODO'C) LI L initial

0.6 0'---0-'.-1 ~-0""".2-----C:0~.3:-----:-0~.4--0-'.::-5 ~--=0:"-.6=----:0'::. 7=---='o.'::"a---=0::'".g=-----'

Time (h)

0.9 I o.a

0.6

#49

#43

T (lODO'C) Relative Density

0.4 0'---0-'.-1--0~.2--0~.3--0~.4--0-'.-5--:0 .... 6-----=-O~.7:---='O.':-8--:0~.9=------'

TIme (h)

FIG. 9. Comparison of the shrinkages (a) and rdative densities (b) of ,amples nos. 43 and 4<) sintered at different temperatures.

explored by use of the noncontacting laser ultrasonic mea­surement of densificatkm, which allows the direct monitor­ing of the sintering process as a function of both time and temperature. This technique should also be applicable to other materials and processing environments. such as heat \r'CaHncnt of mct",b.

ACKNOWLEDGMENTS

The authors thank S. T. Schuetz, R. S. Wallace, and B. H. Park for preparing the zinc oxide sample:; and C.

6082 J. Appl. Phys., Vol. 68, No. 12. 15 December 1990

Stander for assistance with the measurements and a re­viewer for very detailed comments. This work was sup­ported by the Department of Interior's Bureau of Mines under Contract No. 10 134035 through Department of En­ergy Contract No. DE-AC07-76IDOI570.

I C. M. Sayers and R. L. Smith, Ultrasonics 20, 201 (1982). ~E. P. Papadakis and B. W. Petersen, Mater. Eval. 37, 76 (1979). 's. 1. Klima et al., "Ultrasonic Velocity for Estimating Density in Struc·

tural Ceramics," NASA TM·82765 (1981). 4c. B. Scruby, R. L. Smith, and B. C. Moss. NDT Internationa119, 307

( 1<)86) 'J.·p. Monchalin et af.. in Proceedings of the IEEE 1988 Ultrasonics Sympmium (IEEE. New York. 1988). pp. \041-1044.

he. B. Scruby, Ultrasonics 27. 195 (1989). 1 c. B. Scruby, R. J. Dewhurst. D. A. Hutchins. and S. B. Palmer.

Research Techniques in Nondestructive Testing. edlted by R. S. Sharp (Academic. New York. 1982). VoL 2. pp. 281-327.

'D. A. Hutchins, Physical Acoustics, edited by W. p. Mason (Academic Press, New York, 1988), Vol. XVIII. pp. 21-123.

q J. D. Aussel, A. Le Brun. and 1. C. Baboux, Ultrasonics 26.245 (1<)88). l"U. Schleichert, K. J. Langenberg. W. Arnold, and S. Fassbender. Re­

!'iell' 0/ Quantitatil'e /I/ondntructil'e E,'aluation. edited by D. O. Th· ompsun and D. C. Chimenti (Plenum. New York. 1989). VoL SA. pp. 489-4<)6.

"J.-P. Moncha1in and R. Heon. Mater. Eval. 44,1231 (1986). "J.-P. Monchalin. IEEE Trans. UFFC 33. 485 (1986) "L. Piche. B. Champagne. and J.·P. Monchalin, Mater. EvaL 45. 74

(1987 ) 14L. F. Bresse. D. A. Hutcbins. and K. Lundgren. J. Acnus!. Soc, Am.

84. 1751 (1988). "J.-D. Aussel and J.·p. Monchalin. Ultrasonics 27,165 (1989). l"J.·D. Aussel and J.·P. Monchalin. 1. AppL Phys. 65. 2918 (1989) I7R. J. Dewhurst, C. Edwards, A. D. W. McKie. and S. B. Palmer. J.

AppL Phys. 63. 1225 (1988) "M. W. Griggs eI 01 .• J. Phy,. E 19. 1059 (1986). [Cl F. Dm·ant. J. du Mouza. M. Arnould. and R. Struillou. in Proceeding'

0/ the Second International Sympusium of Ceramic Materia/.I and Cumponents for Engines (Deutsche Keramische Gesellschart. Bad Honnef. West Germany. 1986). p. 849.

,II 1. H. Gieske and H. M. Frost. RCl'ieu' of Quantitatit'" iVundestrllctil'<' El'a/uation. edited by D. O. Thompson and D. e. Chimenti (Plenum. New York, 1989). Vol. 8B. p. 1709.

21 K. L. Telschow and R. J. Conant. Nondestructi!'e Jfaterials Choraetcr­ization, edited by P. Holler. V. Hauk. G. Dobmann. C. O. Ruud. and R E. Grecn (Springer, New York. 1989). pp. 799-806.

'~K. L. Telschow and R. J. Conant, J. Acous!. Soc. Am. 88. 1494 ( 19'10) . R A Roberts. Rel'h'u' oj QuantJlatH'1! IVondellructll'e EI'aluatlUn. cd ited by D. O. Thompson and D. C. Chimenti (Plenum. New York, 1987). Vol. 6B. p. 1443.

'"M. P. Jones and G. V. Blessing. in Proceedings of the Secolld Intema­tiollal Symposium on lV'olldestructi!'1! Characterization 0/ ,ltaterials. ed­ited by J. F. Bussiere. J·P. Monchalin. C. O. Ruud. and R. E. Green. Jr (Plenum. New Y"rk. 1987) pp. IJ9-147.

Telschow et al. 6082

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:

75.102.71.33 On: Mon, 24 Nov 2014 19:04:21