Non-destructive characterization and evaluation of thin films by laser-induced ultrasonic surface waves

Non-destructive characterization and evaluation of thin films by laser-induced ultrasonic surface waves

,,rff=.., -,~.-r.w, ELSEVIER Thin Solid Films 290-291 (1996) 305-311 m m Non-destructive characterization and evaluation of thin films by laser-i...

451KB Sizes 0 Downloads 4 Views

,,rff=.., -,~.-r.w,


Thin Solid Films 290-291 (1996) 305-311



Non-destructive characterization and evaluation of thin films by laser-induced ultrasonic surface waves D, S c h n e i d e r n, M . D . T u c k e r b ' FI2GIWS,Pos~ach16, D.OI171Dresden.Germany hSpectrumSciencestat., J050 O~mead VillageDrive,Santa Clara,CA 95031-0808, USA

At~tract Ultrasonicsurface wavesinducedby a pulsed laser act as high-frequencyelasticoscillationspropagatingalongthe target materialssurface. TI~ velocityof these propagatingwaves is eff,'cted by the elasticityand d,'asityof the material.These surfacewaveshave a unique property in that their depth of penatrationin the matedal decreaseswith an increasein theirfrequency,At higherfrequencies,the velocityof the surface waves are affected if the film or near-surfaceregion compositiondiffers from the substratamaterial.G~neratiagsndsce waves using short la~,erpulses has allowed for non-des~ctiv¢ messammantsto be p~ormed over a wide frequencyrange,This in tam resultsin a spccaumof velocityfrequencieswhich, whencomparedto known filmand densitycharacteristicsusing a fit program,can be used to charaet~ze surface composition and thin films, This velocity spectrum can b¢ used to determineYoung's modulus, density, and thickness of the film. The ultrasonicthin film analyzerhas been devalued for use in industrialtest laboratoriesand re~earchinstitutes, Keyword~: Laserirradiation;ElasticprOl~nles;Density;Yo'Jng'smodulus

L Young's modnlus--a promising parameter for thin film characterization The Young's modulus of a given film is related to the microstmctare and bonding condition of the film material, and can be a helpful and interesting macroscopic parameter when used to characterize the state of the material, Represealing the main elastic propeay of the material, Young's modulus of the film determines the mechanical behavior of the material (film), It determines the residual stnssses as well as the elastic energy induced by external loading, which is responsible for initiating cracks and micro-fractures, in the ease of hard film materials, Young'smodulus oReacorrelates with hardness, which may be difficult to measure directly, The properties and microslructuro of films and coatings can be widely varied by deposition technology and methodology. Usually, Yoang's modulus is also affected by these variations. Tabulated modulus values obtained from bulk materials often deviate noticeably from the modulus seen in films composed of the same material, Bulk pro~nicsm not known for many film materials as ~ey have never been obtained as bulk samples, These factors point to Young's modulus as being u favorable method of characterizing the film material and controlling the stability end reproducibility of the deposition process. By using the laser-induced surface wave technique, we are able to provide these parameters 0040-6090196/$15,00© 1996BlsevlerScienceS,A,Allrightsreserved PII S0040-6090 (96) 09029- 3

quickly and without special sample prepamtloa. This fact has been proven many times over several years by using the laser induced surface wave technique in a thin film laboratory, testing a large variety of films and surface layers, For carbide- and nitrite-hardened materials, Youog's modulus is known to depend on stQichiom¢~y, In the case of titanium carbide (TiCz) them is a linear correlation between Young's modulus and carbon content. The dependence on composition may be rather strong, as in the case of TiN~ where Young's modulus is about 180 GPa for low z and 460 GPa for z approaching unity [ 1]. The modulus of carbon films can vary greatly in range from < I0 GPa up to about I000 GPa, Of all materials, diamond has the largest Young's modulus, Consequently, the higher the Young's modulus of the carbon film, the hi,chef the content of diamond bonds {spa bonds), The hardness of the film material can be estimated with sufficient accuracy from the empirical relation H = E I I O , where H represents hardness, and E is Young's modulus, This suggests Ycong's modulus to be an excellent indicator for testing, classifying, and optimizing carbon films. This method has been used very successfully for diamond-like carbon films (DLC) [2-4]. Characterization of boron nilride films is another promising field of application, Cubic boron niuide (e-BN) is expected in have high hardness and excellent wear behavior,


D Schneider, M.D. Tucker/ThinSolidFilm.¢290-291 (1~962305-311

These prol~rties are also atu'iboted to the sp3 bonds [5]. Young's modulus of e-BN is about 600 GPa while that of hexagonal BN is only about 35 GPa. This would show ¥oung's modulus to be a sensitive indicator for BN film

Each of these examples show young's modulus to b¢ a suitable method of film characterization.


2. Ultrasonic sadaae wave analyzer for non.destructive thin lilm characterization

A higher density of micro-defectsreduces Young's modulus. This phenomenon is widely used for testing bulk ceramic martials. It car also be used for plasma-spray~ ceramiccoatings. Dependingupon the plasma sprayingtechnology, Young's modulus of ZrOz coatings have been found to be 70% lower than that of the bulk material [6]. These findingsdepend upon the nattn~and structureof the porosity.

Presently,Young's modalus is not yet a standardparameter for film characterization,although it correlates with relevant film proportles,Measurementproblemsarise for filmshaving a thickness of about l/tin or less. The most conventional

measurementmethods require an extensive prepmation and

Fig, 1, (a} l"]dn Iilm ~a]yzcr, (b) specimen table with acoustic d¢lector and loser optics.

D. $ehneider, M.D. TuckerI Thin Solid Films 2!;O..291 {1996) 305-311

careful time.consuming procedures. The generally simple method of unloading a nano-indenter is frequentlycharacterized by unsatisfactory reliability and comparability. The ultrasonic thin film analyzer (Fig. 1) has shown the ability to perform accurate and continuously reproducible measurements using the laser-induced ultrasonic surface waves, High frequency surface waves allow the Young's modulus to be determined for films down to 100 nm, and no special specimen preparation is required. We have foundthat in order to maintain a reliability level of ± I% with regards to the reproducibility of the modulus measurement, the sampiing area of the ~peeimen should be at least 53<5 mm~, The thickness of the sample can vary, but should not be less than 0,5 ram, In addition to Young's modulus, this technique provides other information as well, The greater the difference between the substrata materiel and the film material, the easier other parameters can be derived. In the case of thicker films, the density and/or thickness can also be determined. Moreover, other gradients of the microstructure can he investigated in addition to those produced by coating, such as damage layers caused by grinding or polishing, of which are usually characterized by higher defect density. 3. Surface waves--an efltdent film test method Yoong's modulus can be determined by measuring the velocity of acoustical bulk waves. However, these longitudinal and transversal modes are not suitable forchamctedzing the thin films representing only a very small part of the test piece, Ultrasonic waves have distinct advantages for testing thin films, They propagate over the surface of the material and behave in a manner similar to the ripples in a pond created by a pebble being chopped into it's surface. The amplitude of this wave mode is usually largest at the surface and decreases exponentially with depth. An important property is that the penetration depth of the surface wave decreases with an increase in frequency. This can be seen in Fig. 2, which shows the wave motion to be more shallow at the higher frequencies. For homogeneous isotropic material the propagation velocity c of the surface waves depends on Young's modulus E, Poisson's ratio v, and the density p: C= 0.87 + 1.12uJ _ E_ l+v V2p(I +v)


The concentration of the wave energy near the surface makes the surface wave very sensitive to coatings and films, even films thinner than the penetration depth of the surface wave. The propagation velocity Of the surface wave is changed by the influence of a film,Therefore, the wave velocity contains information about the properties of the film. As an interesting phenomenon of surface wave propagation in coated materials, the surface wave velocity depends on the frequency, which is called dispersion, This can be seen in

acoui~ ~


w a ~ propaQM~ ~ oozed r n m ~ i

I~fmm~:. m o m ~ . ~ , • wlp w , m ~ ~ ' ~ . e,mUM~loa~l~

e,;~emUr~. - ~ * m l q m .mai~mu~

- l a r i a t e~mfk~

Fig,2, Twosurfacewaveswithdifferentfrequenciespmpagalinginco01ed materials, Fig, 2, Since the penetration depth of the wave motion dependson frequency, surface waves with hider frequencies are more influenced by the film, Therefore, measuring velocity versus frequency can provide useful information. Such an information curve, or velocity spectrum, is shown in Fig, 3 which indicates a measurement taken on a DLC film dspos. ited on a steel substrata. Since the DLC film has a higher velocity than the substratc, the velocity increases with higher hequency. The form of the curve depends upon the Young's modulus, density, and thickness of the film, The DLC film sample shown in Fig. 3 yielded a Young's modulus of 8"/7 GPa, density of 3,44 g cm-3, and thickness 0,93 p.m, The DLC film was created using laser arc deposition. Th~se results were obtained without the need for separate measurements on the substrata. These parameters can be derived from a spectrum which can be extrapolated from the range near the frequency origin where the surface wave propagation is completely governed by the substrat¢, Young's modulus of the substrate can be derived from the extrapolated velocity. A modulus of 203 OPa was obtained from the steel substratatest shown in Fig, 3, k

I °I ///


~oquen~,MHz Fig, 3, Sl~h'um of surface wave velo~fiy raeesored on steel, coate~ with carbon film.

D. S¢llnei~r, M,D, Tbcker/ Thin 8did Films 290-291 (1996)30~-311


of the substrate, a fixed distancn of several millim©tcrsfrom the las©rfocus line [ 7]. The transducer consists of a fine steel edge holding a piezo~.lectric film to the surface, The small size of the photo-acoustic assembly allows for the testing of relatively small samples such as cutting tools. During the measurement procedure, the distance between the laser focus line and the transducer is precisely varied, The spewiman being measured and the transducer are both fixed to a translation stage which in turn moves l~rpendicular to the laser beam, The velocity of the surface waves is easily obtained from this simple equation: C(/)= ( x z - x l ) ' f ' 2 w [~-~'~] Fig, 4, Schematic representation of the measuring eqttipmen¢,

where x~ and x: represent the two distances between the laser focus point and the pick-up of the transducer. At both distances, a laser-induced sound signal is detected by the titans. ducer and then displayed and recorded on the oscilloscope, as is shown in Fig. 5. This figure shows both the signals detected at x, and at x: to have a different form. This phenomenon is caused by the surface wave dispersion, also known as frequency-dependentvelocity, in ooatexlmaterials, As was described earlier, the influence of the film on the surface wave propagation increases with an increase in frequency, The surface wave impulse can be regarded as consisting of an additive suparposition of elastic waves of different frequencies. Since every wave component propagates with its own velocity, the impulse is deformed as it travels along its path. The impulse deformation contains information about the film material (Young's modulus, density, thickness) whose extraction from the data is part of this method. In a homogeneous material without film, the same

4, Measuring equipment for analyzing thin ~lms.-

methodolo~" Although the spectrum of the surface wave velocity has a rnmarkable potential for thin film and surface chal~cturization, it has not been widely used for ,-Jeh. The thin film analyzer is represented in a block schematic in Fig. 4. The measurement requires that the surface waves be generated and detected in a wide frequency range. The short laser pulses (pulse duration, 0.5 us; energy, 0.4 mJ) of the nitrogen laser ate applied for generating wide band surface wave impulses, A cylindrical lens focuses the beam onto the specimen surface. This laser strike at the specimen surface creates a very brief spot of controlled local heating. This short time local heating produces an ultrasonic impulse, which in turn propagates over the surface of the test piece. The impulse is detected by a wideband piezodectric transducer at the surface

posltlon otla~er beam at x z

9o~tlonof le.,er beam atx,


I I I ~°r ]




I .




. ,





sl~1- deted~i it dllian~ x~'

' ei0MI- detscted at esta.ce r,,



. Ilmda





g'~,IJ Fig. 5. Photo-aanustio signals detected at two different distances xl and ~2, between the loser focus I~int and piezoeleeu'ic Iran~ueel' or, coated matedlfl,

13, Schneider, M.D. Tucker/ Tf~lttSolid Ititms 2[~-29! (1996) ,zO5~ l I

impulse shape would be detected at the positions xt and x~in Fig. 5. ~(3') and ~ ( / ) in Eq, (2) arc the phase values of the surface wave frequency ~ They arc, obtained by Fourier transforraing the impulse forms in Fig. 5.


I~OQ 1100 10~0


4 O~mond 4


6~ ,,~"

Ion Bo|m

~. Ultrasonic thin film analyzer ~o~vare

,,.*~*"* • t,. *°*

Pulsed Are I Laser.Arc ...

The software consists of two program packages. The first program is used to measure the velocity spectrum of the surface waves. It uses a special procedure for eliminating the 2¢t ambiguity of the phase values. The seeond program provides the film parameters from the dispersion relation. The exact theory of surface wave propagation in coated materials is used in this program [8]. A fit procedure searches for the best match between the measured and theoretical curves, Both programs are run in Windows environment for cuse of use and operation.

6. Important features ofthe surface wave method in principle, the suffacc wave method can provide three film parameters: Young's modulus, density, and thickness, This method has the advantage of being able to measure the velocity spectrum of the surface waves. The error depends on the accuracy of time and distance measurement, It amounts to:

Ac/c~O.fX 10-~


for a gauge distance ofx~-x~ = 5 ram. The greater the difference between the substrate and the film, the thinner the films that can be investigated, Di~ondlike carbon or boron nitride films can he successfully tested down to 100 am. The ability of determining all three film parameters |~entionod from the surface wave spectrum depends on film and substrata matariel~and upon film thickness. For thicker films, more parameters ,'an be derived from the surface wave spectrum, In ~e case of deriving the Young's modulus of a 100 nm film, the density e~d thickness should he known, However, the density does not need to he exact. This is not an issue for hard thin fiims such as DLC whose density varies in the range between 1,8 and 3.52 g era" 3 while Young's modulus varies between 10 and 1000 GPa, Choosing an intermediatedensity value often provides sufficient accuracy of the ¥oung's modulus for controlling the film quality. Moreover, a numhe.rof hoJ~lfilm materials have an advantageous property in that the Young's modulus and density at~ecorrelated, The larger the ¥oung's modulus, the higher the density, This empirical correlation is shown for carbon films in Pig. 6. The correlation has been integrated into the fit program and allows Ynung's modulus and density to he determined simulta,eously. A similar correlation exists for boron nitride films,


~..'*'" u o

Plasma OVD

1® Fullerene

iF-(: a.e:H'


~..~:./:'~-. ,0 .............................



I.B 3,0 2,2

2.4 2,S 2,S 3,0 3,2 3,4 3S

density, ~¢ms Fig, 5. Correlationbetween Yoaog's n~l~tlus ~nd density for diwond-like carbon film.

7, Applicaflens A certain limitation arises from ultrasonic absorption, It is difficult to investigate films on orconsisting of highly absorbing materials like polymers. The surface wave method has been applied to many different film materials such as protectire films for wear resistance, sensor materiels, and materials formioroelectronicsapplications.TheenclosedTable I gives an overview of the obtained results. The following three examples are described in more detail.

7.1, DLCfilras Fig, 7 shows measured Young's moduli of DLC films deposited at different temperatures on various substfates, These films were produced using laser.are technology [9]. They have a thickness between 100 and 350 am, The diagram Table l Overvlew of results obtsin,cdwith the surface wave ntethod Film mateflal

Film thi¢lmess (tan)


cVD diamond DLC (l~Sercuc) SiN SiC 2~O2 (plasma sprayed) BN

2-13 fi.t-I .S ~2-0.3 0.2--0.3 ~1-105 0.1-0,2

800-1140 150-~00 230-2~5 100-150 12-100 35-284




TiC 'fiAt )N Chromium,ttrid~ Pc|y-silicon Poro-~i

3.5-35 2, I--4 3-4 0.4-0.5 03-0.6

440-.455 260~350 ~,0-120 152-171 2-100




tlard chromium



Nilrlde coatings

about ?.S

about 130


$tdis~te raztmials: stcel~ Haslelloy~ hlcoael, ~,alailtum, WC-Celttetded


D. ,gckneider,M,D. Tucker/Thin Solid Films 2 9 0 - 2 9 1 (1996)305-311























I .o,-

l ,0o




Ill I I I k ~

. I N ......'%.

*o ¢|l~n14niurn "


¢-illetMleO O0lO

t f~0


t sol0

"" " : ~ , "~ ° ~


1 2 a

........ ZZOPI ..... 45GP| 2450P|





2OO ~03O 0

















eu~lrate temperature 'C Rg 7 Youngs mndulasof diamondlike on,ben fikns depositedby laser arctechnologyOhvariousznbstr~es,at differenttemperatures,

reveals the Yonng's modulus of the film increasing with the increasing process temperature. Where 600 OPa was achieved at room temperature, the value decreases to below 100 GPa above 200 °C. An additionalinfluenceof the substrata material can be seen here, Higher mOdulusindicates a higher oontent of sp3 bonds typical of diamond. From the empirical relation between Viekers hardness and Young's modulus for amorphouscarbon films (HV= 0.115E), a hardness of more than 60 OPa for optimally deposited films is found ( 101. This corresponds with the hardness rating of super hard materials. Z2, Boron nitridefilms

Fig. 8 shows results obtainedby IBAD technology,depositing boron nitride film on a silicon wafer. The measurements were performed at different positions on the wafer whorethe film was deposited with different atom-ion ratios. The film has a thickness of 120 am. The aim was to deposit cubic boron nitride, known to have promising wear behavior factors, to be compared with the hexagonal mOdification.The sp3 bonds ate responsible for this property similarto the case of DLC films. Consequently, the cubic BN has a higher Young's modulus (600 OPa) as compared to the hexagonal BN (35 GPa) [5]. The surface wavemethOdprovides awell defined and quick test procedure for films consistingof such materials. The diagram in Fig. 8 shows the surface wave velocity spectra measured in four surface regions. At positions I and 2 the spectrumdescends.This suggeststhe surface wave velocity within the film to be lower than the velocityof the silicon substrata. Young's modulus values of 38 and 45 GPa were' calculated for these films which indicate that hexagonal BN had been deposited in these regions. In regions 3 and 4, the surface wave velocity spectrum ascends.The character of the surfacewave spectrumsuggests that the film deposited here has qualities other than that in regions I and 2. The modulus values of 24,3 and 284 OFa calculated from the results confirm this qualitative assess-


i 40

I , P • , . L , i , ~ , i , G0 80 100 td0 t40 tG0 1SO ~0O fmqtmnWf, MHg

F i g , 8, S p e c t r a o f s u r f a c e w a v e v e l o c i t y m e a s u r e d

at different



a silicon wMer coated with a hot'on hi wide film,

ment. They indicate n considerably higher content of cubic BN in the regionsof 3 and 4,

7,3. Titaniumnitridefilms Fig. 9 shows the Young's mOdUlUS obtained for titanium nitride films with different adhesion levels to a steel surface

These films have a thickness ranging from 135 ~m to 2,~ /~m, Adhesionquality was tested using a mechanical scratch test. Tbequantity of acoustic emission signals detectedduring

the scratch test turned out to be a suitable parameter for characterizingthe adhesion quality [ 11 ], As seen in Fig. 9, intense acoustic emission is related to low modulus, which means that adhesion is related to modulus in this case. Films with a lower mOdulusshow a poorer adhesion capability in the mechanicalscratch test. This result can be explained by the influence of microdefects on the effectiveelasticity as well as on adhesion.This gives rise to the correlation between the non-destructively measured Young's modulus and the resultsof the mechanical test.At the present, we have found films with a modulus above 440 GPa show good adhesion properties, whereas the

adhesion of films with less than 430 GPa is found to be insufficient,



zl~|nlur~n~"i¢loflrnn~on ~1¢ml




to •

• I

¥'m modulus,QRI F i g 9. Con el~tton between Young s modulus and qn~ntity ofl¢oustio ends zion stFIsls during the moclt~icsd scrnteh test for TiN films deposited os steel

D, ~ch~eider, M~D,Tucker/Thin Solid Fil~n~290-291 (1996) $05-~11

References [1] X, Jiang, M, Wens, K, Sehmidt, J. Haapt, and W. Gisslet, J, @pl, Phys,, 69 (1991 ) 3053-3057. [2] B. SchulHch, H.-J, schnibe, D. Schneider,D, Dreschnr,H. Ollenderf, l.ectufe G4,02, ICMCTF,Apfft ~4-28, 1995,San Diego, CA, 1995, [3] B. Sehal~ch, H,-J, schelhn, G. Omdremy, D, Schneider and P, $iemrolh, Thin Solid Films,25.t (19943 125-129. [4-] D, Schneider, H,-J. Schntbe. '1'11.schwar~ and P, Hess, Diamond RelazedMoler.,2 (1993) 1396.1401 [5] H. Hollek, in H, Ftsehmets(erand H. ~ehn (e~,), Hans~g~hfcSten fl)rdi¢ Venchle~miaderung,DGM-lnformationsse~alhehaft.Verlag, 1987,pp,25--44,


[6] D, Schneider,Th, schwarz, H:P, Buchkremer and D. StGveT,Thin ~Ifd gilma,224 (1993) 177-183 [7] H. Could, R. G~gier,P.Hess,~ A. Neubzand,2.Acousr Soc.Am., 92 (1992)2980-2983 [8] G.W. Fro'nelltold13.L.Adl~. inW.P. Muon and R.N.Taunton (~ds.), PhysicalAcou3Jics.Vol. IX, Ac=lemie Press. New York. 1972, [9] H,-|. $cl~lbe ~d B, $chu[Heh, Thin Solid Fi~m~,246 (19941 91-102 [10] X. Jiang, M Reichelt. ond B. Strit~er. J. Appl. P~.. 66 0989) 5805-58~ [I11 D. Schneider,H. OUanderfandTh.schwazz,,ARpLPhys..,A6I (1995) 277