Eur. Phys. J. Appl. Phys. 43, 363368 (2008)DOI: 10.1051/epjap:2008129 THE EUROPEAN

PHYSICAL JOURNALAPPLIED PHYSICS

Investigation of structural and optical properties of sputteredZirconia thin films

F. Rebib1,4,a, N. Laidani1, G. Gottardi1, V. Micheli1, R. Bartali1, Y. Jestin2, E. Tomasella3,M. Ferrari2, and L. Thomas4

1 Fondazione Bruno Kessler-Centro per la Ricerca Scientica e Tecnologica, Via Sommarive 18, 38050 Povo (Trento), Italy2 CNR-IFN, Istituto di Fotonica e Nanotecnologie, via Sommarive 14, 38050 Trento, Italy3 LMI-CNRS, Universite Clermont-Ferrand II, 24 avenue des Landais, 63177 Aubie`re Cedex, France4 PROMES-CNRS, Tecnosud-Rambla de la Thermodynamique, 66100 Perpignan Cedex, France

Received: 16 November 2007 / Received in nal form: 26 May 2008 / Accepted: 3 June 2008Published online: 17 July 2008 c EDP Sciences

Abstract. Zirconium oxide thin lms were deposited by sputtering a ZrO2 target under an argon-oxygengas mixture and dierent total gas pressures. Their composition, structure and optical constants werecharacterised by mean of Auger proles, XRD, XPS, m-line and UV-visible spectroscopies. All the depositswere found to be sub-stoechiometric with O/Zr ratio decreasing from 1.6 to 1.45 when the depositionpressure increased from 0.01 to 0.05 Torr. A SRIM simulation was used to explain this behaviour. TheXRD showed a monoclinic phase for all sample with dierent grain size and residual stress. Finally, theoptical constants were determined. The refractive index decreased slightly when the deposition pressureincreased whereas the optical gap and the Urbach energy were found to be quite constant whatever thesputtering pressure.

PACS. 81.15.Cd Deposition by sputtering 68.55.-a Thin lm structure and morphology 78.20.CiOptical constants

1 Introduction

Preparation and investigation of zirconia (ZrO2) thin lmshave been receiving great attention for the last twentyyears. Indeed, this ceramic oxide is known to combinesthermal, chemical and mechanical stability and to exhibitinteresting optical and dielectrical properties, for example,refractive index close to 2.1, optical bans gap higher than5 eV and permittivity higher than 18 [1]. This resulted,in fact, in numerous applications such as protective andthermal barriers, optical lters, high reectivity mirrors,or insulators in microelectronic devices [2,3].

Among the various techniques available for deposit-ing zirconia lms, reactive sputtering is a highly attrac-tive technique because of the wide experimental possibili-ties. First, it permits to work at temperatures lower than300 C which is often required for optical and electronicstructures processes on sensitive substrates. Secondly, thevariation of several sputtering parameters, such as tar-get material, nature and pressure of the gas atmosphereand discharge power, guarantees a wide range of thin lmcomposition and properties.

It is well known that the properties of the deposits arestrongly aected by the sputtering parameters. Therefore,

a e-mail: farida.rebib@univ-perp.fr

in order to nd suitable deposition conditions for lmswith good optical and dielectrical properties which meetapplication in the various elds cited above, we studiedthe eect of sputtering pressure on the composition, mi-crostructure and optical constants of zirconia thin lms.To achieve this study, a zirconia target was sputtered un-der an argon-oxygen gas mixture at dierent total pres-sures. The layers were then analysed from dierent pointsof view (composition, structure, optical constants) and theresults were discussed as function of the deposition pres-sure.

2 Experimental

Zirconia thin lms were deposited at 13.56 MHz frequencyin a home made capacitively coupled reactor. A zirconiumoxide target (unstabilized, 99.9% purity) was sputteredunder an argon-oxygen gas mixture at xed self bias volt-age of 750 V. The gases ows were kept constant at 24and 6 sccm for argon and oxygen respectively and the totalpressure was varied from 0.01 to 0.05 Torr. Before depo-sition, the sputtering chamber was evacuated to 3 108Torr and then the target was pre-sputtered in an argonatmosphere to eliminate the surface pollution.

Elemental composition and homogeneity of the de-posits were checked by Auger spectroscopy. The depth

Article published by EDP Sciences

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

MANUELResaltado

364 The European Physical Journal Applied Physics

proles were carried out using a Physical Electronics 4200system, equipped with a variable resolution cylindricalmirror analyzer (CMA, energy resolution 0.6%) and acoaxial electron gun. The energy of the electron beam was5 KeV. Depth prolings were carried out on an area of400 m2 by alternating measurements of Auger spectraand sputter-etching the sample surface with 3 KeV Ar+ions. The C KLL, O KLL, and Zr MNN transitions weremonitored for the lm characterization. The minimum-to-maximum intervals of the considered lines were usedin the quantitative analysis, after correction with specicsensitivity factors.

Phase structure of the zirconia lms was deter-mined by X-ray diraction spectroscopy (XRD) usinga Seeman-Bohlin diractometer with Cu K radiation( = 1.5418 A) and grazing angle conguration. The inci-dence angle was equal to 3 in steps of 0.02.

Chemical environment of zirconium and oxygen atomswas analysed by X-ray photoelectron spectroscopy (XPS).The spectra were recorded with a Scienta-ESCA 200 in-strument equipped with a hemispherical analyzer and amonochromatic Al K (1486.6 eV) source. The C 1s, O 1sand Zr 3d core lines were acquired at 0.4 eV energy res-olution. After a Shirley-type background subtraction, thespectra were tted using a non linear least-squares pro-gram.

Concerning the optical measurements, the UV-visibletransmission spectra were carried out using a double beamCary Varian 5000 spectrometer in the range 2001000 nm.In order to determine the optical gap (Eg) and the Urbachenergy (Eu), the Tauc relation [4] was followed. Refractiveindex and layer thickness were measured at both 632.8and 543.5 nm in transverse electric (TE) and transversemagnetic (TM) polarisation modes.

In order to comply with the dierent characterisationtechniques, zirconia thin lms were deposited on varioussubstrates: silicon for composition and structural investi-gations and quartz for optical measurements.

3 Results and discussion

3.1 Deposition rate and film stoechiometry

Figure 1 shows the evolution of the deposition rate (rd)of the zirconia lms as function of the total gas pressure(pt). The dashed lines are drawn as eye guide. One candistinguish two trends: an important increase for pressurevarying between 0.01 and 0.03 Torr followed by a slight de-crease for higher pressures (0.04 and 0.05 Torr). To explainthese tendencies, two main parameters are considered: onone hand, the increase of the argon ions rate with the totalpressure leads to an increase of the amount of sputteredmatter. But, on the other hand, the mean free path of thesputtered particles is lowered when the pressure increases.This leads to a diminution of the amount of species whichare able to reach the substrate surface. So, in our case,the rst parameter is dominant for the lower pressureswhereas at pressures higher than 0.03 Torr, the eect of

0.01 0.02 0.03 0.04 0.05

1.3

1.4

1.5

1.6

1.7

1.8

1.9

r d (n

m m

in-1)

pt (Torr)

Fig. 1. Variation of the deposition rate (rd) of zirconia lmswith the total gas pressure (pt).

0.0 0.2 0.4 0.6 0.8 1.0 1.2

0.5

1.0

1.5

2.0

Si su

bstr

ate

Surf

ace

O/Zr = 2 (ZrO2)O

/Zr

Normalized thickness

pt = 0.01 Torr pt = 0.02 Torr pt = 0.03 Torr pt = 0.04 Torr pt = 0.05 Torr

Fig. 2. AES measured O/Zr atomic ratio for zirconia lmsdeposited at dierent gas pressures.

the mean free path is prevalent. Of course this explana-tion doesnt consider the growing layer density. Indeed,Gao et al. [5] conrmed a decrease of the deposition ratewhen the pressure increases but the mass deposition ratewas found to increase with the sputtering pressure.

Concerning the composition and the homogeneity ofthe dierent deposits, the Auger proles showed some car-bon contamination on the surface and also in the bulk ofall samples. This could be coming from previous carbon-based depositions and/or the pumping system. The atomicratio O/Zr (where O and Zr are the atomic percentages ofoxygen and zirconium respectively) of the zirconia samplesis shown in Figure 2; the erosion depths were normalized tothe lm thickness. All the deposits are sub-stoechiometric;therefore they will be indicated as ZrOx . Moreover,O/Zr ratio is lower when the deposition rate is higher.

Stoechiometry of zirconia deposits is controlled bythe oxygen partial pressure in the plasma [6]. In ourstudy, at constant gases ows, the increase of the totalpressure leads to higher oxygen partial pressure in theplasma; this is going in contradiction with the O/Zr ratiobehaviour. For a better interpretation of these results, a

F. Rebib et al.: Optical potentialities of reactively sputtered zirconia thin lms 365

200 400 600 800 10004.20

4.35

4.50

4.65

4.80

4.95 SO/SZr EO/EZr

S O/S

Zr

EAr+ (eV)

0.22

0.23

0.24

0.25

0.26

0.27

0.28

EO /E

Zr

Fig. 3. SRIM simulation of the sputtering yield and ejectionenergy ratios of O and Zr atoms from ZrO2 target under argonions bombardment.

SRIM 2003 [7] simulation has been realized. The SRIMprogram allowed estimating the sputtering yield (S) andthe ejection energy (E) of Zr and O particles from thesurface of ZrO2 target under argon ions bombardment.Figure 3 shows the variation SO/SZr and EO/EZr ratiosas function of the incident argon ions energy. The amountof ejected oxygen atoms is more important than zirconiumone but the energy of zirconium particles is much higherthan oxygen one. This means that more Zr atoms may beable to reach the substrate surface which could contributeto the lm sub-stoechiometry.

3.2 Structure

Figure 4 shows the XRD patterns of ZrOx samples pre-pared at dierent total gas pressures. Whatever the de-position pressure, the peaks are all attributed to dirac-tion from dierent planes of zirconia monoclinic phase.No peak belonging to the tetragonal phase and no ZrCphase were detected. The latter result means that all car-bon contained in the ZrOx layers is only contamination.

30 40 50 60 70 80 900

50

100

150

200

250

300 m21

1m12

2

m12

2

m21

1 m-1

22m

-220

m-2

02m20

1

m20

0

m11

1

m-1

11

2 (deg)

pt = 0.05 Torr

pt = 0.04 Torr

pt = 0.03 Torr

pt = 0.02 Torr

pt = 0.01 Torr

Inte

nsity

(a.u

.)

Fig. 4. XRD patterns of ZrOx lms deposited at dierent totalgas pressures.

0.01 0.02 0.03 0.04 0.05

-8

-6

-4

-2

0

pt (Torr)

Res

idua

l str

ess (

GPa

)

12

15

18

21

24

27 Residual Stress Grain size

Gra

in si

ze (n

m)

Fig. 5. Variation of crystallite size and residual stress in ZrOxlms.

The average of crystallite dimension of the lms (D) wascalculated using the Scherrer equation [8]:

D =0.9

B cos

where is the X-ray wavelength, is the Bragg diractionangle and B is the full width at half maximum (FWHM)after correction for the instrument broadening. The resid-ual stress of the ZrOx lms was determined using the fol-lowing equation [9]:

= Ecc

d d0d0

where d is the crystallite plane spacing of the lms, and d0is the standard plane spacing from X-ray diraction les.Ec = 170 GPa is the Youngs modulus and c = 0.28 isthe Poisson ratio for ZrO2 [10]. We calculated both D and using the most intense peak (111) of the XRD patterns.The results are reported in Figure 5 assuming an averageerror of about 10%.

The biggest crystallites (D 26 nm) correspond tothe lowest deposition pressure (pt = 0.01 Torr). Whenthe pt increases, D varies only between 14 and 18 nm.At low pressure, the mobility of the condensing particleson the substrate surface is more important; this favoursthe growth of large crystallites [5]. For the residual stress,all the obtained values are negative indicating a com-pressive stress. Moreover, the latter values are inverselyproportional to the crystallite size. This may happen be-cause for bigger crystallite size, there are fewer crystal-lite boundaries and dislocations, and thus a lower residualstress [5,10].

For more structural characterisation, the Zr and Oatoms neighbourhood was studied by XPS. Figure 6 showsthe variation of the Zr 3d and O 1s photoelectron peaksas function of the deposition pressure. Here also, the in-tensities are normalized to their maximum.

For stoechiometric ZrO2, Zr 3d5/2 peak is localized at182182.5 eV with a spin-orbit split of 2.4 eV from theZr 3d3/2 component and the O 1s photoelectron peak

366 The European Physical Journal Applied Physics

180 182 184 186 1880.0

0.2

0.4

0.6

0.8

1.0Zr 3d pt = 0.01 Torr pt = 0.02 Torr

pt = 0.03 Torr pt = 0.04 Torr pt = 0.05 Torr

Nor

mal

ized

Inte

nsity

Binding Energy (eV)528 530 532 534 536

0.0

0.2

0.4

0.6

0.8

1.0O 1s

Nor

mal

ized

Inte

nsity

pt = 0.01 Torr pt = 0.02 Torr pt = 0.03 Torr pt = 0.04 Torr pt = 0.05 Torr

Binding Energy (eV)

Fig. 6. Zr 3d and O 1s photoelectron peaks of ZrOx deposits.

5 10 15 20 25 30 350.0

0.2

0.4

0.6

0.8

1.0 pt = 0.01 Torr pt = 0.02 Torr pt = 0.03 Torr pt = 0.04 Torr pt = 0.05 Torr

Nor

mal

ized

Inte

nsity

Binding Energy (eV)Fig. 7. Valence band XPS of ZrOx layers.

has a binding energy of 530532 eV [11]. Our samplesZr 3d and O 1s peaks are slightly shifted towards lowerbinding energies; they are localized at 181.8182.3 eV and529.9530.1 eV, respectively. This behaviour is probablydue to the sub-stoechiometry of the deposits. Neverthe-less, one can notice that the binding energies of photo-electron peaks corresponding to the layer obtained at thelowest pressure (0.01 Torr) and which possesses the high-est O/Zr atomic ratio are the lowest; this seems contra-dictory. It is important to point out that the XPS is asupercial analysis. The behaviour of the latter samplemay be due to partial bonding of zirconium atoms withthe oxygen [12].

The spin-orbit split of the Zr 3d peak components isconstant at 2.39 eV for all samples; this suggests that allzirconium is bonded to oxygen which is in agreement withthe XRD results. For the O 1s photoelectron peak and inaddition to the main peak, a shoulder appears at higherbinding energies. The latter is attributed to COH or car-bonyl groups coming from surface contamination [13].

In addition to Zr 3d and O 1s core level peaks, thevalence band photoelectron peaks (0-35 eV) of ZrOx sam-ples were recorded and showed in Figure 7. The valenceband structure depends strongly on the material crys-talline structure. Figure 7 represents a typical signatureof the zirconium oxide monoclinic phase [14] which cor-

roborates the former results. From the spectra, one candistinguish:

the O 2p non-bonding and the O-cation hybridizedbonding orbitals between 2 and 8 eV,

the O 2s band which is broad and appears at 19 eVbinding energy,

the Zr 4d states in between 26 and 33 eV.

3.3 Optical measurements

Figure 8 shows an example of m-line spectra representingthe variation of the eective index ( or ne ) of a ZrOxsample.

The intensity loss peaks corresponds to the TE andTM guided modes of the electromagnetic wave and rep-resents the index seen by the considered guiding mode.Using values, we determined the refractive index and thethickness of our ve ZrOx layers. The results, for the twoused wavelengths (543.5 and 632.8 nm), are summarizedin Table 1.

2.1 2.0 1.9 1.8 1.7 1.6

0

15

30

45

60

75 = 632.8 nm

pt = 0.02 Torr

TE TM

Inte

nsity

(a.u

.)

(neff)

Fig. 8. Example of TE and TM guided modes ( = 632.8 nm)for a ZrOx sample obtained at 0.02 Torr.

The refractive index of the ZrOx lms decreasesslightly from 2.11 to 2.00 when the pressure increases

F. Rebib et al.: Optical potentialities of reactively sputtered zirconia thin lms 367

Table 1. Optical constants: refractive index in TE and TM polarisation modes (n (TE), n(TM)), thickness (t), birefringence(n), optical gap (Eg) and Urbach energy (Eu) of ZrOx samples.

m-line = 543.5 nm = 632.8 nm UV-visible

pt t. m n(TE) n(TM) n n(TE) n(TM) n Eg EuTorr 0.01 0.005 0.005 0.01 0.005 0.005 0.01 0.02 ev 0.01 ev0.01 0.30 2.12 2.12 2.11 2.11 5.08 0.430.02 0.42 2.10 2.14 0.04 2.09 2.12 0.03 5.01 0.410.03 0.45 2.04 2.09 0.05 2.03 2.07 0.04 5.02 0.420.04 0.43 2.00 2.06 0.06 2.00 2.05 0.05 5.01 0.410.05 0.40 2.00 2.07 0.07 1.98 2.03 0.05 5.04 0.41

200 300 400 500 600 700 800 9000

10

20

30

40

50

60

70

80

90

pt = 0.01 Torr pt = 0.02 Torr pt = 0.03 Torr pt = 0.04 Torr pt = 0.05 Torr

T (%

)

(nm)Fig. 9. Transmission spectra of ZrOx lms deposited at dif-ferent pressures.

from 0.01 to 0.05 Torr. This behaviour is directly linkedto the O/Zr ratio which varies in the same way. The re-fractive index values of our lms are a bit lower than fromthe stoechiometric and bulk oxide which are in the range2.172.2 [15]. This result could be explained also by thelower density of the lms, compared to that of the massiveoxide, and the eventual porosities of the deposits [16].

When the refractive index is measured using the twodierent wave polarisations TE and TM, one can measurethe modal birefringence (n = nTE nTM ) [17]. Thevalues of n for our samples are reported in Table 1 forboth 543.5 and 632.8 nm wavelengths. ZrOx lms showa birefringence increase with the deposition pressure. Inlow pressure regime, the birefringence is practically negli-gible but it becomes signicant for pressures higher than0.02 Torr. The important birefringence of ZrOx lms couldresult from dierent deposit characteristics such as thegrain size [17], the density and the columnar structure [18].

The optical gap (Eg) and the Urbach energy (Eu) aretwo other optical constants of the zirconium oxide lmswhich were determined from their UV-visible transmis-sion spectra (Fig. 9). The deposits are transparent over alarge wavelength domain; they became absorbent around230 nm. Using this latter part of the spectra and the Taucrelationship [4], one can determine the optical gap usingthe following equation:

(h)1/2 = B (h Eg)

is the absorption coecient and is a constant linkedto the disorder degree of the structure. Urbach tail energy(Eu) is an other parameter which characterizes the disor-der degree in amorphous material. This energy depends ontails in both valence and conduction bands and is linkedto the material absorption by the following relation:

= 0 exp (h/Eu)

Eg and Eu values are reported in Table 1. For all sam-ples, an optical band gap of about 5 eV was found. Thesevalues are similar to those reported in the literature forzirconium oxide thin lms deposited by the same or othertechniques [1,15] and could predict a good dielectrical be-haviour of our deposits. The Urbach energy is around0.4 eV for all samples and doesnt vary signicantly withthe deposition pressure.

4 Conclusion

The properties of zirconium oxide thin lms deposited bysputtering a ZrO2 target under an argon-oxygen gas mix-ture are not too much aected by the elaboration pressure.Indeed, the sub-stoechiometric character of the lms in-creases with the total gas pressure. But, all ZrOx samplecrystallized in the monoclinic phase with dierent grainsize and residual stress. An amount of contaminating car-bon was detected but no ZrC bonding was found usingseveral characterisation methods. The layers were found toexhibit interesting optical properties such as transparencyover a wide wavelength range, refractive index to large op-tical band gap.

An other deposition parameter, whose eects are inter-esting to study, is the ion bombardment. Indeed, when abias voltage is applied on the anode, the growing layer andthe substrate/lm interface properties could be stronglymodied.

F. Rebib is grateful to the Conseil Regional dAuvergne inFrance for the nancial support. This work represents a partof two projects nancially supported by the Provincia Au-tonoma di Trento in Italy: MicroCombi and PAT FaStFAL2007-2009.

368 The European Physical Journal Applied Physics

References

1. S. Harasek, H.D. Wanzenboeck, B. Basnar, J. Smoliner, J.Brenner, H. Stoeri, E. Gornik, E. Bertagnolli, Thin SolidFilms 414, 199 (2002)

2. IBM research Division (Thomas J. Watson ResearchCenter, Yorktown Heights, NY), J. Vac. Sci. Technol. A6 (1988)

3. A.P. Huang, P.H. Chu, H. Yan, M.K. Zhu, J. Vac. Sci.Technol. B 23, 566 (2005)

4. J. Tauc, R. Grigorovici, A. Vancu, Phys. Stat. Sol. 15, 627(1966)

5. P. Gao, L.J. Meng, M.P. Dos Santos, V. Teixeira, M.Andritschky, Appl. Surf. Sci. 173, 84 (2001)

6. J. Ciosek, W. Paszkowicz, P. Pankowski, J. Firak, U.Stanislawek, Z. Patron, Vacuum 72, 135 (2004)

7. N. Inoue, A. Sagara, Y. Nishi, J. Nucl. Mater. 220-222,895 (1995)

8. B.D. Cullity, Elements of X-ray Diraction, 2nd edn.(Addison-Wesley, Reading, MA, 1978)

9. I.C. Noyan, Residual stress (Springer, Berlin, 1978)

10. P. Gao, L.J. Meng, M.P. Dos Santos, V. Teixeira, M.Andritschky, Vacuum 56, 143 (2000)

11. J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben,Handboock of X-ray photoelctron spectroscopy, edited byJ. Chastain, R.C. King Jr. (Physical Electronics, Inc.,Minnesota, USA, 1992 and 1995)

12. K.P.S.S. Hembram, G. Dutta, U.V. Waghmare, G. MohanRao, Physica B 399, 21 (2007)

13. N. Laidani, G. Speranza, A. Nefedov, L. Calliari, M.Anderle, Diam. Relat. Mater. 7, 1394 (1998)

14. R.H. French, S.J. Glass, F.S. Ohuchi, Phys. Rev. B:Condens. Matter 49, 5133 (1994)

15. M. Jerman, Z. Qiao, D. Mergel, Appl. Opt. 44, 3006 (2005)16. A. Diaz-Parralejo, R. Caruso, A.L. Ortiz, F. Guiberteau,

Thin Solid Films 458, 92 (2004)17. D. Navarro-Urrios, F. Riboli, M. Cazzanelli, A. Chiasera,

N. Daldosso, L. Pavesi, C.J. Oton, J. Heitmann, L.X. Yi,R. Scholz, M. Zacharias, Opt. Mater. 27, 763 (2005)

18. S. Wang, G. Xia, H. He, K. Yi, J. Shao, Z. Fan, J. AlloysCompd. 431, 287 (2007)

To access this journal online:www.edpsciences.org

IntroductionExperimentalResults and discussionConclusionReferences