lable at ScienceDirect

Vacuum 86 (2012) 667e671

Contents lists avai

Vacuum

journal homepage: www.elsevier .com/locate/vacuum

A comparative analysis of a-C:H films deposited from five hydrocarbonsby thermal desorption spectroscopy

S. Peter*, M. Günther, F. RichterChemnitz University of Technology, Institute of Physics, D-09107 Chemnitz, Germany

Keywords:Thermal desorptiona-C:HDLCAr effusion

* Corresponding author.E-mail address: s.peter@physik.tu-chemnitz.de (S.

0042-207X/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.vacuum.2011.07.037

a b s t r a c t

Thermal induced gas evolution studies were performed on hydrogenated amorphous carbon (a-C:H)films deposited by plasma enhanced chemical vapour deposition (PECVD). The hydrocarboneargondischarges were powered by asymmetric bipolar voltage pulses. Five different hydrocarbons (acetyleneC2H2, isobutene C4H8, cyclopentene C5H8, toluene C7H8 and cycloheptatriene C7H8) were compared asfilm growth precursors.

The films were heated in vacuum upto 1000 �C and at the same time the gaseous effusion productswere measured with a quadrupole mass spectrometer (QMS). The release of hydrogen, water, argon anddifferent hydrocarbons from CH4 upto C7H8 was proved. Large hydrocarbons were detected only whenheating up soft films. When annealing hard films, they almost only lost hydrogen e mainly as H2 but alsoin remarkable portion as H2O. Possibly, the water in the a-C:H films contains the main fraction ofhydrogen unbounded to carbon.

The characteristic temperatures for the onset of gas evolution of the molecular gases and of argonwerefound to be strictly correlated with the film hardness e independent of the film growth precursor. Thethreshold temperature for the release of the dominating effusion product, molecular hydrogen, issignificantly higher than that of argon only for films with hardness below about 13 GPa.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

In hydrogenated amorphous carbon (a-C:H) films the carbonatoms occupy two substantially different bonding configurations:those associated with threefold plane coordination and sp2

hybridization of the valence orbital and those with fourfold tetra-hedral coordination and sp3 hybridization. Depending on thehydrogen content and the carbon bonding state in these filmsdifferent a-C:H phases exist [1,2]. For a-C:H deposited by PECVD atrelatively low temperature two phases are relevant: under condi-tions of low ion bombardment a soft, polymer-like material isdeposited with a high hydrogen content and predominantly sp3

bonded carbon; with proper ion bombardment a hard hydroge-nated carbon material is formed with predominantly sp2 bondedcarbon and a lower hydrogen content of about 20e35 at.%. Forhistorical reasons the latter has been named diamond-like carbon,DLC. The optimum ion energy flux for the PECVD of DLC is in theorder of 50e100 eV per deposited carbon atom [3]. Above all athigher deposition temperatures a graphite-like a-C:H forms,

Peter).

All rights reserved.

characterized by a low H content (less than 20%), a high sp2 contentand sp2 clustering.

The mechanical properties of a-C:H depend on three correlatingcharacteristics: (a) the carbon bonding structure, (b) the hydrogencontent as well as (c) the size distribution and volume fraction ofvoids in the film. The void formation is associated with the pres-ence of bond terminators like H and CH3 groups in the films [4]. Thevolume fraction of voids was described by a filling factor, definedas the ratio of the total film volume to the volume occupied bycarbon and hydrogen atoms [5].

Thermal desorption spectroscopy (TDS) [6] is an establishedmethod andwas also applied to gain information about the thermalstability, the chemical bonds as well as the microstructure of thinfilms. Relevant results about a-C:H films will be presented in laterparagraphs. The thermally induced changes in a-C:H film structurewere also traced with complementary methods. In a study by Wuet al. [7] about 300 nm thick DLC films, deposited by an ion beamdeposition technique from methane, were analysed after rapidthermal annealing (RTA) for 2 min at several fixed temperatures innitrogen atmosphere. The hydrogen content of the films remainedconstant at 30 at.% upto 500 �C, decreased rapidly from 25% to 9% inthe range 600e800 �C and finally remained at 5% in the range900e1000 �C. The Raman spectra of these films did not change in

S. Peter et al. / Vacuum 86 (2012) 667e671668

the RTA temperature range from 25 �C to 400 �C and began to showthe appearance of crystalline graphitic carbon above 500 �C.

Films with initially 40 at.% hydrogen were found to be struc-turally stable upto 260 �C [8]. A significant conversion to nano-crystalline graphite was detected by Raman spectroscopy alreadyat temperatures above 300 �C and it was completed upon heatingto 450e600 �C. A neutron diffraction study on the structure of harda-C:H upto 1000 �C [9] showed the effect of elevated temperatureson the atomic correlations. The room temperature initial mixture ofsingle bonds and olefinic double bonds became progressivelyaromatic and then graphitic as hydrogen is evolved.

A heating rate of 20 K/min is typical for TDS of thin a-C:H (forexample see Refs. [10] and [11]) but also 60 K/min [12] and as muchas 1200 K/min [13] were used for experimental purposes.

In a previous study [14], the PECVD of a-C:H films from pulseddischarges using five different hydrocarbons as precursors wasinvestigated e mainly in view of their hardness and composition.To get additional information about the influence of the precursoron the structure of these films, thermal desorption mass spec-troscopy was performed.

2. Experimental

The a-C:H films were deposited onto silicon substrates ina parallel-plate reactor using asymmetric bipolar pulsed dis-charges. Their frequency was varied between 50 kHz and 250 kHz(see Ref. [14]).

The precursors acetylene, isobutene, cyclopentene, toluene andcycloheptatriene were used in mixtures with argon. The thicknessof the deposited films was measured with a surface profilometerForm Talysurf 50 (Rank Taylor Hobson). A nanoindentation systemUMIS 2000 (CSIRO) equipped with a Berkovich-type diamondindenter tip was used to measure the hardness of the films. Thethickness of films used for hardness measurements was above 1 mmto prevent any influence of the substrate. The total hydrogencontent in the a-C:H films was measured by nuclear reactionanalysis (NRA).

Thermal desorption was analysed in vacuum with a base pres-sure of the setup below 10�5 Pa.

A heating rate of 3 K/min was applied. Thus full analogue massspectra in the range of mass numbers 1e100 u could be recorded atconstant temperature steps of 17.5 K.

The quadrupole mass spectrometer (QMS) Dycor LC-D 100M(AMETEK) was operated at 70 eV electron energy and at the lowestpossible electron emission current (0.1 mA) to minimize thermalmodification of effusion products. To increase the sensitivity of theTDS analysis, the pumping of the chamber was optimized. Byreducing the rotational speed of the turbo pump from 1500 Hz to900 Hz, the sensitivity of TDS measurements could be drasticallyincreased without approaching the upper pressure limit of theQMS operation. The mass-dependent sensitivity of the whole TDSsystemwas measured using a calibrated mixture of rare gases (He,Ne, Ar and Kr) and absolute partial cross-sections for electron-impact ionization of these gases from the study by Rejoub et al.[15]. The measured QMS signals were corrected for the probeda-C:H film volume (normalised to 1 mm thickness and 1 cm2 area).The gases H2, H2O, CO and CO2 were released also with increasingrate when heating up the empty TDS sample holder. When ana-lysing the thermal desorption from a-C:H, these continuous back-ground signals were fitted and subtracted.

3. Results and discussion

The parameters varied during the deposition experiments [14]were the pressure, the gas flow ratios (precursor to argon), the

frequency and the reverse time. For all precursors deposition at thelowest pressure of 2 Pa resulted in maximum film hardness. Thehardness of a-C:H films deposited from the five precursors undervery different process conditions behaves contrary to the totalhydrogen content of the films (see Fig. 4 in Ref. [14]). That means, ata constant deposition temperature (100 �C) the total hydrogencontent controls the hardness to a large extent.

When heated, the a-C:H films released mainly molecularhydrogen. Also the oxygen-containing gases H2O, CO and CO2 (butno O2) were measured in substantive quantities. Since CO and CO2are again present when a-C:H films, after intermediate exposure toair, are analysed in a subsequent effusion experiment, theirappearance is an indication of chemical reactions of a-C:H withwater or oxygen [16]. The amount of evolved water molecules fromhard a-C:H also in other studies exceeded the hydrocarbon signals[16,17], and water vapour is the second important gas.

Perhaps, just the water molecules may be responsible for thatamount of hydrogen in a-C:H films which is not bound to carbon.Several attempts were made on the quantification of unboundhydrogen in diamond-like carbon films. Empirical, semiquanti-tative relationships have been derived to calculate the content ofbound hydrogen in a-C:H films from the 2900 cm�1 IR band area[18], but the scatter of values is very large. In general, themeasurement of the bound hydrogen fraction based on Fouriertransform infrared spectroscopy (FTIR) [19] seems questionable,since the absorption strengths of individual CeH vibrations are notequal and they also depend on the film structure [11]. In a study byPei et al. [20] the quantification of unbound hydrogen was basedmerely on elastic recoil detection analysis (ERDA) of the hydrogencontent as a function of the annealing temperature. A reduced dropof the H content starting from 350 �C was attributed to a completedeffusion of unbound hydrogen. For the analysed sample, the frac-tion of unbound hydrogen was estimated to be 22.6%. But thevalidity of thesemeasurements is somewhat diminished, since theywere not made on pure a-C:H but on TiC/a-C:H nanocompositefilms, with titanium known to form hydrogen-rich hydride. Thefraction of unbound hydrogen in a-C:H seems to be rather low. Fora-C:H films deposited in DC plasmas from acetylene and cyclo-hexane the amount of hydrogen bound to carbon was found to bebetween 93% and 100% [21].

Analysing the films deposited using five precursors it wasobserved that with increasing film hardness the mass of thedetectable hydrocarbon molecules decreased. The same trend forinstance was observed in a study by Wild and Koidl [22]: filmsdeposited at low bias (jUBj < 300 V) released hydrogen, methaneand higher hydrocarbons (detected by means of C2H4

þ and C3H3þ);

medium bias films desorbed H2 and CH4 whereas from filmsdeposited at jUBj> 500 V only H2 was released. In the study byWildand Koidl [22] it was also experimentally demonstrated, thathydrocarbon diffusion is inhibited in hard/dense films by thereduced pore size of the strongly cross linked hydrocarbonnetwork. But there are also opposite findings: the helium barrierefficacy of a-C:H coatings was found to decrease with an increase infilm density [23]. This unexpected result was attributed to theformation of micro cracks at high intrinsic stress values. Not onlythe transport of species is hindered in hard films, but due to thelower density of CH3 and CH2 groups (sp3 and sp2 bound carbon)the thermally activated formation of higher hydrocarbons shouldalso be reduced. The nature of the diffusing species e radicals ormolecules e is discussed ambivalently. Delocalization of hydrogenatoms followed by surface recombination into hydrogen moleculeswas suggested as the mechanism of H2 evolution [13]. In a study byGerstenberg and Grischke [16] the formation of released moleculesis considered to involve the thermally activated rupture of neigh-bouring bonds of network modifiers (eH, eCH3) to the carbon

S. Peter et al. / Vacuum 86 (2012) 667e671 669

network, the reconstruction of broken C bonds and finally theformation of a hydrogen and/or hydrocarbon molecule.

To give an impression about the diversity of the released species,Fig. 1 shows the results for a polymeric sample deposited fromisobutene. Fig. 1a depicts the mass spectra in the range of 10e100 umeasured at three temperatures (406 �C, 510 �C and 613 �C, resp.),with the hydrocarbon signals peaking at the intermediatetemperature. The signals from molecular hydrogen are orders ofmagnitude larger. Fig. 1c shows the 510 �C results for the massrange above 35 u at higher amplification in order to allow easycomparison to mass spectra of some relevant hydrocarbons fromliterature [24] (Fig. 1b). Hydrocarbons upto benzene were alreadydetected in previous studies [17].

Except for CH4, hydrogen deficient gases are more abundantthan fully hydrogenated ones, in agreement with most reportedresults. In conflict with that, in a study by Ristein et al. [11] thedominant effusion product from a polymeric a-C:H film was iden-tified as the saturated hydrocarbon propane C3H8. But the massnumbers 44 u and 28 u used as an indicative of propane correspondalso to the gases CO2 and CO, which are two typical constituents ofthe residual gas.

Fig. 1. Thermal effusion spectra of a polymeric film deposited from isobutene (a) at406 �C, 510 �C and 613 �C, resp., and a comparison of 70 eV mass spectra of somehydrocarbons [24] (b) to the 510 �C mass spectrum (c).

An analysis of the temperature-dependent contribution ofindividual gases to the hydrogen loss of harder a-C:H films is pre-sented in Fig. 2 for two different magnifications. Two filmsdeposited from isobutene at pressures of 2 Pa (film hardness20.0 GPa) and 10 Pa (17.3 GPa) are compared. Fig. 2a also shows thetotal hydrogen loss of the two films (summed up over all hydrogen-containing species and accumulated over the temperature range) asa function of the temperature (“summed up total H-loss”), whichfinally was 2.1 times higher for the softer film. The total contribu-tions of the most relevant gases are presented in Table 1. The smallsignals from ethylene (C2H4) were not included into the balancebecause of the superimposed much stronger CO background.

The depicted hydrogen signals of individual species areproportional to the number of hydrogen atoms dNH,i transported bya number dNi of molecule i containing nH,i hydrogen atoms (all perunit time dt):

dNH;i

dt¼ nH;i$

dNi

dt(1)

The measured QMS current signal of an ionic species i, Iiþ, isproportional to the desorption rate of the molecules dNi/dt andtheir partial electron-impact ionization cross-section for themeasured parent or fragment ion:

Iþi fdNi

dt$sið70 eVÞ$si (2)

The factor si corrects for the mass-dependence of pumping andof the quadrupole mass analyser. For the partial electron-impactionization cross-section at 70 eV electron energy the data fromstudies by Shirai et al. [25] and [26] were used.

Fig. 2. Temperature-dependent evaluation of individual hydrogen-containing gasesfrom two films deposited at different pressures from isobutene together with thesummed up total H-loss.

Fig. 4. Normalised to the same peak high argon desorption spectra of four films withdifferent hardness deposited in cycloheptatrieneeargon discharges.

Table 1Percentage contribution of gases to the hydrogen release from films deposited at2 Pa and 10 Pa from isobutene.

Pressure H2 CH4 H2O C2H2 C2H6

2 Pa 70.88 0.36 28.62 0.06 0.1010 Pa 78.63 0.60 20.39 0.06 0.32

S. Peter et al. / Vacuum 86 (2012) 667e671670

The very low contribution of hydrocarbons to the hydrogeneffusion (see Table 1) agrees with published data about the carbonrelease: when heated upto 1400 K a hard a-C:H film lost as low asabout 1% of all carbon atoms [16]. The distribution of gaseousproduct was also analysed for only 6 monolayer thick a-C:H filmsdeposited from ethane, ethylene and benzene [17]. The hydrogencontent of these ion gun deposited films varied between about 42%(precursor C2H6) and about 22 at.% (precursor C6H6). From thereleased molecular products, about 90% were hydrogen molecules.

Fig. 3 shows the influence of the pressure during depositionfrom an acetyleneeargon mixture on the temperatures at whichthe evolution of several gases starts. In contrast to the thresholdtemperatures, the temperatures of maximum effusion during TDSmeasurements are affected by the heating rate (for example seeRef. [27]) and hence are not a pure film characteristic. As a quanti-tative measure for a “starting temperature” the temperature atwhich the QMS-signal reaches 1/10 of its maximumvaluewas used.

The predominant effusion products are released above 500 �C.For films deposited at lower pressures, that means harder films, thethermal stability is further increased.

For methane two clearly separated peaks could be distin-guished. The signals from C2Hx species were first detected attemperatures in the range from 300 �C to 350 �C. Due to their lowintensities a second high temperature peak was only occasionallyvisible.

A pronounced two-peak structure was reported for the evolu-tion of methane [4,28] and ethane [4] from RF plasma depositedfilms. The low-temperature evaluation peak was attributed to thedesorption of species from an interconnected void network and thehigh-temperature peak to gas species trapped in isolated voidswhich break up when the pressure inside the voids exceedsa certain value or the films changed their structure [4]. A secondpeak was also found in the effusion rate of helium implanted toa-SiC:H films [29]. It was attributed to He trapped again in isolatedvoids. As another relevant fact, films are not homogeneous: dense(2.4 g/cm3) a-C:H films on silicon substrate were found to consist ofthree layers: surface (z60 nm thick), bulk and interface (z50 nm)

Fig. 3. Influence of the pressure during deposition from an acetyleneeargon mixtureon the onset-temperatures for the effusion of several species, with the methaneeffusion being characterized by separated low temperature (LT) and high temperature(HT) peaks.

with different degrees of carbon cross linking [13] and surely alsowith different thermal stability.

The argon gas added to the hydrocarbon discharges was de-tected in small portions in the TDS experiments, too. Since argonatoms are not bound, their out-diffusion can give information aboutthe voids present in the network. Fig. 4 shows the argon signals enormalised for better visibility e from four films deposited fromcycloheptatrieneeargon discharges. The ratio of the peak valuesbefore normalisation was 1.8:3.9:9.4:100 in the order of increasingfilm hardness. With increasing film hardness the argon peaks areshifted to higher temperatures and they narrow. After all, thequantity of released argon gas increases considerably with thehardness of the a-C:H. The results about the starting temperaturefor argon effusion for films deposited from all five investigatedprecursors are shown in Fig. 5. When these starting temperatures,as in Fig. 5, are plotted over the hardness of the films, the scatter ofdata points is significantly lower than in a temperatureehydrogencontent diagram. The same is valid for the H2 effusion data shownin the following Fig. 6.

The percentage of argon incorporated into a-C:H films and itsthermally activated release was studied only by few authors. Upto0.3 at.% Ar was incorporated to DLC films deposited by ion platingfrom C6H6eArmixtures [30]. FromXPS-spectrae as expectede theincorporated argonwas found to be not bonded with carbon. UsingRF PECVD the films deposited from acetyleneeargon mixtures(60 sccm C2H2 þ 0e50 sccm Ar) contained upto 0.44 at.% Ar [31]. Inthe study by Gerstenberg and Grischke [16] the temperature ofmaximum argon effusion was found to increase as the hydrogenconcentration of a-C:H decreases.

Fig. 5. Correlation of the starting temperature for argon effusion with the hardness ofa-C:H films deposited from five precursors in mixtures with argon.

Fig. 6. Correlation of the starting temperature for molecular hydrogen effusion withthe hardness of a-C:H films deposited from five hydrocarbons.

S. Peter et al. / Vacuum 86 (2012) 667e671 671

The relation of the film hardness to the starting temperature ofmolecular hydrogen effusion is shown in Fig. 6. With increasinghardness of the as-deposited film, the temperature necessary todesorb H2 increases continuously. For films harder than about13 GPa the onset-temperatures for hydrogen are practically thesame as for argon release (see Fig. 5). From softer films argoneffusion sets in much earlier. Similarly, in a study by Schwan et al.[10] it was reported, that the Ar desorption starts at 377 �C, 250 Kearlier than the hydrogen effusion.

As for the Ar desorption (see Fig. 4), the H2 effusion spectra alsonarrow with increasing film hardness. Similarly, analysing a-C:Hdeposited in capacitively coupled 13.56 MHz discharges it wasobserved [22] that low-bias films desorbed hydrogen over a nearly400 K broad temperature range. With increasing UB the effusionspectrum sharpened to about 150 K width.

Summing up, the effusion characteristics of argon should bedetermined by the changes in the porosity of a-C:H films. Theinfluence of a heat treatment on the porosity of a-C:H was char-acterized by a gas absorption method [32]: heat treatment firstopens micropores in the slightly mesoporous film. Further heatingaccompanied by the effusion of gases increases the number andsize of micropores, and also produces mesopores. In contrast toargon, molecular hydrogen first has to be formed by thermal acti-vated reactions.

4. Conclusions

Hydrogenated amorphous carbon (a-C:H) films have beeninvestigated by mass spectroscopic thermal effusion measure-ments. The films were deposited from five hydrocarbons (acety-lene, isobutene, cyclopentene, toluene, and cycloheptatriene) withan argon admixture in mid-frequency pulsed discharges. Depend-ing on the hardness of the as-deposited films, hydrogen andhydrocarbon molecules, water and argon are desorbed at thresholdtemperatures between 300 �C and 650 �C. Molecular hydrogen isthe main product. Analysing the contribution of individual gases tothe total hydrogen release it was concluded, that possibly the watercontent of the films represents the fraction of hydrogen not boundto the carbon in a-C:H films. The threshold temperature for theeffusion of molecular hydrogen increases almost linearly with thea-C:H hardness. The threshold temperature for the effusion of

argon from films with a hardness larger about 13 GPa is practicallyequal to that of H2. From softer films Ar is released at significantlylower temperatures. The argon effusion spectra are related to thechanges in the size and number of micropores in the amorphousfilms.

The effused gases and the threshold temperatureswere found tobe related to the a-C:H structure with the film hardness asa measure e independent of the precursor.

Acknowledgements

The authors gratefully acknowledge the support of this work bythe German Research Foundation (DFG) within the TransregionalCollaborative Research Centre 39 PTPIESA “Production Technolo-gies for light metal and fibre reinforced composite based compo-nents with integrated PIEzoceramic Sensors and Actuators”.

References

[1] Robertson J. Mat Sci Eng 2002;R37:129e281.[2] Casiraghi C, Piazza F, Ferrari AC, Grambole D, Robertson J. Diamond Relat

Mater 2005;14:1098e102.[3] Peter S, Graupner K, Grambole D, Richter F. J Appl Phys 2007;102:1e18.

053304.[4] Jiang X, Beyer W, Reichelt K. J Appl Phys 1990;68:1378e80.[5] Jiang X, Reichelt K, Stritzker B. J Appl Phys 1990;68:1018e22.[6] Redhead PA. Vacuum 1962;12:203e11.[7] Wu RLC, Miyoshi K, Vuppuladhadium R, Jackson HE. Surf Coat Technol 1992;

54/55:576e80.[8] Tallant DR, Parmeter JE, Siegal MP, Simpson RL. Diamond Relat Mater 1995;4:

191e9.[9] Walters JK, Fox DM, Burke TM, Weedon OD, Newport RJ, Howells WS. J Chem

Phys 1994;101:4288e300.[10] Schwan J, Batori V, Ulrich S, Ehrhardt H, Silva SRP. J Appl Phys 1998;84:

2071e81.[11] Ristein J, Stief RT, Ley L, Beyer W. J Appl Phys 1998;84:3836e47.[12] Takabayashi S, Okamoto K, Sakaue H, Takahagi T, Shimada K, Nakatani T.

J Appl Phys 2008;104:1e6.[13] Ralchenko VG, Loubnin EN, Popov AV, Strelnitsky VE. Diamond Relat Mater

1992;1:345e9.[14] Peter S, Günther M, Hauschild D, Grambole D, Richter F. Vacuum 2010;84:

958e61.[15] Rejoub R, Lindsay BG, Stebbings RF. Phys Rev A 2002;65:1e8. 042713.[16] Gerstenberg KW, Grischke M. J Appl Phys 1991;69:736e44.[17] Schenk A, Winter B, Biener J, Lutterloh C, Schubert UA, Küppers J. J Appl Phys

1995;77:2462e73.[18] Buijnsters JG, Gago R, Jiménez I, Camero M, Agulló-Rueda F, Gómez-

Aleixandre C. J Appl Phys 2009;105:1e7. 093510.[19] Martinu L, Raveh A, Boutard D, Houle S, Poitras D, Vella N, et al. Diamond Relat

Mater 1993;2:673e7.[20] Pei YT, Chechenin NG, Chernykh PN, Turkin AA, Vainshtein D, De Hosson JThM.

Scripta Materialia 2009;61:320e3.[21] Donnet C, Fontaine J, Lefèbvre F, Grill A, Patel V, Jahnes C. J Appl Phys 1999;85:

3264e70.[22] Wild Ch, Koidl P. Appl Phys Lett 1987;51:1506e8.[23] Valentini L, Bellachioma MC, Lozzi L, Santucci S, Kenny JM. J Vacuum Sci

Technol A 2002;20:1647e52.[24] Eight peak index of mass spectra. 2nd ed. Aldermaston, England: Mass

Spectrometry Data Centre; 1974.[25] Shirai T, Tabata T, Tawara H. At Data Nucl Data Tables 2001;79:143e84.[26] Shirai T, Tabata T, Itikawa Y. At Data Nucl Data Tables 2002;80:147e204.[27] Su C, Song K-J, Wang YL, Lu H-L, Chuang TJ, Lin J-C. J Chem Phys 1997;107:

7543e58.[28] Bourgoin D, Turgeon S, Ross GG. Thin Solid Films 1999;357:246e53.[29] Saleh R, Munisa L, Beyer W. Appl Surf Sci 2007;253:5334e40.[30] Zhang W, Tanaka A, Wazumi K, Koga Y. Thin Solid Films 2002;416:145e52.[31] Jones BJ, Wright S, Barklie RC, Tyas J, Franks J, Reynolds AJ. Diamond Relat

Mater 2008;17:1629e32.[32] Yin Y, Collins RE, Pailthorpe BA. J Appl Phys 1992;71:3806e11.