effect of deposition parameters on different stages of diamond deposition … · 2005-02-09 ·...

Diamond and Related Materials 13(2004) 74–84

0925-9635/04/$ - see front matter� 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.diamond.2003.09.003

Effect of deposition parameters on different stages of diamond depositionin HFCVD technique

A.K. Dua *, V.C. George , M. Friedrich , D.R.T. Zahna, a b b

Novel Materials and Structural Chemistry Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400085, Indiaa

Institut furPhysik, TU Chemnitz, Chemnitz, Germanyb ¨

Received 2 December 2002; received in revised form 21 May 2003; accepted 23 September 2003

Abstract

The effect of applied bias and system pressure on different stages in the deposition process occurring during diamond depositionon mirror polished Si single crystal substrates in the HFCVD technique has been investigated. For characterization of the deposit,Fourier Transform Infrared Spectroscopy and Laser Raman spectroscopy have been used as the primary techniques while X-raydiffraction and scanning electron microscopy served as supplementary techniques. The findings are that the use of an overall lowsystem pressure;1.3 mbar and application of an appropriate bias to the substrate during the initial stages facilitates the growthof highly oriented cubic silicon carbide film on mirror polished Si single crystal substrates under conditions conventionally usedto grow diamond in HFCVD technique. A critical pressure ‘regime’;26 mbar has been found to exist during the growth stagefor the concurrent growth of SiC and diamond such that only SiC grows below this pressure and diamond starts growing at orabove this value. It is suggested that around this transition region, SiC gets partly converted to nanocrystalline diamond structuredcarbon, which then grows into bigger crystals under appropriate growth conditions.� 2003 Elsevier B.V. All rights reserved.

Keywords: HFCVD; Pristine Si single crystal substrate; Different stages of growth; Effect of bias, pressure; SiC; Conversion of SiC to diamond

1. Introduction

Chemical vapour deposition is a generic name for agroup of processes that involve depositing a solidmaterial on a substrate by activating the precursors ingaseous phase and making them react chemically. Gasphase processes, surface chemistry, nucleation andgrowth phenomena are amongst the important decidingstepsw1x. Each step may have its own specific set ofoptimized deposition parameters and the values actuallyused may affect the different steps thereby influencingthe quality, uniformity and characteristics of the finallayer deposited. In particular, it is during nucleationstage that many important properties of the film, suchas density of crystallites and their alignment are deter-mined. Diamond on silicon forms an intensive researcharea due to its potential applications in microelectronicsand optics. Diamond film deposition by chemical vapour

*Corresponding author. Tel.:q912225593827;fax: q912225505151.

E-mail address: [email protected](A.K. Dua).

deposition methods{ e.g. hot filament assisted chemicalvapour deposition(HFCVD) and microwave assistedchemical vapour deposition(MWCVD)} , using conven-tional parameters and activated methane: hydrogen gasmixture, favours growth of crystalline diamond(by aseries of complex reactions involving hydrocarbon rad-icals in a predominant atomic hydrogen environment)but does not favour nucleation. Diamond nucleation onpristine silicon substrates is usually associated with anincubation period, high localisation and low nucleidensity of;1=10 cm w2x It is a well-known fact4 y2

.

that the nucleation process is strongly affected by thesubstrate surface morphology, which in turn may bechanged by chemical reactions, ion bombardment orother pretreatments. Over a period of time, severaleffective ways have been discovered to prompt diamondnucleation. Scratching the substrate with diamond crys-tallites has been the most popular technique, withsonication in diamond slurry giving more uniformdeposits. However, the resulting mechanical damagenormally produces poor quality diamondysubstrate inter-

75A.K. Dua et al. / Diamond and Related Materials 13 (2004) 74–84

faces—a situation not desirable for device applications.Neverthless, inspite of diamond slurry sonication, Tach-ibana et al.w3x could produce not only good interfacebut also highly oriented{ 111} epitaxial diamond filmson Pt { 111} yIr { 111} yPt { 111} ysapphire { 0001}employing MWCVD. Rather than scratching, a betterway for not only enhancing nucleation but also inducingoriented growth is bias enhanced nucleation and hasspecially been used in MWCVD techniquew4x. As forthe HFCVD technique is concerned, there does not existsubstantial literature on similar bias related work. Thereason for this is that under standard operating condi-tions, there are hardly any charged entities present andsimply biasing the substrate as in MWCVD is notexpected to produce enough ion bombardment for het-ero-epitaxial diamond nucleation. Further, under typicalpressure conditions of several tens of mbar, it is hardlypossible to excite and sustain a glow dischargew5x forgenerating ions and enhancing their concentration. Thereare some reportsw6,7x of negative bias enhanced nucle-ation and textured growth, which has been attributed tosecondary electron emission from the diamond coatedsubstrate holder or the wall of the reaction chamber.However, the reproducibility is very poor, for the simplereason that secondary electron emission properties ofdiamond are highly sensitive to the surface conditionand thermal history. Further, under high bias voltages(approx. 300 V) and currents(1–10 mAycm ), the2

discharge is in abnormal mode and unstablew8x. Nev-erthless, Stubhan et al.w9x and Chen et al.w10x foundthat, under negative bias, diamond nucleates and growson silicon hetroepitaxially, though it was restricted tosmall area. Zhou et al.w11x and Arnault et al.w12x,however, tried to overcome the unstability problem byusing double biasing arrangement and have reportedenhanced nucleation density and epitaxial nucleation.An alternative pretreatment procedurew13x, comprisingof heavy seeding by directly sprinkling diamond powderon the nickel single crystal substrate followed by in-situannealing in pure hydrogen, has also been successfullyused to grow oriented diamond films. Wan et al.w14xtheoretically worked out, employing a non-equilibriumthermodynamic model, the phase diagram of CVDdiamond growth covering the effects of the gas phasecomposition, substrate temperature, system pressure andprovided a means for general prediction about suitableconditions for the process. Brunsteiner et al.w15x, Harriset al. w16x and Schwarz et al.w17x, however, experimen-tally investigated the effect of pressure on growth rate,quality and morphology of the diamond coatings in aHFCVD process. Brunsteiner covered a pressure regionbetween 7 and 660 mbar whereas Schwarz’s studyreferred to an industrial plant, SiC substrates and pres-sures between 1 and 100 mbar. Recently, it has beenshown that the pressures conventionally used in theHFCVD technique is not that suitable for the nucleation

step and that the use of extraordinary low pressure(0.13–1.32 mbar) results in a high nucleation densityon pristine Siw2x. Also, the oxide layer present on theSi surface or getting formed during the initial depositionstage, inhibits nucleationw2x. This calls for the use ofpure processing gases containing minimal amount ofoxygen and getting rid of other sources of oxygen inthe system. The present paper reports our findings onthe effect of applied bias and system pressure ondifferent stages in the deposition process occurringduring diamond deposition on mirror polished Si singlecrystal substrate in the HFCVD technique. Our contri-bution is comparative as well as complementary innature. In particular, we report, in the growth stage, theoccurrence of a transition region in the system pressure(approx. 26 mbar), so that only SiC grows at lowerpressures and diamond starts growing around this orhigher pressures. It is speculated that around this tran-sition region, SiC gets partly converted to nanocrystal-line diamond structured carbon, which then grows intobigger crystals under appropriate growth conditions.

2. Experimental details

Substrates used are one side polished, p-type, Si singlecrystal wafers with(111, 100) orientation from WackerChemitronic, GmbH. These are sonicated first in trichlo-roethylene, then in acetone and finally in isopropylalcohol for 5 min each. The samples are then subjectedto a 2 min etch in 40% HF and a rinse in de-ionisedwater. It is important to note that these substrates arenot subjected to conventional diamond slurry sonicationtreatment. The films are grown on the polished side ina low-pressure HFCVD facility described in our earlierpublicationsw18,19x. Two sets of experiments are per-formed, one restricted to nucleation stage only and theother covering ‘nucleationqgrowth’ stage. Details aboutthe deposition parameters used are given in Table 1. Inthe second set of experiments, three different types ofsamples have been prepared. Bias is present throughoutthe deposition in the first type of samples. However, inthe second type of samples, no bias is present duringnucleation as well as growth stages whereas in the thirdtype of samples, negative bias is present during nuclea-tion stage and absent during growth stage. It is to benoted that for second and third type of samples, differentparameters have been used in the two stages. Pressureis measured using a Strain gauge pressure-indicator,model DPI 260, Druck Ltd., England and the gas flowis controlled using Qualiflow mass flow controllers. Formeasuring the filament temperature, a Raytek two wave-lengths optical pyrometer is used. The substrate temper-ature is chosen keeping in mind the fact that its valuefor maximum nucleation density is lower than that formaximum growth ratew20x. It is measured using a Pt,Pt-14% Rh thermocouple. Film characterization is done

76 A.K. Dua et al. / Diamond and Related Materials 13 (2004) 74–84

Table 1Typical deposition conditions

Parameters Experiments covering

‘Nucleation’ stage Both nucleation and growth stagesonly, i.e. initial

Second type of samples Third type of samples;20 min of film

First type Nucleation Growth Nucleation Growthdeposition

of samples

Bias (V) 0, y350,y415,q300 y350 0 0 y350 0System ;1.3–52 ;1.3–52 ;1.3 5–159 ;1.3 52pressure(mbar)H flow2 ;50–400 ;50 ;50 700 50 400rate(sccm)CH flow4 ;1–4 1 1 7 1 4rate(sccm)Filament ;2100 1950 1950 2100 1950 2020temp.(8C)Substrate ;750 750 750 850 750 750temp.(8C)Filament to ;5 ;5 ;5 ;5 ;5 ;5substratedistance(mm)Deposition ;20 90 15 180 20 up to;1200time (min)

Filament – W coil, wire diameter: 0.5 mm, Substrate – Si(100) and Si(111) wafers, CH – 99.97% pure, H – 99.999% pure.4 2

ex situ and is mainly centred on the detection of SiCand diamond phases using Fourier transform infra-redspectroscopy(FTIR) for the former and Laser RamanSpectroscopy(LRS) for the latter. The two techniquescomplement each other since FTIR has good detectionsensitivity for SiC and rather poor detection sensitivityfor diamond (In fact, its single–phonon absorption isnormally disallowed in pure diamond and becomesactive in CVD diamond as a result of defects whichdisrupts the translational symmetry of the host latticew21x) whereas LRS has a good detection sensitivity fordiamond and poor detection sensitivity for SiC. ABruker model IFS 66 FTIR instrument together with a208 reflection accessory and LABRAM-1, ISA makemicroymacro Raman spectrometer in back scatteringgeometry are used in the present study. Spectral resolu-tion of both the instruments is;2 cm and Ramany1

spectra are taken using 514.5-nm wavelength from anAr laser. X-Ray diffraction(XRD) and scanning elec-q

tron microscopy(SEM) are used to supplement thecharacterisation. XRD patterns are recorded in a PhilipsX-ray diffractometer PW 1710 using CuKa line froman X-ray generator operated at 30 kV and 20 mA. Allmicrographs are recorded in a Philips SEM-515 high-resolution scanning electron microscope.

3. Results and discussion

On activation at or close to the heated filament, thefeed gases form precursor species whose type andamount is governed by the processing parameters such

as filament temperature, feed gas composition, pressureof the precursor gas etc. These are then transported tothe substrate and the mode of transport is affected bythe system pressure and temperature gradients. Substratetemperature and applied bias also affect the nucleationand growth characteristics of the film.The present study distinguishes the two stages of film

formation, i.e. initial ;20 min of film depositionreferred to as nucleation stage and subsequent depositionfrom ;20 min onwards, referred to as growth stage.Investigations cover the effect of bias and system pres-sure on samples, which fall under nucleation stage aswell as those, which have undergone both nucleationand growth.

3.1. Effect of bias

Herein the effect of bias is considered on nucleationstage only. Samples are prepared at;1.3 mbar pressureunder different bias conditions for approximately 25 min(i.e. nucleation stage). Fig. 1 shows the reflection FTIRspectra for bias voltages of 0 V(Fig. 1a), y350 V(Fig. 1b), y415 V (Fig. 1c) andq300 V (Fig. 1d),respectively. Predominant feature is the occurrence of apeak at;800 cm which has been ascribed to the Si–y1

C stretching vibrations inb-SiC w18,19,22x. Changes inits peak intensity, peak shape and its FWHM are appar-ent. Comparing Fig. 1a and b, the FWHM for thenegatively biased sample(approx. 52 cm ) is muchy1

less than that of the unbiased one(approx. 70 cm ).y1

A rather large FWHM may be due to distortions in the


Fig. 1. Reflection FTIR spectrum of the deposit on Si(111) showingthe effect of applied bias during initial deposition(i.e. ‘nucleationstage’) at 1.3 mbar(a) 0 V, 15 min; (b) y350 V, 20 min;(c) y415V, 25 min and(d) q300 V, 20 min. For clarity, the curves(b) and(c) have been shifted alongy-axis so as to avoid overlap.

Fig. 2. Reflection FTIR spectra showing the effect of pressure duringnucleation stage in the absence of applied bias.(a) 1.3 mbar, 25 min,CH : 1 sccm, H : 50 sccm(b) 52 mbar, 30 min, CH : 4 sccm, H :4 2 4 2

400 sccm.

film composition, in both the Si–C bond angle and theSi–C bond length. On increasing the negative bias fromy350 V (Fig. 1b) to y415 V (Fig. 1c), there isappreciable decrease in the peak intensity. This may beattributed to a sputtering effect since the sputtering yieldis generally known to increase with the increase inenergy of the incident ions, having energies in the rangeof present experimentsw23x. X. Jiang et al.’s w24xobservation of decrease in nucleation density withincrease in negative bias voltage beyond 150 V isconsistent with our results. Change of bias polarity fromy350 V (Fig. 1b) to q300 V (Fig. 1d) can be seen todecrease the peak intensity. Thus application of bias tothe substrate(with respect to the filament) does affectthe nucleation stage and its magnitude depends on biasamount as well as its polarity. It means that in thepresent HFCVD environment both positively and nega-tively charged species are present though with differentindividual number density and with the overall densityexpected to be much less than say in MWCVD. In thecase of positive bias to the substrate, electrons, ther-moionically emitted from the filament, will also bebombarding the growing film and playing some role.For instance, these have been shown to facilitate dia-mond growth at substantial lower substrate temperaturein HFCVD techniquew25x. Negative bias of 350 Vdefinitely enhances the amount and crystallinity of thedeposit. This follows from XRD and AFM studies andhas already been reported elsewherew18,19x.Let us consider carbon incorporation in silicon. Car-

bon may go into substitutional sites of silicon. However,its solubility in silicon, at thermal equilibrium, is low((10 ). Larger doses of carbon result in the growthy5

of SiC precipitates, which may be considered precursorsfor the growth of a SiC top layer.

3.2. Effect of pressure

3.2.1. Nucleation stage(a) Without bias – Samples were prepared at several

pressures starting from low;1.3 mbar and going up torelatively high value;52 mbar. These were taken outof the system just after initial deposition stage andanalysed. Fig. 2a,b show the FTIR spectrum of thesamples prepared at;1.3 mbar and;52 mbar, respec-tively. A rather broad peak assigned to SiC and occurringat;800 cm can be seen. Its intensity is much morey1

perceptible in Fig. 1a while in Fig. 1b it is very lowand near the detection limit. Broadness of the peaks alsopoints to the deposit being less crystalline. It followsthat the low pressure of;1.3 mbar is more suited forSiC formation compared to high pressure of;52 mbar.Even the sample of Fig. 2a generally did not show anySiC peak in its X-ray diffraction pattern. If at all presentunder some conditions, signals are very weak andamorphous like. Laser Raman spectroscopy could notdetect the presence of SiC or diamond. It is understand-able because of the poor Raman sensitivity of SiC.However, the technique is quite sensitive for diamondand its non-detection suggests its absence. Fig. 3a,bshow, respectively, scanning electron micrographs of thedeposit prepared at;1.3 mbar and at;52 mbar. It canbe clearly seen that the nucleation density observed atthe lower pressure is quite appreciable whereas it isnegligibly small at the higher pressure.(b) With bias – Here again the samples have been

prepared at a number of pressures. However, Fig. 4shows the Reflection FTIR spectra at two pressures only(a) 1.3 mbar and(b) 52 mbar, the bias applied beingy350 V. In spite of the feed-gas flow rates being morefavourable, the area under the 800 cm peak in Fig.y1

4b is smaller than that in Fig. 4a. This points to higher


Fig. 3. SEM pictures showing nucleation density of typical samples prepared without bias at a pressure of(a) 1.3 mbar and(b) 52 mbar.

pressures being less favourable for nucleation, a resultbroadly similar in nature to what was observed withoutbias. However, here, the nucleation density is compara-tively more and SiC deposit is more crystalline. It is tobe noted that no signals due to diamond are detected inthe Raman spectrum. Data and the analysis of the results,at low pressure and in the presence of bias, have beenpublished elsewherew18,19x.The pressure affects the deposition process in several

ways. Firstly, the concentration of the active speciesproduced at or close to the filament is proportional tothe pressure of its precursor or the reactant gas. Further,since these have to reach the substrate before anynucleation or growth can take place, the fraction thatsurvives collision to arrive at the substrate again dependson the pressure. With decreasing pressure, the mean freepath increases and this leads to an increase in thenumber of active species impinging on the substrate.

The two effects, being in the opposite direction, obvi-ously need balancing and hence there exists an optimumvalue. The present study indicates this optimum valuein the nucleation stage to be;1.3 mbar, where themean free path falls in the millimeter range, which iscomparable to the distance between the filament and thesubstrate. At this pressure, more active species impingeon the substrate thereby creating a more favorablecondition for nucleation. Also, the larger mean free pathat low pressure results in a decrease in the number ofcollisions that these species undergo during their traverseso that their kinetic energy remains close to the localthermal energy of the hot filament(approx. 21008C).Species with a higher kinetic energy have a largersurface mobility, which helps promote the aggregationof the precursors and thus increases nucleation andgrowth on the substrate. In addition, the bombardmentof particles with higher kinetic energy leads to improved


Fig. 4. Reflection FTIR spectra showing the effect of pressure duringnucleation stage in the presence ofy350 V applied bias.(a) 1.3 mbar,20 min, CH : 1 sccm, H : 50 sccm and(b) 52 mbar, 20 min, CH : 44 2 4

sccm, H : 400 sccm. For clarity, the curve(a) has been shifted up2

and the curve(b) down alongy-axis so as to avoid overlap.

Fig. 5. Reflection FTIR spectra of the samples, prepared for 1.5 h(i.e. covering ‘nucleationqgrowth’ stage) with a bias ofy350 V andat pressure of(a) 1.3 mbar,(b) 7 mbar,(c) 13 mbar,(d) 26 mbarand(e) 53 mbar.

surface cleanliness and higher probability of surfacereconstruction or microstructure modification, conditionswhich enhance surface adsorption and aggregation, thusnucleation. Furthermore, the availability of more atomichydrogen and with higher kinetic energy should helpremove the oxide layer on Si, which is known to hinderthe adsorption, aggregation and nucleation. The lowpressure helps the process because under these condi-tions the oxide is more easily decomposed andvapourised.

3.2.2. Nucleationqgrowth stageThis section covers three types of samples. In the first

type, negative bias is present during ‘nucleation’ stageas well as growth stage with pressure remaining thesame throughout the deposition. In the second type ofsamples, bias is absent both during ‘nucleation’ andgrowth stages whereas for the third type of samples,negative bias is present during the ‘nucleation’ stageand is absent during the growth stage. In all the threetypes of samples, nucleation is carried out at lowpressure of;1 mbar and growth at different pressuressuch as 5, 10, 17, 26, 53, 106 and 159 mbar. It may benoted that for the second and third type of samples,deposition parameters including the pressure are differ-ent during the two stages.

3.2.2.1. First type of samples. Fig. 5a–e, respectively,show the Reflection(208) FTIR spectra of the first typeof samples, prepared for;1.5 h, under negative biasand at different pressures of 1.3 mbar, 7 mbar, 13 mbar,26 mbar and 53 mbar. It can clearly be seen that theintensity of;800 cm peak, attributed tob-SiC, isy1

maximum at 1.3 mbar and it decreases as the pressureincreases. It is still perceptible up to 13 mbar but is

hardly detected at 26 mbar pressure. X-Ray diffractionpattern of these samples is shown in Fig. 6. It can beseen that at 1.3 mbar, a sharp peak due tob SiC at2u;368 is present. However, as the pressure increasesto 7 mbar and above, it is hardly perceptible. However,from 13 mbar onwards, diamond(111) peak is moreand more noticeable. Obviously, at lower pressure crys-talline b SiC is present and in the higher pressureregime it is not detectable but diamond starts appearing.

3.2.2.2. Second type of samples. Herein nucleation hasbeen carried out, without bias, at a pressure of 1.3 mbarbut during growth stage different values of pressurewere used. However, Fig. 7 compares the reflectionFTIR spectra at two growth pressures only(a) ;1.3mbar and(b) 26 mbar, the deposition time being;3 h.800 cm b-SiC peak is present in both the cases buty1

it is much more predominant in the former case. Further,the spectrum is quite different in the latter case withoscillations being present. SEM, however, does revealthe difference. Whereas in the former case no diamondis seen(Fig. 8a), the latter does show diamond crystal-lites with well delineated edges and faces(Fig. 8b).Their size is ;4–8 mm, and particle density is


Fig. 6. X-Ray diffraction patterns of the samples, prepared for 1.5 h(i.e. covering ‘nucleationqgrowth’ stage) with a bias ofy350 V andat pressure of 1.3 mbar, 7 mbar, 13 mbar, 26 mbar and 53 mbar,respectively.

;5.5=10 ycm . The observation is supported by the6 2

presence of diamond(111), (220) and(311) diffractionlines in the X-ray diffraction pattern(data not beinggiven) and 1332 cm peak, a signature of cubicy1

diamond, in Raman spectrum(Fig. 9). In addition thespectrum contains a broad hump;1560 cm and thisy1

is attributed to sp -bonded carbon. It should be noted2

that Raman spectroscopy is approximately 75-fold moresensitive to sp bonded carbon sites than it is to2

crystalline diamond when 514.5 nm excitation is usedw26x. However, SiC is not detected because of the poorRaman sensitivity of SiC and small thickness of thefilm w27x. Similar behavior is seen for pressures largerthan 26 mbar and up to 158 mbar. This result is oppositeto that reported by Lee et al.w2x but more in line withthat of Harris et al.w16x. The reason may be that theformer work deals with just the initial 15 min of growth,when two dimensional growth dominates whereas thelatter work and the present work refers to after somehours of growth, when three-dimensional growthprevails.

3.2.2.3. Third type of samples. These samples have alsobeen prepared at different pressures during the growthstage and the behaviour is similar to the second typeexcept that it is much more pronounced, i.e. presence

of bias during nucleation stage at;1.3 mbar helps ingetting many more diamond crystallites during growthstage at pressures around or)26 mbar under otherwiseidentical conditions.It is possible to grow, under optimized conditions, a

22-mm thick high quality continuous polycrystallinediamond film in;20 h. Fig. 10a is the scanning electronmicrograph of such a film showing both its surfacemorphology and cross-section. The crystallites have gotwell-delineated edges and faces and their size is up to;20 mm. The Raman spectrum of the film on Si(100)substrate is shown in Fig. 10b. It is worth noting thatthe FWHM of the 1332 cm peak is;4 cm , they1 y1

non-diamond content is very small and the backgroundis almost flat. This points to better quality of the deposit.The finding is consistent with the reports in the literaturew12x, that bias enhanced nucleation leads to strongincrease in nucleation density.

3.3. Existence of a critical pressure regime and conver-sion of SiC to diamond

From the analysis of the data given above for thethree types of samples, it is clear, without any doubtthat there exists, during the growth stage, a criticalpressure ‘regime’ so that only SiC exists below thisvalue and diamond starts appearing above it. However,in the first type of samples, the situation appears to bea little bit different and the critical pressure ‘regime’value is not exactly the same as that in second and thirdtype of samples. Since in this paper, no claim has beenmade for the precise quantitative determination of this


transition regime value for the concurrent growth of SiCand diamond, the results discussed refer only to twotypical pressures(one approx. 8 mbar and the otherapprox. 26 mbar, supposedly the critical value) ratherthan giving the coating characterization data at all thepressures investigated.It may be worthwhile to understand the occurrence of

this transition regime. Since at a pressure higher thanthat corresponding to the transition value, the mean freepath is appropriately less so that there will be morecollisions and therefore less atomic hydrogen will bereaching the growing film. Atomic hydrogen is knownto erode the nucleation sites thereby making themdisappear. Since at pressures higher than that at thetransition region, less atomic hydrogen is available atthe substrate surface being coated, there will be lessannihilation of diamond nucleation sites. More availa-

bility of these sites is expected to facilitate growth.However, atomic hydrogen has also been reportedw28xto create new surface defects, which could help diamondnucleation but this does not seem to play significantrole under the conditions of the present experiments.Also, because of more number of collisions, precursorspecies have got less energy and therefore there is lessimplantation into the substrate so that more of these areavailable at the surface thereby helping produce super-saturation and hence better growth.Alternatively, SiC may get partly converted to dia-

mond above a certain critical pressure regime. Thisconversion, most probably occurs at or near its surface.It is speculated that under the conditions used in thepresent experiment, Si gets extracted from SiC, leavingbehind nano- and micro-crystalline diamond–structuredcarbon, following the equation


SiC (solid)q4 H (gas)lSiH (gas)4

qC (diamond, solid)

The reaction is feasible as SiH is more stable(Si–H4

bond energys326 kJymol) than SiC(Si–C bond ener-gys275 kJymol) under the specified conditions. In fact,thermo chemical simulation(using fact sage software)at;1120 K (typical substrate temperature used) showsthat the Gibb’s free energy for the reaction is;428 kJ,which indicates that the reaction is highly favourableunder these conditions. Moreover, since SiH product,4

being gaseous, is pumped out of the system therebyfurther accelerating the forward reaction. A furthersupport for this proposition comes from the XPS anddeep level transient spectroscopy measurements(notbeing given here) on some of the samples, grown ataround the transition pressure zone, which indeed showthe presence of SiH at the surface of the film. Based4

on this model, formation of diamond at higher pressurescan be explained using Le Chatelier’s principle, whichstates that an increase in pressure shifts the equilibriumtowards the lower volume side. Accordingly, the con-version of SiC to diamond like carbon structures is morefavoured at higher pressures, which is consistent withour observations. However, at the moment, we do nothave direct experimental proof of SiC being actuallyconverted to diamond. Nevertheless, the probability ofthis occurrence is very high.Recently, similar observations have also been reported

by Yury et al.w29x. They found that SiC, upon treatmentwith flowing chlorine–hydrogen gas mixture(at Cl y2H ratio equal to or larger than 2:1) in argon carrier gas2

at 10008 C, gets partly converted to diamond. Accordingto the original concept for the meta-stable growth of

diamond suggested by Spitsynw30x, it is necessary toconserve the orientational effect of the surface carbonatoms and to use carbon-containing molecules with sp3

bonding that can be attached to the diamond surface ina complementary manner. Both conditions can be satis-fied when silicon is extracted from SiC, forming carbonatoms in sp hybridization.b-SiC has a structure of the3

diamond lattice where 50% of the carbon atoms arereplaced with silicon. This acts as a template for thegrowth of diamond and diamond-structured carbongrows by direct transformation of the SiC lattice becauseof the sp bonding of carbon in SiC. The role of3

hydrogen is primarily in stabilization of dangling bondsof carbon. This helps to maintain sp hybridization of3

carbon and prevent formation of sp -bonded carbon.2

Work of Stoner et al.w2x and Arnault et al.w12x alsopoint to surface carbide to be a precursor stage todiamond nucleation on silicon. Although some studiesw31,32x do report direct nucleation of diamond onsilicon, we believe the results are not general and mighthave arisen because of specific set of parametersyprocedures used.Another more or less analogous situation can be

thought of, namely the novel low pressure solid statesource methodw33x for diamond formation. Herein,under the conditions of the experiment, the ‘metal-carbon-hydrogen’ eutectic contains much more carbonthan its equilibrium solubility limit. When the tempera-ture of the eutectic is lowered or the metal in it isevaporated, excess carbon in the eutectic gets precipi-tated out as diamond. Similar mechanism has also beenconsidered by Zhu et al.w13x to explain the nucleationand growth of oriented diamond films, formed on nickelsubstrates by multistep hot filament chemical vapourdeposition process.Presently, there exist several propositions regarding

the mechanism of diamond formation but no clearconsensus has yet been established and exact mechanismstill remains elusive. The results of the present studymay be considered to provide yet one more feasibleproposition.

4. Conclusions

Use of an overall low system pressure;1.3 mbarand application of an appropriate bias to the substrateduring the ‘nucleation’ stage has been found to facilitatethe growth of highly oriented cubic silicon carbide filmon mirror polished Si single crystal substrate underconditions conventionally used to grow diamond inHFCVD technique. A critical pressure regime has beenfound to exist during growth stage such that only SiCgrows below this value and diamond starts growingaround or above this value. It is suggested that aroundthis transition region, SiC gets partly converted tonanocrystalline diamond structured carbon, which then


grows into bigger crystals under appropriate growthconditions. Some recent discoveries, from the literaturehave been cited to support the proposed speculation.

Acknowledgments

The work has been done under the ages of Indo-German collaboration programme IB No. INI 03. 1999and DAE No, IG 32. 1999.The authors are thankful toDr S. Schulze and K.K. kutty for scanning electronmicrographs. They also acknowledge the experimentalhelp of K.G. Girija and M. Roy in the course of thisstudy.

References

w1x D.S. Dandy, M.E. Coltrin, in: J. Asmussen, D. Reinhard(Eds.),Diamond Thin Films Handbook, Marcel Dekker, Inc, January2002, Chapter 4.

w2x S.T. Lee, Y.W. Lam, Z. Lin, Y. Chen, Q. Chen, Phys. Rev. B55 (1997) 15 937–15 941.

w3x T. Tachibana, Y. Yokota, K. Hayashi, K. Kobashi, DiamondRelat. Mater. 10(2001) 1633–1636.

w4x B.R. Stoner, G.-H.M. Ma, S.D. Wolter, J.T. Glass, Phys. Rev.B 145 (1992) 11 067.

w5x W.L. Wang, G. Sanchez, M.C. Polo, R.Q. Zhang, J. Esteve,Appl. Phys. A 65(1997) 241.

w6x Q. Chen, J. Yang, Z. Lin, Appl. Phys. Lett. 67(1995) 1853.w7x K.J. Liao, W.L. Wang, B. Feng, Phys. Stat. Sol. A 167(1998)

117.w8x Q. Chen, Z.D. Lin, Appl. Phys. Lett. 68(1996) 2450.w9x F. Stubhan, M. Ferguson, H.J. Fusser, R.J. Behm, Appl. Phys.

Lett. 66 (1995) 1900.w10x Q. Chen, L.X. Wang, Z. Zhang, J. Yang, Z. Lin, Appl. Phys.

Lett. 68 (1996) 176.w11x X.T. Zhou, H.L. Lai, H.Y. Pong, C. Sun, W.L. Zhang, N.

Wang, et al., Diamond Relat. Mater. 9(2000) 134–139.w12x J.C. Arnault, S. Pecoraro, F. Le Normand, J. Werckmann, Appl.

Surf. Sci. 212–213(2003) 912–919.


w13x W. Zhu, P.C. Yang, J.T. Glass, Appl. Phys. Lett. 63(1993)1640.

w14x Y.-Z. Wan, D.W. Zhang, Z.-J. Liu, J.T. Wang, Appl. Phys. A67 (1998) 225–231.

w15x R. Brunsteiner, R. Haubner, B. Lux, Diamond Relat. Mater. 2(1993) 1263–1270.

w16x S.J. Harris, A.M. Weiner, J. Appl. Phys. 75(1994) 5026–5032.w17x S. Schwarz, S.M. Rosiwal, M. Frank, D. Breidt, R.F. Singer,

Diamond Relat. Mater. 11(2002) 589–595.w18x V.C. George, A. Das, M. Roy, A.K. Dua, P. Raj, D.R.T. Zahn,

Thin Solid Films 419(2002) 114–117.w19x V.C. George, A. Das, M. Roy, A.K. Dua, P. Raj, D.R.T. Zahn,

Appl. Surf. Sci. 212–213(2003) 287–290.w20x J.W. Kim, Y.J. Baik, K.Y. Eun, in: Y. Tzeng, M. Yoshikawa,

M. Murakawa, A. Feldman(Eds.), Application of DiamondFilms and Related Materials, Elsevier, Amsterdam, The Neth-erlands, 1991, p. 399.

w21x K.M. McNamara, D.H. Levy, K.K. Gleason, Appl. Phys. Lett.60 (1992) 580–582.

w22x H. Nagasawa, Y. Yamaguchi, T. Izumi, K. Tonosaki, Appl.Surf. Sci. 70–71(1993) 542.

w23x G.K. Wehner, G.S. Anderson, in: L.I. Maissel, R. Gland(Eds.),Handbook of Thin Film Technology, McGraw Hill BookCompany, NewYork, 1970, pp. 3–15.

w24x X. Jiang, C.-P. Klages, R. Zachai, M. Hartweg, H.-J. Fuesser,Diamond Relat. Mater. 2(1993) 407.

w25x L. Dong, B. Ma, G. Dong, Diamond Relat. Mater. 11(2002)1697–1702.

w26x R.E. Shroder, R.J. Nemanich, Phys. Rev. B 41(1990)3738–3745.

w27x Z.C. Feng, A.J. Mascarenhas, W.J. Choyke, J.A. Powell, J.Appl. Phys. 64(1988) 3176.

w28x J.C. Arnault, S. Hebert, F. Le Normand, J. Phys. Chem. B 102(1998) 4856.

w29x Y. Gogotsi, S. Weiz, D.A. Ersoy, M.J. McNallan, Nature 411(2001) 283.

w30x B.V. Spitsyn, in: D.T.I. Hurk (Ed.), Handbook of CrystalGrowth, Elsievier, London, 1994, pp. 401–456.

w31x S.T. Lee, H.Y. Peng, X.T. Zhou, N. Wang, C.S. Lee, I. Bello,et al., Science 287(2000) 104.

w32x X. Jiang, C.L. Jia, Appl. Phys. Lett. 67(1995) 1197.w33x R. Roy, K.A. Cherian, J.P. Cheng, A. Badzian, C. Langlade,

H. Dewan, et al., Innov. Mater. Res. 1(1996) 65–87.

effect of deposition parameters on different stages of diamond deposition … · 2005-02-09 ·...

Documents