J. MATER. CHEM., 1994,4(10), 1591-1594 Investigations into the Growth of AIN by MOCVD using Tri-ferf-butylaluminium as an Alternative Aluminium Source Anthony C. Jones,*" John Auld," Simon A. Rushworth," David J. Houlton" and Gary W. Critchlod a Epichem Limited, Power Road, Bromborough, Wirral, Merseyside, UK L62 3QF lnstitute of Surface Science and Technology, University of Loughborough, Loughborough, Leicestershire, UK LE77 3TU Thin films of AIN have been deposited at 500 and 600°C by atmospheric-pressure MOCVD using the precurscirs tri- tert-butylalurniniurn (Bu',Al) and tert-butylarnine (Bu'NH,). Growth rates of 0.5 pm h-' were obtained at 500 "C. Post-growth oxidation of the AIN films was prevented by the deposition of a protective A1 overlayer using Bu',Al.Aluminium nitride (AlN) is an important material with a variety of applications such as passive barrier layers and substrates in silicon integrated circuits, high-frequency acous- tic wave devices, high-temperature windows and dielectric optical enhancement layers in magneto-optic multilayer struc- tures.' In addition, the ternary alloy Al,Ga,-,N has a large potential application in optoelectronic devices operating in the UV- blue spectral region., The development of these various applications is critically dependent on the capability to deposit thin films of AlN at low to moderate substrate temperatures. Conventional ceramic processes, such as the direct nitriding of A1 powder at high temperature (> 1440°C) are unsuitable for the controlled deposition of thin A1N 1aye1-s.~ Therefore, the physical vapour deposition technique of vacuum sputter- ing is generally employed.However, this suffers from the disadvantages of limited scale and poor conformal step cover- age. There has thus been a concerted effort4 to develop metal- organic chemical vapour deposition (MOCVD) techniques which have the advantages of large-area growth capability, excellent conformal step coverage and precise control of layer thickness. The deposition of AlN by MOCVD has traditionally been carried out using mixtures of trimethylaluminium (Me3AI) and ammonia (NH3).5,6 However, the high thermal stability of NH3 necessitates the use of high substrate temperatures (typically >9OO"C).This leads to the problem of nitrogen loss from the AIN film which is only partially alleviated by the use of high V :I11 ratios (e.g.>2000 :1). AIN growth has been achieved at lower substrate tempera- tures (400-800 "C) using a variety of 'single-source' precursors, which already contain an intramolecular (Al-N) bond. These include [A1(NR2)3]2, [HA1(NR2),], (R =Me, Et),7 [Me,AlNH,],,* [Et2A1N,]39 and [Me,AlNR,], (R =Pr').'' However, these precursors have only very low vapour press- ures (< 1Torr at room temperature) which necessitates the heating of source and reactor inlet lines and the use of high- vacuum MOCVD reactors. It is therefore desirable to develop alternative precursors which may be more conveniently utilized in MOCVD, and the volatile nitrogen source hydrazine (N2H4) has been used in combination with Me3AI to grow A1N at temperatures as low as 220°C.'' However, N2H4 is an extremely toxic (TLV(,ki,,0.01 ppm) and unstable compound which has been reported to decompose on contact with stainless steel.These factors are likely to seriously restrict its large-scale application in MOCVD. The successful deposition of A1N from the single-source precursors [Me2A1NR2]," and [Et2A1N3]39 has encouraged us to investigate methods of forming similar species in situ in the vapour phase prior to layer growth. This approach aims to combine the advantages of convenient source temper'atures and high growth rates associated with the use of high l'apour pressure reagents, with the low growth temperatures ,tssoci- ated with single-source precursor molecules.Thus, we have recently demonstrated" the successful deposition of ,!IN in the temperature range 400-600 "C by atmospheric pressure MOCVD using the volatile primary alkylamines, tert-butylamine ( Bu'NH,) and isopropylamine (Pr'NH,) in combi- nation with Me3A1. It was proposed'2 that the directly honded species [Me,AlNHR], was formed in the gas phase prior to AlN deposition, and the recent report13 of AlN growth by high-vacuum CVD using [Me,AlNHR], (R =Bur, Pr') strongly supports this proposal. Similarly, the combination of Me3A1 and trimethylsilylazide (Me3SiN3) proved suitable for the deposition of A1N at 300-450 OC.I4 Significantly. Auger electron spectroscopy (AES) failed to detect silicon in the films, and this was attrib~ted'~ to the formation of diinethyl-aluminium azide (Me2AIN3) in the gas phase, together with tetramethylsilane (Me,%) which allows the efficient tr,tnsport of Si species away from the growth zone.However, the A1N films deposited from mixtures of Me3A1-RNH, or Me,Al-Me,SiN, were found to contain oxygen (2.0-8.0 atom%), due possibly to post-grovl th oxi- dation, together with variable levels of residual carbon (2.7-17.0 atom%). The carbon contamination was attri-buted12,14 to the use of the methyl-based A1 precursor Me3A1, which has been shown to lead to significant levels ot carbon contamination in A1 films15 and AlGaAs epitaxial layers16 grown by MOCVD. Recently, some of the present authors have shown that the new A1 precursor tri-tert-butylaluminium (Bu',Al) can be used to deposit high-purity A1 in the temperature range 300-450 "C by low-pressure CVD.I7 This has encouraged us to investigate But3Al as an alternative precursor to Me3Al for the deposition of AlN by MOCVD, and these results are presented herein.In an effort to prevent post-growth oxidation of the AlN films, But3Al was also used to deposit a protective A1 overlayer, which provides evidence of its usefulness and versatility as a new A1 source for MOCVD. Experimental General Techniques AES was carried out on a Varian scanning Auger spectrometer. The atomic compositions quoted ?re from the bulk of the film (depth from surface >2000 A) and were obtained by combining AES with sequential ion bombardment until com- parable compositions were obtained for consecutive data points.Film thicknesses were estimated by the time taken to sputter through the layer using Ar +-ion bombardment. Proton nuclear magnetic resonance ('H NMR) data were obtained on a Bruker WM 250 spectrometer operating at 250 MHz and microanalytical data (C, H, N analysis) were provided by the Microanalytical Services Department of the University of Liverpool. Scanning electron microscopy (SEM) was performed on a Cambridge Stereoscan 360 microscope. Aluminium Nitride Film Growth The reagents used were Bu',Al, synthesized as described previ~usly'~and Bu'NH,. The Bu'NH, was dried and deoxy- genated prior to use by distillation over sodium under a nitrogen purge.The AlN films were deposited at atmospheric pressure in a simple cold-wall horizontal quartz reactor (Electro Gas Systems Ltd) using radiant substrate heating. The substrates used were Si( 11 1) single-crystal wafers and these were cleaned (20% nitric acid-deionized water), degreased with acetone and dried before use. Trace oxygen and moisture were removed from the hydro- gen carrier gas by passing it through a Nanochem resin purification unit. The Bu'NH, was further purified during use by passage through a Nanochem purifier. The Bu',Al and ButNH2 sources were operated at room temperature (22°C) and were mixed in a 'T-piece' at the reactor inlet. This was heated to 60 "C to prevent condensation of any adducts formed in the gas phase.In order to prevent post-growth oxidation of the deposited AlN films, a protective A1 overlayer was subsequently deposited at low pressure ( 15 Torr) using the But,Al precursor alone. A full summary of growth conditions is given in Table 1. Results and Discussion A1N films were successfully deposited using Bu',Al and Bu'NH, at substrate temperatures between 500 and 600 "C. Below 500"C, the A1N growth rate was found to be prohibi- tively low, whilst at temperatures >600 "C film growth was limited by severe reagent depletion. The atomic composition of the films was determined by AES and these data are summarized in Table 2. These data show that all the films have an A1 :N ratio close to unity, although in film 3 nitrogen is present in slight excess.The most obvious feature of the AES data is the significant reduction of oxygen contamination resulting from the growth Table 1 Growth conditions used to deposit AlN J. MATER. CHEM., 1994, VOL. 4 Table2 Auger electron spectral analysis of AN films grown on Si( 11 1 ) using mixtures of Bu',Al and BuWH, atomic composition %) film no. A1 N C 0 1 (uncapped) 41.4 39.3 6.9 10.4 2 (A1 capping layer) 98.2 - 0.5 1.2 (A1N layer) 3 (AlN layer) 49.7 45.7 44.3 46.6 4.7 7.2 1.3 0.4 Al:N 1.05 -~ 1.21 0.98 of the protective A1 overlayer in films 2 and 3. This strongly suggests that post-growth oxidation has cxcurred in the uncapped A1N film (l),and further suggests that post-growth oxidation was largely responsible for the relatively high levels of oxygen contamination (2.0-8.0 atom%) observed previ~usly'~~~~in A1N films grown using mixtures of Me,Al-RNH, and Me,Al-Me,SiN, .The residual oxygen (between 0.4 and 1.3 atom%) remaining in the capped A1N films and in the A1 overlayer can be attributed to trace oxygen in the relatively unsophisticated MOCVD reactor used in this study. The uncapped A1N films were extremely hard and scratch- resistant and demonstrated specular surface morphology. Scanning electron microscopy (SEM) data for a typical uncapped A1N film grown at 500°C on Si(111) (Fig. 1) Fig. 1 Scanning electron micrograph of an AlN film grown at 500 -C on Si( 111)from a But3Al-Bu'NH, mixture films from mixtures of Bu',Al and Bu'NH," run no.1 2 3 (uncapped) (A1 capped) (a) AIN Growth (cell pressure 760 Torr) H, carrier gas flow through Bu',Al (sccm)b H, carrier gas flow through Bu'NH, (sccm) substrate temperature/T growth ratejpm h-' 200 50 500 0.5 200 50 500 -200 50 600 - approximate V :III ratiod 36 36 36 (h) A1 capping layer (cell pressure 15 Torr) H, carrier gas flow through But3Al (sccm) - 50 50 substrate temperature/^C duration of growth/min -- 400 3 400 1 ~~~ a But3Al and Bu'NH, sources at 22 "C; substrates Si( 111) single-crystal wafers. Standard cm3 min-'. ' Estimated from AES sputter time. Based on an estimated Bu',Al vapour pressure of ca. 2 Torr at 22 "C (vapour pressure Bu'NH, =340 Torr at 25 "C).J. MATER. CHEM., 1994, VOL. 4 indicate that the film is amorphous and structureless, with no grains ekident on a 500 nm scale. A further significant feature to emerge from the AES data (Table 2) is that, despite the use of But3A1 as an alternative to Me,Al, the A1N films still contain residual carbon at a level of between 5 and 7 atom%. These carbon levels are similar to those observed in A1N films grown using Me,Al-Bu'NH, mixtures (Cz3-9 atom%),12 which indicates that But,A1 offers no significant advantage over Me,Al for AlN growth from R3A1-RNH, mixtures. This is a surprising result in view of the marked contrast in the purity of A1 films deposited at 450 "C from But3Al (CFZ 0.2-0.5 atom%)17 compared with A1 films deposited at similar substrate tem- peratures from the methyl-based precursors Me,Al'' or Me,A1H(NMe,)l8 in which carbon levels of up to 39 atom% have been observed.This suggests that the decomposition of the [Al-R] group may not be the only factor controlling carbon incorporation in A1N films grown from R3A1-RNH2 mixtures. The decomposition of the organic radical of the primary alkylamine (RNH,) may also play a role which suggests that carbon incorporation may vary according to the nature and pyrolysis characteristics of the RNH, precur- sor. This proposal is supported by the greatly increased carbon levels (14-17 atom%) observed in A1N films grown using Me,Al-Pr'NH, compared with films grown from Me,Al-Bu'NH, mixtures.', In addition, the carbon contami- nation was shown to increase with increasing V: 111 ratios, in marked contrast to the trend generally observed in the growth of GaAs and AlGaAs by MOVPE;16 this provides further evidence that the RNH, precursor may play a critical role in carbon contamination.Information concerning the possible growth mechanism has been obtained by the ex-situ addition of ButNH, (12.9 g, 0.17 mol) to Bu,'Al (16.0 g, 0.08 mol) in dry pentane solution (25 cm3). Removal of volatiles in ~acuu left a colourless crystalline product which was highly soluble in benzene. This was shown to be the 1: 1 adduct, [But,Al(NH,But)] by 'H NMR data and elemental microanalysis (Table 3). The [Bu',Al(NH,Bu')] adduct was observed to melt at 70-80 C, and at 115 "C a gas was evolved.Further heating of the compound at 115-120°C for 30min led to a white powder which was only sparingly soluble in benzene. This precluded meaningful 'H NMR data and elemental microanal- ysis (Table 3) was also inclusive, although these data suggest that the decomposition product may have the molecular formula [Bu',AI(NHBu')], (n=2, 3). The low solubility of this compound in benzene is consistent with the proposed oligomeric structure. Table 3 Analytical data for 1: 1 adduct formed from the reaction between But,Al and ButNH, 'H NMR data (['H6] benzene) 6 0.85 (s, 9 H, N-But) 1.25 (s, 27 H, But-Al) 2.5 (s, 2 H, N-H) elemental microanalysis C (Yo) H (Yo) N (Yo) found 70.74 14.49 4.83 calcd. for [Bu',Al( NH,Bu')] 70.77 14.14 5.16 elemental microanalysis of decomposition product" C (Yo) H (Yo) N (Oh) found 66.67 13.65 5.74 calcd.for [Bu',Al( NHBu')], 67.54 13.25 6.56 ~~~~ ~ Formed by heating the 1: 1 adduct at 115-120 "C for 30 min. Decomposition product essentially insoluble in c2H6] benzene. During the growth of AlN from But,Al-ButNH, mixtures, a crystalline deposit was observed to form at the reactor inlet if this was left unheated, and it is likely that this is the which[Bu',Al(NH,Bu')] adduct. Previous studie~,'~~~~ are supported by the present work, have shown that such adducts readily form elimination products of the type [R,AlNHR'], on heating, and therefore such species may be expected to form in the hot boundary layer adjacent to the substrate.Subsequent pyrolysis of the directly bonded species [Bu',AINHBut], on or near the substrate surface leads to the deposition of A1N. This proposal is strongly supported by the recently reported growth of AlN by vacuum CVI) using [Me,AlNHR], (R=Pr', The level of carbon contamination in A1N films drrposited from [R,AlNHR], precursors, which have been either pre- synthesized or formed in situ in the gas phase, will depend strongly on the mechanism by which the alkyl radicals bound to A1 or N are desorbed from the growth surface. For Pr' and But radicals there is a ready desorption route viu the p-hydride elimination of alkene;,l however, the continued pres- ence of carbon in AlN grown from Bu',Al-Bu'NH, mixtures indicates that some surface decomposition of the Bu' radical has occurred.The decomposition of the But radical is likely to be promoted by the presence of A1 on or near thcr growth surface.,l This may lead to methyl abstraction from the But radical, leading to surface-adsorbed methyl radicals which subsequently decompose to deposit carbon. The mechanism of AlN deposition from [R,AINHR'], species may be similar to that occurring in the growth of GaAs from the single-source molecule [M~,G~AsBu',],.~~In these studies it was proposed that the facile P-hydride elimin- ation of alkene from the bulky and sterically hindered tert-butyl group leads to the formation of a strong intraniolecular 111-V bond during pyrolysis which facilitates the growth of stoic hiometric GaAs.The precise mechanism of A1N deposition from R3A1-RNH, mixtures has not been established. However, the low growth temperatures (400-600 "C) and low V: IT1 ratios used in this, and previous12 studies, compared with those med for Me,Al-NH, combinations, suggests that 'directly bonded' species of the type [R,AlNHR], are the active precursors to A1N deposition. Conclusions A1N films have been deposited by atmospheric -pressure MOCVD using But3Al in combination with ButNHz. Growth rates of 0.5 pm h-' were obtained at substrate temperatures of 500 "C. The use of a protective A1 overlayer, depoS(ited from But3Al, was shown to lead to a significant reduction in oxygen contamination of the AlN films.However, residuiil carbon was present at levels of between 5 and 7 atom%, and it is suggested that the decomposition of the organic radical in ButNH, may play a role in carbon incorporation. This work was supported by the Department of I'rade and Industry under the LINK/ATP initiative and the Teaching Company scheme. D.H. is a Teaching Company Associate (University of Keele, UK). References 1 L. M. Sheppard, Cerum. Bull., 1990,69, 1801. 2 M. A. Khan, R. A. Skogman, R. G. Schulze and M. Gershenzon, Appl. Phys. Lett., 1983, 43, 492. 3 L. V. Interrante, L. E. Carpenter, C. Whitmarsh and W. Lee, Mater. Res. Soc. Symp. Proc., 1986,13, 359. 4 L. Baixia, L. Yinkui and L. Yi, J. Muter. Chem.. 1993, 3, 117.1594 5 M. Morita, S. Isogai, N. Shimizu, K. Tsubouchi and N. Mikoshiba, Jpn. J. Appl. Phys., 1981,19, L173. 6 M. Morita, M. Useugi, S. Isogai, K. Tsubouchi and N. Mikoshiba, Jpn. J. Appl. Phys., 1981,20, 17. 7 Y. Takahashi, K. Yamashita, S. Motojima and K. Singiyama, Surf. Sci., 1979,86, 238. 8 L. V. Interrante, W. Lee, M. McConnell, N. Lewis and E. Hall, J. Electrochem. Soc., 1989,136,472. 9 K. L. Ho, K. F. Jensen, J. W. Hwang, W. L. Gladfelter and J. F. Evans, J. Crystal Growth, 1991, 107, 376. 10 D. C. Bradley, D. M. Frigo and E. A. D. White, Eur. Put. Appl., 1989 EP 0 331 448. 11 M. Mizuta, S. Fujieda, T. Jitsukawa and Y. Matsumoto, Proc. Int. Symp. on GaAs und Related Compounds, Las Vegas, Nevada, 1986, IOP Publishing Ltd, Bristol, 1987.12 A. C. Jones, J. Auld, S. A. Rushworth, E. W. Williams, P. W. Haycock, C. C. Tang and G. W. Critchlow, Adzi. Muter., 1994,6,229. 13 M. M. Sung, H. D. Jung, J. K. Lee, S. Kim, J. T. Park and Y. Kim, Bull. Koreun Chem Soc., 1994, 15, 79. 14 J. Auld, D. J. Houlton, A. C. Jones, S. A. Rushworth and G. W. Critchlow, J. Muter. Chem., 1994,4, 1245. J. MATER. CHEM., 1994, VOL. 4 15 D. R. Biswas, C. Ghosh and R. I,. Layman, J. Electrochem. Soc., 1983,130,234. 16 T. F. Kuech, E. Veuhoff, T. S. Kuan, V. Deline and R. Potemski, J. Crystal Growth, 1986,77,257. 17 A. C. Jones, J. Auld, S. A. Rushworth and (3. W. Critchlow, J. Crystal Growth, 1994, 135,285. 18 A. C. Jones, S. A. Rushworth and J. Auld, paper to be presented at ICVGE-8, Freiburg, Germany, July 1994, proceedings to be published in J. Crystal Growth. 19 S. Amirkhalili, P. B. Hitchcock, A. D. Jenkins. J. Z. Nyathi and D. J. Smith, J. Chem. Soc., Dulton Trans., 1981, 377. 20 AA Al-Wassil, P. B. Hitchcock, S. Sarisaban, J. D. Smith and C. L. Wilson, J. Chem. Soc., Dalton Trans., 1985. 1929. 21 J. S. Foord, N. K. Singh, E. T. Fitzgerald, G. J. Davies and A. C. Jones, J. Crystal Growth, 1992,120, 103. 22 J. E. Miller, K. B. Kidd, A. H. Cowley, R. A. Jones, J. G. Ekerdt, H. J. Gysling, A. A. Wernberg and T. N. Blanton, Chem. Muter., 1990,2, 589. Paper 4/03214D; Received 31st Muy, 1994