Growth of Boron Nitride Thin Films on Silicon Substrates Using New ...

J.-H. Boo et al.: <strong>Growth</strong> <strong>of</strong> <strong>Boron</strong> <strong>Nitride</strong> <strong>Thin</strong> <strong>Films</strong> on Si 705phys. stat. sol. (a) 176, 705 (1999)Subject classification: 68.55.Jk; 78.20.Ci; S7.14<strong>Growth</strong> <strong>of</strong> <strong>Boron</strong> <strong>Nitride</strong> <strong>Thin</strong> <strong>Films</strong>on Silicon Substrates Using New Organoboron PrecursorsJ.-H. Boo 1 † (a), C. Rohr (b), and W. Ho (b)(a) Institute <strong>of</strong> Basic Science and Department <strong>of</strong> Chemistry, Sungkyunkwan University,Suwon 440-746, Korea(b) Laboratory <strong>of</strong> Atomic and Solid State Physics and Department <strong>of</strong> Physics,Cornell University, Ithaca, NY 14853, USA(Received July 4, 1999)<strong>Thin</strong> films <strong>of</strong> BN have been deposited on silicon substrates in the temperature range <strong>of</strong> 600 to900 C by supersonic molecular jet epitaxy and plasma-assisted MOCVD using new organoboronprecursors such as borane-triethylamine complex (BTEA) and tris(sec-butyl)borane (TSBB). Hydrogenwas used as carrier gas, and additional nitrogen and ammonia as reactive gases were suppliedby either through nozzle or via remoted plasma. Crack-free, polycrystalline hexagonal BNfilms with nano-size crystals were successfully grown at temperature as low as 750 C using BTEA.With increasing substrate temperatures to 900 C, however, the h-BN films chenged polycrystallinewith smoother surface and larger crystals. The changes <strong>of</strong> relative film quality with deposition temperatureand reactive gas have been confirmed with FTIR. Nitrogen plasma enhanced the filmquality and the stoichiometry, and removed the surface carbon effectively rather than ammoniaand hydrogen plasmas molecular beam. Wide band gaps were calculated using optical transmittancemeasurements.1. Introduction<strong>Boron</strong> nitride (BN) has received considerable attention due to its exceptional propertiessuch as extreme hardness, high thermal conductivity, inertness, transparency, anddielectricity [1, 2]. These properties make BN an interesting material for many applicationssuch as protective and optical coatings, mask substrate for X-ray lithography, andinsulating layers in electronic devices [2, 3]. Besides <strong>of</strong> these applications, moreover,thin films <strong>of</strong> BN are useful for the growth <strong>of</strong> GaN and diamond films on Si due to theexcellent lattice mismatch between GaN (or diamond) and BN [4]. Hexagonal BN(h-BN) has good tribological properties that are not degraded in air, otherwise cubicBN (c-BN) has almost the same hardness and thermal conductivity as diamond. Underthe conditions <strong>of</strong> high pressure and temperature, h-BN transforms into a wurtzite w-BNor a cubic c-BN. Up to now, however, it was very difficult to make high quality BN thinfilms on Si due to the large lattice mismatch (about 33%). Because <strong>of</strong> good step coverage,CVD techniques are interesting for the synthesis <strong>of</strong> BN films, since they allows aprecise control <strong>of</strong> the deposition characteristics even on substrates with complex shape[5]. Furthermore, low-pressure synthesis permitting the formation <strong>of</strong> pure and well-crys-1 ) Corresponding author; Tel.: +82-331-290-7072; Fax: +82-331-290-7075;e-mail: jhboo@chem.skku.ac.kr

706 J.-H. Boo et al.tallized c-BN by a CVD method without ion bombardment was not successful so far [6].Thus, a new concept for a low-pressure synthesis must still be found for achieving highquality BN films. Plasma processing is becoming an increasingly important area in materialsand electronics research, and a deposition method with a remote plasma systemto assist the growth seems to be a promising experiment.The CVD preparation <strong>of</strong> BN films is usually carried out using toxic and explosivematerials such as diborane (B 2 H 6 ) or highly corrosive borontrihalides such as BCl 3 . Inrecent years, there has therefore been great interest in the deposition <strong>of</strong> BN with lessor non-toxic organometallic precursors, especially single molecular precursors that containboth boron and nitrogen atoms in the same molecule. Only few papers have beenpublished using boron-organic substances [7 to 10] and single molecular precursors [11to 14], most <strong>of</strong> them using borazine that has the disadvantage <strong>of</strong> instability at roomtemperature [13].To supply a buffer layers for the growth <strong>of</strong> GaN and AlN thin films, therefore, we firstlytried to deposit thin films <strong>of</strong> h-BN on Si substrates in the temperature range <strong>of</strong> 600 to900 C by supersonic molecular jet epitaxy and plasma assisted MOCVD using new organoboronprecursors such as BTEA and TSBB. These precursors have the advantage overmost others <strong>of</strong> being non-toxic and non-explosive and having suitable vapor pressures.2. ExperimentalThe experiments were carried out in a CVD chamber with a remote microwave plasmasystem. BTEA, …CH 3 CH 2 † 2BH 2 : BH 3 , and TSBB, ‰CH 3 CH 2 CH…CH 3 †Š 3B, were used asprecursors, and freeze-pump-thaw cycles were employed for the purification <strong>of</strong> the precursors.H 2 was used as carrier gas, and additional NH 3 and N 2 as reactive gases weresupplied by either through nozzle (orifice size: 150 mm) or via microwave (2.5 GHz)remoted plasma. The substrates were indirectly heated up to 900 C using a BN heater.The substrate temperatures were monitored by an IR optical pyrometer. The base pressure<strong>of</strong> CVD system was 2 10 7 Torr while the CVD pressure was (3 to 4) 10 3 Torr.The typical growth time was 6 h and the average growth rate was approximately200 nm/h. The as-grown films were characterized with AES, XRD, TEM, AFM, andFTIR. Wide band gaps were calculated using optical transmittance measurements. Thedetails <strong>of</strong> experimental set-up and deposition procedure are given in elsewhere [14].3. Results and DiscussionHighly oriented, crack-free, polycrystalline h-BN films were successfully obtained onSi(100) substrates at 900 C using both BTEA and TSBB. The Auger electron spectra<strong>of</strong> the BN films grown on Si(100) with BTEA (Fig. 1a) and TSBB (Fig. 1b) at 900 Cshow two main peaks <strong>of</strong> B(KLL) and N(KLL) appearing at 180 and 380 eV, signifyingthe formation <strong>of</strong> BN film on Si(100) surface. In addition, two satellite peaks (e.g., Augerfine structure) become visible at 25 and 9 eV higher kinetic energy from the180 and 380 eV peaks. These satellite peaks are attributed to a bulk plasmon loss peakand a p plasmon loss peak <strong>of</strong> the sp 2 -bonded h-BN, respectively [15]. Fig. 1a reveals aB :N ratio <strong>of</strong> about 1 :1 near the surface. If the precursor (BTEA) and H 2 carrier gaswere used without the NH 3 supersonic jet or N 2 plasma, the obtained films have anexcess <strong>of</strong> carbon and a deficiency <strong>of</strong> nitrogen in the film. As shown in Fig. 1b, there

<strong>Growth</strong> <strong>of</strong> <strong>Boron</strong> <strong>Nitride</strong> <strong>Thin</strong> <strong>Films</strong> on Si 707Fig. 1. Auger electron spectra <strong>of</strong> h-BN filmsgrown on Si(100) substrates at 900 C witha) borane-triethylamine complex and b) tris(-sec-butyl)boranealso appears a large C(KLL) Augerpeak due to surface contamination fromthe precursor (TSBB) during deposition.The B :N ratio <strong>of</strong> this film is 1 :0.7,suggesting that the ammonia and nitrogencan easily remove the surface nitrogenrather than the surface carbon.Because FTIR spectra provide directinformation about the local bonding environment<strong>of</strong> the constituents on thefilm, the changes <strong>of</strong> relative film qualitywith deposition temperature and reactivegas can easily be confirmed withFTIR. For h-BN films, there are twocharacteristic peaks at about 1400 and800 cm 1 , which are attributed to thein-plane B±N stretching and out-<strong>of</strong>planeB±N±B bending vibrations [16].For c-BN films, however, only one peakwas observed at about 1100 cm 1 . Fig. 2shows the changes <strong>of</strong> FTIR spectrawhich are obtained from the BN filmsgrown on Si(100) with BTEA at variousdeposition temperatures (Fig. 2a) andunder different reactive gas conditions (Fig. 2b). With increasing deposition temperaturesup to 900 C, the FWHMs <strong>of</strong> Fig. 2a become smaller, indicating that high qualityh-BN film can be deposited at higher temperature. With lower deposition temperatures,the FTIR peak at about 1400 cm 1 broadens and slightly shifts towards lower wavenumbers. On the sides <strong>of</strong> the peak shoulders become obvious and one can see `sidepeaks' at 1200 to 1250 and 1500 cm 1 associated with hydrogenated amorphous BNfilms. This indicates that hydrogenated amorphous BN films were obtained with BTEAbelow 750 C. Fig. 2b shows that the N 2 plasma enhanced the film quality and removedthe surface carbon effectively rather than NH 3 supersonic molecular beam or H 2 plasma.This can be explained by the fact that N 2 with additional energy gained from theplasma is a better reactant with boron from the precursor resulting in better films comparedto NH 3 . Because the NH 3 molecule has one nitrogen atom bonded to threehydrogen atoms it is more stable than the excited N 2 which can react more easily.Thus, it can be expected that the best BN films with high quality and good stoichiometrycould be deposited at 900 C with BTEA and N 2 plasma rather than withTSBB and NH 3 or H 2 .

708 J.-H. Boo et al.Fig. 2. FTIR spectra <strong>of</strong> h-BN filmsgrown on Si(100) substrates with a) borane-triethylaminecomplex at various depositiontemperatures, and b) under differentreactive gas conditionsDue to the small X-ray cross-sectionarea <strong>of</strong> boron which is one <strong>of</strong>the light elements, it is very difficultto obtain a good XRD pattern <strong>of</strong>BN thin film. With a Read cameraand the exposure <strong>of</strong> 8 h, however, aclear picture <strong>of</strong> Debye-Scherrer diffraction(see Fig. 3a) could be obtainedfrom the BN film grown withBTEA at 900 C. The existence <strong>of</strong>diffraction rings indicates the presence<strong>of</strong> crystalline material. With atransparency scale, one can identifydirectly the lattice spacings associatedwith the diffraction rings andthe (hkl) planes <strong>of</strong> h-BN. For a singlecrystal one would obtain a diffractionpattern consisting <strong>of</strong> points,but as there are complete rings evenwithout any variation in intensity, itsignifies that the material is randomlyoriented polycrystalline. Inthis study, we found seven different crystal planes, indicating a polycrystalline nature <strong>of</strong>the BN film. The fuzzyness <strong>of</strong> the outer rings is due to the small crystalline size. Inorder to determine the crystallinity <strong>of</strong> h-BN films in more detail, a plan-view TEMimage was obtained using the same film as in Fig. 3a. In Fig. 3b, one can see that thefilm consists mainly <strong>of</strong> crystals <strong>of</strong> 100 to 300 nm size. The black `balls' are crystallites <strong>of</strong>a size <strong>of</strong> roughly 200 nm. With AFM, the morphology could be visualized even better.Fig. 3c shows the film surface seen from an angle to create a three-dimensional image.In Fig. 3c, the surface <strong>of</strong> BN film deposited on Si(100) at 900 C with BTEA showsnano-size crystallites with a root mean square (rms) roughness <strong>of</strong> 7.0 nm and peak-topeakroughness <strong>of</strong> about 100 nm. Only a few grains are as big as 1 to 2 mm. Ichiki etal. [17] reported that on the nanometer scale, the surface <strong>of</strong> h-BN film is rougherthan that <strong>of</strong> c-BN film. Therefore, our plan-view AFM image is consistent with thehexagonal nature <strong>of</strong> the BN film.Optical transmission data were also taken from the h-BN membranes fabricated byetching away a 0.25 inch diameter Si substrates. Fig. 4a shows that the BN film grown at900 C with BTEA has a high transmittance <strong>of</strong> about 80% in the infrared, throughoutthe visible and into the UV, where it drops sharply <strong>of</strong>f to zero indicating the band-edge

<strong>Growth</strong> <strong>of</strong> <strong>Boron</strong> <strong>Nitride</strong> <strong>Thin</strong> <strong>Films</strong> on Si 709Fig. 3. Debye-Scherrer diffraction pattern, TEM and AFM images <strong>of</strong> h-BN films grown on Si(100)at 900 C with borane-triethylamine complex a) Debye-Scherrer diffraction pattern, b) TEM, andc) AFM<strong>of</strong> the material. With the straight-line interception at about 215 nm wavelength, an energyband gap <strong>of</strong> 5.8 eV was deduced. This is in exact agreement with the reportedvalue for the band gap <strong>of</strong> h-BN [2, 18, 19]. On the other hand, Fig. 4b for a h-BN filmgrown at 900 C with TSBB shows rather smaller transmission compared with that <strong>of</strong>Fig. 4a, signifying poor crystallinity. This can reflect a difference <strong>of</strong> carbon contents inthe film layer, that is, the carbon contamination lowers the transmittance considerablyFig. 4. Optical transmittance data <strong>of</strong> h-BN films grown on Si(100) substrates at 900 C with a) borane-triethylaminecomplex, and b) tris(sec-butyl)borane24 physica (a) 176/1

710 J.-H. Boo et al.: <strong>Growth</strong> <strong>of</strong> <strong>Boron</strong> <strong>Nitride</strong> <strong>Thin</strong> <strong>Films</strong> on Siand cuts it <strong>of</strong>f at around 250 nm. Thus, one can estimate a band gap <strong>of</strong> 5.0 eV fromFig. 4b. For h-BN, however, the results from literature differ and it is not clear if theband gap is direct [18] or indirect [19]. In this study, an indirect band gap (5.8 eV) forh-BN in-plane lattice spacing is taken for comparison.4. ConclusionsHighly oriented polycrystalline, crack-free h-BN films in [002] and [101] directions weresuccessfully deposited on Si(100) substrates at temperature as low as 750 C usingBTEA. Below 750 C, on the other hand, hydrogenated amorphous BN films were alsoobtained with both BTEA and TSBB. The changes <strong>of</strong> relative film quality with depositiontemperature and reactive gas have been confirmed by FTIR spectroscopy. Nitrogen plasmaenhanced the film quality and the stoichiometry and removed the surface carboneffectively rather than ammonia molecular beam and hydrogen plasma. The best BNfilms with high quality and good stoichiometry could be deposited at 900 C with BTEAand nitrogen plasma rather than with TSBB and ammonia or hydrogen. Wide band gap(5.8 eV) was calculated from optical transmittance data, and the carbon contamination inthe film layer can lower the transmittance considerably. To our best knowledge, this is thefirst report <strong>of</strong> growing BN films from the precursors <strong>of</strong> BTEA and TSBB.Acknowledgements One <strong>of</strong> authors (J.-H. B.) thanks the Korean Research Foundation(Grant No. D00036) and the Sungkyunkwan University for their financial supports.Support <strong>of</strong> this research by the BMDO via ONR Grant No. N00014-93-0499 and SBIRvia ONR contact No. N00014-95-C-0414 is gratefully acknowledged.References[1] J. H. Edgar, in: Properties <strong>of</strong> Group III <strong>Nitride</strong>s; EMIS Datareviews Series, No. 11, An Inspec.Publication, London, 1994.R. C. DeVries, in: Cubic <strong>Boron</strong> <strong>Nitride</strong>: Handbook <strong>of</strong> Properties, General Electric Rep. No.72CRD178, 1972.[2] S. P. S. Arya and A. D'Amico, <strong>Thin</strong> Solid <strong>Films</strong> 157, 267 (1988).[3] R. F. Davis, Proc. IEEE 79, 702 (1991).[4] J.-H. Boo, C. Rohr, and W. Ho, J. Cryst. <strong>Growth</strong> 189/190, 439 (1998).[5] C. Gomez-Aleixandre, A. Essafti, M. Fernandez, J. L. G. Fierro, and J. M. Albella, J. Phys.Chem. 100, 2148 (1996).[6] A. Bartl, S. Bohr, R. Haubner, and B. Lux, Internat. J. Refractory and Hard Materials 14,145 (1996).[7] A. C. Adams, J. Electrochem. Soc. 128, 1378 (1981).[8] W. Schmolla and H. L. Hartnagel, Solid State Electronics 26, 931 (1983).[9] J. Kouvetakis, V. V. Patel, C. V. Miller, and D. V. Beach, J. Vac. Sci. Technol. A 8, 3929 (1990).[10] G. S. Yuryev, E. A. Maximovskiy, Y. M. Rumyantsev, N. I. Fainer, and M. L. Kosinova, J.Physique IV C5, 695 (1995).[11] A. R. Phani, S. Roy, and V. J. Rao, <strong>Thin</strong> Solid <strong>Films</strong> 258, 21 (1995).[12] T. P. Smirnova, L. V. Jakovkina, I. L. Jashkin, N. P. Sysoeva, and J. I. Amosov, <strong>Thin</strong> Solid<strong>Films</strong> 237, 32 (1994).[13] A. Weber, U. Bringmann, R. Nikulski, and C.-P. Liages, Diamond Relat. Mater. 2, 201 (1993).[14] C. Rohr, J.-H. Boo, and W. Ho, <strong>Thin</strong> Solid <strong>Films</strong> 322, 9 (1998).[15] T. Mega, R. Morimoto, M. Morita, and J. Shimmomura, Surf. Interface Anal. 24, 375 (1996).[16] R. Geick, C. H. Perry, and B. Rupprecht, Phys. Rev. 146, 543 (1966).[17] T. Ichiki, S. Amagi, and T. Yoshida, J. Appl. Phys. 79, 4381 (1996).[18] J. J. Jia, T. A. Callcott, E. L. Shirley, J. A. Carlisle, L. J. Terminello, A. Asfaw, D. L.Ederer, F. J. Himpsel, and R. C. C. Perera, Phys. Rev. Lett. 76, 4054 (1996).[19] M. I. Erements, M. Gauthier, A. Polian, J. C. Chervin, and J. M. Besson, Phys. Rev. B 52,8854 (1995).