Corrosion resistant coatings (A1203) produced by metal organic
chemical vapour deposition using aluminium-tri-sec-butoxide V. A. C. Haanappel, H. D. van Corbach, T. Fransen*, and P. J. Gellings University of Twente, Department of Chemical Technology, P.O. Box 217, 7500 AE Enschede (Netherlands) (Received February 9, 1993; accepted March 25, 1993) Abs t r ac t The metal organic chemical vapour deposition (MOCVD) of amorphous alumina films on steel was performed in nitrogen at atmospheric pressure. This MOCVD process is based on the thermal decomposition of aluminium- tri-sec-butoxide (ATSB). The effect of the deposition temperature (within the range 290-420 C), the precursor vapour pressure (5.33 x 10 - 3_2.67 x 10 - 2 kPa), and the gas flow (6.5-12.5 1 min - t) of the MOCVD process have been studied in relation to corrosion properties at high temperatures. The corrosion experiments were performed at 450 C in a gas atmosphere containing 1% H2S, 1% H20, 19% H2, and balanced Ar. It was found that the amount of corrosion products on an alumina film (0.20__. 0.05 mg cm- 2) - AI SI 304 combination decreased with increasing deposition temperature of the coating. This was more pronounced for the products formed through the coating owing to a certain porosity. The crack density, where products were also formed, was almost unaffected. 1. I nt roduct i on For the deposi t i on of al umi na several react ant s have been investigated [ 1- 11] , f or exampl e A1C13 which can react with car bon di oxi de and hydr ogen under the f or mat i on of al umi ni um oxide. These layers are most l y used in the el ect roni c i ndust r y f or t he fabri ca- t i on of met al oxi de semi conduct or - f i el d- ef f ect transis- t ors [12-14]. The process has t o be carri ed out at high t emperat ures, nor mal l y bet ween 850 and 1600 C, which demands a high energy consumpt i on. Fur t her - more, the experi ment al pr ocedur e needs several gas streams, leading to an i ncreased amount of par amet er s influencing the propert i es of t he al umi na films. For example, owing to different deposi t i on t emperat ures, several well-defined oxi de films can be obt ai ned with a desired structure. At t emper at ur es of about 850 C an amor phous l ayer [12] can be obt ai ned, as can a fine grai ned pol ycryst al l i ne st ruct ure [13] at t emper at ur es up to 1000 C, an al pha phase at t emper at ur es up to 1400 C, and at hi gher t emper at ur es a single crystal st ruct ure [14] can be obt ai ned. The deposi t i on by means of compounds such as AICI3, is called chemical vapour deposi t i on ( CVD) . It is also known t hat met al al koxi des are somet i mes excel- lent precursors f or the deposi t i on of met al oxides such as SiO2, TiO2, Cr203 and A1203 [ 15- 18] . The use of * Author to whom all correspondence should be addressed. these precursors offers several at t ract i ons: the met al al koxi des can be purified to a high degree, t hey cont ai n mor e t han enough oxygen for the f or mat i on of the desired oxide, and, last but not least, the much l ower deposi t i on t emper at ur e compar ed with nor mal CVD can also be ver y at t ract i ve i f the subst rat e does not suffer under high t emperat ures. The deposi t i on of thin oxi de films by t hermal decomposi t i on of organomet al l i c compounds is nor mal l y called met al organi c chemical vapour depo- sition ( MOCVD) . The t ri val ent met al al koxi des with no alkyl groups are frequent l y consi dered to be rela- tively stable at elevated t emperat ures. They can be heat ed wi t hout decomposi t i on. Thus for our experi ment s onl y met al (al umi ni um) al koxi des are rel evant such as A1- t ri -t ert -but oxi de ( ATTB) , Al - t ri - sec- but oxi de ( ATSB) , Al - t r i - n- but oxi de ( ATNB) , Al - t r i - i s opr opoxi de ( ATI ) , and Al - acet yl - acet onat e ( AAA) . The use of ATSB was r epor t ed onl y by Okuyama et al. [19] and Kodas et al. [20] f or the pr oduct i on of ultrafine al umi ni um oxi de aerosol particles by t hermal decomposi t i on of ATSB. Physical propert i es of ATSB and some ot her al umi ni um al koxi des were r epor t ed by Sladek [2] and Wi l hoi t [21] regardi ng the equi l i bri um part i al pressure. Fr om t he l i t erat ure it is clear t hat until t oday not hi ng has been r epor t ed about MOCVD using ATSB as the met al al koxi de for the f or mat i on of an adher ent and dense al umi na film. The aim of our st udy was the devel opment and testing of dense and adher ent t hi n al umi na films, ap- 0040-6090/93/$6.00 1993 - - Elsevier Sequoia. All rights reserved 11".A. C. Haanappel et al. / MOCVD production of A1203 coatings using ATSB 139 plied on metallic materials. These cerami c coat i ngs shoul d pr ot ect the underl yi ng subst rat e agai nst t he corrosi ve envi r onment s at high t emper at ur es such as coal gasification and gas t urbi ne at mospheres. In this paper the results are present ed regardi ng t he deposi t i on of al umi na by MOCVD f r om ATSB. The kinetic as- pects in rel at i on to the prot ect i veness of t he coat i ng in si mul at ed coal gasification at mospher es are discussed. 2. Experimental details Deposi t i ons were per f or med on AI SI 304 (18% Cr, 8% Ni, 0.08% C, 1% Si, 2% Mn, 71% Fe). The AI SI 304 samples were cut f r om an el ect ropol i shed sheet and subsequent l y ul t rasoni cal l y cl eaned f or 30 mi n in RBS soap, hexane and et hanol . Finally, t he samples were i mmersed in St ruer' s et chi ng fluid (5% sol ut i on of 3 M nitric acid in et hanol ) f or 15 min, washed with pur e et hanol and dri ed in ai r at r oom t emper at ur e. For t he cor r osi on experi ment s, t he samples were coat ed with 0 . 2 0 _ 0.05 mg c m- 2 al umi na. Al umi na films were pr oduced by decomposi ng A1- tri-sec-butoxide ( ATSB, Janssen Chi mi ca) under at mo- spheric circumstances. A schemat i c view of t he experi ment al appar at us is given in Fig. 1. A st ream of pur e ni t rogen gas was passed t hr ough a coil in a silicon oil bat h at 138 C where t he ni t r ogen gas was sat ur at ed wi t h the ATSB. The resulting gas was mi xed with anot her ni t rogen gas st ream t o achi eve t he requi red concent r at i on of the pr ecur sor before ent eri ng t he reac- t i on tube. The sample was at t ached t o a cerami c t ube wi t h a t her mocoupl e on the inside f or accur at e mea- surement of t he sample t emperat ures. The st andar d condi t i ons f or t he MOCVD process were also used f or Figs. 5- 12, and except i ons are ment i oned in t he figure capt i ons: silicon oil bat h wi t h ATSB, 138 C; car r i er gas flow (N2), 1300 ml mi n - 1 ( STP) ; di l uent gas flow (N2), 5200 ml mi n- ~ ( STP) ; subst rat e t emper at ur e, 330 C. The cor r osi on experi ment s were carri ed out f or 24 h in a closed system. Before st art i ng t he experi ment , t he system was flushed with ar gon ( sat ur at ed with wat er at 15 C) f or 20 h with a flow rat e of 7 1 h- ~ ( STP) . The mi xt ur e of gases (1% H2S, 1% H20, 19% H2, bal. Ar. ) was i nt r oduced aft erwards, with a flow rat e of 16 1 h - 2. Af t er 2 h the f ur nace was heat ed t o the t emper at ur e at whi ch the experi ment s were per f or med. The whol e sys- t em was flushed f or anot her 2 h and t hen closed. It was previ ousl y f ound t hat no rel evant difference in cor r o- sion rat e and cor r osi on pr oduct s was obt ai ned bet ween a closed system and a system with a cont i nuous flow, if the reactive gas cons umpt i on di d not exceed mor e t han 18%. The mor phol ogy and t he composi t i on of t he layers were investigated by means of opt i cal mi cr oscopy and At mos pher i c MOCVD 1 =__= 1: Thermocouple with specimen 2: Furnace 3: Nitrogen 5.0 carder gas 4: Container with ATSB at constant tempera- ture 5: Nitrogen dilution gas =-== 3 4 Fig. 1. Schematic view of the MOCVD apparatus: 1, thermocouple with specimen; 2, furnace; 3, nitrogen 5.0 carrier gas; 4, container with ATSB at constant temperature; 5, nitrogen dilution gas. scanni ng el ect ron mi cr oscopy ( SEM) ( J EOL M 35 CF) equi pped with an EDX analysis system ( Kevex Del t a, class III). The deposi t i on rat e of the films on the metallic subst rat e was det er mi ned by weighing t he samples be- fore and af t er the deposi t i on. 3. Results and discussion 3.1. Deposition experiments The effect of t he deposi t i on t emper at ur e ( 290- 420 C) on t he rat e ( mg c m- 2 h-1) is shown in Fig. 2. The overal l act i vat i on ener gy f or t he het er ogeneous r eact i on is 83 _+ 5 kJ mo l - t (cal cul at ed f or the r eact i on rat e limited film growt h). At hi gher t emper at ur es the deposi t i on rat e increases mor e slowly t han at l ower t emperat ures. A f ur t her increase in t emper at ur e will result in a decreased deposi t i on rat e and becomes i rreproduci bl e wi t h increas- ing t emper at ur e. Thi s means t hat t hree regi ons can be distinguished: region 1, represent i ng t he deposi t i on be- havi our at low t emperat ures, shows a deposi t i on rat e cont r ol l ed by react i ons bet ween t he subst rat e and t he pr ecur sor ( r eact i on rat e limited film gr owt h) ; region 2, at hi gher t emperat ures, represent s t he deposi t i on rat e cont r ol l ed by mass t r ans por t of t he react ant s in t he gas phase (di ffusi on rat e limited film gr owt h) ; and region 3, at the highest t emperat ures, represent s mass 1 4 0 V. A. C. Haanappel et al. / MOCVD production of A1203 coatings using ATSB 1 . 2 - 1 i 0.0 ~. 0 . 6 - ! 0 . 4 - 0.2 1 ', 1,4 o ~ t ~ \ \ o o 125 116 117 120 ~ . . 9 1000/T(~ Fig. 2. Deposition rate (mg cm-2 h-~) of alumina as a function of the deposition temperature (K). 2 1.75 - 1 . 5 . ~ 1.25. ~ 0 . 7 5 O.S~ 0 . 2 g ~ I 0 0 2 [ ] [ ] [ ] ~ - Dep. Temp. (C) [ ] 370 4OO ,~, 420 7 8 9 10 11 12 13 Gas Flow (I.min-1) Fig. 4. Deposition rate (mg cm-2 h-~) of alumina as a function of the gas flow (1 min-~) and the deposition temperature (C). t r anspor t of the react ant s as well as homogeneous gas-phase reactions. The deposi t i on rat e ( mg cm -2 h -~) as a funct i on of the part i al pressure of ATSB (5.33 x 10- 3- 2. 67 10 - z kPa) at vari ous t emper at ur es (330, 370, 400 and 420 C) is given in Fig. 3. The react i on or der of the deposi t i on of al umi na ( f r om ATSB) coul d be cal cul at ed f r om the equat i on: In[deposi t i on rat e ( mg c m- 2 h- l ) ] = k l + n In[part i al pressure ATSB (kPa)] (1) where n is t he react i on or der and k~ is a const ant . It was f ound t hat in the t emper at ur e range 330- 420 C the deposi t i on rat e of al umi na is about first or der (1.0 + 0.2) with respect t o t he initial concent r at i on of ATSB. The overall r eact i on includes several steps, such as: ( I ) diffusion t o and adsor pt i on of t he react ant s on the subst rat e surface; (2) chemi cal react i on and diffu- sion on the surface; and (3) desor pt i on and diffusion of product s (gases) f r om the surface. In general one step pr edomi nat es the overal l react i on rat e and, hence, represents the t ot al resistance. 0 . 5 . [ I / - r 0 ~ ' ~ / i i -0.5 / O / / ~ / . / / '5 1 ~ D e p . T e m p . (*C) ] - I . 5 -2.5 -3.2 -2.8 -2.4 -2 -1'.6 -1.2 In (Vlkoour Pr ~ s u r e ATSB (ram Hg)) F i g . 3 . D e p o s i t i o n r a t e ( r a g c m - 2 h ) o f a l u m i n a a s a f u n c t i o n o f t h e concentration ATSB (kPa) and the deposition temperature (C). In Fig. 4 the effect of the gas flow (1 mi n - t) on the deposi t i on rate ( mg cm -2 h -~) is shown. At high t emper at ur es the gas flow has a mor e prongounced effect on the deposi t i on rat e t han at l ower ones ( < 370 C), where t he deposi t i on rat e was, within t he accur acy of the experi ment s, pract i cal l y i ndependent of t he gas flow. A quant i fi cat i on of the MOCVD process can be per f or med by the ' film model ' , set up by Bird [23]. Thi s model implies: (1) a st agnant boundar y diffusion l ayer f or med ar ound the subst rat e with a thickness 6; (2) react i ons onl y t aki ng place on the subst rat e surface; and (3) a first or der process wi t hout homogeneous react i ons t aki ng place in the boundar y layer. The depo- sition rat e can now be descri bed using t he fol l owi ng simplified equat i on: J - [ATSB] (2) ( 1 / k r + 1/kg) where D is the diffusion coefficient, J is t he deposi t i on rat e, [ATSB] is t he concent r at i on of ATSB, kr is the rat e const ant of the surface react i on, kg = ( D / 6 ) is the mass t ransfer coefficient in the gas phase, and 6 is the thickness of t he diffusion boundar y layer. Two ext reme si t uat i ons can be observed: f or k r / k g >> 1 the deposi t i on rat e is det er mi ned by diffusion-limited growt h, and f or k r / k g << 1 t he deposi t i on rat e is limited by the react i on rate. Thi s equat i on implies t hat under react i on-rat e-l i mi t ed ci rcumst ances, the deposi t i on rat e is onl y a funct i on of kr, whereas it is onl y a funct i on of k, i f the react i on is compl et el y di ffusi on-cont rol l ed. Thi s means t hat i f l i mi t at i on by diffusion t akes place, the deposi t i on rat e is also a funct i on of the t ot al flow rate. At hi gher t emper at ur es the deposi t i on rat e will increase pr opor t i onal to t he flow rate. This is in agreement with Fig. 4, where t he deposi t i on rat e (rag cm -2 h -~) is given as a f unct i on of t he gas flow. Her e it was f ound t hat at t emper at ur es above 370 C, the deposi t i on rat e increased by increasing the gas flow. Fr om this figure it is also cl ear t hat the difference of the V. A. C. Haanappel et al. / MOCVD production of AI203 coatings using ATSB 141 deposition rates between 370 and 400 C is much higher than between 400 and 420 C. This can be explained by: (I) a change of the deposition regime from reaction limitation to diffusion limitation; and (2) a homo- geneous reaction (alumina powder formation in the gas phase) which will become more pronounced at higher temperatures. 3.2. Corrosion experiments In Figs. 5 - 7 the relative weight gain is shown, defined as 100% x (the ratio between the weight gain of a coated and an uncoated sample). Fr om Fig. 5 it is clear that the amount of corrosion products (relative weight gain) tends to decrease with increasing deposi- tion temperature. At deposition temperatures above 370 C the relative weight gain scatters somewhat. In Figs. 6 and 7 the relative weight gain is shown as a function of the initial precursor concentration and the gas flow. The deposition rate is directly related to the partial pressure of the precursor, as seen in Fig. 3. From these figures it is clear that the relative weight 100 8O (.0 6O i . o 20 290 ] 310 r 330 ' 350 ~ 370 390 blank 300 320 340 360 380 Deposition temperature (" C) 420 400 Fig. 5. Bar diagram of the relative weight gain (%) of the specimens (after 24 h sulphidation at 450 C) as a function of the deposition temperature (C). 100 blank Dep. Temp. (*C) W ~ 3to [ ] 4ao 0.2 0.4 0.6 0,8 1 Vepour Pressure ATSB (ram Hg) Fig. 6. Bar diagram of the relative weight gain (%) of the specimens (after 24 h of sulphidation at 450 C) as a function of the vapour pressure of ATSB (kPa) and the deposition temperature (C). 100 ~ ~ ' i blank Dep.Temp. (' C) ~ 370 I I ~ o E3 ~o 6. 5 8 10 12 Gas Flow (Lmin-1) Fig. 7. Bar diagram of the relative weight gain (%) of the specimens (after 24 h of sulphidation at 450 C) as a function of the gas flow (1 mi n- t) and the deposition temperature (C). gain is almost unaffected by changing the partial pres- sures of the precursor (concentration) between 5.33 x 10 -3 and 2.67 10 -2 kPa and a gas flow between 6.5 and 12.5 1 min-~. In Figs. 8( a) - 8( d) SEM pictures of the sulphidized samples are shown. These figures represent the surface of a sample coated under standard conditions and at deposition temperatures ranging from 330 C to 420 C. At higher deposition temperatures the corrosion attack decreases, especially by a reduced scale porosity. Espe- cially, the FeS formation between the chain-like prod- ucts is reduced. Cracks in the alumina films are responsible for the chain-like corrosion products, and are more pronounced near the edges. Higher magnifications (Fig. 9) show that Fe sulphide is also formed through the alumina film, indicative of a porous structure of the film (deposition under standard conditions and after 24 h sulphidation). This is more pronounced at lower deposition temperatures. Experiments were also :performed by changing the ATSB concentration and gas flow. Fr om the weight gain bar diagrams it is clear that the relative weight gain was not strongly affected by those variations. On the contrary, SEM observations show that at lower deposition temperatures the surface morphol ogy of the corroded specimens has changed somewhat with respect to different ATSB concentrations. Depositions per- formed at low deposition rates result in a surface where corrosion products are mainly concentrated near the edges. At higher deposition rates corrosion products were found all over the specimens, and this was more pronounced at lower deposition temperatures than at higher ones. See Figs. 10 and 8(c) showing the surface morphol ogy after sulphidation (deposition at 400 C and a partial pressure of ATSB of 5.33 x 10 -3 and 2.67 x 10 -2 kPa, respectively). The best protection of the alumina films against high temperature corrosion was obtained at high deposition 142 V. A. C. Haanappel et al. / MOCVD production of Al203 coatings using ATSB (a) (b) (c) (d) Fig. 8. SEM image of the surface morphology of coated samples after 24 h of sulphidation at 450 C: (a) deposition temperature, 330 C; (b) 370 C; (c) 400 C; and (d) 420 C. Fig. 9. SEM image of the surface morphology of a coated sample after 24 h of sulphidation at 450 C showing Fe-rich sulphides products which were formed throughout the alumina scale. temperatures bot h considering the relative weight gain of the specimens and the surface morphol ogy after the corrosion experiments. Two types of corrosion products could still be found on the surface of the sulphidized specimens: ( l ) chain-like corrosion products (Fig. 8(d)); and (2) single crystals, all consisting of Fe-rich sulphides (Fig. 11). Cracks in the alumina films are responsible for the chain-like corrosion products. Owing to these cracks in the alumina coating the aggressive gas components can reach the underlying imetallic surface. The existence of the cracks is generally related to the internal stress in the thin oxide films [24]. Oxide films applied by MOCVD suffer in general from thermal and intrinsic stresses, according to: O ' i n t e r n a l = O ' i n t r i n s i c - A i - O ' t h e r m a I These internal stresses may be generated by different sources: during film formation and/ or owing to thermal mismatch. The thermally induced part of the stress in alumina on steel can be calculated from the difference between the thermal expansion coefficients of the film and the metal, the difference between the deposition and the actual temperature, and the Young' s modulus of the film. The intrinsically induced part is caused by the growth mechanism of the amorphous alumina film. ~ N . - . ' ~ ' ) . - : ' ~ ' ~ . . . , , . ~ > 7 ) ' ; . . . . . . Jt " : " ' < ~ 7- " . . . . . . . . . . . . . . . . "" " ~ - X ~ V. A. C. Ha a n a p p e l et al. [ MOC V D pr oduc t i on o r A l 2 0 3 coat i ngs us i ng A T S B 143 ( a ) ' : ' , , - . - ~ , ~. - " , ' < . - - " z,~",'1",~ ..-E~.?"~-.,' \ %:~7~"-~-,. " ' " , (~" " Lk--<- - - ,.~',~ ,.--, : 7;z,: L- " " ,". --r,:. . . . . . :" . : . --,..,.,:., . .:.,," . . . . _ Fig. 12. SEM image of the surface morphology of a coated sample after 24 h of sulphidation at 450 C, showing cracks near the edge of the specimen. Deposition temperature, 400 C; partial pressure of ATSB, 2.67 x 10 -2 kPa. (b) Fig. 10. SEM image of the surface morphol ogy of coated samples after 24 h of sulphidation at 450 C showing Fe-rich sulphide products. Partial pressure of ATSB, (a) 5.33 x 10 -3 kPa, (b) 2.67 x 10 -2 kPa; deposition temperature, 400 C. 1 r ! , _ I Fig. 11. SEM image of the surface morphol ogy of a coated sample after 24 h of sulphidation at 450 C showing Iocalised Fe-rich sulphide products. Deposition temperature, 400 C; partial pressure of ATSB, 5.33 x 10 -3 kPa. The origin of the intrinsic stresses is yet unclear. Mor ssi nkhof [25] report ed t hat the large intrinsic stress may be explained by the presence of hydroxi de groups in the oxide scale. However, the relationship between the i ncorporat i on of elements or groups of at oms and the critical coat i ng thickness is not yet clearly under- stood. Tucker [26] ment i oned t hat the residual stress increased linearly with coat i ng thickness but an exact mechani sm explaining this relation was not formul at ed. Crack f or mat i on was also f ound by Evans e t al . [27], who report ed t hat a curved substrate or a surface showing sinusoidal interfaces, will lead to an increased susceptibility to crack format i on. This explains the exis- tence of subst ant i al shear and tensile stress concent ra- tions near edges and film t ermi nat i ons (see also Fig. 12). The porous st ruct ure of t he al umi na films, which was more pr onounced for low t han high deposi t i on temper- atures, can be explained by the low conversions of the met al alkoxides or i ncorporat i on of A10( OH) or H20 into the film. For example, Baryshni kov [28] report ed t hat at low conversion of the met al alkoxides, the correspondi ng alcohol was identified in amount s of up to 1 mole per mole of the substance being decomposed. In a later stage the alcohol decomposes at the surface of the specimen. To increase the protectiveness of the al umi na coat- ings, a crack- and pore-free oxide film wi t h reduced stresses shoul d be produced. This can be obt ai ned by, amongst others, the f or mat i on of an i nt ermedi at e oxide layer between the coat i ng and the metal, lowering of the deposi t i on rate at high deposi t i on t emperat ures, the i ncorporat i on of elements in the oxide lattice such as Si [29] which will result in a higher density of the film, and the addi t i on of ot her compounds to the gas st ream sat urat ed with the precursor. Also, t hermal anneal i ng by a heat -t reat ment aft erwards may result in reduced stresses [5, 30]. In the near fut ure more research will be performed in order to obt ai n al umi na films with well-defined chemi- cal and mechanical properties. 144 V. A. C. Haanappel et al. / MOCVD production of AIzO 3 coatings using ATSB 4. Conclusions Thin films of aluminium oxide can be obtained by pyrolytic decomposition of Al-tri-sec-butoxide, The activation energy for the heterogeneous reaction is 83 + 5 kJ mol -1. The surface morphol ogy of the thin films, deposited on metallic substrates, showed cracks and a porous structure, which were more pronounced at low deposition temperatures. Eventhough the coated speci- mens are much less attacked than uncoated ones, the application of these layers for corrosion protection of the underlying metallic materials is as yet suffering from fast diffusion paths such as cracks and pores. The aggressive gas components can consequently easily reach the underlying substrate, forming, for example, sulphide products. Higher deposition temperatures, correspond- ing to the region of diffusion-limited film growth, leads to a less uniform but a more protective behaviour of the alumina film owing to a denser structure. More research is necessary for the preparation of uniform thin alumina films with well-defined chemical and mechanical properties by several methods such as the addition of foreign elements, thermal annealing and the addition of other compounds to the gas stream. Acknowledgments This research was supported by the Innovative Re- search Program on Technical Ceramics (IOP-TK) with the financial aid of the Dutch Ministry of Economic Affairs. The authors thank J. Wassens, A. Thiel, and J. B. Rem for the technical assistance. References 1 L. Hall and B. Robinette, Properties of Al umi ni um Oxide Films obt ai ned from Ni t rous Oxide and Al umi ni um Trimethyl, in J. M. Blocher, Jr., and J. C. Wi t hers (eds.), Chemical Vapour Deposi- tion, Sec. Intern. Conf., Electrochem. Soc., Inc., New York 1970. 2 K. J. Sladek and W. W. Gibert, Proc. Eur. Cong. CVD, (1972) 215. 3 A. A. Barybin and V. I. Tomilin, Zh. Prikl. Khim., 49(1976) 1699. 4 V. F. Korzo, Zh. Prikl. Khim., 49( 1975) 74. 5 S. B. Desu, J. Am. Ceram. Soc., 72(1989) 1615. 6 V. M. Koleshko, V. V. Sviridov, V. P. Boldyrev, B. S. Reznikov, I. V. Nekaryuki n, K. D. Yashi n, N. N. Gor os hko, and A. A. Kovalevskii, lzv. Akad. Nauk SSSR, Neorg. Mater., 12 (1976) 1780. 7 A. A. Barybin, V. I. Tomilin, V. A. Kempel and E. A. Makhot i n, Prib. Tekh. Eksp., 3 (1975) 238. 8 J. Saraie, J. Kwon and Y. Yodogawa, J. Electrochem. Soc., 132 (1985) 890. 9 M. Mat sushi t a and Y. Koga, J . Electrochem. Soc., 115 (1968) 69C. 10 N. Y. Turova and M. I. Yanovskaya, lzv. Akad. Nauk SSSR, Neorg. Mater., 19 (1983) 703. 11 G. P. Schulman, M. Trust y and J. H. Vickers, Thermochem. Acta, 85 (1963) 907. 12 S. W. Choi, C. Kim, J. G. Kim and J. S. Chum J. Mat. Sei., 22 (1987) 1051. 13 P. Wong and McD. Robi nson, J. Am. Ceram. Soe., 53 (1970) 617. 14 D. R. Messier and P. Wong, J. Electrochem. Soc., 118(1970) 772. 15 K. Natesan, Corrosion (Houston), 41 (1985) 646. 16 P. Kofstad, High Temperature Corrosion, Elsevier, Amsterdam, 1988. 17 D. C. Bradley, Chem. Rev., 89 (1989) 1317. 18 D. C. Bradley, R. C. Mehr ot r a and D. P. Gauer, Metal Alkoxides, Academic Press, London, 1978. 19 K. Okuyama, Y. Kousaka, N. Tohge, S. Yamamot o, J. J. Wu, R. C. Fl agan and J. H. Seinfeld, AI ChE J., 32 (1986) 2010. 20 T. T. Kodas, A. Sood and S. E. Pratsinis, Powder Technol., 50(1) (1987) 47. 21 R. C. Wilhoit, J. Phys. Chem., 61 (1957) 114. 22 K. R. Westerterp, W. P. M. van Swaaij and A. A. C. M. Beenack- ers, Chemical Reactor Design and Operation, Wiley, New York, 1984. 23 R. Bird, W. E. Steward and E. N. Lightfoot, Transport Phenom- ena, New York, Wiley, 1960, p. 658. 24 M. G. Hocking, V. Vasantasree and P. S. Sidky (eds.), Metallic and Ceramic Coatings, Longman, London, 1989. 25 R. W. J. Morssinkhof, The deposition of t hi n al umi na films on steels by MOCVD, PhD Thesis, University of Twente, Enschede, Netherlands, 1991. 26 R. C. Tucker, J. Vac. Sci. TechnoL, 11 (1974) 725. 27 A. G. Evans, G. B. Crumley and R. E. Demaray, Oxid. Met., 20 (1983) 193. 28 Yu. Yu. Baryshnikov, I. L. Zakhar ov and G. I. Makin, Zh. Obshch. Khim., 60 (1990) 1350. 29 F. C. Eversteijn, Philips Res. Rep., 21 (1966) 379. 30 S. B. Desu, Jpn. J. AppL Phys., 30( 1991) L2123.