J. MATER. CHEM., 1994.4( 5), 669-673 Influence of Methane on the Nitriding Gas Reduction of Kaolinite Alain Seron,a Jacques Thebaulf‘ and Franqois Beguin” a CRMD, UMR CNRS Universite, 16 rue de la Ferollerie, 45071 Orleans cedex 2, France SEP, les cinq chemins, le Haillan, BP 37, 33165 St Medard en Jalles, France A new synthesis of P’-SiAION via the reduction and nitridation of aluminosilicates in the presence of a carbon source by a hydrogen-nitrogen gas mixture is presented. The formation of p-SiAION on surfaces (Sic, alumina) devoid of carbon was found to be impossible. Thermogravimetric analysis and mass spectrometry showed that, in a carbon crucible, methane was formed from the reaction of hydrogen with carbon. p-SiAION was obtained in the absence of free carbon by adding a small amount of methane to the hydrogen-nitrogen gaseous mixture.The formation of P’-SiAION is thus the consequence of reduction by methane rather than by hydrogen. Parasitic reactions of hydrogen lead to silicon elimination as SiO in the gas phase, and to a Si/AI ratio lower than in the starting kaolinite. Oxide ceramics have undergone important developments for centuries. Traditional ceramics have been used to make pot- tery, glass, bricks or cement which play an important role in everyday life. New oxide ceramics have been developed for many applications, but non-oxide ceramics are now widely used in specific cases. Of the non-oxide ceramics, carbides and nitrides are currently the most attractive. In addition, intermediates between oxide and non-oxide ceramics, new compounds such as oxycarbides and oxynitrides have now been developed.Our interest is in the synthesis of P‘-SiAlONs, which are silicon and aluminium oxynitrides.’ Compared with silicon nitride (P-Si3N4), from which they are derived, the main interest of these ceramics is their possibility of being densified. Indeed, obtaining dense materials from silicon nitride is impossible without additives such as A1203, MgO or Y203.’ Among the various preparation processes, p-SiAlONs can be formed from clay minerals at high temperature, using carbon as a solid reducer under a nitrogen atmosphere. Higgins and Hendry2 demonstrated that at least 4 h heating was necessary at 1400°C to obtain a significant amount of p-SiAlON.The powders obtained, after reaction, generally contain residual carbon which hinders sintering and prevents the formation of film coatings. On the other hand, even if obtaining P‘-SiAlON is easy, this compound is always synthe- sized together with by-products such as mullite or aluminium nitride.3,4 Such problems seem to be related to the inhomogen- eity of the solid-solid mixture even if some improvements can be obtained using carbon-oxide nanocomposites in which the reagents are highly di~persed.~.’ Better contact between the clay mineral and the reducing agent can be realized using a solid-gas reaction. In a previous paper: we described the synthesis of p-SiAlON using nitriding hydrogenoreduction of kaolinite in a nitrogen-hydrogen atmosphere, according to the theoretical equation (1) This is an easy method of producing P’-SiAlON at fairly low temperature (llOO°C), in a reasonable time (65 h).Even if 1450“C is the best temperature for carbored~ction,~ reaching such a temperature with hydrogen leads to products contain- ing a major portion of aluminium nitride which is the conse- quence of the reduction of P’-SiAlON. The thickness of the heat-treated samples strongly influences the nature of the products, owing to diffusion phenomena. Moreover, we found that the heat-treatment of aluminosilicates in alumina or silicon carbide crucibles under nitrogen-hydrogen flow ( 1:3) did not yield P’-SiAlON, as did the same treatment in carbon crucibles.Because gas reduction occurred at a very low temperature compared with carboreduction, we suggested the existence of a carbonaceous species in the gas phase6 These results seem to contradict those reported by Wild; who showed that a mixture of P’-SiAlON and A1N could be obtained by heating clay minerals in an alumina crucible under ammonia flow. This paper will demonstrate that carbon is absolutely necessary in the nitriding gas reduction process of kaolini te. Experimental The clay mineral used in this work for producing P-SiAlON was a kaolinite (‘kaolin supreme des Charentes’). The chemical composition given by elemental analysis was: Si 20.35%, A1 18.70%, Fe 0.40%0, Ca 0.10%, Mg 0.20%, Na 0.10%, K 0.85Y0,Ti 0.02%. In a typical experiment, a thin layer of kaolinite (200mg) was heat-treated in alumina, Sic or graph- ite crucibles under nitrogen-hydrogen (N2, 100cm.‘ min-’; HZ, 300 cm3 min-’, both gases from ‘Air Liquidc’: H20, 0,<5 ppm) at 1100 “C6 Tb 1450 “C, with a heating rate of 300°C h-’ and 0.5 to 100h at the highest temperature.We used a vertical tubular furnace with a sintered Al,03 tube (diameter 7 cm). The lower and upper parts of the Al,03 tube were filled with cement cylinders. A 90cm3 free space was available in the centre of the tube to place the crucible containing the sample on the lower cylinder. The temperature of the specimen was determined by a Pt/PtRh 10% thermo-couple under the crucible. Before heating, the furnace was evacuated for 0.5 h to lO-’mbar and then swept by the nitrogen-hydrogen mixture.The synthesized products wer? characterized by powder X-ray diffraction at 3, =1 S405A (Siemens D 500 goniometer) in a reflection set-up. Thermal treatments were also carried out in a thermo-gravimetric analysis (TG) apparatus coupled with a mass spectrometer.8 Such a set-up enables the analyses of the components in the outflowing gases and allowed us identify the different steps of the reaction. For the other experiments, we used a furnace with a vertical A1203 tube. For these analyses, the aluminosilicate powder (ca. 20 mg) was In either alumina or carbon crucibles. The thermobalance wa:, evacu- ated (P=lo-’ mbar) before heat-treatment, then swept by nitrogen-hydrogen flow (1 :3).The temperature of the samples was determined, by a Pt/Pt Rh 10% thermocouple under the 670 crucible and the mass was determined by an electromagnetic balance (10-8 MTB Setaram). Evolved gases were sampled by a heated alumina capillary set just near the crucible and continuously analysed by a mass spectrometer (Balzers QMG 420 C), from which selected masses were plotted as a function of time, i.e. of sample temperature. Results and Discussion When kaolinite was heat-treated in alumina or silicon carbide crucibles, under nitrogen-hydrogen flow (1 :3), only mullite and cristobalite were formed (Fig. l).' We obtained the same phases in the presence of ammonia, contrary to the results reported by Wild.7 Indeed, these authors reported that a mixture of P'-SiAlON and AlN could be obtained by heating clay minerals in an alumina crucible under ammonia flow, and they do not mention the existence of carbon in their experimental set-up.On the other hand, we observed that heat-treatment of kaolinite in carbon crucibles under N,-H, flow led to the formation of /l'-SiAlON after 3 h heating at 1350°C [Fig. 2(b)].Apparently, the presence of carbon near the clay sample is necessary for the production of /l'-SiAlON. After heat-treatment of clay mineral samples in carbon crucibles under N,-H,, scanning electron microscopy showed that the carbon support was corroded near the oxinitride powder (Fig. 3). This was at first attributed to a carboreduction I I 12 ' 16 ' 20 24 28 32 36 40 2#degrees Fig.1 X-Ray diffractogram of the solid phase obtained after thermal treatment of kaolinite at 1200°C (48 h plateay) in an alumina crucible H,-N, =3 : 1, 400 cm3 min-'; (A = 1.5405 A); 0=cristobalite, H =mullite I' , . . . . '+ ' ,.... 'I 1, I,,,,,12 ' 16 20 24 28 32 36 40 2BJdegrees Fig. 2 X-Ray diffractograms of the solid phases obtained after 3 h heat-treatment of kaolinite at 1350 "C. (a) nitrogen, 100 cm3 min-'; =cristobalite, H =mullite. (b) Nitrogen-hydrogeq, H,-N, =3 :1, 100 cm3 min-'; *=P'-SiAlON, 0 =AlN (A= 1.5405 A). J. MATER. CHEM., 1994, VOL. 4 Fig.3 Corrosion of a carbon crucible (arrows) after the nitriding hydrogenoreduction of kaolinite, as seen by scanning electron microscopy phenomenon allowing the formation of P'-SiA10N,3 but this hypothesis was contradicted by the result of heating pure kaolinite in a graphite crucible at 1350 "C under nitrogen.Under these conditions, and even after 3 h heating, neither AIN nor P'-SiAlON was formed [Fig. 2(u)]. The only phases detected were mullite and cristobalite, whereas /i'-SiAlON and A1N were easily synthesized in a nitrogen-hydrogen flow. Slight carboreduction could only be observed after 48 h heating in a graphite crucible at 1450"C, whereas P'-SiAlON was totally reduced into A1N after only 10 h under N2-H2 flow. It is therefore likely that P'-SiAlON is not formed by a carbothermal process through a solid-solid interaction between the clay mineral and the carbon support.To demon-strate the influence of a gas formed from the carbon support, a kaolinite sample in an alumina crucible was put on a graphite plate, surrounded by a graphite cylinder, and heated under nitrogen-hydrogen (Fig. 4). Under these conditions, p'-SiAlON with a small quantity of mullite were obtained after heating at 1350°C for 3 h [Fig. 5(b)].On the contrary, when the alumina crucible was only put on the carbon plate without any graphite cylinder, the major phase obtained was mullite [Fig. 5(u)]. Another proof that a gaseous carbon species could influence the reduction process was given by the reaction of silica under N,-H2 flow in an alumina crucible surrounded by a graphite cylinder (Fig. 4). The only phase detected (Fig.6) was silicon carbide, showing that a sufficient amount of a carbonaceous species was carried by the gaseous phase to reduce all the powder. Indeed, Shickg and Lee and Cutler"' showed that carbon could reduce silica in this temperature range and convert it to silicon carbide. We therefore propose for the reduction of SiO,: SiO, +3C(g)+Sic +2CO (2) Fig. 4 Experimental set-up design to clarify the role of carbon J. MATER. CHEM., 1994, VOL. 4 671 ~ , , . . , .,..., ...,...,.. , .,, , .., ,.. , ,.J I.. ..,. ,,.. ,.. , .,,...,...,...,., /.. .,. ,,..., ..,,.., ../,I12 16 20 24 28 32 36 40 2Wdegrees Fig. 5 X-Ray diffractograms of the solid phases obtained after 3 h heat-treatment of kaolinite at 1350'C in an alumina crucible (u) without graphite cylinder, W =mullite, *=p-SiAlON; (b) with graphite cylinder, Hz-N2=3 :1; 400 cm3 mi;-';W =mullite, *=p'-SiAlON, +=Al,O,N, W =AlN; (3.=1.5405 A) 1Ii iI i i ~ I 12 16 20 24 28 32 36 40 2Hdegrees Fig. 6 X-Ray diffractogram of the solid phase obtained after thermal treatment of silica surrounded by a graphite cylinder Jplateau at 1200"C, 48 h); Hz:Nz=3 :1, 400 cm3 min-'; (i=1.5405A); V=Sic where C(g) represents a carbon-bearing gaseous species. To confirm this hypothesis we built an apparatus consisting of a thermogravimetric (TG) analysis apparatus and a mass spec-trometer, thus giving us simultaneously the mass loss and the molecular mass of the evolved gases (m/z us. time). It was shown previously that when kaolinite is heat-treated in an alumina crucible under neutral atmosphere (N,, Ar), only one mass loss is observed, close to 500°C.6 The XRD analysis of the resulting powder indicated that it contained non-reduced phases: mullite and cristobalite.The mass loss observed by TG was confirmed by mass spectrometry, which showed peaks at m/z= 17 (OH') and m/z= 18 (H,Of) cor-responding to the dehydroxylation of kaolinite. A similar heat-treatment of a kaolinite sample in a graphite crucible led to two mass losses at 500 and 1100"C,respectively (Fig. 7). The first is attributed to the loss of hydroxy groups whereas the second, which reaches 74% in mass after a 1 h plateau at 1350"C,seems to be due to the reduction of a part of silica by the carbon substrate'' generating the loss of silicon as SiO and carbon as CO: SiO, +C+ SiO +CO (3) The SiO loss is confirmed by the X-ray diffraction spectrum of the powder obtained, which only contains a small amount of silica as cristobalite (Fig.8) compared to the one heat-treated in the same conditions in an alumina crucible (Fig. 1). On the other hand, mass spectrometry of the gas evolved (Fig. 9) showed a loss of carbon monoxide and carbon dioxide, isotherm I I 11 I 250 500 750 1000 1250 1360 V0C Fig. 7 Thermogravimetric analysis of kaolinite in a graphite crucible under nitrogen flow; heating rate 200°C h-l, plateau at 1350°C; 1 h, DN2=100 cm3 min -II II 'I2 ' 16 ' 20 ' 24 28 ' 32 ' 36 ' 4'0 28Jdegrees Fig.8 X-Ray diffractogram of the solid phase obtaincd after thermal treatment of kaolinite at 1350"C (1 h plateau) in a graphite crucible under nitrogen flow: N,, 100 cm3 min-'; (A= 1 5405 A); =cristobalite, W =mullite .12I I '\II i I I I----------i i I 1 I I I,,isotherm 250 750 125011350 T/"C Fig.9 Mass spectrometry (m/z us. TIT) of the gas generated by the heat treatment of kaolinite in a graphite crucible under nitrogen flow. Same experimental parameters as in Fig. 8 characterized by peaks at m/z= 12 (C') and m/z=44 (CO;) and confirmed by m/z= 16 corresponding to the 0' ion. The existence of these gases must be attributed to the zlassical reaction which favours C02 at low temperature and CO at high temperature (>1000OC): 2CO+-CO,+C (4) 672 The characteristic ion for CO (m/z =28) could not be detected because of the presence of a large amount of nitrogen in the atmosphere, nor could silicon monoxide be detected by mass spectrometry because of its condensation in the cold part of the apparatus.B'-SiAlON was readily obtained by heat-treatment of kao- linite under H,-N, flow in the graphite crucible of the TG apparatus. The TG curves showed two mass losses, in the range 300-600°C (elimination of hydroxy groups) and in the range 1000 to 1350°C (Fig. 10).The last mass loss is almost independent of the initial mass of clay mineral but only influenced by the residence time at 1350 "C. Mass spectrometry (Fig. 11) points to the formation of a small amount of ammonia close to 800°C [NH,' (m/z=17), NH,f (rn/z=16), NH' (m/z = 15)and N+ (m/z = 14)], because these peaks were non-existent when the treatment was performed under a hydrogen-argon flow.Above 1000 "C, symmetric peaks for C' (m/z=12), CH' (m/z= 13), CH,f (m/z=14), CH,' (m/z= 15) and CH,f (rn/z=16), all corresponding to methane, appeared. As expected from the equilibrium constant K, [corresponding to eqn. (5)] the formation of this gas reaches 1.0. F"Ed -isotherm 1350 I I I I I , , I I I I 250 750 1250 1250 750 TI0C Fig. 10 Thermogravimetric analysis during heat-treatment of kaolin- ite under nitrogen-hydrogen flow in a graphite crucible; heating rate 200°C h-', plateau at 1350°C, 1 h, H,:N,=3:1,400cm3 min-' 500 1000 1000 500 TI"C Fig.11 Mass spectrometry of the gas generated by the heat treatment of kaolinite in a graphite crucible under nitrogen-hydrogen flow. Same experimental parameters as in Fig. 8 J. MATER. CHEM., 1994, VOL. 4 a maximum rate at a temperature close to llOO°C, whereas increasing the temperature above 1100"C lowers its partial pressure, which becomes stable during the isothermal plateau. Besides the signal corresponding to C+ (m]:= 12), another gas is detected during the isothermal plateau which seems to be CO. It is likely that hydrogen is not directly responsible for the formation of p-SiAlON, but that it rather reacts with the carbon support to form methane [eqn.(5)]which then reduces kaolinite [eqn. (6)]: C +2H, + CH, (5) 3 (A1,03,2Si02)+ 15CH, + 5N, +2Si3Al3 O3N, + 15CO+30H2 (6) In equilibrium (5), the partial pressure of CH, at 12OOcC, deduced from K, (Kp=3.16 x is 300 Pa. By mixing 74.78% hydrogen, 0.22% methane and 25% nitrogen, a mixture with the same partial pressure of CH4 was prepared, which was then allowed to react with kaolinite in alumina crucibles at 1200°C for 48 h. Fig. 12 shows that, under these conditions, almost pure /3'-SiAlON was formed. Because all the experimental data are the same as in Fig. 1, except the small amount of methane incorporated in the vapour phase, this experiment clearly demonstrates that methane is respon- sible for the nitriding reduction of kaolinite.We tried to increase the methane content of the gaseous mixture. For instance, heat-treatment of kaolinite under a flow of nitrogen (100cm3 min-I), hydrogen (300 cm3 min-l) and methane (10 cm3 min-') at 1200"C for 48 h did not yield B'-SiAlON because of the formation of a carbon film on the surface of the sample: the partial pressure of methane was higher than the equilibrium value, and a part was dissociated, accordingly to eqn. (5). Thus, even if methane is solely responsible for P'-SiAlON formation, hydrogen is necessary to avoid the dissociation of the methane. Even if eqn. (6) is fairly representative of B'-SiAlON forma- tion by the gas reduction of kaolinite, it is likely that the real process is more complicated because of side-reactions such as the reduction of silica by hydrogen according to: Si02+H, +SiO +H20 (7) This reaction certainly occurs, because we always synthesized B'-SiAlON with Si :A1 < 1 (Table 1), the theoretical value which should be obtained if only methane reduction occurred Ceqn.To describe the formation of P'-SiAlON from the reaction 12 16 20 24 28 32 36 2Bldegrees Fig.12 X-Ray diffractogram of the solid phase obtained after nitriding reduction of kaolinite in an alumina crucible under hydrogen-nitrogen-methane flow with a plateau at 1200"C (48 h): HZ,?4.78%; N,, 25%; CH,, 0.22%; Dtota,=400 cm3 min-'; (A=1.5405 A); *=/?'-SiAlON, =mullite J. MATER. CHEM., 1994, VOL. 4 Table 1 Si: A1 ratios given by elemental analysis of kaolinite and of samples obtained after nitriding gas reduction of kaolinite on graphite crucibles samplea experimental parameters Si : A1 kaolinite -1.05 H2:N,=3: 1 1 plateau at 1450 “C (10 h) 0.11 H2:N2=3:1 2 plateau at 1350°C (3 h) 0.87 “Sample 1 was identified as p-SiAlON whereas sample 2 is essentially AlN by X-ray diffraction.between kaolinite and ammonia, Wild proposed a reaction scheme very similar to eqn. (l),which does not involve any carbonaceous species. In fact, he described in detail neither the experimental set-up used nor the purity of the ammonia in term of ‘carbon’ traces. Our results suggest that this author was able to obtain p’-SiAlON owing to the presence of unidentified methane, either already existing in ammonia, or formed from the reaction of ammonia with any carbon substrate.Conclusions The synthesis of p’-SiAlON by heat-treatment of kaolinite under hydrogen-nitrogen flow is impossible without a carbon source. Using graphite crucibles, the analysis by mass spec- trometry of the gas evolved showed the formation of methane which acted in the formation of p-SiAlON. Pure P’-SiAlON could be obtained by heat-treatment of kaolinite in alumina crucibles under nitrogen-hydrogen-methane ternary mixture, without another source of carbon. The use a nitrogen-methane gas mixture is, however, impossible because its results in the formation of carbon coating on the surface of the samples which inhibits the reduction. The main reaction generating p’-SiAlON from kaolin is the methane nitriding reduction.However, side-reactions such as the elimination of SiO by hydrogen reduction cannot be rejected. This treatment of kaolinite, at relatively mild temperatures, by the hydrogen- nitrogen-methane mixture, allows the formation of thin films with low porosity at the surface of a carbon substrate. This process could be applicable for the ceramic coatings of carbon materials. Thanks are due to the ‘Societe Europeenne de Propulsion’ for financial support. References 1 K. H. Jack, J. Muter. Sci.,1976, 11, 1135. 2 I. Higgins and A. Hendry, Br. Ceram. Trans. J., 1986,85, 161. 3 J. B. Baldo, V. G. Pandofelli and J. R. Casarini, Ceram. Powd., ed. P. Vincenzini, Elsevier, Amsterdam, 1983, pp. 437. 4 Y. Sugahara, J. Mitamoto, K. Kuroda and C. Kato, Appl. Clay Sci.,1989, 4, 11. 5 A. Seron, I. Ben Maimoun, M. Crespin and F. Beguin, Marer. Sci. Eng., 1993, A168,239. 6 A. Seron, J. Thebault and F. Beguin, J.Muter. Res., in the press. 7 S. Wild, J. Muter. Sci., 1976,11, 1972. 8 A. Seron, PhD Thesis, 1993, Orleans. 9 H. L. Shick, Chem. Rev., 1960,60,331. 10 J. G. Lee and I. B. Cutler, Ceram. Bull., 1975,54, 195. 11 P. D. Miller, J. G. Lee and I. B. Cutler, J. Am. Ceram. So( ., 1978, 62. 147. Paper 3/07414E; Received 16th December, 1993