`
`Journal of The American Ceramic Society-Jorgensen et al.
`Vol. 44, No. 6
`continued to yield 6-alumina as long as the nitrogen content
`IV. Discussion
`was not reduced to a very low value. The 6 phase will trans-
`It has been shown that aluminum nitride additions to alu-
`form to the a phase on removal of the nitrogen.
`mina brought about the formation of a stable high-tempera-
`The formation of I-alumina with the addition of larger
`ture 6-alumina phase. The fact that this phenomenon
`amounts of aluminum nitride to fused alumina completes the
`escaped notice previously can be attributed to the fact that
`high-temperature crystal phases between A1208 and A1N.
`the usual thermal treatment of alumina is in an oxidizing
`There appears to be no other intermediate phase inasmuch as
`atmosphere where aluminum nitride, if present, is readily
`a mixture of I and AlN was observed when the nitrogen con-
`oxidized. Nitrogen per se will not affect the alumina phase
`tent exceeded the 3.9% required for I . The < phase, like the
`because the reaction between aluminum oxide and nitrogen is
`6 phase, will revert to the a phase on removal of the nitrogen
`not favored thermodynamically.
`by oxidation.
`Foster et aL6 reported 6-alumina as the alumina phase in
`The mechanism of these phase transformations in alumina
`aluminum oxide-aluminum carbide melts. These melts were
`containing aluminum nitride is not understood. Recognizing
`studied in the temperature range 1800’ to 200OoC. and the
`these effects may aid in the understanding of the complex
`presence of a 6-alumina phase was unexpected. They were
`aluminum oxide system.
`unable, however, to establish responsibility for this 6-alumina
`phase. Schneider and Gattow’ observed a similar phase on
`burning aluminum in the presence of a carbonaceous promoter
`and attributed the stabilizing effect to carbon. None of
`these experiments was explicit enough to establish responsi-
`bility for the formation of 6-alumina at high temperatures.
`The addition of aluminum nitride to fused alumina yielded
`the 6-alumina phase in the absence of a carbonaceous pro-
`moter. About 3% A1N was adequate to convert all the alu-
`mina to the 6 phase at 205OOC. Repeated fusion of this phase
`
`OL. M. Foster, G. Long. and M. S. Hunter, “Reactions Be-
`tween Aluminum Oxide and Carbon-The AlrO,-Al,C~ Phase
`Diagram,” J. Am. Ceram. SOC., 39 [l] 1-11 (1956).
`7 A. Schneider and G. Gattow, “Zur Bildungswkme des Alumin-
`ium Oxyde” (Heat of Formation of Aluminum Oxide), Z . anorg.
`u. allgem. Chem.. 277 [1-2] 41-48 (1954); Ceram. Abstr., 1958,
`September, p. 250c.
`
`Effects of Water Vapor on Oxidation of Silicon
`Carbide
`by PAUL J. JORGENSEN, MILTON E. WADSWORTH, and IVAN B. CUTLER
`College of Mines and Mlneral Industries, University of Utah, Salt Lake City, Utah
`
`The rate of oxidation of silicon carbide was
`studied at different partial pressures of water
`vapor. The difision-rate constant was found
`to vary with the logarithm of the partial pressure
`of water vapor according to the theory of thin-
`film oxidation as proposed by Engell and H a d e .
`The products of oxidation were cristobalite and
`tridymite, depending on the temperature. The
`dfising species appeared to be the same in the
`presence of partial pressures of water vapor and
`in the presence of partial pressures of oxygen.
`
`1.
`Introduction
`HE kinetics of the oxidation of silicon carbide have been
`studied quite extensively in the presence of air and
`
`T oxygen,’ and these studies have pointed out that the rate
`
`of oxidation follows a parabolic rate law. Jorgensen, Wads-
`worth, and Cutler2 found that the parabolic rate constant
`
`Received August 19. 1960; revised copy received December
`2,1960.
`Part of a thesis submitted by Paul J. Jorgensen in partial ful-
`fillment of the requirements for the degree of Doctor of Philoso-
`phy, University of Utah.
`This work was made possible by a financial grant from the
`National Science Foundation.
`
`varied with the logarithm of the partial pressure of oxygen
`for thin-film oxidation and also found that a small amount
`of water vapor present in the oxidizing atmosphere greatly
`accelerated the rate of oxidation. Leaa made a brief survey
`
`At the time this work was done, the writers were. respectively,
`graduate student, professor and head, Department of Metallurgy,
`and professor and head, Department of Ceramic Engineering,
`College of Mines and Mineral Industries, University of Utah.
`Paul J. Jorgensen is now research staff member, Ceramic Studies
`Section, Metallurgy and Ceramics Research Department, Re-
`search Laboratory, General Electric Company, Schenectady,
`New York.
`(a) A. C. Lea, “Silicon Carbide and Its Use as a Refractory
`Material,” Trans. Brit. Ceram. SOC., 40 [4] 93-118 (1941);
`Ceram. Abstr..21 [l] M(1942).
`(b) T. H. Elmer and W. J. Koshuba, “Investigation of Oxida-
`tion of Silicon Carbide,” NEPA Division Report No. 1768,
`Oak Ridge, Tennessee, March 1951 (unclassified).
`( c ) W. A. Lambertson, “Oxidation of Silicon Carbide.”
`Carborundum Company report (unpublished).
`(d) Guy Ervin, Jr., “Oxidation Behavior of Silicon Carbide,”
`J. Am. Ceram. Soc.. 41 191 347-52 (1958).
`(e) Hiro hige Suzuki. “Study of Oxidation of Silicon Carbide
`Powder,” Yogyo KyoRai Ski, 65 [i36] 88-93 (1957); Ceram.
`Abstr., 1958, January, p. 17a.
`* P. J. Jorgensen, M. E. Wadsworth, and 1. B. Cutler, “Effects
`of Oxygen Partial Pressure on Oxidation of Silicon Carbide,” J.
`Am. Ceram. Soc.. 43 [4] 209-12 (1960).
`a A. C. Lea, “Oxidation of Silicon Carbide Refractory Mate-
`rials,” J. SOC. Glass Tecknol., 33 [la] 27-50T (1949); Ceram.
`Abstr., 1951, November, p. 198J
`
`GE-1020.001
`
`
`
`Eflects of Water Va$or on &&wn
`June 1961
`into the effects of different gases on the rate of oxidation of
`silicon carbide and noted that steam accelerated the rate of
`oxidation, but no satisfactory explanation was given. Three
`possible explanations were advanced, i.e.
`“(a) The volatilisation of silica in an air-steam atmos-
`phere, or,
`the hydrolysis of the coating, or,
`(b)
`the physical
`(c) modification by the atmosphere of
`condition of the silica formed by the oxidation.”
`Lea favored the explanation of hydrolysis of the oxide film.
`Suzuki‘ also studied the role of steam in the oxidation of
`silicon carbide; he found that, in general, water vapor was
`an oxidizing agent for silicon carbide. The rate of oxidation
`was accelerated when water vapor was added to oxygen, and
`Suzuki proposed that possibly water vapor was accelerating
`the diffusion of oxygen, carbon monoxide, or carbon dioxide
`through the layer of SiOz or that water vapor promoting the
`transition of amorphous silica to cristobalite might account
`for the accelerated rate.
`Water vapor undoubtedly has a marked effect on the rate
`of oxidation of silicon carbide. The effects of the partial
`pressure of water vapor on the oxidation of silicon carbide
`therefore were studied.
`II. Experimental Procedure
`A thermogravimetric apparatus was used to measure the
`rate of oxidation of silicon carbide; it consisted of a helical
`quartz spring, a Gaertner cathetometer, and an electric
`furnace heated by Kanthal-Super elements.
`The equipment has been described6; for the present in-
`vestigation the apparatus was modified, however, so that a
`controlled mixture of water vapor and argon could be passed
`through the furnace. The amount of water vapor present
`in the gaseous atmosphere was controlled by means of
`bubbling argon through water held at a constant temperature.
`The use of water vapor as an oxidizing agent necessitated
`the heating of the quartz spring assembly to prevent condensa-
`tion on the spring. The temperature of the spring assembly
`was held constant to f l0C. by means of a separate controller.
`The exit gases were bubbled through water in order to
`exert a slight positive pressure on the system. Fiberfrax
`was placed in the beginning of the hot zone of the furnace to
`heat the gases to the sample temperature.
`The procedure for preparing the silicon carbide as a -325
`+400-mesh powder was the same as that used in the previous
`work.6
`
`of Silicon Carbide
`
`259
`
`1 -
`
`W C .
`
`
`
` ’
`
`I -
`
`I
`
`’
`
`2 4 6 8 D L H l S M 2 0 2 2 2 4
`wb)r ma ant,.io4
`Fig. 1. Rate of oxidation at five different temperatures at
`various partial preuurer of water vapor. Each point represents
`an experimental determination of the diffusion constant of
`equation (1 ) measured in reciprocal seconds.
`
`The maximum value of R calculated from the thermogravi-
`metric data was approximately 3%.
`The curves of Fig. 1 are very similar to the curves obtained
`for the effects of the partial pressure of oxygen. A careful
`examination of the data pointed out that the rate of oxida-
`tion varied linearly with the logarithm of the partial pressure
`of water vapor. This indicated that water vapor acts simi-
`larly to oxygen in the oxidation of silicon carbide, and these
`results may be explained by the theory of thin-film oxidation
`proposed by Engell and HauffeO or by the theory that the
`free energy for adsorption changes with the amount of surface
`covered.’
`The theory of Engell and Hauffe describing the formation of
`thin oxide films involves the transport of cations through a
`film of thickness x to the surface and follows the parabolic
`law
`
`dx/dt = k / x
`in which the constant k is defined as
`n.vnA*k‘(ln PO, + In n4K’*)
`k = !!F
`4N
`v* =
`molar volume.
`Avogadro’s number.
`
`N =
`number of diffusing cations.
`nn =
`charge on cations.
`vn =
`jump distance.
`
`A
`=
`k’ =
`specific rate constant.
`partial pressure of oxygen.
`Por =
`number of, quasi-free electrons in system.
`
`n
`=
`K” =
`constant.
`This equation was developed on the basis of equilibrium
`according to the following equation :
`i s.o--
`j s + 02 + 2e-
`‘
`1
`
`= a surface site.
`I S
`I S.O-- = a surface site with an adsorbed oxygen ion.
`
`(2)
`
`(3)
`
`(4)
`
`111. Discussion
`The rate of oxidation was measured at five different tem-
`peratures at various partial pressures of water vapor. Figure
`1 shows the data obtained; each point on this figure represents
`the slope of a plot of the following equation which was found
`to be applicable to the oxidation of silicon carbide?
`
`R = fraction of reaction completed.
`A = jump distance of diffusing species.
`k‘ = specific rate constant.
`t = time.
`ro = initial radius of silicon carbide particles.
`Equation (1) is a modified form of th& parabolic rate equa-
`tion (x2 = kt), where x is the thickness of the oxide film.
`
`4 Hiroshige Suzuki, “Study on Oxidation of Silicon Carbide
`Powders: 11, Effects of Steam on Oxidation of Silicon Carbide
`of Various Colors and Crystal Structures,” Yogyo Kyokui Shi,
`67 [76l] 157-64 (1959); Cerum. Abstr., 1960, May, p. 113e.
`6 P. J. Jorgensen, M. E. Wadsworth, and I. B. Cutler, “Oxi-
`dation of Silicon Carbide,” J . Am. Cerum. Soc., 42 [12] 613-16
`(1959).
`
`6 ( a ) H. J. Engell and K. Hauffe, “Influence of Adsorption
`Phenomena on Oxidation of Metals at High Temperatures,”
`Metull,6 [11/12] 285-91 (1952):
`(b) Progress in Metal Physics, Bruce Chalmers (editor), Vol.
`Interscience Publishers, Inc .,,, New York, 1953.
`IV, p. 97.
`‘ ( a ) M. A. Cook and A. G. Oblad, Dynamic Mechanism
`of Heterogeneous Catalysis,” Id. Eng. Chem., 45 [7] 1456-61
`(1953).
`(b) Izumi Higuchi, Taikyue Ree, and Henry Eyring, “Ad-
`sorption Kinetics: I, System of Alkali Atoms on Tungsten,”
`J. Am. Chem. SOL, 77 [lo] 4969-75 (1955).
`
`GE-1020.002
`
`
`
`260
`
`Journal of The A w h n Ceramic Society-Jorgensen et al.
`
`Vol. 44, No. 6
`
`16
`n
`
`12
`
`-4
`
`i
`
`'
`
`'
`
`'
`
`I
`
`I
`~
`I
`~
`~
`C
`I
`5.8 5 9 6.0 6.1 6.2 63 6.4 6.5 6s 6.7
`.+ 1104
`Fig. 3. Eyring plot used to calculate an energy of
`activation according to equation ( I 2).
`
`,
`
`1
`
`1
`
`1
`
`1
`
`1
`
`1
`
`-3.5 -34 -a3 -3.2 -3l J1) -29 -28 -2.7 98-25
`LW-MPacar)
`
`1
`
`1
`
`
`
`Fig. 2. Data from Fig. 1 showing logarithmic dependence on
`the partial pressure of water vapor, where k represents the
`experimentally determined rate constant in reciprocal seconds.
`
`Let So be the total number of surface sites, No-- the number
`of surface sites with an oxygen ion adsorbed on the site, and
`n the number of electrons; writing the equation for the equilib-
`rium constant, one then obtains
`
`(5)
`
`In the case of watpr vapor, similar equations may be
`written, so that
`I S + H a + 2e-
`i S . O - - + HZ
`The equation for the equilibrium becomes
`
`(6)
`
`Thus in developing the Engell and Hauffe theory for the
`oxidation of thin films in the case of water vapor, one must
`assume that the logarithm of the pressure of the hydrogen
`gas produced is negligibly small with respect to the logarithm
`of the pressure of the water vapor. Small amounts of H2
`and C02 were found in a sample of exit gases by using a
`If, however, a steady state is assumed,
`mass spectrograph.
`one may write
`: S + H I O + ~ ~ - + ~ S . O - - + H Z
`
`( 8 )
`
`and
`
`j s.o-- 4 : s + 0--
`(9)
`If kl is the rate constant for equation (8), and k2 is the
`rate constant for equation (9), neglecting the reaction of
`hydrogen with the oxygen ion adsorbed on the surface site,
`one then has
`
`Equation (10) may be set equal to zero because of the
`assumption of the steady state, hence
`
`Thus by using equation ( i ) assuming SO >> NO-- or
`by using equation (ll), the Engell and Hauffe expression for
`the parabolic rate constant in the presence of water vapor
`becomes
`k = - nKvKX*kk' (In pH,o + In n*K")
`Ir,
`2N
`
`(12)
`
`(13)
`
`This equation may be written as
`k = A In pH,o + B
`A and B = constants.
`Thus a straight line should be obtained when plotting k
`This plot is shown in Fig. 2.
`versus In @E,o.
`The slopes
`of the lines shown in Fig. 2 contain the specific rate constant
`k I , and thus an activation energy may be obtained by plotting
`In (slope/T) versus 1/T. The plot is shown in Fig. 3. The
`value of the activation energy for diffusion was 24.4 kcal.
`per mole.
`The data for the 1514OC. isotherm in Fig. 2 were extremely
`difficult to obtain; therefore a dashed line is drawn for this
`isotherm. The slope of this line was obtained by extrapolat-
`ing the data of Fig. 3.
`Almost the same results as those from the Engell and Hauffe
`theory may be derived on the basis that the free energy for
`the adsorption process depends on the surface coverage as
`follows :
`A F = AFo + aB
`(14)
`where a is a proportionality constant with the units necessary
`to convert 0, the fraction of the surface covered, to units of
`free energy.
`This theory was derived in detail previously* for oxygen;
`for water vapor, the parabolic rate constant becomes
`
`& and KO = constants.
`X = jump distance of diffusing ions.
`A = surface area.
`R = gasconstant.
`The slope of equation (15) contains Tp; hence a plot of
`In (slope/T2) versus 1/T allows the activation energy to be
`calculated. The plot is shown in Fig. 4, and the value of
`the activation energy according to this model was 21.1
`kcal. per mole. Equation (15) can be used to obtain the
`value of the energy of adsorption of the water vapor on the
`surface sites. The intercept of a plot of equation (15)
`divided by the slope gives the following relation:
`Intercept/slope = In KO
`
`The energy of adsorption therefore may be obtained by
`plotting the intercept/slope versus 1/T as shown in Fig. 5.
`Note that the slope of the line is negative; hence a positive
`
`GE-1020.003
`
`
`
`June 1961
`
`t
`#- -ms .
`5
`4 -11.1
`-11.2 1
`
`-la7
`-IQ8
`-m ’
`
`-11.0 ’
`
`’
`
`Effects of Water Vapor on Oxidation of Silicon Carbide
`
`261
`
`I 0
`
`5.7 5B 59 6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8
`!+ 104
`Fig. 4. Eyring plot used to calculate an energy of activation
`according to equation ( I 5).
`
`Fig. 6.
`Cornpariaon of the rates of oxidation of silicon carbide in
`the presence of water vapor and in the presence of oxygen ot
`1398OC.
`
`value of 19.2 kcal. per mole was obtained for the energy
`of adsorption on the silica film. The positive enthalpy of
`adsorption is very unlikely and casts much doubt on this
`model as an explanation for the oxidation of silicon carbide.
`In Fig. 5, the point corresponding to the temperature of
`1218OC. does not fall on the line, which indicates a change
`in the system. Samples from each of the isotherms were
`analyzed using an X-ray diffractometer ; the samples oxidized
`at 1218OC. were found to be the cristobalite form of silicon
`dioxide, whereas samples from the other isotherms oxidized
`to tridymite.
`The main problem involved in studying a diffusion problem
`is to Sdentify the diffusing species. The diffusing species
`in the case of the oxidation of silicon carbide may be 02,
`CO, or C& as suggested by Suzuki‘ or may be oxygen ions
`diffusing in, carbon ions diffusing out, or silicon ions diffusing
`out. Diffusion of 02, CO, and C& was ruled out on the
`basis that the activation energy obtained in this study was
`too great for molecular diffusion. The diffusion of oxygen
`atoms was eliminated because the kinetic data did not
`correspond to the correct partial pressure of oxygen de-
`pendence. Carbon ions do not seem to be the diffusing
`species when the value of the energy of activation for the
`is compared with that
`oxidation of silicon, 45 kcal. per mole:
`of silicon carbide, 48.8 kcal. per mole. The diffusing species
`
`’
`’
`‘
`01
`5.8 5 9 60 6.1 6.2 6.3 6.4 6.5 6.6 6.7
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`1
`
`1
`
`
`
`therefore may be oxygen ions or silicon ions and more data
`are necessary to distinguish between them.
`The difference in the activation energy between silicon
`carbide oxidized in partial pressures of water vapor and
`silicon carbide oxidized in partial pressures of oxygen may be
`attributed either to a change in the film from an amorphous
`phase to a crystalline phase or to a change in the diffusing
`species. An experiment therefore was conducted to compare
`the rate of oxidation of Sic in water vapor with the rate of its
`oxidation in oxygen when a crystalline film was present.
`This was accomplished by first oxidizing the silicon carbide
`in water vapor to produce a tridymite film and then changing
`to oxygen at the same partial pressure as the water vapor.
`A plot of the data obtained is shown in Fig. 6. The circles
`indicate the data obtained in water vapor and the triangles
`indicate the data obtained in oxygen. Note that the rate,
`which is the slope of the line, is constant. It is most likely,
`therefore, that the diffusing species is the same in the presence
`of water vapor and in the presence of oxygen, and the differ-
`ence in the rate of oxidation in water vapor and in partial
`pressures of oxygen is due to the change in the nature of the
`oxide film. This may possibly be accounted for by a change
`from bulk diffusion to grain-boundary diffusion.
`
`IV. Conclusions
`The following conclusions may be drawn from the data
`presented in this paper: (1) The rate of oxidation of silicon
`(2) The
`carbide depends on the water-vapor pressure.
`silicon dioxide film in a water-vapor atmosphere is either
`tridymite or cristobalite. (3) The mechanism of thin-film
`oxidation proposed by Engell and Hauffe correlates the
`kinetic data for the oxidation of silicon carbide in the presence
`(4) The model of change in free energy
`of water vapor.
`with the amount of surface covered does not fit the oxidation
`of Sic in water vapor and tends to discount this model for
`the oxidation of Sic in the presence of partial pressures of
`(5) The diffusing species is the same in the presence
`oxygen.
`of partial pressures of water vapor and in the presence of
`partial pressures of oxygen.
`
`Fig. 5. Ratio of the intercept/slope of equation (1 5) vs. 1 / T
`used in calculating a value of the energy of adsorption of
`water vapor.
`
`* (a) J. T. Law, “High-Temperature Oxidation of Silicon,”
`J. Phys. Chem., 61 [9] 1200-1205 (1957).
`( b ) M. B. Brodskv and D. D. Cubicciotti. “Oxidation of Sdi-
`con at High Temperkres.” J. Am. Chem. Soc., 73 [7] 3497-99
`(1951).
`
`GE-1020.004