Reactors.

Kinetics and Catalvsis

Kinetics of Carbothermal Reduction Synthesis of

Beta Silicon Carbide
Alan W. Weimer, Kevin J. Nilsen, Gene A. Cochran, and Raymond P. Roach
Ceramics and Advanced Materials Research, Dow Chemical U.S.A., Midland, MI 48667

Carbothermal reduction kinetics to synthesize S i c is studied under conditions of

high carbon/silica precursor heating rates (Id K / s ) and minimized reaction times
(s) over a wide temperature range (1,8485 T52,273 K ) . The reaction mechanism
includes rapid formation of a gaseous SiO intermediate. Further carbon reduction
of the SiO to S i c is reaction-rate-controlling.Carbon crystallite diameter, d, has a
substantial influenceon the rate of reaction and thesize of synthesized Sic. Fractional
oxide conversion, X , can be described by a contracting volumeshrinking core model:

I - ( I -X)” k,
k= =- exp( - E / R T )
t d

where ko=27.4 m / s and E=382&34 kJ/mol.

Background
Thermochemistry
Beta silicon carbide (Sic) is manufactured by the carbo- la. Calculations for these free energy changes have been made
thermal reduction of silica (SiO,) according to the overall re- using the FACT (Thompson et al., 1985) database.
action: The overall carbothermal reduction reaction to synthesize
Sic, reaction 1, has a positive standard free energy change
3 C(s) + SiO,(s) =) SiC(s) + 2 CO(g) (1) (Figure la) up to about 1,793 K; so, unless carbon monoxide
(CO) produced is removed from the process, a higher tem-
Reaction 1 is highly endothermic, requiring approximately 572 perature is needed to promote reaction at a reasonable rate.
kJ/mol S i c at 2,100 K. The inhibiting effect of CO on kinetics at lower temperatures
The generally accepted reaction mechanism is that S i c is has been reported to be significant (Lee and Cutler, 1975;
synthesized through intermediate silicon monoxide (SO) gas. Khalafalla and Haas, 1972).
Based on this scenario, overall reaction 1 can be divided into The generation of SiO via reaction 2 requires temperatures
two elementary reaction steps: above approximately 2,025 K to proceed under equilibrium
conditions at atmospheric pressure. Since reaction 3 is favor-
C(s) + Si02(s)* SiO(g) + CO(g) (2) able over the entire temperature range, a reduced CO partial
pressure, pco, allows overall reaction (Eq. 1) to occur at lower
2 C(s) + SiO(g)=) SiC(s) + CO(g) (3)
temperatures by improving the favorability of SiO generation
via reaction 2.
The variation of free energy with temperature for reactions 1 The generation of SiO can result directly via solid-solid re-
to 3 at atmospheric pressure over the temperature range of action (Eq. 2), S O 2dissociation (Klinger et al., 1966; Lee and
interest for synthesis (1,500 IT r 2,500 K) is shown in Figure Cutler, 1975):

SiO,(s)= SiO(g) + 1/2 O2 (4)

Correspondence concerning this article should be addressed to A. W. Weimer, 52 Building,
Dow Chemical U.S.A.. Midland, MI 48667.

AIChE Journal March 1993 Vol. 39, No. 3 493

 200 r 500 r

i ;iil-.- --.- (iiT-*-*-.-,-

-*-,-,
-.-.-,_
-200 1 \ -500
-600
'-.I.-
'-.-.-.-.,

-250 1500 1700 1900 2100 2300 2500

1500 1700 1900 2100 2300 2500
Temperature (K) Temperature (K)
Figure lc. Free energy change for reactions 3 and
Figure la. Free energy change for Reactions 1,2 and 3.
8-11.

or gas-phase carbon reduction (Lee et al., 1976):

SiO(g) + 3 CO(g) * SiC(s) + 2 CO,(g) (10)
SiO,(s) + CO(g) * SiO(g) + CO,(g) (6)
2 C(S) + 2 CO,(g) * 4 CO(g) (11)

Free energy considerations regarding the synthesis of S i c from

The free energy change of SiO generation via a solid-solid SiO are summarized in Figure lc for reactions 3 and 8 to 11.
mechanism (reaction 2), SiOz dissociation (reactions 4 and 5 ) , Gas-phase S i c synthesis via CO and SiO (reaction 10) is not
and gas-phase carbon reduction (reactions 6 and 7) is shown favorable over the entire temperature range of interest
in Figure lb. Clearly, free energy considerations indicate that (1,5005 Ts2,500K). It is, however, thermodynamically fa-
SiO is most likely generated by solid-solid reaction (Eq. 2). vorable for approximate temperatures of T < 1,200 K. This
However, in the absence of direct carbon and silica contact, explains why several investigators (Kurosawa et al., 1966; Vis-
SiO could be generated by either reaction 4 or 6. Gas-phase comi and Himmel, 1978) found S i c product downstream of
carbon reduction via reaction 6 is favored on the basis of both the reaction hot zone in the cooler parts of the system. Al-
free energy and enthalpy considerations. though reactions 8 and 9 are both favorable over the entire
Likewise, SIC may result from the direct carbon reduction temperature range of interest, neither is as favorable as the
of SiO (reaction 3), a silicon (Si) metal intermediate: direct carbon reduction of SiO (reaction 3). In addition, PSic3
is only about 1.1 Pa at 2,038 K (Klinger et al., 1966) indicating
C(s) + SiO(g) * Si(s,l,g) CO(g) + (8) that Si(g) should not constitute a significant gaseous inter-
mediate to overall reaction 1.
As indicated, overall reaction 1 may be conveniently de-
+
Si(s,l,g) C(s)* SiC(s) (9)
scribed by consecutive reactions (Eqs. 2 and 3), each of which
may proceed via more complex mechanisms as discussed.
or gas-phase carbon reduction (Kurosawa et al., 1966; Viscomi
and Himmel, 1978):
Kinetics
A number of researchers (Blumenthal et al., 1966; Hender-
500 r son and Tant, 1983; Kennedy and North, 1983; Kevorkijan et
al., 1989, 1992; Khalafalla and Haas, 1972; Klinger et al.,
400
1966; Krstic, 1992; Kurosawa et al., 1966; Kuznetsova et al.,
-3
h

a,
300 1980; Lee et al., 1976; Lee and Cutler, 1975; Miller et al.,
p 200 1979; Ono and Kurachi, 1991; Papin et al., 1972; Shimoo et
m al., 1990; van Dijen and Metselaar, 1991; Viscomi and Himmel,
6 100 1978; and Wei et al., 1984) have investigated the reaction
h
P O kinetics of the overall carbothermal reduction reaction (Eq.
2
w -100 1). In general, it was found that higher temperature, finer
a, precursor particle size, and the removal of CO increased the
-200 -.'.*.'.--*-........
.........* ................................................
a,
t (5) reaction rate.
-300 - There was some controversy as to whether or not iron cat-
-400 alyzed the reaction. Reported activation energies were in the
1500 1700 1900 2100 2300 2500 range of 230 to 552 kJ/mol, depending primarily on the pre-
Temperature (K) cursor silica and carbon sources and particle sizes.
Much controversy exists regarding the exact mechanism and
Figure lb. Free energy change for Reactions 2 and
4-6. controlling step in the process. Lee et al. (19761, Lee and Cutler

AIChE Journal
March 1993 Vol. 39, NO. 3
494
(1975), Kennedy and North (1983), and Krstic (1992) found Table 1. Chemical and Physical Analysis of Carbon and
reaction 1 to be controlled by the generation of SiO, reaction Silica
2, while Viscomi and Himmel (1978) reported the reduction ~

Carbon Source Silica Source

of SiO to Sic, reaction 3, to be rate-controlling. Several re-
searchers (Klinger et al., 1966; Ono and Kurachi, 1991) have Element’ c1 c2 s1 s2
reported nucleation growth to be of importance, particularly A1 < 10 < 10 760 640
for temperatures near or above the transformation temperature Ca < 10 < 10 < 10 290
of quartz to cristobalite (Klinger et al., 1966). Carbon diffusion C1 < 10 < 10 < 10 42
Fe 12 7 123 360
effects have been reported to be important in several studies Na < 200 < 200 0.19% <500
(Ono and Kurachi, 1991; van Dijen and Metselaar, 1991). Van S < 10 < 10 470 25
Dijen and Metselaar (1991) have recently rejected an SiO in- Ti < 10 < 10 170 79
termediate and suggested that overall reaction (Eq. 1) occurs Zr - - 83 < 10
-
through direct solid state diffusion, in which SiOz migrates c (wt. 070) 98.6 98.9 - -
over the Sic surface to the carbon. Wei et al. (1984) cited heat- 0 (wt. 070) 0.3 0.3 - -
transfer effects as being rate-controlling. -
Product S i c crystal size is reportedly governed by the starting d (ccm) 0.030 0.218 0.022 1.1
carbon crystal size, reaction temperature, and time at tem- S.A. (m2/g) 72 8 - 2.2
~~ ~

perature (Kennedy and North, 1983; Kevorkijan et al., 1989, ‘ppm unless otherwise noted.
1992; Klinger et al., 1966; and Wei et al., 1984). Klinger et al.
(1966) noted that product S i c particles could not be distin-
guished from the starting graphite particles, while Kennedy the silica source was either a fine colloidal (Sl) or milled 2.2
and North (1983) and Kevorkijan et al. (1989,1992) noted that m2/g naturally crystalline (S2) powder. As indicated in Table
Sic surface area decreased with higher reaction temperatures 1, both carbon sources were of ultra high chemical purity,
and longer reaction times, but that product S i c morphology while both silica sources contained similar impurity levels. By
resembled that of the starting carbon. However, Kevorkijan choice of these starting materials it was possible to determine
et al. (1989, 1992) also found that Sic crystallites synthesized the effect of reactant carbon and silica size on reaction kinetics
at low temperature (1,443 K) corresponded to the size of the and S i c morphology.
starting carbon crystallites. An aqueous slip of carbon black was made using alkyl phen-
In general, the reported kinetic studies for the overall car- oxy polyethoxy ethanol as a wetting agent. The silica source
bothermal reduction reaction (Eq. 1) were carried out with was added to the carbon slip for a 50 wt. 070 solids loading,
larger samples in slowly heated systems (< 100 K/s) at lower and the combined slip was spray-dried. Colloidal silica was
temperatures (typically below 1,873 K) for extended reaction used as the binder for all systems. Product collected from the
times (minutes to hours). In light of the inability to heat samples spray dryer was calcined at 673 K for 12 hours in a nitrogen-
rapidly in previous studies, the reported intrinsic kinetics have purged tray oven. The calcination step removed hydrated water
been limited primarily to lower temperatures where heat-trans- from the silica. The spray-dried and calcined “microcontain-
fer resistances did not dominate. Heat-transfer resistances are er” C/Si02 precursor particles had an average volume median
of particular concern when carrying out intrinsically fast, highly diameter of approximately 60 microns. Silica was insured as
endothermic reactions such as overall reaction (Eq. 1). the limiting reactant by maintaining a molar ratio of
It is the intent of this article to better quantify the intrinsic C/Si02 = 3.1 according to overall equation (Eq. 1).
kinetics of overall reaction (Eq. 1) and to obtain a better
understanding of the reaction mechanism and the rate-limiting Reaction
step. Of particular interest is whether or not overall reaction
Reaction was carried out in a rapid heating mode using a
(Eq. 1) proceeds through gaseous SiO and, if it does, whether
dilute transport vertical reactor process for reaction times t < 5
reaction 2 or 3 is rate-limiting.
s at temperatures in the range of 1,848 K S T ~ 2 , 5 7 3K. The
Here, kinetic studies are carried out in such a manner that
spray-dried precursor was entrained with argon gas through a
heat-transfer resistances are effectively eliminated. Fine “mi-
cooled transport tube which extended directly into the upper
crocontainers” of reactant C/Si02 precursor are rapidly heated
reaction zone of the heated graphite flow reactor. Reaction
within milliseconds, held at reaction temperature for only sev-
residence time was governed by the argon flow rate, the CO
eral seconds, and immediately quenched. This is the first fun-
generated by reaction, the tube temperature, and the volume
damental kinetics study for carbothermal reduction S i c
of the reaction zone. The use of dilute flow vertical reactors
synthesis to be carried out under such conditions. Intrinsic
for studying heterogeneous reactions has recently been dis-
kinetics can be determined for higher reaction temperatures.
cussed (Bjerle et al., 1992).
As fine cooled precursor particles were injected directly into
Experimental Procedure the hot reaction zone, they were heated rapidly at the surface
by radiation and gas conduction. The surface heating rate and
Preparation of reactants internal particle conduction are highly dependent on particle
An intimately mixed C/Si02 precursor was prepared by spray size as well as the temperature and physical properties of the
drying and calcining an aqueous carbon black/silica slip. Car- solid precursor and surrounding gas. Since the flow rate of
bon and silica sources for the precursor are summarized in cooled argon gas carrying the precursor particles directly into
Table 1. The carbon source was either a fine (Cl) 72 m2/g or the reaction zone was small relative to the total flow rate of
slightly larger (C2) 8 m2/g sized high-purity gas black, while hot gas through the reaction zone, a reasonable assumption is

AIChE Journal March 1993 Vol. 39, NO.3

495
 Analytical
Product S i c powders were analyzed chemically for carbon
and oxygen using a LECO IR-412 Carbon Determinator and
LECO TC-436 Oxygen Determinator , respectively. For carbon
analysis using the LECO Carbon Determinator, sample ma-
,*+*’
Gas Phase
Conduction +* ,.”’ terial is combusted with released carbon monoxide converted
** ,/ to carbon dioxide in a rare earth copper oxide catalytic heater.
.** ‘., <.0*
*** Carbon is then measured as carbon dioxide by IR analysis.
,***
$ . , ,
,./ ,0”
Internal
Particle
For oxygen analysis using a LECO Oxygen Determinator, the
**** Conduction sample is pyrolyzed and released oxygen combined with carbon
to form carbon monoxide. The carbon monoxide is converted
to carbon dioxide as described above and the oxygen is meas-
/I
,/-
I 111111 I I Illll I I I Ill
ured by IR in the form of carbon dioxide.
Crystal-phase purity was determined by X-ray diffraction
10 10‘
Particle Radius (m) x lo6 (Rigaku “Miniflex” X-ray diffractometer). Product mor-
phology was identified by scanning electron microscopy (IS1
Figure 2. Precursor particle characteristicheating times. model SS40), and surface area was determined by the BET
nitrogen adsorption method using standard manufacturer’s
procedures (Quantochrome Corp. Autosorb-1). Trace metal
that cooled particles instantly contact surrounding gas which impurities were quantified by X-ray fluorescence.
is maintained at reactor wall temperature. Silica (or oxygen) conversion in a reacted product sample
The characteristic heating times for spherical particles of was calculated as follows. Product was analyzed for total car-
radii between 1 and 1,000 microns are shown in Figure 2 for bon and total oxygen wt. %. The remaining sample weight
radiation, gas-phase conduction, and internal particle con- was assumed to be silicon. Having the total silicon and a;-
duction for particles increasing in temperature from ambient suming that all of the oxygen was in. the form of S O z , the
to 2,273 K. The method of calculation has been described amount of Si as S i c and subsequently the amount of S i c were
previously (Weimer and Clough, 1980) and is modified here calculated. The amount of SiOz reacted to yield the calculated
to account for different physical properties. It is clear that the amount of S i c could then be determined. Fractional silica (or
dominant mechanism of heating is gas-phase conduction with 0) conversion was then easily calculated. This method for
radiation playing a minor, but increasing, role as particles calculating oxygen conversion was suitable because unreacted
increase in size. Also, internal particle conduction is almost volatile oxide condensed by thermophoresis on cooled product
two orders of magnitude faster than surface heat transfer. powder exiting the reaction zone. A sample calculation is shown
Hence, particles are heated uniformly throughout and may be in the Appendix.
assumed to be isothermal.
Estimated particle heating rates have been calculated (Figure
3) using the dominant gas-phase conduction characteristic Results and Discussion
heating times (Figure 2). It can be seen that a characteristic Product exiting the transport reactor was collected in an in-
heating rate of approximately lo5 K/s is estimated for 60 mi- line sampling device downstream of the reactor cooling zone
cron average diameter precursor particles. This heating rate is and analyzed as described previously.
consistent with calculations made by Celik et al. (1990) for 20 Porous product particles (Figure 4b) maintained the spher-
to 100 micron diameter coal particles entering an entrained ical shape and size of the starting precursor microcontainers
flow coal devolatilization reactor. (Figure 4a) and contained synthesized S i c crystallites through-
out. Porosity in a completely reacted precursor microcontainer
particle was the result of all of the oxygen and 2/3 of the
starting carbon being removed as gaseous CO according to
overall reaction (Eq. 1). A realistic macroscopic scenario of
the reacting microcontainer particles appears to be one in which
the particles remain a constant size, but change in density as
the reaction occurs uniformly throughout.
Raw product was then calcined in air to remove residual
carbon, ultrasonically deagglomerated (Figure 4c), and acid
treated by conventional methods (Shaffer et al., 1987) to re-
move residual oxygen and metal impurities.
Experiments were carried out over a wide temperature range
to determine the effect of reaction parameters on crystallite
size, the reaction mechanism, and the intrinsic reaction rate.

Effect of reaction parameters on S i c crystallite size

Particle Radius (m) x 1O6 Photomicrographs (SEM) of product S i c (Figure 5 ) syn-
Figure 3. Precursor particle characteristic heating rate. thesized at temperatures between 2,073 and 2,573 K clearly

496 March 1993 Vol. 39. No. 3 AIChE Journal

Figure 4. Spray dried precursor (Cl/Sl)and product Sic.
a. Precursor (CI/SI) microcontainers.
b. Product SIC particles synthesized at 2,073 K .
c . Product SIC particles after 10 min in ultrasonic bath.

indicate that S i c crystallite size increased with an increase in

reaction temperature. All crystallites synthesized at tempera-
tures below 2,573 K were less than 0.5 micron in size.
Crystallites synthesized at 2,073 K appear to be in the 0.1
to 0.2 micron size range, while those synthesized at 2,473 K
appear to be in the 0.3 to 0.4 micron size range. The 2,573 K
product i s sintered (Figure 5d), indicative of reaction temper- Figure 5. Product Sic particles synthesized from (C11
ature exceeding the sintering temperature of Sic. S1) precursor.
Crystallite size was quantified through extensive TEM char- a. 2,073 K; b. 2,173 K; c. 2,473 K; d. 2,573 K.

AlChE Journal March 1993 Vol. 39, No. 3 491

 V.""

2000 2100 2200 2300 2400 2500

Reaction Temperature, K
Figure 6. Effect of temperature on mean Sic crystallite
size synthesized from (ClISl) precursor.

acterization and subsequent crystal size distribution analysis.

Martin's diameters (Martin et al., 1924) were measured from
randomly obtained TEM's and statistically analyzed to quan-
tify the mode, median, mean and standard deviation. The mean
crystallite size varies with temperature as shown in Figure 6.
The crystaliite-size distributions for these powders are very
narrow as shown in Figure 7 for the 2,173 K synthesized prod-
uct which has a coefficient of variation of s,=O.34.
Starting carbon crystallite size also has a substantial influ-
ence on the resulting S i c crystallite size as shown by the pho-
tomicrographs of Figure 8. Here, the product Sic synthesized
at T = 2,323 K (Figure 8b) from the C2/S2 precursor (Figure
8a) comprises crystallites which resemble the starting C2 carbon
(Figure 8a).
Also, a comparison of the photomicrographs of the C2 car-
bon derived S i c product at T=2,323 K (Figure 8b) with the
Figure 8. (C21S2) precursor and product Sic.
Cl carbon derived S i c product at T=2,173 K (Figure 5b) and a. (C2/S2) precursor.
T = 2,473 K (Figure 5c) further substantiates the conclusion b. SIC Product ( T = 2,323 K)
that Sic crystallite size resembles the starting carbon size. Here,
the C1 carbon derived S i c (Figures 5a,b,c) is finer than the
C2 derived S i c (Figure 8b) at comparable temperatures. These A realistic physical scenario of chemical reaction within the
results are consistent with reported observations of Kennedy microcontainer particles is one in which the initial S i c crys-
and North (1983), Kevorkijan et al. (1989, 1992), Klinger et tallites resemble the starting carbon crystallites and then, once
al. (1966), and Wei et al. (1984). formed, grow by a sintering process to larger crystallites. Finer
Starting silica size appears to have no influence, as the sub- carbon crystallites such as C1 are more reactive (to be dis-
micron-sized S i c crystallites (Figure 8b) show no resemblance cussed) and may allow grain growth to occur at lower tem-
to the angular 1.1 micron mean sized S2 silica reactant (Figure peratures (since S i c has been synthesized) relative to larger
8a and Table 1). carbon crystallites such as C2. It appears that S i c grain growth
does not substantially occur as long as carbon is available for
reaction to Sic. This indicates that either the carbothermal
8
20.0 I Martin's Diameter, wm
reduction reaction kinetics to S i c is much faster than S i c grain
growth or that free carbon inhibits S i c grain growth.
s - Average: 0.085
$ 15.0 Std Dev: 0.03
a
3 Median: 0.078
D - Mode: 0.073 Reaction mechanism
? 10.0 Count: 668
L Photomicrographs (Figure 9) and X-ray diffraction studies
(Figure 10) of partially reacted particles exiting the reactor
cooling zone suggest that reaction occurs throughout the pre-
cursor microcontainer particles and that, at low silica (or ox-
u.u ygen) conversion, volatile, unreacted oxides condense as a
0.005 0.055 0.105 0.155 0.205 0.255 0.305 fibrous species on the exiting microcontainer particles. The
Martin's Diameter pm condensation most likely occurs as a thermophoresis phenom-
Figure 7. Martin's diameter SIC crystallite size distri- enon during rapid cooling of the incompletely reacted particles
bution. upon entering the cooling zone. The extent of this condensation
7=2,173 K; Cl/SI precursor. is shown in Figure 9 at various magnifications for partially

498 March 1993 Vol. 39, No. 3 AIChE Journal

 X-Ray DinractorneterScan

0-SIC
I s i
T = 1898K i,
I xi022
I u
80 68 56 44 32 20
Angle

(a)
X-Ray DinractometerScan
.. __ -
0.d T = 1948 K
X=052

0-SIC
1 "n"' SI

I
68 56 44 32

Figure 10. X-ray diffraction pattern for partially con-

verted raw SIC powders (from ClIS1 precur-
sor).
a. T=1,898 K; X=O.22.
b. T = 1,948 K; X = 0 . 5 2 .

portionation of gaseous SiO on cooling according to the re-

action:

2 SiO(g)=) Si02(s)+ Si(s,l) (12)

Free energy change considerations for reaction 12 indicate that

Si(s,l) formation is favorable at temperatures below approx-
imately 2,132 K.
Figure 9. Partially reacted raw product SIC powders from These results provide substantial evidence that overall re-
(ClISl) precursor. action (Eq. 1) occurs via a gaseous SiO intermediate and that
T = 1,898 K; X = 0 . 2 2 . the formation of SiO(g) is fast. Initially, SiO(g) is most likely
a. 2 , 5 0 0 ~magnification; b. 3 , 0 0 0 ~magnification; c. 10,OOOx formed rapidly according to reaction 2 when carbon and silica
magnification.
particles are in contact. The high rate of SiO(g) synthesis sug-
gests that the CO produced by reaction 2 reacts with SiOz to
converted ( X = 0.22) microcontainer particles reacted at 1,898 synthesize SiO(g) according to reaction 6. The C 0 2 which is
K for approximately 4 s. formed then reacts with carbon (reaction 7) to synthesize CO,
X-ray diffraction patterns (Figures IOa,b) for low conver- which, in turn, reacts with silica according to reaction 6 to
sion/temperature products indicate the presence of crystalline continue the cycle. Reaction 2 is feasible only at points of
silicon (Si) metal. Free Si metal was not detected in the X-ray direct contact between C and SOz, and hence takes place
diffraction patterns of highly converted products and is as- during the initial period only. After the direct contact points
sociated with the fibrous condensed species as determined by are consumed, reaction 2 most likely ceases. Hence, reactions
X-ray microprobes. 6 and 7 are the ones which make reaction 3 possible.
It is believed that the Si was most likely formed by dispro- Experiments were carried out to determine the effect of

AIChE Journal March 1993 Vol. 39, No. 3 499

 powders synthesized from the Cl/Sl and C2/S2 precursors
X over the entire temperature range clearly indicates that larger-
sized carbon crystallites are substantially less reactive. At
T = 2,173 K, the C2 containing precursor only achieves a silica
conversion of X=0.63, while the C1 containing precursor
achieves a much higher X=0.91 conversion level. This dif-
ference is reactivity holds over the entire temperature range,
although the degree of difference decreases with increasing
temperature. High silica conversions can be achieved for the
0.6 I I I I I I larger sized carbon crystallite containing precursors at higher
2200 2300 2400 2500 temperatures (that is, X>0.90 for T>2,473 K).
Temperature, K In summary, rapid reaction rates which complete overall
reaction (Eq. 1) in seconds, substantial quantities of unreacted
Figure 11. Effect of carbon and silica crystallite size on
SiO(g) exiting the reactor under more mild conditions, the
reaction kinetics.
increase in reaction rate with finer carbon crystallite size, the
Equivalent gas/solid feed rates & precursor composition.
0 (CI/SI) precursor; (CI/S2) precursor; (C2/S2) pre- independence of rate on silica size, and the fact that S i c crys-
cursor. tallites resemble the starting carbon crystallites suggest that
overall reaction (Eq. 1) most likely occurs through an SiO(g)
intermediate via reactions 2, 6, 7 and 3, and that reaction 3 is
carbon and silica crystallite size on the rate of reaction. Mi- rate-limiting.
crocontainer precursors of identical molar C/Si02 = 3. l com-
position were prepared using Cl/Sl, C1/S2, and C2/S2
reactants. These precursors were reacted under identical single Intrinsic reaction rate
pass rapid heating conditions (identical gadsolid feed rates at To quantify the intrinsic reaction rate, kinetic experiments
various temperatures) in order to quantify the effect of starting achieving a wide silica conversion level ( 0 . 1 4 ~ X ~ 0 . 8were
3)
carbon and silica crystallite size on reaction rate. The effect carried out over a wide temperature range (1,848 K i T i 2,273
of crystallite size on conversion at various temperatures (2,173 K) for various short residence times (2.5 s t 5 4.3 s).
K c Ts2,473 K) is shown in Figure 1 1 . In view of the fact that Sic crystal size and morphology
It appears, from a comparison of the Cl/SI and Cl/S2 initially (prior to grain growth) reflect that of the starting
results at T= 2,373 K (Figure 1 l), that starting silica crystallite carbon and the fact that SiO(g) is an intermediate in the re-
size has no effect on the rate of reaction. Here, an increase in action as discussed previously, a suitable and simple gas/solid
silica crystallite size of approximately 37 times results in single- kinetic model appears to be one in which reaction proceeds at
pass product of nearly equivalent, but unexpectedly higher, the surface of the carbon according to a shrinking core model
oxide conversion. (Yagi and Kunii, 1955). As the Sic product layer grows, the
Starting carbon crystallite size, however, appears to have a reaction surface decreases and leads to a decrease in the rate
significant influence on the reaction rate. A comparison of of reaction. A schematic of this mechanistic model for SIC
synthesis according to reactions 2 , 3 , 6 and 7 is shown in Figure
12.
For an irreversible phase boundary controlled reaction of
spherical carbon particles for which the thickness of the re-
action zone is small compared with the dimensions of the
particle:

1 -(1 - X ) ” 3
k= = 5 exp ( - E/RT)
t d

-16I
-1 7
h -1 8
n
s.-19-k, = 27.35 mls
C
- -20-1848 < T < 2273 K
‘Sic 0.14 c X c 0.83
-21 -2.5 c t < 4.3 S

C + SiO, --+ SiO + CO -33

LL

CO + SiO, +SiO + CO2 4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6
c+co, + 2 c o 1 / l (K) x lo4
C + SiO --+SIC + CO Figure 13. Reaction kinetics for contracting volume
model with phase boundary control-Eq. 13.
Figure 12. Proposed reaction mechanism. 8 CI carbon: 0 C2 carbon.

500 March 1993 Vnl. -9-.,No.

. --- 3 . 3
- AIChE Journal
where Comparison to reported results
The reaction rates, k, and activation energy, E (382 k 3 4 kJ/
k =:- 4ks0csi0 (14)
mol), reported in this work are generally consistent with pre-
P C viously reported results as shown in Table 2. The broad range
of activation energies reported (251 IE I 552 kJ/mol) is most
for reaction 3 being rate-controlling. Since the surface con- likely due to the broad range of rate equation forms which
centration, Csio, is unknown, the preexponential factor, k,,, have been fit to the data. The differences in reaction rates are
cannot be determined. However, the activation energy, E, and consistent with differences in carbon source reactivity and crys-
rate parameter, k,, can be determined as shown in Figure 13 tallite size. For example, the lower rates reported by Klinger
where the low temperature/conversion kinetic results obtained et al. (1966) are expected for the reaction of large, low reac-
for the C1 and C2 containing precursors are linearly fit to Eq. tivity, graphite particles. In addition, petroleum coke generally
13. contains metal impurities which are known to increase reaction

Table 2. Comparative Reaction Rate and Activation Energy

Carbon
Reference Source d (am) T (K) k (s-l) E (kJ/mol)
Blumenthal et al. (1966) carbon black 0.017 1,573 LOX 1 0 - ~ 287
1,673 5 . 7 I~O - ~
1,773 1.5 x lo-’
Khalafalla and Haas (1972) graphite - 1,673 1 . 4 lo-)
~ 322
1,698 I .7 x 10-3
1.723 1 . 9 1~ 0 - ~
1;788 2.8 x 10-3
Klinger et al. (1966) graphite <44 1,818 3.2~ lo-’ 510
1,863 4.5 x lo-’
1,913 1.7 x
1,943 2.3 x
1,983 4.5 x 10-6
2,038 1 . 4 1~0 - ~
Kuznetsova et al. (1980) carbon black 0.187 1,773 1 . 7 lo-’
~ 25 I
1,873 5.8 x lo-’
1,973 7.6~
2,073 2.1 x lo-’
carbon black 0.385 1,773 7.8~ lo-) 288
1,873 2.8 x
1,973 6.2 x lo-’
2,073 1 . 3 10-1
~
coke 0.198 1,773 9.2 x 10-3 299
1,873 3.1 x lo-’
1,973 6.2 x l o T 2
2,073 1 . 6 lo-’
~
Lee and Cutler (1975) charcoal 70 1,623 5 . o 1~ 0 - ~ 544
1,643 L O X 10-3
1,673 1.5 x lo-’
1,703 3 . 9 1~ 0 - ~
1,713 4 . o ~lo-’
Ono and Kurachi (1991) pyrolyzed polymer - 1,795 3 . 2 lo-’
~ 391
1,813 3.5 x 10-3
1,821 5.5 x 1 0 - ~
1,881 1.2x lo-’
1,962 2.7 x lo-’
Shimoo et al. (1990) pyrolyzed polymer - 1,773 1.5 x 352
1,873 2.7 x lo-’
1,973 4.5 x 10-2
Visromi and Himmel (1978) coke breeze - 1,673 1.1 x 552
1,723 1.4 x lo-’
1,773 2.5 x lo-’
This work carbon black 0.030 1,848 1.2x 10-2 382
1,898 2.0x 10-2
1,948 6.2 x
1,973 8.9x lo-’
1,998 1 . 3 lo-’
~
0.218 2.173 9 . 2 ~lo-’
2,273 1 . 5 lo-’
~

AIChE Journal March 1993 Vol. 39, No. 3 501

rates as shown by the data of Kuznetsova et al. (1980) and Susnitzky, Charles Wood, Harold Klassen and Don Beaman for their
Viscomi and Himmel(l978). Carbon from pyrolyzed polymers analytical support.
is generally very reactive as indicated by the rates reported by
Ono and Kurachi (1991) and Shimoo et al. (1990). Notation
The reaction rates estimated for previously reported carbon c,,, = concentration of SiO, mol/m3
blacks (Blumenthal et al., 1966; Kuznetsova et al., 1980), using d = carbon crystallite diameter, m
E = activation energy, kJ/mol
the rate equation derived in this work, are slightly lower than k = reaction rate, Eq. 13, s - ’
those reported by the previous investigators. This may be the k, = rate parameter, Eq. 13, m/s
result of differences in metal impurities within the carbon k,, = preexponential factor, m/s
blacks. The carbon sources used in this work are of ultra high s , = coefficient of variation
purity (Table 1). The purity of carbon in the previous studies t = time, s
T = temperature, K
was not reported. R = gas law constant, 8.314 J1rnol.K
The rates reported in this work include higher temperature X = silica (or oxygen) conversion
(up to T=2,273 K) reaction rates which have never been re- pc = carbon molar density, mol/m3
ported previously. The rapid heating of small samples allowed
heat-transfer resistances to be effectively eliminated and in- Literature Cited
trinsic kinetics to be determined at the higher temperatures. Bjerle, I., F. Xu, and 2. Ye, “Useful Experimental Technique for the
In addition, the effect of carbon crystallite diameter, d, is, for Study of Heterogeneous Reactions,” Chem. Eng. Technol., 15, 151
the first time, explicitly contained in the reaction rate expres- (1 992).
Blumenthal, J . L., M. J. Santy, and E. A. Burns, “Kinetic Studies
sion. of High-Temperature Carbon-Silica Reactions in Charred Silica-
Reinforced Phenolic Resins,” AIAA J., 4(6), 1053 (1966).
Conclusions Celik, I., T. J. O’Brien, and D. B. Godbole, “A Numerical Study of
Coal Devolatilization in an Entrained-Flow Reactor,” Chem. Eng.
Overall reaction (Eq. 1) is intrinsically fast and may be Sci., 45, 65 (1990).
completed in several seconds at temperatures in excess of ap- Henderson, J. B., and M. R. Tant, “A Study of the Kinetics of High-
Temperature Carbon-Silica Reactions in an Ablative Polymer Com-
proximately 2,100 K. The reaction appears to proceed through posite,” Poly. Composites, 4(4), 233 (1983).
a rapidly formed SiO(g) intermediate synthesized according to Kennedy, P., and B. North, “The Production of Fine Silicon Carbide
reactions 2, 6 and 7. Solid/solid reaction (Eq. 2) most likely Powder,” Fabrication Science 3, D. Taylor, ed., British Ceram.
occurs at the contact points of the carbon and silica. However, SOC.,Shelton, No. 33, 1 (1983).
the rapid reaction rates indicate that the gadsolid mechanism Kevorkijan, V., M. Komac, and D. Kolar, “The Influence of Prep-
aration Conditions on the Properties of Beta SIC Powders Synthe-
through reactions 6 and 7 is most important. sized by Carbothermic Reduction,” Ceram. Powd. Proc. Sci. Proc.
The presence of substantial unreacted SiO(g) exiting the Int. Conf., 327 (1989).
reactor indicates that the generation of SiO(g) is fast and that Kevorkijan, V., M. Komac, and D. Kolar, “Low-Temperature Syn-
the overall reaction is controlled by the reduction of SiO(g) to thesis of Sinterable Sic Powders by Carbothermal Reduction of
Colloidal SiOZ,’’J. Mat. Sci., 27(10), 2705 (1992).
synthesize S i c (reaction 3). Further evidence for reaction 3
Khalafalla, S. E., and L. A. Haas, “Kinetics of Carbothermal Re-
being rate controlling is provided by the fact that carbon crys- duction of Quartz Under Vacuum,” J. Amer. Ceram. SOC., 55,414
tallite size has a substantial effect on the rate of reaction while (1972).
the size of the starting silica is of no influence. Klinger, N., E. L. Strauss, and K. L. Komarek, “Reactions Between
Initially synthesized S i c crystallites resemble the starting Silica and Graphite,” J. Amer. Ceram. SOC., 49, 369 (1966).
Krstic, V. D., “Production of Fine, High-Purity Beta Silicon Carbide
carbon crystallites in size and morphology and, once formed, Powders,” J. Amer. Ceram. SOC.,75(1), 170 (1992).
grow by a sintering process to larger crystallites. Substantial Kurosawa, T., T. Kikuchi, and T. Yagihashi, “Carbothermic Reduc-
SIC grain growth does not appear to occur as long as sufficient tion of Silica,” Proc. Meml. Lect. Meet. Anniv. Found. Natl. Res.
carbon is available for reaction to Sic. Inst. Mer., National Research Institute of Metals, Nakameguro,
The macroscopic reaction of spray dried precursor micro- Japan, 114 (1966).
Kuznetsova, V. L., V. A. Dmitrenko, and A. D. Kokurin, “Kinetics
containers of starting carbon and silica may be described as of Formation of Silicon Carbide,” Proc. of the Mendeleev Chem.
spheres of constant size which change in density as reaction Soc., Zh. Vses. Khim. 0 - V A , 25(1), 1 18 (1980).
occurs uniformly throughout. Internally, a realistic kinetic Lee, J.-G., P. D. Miller, and 1. B. Cutler, “Carbothermal Reduction
model appears to be one in which reaction proceeds at the of Silica,” Reactivity of Solids, J. Wood, 0. Lindqvist, C. Helges-
son, and N.-G. Vannerberg, eds., Plenum, 707 (1976).
surface of the carbon. The reaction rate decreases as the grow- Lee, J.-G., and I. B. Cutler, “Formation of Silicon Carbide from Rice
ing S i c product layer decreases the reaction surface area and Hulls,” Ceramic Bulletin, 54, 195 (1975).
may be described by a shrinking core model: Martin, G., C. E. Blyth, and H. Tongue, “Researches on the Theory
of Fine Grinding: VII,” Trans. Ceram. SOC. ( E n g . ) ,23,61 (1924).
Miller, P. D., J.-G. Lee, and I. B. Cutler, “The Reduction of Silica
1 -(1 - X ) ” 3 --
- 27.4 exp( - 382,000/RT) (15) with Carbon and Silicon Carbide,” J. Amer. Ceram. Soc., 62, 147
k=
t d (1979).
Ono, K., and Y. Kurachi, “Kinetic Studies on Beta SIC Formation
The intrinsically fast reaction kinetics of highly endothermic from Homogeneous Precursors,” J. Mat. Sci., 26, 388 (1991).
Papin, G. G . , I. v. Ryabchikov, v. G. Mizin, G. v. Serov, and N.
overall reaction (Eq. 1) will force Slow heating rate s i c Pro- M. Dekhanov, “Reaction of Quartz with Carbon Reducing Agents,”
duction processes to be throughput-limited by heat transfer. Khimiya Tverdogo Topliva, No. 1, 101 (1972).
Shaffer, P. T. B., K. A. Blakely, and M. A. Janney, “Production of
Fine, High-Purity, Beta Sic Powder,” Ceramic Powder Science,
Acknowledgment G. L. Messing, K. S. Mazdiyasni, J . W. McCauley, and R. A. Haber,
The authors would like to thank Dow Chemical employees David eds.* Amer. Ceram. 257

502 March 1993 Vol. 39, No. 3 AIChE Journal

Shimoo, T., M. Sugimoto, and K. Okamura, “Synthesis of Silicon wt. Vo total Si=lW-wt. To C-wt. Vo 0
Carbide Powders from Organosilicon Polymers,” Funtai Oyobi
Funmalso Yakin, 37(11), 1132 (1990). =36.11
Thompson, W. T., A. D. Pelton, and C. W. Bale, Facility for the
Analysis of Chemical Thermodynamics [FACT], Thermfact Ltd.,
Mount-Royal. Quebec, Canada (1985). wt. 070 Si02= (60.09/32) x (wt. Vo 0)
Van Dijen, F. K., and R. Metselaar, “Chemical Reaction Engineering
Aspects of a Rotary Reactor for Carbothermal Synthesis of Sic,” = 37.24
J . Eur. Ceram. SOC., 5, 55 (1989).
Van Dijen, I:. K., and R. Metselaar, “The Chemistry of the Car- wt. Vo Si as SiO, = (28.09/60.09) x (wt. Vo SO2)
bothermal Synthesis of Beta Sic: Reaction Mechanism, Reaction
Rate and Grain Growth,” J. Europ. Ceram. SOC.,7 , 177 (1991). = 17.41
Viscorni, F., and L. Himmel, “Kinetic and Mechanistic Study on the
Formation of Silicon Carbide from Silica Flour and Coke Breeze,”
J. Metals, No. 6 , 21 (1978). wt. Vo Si as SiC=wt. 070 total Si-(wt. Vo Si as SiOz)
Wei, G. C., C. R. Kennedy, and L. A. Harris, “Synthesis of Sinterable
SIC Powders by Carbothermic Reduction of Gel-Derived Precursors = 18.7
and Pyrolysis of Polycarbosilane,” Ceram. Bulletin, 63,1054 (1984).
Weimer, A. W., and D. E. Clough, “Modeling of Char Particle Size/ wt. Vo SiC=(40.1/28.09)x(wt. Vo Si as Sic)
Conversion Distributions in a Fluidized Bed Gasifier: Non-Isother-
ma1 Effects,” Powd. Technol., 27, 85 (1980). = 26.7
Yagi, S., and D. Kunii, Proc. Int. Symp. Combustion,Van Nostrand
Reinhold, New York, 231 (195s).
wt. Vo Si02 required to yield Sic = (60.09/40.1)
~ ( w t Vo
. Sic)

Appendix: Example Calculation of Silica (or Oxy = 40.01

gen) Conversion
wt. Vo Initial SiO,=(wt. Vo SiO,)+(wt. Vo SiO,
Assumptions. Product composition is Sic, SiO,, and free C
only. All unreacted volatile oxides deposit by thermophoresis required to yield SIC)
on the raw product on entering the cooling zone (see Figure
= 77.25
9) and are accounted for as SOz. The amount of free Si metal
is negligible (crystalline Si metal strongly diffracts and even
minute quantities [ <OSVo] have relatively large X-ray dif- Vo Si02 Conversion = 100x (wt. Vo Initial Si02
fraction peak intensities, Figure 10). -wt. Vo SiO,)/(wt. Vo Initial SO,)
=51.79
Analytical composition of raw product: 44.06 wt. Vo total C
19.83 wt. Vo total 0 Manuscript received Mar. 27, 1992, and reuision received Sept. 2, 1992.

AIChE Journal March 1993 Vol. 39, No. 3 503