Lrrh,
I!J74. Vol
IO. pp Yi-11 I
Pergamon
Pless
CATALYTIC
Pruned
m Great
Bntam
OXIDATION
OF CARBON
E. T. TURKDOG AN and J. V. VINTERS Edgar C. Bain Laboratory For Fundamental Research, United States Steel Corporation, Research Center, Monroeville, Pennsylvania 15116, U.S.A. (Received 16June 1971) Abstract-The rate of oxidation of iron-impregnated electrode graphite granules (< O-01 to 2.1% Fe) in CO*-CO mixtures was measured at temperatures 700 to 1000°C and at pressures 0.03 to 1.0 atm. Some measurements were made also with iron-impregnated coconut charcoal and coke. In air oxidation experiments, graphite granules impregnated with silver, copper, chromium, zinc, nickel, cobalt and iron were used. The rate of oxidation in C02-CO mixtures increases by several orders of magnitude when graphite is impregnated with iron, e.g. with 2.1% Fe-graphite the rate in CO, is a factor of about one million greater than that for iron-free graphite. At 8W”C, the rate increases almost linearly with the iron content. After 20 to 50% oxidation the catalytic effect of iron diminishes even in gas mixtures where iron cannot be oxidized. However, upon Hztreatment the catalytic effect can be revived. The dependence of the rate on gas composition suggests that there might be strong CO, and CO adsorption on the pore walls of graphite impregnated with iron and that for a given total pressure the rate of oxidation in CO,-CO mixtures is proportional to the partial pressure of CO*. That is, the rate of oxidation is controlled by the dissociation of CO2 with an apparent heat of activation of 87.6 kcal. 1. INTRODUCTION
librium between graphite and hydrogenmethane mixtures could be obtained rapidly at temperatures below 1000°C only when graphite was impregnated with a small amount of iron. Long and Sykes [5] observed that the rate of oxidation of coconut charcoal in steam or carbon dioxide decreased by about an order of magnitude upon acid extraction of impurities from the charcoal. However, relative dependence of the rate on the gas composition was not affected by this purification treatment. In a series of ud hoc experiments done by Wynne-Jones[6] a marked decrease was noted in the ignition temperature of cellulose chars by the addition of alkali carbonates. For example, the ignition temperature in air decreased from about 500 to 300°C by impregnating the char with 1% Na2C03. A marked increase was also noted in the rate of oxidation of Ceylon graphite in CO, at 800°C with iron impregnation (up to 0.25% Fe). Similar observations
As early as 1903, Schenck and co-workers [l] found that soot formation from carbon monoxide was catalyzed by the presence of metallic iron, and that iron oxides have a lesser effect. In a review paper in 1923 Bancroft[2] cited interesting observations made by many investigators on the catalytic oxidation of coal, charcoal and other carbonaceous materials. A variety of additives was noted to enhance the steam-carbon reaction, e.g. alkali and alkaline earth salts (chlorides, sulphates, carbonates, etc.), metal oxides and metals. Apparently, common salt or spelter (commercial zinc) has long been recommended for the removal of chimney soot. In a detailed investigation of the reactivity of coke, Jones et al. [3] observed that metallic iron (up to about 1%) was more effective than iron oxide in accelerating the oxidation of coke in carbon dioxide. It was also noted by Smith [4] that the equi97
Carbon Vol. 10. No I-G
98
E. T. TURKDOGAN and J. V. VINTERS
are reported in more recent papers by Walker et al. [7] and by Gallagher and Harker[S] demonstrating the catalytic effect of iron and compounds of transition metals on the oxidation of graphite. For further information on previous investigations of this subject reference should be made to a review paper by Walker et al. [g]. In the course of investigation of the rate of oxidation of graphite in hydrogen-steam mixtures in this laboratory, McKewan and Rice* observed a remarkable catalytic effect of iron and nickel on the oxidation of graphite. During oxidation in water vapor for 30min at lOOO%, a point contact was maintained on the surface of an electrode graphite sphere by a 0.1 mm dia. nickel wire. Severe erosion of graphite over an extended area (I cm dia.) the center of which was the point of contact with nickel wire, is indicative of enhanced rate of oxidation. Study of published papers on the catalytic oxidation of carbons leaves one bewildered. At present there appears to be no clear understanding of the reaction mechanisms which may be responsible for the catalytic effect. Much of the previous work was of an exploratory nature, illustrating the phenomenon, but not quantitative enough for better assessment of the causes of enhancement of the reaction rates. For the purpose of furthering our knowledge of this phenomenon, the present experimental work was undertaken to investigate more systematically effects of various metals on the rate of oxidation of graphite, charcoal and coke in C02-CO mixtures and in air. 2. EXPERIMENTAL
In the initial exploratory experiments the possible effect of iron and ferric oxide on the oxidation of graphite in carbon dioxide was investigated using the following procedure. A hole (- 0.05 mm dia.) drilled into a 12 mm “Private communication U.S.S. Fundamental Research Laboratory, Monroeville, Pa.
dia. electrode graphite sphere was filled with a small quantity of electrolytic iron powder or reagent grade ferric oxide and the rate of oxidation in COz at 900°C was measured. Typical examples of the results are given in Fig. 1 showing weight change with time. The top curve is for 29 mg Fe packed into the hole in the graphite sphere. Initially, a weight increase was observed and subsequently weight decreased linearly with time for the period investigated. It is evident that in the initial stages of reaction the iron is first oxidized. The initial weight gain estimated by linear extension of the curves, as shown by dotted lines in Fig. 1, corresponds to about 8 mg 0; this is very close to 8.6 mg 0 needed to convert 29 mg Fe to wustite. After about 40 min reaction time, the graphite is oxidized at the rate R,-,= 1.5 X 10m4 fractional mass loss per minute. When the hole in the graphite sphere I..
-601
0
I
I 40
I
, 60
, 120 TIME, rn,n
1 I60
, 200
I 240
Fig. 1. Oxidation of a sphere of electrode graphite (12 mm dia.) in 0.96 atm CO2 at 900°C: (a) hole filled with 29mg Fe and oxidized in CO,; (b) hole filled with 26 mg Fe203 and oxidized in CO, with intermitten H2 treatment; (c) iron-free electrode graphite.
was filled with 26 mg Fe24 powder, the lower curve was obtained. During the first 40 min the rate of oxidation was R. = I.4 X 1O-4 min-‘. After 80 min reaction time, CO, was replaced by H2 which brought about a rapid drop in weight by 6 mg; subsequently no further change in weight was observed. Upon re-introduction of COP, there was initially some weight increase followed by
CATALYTIC
OXIDATION
the oxidation of graphite. It is evident that upon H, treatment the iron oxide in the hole of the sample is reduced to Fe which is again oxidized to wustite when CO, is reintroduced. The line c in Fig. 1 is for a 12 mm dia. ironfree electrode graphite sphere oxidized in atmospheric pressure of CO, at 900°C at the rate of R, = 1.0 X 10e4 min-l which is close to those for graphite in contact with iron oxide. Similarly, when the graphite sphere was suspended in the reaction tube with a nickel or an iron wire, there was no appreciable effect on the rate of oxidation in CO,. It should be noted, however, that as observed by McKewan and Rice (mentioned earlier), oxidation in steam is enhanced considerably if graphite is in contact with iron or nickel wire; this effect is not observed during oxidation in CO,. In order to achieve a uniform distribution of the desired metal in carbon samples, the following procedure was employed. Three types of carbon were used: electrode graphite, National AUC with an average ash content of 0.03%; coconut charcoal with an ash content of about 1% containing primarily alkali aluminosilicate and iron oxide; metallurgical coke with an ash content of 7.5% containing 53% SiOz, 32% A1203, 10% (CaO, MgO, alkaline oxides), 3% Fez03 and 0.4% S. To ensure uniformity of the samples, adequate batches of these carbons were crushed and sized to within - 30 + 40 screen mesh. About 20 to 30 g of the carbon granules contained in a platinum basket was suspended over an aqueous salt solution of the desired metal in a flask which was evacuated to about 10e3Torr. Then the granules were submerged into the solution and air was introduced into the flask. The granules were allowed to soak in the salt solution for about 20 hr. Subsequently, the granules were rinsed in water and dried; finally they were heat treated in a stream of hydrogen at 750°C for about 20 hr. The batches of samples thus treated, and rescreened to remove fines, were analyzed for the content of the metal with which they were
OF CARBON
99
impregnated. In this impregnation treatment using 10% or 20% of the salt solution the following reagent grade chemicals were employed: FeCl,, Fe(N03)3, NiC&, CoCl,, CrCl,, ZnCl,, Cu(NO,), and AgNO,. Most of the experiments were carried out in CO,-CO mixtures using graphite impregnated with iron to < O-01%, 0*03%, 0.30%. 0.6% and 2.1%; the sample containing 2.1% Fe was treated with 20% Fe(NO,), solution, other batches were prepared using lo%, or 20% FeC13 solution. The experimental technique and the apparatus used for,oxidation were the same as those described previously [lo]. That is, one or two layers of about 100 mg granules (- 0.4 mm dia.) were placed on a platinum gauze suspended, within the uniform temperature zone of the reaction tube, by a platinum wire from an automatic recording semimicrobalance. The rate measurements were continued, at a constant temperature (? 1°C) up to 50-80 per cent weight loss or higher. In several experiments the residual sample was analyzed for iron, showing no loss of the metal during oxidation. In all experiments the sample was heated in hydrogen for 30 min prior to the introduction of the reacting gas into the reaction chamber. The gas flow employed in these experiments was 1500 cm3 (STP)/min; as shown later, there is no CO,starvation at this flow rate. 3. OXIDATION
IN CO,-CO
MIXTURES
In a few experiments the spheres of graphite (up to 12 mm dia.) impregnated with 0.4% Fe were oxidized in 100% CO, and CO&O = 1 at atmospheric pressure. The weight loss data, as a semilog plot, are shown in Fig. 2. Similar type of rate curves were obtained for iron impregnated granules oxidized in CQCO mixtures at and below atmospheric pressures; the typical examples of the weight loss data are shown in Fig. 3 for 0.30% Fe samples. The linear parts of the curves for initial fast rates, denoted as stage (I), extend over the range lo-50 per cent mass loss, depending on
E. T. TURKDOGAN
100
and J. V. VINTERS
44 II
-G E I 5
30 42
.” i+
40 40
9
b -I
; %
I 30
38 I 3L-___L
I
0
TIME. m,n
Fig. 2. Oxidation of electrode graphite spheres (containing 0.4% Fe) in CO,-CO mixtures: CO&O = I at O-5 atm and 1 atm CO*.
gas composition. At later stages of oxidation, the rate ultimately reaches a steady state at a lower level, designated as stage (II). Similar behavior was observed in the oxidation of charcoal and coke impregnated with iron. 3.1. Oxidation of graphite The effect of particle size on the rate of oxidation of 0.4% Fe-graphite spheres at 800°C is compared in Fig. 4 with that of graphite (no iron added) oxidized at 900°C. Despite this temperature difference, we see a very strong effect of iron in enhancing the rate of oxidation in COZ. Even the slower stage II-oxidation is very much faster than that observed in the absence of iron. As would be expected, the particle size has a marked effect on the initial (state I) rate of oxidation. In order to minimize the pore diffusion effect which is appreciable for large particles, the remainder of the rate measurements were made with granules having an average particle size of 0.4 mm.
40
U
I j 80
120
I
I60
60
200
Fig. 3. Semi-log plot of weight less data for oxidation of 0.3% Fe-graphite granules (- 0.4 mm dia.) at 800°C in CO,-CO mixtures.
lo-’ I e
2
z 10-2
latm
4 z
s
co,/co= 100%co2I >
,
6WC
0 4% Fe
10-3
4’
$ 4u E r
lo-’
u?
9OOT
NO IRON
10-51’
0
8 DIAMETER,
16
24
mm
Fig. 4. Effect of iron impregnation on the rate of oxidation of electrode graphite spheres in COz.
The weight loss data in Fig. 5 show that when graphite is impregnated with cobalt or
CATALYTIC
OXIDATION
nickel, there is again a marked decrease in the rate with the progress of oxidation. This behavior cannot be attributed to the oxidation of the metal. For example, NiO does not form at 800°C in a gas mixture with pcO,/pcO= 1 used in some of the oxidation experiments. Similarly, in gas mixtures havingpco/pcol > 1 used in many oxidation experiments iron cannot be oxidized to wustite, and yet, the catalytic effect becomes smaller at later stages of oxidation of the iron-impregnated graphite.
L.__!k 0 6 % Fe
16
0
20
40 TIME, mtn
OF CARBON
101
from the initial rate at the start of the experiment. There is no doubt that the catalytic effect of iron (and presumably of other metals) on the oxidation of carbon can be revived by hydrogen treatment. The weight loss data are given in Fig. 7 for oxidation of 0.6% Fe-graphite in II,-CO, mixtures at 800°C. Beyond about 60 per cent mass loss there is again a decrease in the rate of oxidation, despite the fact that the catalytic effect of iron is revived when the sample is treated in 100% Hz (Fig. 6). The reason for the decrease in the catalytic effect of iron during oxidation in COB, CO,--CO and CO,-E-I, gas mixtures is not understood; however, this behavior does not appear to be related to the oxidation of the iron.
60 60
80
Fig. 5. Oxidation of granular electrode graphite (- 0.4 mm dia.) impregnated with Fe, Ni or Co in CC&/CO = 1 at 1 atm and 800°C.
The catalytic effect of iron on the oxidation of graphite can be revived by hydrogen treatment; this is demonstrated in Fig. 6 for a 2.1% Fe sample at 800°C. After 60 min oxidation in CO, the rate decreased considerably, Noticeable weight loss during the 30 min H,treatment is attributed to the formation of CH+ Upon re-introduction of CO,, the rate of oxidation R0 = 0.034 fractional mass loss per minute is similar to R. = 0.035 mm-’ measured at the beginning of the experiment. When the rate of oxidation decreased, a second Hz-treatment was applied. Once again we see that the catalytic effect is recovered. After the third Hz-treatment, the rate of oxidation (at about 86 per cent mass loss) in COz, R0 = 0.02 min-I, is not much different
TIME,
mtn
Fig. 6. Oxidation of granular electrode graphite (O-4 mm dia. and impregnated with 2.1%, Fe) in 1 atm CO, at 800°C with intermittent H, treatment.
In a series of experiments, graphite granules impregnated with 0.3 and 06% Fe were oxidized in CO&O mixtures at 800°C. The results given in Figs. 8 and 9 show the effect of gas composition on the rate of oxidation. It should be noted that the rate for stage II oxidation is not much affected by gas composition. Because of the many uncertainties involved with the possible ‘poisoning’ of the iron catalyst in the later stages of oxidation, our study is confined primarily to the initial 10 to 30% oxidation (stage I) where the cataly-
E. T. TLJRKDOGAN and J. V. VINTERS
102
25
16
1 I
'2
5
@?,@ 10-S 06
I 04 0
40
60
120
160
,I,
;
10-Z
pco, 8atm
10-I
100
Fig. 8. Effect of gas composition on the rate of oxidation of electrode graphite granules (- 0.4 mm dia., impregnated with 0.3% Fe) in CO,-CO mixtures at 800°C.
TIME, mm
Fig. 7. Oxidation of 0.6% Fe-impregnated graphite granules (- O-4 mm dia.) in Hz-COP mixtures showing decrease in the catalytic effect beyond 60% mass loss. tic effect of the metal is most striking. In these experiments with C02-CO mixtures ranging from 0 to 90% CO, total pressure was varied from about 0.03 to 0.96 atm. In 100% COP, the rate of oxidation becomes essentially independent of pressure beyond about 0.1 atm COZ. In the presence of CO, the rate of oxidation first increases with increasing pco2 reaching a maximum rate then decreases with further increases in pcoZ. This effect becomes pronounced at higher concentrations of CO. This behavior is due in part to the effect of the reverse reaction (2C0 = C + CO,) on the rate of oxidation. The results for other temperatures and other samples are given in Fig. 10: (a) 0.03 and 0.6% Fe-graphite oxidized in CO%-CO mixtures at 900°C; (b) < 0.01, O-3 and 0.6% Fe-graphite oxidized in CO,-CO (1 atm) at 1000°C and 0.6% Fe-graphite in CO, (0% CO) at 750°C. It is important to note that the rate of oxidation in 100% CO2 becomes essentially independent of pressure at pressures above 0.1 atm, an effect that is a manifestation of relatively strong adsorption of CO, on the
Ir” lo-’
10-s
10-z
,Blln P CO2
10-I
100
Fig. 9. Effect of gas composition on the rate of oxidation of electrode graphite granules (- 0.4 mm dia., impregnated with 0.6% Fe) in COB-CO mixtures at 900°C.
reaction surface of graphite impregnated with iron. The highest rate observed in these experiments was R, = 0.12fractional mass loss per minute. For the initial sample weight w,, = 100 mg, this rate corresponds to the usage of 0.001 mole CO,/min which is less than 2 per cent of the ingoing CO, flow. It is therefore reasonable to state that under the experimental conditions employed there was no CO,-starvation.
CATALYTIC
OXIDATION
103
OF CARBON
bl
to~“C
06%Fe
Fig. IO. Effect of gas composition and iron content of oxidation of electrode graphite granules
on the rate (- 0.4 mm
dia., impregnated with Fe) in CO,-CZO mixtures at 750, 900 and 1000°C. 3.2. Oxidation of coke A few oxidation experiments were carried metallurgical coke out using acid-purified granules impregnated with 1.8 % Fe. About 0.4 mm dia. coke granules were treated with boiling HCl solution then impregnated with iron as described earlier. Upon acid purification the iron content decreased to 0.21% from 0.38% Fe in the unpurified coke. The rate of oxidation of the acid purified coke in CO,/CO = 1 at 1000°C was about the same as that measured previously [lo] for unpurified coke; this is shown by the dot-dash curve in Fig. 11. Upon impregnation with 1.8% Fe, the rate of oxidation increased by a factor of about 3 to 4. Compared to graphite, the catalytic effect of iron on the rate of oxidation of coke is almost negligible. 3.3. Oxidation of coconut charcoal Unpurified coconut charcoal used in the previous [ lo] and present experiments containing 0.11% Fe; upon acid purification the iron content was reduced to 0*010/c. As indicated in Fig. 12, at 800°C and with COJCO = 1 at atmospheric pressure the rate R. = lop5 fractional mass loss per minute for acid purified charcoal is about l/8 of that for unpurified charcoal measured previously [lo]. This is similar to the observations made by Long
ti
_. ACID PURIFIED 8 IRON IMPREGNATED (1.8%Fe)
:
z 2 2 5 F ::
.H
lo-3-
_I-
ff
_ a?
0’ 0’
IO-q-J 0.01
/-
UNPURIFIED (0.38%Fe) PURIFIED (0.2l%Fe) _
0.1
I .o
pco, , atm. Fig. 1 I. Effect of CO,-pressure on the rate of oxidation of acid purified (1.85% Fe) and unpurified (0.38% Fe) coke granules (- 0.4 mm dia.) in CO,/ CO = 1 at LOOO”C.
and Sykes [5] mentioned earlier. Whether acid-purified or unpurified, impregnation of the charcoal with 1 to 1.7% Fe increases the
E. T. TURKDOGAN
104
and J. V. WNTERS 10-l
t
!-
1 4
I
I
-
z zz 1 g F :: E
-
I
/I/
I
I/
- - - -UNPURIFIED (O.li%Fe) -ACID PURlFliD (O.OI%Fe)
P z 5
-1
I
900°C, COCONUT CHARCOAL
-
!
UNPURIFIED 8 IRON IMPREGNATED (I%
Fe)
10-4 _ UNPURIFIED B NO IRON ADDED;!
.-@
_rC--
- (0.11 % Fe)_,’
10-43 0.01
ho,
’
atm.
0.1 Pc02,
1.0
otm
Fig. 12. Effect of COa-pressure on the rate of oxidataion of acid purified and unpurified coconut charcoal granules (with or without iron impregnation) in COJCO = 1 at 800°C.
Fig. 13. Effect of CC&-pressure on the rate of oxidation of acid purified (1.7% Fe) and unpurified (0.1 1% Fe) coconut charcoal granules (- 0.4 mm dia.) in 100% CO, and CO&O = 1 at 900°C.
rate of oxidation
of the results is that, unlike the behavior observed in COz-oxidation, the rate of oxidation remains essentially unchanged, or if anything increases, with the progress of
by over two orders of magnitude, compared to the acid-purified charcoal. Although the rate of oxidation of coconut charcoal at 800% is enhanced considerably by iron impregnation, the magnitude of the effect is much smaller than that observed for iron-impregnated graphite.- In Fig. 13, the effect of pressure on the rate of oxidation of acid-purified charcoal is compared with that of unpurified charcoal at 900°C. The rates differ by about a factor of 3 to 4 which is less than that observed at 800°C. 4. OXIDATION
IN AIR
Granules of electrode graphite impregnated with various metals were oxidized in 2200cm3 (STP)/min flow of air at temperatures within the range 500~800°C. Typical examples of the weight loss data are shown in Fig. 14 for 6~-8~OC. Most striking feature
100
0 40 60 60
IO
90 94
F 96 f *z
I
8 P (L (1 4 ; b:
Fig. 14. Oxidation of metal-impregnated electrode graphite granules (- O-4 mm dia.) in air.
CATALYTIC
OXIDATION
oxidation. If the decrease in the catalytic effect of metals on the oxidation of graphite in COP were attributed to the oxidation or oxygen poisoning of the catalyst, such an effect would have been more pronounced during oxidation in air. The rates derived from the initial linear parts of the log w vs. t plots for air oxidation extend to 50% or more mass loss. The temperature dependence of the rate of oxidation in air is shown in Fig. 15 for graphite granules impregnated with 1% Ni, 0.1% Zn, 0.54% Co, 0.6% Fe, 1.0% Cr, 1.1% Cu or 1.7% Ag. The rate measurements were also made with untreated (no metal impregnation) graphite granules. For comparison, the rate of oxidation in 1 atm COz is included in Fig. 15 for zero and 0.6% Fegraphite.
It is seen that the effect of iron on
1000
100
900
800
700
5OOT
600
OF CARBON
105
the relative rate of oxidation in air is less than that observed in COz. There are variations in the catalytic effect of different metals. For example, while 1% Ni hardly affects the rate of oxidation in air, 1*7%Ag and 1.1% Cu increase the rate at 600°C by almost three orders of magnitude. Above 700°C the rate of oxidation in air becomes essentially independent of temperature and the type of metal impregnated into graphite. This is believed to be due to side effects arising from the lack of heat and mass transfer at these very high reaction rates. For the initial sample weight w. = 100 mg, the rate R. = 0.5 fractional mass loss per minute corresponds to usage of about 20% oxygen (overall reaction product being CO,) of the ingoing air flowing at 2200 cm3 (STP)/min. The graphite impregnated with Ag, Cu, Cr, etc., was also oxidized in C02-CO mixtures at 800°C. Except for Fe, Ni and Co, impregnation of graphite with other metals had no perceptible effect on the rate of oxidation in COz, although the rates in air were much enhanced. 5. DISCUSSION
When discussing the rate of a reaction between a gas and a porous medium, due consideration should be given first to the possible effect of gas diffusion within the pores of the solid medium. In our previous work [lo] on the oxidation of electrode
I
O6 7
8
I
1
IO
9 104/T,
II
12
13
K-’
Fig. 15. Temperature dependence of the rate of oxidation of metal-impregnated electrode graphite in air or CO, is compared with that for graphite to which no metallic addition was made.
graphite (containing less than 0.03% ash) in CO,-CO mixtures we found that there was essentially uniform internal oxidation with granules less than 1 mm dia. over the temperature range 800-1200°C. In a subsequent mathematical analysis of the pore diffusion effect[ll], it was shown that the depth of internal oxidation, for a given type of carbon, depends on temperature and particle size. For a given type of carbon, i.e. given usable pore area and effective gas diffusivity, the rate of oxidation relative to that for uniform internal oxidation, R$R,, is a function of the product r2k’ where r is the radius of the par-
106
E. T. TURKDOGAN
title and k’ is the specific rate constant. From the mathematical analysis it was found that for r*k’ < 5 X lobs g-atom C/cm set the relative rate is within the range 0.90 < Ri/R, -=z 1.0, i.e. essentially uniform internal oxidation. For the oxidation of electrode graphite in CO,-CO at 8OO”C, from our previous work, r*k’ = 2 X IO-‘” g-atom C/cm set for 0.04 cm dia. granules. Using this value, it is estimated that an increase in the rate of oxidation at 800°C by 4 to 5 orders of magnitude, as observed for iron-impregnated graphite, would still ensure essentially uniform internal oxidation in O-4 mm dia. granules. At lOOO“C, however, Ri/R, is estimated to be about 0.4 for O-4 mm dia. granules. When porous carbon is impregnated with a metal using a suitable salt solution (as employed in this work) followed by drying and hydrogen treatment, the pore walls of the carbon may be partly or completely covered by a monolayer of the adsorbed metal at the reaction temperature. For the common arrangement of six nearest neighbors, the cross-sectional area, A, of the adsorbed iron is estimated from A = (V/N)2’3, where V is the molar volume of iron and N is the Avogadro’s number,. thus A = 5 A*. Using this, the amount of iron required for monolayer coverage is calculated from wt.% Fe = 5585 S/ AN where S is the pore area per unit mass of graphite. For S = 1 m’/g, 0.17% Fe is needed for the monolayer coverage of pore walls of graphite. If the amount of iron impregnated into graphite is more than 0.17% the remainder is considered distributed evenly in the pores of the carbon as submicron size droplets. Of course, it is equally possible that there may be little or no chemisorption of iron, all of which may be distributed within the pores as minute particles. The fracture surfaces of unoxidized graphite granules impregnated with 0.4 and 2.1% Fe, as viewed by scanning electron microscopy, are shown in Fig. 16. In the sample with 0.4% Fe, the iron particles are not visible; however, the presence of iron was confirmed
and J. V. VINTERS
by fluorescent X-ray scanning of the surface. The ‘dusty’ appearance of the fracture surface in Fig. 16(b) is believed to be caused by the presence of iron droplets less than 1 pm size. As the oxidation progresses, the concentration of the residual iron in the pores increases and the iron droplets grow. The scanning electron micrographs in Fig. 17 are from different area of 0.6% Fe-graphite oxidized to 80 per cent mass loss in 50% CO,50% CO at 9OO”C, showing growth of iron particles (white) during oxidation; in some regions (lower photograph) large clusters of iron were observed. The scanning electron micrograph in Fig. 18 is that of the iron residue after complete oxidation of 0.6% Fegraphite in 50% COZ-50% H2 at 900°C. The pore structure is similar to that of porous iron formed by gaseous reduction of iron oxide. Accumulation and growth of iron particles during oxidation do not affect the rate of oxidation of graphite in a consistent manner. For example, the catalytic rate of oxidation in CO,-CO, and even in CO*-Hz mixtures, is retarded beyond about 20 to 50 per cent oxidation. However, by applying intermittent H,-treatment, the catalytic effect of iron on the oxidation of graphite in CO, can be revived (Fig. 6). On the other hand, when oxidized in air, there is little change, or a slight increase, in the rate during the progress of oxidation up to 90 per cent or more weight loss (Fig. 14). These observations, though apparently not interconsistent, are found to be reproducible, but remain unexplained at present. The surface areas of some of the metalimpregnated graphite granules (after Hztreatment) were measured by nitrogen adsorption (BET) technique. In the results given in Table 1 a modest increase in the pore area resulting from metal impregnation cannot account for the dramatic increase in the rate of oxidation. In our previous work[ro] on the oxidation of iron-free graphite (< 0*03% ash) in CO*-
Fig. 17. Scanning electron micro~aphs showing growth of iron oarticles aft.er 80% oxidation of 0.6% Fegraph’ite granules in 50% CO,-50% CO at 000°C.
Fig. 18. Scanning electron micrograph of’ iron residue after complete oxidation of’ O+.?% Fe-graphite in 507% CO,-50% Hz iIt 900”~:.
CATALYTIC
OXIDATION
Table 1. Pore surface area (BET) of metal-impregnated electrode graphite Metal 0 0.3% 0.6% 1.1% 1.7%
S(m*/g)
Fe Fe cu Ag
1.0 1.5 2.5 2.0 1.7
CO mixtures, the experimental rate data for uniform internal oxidation was interpreted in terms of two consecutive reactions in series: I. Dissociation CO&) II. Oxidation
of COz,
= CO&ad)
-+ C+20(ad)
(1)
of carbon,
C + O(ad) = COS(ad) -+ CO(g)
(2)
where COz$ and CO$ are activated complexes occupying single sites, g and ad indicate gaseous and adsorbed species. The previous experimental results support the view that in the presence of CO, which is adsorbed strongly on the surface of carbon, the resistance due to reaction (1) is much greater than that due to reaction (2). Hence, for a given pco, the rate is proportional to pcoz, and for the limiting case of almost complete surface coverage by CO, i.e. fraction of covered sites Oco -+ 1, the apparent heat of activation is 87.6 kcal. In the absence of CO, the resistance of reaction (2) appears to be much greater than that of (1) as evidenced by the proportionality of the rate to (~,,)“’ with an apparent heat of activation of 69 kcal; there was no evidence of CO, adsorption on the carbon surface. The change in the relative resistances of reactions (1) and (2) by CO was attributed to the changes in the thermodynamics of the carbon surface brought about by the adsorption of CO. We see from the results in Figs. 8, 9 and 10
107
OF CARBON
for oxidation in IOO% COz that the rate becomes essentially independent of pressure beyond about 0.1 atm CO,. This indicates strong CO, adsorption on the pore surface of iron-impregnated graphite; an effect that was not observed for iron-free graphite. Furthermore, a marked decrease in the rate of oxidation with increasing CO-pressure (Figs. 8 and 9), even when the reverse reaction is insignificant, is a manifestation of strong CO-adsorption on the pore surface of iron-impregnated graphite; this is similar to that observed for iron-free graphite. The role of adsorption on the kinetics of heterogeneous reactions was demonstrated by a number of examples in a review paper by Darken and Turkdogan [ 121. It was shown that a rigorous phenomenological treatment of the reaction kinetics (for an assumed ratecontrolling reaction) is possible only for the limiting case of either (i) essentially bare surface, i.e. (1 - 0) + 1, or (ii) almost complete fillage of the surface sites, i.e. (1- 0) -+ 0. The present results for iron-impregnated graphite indicate strong CO, as well as COadsorption, thus complicating the interpretation of the rate data. If the thermodvnamics of CO, and CO adsorption on iron-impregnated graphite turns out to be similar, for the limiting case (1 - f3) = [l - (0e02+ &,)l -+ 0,
where the activity coefficient-like term cp’ is assumed to be approximately the same for adsorbed CO, and CO. With this approximation, as a test case, the rate for dissociation of CO, is approximated by
R
0
~
@‘4[pco* - (Pco,M Pent + PC0
(4)
where a’ is the rate constant (product of various constants) dependent of temperature, iron content and the area of the reaction surface; (pcot), is that for CO, + C = 2C0
108
E. T. TURKDOGAN
equilibrium for given temperature and PC0in the gas; R. is the rate in terms of fractional mass loss per minute. By rearranging equation (4),
and J. V. VINTERS
case of almost complete surface coverage by CO, and CO. We see from equation (4) that for pcO,+pcO = 1, or for any other constant total pressure, the rate is proportional to CO,-pressure. The results for < 0.01% Fe-graphite (Fig. 11) oxidized in C02-CO mixtures at atmospheric pressure show that within the range 0.05 to 0.4 atm COP, the rate is proportional to pcoz. The decrease in the rate at higher CO,-pressures is attributed to the partial internal oxidation discussed earlier. The product of the rate parameter @‘cp’ derived from the experimental results using equation (5) is given in Table 2, together with graphite the values Q1cpco for iron-free taken from the previous paper[ro] for the limiting case of @c, + 1. The values given in parentheses are derived from measure-
pch -Ro(pcch)e =&i(Pco2+pc!o) (5) In Figs. I9 and 20, the experimental results for 800 and 900°C are plotted in accordance with equation (5). Within the limits of uncertainty of the data the results seem to fit the approximate rate equation (5) for the special I5
Table 2. The rate constant Wcp’ (min-‘) for oxidation of iron-impregnated graphite in CO,-CO mixtures derived from experimental results by using equation (5). Values of WV’ for zero iron are from the previous [lo] for 0c, + 1 “C Fig.
19. Rate data plotted in accordance equation (5). e,
0
I
I
I
I
02
04
06
08
with
I
~PCo2+P~~La~m
Fig.
20. Rate
data plotted in accordance equation (5).
with
700 700 750 750 800 800 800 800 800 900 900 900 900 900 1000 1000 1000 1000
% Fe 0
0.01 0 0.6 0 0.01 0.3 0.6 2.1 0 0.03 0.03 0.6 2.1 0 0.01 0.3 0.6
log Wp’ (min-‘) - 8.52 (1;‘;;) - 1.40 - 6.70 (- 3.35) - 1.17 - 0.88 (- 0.57) - 5.22 - 0.88 (- 0.40) - 0.22 (-0.10) - 3.90 - 0.60 (- 0.24) (-0.12)
Values in ( ) are estimated from rate measurements in 50% CO,-50% CO mixture at 1 atm.
CATALYTIC
OXIDATION
ments in 50% CO,-50% CO mixture at atmospheric pressure. The temperature dependence of the rate parameter is shown in Fig. 21 for iron-free and iron-impregnated graphite. The departure from linearity at higher iron contents and higher temperatures is characteristic of partial internal oxidation. The line drawn for < 0.01% Fe-graphite is probably for uniform internal oxidation for which the apparent heat of activation, 87.6 kcal, is the same as that found previously [ lo] for iron-free graphite (lower line in Fig. 21). The effect of iron on the rate of oxidation of graphite in CO&O mixtures relative to that for iron-free graphite is shown in Fig. 22. The lowest iron content in the impregnated graphite was < 0.01% Fe; this is indicated by an arrow in Fig. 22. The dotted line drawn through the points for 800°C with a slope of unity suggests that when there is essentially 1000 900
6
600
8
700
IO 104/T,
0 01
01
IO
IRON,
wt. %
Fig. 2’L. Effect of iron on the rate of oxidation of graphite in CO,-CO relative to that for iron-free graphite.
6OO’C
12
K-’
Fig. 21. Temperature dependence of the rate constant WV’ for oxidation of iron-impregnated graphite in COP-CO mixtures.
109
OF CARBON
uniform internal oxidation, the rate mav be proportional to the iron content of the graphite, at least beyond about 0.03% Fe. As pointed out previously, the lesser effect of iron on the rate at higher temperatures may be attributed to partial internal oxidation. The kinetics of the dissociation of CO, on the surface of iron or wustite is different from that on the pore surface of iron-free or ironimpregnated graphite. According to the work of Grabke [13] and of Fruehan and Martonik [14], the rate of decarburization of ironcarbon alloys in CO,-CO mixtures is proportional to CO,-pressure. Similarly, the rate of reaction of CO, with wustite is proportional to CO,-pressure[l5, 161. For both reactions the apparent heat of activation is 46 to 48 kcal which is about one half of that for oxidation of graphite (with or without iron). The catalytic effect of iron on the rate of oxidation of electrode graphite is much greater than that for charcoal and coke. For example, at 800°C the rate of oxidation of acid-purified charcoal increases bv a factor of
110
E. T. TURKDOGAN and J. V. VINTERS
270 when impregnated with 1.7% Fe (Fig. 12); the same amount of iron increases the rate of oxidation of graphite under similar conditions by six orders of magnitude, despite the fact that the pore area of the graphite used is about l/500 of that of the charcoal. In coke, the effect of iron is almost negligible (Fig. 11). The diminishing catalytic effect of iron in charcoal (even acid purified) and coke may probably be attributed to the ash contents of these materials. 6. CONCLUSIONS When graphite is impregnated with iron, nickel or cobalt, the rate of oxidation in COZCO mixtures increases by several orders of magnitude. With < 0.01% Fe the rate of oxidation at 700-1000°C is enhanced by a factor of 2000; with about 2% Fe the rate at 800°C is increased by a factor of about one million. At BOO”C,the rate is proportional to the amount of iron impregnated. Beyond about 20 to 50 per cent oxidation there is a marked decrease in the catalytic effect of transition metals when metal-impregnated graphite granules are oxidized in CO,-CO or CO,-H, mixtures. This behavior persists even when the gas used is reducing to wustite. Upon HZ-treatment of partially oxidized graphite, the catalytic effects of various metals do not diminish during the progress of oxidation. This is in accord with the observations reported by Walker et al. [9]. The kinetics of oxidation of iron-impregnated graphite in CO,-CO mixtures is similar to that of iron-free graphite. However, the results indicate that both CO and COZ are strongly adsorbed on the pore walls of graphite when impregnated with iron, while in iron-free graphite the previous work did not indicate any appreciable CO,-adsorption. In both cases the dissociation of CO, on the pore walls appears to control the rate of oxidation. For a given total pressure, i.e. PC,% + PC07the rate is proportional to pcoZ, suggesting that at pressures above about 0.1 atm there is almost complete fillage of surface
sites by adsorbed CO, and CO. However, these conclusions are tentative, because of the approximations involved in the derivation of the rate equation. For both iron-free and iron-impregnated graphite the apparent heat of activation for the dissociation of CO, (hence for oxidation) is 87.6 kcal. The kinetics of this reaction on graphite (with or without iron) appears to be different from that on iron or wustite surfaces. Silver, copper, zinc and transition metals also enhance the rate of oxidation of graphite in air. The effect of transition metals in air oxidation is much less than that observed when oxidized in CO,. On the other hand, silver, copper, zinc and chromium have virtually no effect on the rate of oxidation of graphite in COe. For metal-free and metalimpregnated graphite the apparent heat of activation for oxidation in air is in the range 50 to 60 kcal. Numerous postulates have been made by many previous investigators (reviewed by Walker et al. [9]) in an attempt to explain the catalytic effect of additives on the oxidation of carbon. However, there seems to be little consistency between these postulates and many of the experimental observations. Although some useful information is obtained from the present investigation, there remains a great deal to be learned about the kinetics of the catalytic effect of metals on the reactivity of carbons.
Acknowledgements-The authors following members of this Bombacvk and C. E. Brickner scanning electron microscope, measuring pore surface areas.
wish to thank laboratory: 1. for work’wi& and B. B. Rice
the L. the for
REFERENCES 1. Schenck (1903); Schenck 2. Bancroft 3. Jones J. Research
R. and Zimmermann F., Ber. 36, 1231 Schenck R., Ber. 36, 3663 (1903); R. and Heller W., Ber. 38,2139 (1905). W. D., J. Phys. Chem. 27,801 (1923). H., King J. G. and Sinnatt F. S., Fuel Tech. Paper No. 25 (1930).
CATALYTIC
OXIDATION
4. Smith R. P., J. Am. Chem. Sot. 68, 1163 (1946). 5. Long F. J. and Sykes K. W., J. Chem. Phys. 47, 361 (1950); Proc. Roy. Sot. (London), 215A. 100 (1952). 6. Wynne-Jones W. F. K., 6th Conf. Britirh Coke Res. Assoc. 8 (1953). 7. Rakszawski J. F., Rusinko F. and Walker P. L., Proc. 5th Conf. on Carbon 2,243 (1961). 8. Gallagher J. T. and Harker H.. Carbon 2, 163 (1964). 9. Walker P. L., Shelef M. and Anderson R. A., Physics and Chemistry of Carbon 4,287 (1968). 10. Turkdogan E. T. and Vinters J. V., Carbon 7, 101 (1969); 8,39 (1970).
OF CARBON
111
11. Tien R. H. and Turkdogan E. T., Carbon 8.607 (1970). 12. Darken L. S. and Turkdogan E. T., Proc. Int. Con$ zn Metallurgy and Materzab Scaence, pp. 25-101, Plenum Press (1970). 13. Grabke H. -J., Proc. 3rd Int. Congress on Catalyszs, Amsterdam, 1964, pp. 928-38, North Holland, Amsterdam (1965). 14. Fruehan R. J. and Martonik L. .J., Nzgh Temp. Sczence, 3,244 (1971). 15. Hauffe K. and Pfeiffer H. Z.. 2. Metullk. 44, ?7 (1953). 16. Pettit F. S. and Wagner J. B., Acta Met. 12, 35 (1964).