486
Ind. Eng. Chem. Fundam. 1983, 22, 486-494
Kinetic Study of Sulfate Reduction with Carbon John H. Cameron' and Thomas M. Grace The Institute of Paper Chemistry, Appleton. Wisconsin 549 12
One of the key reactions occurring in a kraft recovery furnace is the reduction of sodium sulfate in the presence of carbon. This paper contains a kinetic study of sulfate reduction in a sodium carbonate melt. This reaction was followed through infrared analysis of the carbon dioxide and carbon monoxide evolved during reduction. Sulfate reduction was found to be autocatalytic, to vary from first order in sulfate at low sulfate concentrations to zero order at high sulfate concentrations, to be proportional to the carbon surface area, and to be independent of the sulfide concentration. To explain these results, the foliowing mechanism is proposed: sulfate absorbs on an active carbon surface site and reacts with the carbon to form a reduced sulfur compound that then desorbs from the carbon. The autocatatytic nature of sulfate reduction with carbon is due to an increase in active sites on the carbon as the carbon is oxidized by sulfate.
Introduction One of the principal steps in the kraft pulping process is the recovery of the pulping chemicals. In the recovery of the pulping chemicals, black liquor from the pulping procw is burned in a kraft recovery furnace and the sulfur compounds present are converted to sulfide. The firing of black liquor in a recovery boiler is somewhat more complex than the firing of a fossil fuel. Overall objectives are different, since recovery of the inorganic salts in a form suitable for regeneration of pulping chemicals is required as well as combustion of the organic matter for steam generation. The conversion of the inorganic material to the desired Na2C03and NazS takes place in the char bed and occurs simultaneously with the combustion of the carbonaceous char. Optimization of furnace performance requires a better understanding of these processes occurring in the char bed. One of the key reactions in kraft chemical recovery is the reduction of sodium sulfate to sodium sulfide. Although this reaction is industrially important, little information is available concerning its mechanism or controlling parameters. In a kraft recovery furnace, reduction occurs in a melt consisting principally of sodium carbonate, sodium sulfate, sodium sulfide, and carbon at temperatures in excess of 760 "C. Here molten sulfate reacts in the presence of carbon to form sulfide, carbon monoxide, and carbon dioxide. The lack of information on this reaction is due to the difficulty of studying a reaction involving three phases (solids, liquid, and gas), the high temperature required, and the corrosive nature of the reaction system. Previous Work Using carbon monoxide, hydrogen, and carbon as reducing agents, Budnikoff and Shilov (1928) conducted one of the earliest sulfate reduction studies. Their experimental technique consisted of heating a platinum boat containing sodium sulfate in an electric furnace while purging the system with either inert or reducing gases. The sample was then removed from the furnace and analyzed for sulfide and sulfate. One of the more interesting results from this study was that at 850 "C or below, carbon would not reduce sulfate under an inert atmosphere. However, when either carbon monoxide or carbon dioxide was used as the purge gas, the sulfate was reduced with yields of 80% sulfide being obtained in less than an hour at 850 "C. In addition to the work of Budnikoff and Shilov (1928), several studies have been conducted on sulfate reduction using a variety of reducing agents. With carbon as the
reducing agent, the rate of reduction is strongly dependent on the nature of the carbon. White and White (1936) determined that coal and charcoal had the greatest reactivity, followed by lampblack and Acheson graphite, respectively. With gaseous reducing agents, Nyman and O'Brien (1947) determined that H2was the most effective, followed by H a , NH,and CH4,respectively. Budnikoff and Shilov (1928) reported that the reactivity of H2was much greater than that of CO. Several authors have reported that certain metals will catalyze the reduction of sodium sulfate. With hydrogen as the reducing agent, iron has been found to be the most effective catalyst. The catalystic effect of iron in the presence of hydrogen was clearly demonstrated in the work of Nyman and OBrien (1947) and later confirmed by Birk et al. (1971). Nyman and OBrien (1947) also reported that copper(I1) sulfate was an effective catalyst when hydrogen was used as the reducing agent. Budnikoff and Shilov (1928) reported that carbon catalyzes sulfate reduction in the presence of carbon monoxide, and Kunin and Kirillov (1968) reported that carbon catalyzes sulfate reduction in the presence of hydrogen. The work of Kunin and Kirillov (1968) indicated an apparent 25% increase in the reduction rate with a small amount of lampblack present. Although this indicates some catalytic effect of carbon, this increase in rate is small compared to the effect of iron. Sulfate reduction has been found to be autocatalytic with both gaseous and solid reducing agents, as indicated by Birk et al. (1971) and Budniioff and Shilov (1928). Birk et al. (1971) reported that a large initial amount of sulfide increased the initial reduction rate for sulfate from 28% per hour without sulfide present to 37% per hour. Birk et al. (1971) proposed that the catalytic nature of sulfide was due to a reaction between sulfide and sulfate to form a more reactive intermediate. Birk et al. (1971) reported that the reduction of sulfate with hydrogen is zero order in sulfate. Nyman and O'Brien (1947) presented data indicating that sulfate reduction by hydrogen without iron present was independent of sulfate until approximately 90% of the sulfate was reduced. A t this point, the reduction rate decreased, indicating the dependence of reduction on sulfate concentration. The determination of the reduction rate dependence on sulfate is complicated by the autocatalytic nature of the reaction. As reduction proceeds, the increase in the rate due to the reaction's autocatalytic behavior is greater than the decrease in rate due to the depletion of sulfate. However,
0196-4313/83/1022-0486$01.50/00 1983 American Chemical Society
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983
487
Sample + Carbon Addition Pon
44 Figure 1. Experimental system.
once the sulfate is reduced to a low level, its effect should be observed and the reaction rate should decrease. Experimental System
In a kraft recovery furnace, sulfate reduction occurs in a melt consisting principally of sodium sulfate, sodium sulfide, sodium carbonate, and carbonaceous char. To obtain kinetic data on sulfate reduction, two experimental systems were designed to monitor sulfate reduction in an environment similar to that found in a kraft recovery furnace. These experimental systems were similar in design, differing only in reactor size and furnace design. Figure 1 illustrates one of the experimental systems employed to study the reduction of sodium sulfate in the presence of carbon. The reaction vessel consists of an alumina crucible contained in an electric furnace. The reduction reaction was followed by measuring the evolution of carbon monoxide and carbon dioxide, which were removed from the reactor with a nitrogen purge stream. Knowing the nitrogen purge rate to the reactor and the percentages of carbon monoxide and carbon dioxide in the purge stream from the reactor allowed the rate of sulfate reduction to be determined. To accurately measure the volumetric flow rate of the nitrogen purge stream, the nitrogen was metered from a pressurized gas cylinder through a dry gas meter. Next the purge stream passed through an orifice meter which gave a direct reading of the flow rate and enabled approximate adjustments to be made in the flow rate. Before the purge stream entered the reactor, ita temperature was measured with a thermocouple. A mercury manometer connected to the purge line served as a pressure release valve. If the purge line from the reactor became plugged, this manometer would blow, releasing the purge pressure and preventing overpressurization of the reactor. The nitrogen purge stream plus any carbon dioxide and carbon monoxide generated by the reaction was conveyed from the reactor in a lI4-in.steel tube. This gas stream then passed through a filter to remove any particles and to a carbon monoxide-carbon dioxide infrared gas a n a l p r . The infrared analyzer was capable of simultaneously measuring both carbon monoxide and carbon dioxide over a 0 to 30% range. Before each experimental run, the infrared analyzer was calibrated with a standard carbon dioxide, carbon monoxide, and nitrogen mixture. Checking the calibration after each run indicated that the infrared system was very stable during the reduction runs. Figure 2 illustrates the alumina crucible-electric furnace section of the reactor system. The cylindrical crucible has an inside diameter of 4lI8 in. a wall thickness of 'I4 in., and a height of 18 in. During a typical reduction run, the crucible would contain approximately 3000 g of smelt occupying 1.5 L. in. in diameter was located between A steel retort the alumina crucible and electric furnace. If a leak developed in the alumina crucible, the smelt would be con-
ik 't
Support Cradle
N2 Reton Purge 1
N 2 Reton Purge 2
Figure 2. Experimental reactor.
tained by this steel retort protecting the furnace from corrosive attack. When the furnace was operating, both sides of this retort were swept with nitrogen to prevent oxidation of the retort. However, to avoid any unknown dilution of the gas sample from the reactor, the nitrogen purge to the interior of the steel retort was discontinued during a reduction run. The nitrogen purge gas entered the reactor through lI8 in. inside diameter alumina tube with the open end located 1to 2 in. from the bottom of the crucible. This ensured that the carbon and smelt were well-mixed and helped strip the carbon monoxide and carbon dioxide from the smelt. A chromel-alumel thermocouple sheathed in an alumina tube was used to monitor smelt temperature. The smelt temperature was maintained at the desired temperature using an on-off temperature controller to regulate furnace temperature. To prepare for a reduction run,anhydrous, reagent grade alkali carbonates and sodium sulfate were premixed and added to the alumina crucible. A typical charge consisted of 25 mol of sodium carbonate and 1mol of sodium sulfate. This ratio of sodium sulfate to sodium carbonate produced a reduction rate that was easily monitored. Also, considering the sodium sulfide present as inert, the mole ratio of sulfate to inerts in a recovery furnace is close to the ratio used here. To prevent damage to the furnace, the temperature of the system was slowly increased until a temperature a few degrees above the desired reduction run temperature was reached. The reduction reaction was then initiated by adding graphite rods 1in. in length and either 1/8 or 3/a in. in diameter into the smelt. Once the rods were added, the reactor was sealed and the nitrogen purge to the reactor was started. The reaction was then followed through the evolution of carbon monoxide and carbon dioxide. The reactor design also allowed smelt sample to be removed from the reactor during a run. These samples could be analyzed for sulfate and sulfide, and the extent of reduction could be compared to that predicted through
488
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983 2 0 0
18 0
16
14
N
8
$
1
w
I
-c X
IA’
CO E ~ o I u t i o n 1
10
30
.. p
~o~o~o~o-o-o-o-o--a-~-~
0
ElectrisFurnace 4 MoBsY, carbon rods 25 MolasNa,CO, (MokNa,SO, 1750*F lnduciionFurnaceX28 3 0 12 moles V, carbon rode 0 078 mobsNa,CO, 0 0351moleiNa,SO, 1750‘F
1
I
40
50
TIME, man
Figure 3. Sulfate reduction with carbon.
carbon monoxide and carbon dioxide evolution. The second experimental system was very similar to the one described. The major differences were that the reactor contained only 80 g of salts and was heated in a radiofrequency induction furnace. Experimental Results The effects of the following parameters on the rate of sulfate reduction were determined (a) carbon loading and carbon surface area, (b) sulfate concentration, (c) sulfide concentration, (d) temperature, (e) possible catalysts, and (f) smelt composition. Rates of carbon monoxide and carbon dioxide generation and sulfate reduction obtained during a typical reduction experiment using the large reactor are shown in Figure 3. Although the reduction rates vary, dependent on the initial level of reagents and temperature of the system, the basic shape of this reduction curve is representative of the sulfate reduction rates seen in this study. As illustrated in Figure 3, sulfate reduction with graphite rods is a highly autocatalytic reaction. The reaction begins with an induction period of about 6 min, where no reaction is observed, followed by a sharp rate increase, a peak, and, depending on the reaction parameters, either a gradual or sharp rate decrease. Effect of Reactor Configuration on Experimental Results. The reactor heated by the induction furnace contained only 3.5% of the smelt contained in the large electrically heated reactor, whereas the product gases from the reduction reaction were removed with 10% of the nitrogen purge rate used with the large electrically heated reactor. Therefore, the gas flow rates per smelt volume are three times as great in the smaller reactor. To determine if the small reactor would produce experimental data similar to those obtained with the relatively large electrically heated reactor, several of the reduction experiments with graphite rods previously conducted in the large reactor were repeated using this small reactor. Since the reactor configurations, melt mixing patterns, and purge rates are significantly different in the two readors, comparison of the experimental data from the two systems will indicate what effect these variables have on the observed sulfate reduction rates. Any mass transfer influences on the reaction kinetics should also be evident in comparing the experimental data from the two reactor systems. Sulfate reduction rates determined from carbon dioxide and carbon monoxide generation for the two reactors are shown in Figure 4. Here, the level of carbon rods used in the small induction heated reactor was selected to maintain the sarne ratio of carbon surface area per volume of melt to sulfate concentration in the two reactor systems. Therefore, the only experimental variable was the size of the reactors, with the large reactor containing 28.3 times
-m
12
0
E. 1 0 m
K C
0
g
0 8
0
K
m
w
06 J
cn 04
0 2
A
00 2
4
6
8
10
12
14
16
18
20
24
26
28
TIME ( M i n )
Figure 4. Effect of experimental system on sulfate reduction.
more reactants than the smaller reactor. In Figure 4, the sulfate reduction rate in the large electrically heated reactor is plotted directly against time, and the sulfate reduction rate in the smaller reactor is multiplied by the ratio of the two reaction systems 28 to 1 and then plotted vs. time. The relative reduction rates in the two reactor systems are nearly equal with the maximum rate being only 3.7% higher in the smaller reactor. This difference in reduction rates is within the normal experimental variation with either reactor system. One difference that is evident in the sulfate reduction rates in Figure 4 is the lack of an induction period with the small reactor. With the large electrically heated reactor, a period of 5 to 7 min was required before any carbon dioxide or carbon monoxide generation was detected. With the small reactor in the induction furnace, carbon dioxide is observed almost immediately after the addition of the graphite rods. This difference in induction period is due to the initial level of carbon dioxide in the melt. With the large furnace several hours was required to melt the salts and reach the desired reaction temperature. During this period, any carbon dioxide generated from the decomposition of the alkali carbonates was removed from the melt. With the small reactor in the induction furnace, only 10 to 15 minutes was required to melt the smelt and reach the desired temperature. Therefore, a considerably greater amount of carbon dioxide was being evolved from the s m d reactor when the carbon rods were added. In both reactor systems, the dead time (volume of reactor, lines, and filter divided by the volumetric flow rate) is less than 30 s and has little effect on either rate curve in Figure 4. To verify the cause of this induction period, in one experiment, the molten salts in the small reactor were held at temperature for 2 h before introducing the graphite rods. Here, a 5-min induction period was observed, demonstrating that the induction period results
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983
488
-
. e
E
Y)
$ -1.8-
initial Levels Na,CO1= 0 51 mol K,COI 0.26 mol Na,SO, = 0 03 mol Temperature = 1750.F 0 REACTEDROD 0 UNREACTER ROD
9
E
&i
2 -2.0z 0
-
g -2.2L
LOG (SO,,
g-fnoles/liter)
Figure 6. Effect of sulfate on sulfate reduction with carbon. -2.3
-2.4
-2.5
.E -2.6
E
'a
2
2
6
8
10
12
14
TIME (Min)
Figure 5. Effect of graphite rod surface oxidation by sulfate on sulfate reduction.
from stripping the melt of carbon dioxide. In comparing the two reactor systems, no differences were found in the experimental results from sulfate reduction with carbon rods. Therefore, any mixing or mass transfer effects appear to be minimal. Autocatalytic Nature of Sulfate Reduction. With the reactor configuration used in the induction furnace, the graphite rods can be removed from the melt during a reduction experiment. By using graphite rods that have been previously reacted with sulfate, any changes in reaction rate due to surface changes on the graphite can be identified. In Figure 5, the sulfate reduction rate for a previously reacted graphite rod is compared with the rate for an unreacted rod. Since the melt in these experiments was rapidly brought to the desired reaction temperature, a considerable amount of carbon dioxide from the decomposition of the alkali carbonates was being generated a t the time of the carbon addition. Therefore, both reduction rates in Figure 5 begin at relatively high levels. However, the sulfate reduction rate for the previously reacted graphite rod initially increases very rapidly followed by almost a plateau, whereas the sulfate reduction rate with the unreacted rod gradually increases. This demonstrates that the autocatalytic nature of sulfate reduction with graphite rods is due to changes in the nature of the graphite surface. Sulfate Effect. To determine the effect of sulfate concentration on reduction, a series of reduction experiments were conducted with various levels of sulfate present. In these experiments, the melts initially contained 25 mol of sodium carbonate, 4 mol of graphite rods, and sodium sulfate levels varying from 0.25 to 3.5 mol. By comparing the reduction rates in these melts with 0.1 mol of sulfate reduced, the effect of the sulfate concentration
-2.7
i a a
.
.-B
-2.8
a
1 a0
-2,s
c
= a
u, -8
-3.0
-3.1
-3.2
-3.3 -0.2
-0.1
0.0
0.1
0.2
0.3
0.4
0.6
log (Carbon Surface Ares, Sq Crn)
Figure 7. Effect of carbon surface area on sulfate reduction with graphite rods.
on the reduction rate was determined. Figure 6 illustrates the effect of sulfate on the rate of sulfate reduction. Since the reduction rates in Figure 6 were measured with the same amount of sulfate reduced, the carbon surfaces at the time these rates were measured were all oxidized to the same extent. This ensured that all carbon surfaces had the same degree of reactivity. As seen in Figure 6, sulfate reduction with graphite rods varies from first order in sulfate at low sulfate concentrations to zero order in sulfate at high sulfate concentrations. Carbon Surface Area Effect. As shown in Figure 5, the reactivity of the graphite rods is dependent on the extent of oxidation of the graphite. Therefore, to determine the effect of carbon surface area on the sulfate reduction rate, it is necessary to measure the rate at the point in time where the graphite rods have experienced the same degree of oxidation per unit surface area. Figure 7 illustrates the effect of carbon surface area on the rate of
490
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983
Table I. Effect of Composition on Activation Energy runs 55-60,g-mol
melt composition
Na,CO, KC03 Li,CO, Na,SO,
carbon JIs-in.
temperature range activation energy
sulfate reduction with graphite rods. Here, each unit surface area on the graphite rods has been oxidized to the same extent by the sulfate. The melts in Figure 7 initially contained the same level of sulfate but different levels of graphite rods. For each unit surface area on the graphite rods to be oxidized to the same extent a t the time the reduction rate was measured required that the rate be measured with the extent of sulfate reduction proportional to the graphite surface area in the melt. Therefore, if the number of carbon rods for a particular run was doubled, the reduction rate was measured with twice the amount of sulfate reduced. The melts in Figure 7 contained a relatively high level of sulfate, and it was found that at high levels of sulfate the reduction rate is independent of the sulfate concentration. Therefore, although the rate measurements were made with different levels of sulfate, this had no effect on the sulfate reduction rate. In Figure 7, the log of the sulfate reduction rate is plotted vs. the log of the carbon surface. Since the slope of the curve in this plot is one, the sulfate reduction rate is directly proportional to the carbon surface area. Sulfide Effect. With hydrogen as the reducing agent, Birk et al. (1971) reported that the autocatalytic nature of sulfate reduction resulted from the increase in sulfide as the reaction proceeded. To determine the effect of sulfide on sulfate reduction with carbon, several reduction experimenb were conducted with initial amounta of s a i d e added. Two of these experiments are runs 87 and 90 in Figure 6. Anhydrous sodium sulfide was prepared by dehydrating crystallized Na2S.9H20 under a nitrogen atmosphere at 180 "C. Laboratory analysis of anhydrous sulfide prepared in this manner determined that the samples contained 95% to 97% sulfide and 1 to 2% sulfate. Sulfate reduction with initial sulfide present indicates that sulfide has no effect on the sulfate reduction rate. Although runs 87 and 90 have initial amounts of sulfide present, their reduction rates fall on the same curve in Figure 6 as the runs with no initial sulfide present. Iron Catalyst Effect. Various transition metals have been shown to catalyze sodium sulfate reduction with hydrogen or carbon monoxide. For sulfate reduction with hydrogen, iron in the form of FeSO, or Fe203was the most effective catalyst, as reported by Birk et al. (1971). If the mechanism of sulfate reduction with carbon is similar to that with carbon monoxide or hydrogen, iron should also catalyze sulfate reduction with carbon. To determine the effect of iron on sulfate reduction with carbon, mveral reduction experiments were conduded with various levels of FeSO, present. The effect of iron on sulfate reduction with carbon was studied a t different temperatures and smelt compositions. Figure 8 is representative of the results of these experiments. Here, the rates of reduction at 955 "C in Na2C03 are shown with different amounts of iron added. The shape of these curves demonstrates that iron has no effect on sulfate reduction with carbon. The slight differences in the reduction rates shown in Figure 8 are due principally
25.0 0.0 0.0 1.0 4.0 900-983 "C 49 kcal
runs 61-66,g-mol
runs 67-73,g-mol
15.0 10.0 0.0 1.0
3.61 7.48 13.92 1.0 4.0 816-983 "C 71 kcal
4.0
871-983 "C 53 kcal
0.024
0
0.0 % lion in Melt
TIME, rnm
Figure 8. Effect of iron on sulfate reduction with carbon.
to temperature variations during the reaction. Temperature. To determine the effect of temperature, sulfate reduction was studied over a temperature range of 816 to 983 "C. Due to freezing of the melt, 900 "C was the lowest temperature at which sulfate reduction in a melt consisting primarily of Na2C03was studied. To study reduction below this temperature, Li2C03and K2C03were added to the Na2C03melt. The temperature ranges and melt compositions studied and the calculated activation energies for reduction in these melts are shown in Table I. The activation energies in Table I are based on the reduction rate at 40% of the sulfate converted to sulfide. Since the initial concentrations of reagents in Table I were the same for all runs, all the systems contained the same levels of sulfate, sulfide, and carbon at 40% conversion. For the systems in Table I, the reaction rate at 40% conversion is near the maximum rate, and similar activation energies can be obtained using the maximum rate at each temperature. The activation energies were calculated assuming that the temperature dependency can be represented by Arrhenius' law, eq 1 rate = KO X e-hEIRTX f(composition)
(1)
where KO is the frequency factor, AE is the activation energy, R is the ideal gas constant, T i s the absolute temperature, and f(composition) represents the compositiondependent terms. The activation energies in Table I were calculated by a least-squares fit of the natural logarithm of the reduction rate, mol/min, vs. the reciprocal of the absolute temperature, K-l, as shown in Figure 9. During the temperature study with the melt containing K2C03,small explosions occurred in the reactor. A material was also deposited in the purge line from the reactor which exploded when contacted. It is suspected that the unstable compound was potassium carbonyl, K O C 4 O K . Mellor (1927) states this compound may be formed when potassium salts are exposed to high temperatures and strong reducing conditions. Effect of Carbonate on Sulfate Reduction. To determine what effect the alkali carbonate melt may have
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983 491
Table 11. Solubility of Carbon Dioxide in Sodium Chloride and Sodium Carbonate
o'iol
ref Na,CO, NaCl
Figure 9. Effect of temperature on sulfate reduction with carbon. 4.0-
3.0-
-
$ X
C
i
0 CarbonsteMell
\
D
Temperature = 1760.F Initial Concentrations N.,CO, = o . n m o l Na,SO. = 0 . 0 3mal Qriphite=O 12mol,3/8"rcd
0
E d r a
2.0-
0 Sodium Chloride Men
Y I 0
2
4
8
S
10
1
I
12
14
TIME (Min)
Figure 10. Effect of carbonate on sulfate reduction with graphite rods.
on reduction, sulfate reduction rates were also measured in sodium chloride-sodium sulfate melts. In Figure 10, the sulfate reduction rate in sodium chloride is compared with the reduction rate in sodium carbonate. As illustrated in this figure, sulfate reduction is considerably slower in a sodium chloride compared with a sodium carbonate melt. Fourteen minutes after the addition of the graphite rod, the sulfate reduction rate in sodium chloride is approximately only 10% of the rate in sodium carbonate. To determine if this decrease in reduction rate is due to inhibition by sodium chloride or the lack of catalytic action of sodium carbonate, sulfate reduction was studied
880 (9.6t 2.3)X 950 (0.99f 0.18)X 1000 (1.01t 0.12)x
Andresen (1979) Sada et al. (1981)
in various combinations of sodium carbonate, sodium chloride, and sodium sulfate. As the level of sodium chloride in a sodium carbonate-sodium sulfate melt was increased, the reduction rate gradually decreased. Sulfate reduction was also studied in a pure sodium sulfate melt. Here, the reduction rate was also considerably slower than in a sodium carbonate-sodium sulfate melt. These results demonstrate that sulfate reduction with carbon is catalyzed by carbonate. This catalytic action of carbonate may be due to the dissolved carbon dioxide in the carbonate melts. In the comparison of the two reactor systems, it was demonstrated that stripping carbon dioxide from the melt significantlydelayed the initiation of sulfate reduction. As carbon dioxide is stripped from the melt, the equilibrium of eq 2 is shifted toward sodium oxide and Na2C03a Na20 + C02(g) (2) an observed decrease in the initial reduction rate may occur due to the reaction of the carbon dioxide generated from reduction with sodium oxide. However, this shift in equilibrium would result in an apparent decrease in the reduction rate but not the delay in the initiation of reduction that is observed. Therefore, this delayed rate is not due to a shift in equilibrium but to the lack of carbon dioxide generation from reduction, indicating that sulfate reduction with carbon is catalyzed by dissolved carbon dioxide in the melt. With alkali carbonate melt, there is a relatively high level of carbon dioxide dissolved in the melt from the decomposition of the carbonate (eq 2). If sulfate reduction is catalyzed by carbon dioxide the rate of sulfate reduction with carbon should be considerably reduced in a noncarbonate melt where little carbon dioxide would be present. The solubility of carbon dioxide, reported as Henry's Law constants, Table 11, is significantly lower in sodium chloride than in sodium carbonate and no carbon dioxide would be present from decomposition reactions in sodium chloride. Although carbon dioxide does catalyze sulfate reduction with carbon, it appears to be only an initiating catalyst. If increasing levels of carbon dioxide continued to produce a catalytic effect, the first-order rate dependence of reduction on carbon surface area would not be observed. The reduction rate in Figure 7 would increase due to both the increase in surface area and increase in carbon dioxide as the surface area increased. Therefore, only an initiating level of carbon dioxide is required to produce the full catalytic effect. Once this level is exceeded, no additional increase in reduction rate is observed with increasing carbon dioxide level. This limit is apparently not reached in the noncarbonate melts, due to the lower carbon dioxide solubility in these melts and the slower reduction rate. The Henry's law constant for carbon dioxide is an order of magnitude lower in sodium chloride than sodium carbonate and the rate of carbon dioxide generation from reduction is also an order of magnitude lower. Therefore, the level of carbon dioxide present in sodium chloride during these reduction reactions would be approximately two orders of magnitude lower than the level in a carbonate melt. At these levels, apparently not enough carbon dioxide is present to fully catalyze reduction.
492
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983 0.9
Temp = 175OOF Gas Purge Rale=l2.2Llmln. 10.7% O2 Oxidolion Rate = 0 . 0 2 6 moles Na2S converted 10 No2S04/min Moles Oxygen Supplicd=0.048 moles/min
0.7-
0.6
-
W -I
I 0.5-
O \'
v)
N
s 0.4uI
OL- J0
10
TIME, min I5 20
25
30
35
Figure 11. Oxidation of sodium sulfide.
Oxidation of Sulfide. Any time smelt in a recovery furnace is exposed to air, oxidation of sulfide occurs. The degree of sulfate reduction achieved depends not only on reduction but also on the oxidation of sulfide. Therefore, a study on the oxidation of sulfide was included in this project. Oxidation of sulfide was studied by purging a melt containing sodium sulfide with a mixture of air and nitrogen and analyzing melt samples for sulfate and sulfide. Two different methods were employed to provide the initial sullide. In the first method sodium sulfate was added to the melt and then reduced with carbon monoxide to yield sodium sulfide. In the second method, dehydrated sodium sulfide was added directly to the melt. Both methods produced similar results on oxidation. Figure 11 illustrates typical results obtained from the oxidation of sulfide. During this experiment, a gas purge stream containing 10% oxygen was used to oxidize the sulfide. All the oxygen supplied was consumed by the oxidation reaction. Here, the rate of oxygen addition was measured at 0.048 mol/min and the rate of sulfide depletion at 0.026 mol/min. The slight difference in oxygen addition and sulfide depletion was probably due to inaccuracies in the measurement of the air addition rate. Here, the essential feature is that a t the temperature range of this study, oxidation is only limited by the availability of oxygen. In comparison to reduction with either carbon or carbon monoxide, oxidation is a considerably faster reaction. Discussion The autocatalytic nature of sulfate reduction with graphite rods was found to be due to increased reactivity of the carbon surfaces resulting from oxidation by sulfate. This increase in reactivity may be the result of increasing edge carbons on the graphite rods. Walker et al. (1959) and Rodriquez-Reinoseet al. (1963)have shown that essentially all surface reactions with pure carbons occur at the carbon edges rather than basal planes. Walker (1959) and Laine et al. (1963)have shown that, depending on the degree of graphitization, the reactivity of the carbon edges is two to three orders of magnitude greater than that of the basal carbon. Therefore, if the oxidation of the graphite rods by sulfate results in increased edge carbons, one would expect to observe an increase in reactivity.
With graphite rods, sulfate reduction was found to be proportional to the number of active sites on the graphite surface. However, as sulfate reacts with the graphite rods, the number of active sites increases resulting in sulfate reduction with carbon being an autocatalytic reaction. An increase in active sites has been observed with other reactions involving carbon. Walker et al. (1959)state that in the reaction of carbon dioxide with carbon, the number of active sites formed for each carbon monoxide molecule desorbed from the carbon may be 0, 1,or 2. If one to two active sites are formed per carbon monoxide molecule desorbed, an autocatalytic effect will be observed. When all surfaces of the graphite rods are oxidized to the same extent, the sulfate reduction rate is proportional to the surface area. If the surface area of the graphite rods is doubled, the sulfate reduction rate also doubles. The shift in the rate dependence on sulfate from zero order to first order can be described by an adsorption isotherm, eq 3 indicating that sulfate reduction with carbon occurs through adsorption of sulfate on an active carbon site. (3)
Here, C(S04)& is adsorbed sulfate, and [SO4]is the sulfate concentration in the melt. At high sulfate concentrations, the level of adsorbed sulfate is independent of the sulfate concentration, and at low levels of sulfate, the adsorbed sulfate is directly proportional to the sulfate concentration. The effect of sulfate on reduction can be described by reduction rate being dependent on the level of adsorbed sulfate. The catalytic effect of carbon dioxide may be due to the reaction of carbon dioxide with the carbon to form two adsorbed oxygen atoms. These oxygen sites then serve as active sites for the adsorption of sulfate. Blackwood et al. (1959)and Duttal et al. (1977)have reported that the reactivity of coal char with oxygen and hydrogen is correlated to ita oxygen content and that active carbon sites result from the inclusion of oxygen in the coal char. If the carbon atom in the adsorbed carbon monoxide is considered to be part of the carbon matrix, the adsorbed carbon monoxide is then equivalent to an adsorbed oxygen. This adsorbed oxygen in kraft black liquor char may have an effect similar to oxygen in coal char and catalyze reduction. Since these oxygen sites result from the reaction between carbon dioxide and carbon, melts with a low carbon dioxide content would produce few oxygen sites on the carbon and reduction would occur at a slower rate. Only in melts where there is a relatively low carbon dioxide level is the carbon dioxide effect observed. With sulfate reduction in carbonate melts, there is no effect of increasing carbon dioxide. Therefore, carbon dioxide is only an initiating catalyst and at high levels has no additional effect on reduction. Since a relatively high level of carbon dioxide normally is present in carbonate melts, reduction in these melts is independent of the carbon dioxide level. Proposed Mechanism for Sulfate Reduction with Carbon. Based on the experimental results presented in this report, the following mechanism for sulfate reduction with carbon is proposed. Dissolved carbon dioxide in the melt reacts with a carbon site to form two carbon monoxide molecules on the carbon surface. These carbon monoxide molecules then serve as active sites for the adsorption of sulfate. Sulfate adsorbs on these carbon monoxide sites and reacts, forming a reduced sulfur compound and carbon dioxide. The reduced sulfur compound then undergoes
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983
a series of fast reduction reactions to form sulfide. This proposed sulfate reduction mechanism is described in more detail in the following paragraphs. Sulfate reduction with carbon was found to be catalyzed by carbonate. This catalytic effect appears to be due to the high carbon dioxide level in carbonate melts, suggesting that carbon dioxide has a key role in sulfate reduction. One method by which carbon dioxide could influence reduction is through the reaction of carbon dioxide with carbon forming adsorbed carbon monoxide on the carbon surface. These adsorbed carbon monoxide molecules could then act as active sites for the adsorption of sulfate. In the reaction between carbon dioxide and carbon, one molecule of carbon dioxide could be adsorbed on the carbon and react to form two molecules of carbon monoxide (eq 4 and 5).
Here, C02 is carbon dioxide in the melt, Cf is a free carbon site, (CO),& is adsorbed CO, and C(CO,),& is adsorbed COP This reaction of one molecule of carbon dioxide with carbon to form two adsorbed carbon monoxide molecules is consistent with the mechanism proposed by Darken and Turkdogan (1970) for the reaction of gaseous carbon dioxide with carbon. The adsorbed carbon monoxide molecules serve as active sites for the adsorption of sulfate and in the future will be referred to only as active sites. The adsorption of sulfate on these active sites can be represented by eq 6. [so,]
+ cf
C(S04)e&
(6)
with C(S04)ads= K 1 X [SO,] X C;
(7) Here, [SO,] is the sulfate in the melt, C{ represents an active site resulting from the adsorbed carbon monoxide, (CO)a&,and C(S04),& is adsorbed sulfate. The total number of active carbon sites available for the adsorption of the sulfate is the sum of the free carbon sites and those occupied by the sulfate (eq 8). CT = c; + C(SO4)ad, (8) By combining eq 7 and 8 and eliminating the free carbon site term C{, the adsorbed sulfate can be expressed in terms of sulfate concentration and total active carbon sites (eq 9).
At low sulfate concentrations, the adsorbed sulfate is proportional to the sulfate concentration, but a t high sulfate concentrations, the adsorbed sulfate becomes independent of the sulfate concentration. Therefore, the effects of the sulfate concentration on sulfate reduction with carbon are described by the rate of sulfate reduction being dependent on the level of adsorbed sulfate. The rate-limiting step in sulfate reduction is then the reduction of the adsorbed sulfate on the carbon surface forming carbon dioxide and a reduced sulfur compound. Since the rate-limiting step is the reduction of adsorbed sulfate, the reduction rate is proportional to the amount of adsorbed sulfate and can be described by eq 10.
483
Here, [SO,] is the adsorbed sulfate, CT is the number of active carbon sites, bE is the activation energy, R is the gas constant, and Tis the absolute temperature. Here, the rate of sulfate reduction with carbon in a carbonate melt is proportional to the sulfate concentration at low sulfate levels, independent of the sulfate concentration at high sulfate levels, and proportional to the number of active carbon sites. To describe the increase in the reactivity of the graphite resulting from sulfate reduction, the surface of the graphite rods was considered to consist of two types of carbon sites: active sites and inactive sites that were capable of becoming active. The sum of the inactive and active sites per unit surface area is referred to as total sites and is assumed to be constant (eq 11). NT = NA + NI (11) Here, NT is total number of sites per unit surface area, NA is the number of active sites per unit surface area, and NI is the number of inactive sites per unit surface area. The increase in active sites results from the reduction of sulfate on the carbon surface. This reduction reaction is assumed to occur on an active site and in the reduction process may convert an inactive site to an active site. The rate of increase of active sites per unit carbon surface area is then proportional to the rate of sulfate reduction multiplied by the fraction of inactive sites per unit surface area (eq 12).
Here, K’is the proportionality constant, FI is the fraction of inactive sites = NI/NT, and SO4 is sulfate per unit surface area of carbon. Since the fraction of inactive sites plus the fraction of active sites equals one, eq 12 can be written in terms of the fraction of active sites as
By multiplying both sides of eq 14 by dt, integrating, and applying the initial conditions, the fraction of active sites can be expressed in terms of sulfate reduced or sulfide formed per unit carbon surface area (eq 17).
Initial Conditions. At [SO,] = [S041] (initial’sulfate concentration per unit surface area), and FA = FAI (initial fraction of active sites). Applying these limits, eq 14 becomes
[so,- ~ 0 ~ 1 1(15) NT Since [S041- SO4]is the moles of sulfide formed per unit carbon surface area, eq 15 can be written as K’X SI In (1- FA) - In (1- FAI) = NT Here, SI is the moles of sulfide formed per unit surface area. The fraction of active carbon sites can now be expressed as FA = 1 - (1- FAI) e-K’xS*/NT (17)
In (1 - FA) = In (1- FAI) =
The nature of the active carbon site was previously ascribed to adsorbed carbon monoxide on the carbon surface. The active sites described here would also be adsorbed
Ind. Eng. Chem. Fundam., Vol. 22, No. 4, 1983
494
Here, SI is the amount of sulfide formed per unit surface area and CSA is the carbon surface area per liter of melt. With five experimentally determined parameters, K , K,, AE,(1 - FAI), and K’INT, eq 18 describes sulfate reduction with graphite rods over a wide range of experimental conditions. To determine the validity of this rate expression and the experimental parameters, data from six representative reduction experiments were fit to this rate expression using a nonlinear regression analysis computer program. These parameters and their estimate of standard deviation obtained from this analysis are shown in Table 111. The six reduction experiments used to determine the other parameters represent a large range of sulfate concentrations, carbon surface areas, and temperatures. 11lustrated in Figure 12 is a typical reduction experiment showing the actual and predicted reduction rates. As illustrated by this figure, the rate expression for sulfate reduction with graphite rods, eq 18, accurately describes sulfate reduction. The treatment of the active sites results in a rate expression that closely fits the observed autocatalytic behavior of sulfate reduction with graphite rods. Registry No. Sodium sulfate, 7757-82-6;carbon, 7440-44-0;
Table 111. Experimental Parameters Describing Sulfate Reduction with Graphite Rods linear est of std dev
parameter 9.96 x 107 10.4 0.832 1.30 x l o 3 51 050
K ( L/cmZ-s) K , (L/mol)
- FA)
(1
K‘/NT (cm2/mol) A E (cal/mol)
1.26 x 107 2.0 0.016 0.02 x l o 3 1670
I W
a 4
I Run142 Tempsrmure = l750’F Initial Concentrations NWX, = 0.nmi NatSOl = 0.03 mol Carban = 0.12 mol 3/8 graphite rod
l
0
1
I
I
~
2
4
~
6
S
’
10
graphite, 7782-42-5.
Literature Cited
12
i
14
16
~
IS
20
4
22 24 26
t
28
TIME (mini
F i g u r e 12.
Model for sulfate reduction with graphite rods.
carbon monoxide. Sulfate reduction could either result directly in adsorbed carbon monoxide at a neighboring carbon or in a carbon site that would react with carbon dioxide to form adsorbed carbon monoxide. In either case, these active sites would be the same as those previously described. Combining eq 12 and 17, the rate of sulfate reduction with graphite rods can be described by dISO4l --
~
Andresen, R. E. J . EkWochem. Soc. 1979, 238. Birk, J. R.; Larsen, C. M.; Vaux, W. G.; Oldenkamp, R. D. Ind. Eng. Chem. ~ ~ ~ ’ l Process Des. Dev. 1971, * 10, 7-13. Blackwood, J. D.; McTaggart, F. K. Aust. J . Chem. 1959, 12, 533. Budnikoff, P. P.; Shilov, E. J . Soc. Chem. Ind. 1928, lllT-13T. Darken, L. S.;Turkdogan, E. T. “Heterogeneous Kinetics at Elevated Temperatures”; Plenum Press: New York, 1970;p 25. Dutta, S.; Wen, C. Y. Ind. €178.Chem. Process Des. D e v . 1877, 16, 1. Kunin, V. T.; Kiriiiov, I . P. Izv. Vyssh. Vcheb. Zeved., Kim Teknol. 1968, 1 1 , 569. bine, N. R.; Vastola, F. J.; Walker, P. L., Jr. J . Phys. Chem. 1963, 6 7 ,
2030. Melior, J. W. “A Comprehensive Treatise on Inorganic and Theoretical Chemistry”; Longmans, Green and Co. Ltd.: New York, 1927. Nyman, C. J.; O’Brien, T. C. Ind. Eng. Chem. 1947, 8 , 1019-21. Rodrlguezdeinosa, F.; Thrower, P. A.; Walker, P. L., Jr. Carbon 1963, 12, 63. Sada, E. et ai. J . C h m . Eng. Data 1981, 2 6 , 281. Walker, P. L., Jr.; Rusinko, F.,Jr.; Austin, L. G. “Advances in Catalysis”; Elay, D. D.; Selwood, P. W.; Welsz, P. B., Ed.: Academic Press: New York, 1959;p 133. White, J. F. M.: White, A. H. Ind. €178.Chem. 1936, 28, 244.
Received for review August 5, 1981 Revised manuscript received July 8, 1983 Accepted August 3, 1983
