Ind. Eng. Chern. Process Des. Dev. 1980, 79, 108-113
108
Reduction of Nitric Oxide with Metal Sulfides F. P. McCandless,” Kent M. Hodgson, Richard H. White, and Jerry D. Bowman Department of Chemical Engineering, Montana State University, Bozernan, Montana 597 17
Nineteen different metal sulfides were investigated for the reduction of nitric oxide in an initial screening using a test gas of pure NO. Depending on the sulfide, the reduction reaction proceeds at temperatures ranging from about 100 to 500 OC. Although most sulfides investigated resulted in some NO reduction, Bas, Cas, SrS, and FeS were judged to be the most promising. Several catalysts were found which reduced the temperature at which the reaction proceeds by as much as 200 OC, the most active being NaF, NiCI,, and FeCI,. A further temperature reduction was obtained by impregnating the sulfide on high surface area supports of activated alumina or molecular sieves. Continuous electrobalance studies showed that in general, 0,reacts with the sulfides at rates higher than with NO. All combinations of the most promising sulfides and catalysts were tested using a synthetic flue gas containing 1000 ppm of NO and 1% 0,at 400 OC. The capacities of the six best were FeS-FeCI, > SrS-NaF > Cas-NaF > Bas-FeCI, > FeS-NiCI, > Cas-FeCI, and ranged from 0.037 to 0.013 g of NO reduced per gram of initial sulfide present. Capacities of 0.76 and 0.91 were obtained when using 5% Cas (onty) impregnated on alumina and molecular sieves, respectively.
Introduction and Background This research program was initiated to investigate the technical feasibility of using metal sulfides for the reduction of NO, with the aim of controlling emissions of these pollutants from stationary sources. Oxides of nitrogen are one of the most prevalent air pollutants and their emissions must be controlled because of their adverse effects on human health and plant life (EPA, 1971). The literature of 60 years ago (Fulton, 1915; Young, 1915) discusses the dry “Thiogen” process for the recovery of sulfur from SOz according to the reactions CaS + 2S02 CaS04 + S2 4CaS
-
+ 6S02
4CaS03 + 3S2
Barium sulfide was also used. Preliminary research by the authors indicated that the reaction CaS + 4 N 0 CaSO, + 2N2 proceeds rapidly at temperatures above 450 “C when pure NO is passed through a reactor containing powdered calcium sulfide. Based on the promise of this type of reaction for NO, emissions control, a research program was initiated. This paper presents the results of a preliminary investigation of various metal sulfides for reducing NO together with tests of various means to increase the reaction rate. Thermodynamic Study Thermodynamics gives a good indication of the potential of metal sulfides for NO, reduction and so a brief study was made to compare the reduction reactions using various metal sulfides. Several possible side reactions of CaS were also considered together with the thermodynamics of reducing NO with hydrogen. Possible side reactions with other gaseous compounds which may be present in flue gases were not considered. The study was limited to systems for which literature data on both the sulfide and corresponding sulfate were available. Approximate methods were used to calculate the effects of temperature on the free energy of reaction where heat capacity data was not available. The results are summarized in Table I. Several interesting and important conclusions can be drawn from these data. First, the reduction reactions of NO, using metal sulfides are very favorable a t all tem0019-7882/80/1119-0108$01.00/0
peratures of practical interest and in particular a t temperatures below 1000 K. In addition, from a standpoint of free energy driving force, the use of metal sulfides compares favorably with reduction using a conventional reducing agent such as hydrogen. This includes the reduction of both NO and NO2 with the formation of either N2 and/or N20. In addition from a thermodynamic standpoint, it appears that the reduction of NO or NO2 is favored over the oxidation of the sulfide with free oxygen and indicates that the formation of the sulfate is favored over the undesired reactions which liberate SO2. However, the same data also show that these undesired side reactions are also possible. Unfortunately, thermodynamics only indicates which reactions are theoretically possible and gives no indication as to how fast the reactions will proceed or the relative magnitude of the rates. Experimental Section A simple semi-batch reactor system (continuous flow of the test gas to the reactor but batch-wise sulfide reactant addition) was used for most of the tests including initial screening of various metal sulfides, catalyst tests, and more detailed tests of the most promising sulfide-catalyst combinations using a synthetic flue gas. Several reactors were used at various times during the course of the research but all had the same basic design. The reactors were constructed from 35 to 45 cm lengths of schedule 40 type 304 stainless steel pipe with nominal diameters of 3.8,1.9, and 1.3 cm with appropriate inlet and outlet fittings. The bottom half of the pipe was packed with stainless steel wire rings (“Fenske rings”) to increase heat transfer and acted as a gas preheat section. The top half constituted the reaction chamber where the powdered sulfide or sulfide pellets were contained between two porous stainless steel plates. A thermowell made from 6.4-mm stainless steel tubing was mounted axially in the pipe where thermocouples were mounted a t three locations in the reaction zone, In operation the reactor was mounted in a 10 cm diameter cylindrical stainless steel block with a slightly oversize hole through it. This block was wrapped with three nichrome heating coils in ceramic beads which were controlled by variable transformers. The metal block ensured near isothermal operation of the reactor. The outlet gas passed through a small water cooler and then through a tee containing a septum and thence to either a contin0 1979 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980
109
Table I. Calculated Gibbs Free Energy Change and Heat of Reaction for t h e Reduction of NO with Various Metal Sulfides ______
AH,, kcal/mol of sulfide
AG,, kcal/mol of sulfide reaction C a s I 4 N 0 + CaSO, + 2N, MnS + 4 N 0 + MnSO, + 2N2 PbS + 4 N 0 + PbSO, + 2N, ZnS + 4 N 0 + ZnSO, + 2N, Cds t 4 N 0 CdSO, + 2N, CuS A 4 N 0 + CuSO, A 2N2 Ag,S + 2 N 0 -+ Ag,SO, I 2N, I/,AI,S3 + 4 N 0 + ' / ? A 1 2 ( S O I )+I 2N, 4H2 4 N 0 + 4H,O I- 2N, C a s + 1 . 5 0 , CaO I- SO, CaS + 2 0 , + CaSO, C a s + 3 N 0 + CaO + SO, 1. 1.5N2 N, C a s t 2 N 0 2 CaSO, C a s + 8 N 0 + CaSO, I- 4N,O +
J
--f
-.
T
298 K
500 K
1000 K
1500 K
284.3 261.4 - 253.6 - 243.7 - 242.0 - 229.3 - 220.3 289.9
264.3 -241.0 - 234.2 224.5 - 222.8 - 209.5 200.9 - 270.3
--215.7 190.0 - 186.1 177.0 - 174.8 - 160.2 - 152.4 - 221.1
166.7 --139.2 -- 138.0 129.5 -- 116.8
-301.3 - 102.0 201.5 - 164.3 - 226.2 - 268.2
-
290.4 98.3 - 184.1 - 158.9 -~219.6 - 231.8
-263.2 - 89.1 -141.1 145.4 - 195.9 - 141.6
-
-
uous analyzer or the vent hood. Stainless steel was used throughout the reactor system. A typical test was made by placing 2 to 20 g of the powdered sulfide or sulfide-catalyst mixture, or sulfide impregnated on high surface area support, in the reactor and the test gas fed to the reactor at a constant rate while the reactor was heated. Samples of the exit gas were taken periodically in a gas-tight syringe and analyzed by gas chromatography. In some of the tests, rates of reaction were compared using a Cahn R-100 continuous recording electrobalance. For these tests the sulfide sample was placed on the weighing pan of the balance which was suspended in a heated Flo-thru Vycor tube. The test gas was introduced through a rotameter and the weight change of the sulfide continuously recorded as the gassolid reaction proceeded. The slope of the weight vs. time curve represents the rate of reaction and can be calculated graphically, or by use of the time derivative computer which was also part of the system. Since the rate of reaction gradually decreases with time, probably because of diffusional resistance through the outer reacted layer of solid, the weight change during the first hour was usually used to determine the reaction rate for comparison purposes. Detailed descriptions of the experimental apparatus have been published (Hodgson, 1978). Analyses. Depending on the test gas and the reaction conditions, the exit gas could contain varying amounts of Oz, N2, NO, NO2, N 2 0 , SOz, and COz, and thus the analytical problem is quite complex. For much of the preliminary work, only a gas chromatograph was available for analysis and this put limitations on the specific combinations of compounds that could be quantitatively determined. Thus, for much of the work reported in this report, a test gas containing NO in helium was used, and the reacted gas stream was analyzed for Nzto determine reduction of the NO. A two-column system was used for most of the gas analysis in a Varian Aerograph Model 1420 low volume thermal conductivity chromatograph. The columns and conditions listed in Table I1 were used. In the first (internal) column, Nz and NO emerge as one peak while NzO and SOz are eluted separately. Nz and NO are separated in the low-temperature (external) column which was enclosed in a Dewar flask containing ice. Equal gas volumes are injected into each column using a gas-tight syringe and the chromatograph polarity switched manually at the proper time to obtain both sets of peaks. During the late stages of the project a Thermo-Electron Model 10A Chemiluminescent NO, analyzer became
-
--
~
~
298 K
104.0 -~171.8
313.5 291.8 282.3 - 273.1 270.8 258.8 249.3 - 319.5
~- 236.1
-
79.8 --98.1 - 131.9 - 172.2 -- 51.4
-
~
-
-
110.8
-
--
-~
-
~
~
317.5 107.6 -- 227.5 - 172.4 -- 243.3 322.0 -
Table 11. Columns and Conditions
column temperature detector temperature He flow
internal column'
external columnb
130 'C 130 "C 10 cm3/min
0 "C 130 "C 1 0 cm3/min
' 4 m X 3 mm stainless steel Porapak Q-S. mm stainless steel Porapak Q.
&I
8m
X
3
available and this was used to determine NO and NO, concentrations, especially when working with NO, concentrations below lo00 ppm when O2 was also present. N2 and O2 concentrations were determined using a molecular sieve 13X column in the chromatograph. Reacted solids were analyzed for sulfate using simple barium precipitation techniques. Experimental Results Preliminary Screening. For the preliminary tests, approximately 20 g of the sulfide were placed in the reactor and then pure NO was passed through the reactor a t a rate of about 0.3 L/h while the reaction zone was being heated. Samples of the exit gas were taken periodically and analyzed by the gas chromatograph for NO, N20, and Np The temperature a t which NzO or N2 was first detected was noted and the heating was continued until complete conversion (within the sensitivity of the chromatograph) of the NO to Nz was obtained. The reactor was then purged with helium while cooling and then the reacted sulfide was weighed and when positive weight gains were noted, tested for sulfate by the barium precipitation method. The results of these preliminary tests are presented in Table 111. As can be seen from these data, virtually all of the metal sulfides tested resulted in the reduction of pure NO to N2 but the temperature range in which the reaction proceeds varies widely. Actually, the temperature a t which the reaction will proceed also probably depends on a number of uncontrolled variables such as particle size (surface area), bed packing, and fluid-particle dynamics, but these data give an indication of overall reactivity of the various sulfides and were sufficient for the preliminary tests. Many of the sulfides reacted with pure NO to form a t least some sulfate, but analysis for other possible reaction products was not carried out and other oxidation products are also possible. A12S3,Sb2S3,BiS, CuS, FeS, MnS, HgS, MoS2,KzS, WS2, and T12S all lost weight during reaction. Visual inspection showed that considerable elemental mercury was formed when HgS was tested, but the final form of the other materials was not considered further.
110
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980
Table 111. Preliminary Tests of Various Metal Sulfides for N O Reduction reducing agent A&S, SblS, Bas BiS CdS Cas
cus cu,s
FeS PbS hlnS HgS MoSl SrS K,Sa
WS ZnS NIS
T1,S a
temp of temp for 100% react. initn, C convn 250 200 475 350 650 450 250 450 300 250 475 450 400 400 100 250 550 90
-
400 550 650 600 800 650 500 700 525
550 550 650 500 650 450 500 800 360 -
wt change loss loss gain loss gain gain loss gain loss gain loss loss loss gain loss loss gain gain loss
acted layer. The solid reaction product analyzed 87.2% CaS04. sulfate test
? I
4
t i
t
t
Sulfurated potaSh
Al2S3 is unstable
and reacts with atmospheric water to form H2S and, as a result, probably would not be practical for use under any conditions. Based on the success of the initial tests, several of the sulfides which appeared to have promise were tested further to see if the reduction reaction could be catalytically promoted. The first candidate catalysts tested were Fe203and Cr203since these have been reported to increase the rate of reduction of NO with carbon monoxide (Bartok et al., 19691, but the catalytic effects of these materials were quite small. However, the addition of about 5 w t 70CoC12 or FeC1, to CaS resulted in a pronounced (about 200 "C) decrease in the temperature at which the reduction reaction proceeded. Based on the success of these tests, other compounds were tried on a trial and error basis with the tests being limited to materials on hand. For these tests 20 g of powdered sulfide were physically mixed with 20 wt % of the candidate catalyst using a mortar and pestle. Again, the test gas was pure NO a t a rate of 0.3 L/h. The important results of these tests showed that the addition of CoCl,, NiC12,FeCl,, and K3CrF6to CaS significantly reduced the temperature at which the reduction reaction proceeded while Fe203, Cr203, CaCl,, CuCl,, FeS04, KzZrF6,PdC12,and PtC12had little or no effect on the reaction. K3FeF6and NiC12 also reduce the reaction temperature with Bas, SrS, ZnS, MoS2, and FeS. NaCl and CaC1, both appeared to enhance the reaction with BaS while CuCl appeared to exhibit a catalytic effect with Cu2S, MoS2, and FeS. A violent, highly exothermic reaction occurred when argentic (silver) fluoride, AgF2, was mixed with Cas, and so this material was not tested further. These tests showed that a t higher temperatures, when using some of the promoters, another peak showed up in the gas analysis using the internal column. This was identified as SO2. Further tests were carried out to better define the conditions for SO2formation but significantly, the reduction reaction proceeds at temperatures below the temperatures a t which SO2 is produced. To further characterize the solid product of reaction, a run lasting 9 days was made using calcium sulfide. Pure NO was passed through the reactor while it was maintained at temperatures between 450 and 700 "C. After 9 days, Nz was still being produced but a t a very low rate thought to be limited by diffusion through the outer re-
Discussion of Preliminary Results The preliminary results were very interesting and encouraging. Nineteen sulfides were tested for NO reduction and all gave positive results. Of these, Bas, Cas, CdS, Cu2S, NiS, PbS, SrS, and ZnS all gained weight during reaction indicating that undesirable side reactions producing gaseous sulfur compounds may not be occurring. The alkaline earth sulfides Bas, Cas, and SrS looked particularly attractive. Of these, CaS probably would be preferable because of availability and cost. In addition, CaS can easily be regenerated from the sulfate by reduction with CO, H2,or coke (George et al., 1968; Zadick et al., 1972). The temperature required for the reduction reaction to proceed was lowered considerably by mixing certain materials with the powdered sulfide. This effect may or may not be catalytic in nature and the mechanism is not known. It is possible that the promoter takes part in the reaction and is chemically altered and thus not a true catalyst. The most active materials found were metal chlorides and fluorides with K3FeF6,NiC12, NaF, CoCl,, and FeC1, being the most active. It is interesting to note that certain metal chlorides form metal complexes (FeNOCl,, for example) with NO (Partington, 1961). The 9-day run with CaS indicated that diffusion through the outer layer probably controls the rate of reaction after the sulfide on the outer surface has reacted. This indicates that a fluidized bed of very fine particles or sulfide impregnated on a high surface area support may be necessary to obtain high rates of reaction and efficient use of the reactant. Also, although not investigated in the preliminary screening, sulfide consumption by free oxygen may limit the use of sulfides if it were employed in a flue gas containing oxygen. Hence, subsequent tests investigated the relative rates of reaction of NO and O2 using a continuous recording electrobalance and the most promising sulfide-catalyst combinations were tested using a synthetic flue gas containing 0,. Finally, tests were made using CaS impregnated on high surface area supports. Rate of Reaction of Metal Sulfides with NO and 0,-Electrobalance Studies. For these tests, 0.8 g of the powdered sulfide was distributed evenly on the weighing pan of the balance and the weight change was noted when a test gas mixture of either 2.5% NO or 02 in helium was passed through the hang down tube at a rate of about 13 cm3/min. Temperatures of 300,400, and 500 "C were investigated with a fresh sample of sulfide being used for each temperature and each gas mixture. The reactor was purged with pure helium while heating the reactor to the desired temperature and until the weight remained constant (due to moisture being removed from the sample). Table IV is a summary of the rate of weight gain or loss for the various sulfides listed. In these tables, the negative sign indicates a weight loss when the test gas containing NO or 0, is passed through the reactor tube while the "d" indicates a weight loss while purging with pure He a t that temperature indicating a decomposition. As can be seen from data, the metal sulfides react faster with oxygen in every case except one. At 300 "C the rate of reaction of SrS was slightly faster with NO than with O2 BaS, FeS, and CaS atl lost weight when NO was passed through the reactor a t 300 "C indicating the possibility of undesirable side reactions taking place. Analysis of the dilute exit gas stream from these runs was impossible, so these possible side reactions were not investigated further. Bas, FeS, SrS, CdS, Cas, and K2S all reacted both with
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980
Table IV. A Summary of the Rate of Weight Gain or Loss for the Various Sulfides Reacting with NO and 0,' rate x 300 " C sulfide
NO
Bas FeS ZnS SrS
2.08 2.27 1.50 2.22
cus
d
CdS cu,s PhS
0.98
ws,
MoS, K,S. T1S Cas MnS
2.53 0 0 38.8 d - 0.49 d
l o 3 , g of
0, 0 20.80 1.77 2.15 d 3.80
NO 4.09 1.17 0 2.73 d 3.62
2 GR C A S I N S 4 0 0 'C FEED GAS: 1 5 2 MLlMlN 1000 P P U NO
9
500 "C NO
O?
1SI. 18 " I n
53 27.58 4.53 5.29 d 5.50
182.00 144.0 7.97 7.81 d 27.6 8.07 108.00 -25.20 10 20
32.40
(I
100
promoter N O 0, N O
0,
NO
O2
I
I
5
I
10 Ti I.'E.HOdRS
500 "C
NO
O2
2 GR CaSrNiC12 400 'C FEED G A S : 1000 100 MPLFlNM lNo N
NO and 02. MnS, TIS, CuS, and MoS2 all decomposed in the helium purge. WS2 lost weight when reacting with both O2 and NO. ZnS did not react with NO a t 400 and 500 "C but did react with 02. This is in line with the preliminary tests which showed that NO does not react with ZnS a t temperatures below about 550 "C. The fact that the rate of weight change when the sulfides react with O2 is faster than with NO is slightly misleading since it takes twice as much NO as O2 on a molar basis (assuming sulfate is formed) to produce the same weight change. On this basis, the rate of reaction of NO is greater than O2 for SrS a t 300 and 400 "C and BaS and CdS a t 400 "C. Also, significantly, there is little difference between the rates of reaction for NO and O2 for SrS at all temperatures. The next experiments investigated the relative rates of reaction of CaS with NO and O2 when the sulfide was mixed with NaF or K3FeF6. These data are shown in Table V. As can be seen from Table V, the presence of NaF or K3FeF, has a pronounced effect on the rate of reaction of CaS with both NO and O2 but there is a greater increase in the rate of reaction of NO, especially a t the lower temperatures of 200 and 300 "C. T o test the possible catalyst for reactivity with NO and 02,a series of runs were made using the pure promoter on the electrobalance. NaF, NiC12, FeCl,, CaCl,, and Fe203 showed no weight change a t 300 or 400 "C with either NO or O2 but K3FeF6 reacted with NO a t all temperatures investigated thus eliminating it from further consideration as a catalyst. In addition, it undergoes a color change from white to brown as it is heated in helium indicating that it is unstable a t higher temperatures. Reduction of No, in a Synthetic Flue Gas. The reduction of NO, in a synthetic flue gas was investigated
1 'I.
80
o2
18 * I O CQ BALANCE V2
70
\
0
2
63
CK
NaF 3.35 0 1 . 9 3 1.02 5.99 1 8 . 5 43.5 73.9 2.99 0 3.73 1 . 3 9 10.80 21.60 25.9 24.2 K,FeF, pure 1 . 6 4 2.56 3.20 5.71 -no reactionsulfide
15
I
90
~-
400 " C
\'
Figure 1. The percent NO, and O2 reacted by CaS/NaF at 400 "C.
rate x 10' [g-mol of gas reacted/(min)(gof initial C a s ) ] 300'C ___
r
01
Table V. Rate of Reaction of Cas at Various Temperatures Using Different Promoters
200 'C
'i
0
Notes: a minus sign represents a weight loss; d indicates weight loss while purging the reactor tube with helium.
~
co,
01
7.22 70.9 0 3.68 d 2.43 0 8.26 3.52 31.70 1.59 - 3.08 8.33 -22.90 -6.54 d d 185.00 79.8 88.70 127.00 d d 0 2.08 6.50 3.85 d d
-
100
wt changeimin
400 'C
-
111
350 cr
-
O 40-
$ 2 302G
-
10
I
0;
I
I
I
2 3 TiME ,HOURS
4
Figure 2. The percent NO, and O2reacted by CaS/NiCI2 at 400 "C.
in the reactor system previously described. Two grams of the sulfide-catalyst material was added to the reactor and the test gas mixture passed through the sulfide bed a t a rate of 100 mL/min. The test gas contained 1000 ppm of NO,, 1% 02,18% COS, and the balance N2. A concentration of this magnitude would be expected in a flue gas from the burning of coal using about 10% excess air. Although most flue gases probably would contain more O2 than this, 1%was chosen because it represented the lower limit likely to be encountered and reasonable run times could be obtained. Sulfide-catalyst combinations that looked to be the most promising in the electrobalance studies were investigated a t 400 "C. Mixtures of Bas, CaS, SrS, and FeS with NaF, NiC12, CoCl,, FeCl,, and Fe203 were investigated using 20 w t % promoter mixed with the sulfide. NO, analysis was accomplished using the chemiluminescent analyzer and O2 concentrations were determined using the molecular sieve column in the chromatograph. The results of the tests with Cas, which are typical of the data obtained, are shown in Figures 1 through 5 . As can be seen by comparing these figures, the various sulfide-promoter mixtures behaved quite differently under the test conditions, but most important, in certain cases NO, removal was a t a high level for significant periods of time even in the presence of 02.
112
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 1, 1980
Table VI. The Capacities of the Metal Sulfide-Catalyst Mixtures and of the Metal Sulfides
-_
[(total wt of NO reduced to 600 p p m ) / ( w t of MeS initially present)] metal sulfide
NaF
Cas SrS Bas FeS
18.6 31.7 6.3 1.9
NiC1,
400 "C COCI,
FeC1,
Fe,O,
unpromoted metal sulfide, 400°C
5.5 1.4 4.8 13.6
0.9 2.1 2.0 2.2
13.9 7.2 16.8 31.2
1.0 1.1 3.4 1.3
0 0 2.3 0.5
X
lo3 300 "C
NaF
FeCI,
0.1 2.0
100 2 GR. U S / C o C I 2 400 FEED *CG A S :
-
80
0
3
2
1
4
TIME, HOURS
TIME,HOURS
Figure 3. The percent NO, and O2 reacted by CaS/CoCI2at 400 "C.
Figure 5. The percent NO, and O2 reacted by CaS/Fe203at 400 "C. 100 r
1 n
:
90
I80
:
0 01. g 2 GR CaSrNnF
80 0 W
-
70
0
$
60-
LT
d
-
50
02
0
2 GR CaS/FeCI2 400 * c FEED GAS: 100 MLlMlN W P P M NO 1 .I. 9 18.1. c q WANCE Up
2
FEED GAS! 00 MLiMlN 1 0 3 3 PPM NO 18 ' I . c 0 2 BALANCE N2
70
-\
E
6 z p
40-
-
30
3 20 0
0
K
10
*O 10 0
1
1
2
3 TIME .HOURS
0 4
0
1
5
I
10 TIME, HOURS
15
Figure 4. The percent NO, and O2 reacted by CaS/FeC12at 400 "C.
Figure 6. The effect of O 2 concentration on NO, removal.
The capacity (defined as the weight of NO, reduced per unit weight of sulfide initially present from the start of the run until the NO, concentration at the reactor outlet exceed 600 ppm) was calculated for each combination tested. This gives a quantitative indication of how good the mixtures are for NO, reduction. These are listed in Table
shows the effect of the amount of 0, in the gas stream on NO, reduction. As can be seen, increasing the amount of O2 also dramatically decreases the capacity of the CaSNaF mixture for NO, reduction. Finally, two runs investigated the removal of NO, from the synthetic flue gas by 5% C a s (only) impregnated on the high surface supports Harshaw 1602-T and Linde TM-0-1114. The Harshaw pellets were in the form of 3-mm cylinders and had a surface area of about 240 m2/g. The Linde molecular sieve was in the form of 1.5-mm extrusions and had a surface area of about 350 m2/g. The results are shown in Figure 7 . As can be seen, perforniance curves of the pellets are much better than the best sulfide-promoter mixtures in bulk form. The capacity of the pellets for NO, reduction was calculated to be 0.76 and 0.91 g of NO/g of initial sulfide in the pellets for the Harshaw
VI. As can be seen from Table VI, the capacities varied from 0 to 37.2 X g of NO, reduced/g of sulfide initially present. The six best were FeS-FeC1, > SrS-NaF > Cas-NaF > BaS-FeC12 > FeS-NiC1, > Cas-FeC1,. The FeS-FeC1, and Cas-NaF mixtures were also tested at 300 and 200 "C but both showed a low capacity a t 300 "C and no detectable reaction at 200 "C. This was disappointing since the electrobalance study showed that a t 200 "C the NaF-CaS mixture reacted with NO but not 02.Figure 6
Ind. Eng.
01 0
I
5
I
10 TIME ,HOURS
I
15
Figure 7. The removal of NO, by Harshaw pellets and Linde molecular sieves impregnated with 5% Cas.
and Linde pellets, respectively, over 20 times that of the best bulk sulfide-catalyst mixture. Discussion Of the 19 metal sulfides tested for NO reduction only eight (Bas, CdS, Cas, Cu2S, PbS, SrS, ZnS, and NiS) showed a weight gain. Of these, the alkaline earth sulfides (Cas, B a s , SrS) were thought to have the most promise because of the stability of the sulfides and corresponding sulfates. Other factors being equal, CaS would be the most desirable sulfide because of its potential abundance and low cost. It can readily be produced from gypsum by reduction using a number of reducing agents including coke, carbon monoxide, and hydrogen (George et al., 1968; Zadick et al., 1972). Because of these considerations, much of the research effort concentrated on using C a s as the reducing agent. However, some of the data in the research program indicated that SrS may be superior to C a s when reducing NO in the presence of 02.In addition, FeS was investigated further because of its abundance and high rate of reaction even though SO, was formed in the reduction of NO by FeS. The rate of reaction is significantly increased by intimately mixing certain materials with the sulfides. K3FeF6 was one of the most active promoters but electrobalance studies showed that it reacted with NO a t temperatures between 200 and 500 "C resulting in a weight gain. NaF, NiC12, FeCl,, CoC12,and Fe203did not react with either NO or O2 at 300 and 400 "C. These materials significantly increased the rate of reduction with some of the sulfides a t these temperatures and thus the action appears to be catalytic in nature. The electrobalance study showed that Fe203mixed with FeS, SrS, and Cas significantly increased the rate of reaction with NO in the absence of O2 but results using the synthetic flue gas containing 1% O2 were poor with the sulfide-Fe203 mixtures. This indicates that O2 may interfere with any catalytic activity exhibited by Fe203under these conditions. A very active form of sulfide is obtained by impregnating it on a high surface area support but these are so active that there is a rapid oxidation of the pellets in air at room temperature. The reaction of the sulfides with O2 is very important in determining the usefulness of the sulfides for the control of NO, emissions from flue gases. Flue gases from power plants typically would contain much more O2 than NO,
Chem. Process Des. Dev., Vol. 19, No. 1, 1980
113
( 0 2 / N 0 ratios as high as 50:l are probable) and so when O 2 reacts with the sulfide most of it would be consumed by a nonpollutant. Tests of Cas-NaF at 400 "C increasing the O2concentration to 10% greatly decreased the capacity of the mixture for NO, reduction. However, the electrobalance studies indicated that, for Cas-NaF mixtures a t 200 "C, NO reacted a t a finite rate while no reaction with O2 was detected. This could not be verified using the Cas-NaF mixture in the tubular reactor with the simulated flue gas, however, probably because the reaction rate was too low to detect a change in NO, concentration under the conditions of the test. Apparently the reaction rate is higher a t the higher NO, concentration used for the electrobalance tests. The tests using the synthetic flue gas indicate that each sulfide-catalyst mixture behaves quite differently. The capacity (defined as the weight of NO reduced per unit weight of sulfide initially present from the start of the run until the exit concentration exceeds 600 ppm) was quite low for all of the bulk sulfide-catalyst mixtures under the conditions of the tests using the synthetic flue gas test mixture. The two highest were 0.037 and 0.032 g of NO/g of sulfide for FeS-FeC1, and SrS-NaF mixtures, respectively. The best C a s mixtures were 0.019 and 0.013 g of NO/g of sulfide for Cas-NaF and CaS-FeC12 mixtures, respectively. These compare with the capacity of the 5% C a s (only) high surface area pellets of 0.76 and 0.91 g of NO/g of sulfide in the Harshaw and Linde pellets, respectively. Thus, it appears that the capacity of the bulk sulfide beds is being limited by poor gas-solid contacting. For comparison, assuming the reaction Cas + 4N0 CaSO, + 2N2
-
the maximum capacity for C a s would be 1.66 g of NO reducedlg of sulfide. The pellet tests indicate that the capacity can be greatly increased by impregnating the sulfide on high surface area supports and it may be possible to further increase it by including a catalyst on the pellets. Reaction in a fluidized bed reactor should also greatly increase the capacity if the bulk material were used. Acknowledgment Many individuals were very helpful in carrying out this project. Montana State University Chemical Engineering students Russell Erickson and John McIntyre contributed greatly to the experimental investigation. In addition, the invaluable assistance of the late Silas Huso and Jim Tillery, who constructed much of the laboratory equipment, is gratefully acknowledged. The work was carried out in fulfillment of Grant No. R800682 by Montana State University under the partial sponsorship of the Environmental Protection Agency. Literature Cited Bartok, W.. Crawford, A. R., Hall, H. J., Maury, E. H., Skopp, A,, "Systems Study of Nnrogen Oxide Control Methods for Stationary Sources", Clearing House for Federal Scientific and Technical Information, PB 184479, May 1969. Environmental Protection Agency Publication No. Ap-67, "Air Quality Criteria for Nitrogen Oxides", Jan 1971. Fulton, C. H., "Metallurgical Smoke", U.S. Bureau of Mines Bulletin 84, 1915. George, D., Crocker. L.. Rosenbaum. J. B., "Current Research of the Production of Sulfur from Gypsum at the Salt Lake City Metallurgy Research Center", US. Bureau of Mines, 1968. Hodgson, K., Ph.D. Thesis in Chemical Engineering, Montana State University, June 1978. Partington, J. R., "A Textbook of Inorganic Chemistry", 6th ed. p 544, MacMillan. London, 1961. Young, S. W.. Trans. AIChE, 8, 81 (1915). Zadick, T. W., Zavaleta, R., McCandless, F. P., Ind. Eng. Chem. Process Res. Dev., 11, 283 (1972).
Received for review December 26, 1978 Accepted August 3, 1979