Ind. Eng. Chem. Prod. Res. Dev. 1982, 21, 48-52
48
Becke, F.; Swaboda, 0. P. (BASF) Ger. Offen. 1 117 121; CI. Int. C07C,d, Jan 8, I960a. Becke, F.; Swaboda, 0. P. (BASF) Brevet francais No. 1250 165; CI. Int. BOIJCOIC, Nov 28, 1960b. Becke, F.; Swaboda, 0. P. (BASF) Brevet francais No. 1586750; CI. Int. C07C. July 8, 1988. Courty, P.; Ajot, H.;Marcilly, C.; Delmon, B. Powder Techno/. 1973, 7 , 21. Muench, W.; Ruoti, V.; Slhrestrl, G. Commonwealth of Australia, PS 214, 473 (Apr 14, 1958).
Rodriguez, M. V. E.; Delmon, 6.; Damn, J. P. “hoceedings, 7th International Congress of Catalysis”; Selyama, T., Tanabe, K., Eds.; Kodansha and Elsevier: Tokyo and Amsterdam, 1981; pp 1141-1153. Ruger, C.; Schwetlich, K.; Krammer, H. 2.Chem. 1974, 14, 152.
Received for review January 23, 1981 Revised manuscript received June 15, 1981 Accepted August 31, 1981
Deposition of Ammonium Bisulfate in the Selective Catalytic Reduction of Nitrogen Oxides with Ammonia Shlmpel Matsuda,’ Tomolchl Kamo, Aklra Kato, and Fumito Nakajlma Hitachl Research Laboratoty, Hitachi Ltd., Hitachkshi, Ibarakkken, 3 19- 12 Japan
Teruo Kumura and Hlroshl Kuroda Kure Work, Babcock-Hitachi K.K., Kure-shi, Hiroshima-ken, 737 Japan
The selectiie catalytic reduction of NO, with ammonia in the presence of SO,, especially sulfur trioxMe (SO,),has been investigated. Although a catalyst composed mainly of titania Is resistant to the SO, poisoning, the catalyst activity decreases to ahnost zero in the presence of NH, and SO, below a certain temperature due to the deposition of sulfate of ammonia. A gas mixture containing varied amounts of NH, and SO, was pass& through a packed reactor placed in a fumace with a temperature gradient. It was found that ammonium bisulfate was the only product above 240 OC. The vapor pressure of ammonium bisulfate was found to be expressed as Pw;Pym = 1.14 X lo1* e x p ( 4 3 000/RT) (P in atm). In the actual NO, removal process the deposition of ammoniumbisulfate occurred at a higher temperature than that expected from the vapor pressure due to the capillary condensation of ammonium bisulfate in micropores of the catalyst. In the presence of NH, (200 ppm) and SO, (10 ppm) the catalytic activity decreases to zero below 282 O C considering the distribution of micropore radius of the catalyst.
Introduction Nitrogen oxides (NO,) from stationary combustion facilities such as power plant boilers comprise a considerable part of the total NO, emitted to the atmosphere. Several methods for the control of NO, have been proposed and tested using pilot plants (Bartok et al., 1969). It has been found that the selective catalytic reduction (SCR) process is most feasible for industrial application. By the end of 1980 several tens of commercial plants based on SCR process have been constructed in Japan, the largest one treating 2 million normal cubic meters per hour (Nm3/h) of flue gas at a power station (700 MW) (Kuroda and Nakajima, 1978; Nakajima et al., 1979). It has been known that NO, are selectively reduced by NH3 in the presence of a large excess of oxygen over various catalysts. The reaction is expressed as (Matsuda et al., 1978; Kasaoka et al., 1977; Kat0 et al., 1980)
NO
+ NH, + ‘/40z= Nz + 3/2Hz0
(1)
A catalyst used in a commercial plant must possess high activity and selectivity, since volume of flue gas to be treated is extraordinarily large. In addition the catalyst must be resistant to the SO, poisoning, since sulfur dioxide (SO,) and sulfur trioxide (SO,) are usually contained in an oil or coal-fired boiler flue gas. In the early stage of the SCR process development several catalysts, for example, V, Mo, and W oxides supported on A1203 carrier, and FepOs based catalysts, were tested. The life of these catalysts was found to be short because they were susceptible to the ois6-432i/a2/i221-004a$o 1.2510
SO, poisoning. A series of catalysts consisting mainly of titania have been developed. The Ti0,-based catalysts show a high activity, selectivity, and resistance to the SO, poisoning over a wide range of temperatures, 200-450 “C (Nakajima et al., 1979; Matsuda et al., 1978). In a boiler flue gas SO3 is contained in a few percent, usually 2-5%, of the total SO,. This renders a serious problem for the selection of the reaction condition of the SCR process because SO3 reacts with NH3 to form ammonium bisulfate (NH4HS04,denoted ABS in this paper) in the temperature range. Deposition of ABS on the surface of catalyst occurs below certain temperature depending on the concentrations of NH3 and SO,. We have experimentally determined the vapor pressure of ABS and established the reaction condition (temperature) of the SCR process according to the gas composition. The vapor pressure of ABS can not be measured directly. Therefore, it is defined as a product of the vapor pressure of NH3 and H8O4which are in equilibrium with liquid ABS. We have observed’that the decrease of catalytic activity by ABS deposition occurs at a higher temperature than that expected from the ABS vapor pressure due to the capillary condensation in micropores of the catalyst. Experimental Section Catalyst. The Ti02-based catalyst which consists of more than 70 atomic % TiOz and the remainder second component was used in the present study. The second components are selected from transition metals in group 5B (V), 6B (Cr, Mo, W),8 (Fe, Co, Ni), and 1B (Cu), and 0 1982 American Chemical Society
Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 1, 1982
49
Nz
Figure 1. Experimental apparatus.
added in 1-30 atom %. The main component Ti02 is considered to be not only a support but also an active component, since the Ti02-based catalyst has much higher activity than A1203-basedcatalyst with the same second component . Powder of the catalyst mixture was molded into 4-6 mm diameter sphere pellets by the rotating-molding method. The pellet was calcined in an electric furnace at 500 "C for 2 h. The specific surface area (BET method) and pore volume (mercury porosimeter) were found to be 85 m2/g and 0.390 mL/g, respectively. Reaction Apparatus. The catalytic activity on the NO-NH3 reaction was studied in a flow system illustrated in Figure 1. A vertical 36 mm inner diameter quartz tube reactor which contained 20 mL of the catalyst was heated to 200-600 "C with an electric furnace. A simulated flue gas containing 200 ppm NO, 240 ppm NH3,500 ppm SO2, 3% 02,12% C02, 12% H20, and the remainder N2 was passed through the catalyst bed at a rate of 200 NL/h, producing a gas space velocity of 10000 Nm3/(m3 h). When sulfur trioxide was contained in the gas mixture, a dilute sulfuric acid solution was pumped into the upper region of the reactor and vaporized at about 550 "C. Measurement of Ammonium Bisulfate Vapor Pressure. The vapor pressure of ABS was determined by measuring the temperature of ABS deposition from NH3 and SO3 (or H2S04)gas. The reaction tube for the measurement of ABS vapor pressure is shown in Figure 2. At 17 mm i.d. and 900 mm length reactor was heated by three electric furnaces to produce temperature gradient along the reactor. The uppermost part of the reactor was heated to 550 "C to evaporate the dilute sulfuric acid solution. The lower part of the reactor was composed of 12 quartz plates placed every 50 mm and quartz wool between the plates. The temperature of the top plate was set about 360 "C and the bottom one at 100 "C. A gas mixture containing varied concentrations of NH3 and SO3 was passed through the reactor. Ammonium bisulfate was deposited on the quartz wool, plate, and wall of the reactor. After passing the gas mixture for 20 h, the plates and wools were removed from the reactor very carefully and supplied for the analysis of NH4+and SO:-. Each set of plate and wool was washed with 200 mL of water. The concentration of NH4+and SO4" in the solution was determined by the Nessler method and the precipitation method using barium chloranilate, respectively. Results and Discussion Vapor Pressure of Ammonium Bisulfate. Condensation of ABS may be represented by the reactions NH3 + SO3 + H20 = NHIHSOd(1) (2) AHoZg8= -78 kcal/mol NH3 + H2S04 = NH4HS04(1) (3) = -53 kcal/mol depending on SO3 or H2S04is dominant in the gas phase
Electric Furnace
Quartz plate Quartz wool
Gas o u t
-SiIU Thermocouple
Figure 2. Apparatus for measurement of ammonium bisulfate deposition.
under the reaction conditions. Equilibrium between SO3 and H2SO4 is, therefore, considered first.
SO3 + H 2 0 = H2S04
(4)
= -25 kcal/mol
K = W2SO41/([So31[H,OI) (5) Equilibrium constant K can be calculated from AF of reaction 4 using thermodynamic data given in the literature (Stull and Prophet, 1971). The K values are found to be 3.57, 18.6, and 138 at 350, 300, and 250 "C, respectively. In the gas mixture containing 12% H20the [H&304]/[S03] ratio is 0.43, 2.2, and 16.6 at 350, 300, and 250 "C, respectively. From the chemical kinetic point of view, the rate of SO3 association with H20 of form H2S04 seems to be very fast, taking into account the heat of reaction 4, -25 kcal/mol. Therefore, H2S04 is the predominant species below 300 "C under the present conditions. Thus we have adopted reaction 3 for the condensation and vaporization of ABS in the discussion given below. The vapor pressure of ABS was determined by measuring the temperature at which ABS was deposited at certain concentrations of NH3 and HzS04using the apparatus shown in Figure 2. Five eFperiments were performed using a gas mixture containing NH3 and H2S04by amounts (833, 100), (83.3,100), (83.3, lo), (30,65), and (30, 10) (units in ppm). Distribution of deposits in one experiment is given in Table I. Amounts of NH4+and S042deposited in each region of temperature are given in units of millimoles per normal cubic meter of gas passed through the reactor. The ratio of NH4+/S04z-is also listed in the table, showing clearly the formation of ABS (NH4S04) above 240 "C.In the regions below 241 "C the NH4+/S0tratio was less than unity, since the gas mixture contained less NH3 than SO3. The total amounts of deposited NH4+ and accounted for 91% of NH3 and 93% of SO3 in the gas mixture, respectively. The deposition of ABS is first observed in region (323-291 " C ) and maximum de-
Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 1, 1982
50
Table I. Distribution of Deposited Ammonium Bisulfate a Temperature NHi/SOi-
Region
(TC)
*
Amount of Deposition (mmol/ Nm3)
(mole ratio)
0
i,O
05
1,5
ppre
2,O
I
I
-
1
356-323
2
323
0 91
3
- 291 291 - 265
4
265- 241
0 95
1I
5
241
-
222
ENsot-H ~
1 00
I
0 77
6
2 2 2 - 203
0 26
7
2 0 3 - 183
0 14
8
183-158
0 i1
Table 11. Vapor Pressure Decrease of Ammonium Bisulfate in Micropores
I I
r
I
PlPe g
radius, A
200 "C
300 "C
400 "C
400 200 100 50 25 20 10
0.88 0.78 0.61 0.37 0.14 0.085 0.0073
0.90 0.82 0.67 0.44 0.20 0.13 0.017
0.92 0.84 0.71 0.50 0.25 0.18 0.031
a Pes = equilibrium vapor pressure of NH,HSO,. vapor pressure of NH,HSO, in micropores.
P=
-"
io-'
-
5
10-8
-
c
10-9-
1
-E
2 10-10
-
\
\ \
15
16
17 1000/1'
18
19
20
21
'h-' I
Figure 3. Vapor pressure of ammonium bisulfate.
200
position in region 3 (291-265 "C). From the distribution of ABS deposition we guessed that the temperature at which ABS is condensed at NH3 83.3 ppm and H$04 100 ppm is about 300 "C. In the same manner, the vapor pressure of ABS was estimated from five experiments and plotted against T 1 in Figure 3, where the temperature gradient was set to the heat of vaporization of ABS (53 kcal/mol) considering the Clausuis-Clapeyron equation. The vapor pressure of ABS reported by Ikeda and Koyata (1978) is also shown by a dashed line which is about a factor of 10 higher than the present study. Their temperature gradient is much higher than ours, since they employed eq 2 for ABS condensation = -78 kcal/mol). We assume that the use of eq 3 is justified, because SO3 exists as H$O4 gas in the temperature range of interest. From Figure 3 the vapor pressure of ABS is represented by
P N H ; P ~=2 1.41 ~ ~ ,X
exp(-53000/RT)
(6)
Capillary Condensation of ABS in Micropores. It has long been known that a liquid in micropores has lower vapor pressure than that of free liquid, the phenomenon known as "capillary condensation". A catalyst is a porous material whose catalytically active sites are located in micropores ranging from 10 to 100 A in radius. Therefore, the deposition of ABS in catalyst is expected to occur at a higher temperature than that calculated from the equilibrium vapor pressure. According to Thomson's theory of capillary condensation, the vapor pressure (P) in the micropore of radius y is given by P = --2aM 1n (7) peq PTRT
250 Tempe r a t u r e
300
350
("C 1
Figure 4. Vapor pressure of ammonium bisulfate in micropores; effect of capillary condensation.
where Pq is equilibrium vapor pressure, a is surface tension, M is molecular weight, p is density, R is the gas constant, and T i s the temperature (K). Assuming u = 150 dyn/cm, and p = 1.78 g/cm3 for ABS, eq 7 is reduced to
The decrease of vapor pressure ( P / P , ) of ABS as a function of pore radius is calculated and listed in Table 11. The effect of the capillary condensation is more pronounced at the lower temperature. At pore radius larger than 200 A the effect is relatively small, but P/Pw becomes less than one-half at about 50 A. Using eq 7 the vapor pressure of ABS in micropores is calculated and shown in Figure 4. Any catalyst would lose their activity completely, if they are used under such conditions that ABS is condensed on free surface, i.e., [NH,] X [H2S04]is larger than P . A catalyst would lose its activity partly in case [NH3][Hz041 < P , since a part of the micropore is filled with a liquid of d S . Given a pore distribution of a catalyst it is possible to estimate the decrease of catalytic activity due to the ABS condensation under certain assumptions. Thiele has derived an equation relating the rate of reaction in one micropore with the intrinsic rate constant, diffusion coefficient, and pore radius on a first-order reaction which is controlled by reactant diffusion in a micropore (Wheeler, 1955). Based on the Thiele's treatment it is derived that the catalytic activity (overall rate constant) is linearly pro-
Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 1, 1982
51
100,
I
10
50
I
I
100
500
I
I
~
0
10
1000
20
30 Time
(b) Figure 5. Micropore distribution of a sample catalyst. pore radius
40
50
lh)
Figure 7. Change of catalytic activity in presence of SOB. Gas mixture: 200 ppm NO, 240 ppm NH3, 100 ppm SO3, 500 ppm SOz, 3% Oz, 12% HzO, and the remainder Nz. Space velocity: loo00 h-l.
>
Y
250
300
350
400
T e m p e r a t u r e ('C)
I
I
I
1
200
250
300
350
React ion Temperature
('C)
Figure 6. Change of catalytic activity due to ammonium bisulfate deposition.
portional to the pore volume (see Appendix). It has been known that the rate of NO-NH3 reaction is approximately proportional to first order of NO concentration (Imanari et al., 1980). -d[NO]/dt = k[NO]
(9)
In the derivation it is assumed that length of pore is independent of pore radius. We were forced to use the assumption to derive a simple equation, though it was not completelyjustified. The assumption seems to be a rough approximation for a very fast reaction such as NO-NH, reaction which proceeds almost exclusively in very thin layer of the catalyt surface (only less than l/laoof the total thickness is effectively used for a 6-mm sphere catalyst). We have often observed experimentally that the catalyst activity is approximately linearly proportional to pore volume when catalysts with varied pore volume are prepared, An example of pore distribution is shown in Figure 5. Catalyst B was used in the present study. The catalyst had a total pore volume of 0.389 mL/g of which about 80% located below 300 A. When the NO-NH3 reaction is conducted at [NH3][H2S04]less than P ABS is condensed in pores of small radius to large r a i u s in succession. The catalytic activity decreases to a constant value after partial filling of ABS in micropores. The steady-state
Figure 8. Steady-state catalytic activity after ammonium bisulfate deposition: -, initial activity; -o-, experimental (SO3100 ppm); ---, calculated. Gas mixture: 200 ppm NO, 240 ppm NH3, 100 ppm SO3, 500 ppm SOz, 3% Oz, 12% HzO, and the remainder Np Space velocity: loo00 h-l.
activity k relative to the initial activity ko is calculated for catalyst B shown in Figure 5. The result is shown in Figure 6. In the upper-left region the catalytic activity reduces to zero due to the complete coverage of catalyst surface by the ABS deposition. In the lower-right region essentially no decrease in activity will be observed. In case the reaction conditions are between the two regions, the catalytic activity is reduced to a lower level and stays constant after a certain period of time. The change of the catalytic activity on the NO-NH, reaction in the presence of SO3 was studied between 310 and 350 "C. The result is summarized in Figure 7. In experiments performed above 325 "C the NO conversion seems to reach a constant value after 40 h. On the other hand, in an experiment performed at 310 OC the NO conversion seems to be still decreasing at 40 h. The steady-state NO conversion in the presence of SO3, i.e., the catalytic activity after the ABS condensation, is calculated for the case SO3 1, 10, and 100 ppm using the data of Figure 5. The calculated NO conversion as well as the experimental data for the case SO3 100 ppm is shown in Figure 8. The temperature at which the NO conversion reduces to zero is in good agreement, while a considerable discrepancy between the calculated and experimental NO conversion is seen above 320 "C. A number of experiments were performed to find reasons for the discrepancy. The catalyst was treated by the same gas mixture as in Figure 7 except that no NH3 was contained. After 20-40 h the
Ind. Eng. Chem. Prcd. Res. D ~ V .~
52
NO conversion was measured by introducing NH3 into the gas mixture for a short period of time, 10-20 min. It was found that the catalyst lost activity to some extent, larger extent at the lower temperature. The loss of activity could be ascribed to the SO3 adsorption of the active sites of catalyst. The loss of activity was found to be temporary, since the catalytic activity was recovered quickly when SO3 was removed from the gas mixture. Taking into account the poisoning by the SO3 adsorption and the ABS condensation, the calculated activity was in good agreement with the experimental value. An extensive study on the SO3adsorption has been in progress in our laboratory and a quantitative treatment will be reported in the future. Appendix Reaction Rate and Pore Volume. Based on the Thiele model for a catalytic reaction which is controlled by reactant diffusion in micropore, the reaction rate R in one pore of radius r and length 2L is given by
R(r) = C o 2 a r m tanh(h) h =
Ldm
(AI)
(A21
where co is the reactant concentration to which the rate is linearly proportional, k is the intrinsic rate constant per unit area, and D is the diffusion coefficient. In the case of NO-NH3 reaction on the catalyst used in the present study, h is in a range of lo2-lo3, and tanh (h) = 1. Equation A1 is reduced to
R(r) = C 0 2 7 r r m (-43) The mean free path of NO is calculated to be about 1300 A at 350 “C and normal pressure. The transfer of NO in micropores of the catalyst is controlled by the Knudsen diffusion, since the mean pore radius is about 200 A.
D = 9.7 x
1 0 3 ~ m a r ~
22 1 , , ~ s
where T is temperature in K and M is the molecular weight. From eq A3 and A4,
R(r) 0: r2 (A5) On the other hand, pore volume V of radius r is given by V(r)= 2L.7rr2
(A6) Under the assumption that the length of pore L is independent of r, we derive from eq A5 and A6
R(r) a V(r)
(A7) Then, the total rate of reaction on one catalyst particle will be
Literature Cited Bartok, W.; Crawford, A. R.; Cunningham, A. R.; Hail, H. J.; Manny, E. J.; Skopp, A. “Systems Study of Nitrogen Oxldes Control Methods for Stationary Sources”, National Technical Information Services, Springfield. VA, Government Report PB-192-789, 1969. Ikeda, T.; Koyata, K. Karyoku Genshkyoku Hatsuden 1978, 29, 59. Imnari, M.; Watanabe, Y.; Matsuda, S.; Nakajlma, F. Preprints of Contrlbuted Papers, 7th International Congress on Catalysis, 5 9 , Tokyo, Japan, 1980. Kasaoka, S.;Sasaoka, E.; Yamanaka, T. “ y o Kyokaishi 1977, 56, 834. Kato, A.; Matsuda, S.;Kamo, T.; Nakajlma, F.; Kuroda, H.; Nab, T. J . phvs. Chem., 1981, 85. 1710. Kuroda, H.; Nakajima, F. “Some Experlences of NO, Removal In Pilot Plants and Utility Boilers”; presented at 2nd EPRI NO, Control Technology Seminar, Denver. CO, 1978. Matsuda, S.; Takeuchi, M.; HlsMnwna, T.; Nakajlma, F.;Narlta, T.; Watanabe, Y.; Imanari, M. J . Air Pollut. ControlAssoc. 1978, 28,350 (presented at National Meetlng of APCA, Portland. 1976). Nakajima, F.; Hishinuma, T.; Kumura, T.; Arikawa, Y.; Narita, T. “Catalytic Reduction of NO, In Stack Gases”; presented at ACSlCSJ Chemical Congress, Honolulu, HI, 1979. Stuli, D. R.; Porphet, H. “JANAF Thermochemical Tables”, 2nd ed.; Natbnai Bureau of Standards, Washington, DC, 1971. Wheeler, A. “Cetalysis”; P. H. Emmett, Ed., ReinhoM Publishing Corp., New York, 1955; Voi. 11, Chapter 2, p 133.
Received for review M a r c h 31, 1981 Revised manuscript received A u g u s t 31, 1981 Accepted S e p t e m b e r 17, 1981
(A4)
Effect of Sulfur Poisoning on the Hydrogenolysis Activity of Pt in Pt-AIzOa CataIysts P. Govlnd Menon, Guy B. Marln,‘ and Gilbert F. Fromenl’ Laboratorlum voor Petrochemische Techniek, Rijksuniversitek Gent, Krijgshan 27 1, 9000 Gent, Eelglum
Using the hydrogenolysis of n-pentane at 300-380 O C as an indicator reaction, the S poisoning of Pt-AI,03 catalysts (0.3-2.0% Pt) was studied by the gas chromatographic pulse titration technique. The hydrogenolysis activii was reduced to zero only when the atomic ratio of S per exposed Pt atom, SIR,, was about unity. The breakthrough of H,S from the catalyst bed can also be used to indicate the end point in the S titration and hence to estimate the Pt dispersion on the catalyst surface. A fresh catalyst and one containing only irreversibly held S have practically the game activity and selectivity for reforming reactions of n-hexane at 423 OC and 10 bar pressure. The suppression of hydrogenolysis and enhancement of isomerization and aromatization can be achieved only by reversibly adsorbed S on the Pt over and above that held irreversibly. Here also, the hydrogenolysis is effectively suppressed only when all surface Pt atoms are covered by S.
Introduction Poisoning of the activity of supported platinum catalysts by traces of S is well known (cf. Smith et al, 1971). SoDepartment
of C h e m i c a l Engineering, Stanford U n i v e r s i t y ,
Stanford, CA 94305. 0196-4321/82/1221-0052$01.25/0
morjai (1972) suggests that S may catalyze the recrystalization of (111)crystal planes of Pt characterized by sixfold rotational symmetry to (100) planes with fourfold rotational symmetry, and that sensitivity to poisoning by S indicates that a particular reaction is structure-sensitive and not facile, according to Boudart’s (1969) classification. 0 1982 American Chemical Society