62
Ind. Eng. Chem. Prod. Res. Dev. 1980, 19, 62-65
Deactivation of Iron Oxide Catalysts during NO, Reduction with NH, in Flue Gases Yoshihiro Naruse, * Takeshi Ogasawara, Toshihiko Hata, and Hisashi Kishitaka Research Laboratories, Kawasaki Steel Corporation, Ka wasaki-cho, Chiba 260, Japan
in steel works have been studied in real flue gases containing SO, and dust on a bench scale. Samples of catalysts were taken out of the reactors in certain periods and examined by means of a laboratory reactor. When the catalyst was used in the flue gas from a coke oven, its activity decreased to ca.0.6 in terms of k l k , after 5470 h, and the deactivation was considered due to SO, in the gas. On the other hand, when the catalyst was used in the flue gas from a sintering furnace, it showed a fairly drastic deactivation which could not b e ascribed to SO, alone. Experimental results made it clear that the dust component of K had a vital influence on the catalytic activity. NO, reduction with NH, on two iron oxide catalysts which were obtained from a byproduct
Introduction It has been reported in our previous paper (Naruse et al., preceding paper in this issue) that an iron oxide catalyst obtained from a byproduct in steel works may be used as an inexpensive catalyst for the reduction of NO, with NH3. In this study, two bench scale tests have been conducted with one of the iron oxide catalysts and one of the supported iron oxide catalysts (5% FeS04) using two kinds of flue gases, the one from an iron ore sintering furnace and the other from a coke oven, respectively. Deactivation and particle properties of the catalysts were observed during long runs. The causes which led to the deactivation were discussed. Experimental Section ( 1 ) Tests in Real Flue Gases. The catalysts and characteristics of the flue gases used in this study are shown in Table I. Catalyst A is 5% FeS04/Fe203(eq. diameter 10 mm) and catalyst B is an iron oxide catalyst (N-1, eq. diameter, 21 mm). They are the same catalysts referred to as no. 8 and no. 7 in Table I in the previous paper. Two reactors were used for the tests on a bench scale. Flue gas I with a flow rate of 25 Nm3/h from the coke oven and flue gas I1 with a flow rate of 1000 Nm3/h from the sintering furnace were introduced to each reactor. In the test in flue gas 11, inlet and outlet concentrations of NO, were monitored continuously. On the other hand, catalyst A had only been exposed to the flue gas without any measurement in the case of flue gas I. (2) Analytical Procedures. The surface area of the catalysts was measured by a continuous flow method (Nelsen and Eggertsen, 1958) using a sorptgraph (Shimazu ABS-1D) and was evaluated by a one-point method. Surface area which is given by pores larger than 100 b, in radius was also evaluated by integration of mercury penetration curves. Analysis of the catalysts was conducted with catalyst bulk by a conventional method and with surface by EPMA, being carried out on five points arbitrarily picked up from the catalyst B surface with an electron beam width of 70 pm. Semiquantitative results were obtained as an average from these points. X-ray diffraction was carried out with the catalyst B itself and the dust deposited on the catalyst. (3) Activity Test in Laboratory. Activity tests were carried out using small pieces of catalyst A (2-4 mm in diameter) and catalyst B (a cubic form with ca. 8 mm3 volume) in a simulated flue gas. Space velocities were 40000 h-' for B and 18000 h-' for A, respectively. Inlet 0196-4321/80/1219-0062$01 .OO/O
gas of the simulated gas was composed of 130-200 ppm of NO, 130-200 ppm of NH3, 130 ppm of SO2,13.4% 02, and N2 as the diluent. Experimental apparatus and procedures were described in detail in the previous paper. Results Change in Catalyst A Used in Flue Gas I. Samples of catalyst A exposed at 350 "C in flue gas I containing less than 20 ppm of SO, were taken out of the catalyst bed at certain periods. They were examined to check the life of the catalyst. Figure 1 shows change in activity and surface area with running time. The solid line indicates the conversion of NO at 350 "C in the simulated gas. It was shown that the catalytic activity decreased relatively slowly, and the NO conversion decreased from 0.916 to 0.763 in 5470 h. The dot-dashed line indicates how the ratio k / k o [ k o = intitial value of the rate constant of a pseudo-first-order reaction obtained by an equation k = -SV In (1- x)] varied with the time. The ratio k / k , decreased rapidly in the first 1000 h, followed by a small change. The ratio fell to a level of 0.6 in 5470 h. The dotted line indicates change in surface area of the catalyst in terms of S / S o (So = initial value of the surface area). The content of SO, in the catalyst increased from the initial value of 4.46% to 6.0% in 5470 h. Sulfur was the only one which had experienced the change in the chemical composition of the catalyst. Change in Catalyst B Used in Flue Gas 11. Change in Activity. The decrease of the k / k , value with time is shown in Figure 2a. It was demonstrated that the catalyst used in flue gas I1 was deactivated fairly rapidly. Activity tests were carried out with the catalysts which were taken out of top and middle layers of the catalyst bed after about 500 h. Fresh catalyst B was also examined under the same conditions for comparison. Figure 2b shows these results obtained in the simulated gas. The catalyst from the top layer [B-Used(Top)] showed the lowest activity, but the ratio k / k o was maintained at a level of 0.9 during the activity test, when the crushed catalysts were used. As shown in Figure 2a, however, the ratio k / k o with the original size of the catalyst fell down to 0.6 in the first 500 h. This suggests that the deactivation of catalyst B occurred at its outer surface alone. Change in Chemical Composition. It is supposed that such a rapid decrease in the activity was due to SO, and dusts in the flue gas because flue gas I1 contains a considerable amount of them. Both bulk and surface analyses were conducted with the catalyst before and after use. 0 1980 American
Chemical Society
Ind. Eng. Chem. Prod. Res. Dev., Vol. 19, No. 1, 1980 63 Flue Gases and Catalysts Used
Table I.
test conditions -
SO,, ppm
no.
flue gas source
I I1
coke oven sintering furnace
dust, mg/Nm3
reac. temp, "C
SV, h-l
trace ca. 50
350 400
10 000 3 000
< 20 ca. 200
catalyst A: FeSO,/N-1 B: N-1 (Fe,O,)
Table 11. Analytical Results of Fe,O, Catalyst before and after Use ( a ) Bulk Analysis ~~~~~~~
~~~~
~
chemical composition, wt %
sample (catalyst)
--
~
Fe
S
Na
K
SiO,
A1,0,
CaO
MgO
C
NH,
T.N
67.5 66.7 67.2
0.54 0.72 0.81
0.01 0.02 0.01
0.01 0.02 0.02
0.28 0.44 0.26
0.05 0.07 0.05
0.06 0.08 0.08
0.02 0.06 0.02
0.04 0.02
0.01 0.01
0.01 0.01
0.02
0.01
0.01
.~
B-Fresh B-Used ( T o p ) B-Used (Mid)
( b ) Surface Analysis by EPMA (Beam Width 7 0 Fm)" chemical composition, wt % sample
N
B-Fresh B-Used ( T o p ) B-Used (Mid)
trace trace trace
S
Na
K
1.7
0.1 0.2 0.2
trace
10.2 4.7
0.8 0.3
Average value of five points measured. ( T o p ) and (Mid) stand for t o p and middle layers of a reactor, respectively. Table 111. Analytical Results of t h e Dust Deposited o n Catalyst B chemical composition, % T.Fe 31.7
so
T.S 10.4
4
29.4
Na
K
SiO,
Al,O,
CaO
MgO
C
0.28
3.15
6.12
3.87
9.08
1.39
0.40
4 "
sample sample V, r d i u s , 100 .i, I I1 cm3/e r . A m’ie
$
surface area
catalvst B-Fresh B-Used(Top) B-Used (Mid)
30.7 20.4 16.7
33.6 22.0 17.3
0.304 0.306 0.286
240 283 336
25.0 21.6 17.1
Sample I, from outer part of the catalyst; sample 11, from central. a
5x
L
lax
me
1
m
U
5?W
( $ )
Table VI. Effective Diffusion Coefficient Estimated at 600 K , 1 atma
a
ponent KC1 in the flue gas. Since no change was detected with nitrogen before and after use, the adhesion of ammonium salt could be ruled out. Table IV shows the compounds both identified and supposed to exist (marked with asterisks) by X-ray diffraction. Such compounds as SiOz and KFe(S04),, which were identified in the deposited dust, were also detected on the surface of the used catalyst [B-Used(Top)]. Change in Particle Properties of Catalyst B. A surface area measurement by the continuous N2adsorption method was conducted with two speciments for each used catalyst; one was a piece of the catalyst from the outer part of the original catalyst (sample I), and the other was three pieces of the catalyst from the central part (sample 11). Another surface area given by pores with radius greater than 100 A was evaluated. Table V shows surface areas given by the two methods and particle properties estimated by mercury porosimetry. This indicates that the used catalyst showed a lower surface area and a larger mean pore radius than those of the fresh catalyst. Comparison of the results obtained by these two methods told us that the N2 adsorption method brought about a larger area to fresh catalyst, while there was good agreement in the results obtained by these two as far as used catalysts were concerned. This implies that there were considerable amounts of micropores ( r < 100 A) in the fresh catalyst and that most of them disappeared after being used in flue gas 11. The surface area of the used catalyst decreased roughly lineally with the increase of S in catalysts as was shown in Figure 3. It clearly shows that iron(II1) sulfate was formed by the reaction of SO,, which was supposed to diffuse deeply into the catalyst pores, with a catalyst component Fe203,and consequently most of the micropores of the used catalysts disappeared. Discussion It is true that iron sulfate supported on Fe203works as an effective component as was described in the previous paper. However, as was shown in Figure 1, SO, species formed by the reaction with SO, may eventually have an adverse effect on catalytic activity. As is shown in Figure 4, the two iron oxide catalysts showed utterly different
Test plant
L
Figure 4. Change in activity of catalyst A and B.
catalyst
Figure 3. The relationship between sulfur increase and surface area of Fep03catalyst used in flue gas I1 (sinter).
G’C
Running
B-Fresh B-Used ( T o p ) B-Used (Mid)
01 02 03 Increase of 5 content ( % )
L1
231
1;
1
Cat T e m p 4 Sinter A 350.C Lab
2 I1 A 3 3 L L
401
0
0
Notes:
mean Knudsen pore diffus., diffus., radius, 2)&, D, 7 , .i cmA/s cm’/s 243 283 336
0.1051 0.1228 0.1158
0.0913 0.1041 0.1201
effect. diffus.,
De =
porosity,
Dc’,
cm2/s
E
0.304 0.306 0.286
= 0.06825 c m 2 / s ;110 = l / D A B
0.00844 0.00975 0.00982
+
llDK.
behavior in the flue gases containing different amounts of SO, and dust. It was observed that the activity of catalyst A decreased with a decrease of the surface area of the catalyst as shown in Figure 1. It was considered that the decrease of surface area of the catalysts was mainly due to SO, in the flue gases. The deactivation of catalyst A was supposed to be ascribed to SO,, because the flue gas from the coke oven contained little dust. On the other hand, a fairly drastic decrease in activity was observed with catalyst B. Although the level of SO, in flue gas I1 was ten times more than that of flue gas I, the deactivation of catalyst B could not be explained only by the increase of sulfur in catalyst as was described above (Figures 2 and 3). It was assumed that dust components had a vital influence on activity in the case of flue gas 11. We will discuss the causes of deactivation of catalyst B based on the results obtained so far. Although the surface area of catalyst B decreased with the increase of S in the catalyst as shown in Figure 3, o sufficient explanation for the deactivation of catalyst B w‘gs given only by the surface area decrease. The effective diffusion coefficient (De)of a catalyst may change due to changes in pore characteristics. Table VI was obtained when De was evaluated according to a random-pore model (Wakao and Smith, 1962) using the data in Table V and assuming the pore structure was of the mono-disperse type. Since the ratecontrolling step of NO reduction with NH3 on the catalyst was intraparticle diffusion as was reported in the previous paper, the apparent rate was proportional to (De)l/z.Table VI shows that the used catalysts had larger De values. Thus, the deactivation of the catalyst was not explained sufficiently only by the physical changes of the catalyst. As shown in Table 11, the dust components, which increased most remarkably with respect to the used catalysts, were S and K. Potassium, which was supposed to exist as KzS04,was very reactive and liable to react with Fez(SO,), on the catalyst to form such a double salt as KFe(SO,), which was actually identified in the dust deposited on the catalyst (Table IV). Therefore, it was considered that the main cause of deactivation of the catalyst was deterioration of an active site which was resulting from the reaction of the catalyst component with the dust components. Nishijima et al. (1979) have also reported that the influence of K was very vital with alumina-supported Vz05 and Fe203catalysts used in flue gas from a sintering furnace. Activity tests were carried out to determine the effect of K2S0, on the catalyst activity. Figure 5 shows how the
Ind. Eng. Chem. Prod. Res. Dev. 1980, 79, 65-70
08 -
SV
18000h-'
Cat
N-1, 2-4mm
As was discussed so far, the main cause of the catalyst deactivation in a dust-free gas such as flue gas I was SO, in the gas. In the flue gas containing reactive dust components, however, the dust component, in particular K, had a great influence on catalytic activity. Therefore, these iron oxide catalysts could be used practically for the removal of NO, in flue gases without such a dust component as K.
I
Temp 3 5 0 'C
65
Literature Cited 0
05
10
15
Naruse, Y . , Ogasawara, T., Hata, T., Kishitaka, H., Ind. Eng. Cbem., Prod. Res. Dev., preceding article in this issue, 1980. Nelsen, F. M.. Eggertsen. F. T., Anal. Chem., 30(8), 1387 (1958). Japanese Patent, Application No. 50-127081, Research Association for Abatement and Removal of NO, in the Steel Industry in Japan, 1975. Nishijima, A., Kurita, M., Sato, T., Kiyozumi, Y., Hagiwara, H., Ueno, A,, Todo, N., Nippon Kagaku Kaisbi, 276-282 (1979). Wakao, N., Smith, J. M.. Cbem. Eng. Sci., 17, 825-834 (1962).
20
K,SOL a d d e d ( % )
F i g u r e 5. E f f e c t o f K2S04o n a c t i v i t y catalyst size, 2-4 mm; S V , 18 000 h-l.
of a n d iron oxide catalyst:
ratio k / ko declined with varying amount of K2S@,added. The k/ko ratio decreased below 0.3 with an addition of ca. 2% K2S04;it was realized that K2S04played a vital role on the catalyst activity.
Received f o r review April 16, 1979 Accepted S e p t e m b e r 23, 1979
Deposit Formation from Deoxygenated Hydrocarbons. 4. Studies in Pure Compound Systems John W. Frankenfeld' and William F. Taylor Exxon Research and Engineering Company, Linden, New Jersey 07036
The effects of hydrocarbon type on deposit formation in deoxygenated fuels was studied using purified hydrocarbon blends. The rate of deposit formation was determined at 150-650 OC in fuel blends with molecular oxygen levels reduced to below 1 ppm. Deposit formation rates with deoxygenated pure compound blends that did not contain olefins were low at low temperatures but accelerated rapidiy above 500 O C . Most olefins added to the fuels promoted deposit formation even at low temperatures but the effect varied widely with compound type. The morphology of the deposits obtained from deoxygenated blends was different from that observed in air-saturated fuels. The results are consistent with a dual mechanism for deposit formation: autoxidative oligomerizationat low temperatures and pyrolytic breakdown at high temperatures.
Introduction The deposit formation tendencies of jet fuel range hydrocarbons have been the subject of considerable research (Nixon, 1962). Initial work was carried out with air-saturated hydrocarbons in a narrow, near ambient temperature range in order to investigate storage stability characteristics. Subsequent studies were extended to higher temperatures in order to investigate the stability of such fuels when used in high-speed supersonic aircraft (Nixon, 1962; Churchill, et al., 1966). Such studies were carried out mainly with fuels saturated with molecular oxygen via exposure to air although some limited work has been reported with reduced oxygen containing fuels (Taylor and Wallace, 1967). This laboratory has conducted an extensive study of the variables which control the kinetics of deposit formation from hydrocarbons exposed to such high-temperature stress and the factors which may help overcome this instability. Initially these studies were carried out with air-saturated jet fuels, (Taylor, 1969, Taylor and Wallace, 1967). More recently they have been extended to deoxygenated systems. The improvements in fuel stability which accrue on deoxygenation were pointed out by Taylor (1974). However, in certain poor quality fuels the expected 0196-4321/80/1219-0065$01.00/0
enhancement of stability by deoxygenation did not occur. This observation led to a study of the effects of trace impurities, likely to be present in the poor quality fuels, on deposit formation to determine whether such impurities were negating the beneficial effects of molecular oxygen removal. Taylor (1976) found that certain sulfur containing compounds could be highly deleterious to hightemperature stability in deoxygenated fuels. Taylor and Frankenfeld (1978) studied nitrogen and oxygen containing impurities and found that the nitrogen compounds studied were nondeleterious at high temperature but certain of them led to sludge formation during storage under ambient conditions. Many of the oxygen compounds, on the other hand, were found to be moderately to severely deleterious to high-temperature stability in deoxygenated JP-5. In this paper the effects of hydrocarbon type on deposit formation in deoxygenated fuels are discussed. The fuels used in these studies were blends of pure hydrocarbons, representative of those found in actual jet fuels. In addition to the influences of individual hydrocarbons, the effects of interactions between hydrocarbon types were investigated. Finally, the effects of dissolved oxygen (0,) level on the morphology of high-temperature deposits in pure hydrocarbon blends are discussed in light of the C
1980 American Chemical Society
