- Email: [email protected]
Supported rhodium catalysts for nitrous oxide decomposition in the presence of NO, CO2, SO2 and CO
ELSEVIER
Applied Catalysis B: Environmental
6 (1995)
l-10
Supported rhodium catalysts for nitrous oxide decomposition in the presence of NO, CO;?, SO2 and CO Todd W. Dann a, Kirk H. Schulz a7*, Michael Mann b, Mike Collings b aMolecule-interface Science and Engineering Research Laboratory, Department of Chemical Engineering, Box 7101, University of North Dakota, Grand Forks, ND 58203, USA ’ Energy and Environmental Research Center, Box 9018, University of North Dakota, Grand Forks, ND 58203, USA Received 18 August 1994; revised 6 January 1995; accepted 7 January 1995
Abstract Nitrous oxide ( N20), a greenhouse gas and ozone destroyer, is a by-product of coal combustion in fluidized coal combustors (FBC’s) . We have examined the viability of using supported rhodium catalysts at various reaction temperatures to reduce nitrous oxide into nitrogen and oxygen in the presence of CO, NO, SOZ, and CO,, which are also present in fossil fuel emissions. The rhodium catalyst was oxidized in a 20% 02-80% N2 mixture prior to each experimental trial. EDS studies suggest that rhodium concentrates near the catalyst surface following oxidation for 60 min. A space time of 0.07 min was sufficient to ensure 95% conversion of N20 into nitrogen and oxygen at 730 K. No statistically significant effects on N,O conversion were apparent when CO2 ( 15 volume-%), SO, ( 1000 ppm), or NO (370 ppm) were added to the reactor feed streams. However, concentrations of CO as low as 0.4 volume-% lowered the temperature for 95% N20 conversion from 730 to 600 K. Keywords: Rhodium catalysts; Fhridized bed combustors;
Nitrous oxide; Nitrogen oxides
1. Introduction Nitrous oxide, N20, contributes to stratospheric ozone destruction and is considered a green house gas [ 11. The environmental impact of this gas is increasing because the atmospheric nitrous oxide concentration is increasing at an annual rate * Corresponding 7774838.
author. E-mail: [email protected],
tel. ( + l-701)
0926-3373/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSD10926-3373(95)00006-2
7773852,
fax.
( + l-701)
2
T. W. Dann et al. /Applied
Catalysis B: Environmental 6 (1995) I-IO
of 0.2-0.3% [ 21. A complete accounting of all natural and artificial sources of nitrous oxide is not currently available; however, it is accepted that the main source of atmospheric nitrous oxide is microbial action in soils [3], while secondary sources are the manufacture of nylon and adipic acid [ 41 and fossil fuel combustion [ 561. Because of the potential environmental impact of increasing concentrations of atmospheric nitrous oxide, economically feasible strategies are sought to decrease its emission from artificial sources. One artificial source of nitrous oxide is stationary fossil fuel, particularly coal combustion [ $6 3. Pluidized bed coal combustors are operated at low temperatures (1150-1500 K) to minimize NO, emissions and to improve the capture of SO,; however, a side effect is an increased production of nitrous oxide [ 71. Because of the environmental impact of increasing concentrations of nitrous oxide, serious attention has been directed to find methods to reduce nitrous oxide emissions from coal combustion [ 61. No EPA regulations control nitrous oxide emissions, and since operating parameters for coal combustors are set to minimize NO, formation, post-combustion catalytic schemes to convert nitrous oxide to nitrogen and oxygen are being investigated. Li and Armor screened a wide variety of both alumina-supported and zeolitesupported metal and metal oxide catalysts for N,O reduction to N2 and OZ [ 81. They report that rhodium and ruthenium-based catalysts gave nearly complete conversion of N20 to N2 and O2 at low temperatures (625 K). One of the difficulties in using catalysis for nitrous oxide reduction is the presence of additional gases in the stream from the coal combustor. In addition to N20, CO, COZ, NO,, and SO* are present in gases from fluidized bed coal combustion [ 93. Thus, it is critical to study the catalytic conversion of nitrous oxide in the presence of these other gases. Compositions of exit gases were determined using published exit gas concentrations from fluidized bed combustors [ 93. An alumina supported rhodium (Rh/Al,O,) catalyst was chosen for this study because it provided nearly complete conversion of nitrous oxide into nitrogen and oxygen at the lowest temperature [ 81. Although effective catalysts, rhodium-based catalysts have a prohibitively high cost. Even though gases leave fluidized coal combustors at temperatures (ca. 1000-1200 K; [7] ) much higher than the temperature range under investigation (ca. 500-700 K),increased flexibility in placing the catalytic reactor (i.e., closer or further from the PBC) during scale up can be accomplished by finding a suitable nitrous oxide reduction method at the lowest possible temperature. 2. Experimental 2.1. Apparatus All experiments were done in a stainless steel flow reactor system. A 50 ml Pyrex reactor containing 45 g of catalyst was used. The feed gases were passed through
T. W. Dunn ef al. /Applied Catalysis B: Environmental 6 (1995) l-10
3
l/4 in. plastic tubing prior to entering the stainless steel gas manifold. After mixing, the gases entered the reactor, which was mounted inside a conventional tube furnace. Temperatures in the catalyst bed were monitored using a type K thermocouple in contact with the center of the packed bed. Mass flow controllers were used to regulate flow rates, which were typically ca. lOOO( ) 25) cm3/min. Total pressure in the reactor was maintained near atmospheric. The conversion of nitrous oxide was monitored using a Siemens Ultramat 5e nitrous oxide analyzer which could read between 0 and 550 ppm of nitrous oxide. No other reaction products were monitored. 2.2. Electron microscopy studies The electron microscopy and energy dispersive X-ray spectroscopy studies were done at the NSF Center for Interfacial Engineering (CIE) at the University of Minnesota-Twin Cities. A JEOL 84011 scanning electron microscope (SEM), a JEOL JEM-1210 transmission electron microscope (TEM), and an ElectroScan environmental scanning electron microscope (ESEM) were all used in this study. Catalyst samples were coated with chromium to minimize charging for the SEM and TEM studies; no chromium [ lo] was need in the environmental SEM. EDS was run using the TEM. For these studies, a 9 keV electron beam was used with a take-off angle of 40”. The detector resolution was 148.5 eV. Identical areas were measured on all samples. Chemicals Feed gases were supplied to the reactor using nitrogen as the carrier gas. N,O ( 1% in N2) and NO (370 ppm in N2) were supplied by Matheson; other gases used included CO* (Acme-Dow-Western), 02, CO (Linde), and NZ. The rhodium catalyst (5 wt.-% rhodium on 0.125 in. A&O3 spheres), which was purchased from Aldrich, was used as received. Catalyst preparation Using a procedure similar to that of Li and Armor [ 81, the catalyst (Rh/Al,O,) was oxidized in an 80/20 volume percent mixture of nitrogen and oxygen for between 30 and 60 min prior to each experimental run. Feed gas compositions The composition of the feed gas to the reactor was set to reflect typical FBC exit gas concentrations. Typical concentrations in FBC exit gases for the combustion of center lignite and Blacksville bituminous coal samples are given in Table 1. Using these values, feed gas concentrations were selected for N20, CO, NO, SOZ, and COZ, which are listed in Table 2. It should be noted that these compositions are for flue gas emissions during coal combustion with 3% excess oxygen. The exiting compositions were only used to give appropriate concentration ranges for
4
T. W. Dann et al. /Applied
Catalysis B: Environmental 6 (1995) I-IO
Table 1 Concentrations of NzO, NO, CO, COa and SO2 in fluid&d bed combustion from the pilot-scale fluidized-bed combustion evaluation facility [9]
emissions from center lignite coal
Hue gas component
Center lignite coal Low-high
Blacksville Low-high
Nitrous oxide (NaO) Nitrogen oxide (NO) Carbon monoxide (CO) Carbon dioxide ( COZ) Sulfur dioxide (SOP)
21-275 69-278 11430 16.617.7% 3459
52-264 35-264 12-601 15.2-17.6% 172-1703
bituminous
All values have been corrected to 3% oxygen, and are given in ppm, except for COa which is given as a percent of the Rue gas.
our catalytic study. The experiments reported in this paper were not done in the presence of a 3% excess of oxygen.
3. Results and discussion 3. I. Microstructure of rhodium-based catalysts Environmental
SEMphotographs
Fig. 1 shows an ESEM photograph of the rhodium catalyst prior to oxidation. The lightest colored ‘specks’ are rhodium present on the catalyst surface. Energy dispersive X-ray spectroscopy measurements
Energy dispersive X-ray spectroscopy (EDS) , a useful technique for microelemental analysis of materials, is often done in conjunction with either TEM or SEM [ 131. To aid in the catalyst characterization studies, EDS measurements were made of both non-oxidized and oxidized Rh-A1,03 catalyst samples. The results are shown in Fig. 2. EDS is a surface sensitive technique, so it only detects elements near the catalyst surface. Fig. 2a shows the catalyst after heating to 775 K in the nitrogen carrier gas. No significant Rh L,, peak is present, which suggests that the Table 2 Concentrations
of feed gases to the reactor
Flue gas component
Concentration
Nitrous oxide (N,O) Nitrogen oxide (NO) Carbon monoxide (CO) Carbon dioxide (CO,) Sulfur dioxide ( SOZ)
200-300 20&370 2ooo-4ooo 15%-30% 200-500
range selected
All values are given in ppm, except for CO, which is given as a percent of the flue gas.
T.W. Dunn et al. /Applied Catalysis B: Environmental 6 (1995) I-10
Fig. 1, Scanning electron micrograph
of the non-oxidized
Rh-A&O3 catalyst.
A
c
.oo
KeV
4.00
1.00
B
8.00
0
0.00
2.00
4.00
*.oo
a.00
KeV
Fig. 2. Energy dispersive X-ray spectroscopy plots of 5.0 wt.-% rhodium supported on alumina as a function of catalyst oxidation treatment. (a) The catalyst heated to 500°C in Nz. (b) the catalyst oxidized in a 20% oxygen stream for 60 min.
6
T. W. Dann et al. /Applied Catalysis B: Environmental 6 (1995) I-IO
rhodium is either not concentrated at the surface or that no rhodium is present at the surface within the detection limits of the instrument. However, after heating to 775 K in a 20% oxygen stream for 60 mitt, a substantial Rh Lcrl peak is observed, demonstrating that significant amounts of rhodium are present near the surface of the catalyst pellet. A comparison of the intensities of the peaks ( Al-Kcr to Rh La1 ) is 0.05 in Fig. 2a which increases up to 0.29 in Fig. 2b. These results are consistent with an ‘egg-shell’ catalyst distribution, where the vast majority of the metal loading occurs near the surface. Schwartz and Schmidt [ 121 examined the microstructure of rhodium+erium and platinum-cerium particles on alumina and silica particles using TEM and XPS. They report no clear evidence in the TEM studies that any changes in the rhodium particle size occurred from a preoxidized metallic state (only Rh) to oxidized rhodium particles ( Rh203). Our results, although inconclusive, tend to suggest that some agglomeration of rhodium occurs following our oxidation treatments. Unfortunately, EDS is not able to distinguish between rhodium metal and Rh203, so we cannot definitively suggest the microstructural changes that occur during oxidation. 3.2. Space time Initial experiments focused on determining the appropriate space time (T) to achieve greater than 95% conversion of N20 to Nz and OZ. The Siemens N20 analyzer required a minimum flow rate of 720 cm3/min to ensure accurate nitrous oxide concentration readings; thus, space time could be varied only by changing the reactor volume. A space time of 0.025 min (25 ml reactor) gave less than 95% conversion at 700 K as expected from published literature results [ 8 1. After increasing the space time by a factor of two to 0.05 min (50 ml reactor), conversion increased to 95% at 730 K. One difference between our study and other Rh-A1203 nitrous oxide decomposition studies [ 8,111 is the temperature in which approximately 95% conversion is achieved. According to Li and Armor [ 81, 95% nitrous oxide conversion began occurring at approximately 600 K, while McCabe and Wong [ 111 report only 10% conversion of nitrous oxide near 700 K. Our studies show that 95% conversion did not occur until 730 K, an intermediate value between the temperatures in the two other studies. These discrepancies may be caused by differences in reactor configuration in the various studies as well as different catalyst oxidation treatments. McCabe and Wong [ 1l] calcined their rhodium catalyst at 775 K for 2 h, and then reduced the catalyst with hydrogen at 473 K prior to each experimental run. Schwartz and Schmidt [ 121 showed that reduction in hydrogen transforms the oxidized rhodium ( Rh203) to rhodium metal. Thus, McCabe and Wong produced Rh203 during oxidation, and then reduced some of the rhodium oxide back to metallic rhodium during reduction with hydrogen. Assuming that Rh203 is the catalytically important compound for nitrous oxide reduction to nitrogen and oxygen, the large temperature discrepancy between the McCabe study ( 10% conversion
T.W. Dam et al. /Applied Catalysis B: Environmental 6 (1995) I-10
100 C
._0
; 2
80
6 60 8 ‘Z 0 40 v) z *g 20
550
600
650
Temperature (K) Fig. 3. N20 conversions on l&-A&OS as a function of temperature. Reactions were with a total flow-rate of 1000 ml/min and N,O= 330 ppm. Reactions were run with 0.4% CO, 0.04% NO, 1000 ppm SO2 and N,O only. All trials used N, as a carrier gas.
at ca. 700 K) and Li and Armor (95% conversion at ca. 625 K) and our results (95% conversion at ca. 730 K) is most likely due to differences in catalyst oxidation and preparation procedures. Additional differences in catalyst loading and reactor space time may also have contributed to the large temperature difference observed. The difference in reaction temperature between the present study and Li and Armor’s work is not entirely clear. At high conversions, parameters such as reactor geometry and mass and heat transfer effects will affect overall reactor performance, including conversion. Our reactor volume and sample size (45 g of catalyst) and flow rate of gases ( 1000 cm3 /mm) are significantly different from the experimental setup employed by Li and Armor (0.1 g catalyst sample, flow rate 100 ml/min). We believe that the results of our study and those of Li and Armor are different because of the, difference in scale. The observed temperature difference is not due to different temperature measurement techniques; both systems recorded temperatures in the catalyst near the end of the bed. 3.3. Reaction studies with rhodium-based
catalysts
In order to determine the effects of different common flue gas compounds on nitrous oxide conversion over supported rhodium catalysts, each of the common flue gas components ( SOZ, NO, CO, and COT) were separately added to the reactor feed stream. The effect of each of the gases on nitrous oxide conversion is shown in Fig. 3 as a function of reactor temperature. We chose to set the upper limit for nitrous oxide conversion at 95 + %. Higher conversions are certainly possible at elevated temperatures and decreased space times; and thus, nearly 100% nitrous oxide conversion may be possible. However, for initial catalyst screening studies, 95% conversion was deemed sufficient to test the effects of commonly encountered flue gas components on catalytic performance.
8
i? W. Dann et al. /Applied Catalysis B: Environmental 6 (1995) l-10
The following sections detail the results of each of the experiments illustrated in Fig. 3. Nitrogen oxide (NO), sulfur dioxide (SO,), carbon dioxide (CO,) NO was selected to represent the class of molecules normally designated as NO,.
NO was added to nitrous oxide and the carrier gas, NZ, to determine its effect on conversion. No significant changes in N20 conversion were observed after adding 370 ppm (0.037 volume-%) NO to the gas stream. Similarly, no significant changes in conversion were observed after adding approximately 1000 ppm of SO* or 15% CO, (by volume) to the N2/N20 gas stream. However, differences were apparent for SO2 between 600 and 650 K, where a decrease in nitrous oxide conversion is observed. The reasons for this decrease over a very narrow temperature range are not completely understood. However, compared with the significantly larger changes observed after adding carbon monoxide, the change in nitrous oxide conversion after adding SO2 is relatively modest, and does not adversely affect catalytic performance at higher temperatures. Carbon monoxide (CO)
Varying concentrations of carbon monoxide were also added to nitrous oxide and the carrier gas. For carbon monoxide concentrations below 0.2% by volume, no significant effect on nitrous oxide conversion was observed. However, upon increasing the carbon monoxide concentration up to 0.4% by volume, a significant downward shift in the temperature [from approximately 730 K (no carbon monoxide present) to 600 K (0.4% CO)] at which 95% nitrous oxide conversion occurred was observed. This phenomenon is shown in Fig. 3, where a clear downwards shift in the reaction temperature for 95% conversion is seen. Further increasing the carbon monoxide concentration above 0.4% had no further effects on the temperature for 95% conversion. According to McCabe and Wong, carbon monoxide is a ‘major removal pathway’ for removal of nitrogen oxides [ 111. In fact, there seems to be a synergistic side reaction occurring simultaneously involving carbon monoxide that aids in the conversion of nitrous oxide. The key reaction between nitrous oxide and carbon monoxide which contributes to the enhanced nitrous oxide conversion is [ 141: CO+N,O
+ N,+CO,
(1)
Cho and Shanks report that despite the low rate of the CO-N20 reaction, N20 is converted more rapidly when CO is present along with the FUr-A&O3catalyst. The results of our experiment agree with those of McCabe and Wong, even though their experiments were conducted at significantly lower nitrous oxide conversions. The key reason for the enhanced conversion of nitrous oxide in the presence of carbon monoxide can be attributed to the dissociative adsorption of nitrous oxide on clean metal surfaces. Daniel et al. studied nitrous oxide decomposition on the Rh( 100) surface and observed that nitrous oxide dissociatively adsorbed, leaving
T. W. Dann et al. /Applied
Catalysis B: Environmental
6 (1995) I-10
9
an oxygen overlayer on the Rh( 100) surface, and giving off N2 as a desorption product [ 151. It has been clearly shown that oxygen-covered metal surfaces are not as catalytically active as bare-metal surfaces, and thus, any mechanism that consumes surface oxygen and keeps catalytically active sites open for nitrous oxide dissociation will enhance the rate of conversion of N,O to Nz and OZ. Since carbon monoxide acts as a reducing species to form carbon dioxide with surface oxygen via reaction ( 1) , an enhanced rate of nitrous oxide conversion will result. An analogous mechanism is suggested for nitrous oxide decomposition on oxidized rhodium catalysts (Rh,O,) .There is strong evidence rhodium is fully oxidized to Rh,03 following oxidation, as reported by Van’t Bilk et al. [ 161 and Schwartz and Schmidt [ 121. Nitrous oxide dissociation has been reported on CaO [ 171, and thus it is plausible that nitrous oxide dissociation also occurs on rhodium oxide. 4. Conclusions In low concentrations, SO*, CO,, and NO do not significantly effect the conversion of nitrous oxide to nitrogen and oxygen over supported rhodium catalysts. These results suggest that rhodium catalysts may provide a basis for significantly reducing nitrous oxide emissions from coal combustion facilities using catalytic post processing of flue gases. Additionally, adding carbon monoxide significantly reduces the temperature at which 95% conversion of nitrous oxide occurs. It may be practical to feed carbon monoxide or another reductant gas into the reactor feed stream just as ammonia is fed into the catalytic reactor bed for the reduction of NO, emissions [ 18,191. Several more obstacles need to be cleared before a scaled up catalytic reactor can be installed and operated. First, and most importantly, will be the question of cost. Although an effective catalyst for nitrous oxide control, the prohibitive cost of rhodium catalysts when scaled up to an industrial sized reactor precludes its practical use. The second problem will be the positioning of the reactor. Since exit gases leave the stack at approximately 373-473 K, the reactor will have to be positioned closer to the FBC. This will require redesigning or relocating the emission equipment after the exit gases leave the combustor. Catalytic reduction of nitrous oxide has many possibilities. Understanding the reaction mechanisms and kinetics of nitrous oxide decomposition on rhodium based catalysts in the presence of other flue gas compounds will be of great importance for continued progress in this area. Acknowledgements
The authors acknowledge the UND-Energy and Environmental Research Center (UND-EERC) and the Chemical Engineering Alumni Fund at the University of
10
T. W. Dann et al. /Applied
Catalysis B: Environmental 6 (1995) l-10
North Dakota for financial assistance with this project. The Association of Western Universities (AWU) summer fellowship to KS is also gratefully acknowledged. Funding for the electron microscopy studies was through a grant from the NSF Center for Interfacial Engineering at the University of Minnesota-Twin Cities.
References [l] [2] [3] [4] [5] [6] [7] [S] [ 91 [lo] [ 1 l] [ 121 [ 131 [ 141 [ 151 [ 161
[ 171 [IS] [ 191
R.E. Dickinson and R.J. Cicerone, Nature, 319 (1986) 109. M.A.K. Khalil and R.A. Rasmussen, Tellus, 35B ( 1983) 161. A.R. Mosier and D.S. Chime], Chem. Ind., (December 2, 1991) 874. M.H. Thiemens and W.C. Trogler, Science, 251 (1991) 932. M. Takeshita, L.L. Sloss and I.M. Smith, IEA Coal Research, November (1993) 1. M.A. W6jtowicz, J.R. Pels and J.A. Moulijn, Fuel Process. Technol., 34 ( 1993) 1. M.D. Mann, M.E. Collings and P.E. Botros, Prog. Energy Combustion Sci., 18 (1992) 447. J.N. Armor and Y. Li, Appl. Catal. B, 1 ( 1992) L21. Electric Power Research Institute (EPRI) , Pilot-Scale Circulating Fluidized-Bed Combustion Evaluation Facility, EPRI TR-100829, Project 1179-32, Final Report, December, 1992. R.E. Cameron, USA Microsc. Anal., May ( 1994) 17. R.W. McCabe and C. Wong, J. Catal., 121 (1990) 422. J.M. Schwartz and L.D. Schmidt, J. Catal., 138 ( 1992) 283. B.W. Rossiter and J. F. Hamilton, Physical Methods of Chemistry: Volume IV Microscopy, Wiley, New York, 1991, pp. 203-238. B.K. Cho and B.H. Shanks, General Motors Research Laboratories Research Report PC-421, March 1988. W.M. Daniel, Y. Kim, H.C. Peebles and J.M. White, Surf. Sci., 111 (1981) 189. F.J. Van’t Blik, J.B.A.D. van Zon, T. Huizinga, J.C. Vis, D.C. Koningsberger and R. Prim, J. Am. Chem. Sot., 107 (1985) 3139. M. Nakamura, H. Mitsuhashi and N. Takesawa, J. Catal., 138 ( 1992) 686. S.T. Darian, J.W. Eldridge and J.R. Kitrell, Environ. Prog., l-2 ( 1982) 84. J.W. Eldridge, J.R. Kittrell and F.G. Medros, Ind. Eng. Chem. Res., 28 (1989) 1171.
Applied Catalysis B: Environmental
6 (1995)
l-10
Supported rhodium catalysts for nitrous oxide decomposition in the presence of NO, CO;?, SO2 and CO Todd W. Dann a, Kirk H. Schulz a7*, Michael Mann b, Mike Collings b aMolecule-interface Science and Engineering Research Laboratory, Department of Chemical Engineering, Box 7101, University of North Dakota, Grand Forks, ND 58203, USA ’ Energy and Environmental Research Center, Box 9018, University of North Dakota, Grand Forks, ND 58203, USA Received 18 August 1994; revised 6 January 1995; accepted 7 January 1995
Abstract Nitrous oxide ( N20), a greenhouse gas and ozone destroyer, is a by-product of coal combustion in fluidized coal combustors (FBC’s) . We have examined the viability of using supported rhodium catalysts at various reaction temperatures to reduce nitrous oxide into nitrogen and oxygen in the presence of CO, NO, SOZ, and CO,, which are also present in fossil fuel emissions. The rhodium catalyst was oxidized in a 20% 02-80% N2 mixture prior to each experimental trial. EDS studies suggest that rhodium concentrates near the catalyst surface following oxidation for 60 min. A space time of 0.07 min was sufficient to ensure 95% conversion of N20 into nitrogen and oxygen at 730 K. No statistically significant effects on N,O conversion were apparent when CO2 ( 15 volume-%), SO, ( 1000 ppm), or NO (370 ppm) were added to the reactor feed streams. However, concentrations of CO as low as 0.4 volume-% lowered the temperature for 95% N20 conversion from 730 to 600 K. Keywords: Rhodium catalysts; Fhridized bed combustors;
Nitrous oxide; Nitrogen oxides
1. Introduction Nitrous oxide, N20, contributes to stratospheric ozone destruction and is considered a green house gas [ 11. The environmental impact of this gas is increasing because the atmospheric nitrous oxide concentration is increasing at an annual rate * Corresponding 7774838.
author. E-mail: [email protected],
tel. ( + l-701)
0926-3373/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSD10926-3373(95)00006-2
7773852,
fax.
( + l-701)
2
T. W. Dann et al. /Applied
Catalysis B: Environmental 6 (1995) I-IO
of 0.2-0.3% [ 21. A complete accounting of all natural and artificial sources of nitrous oxide is not currently available; however, it is accepted that the main source of atmospheric nitrous oxide is microbial action in soils [3], while secondary sources are the manufacture of nylon and adipic acid [ 41 and fossil fuel combustion [ 561. Because of the potential environmental impact of increasing concentrations of atmospheric nitrous oxide, economically feasible strategies are sought to decrease its emission from artificial sources. One artificial source of nitrous oxide is stationary fossil fuel, particularly coal combustion [ $6 3. Pluidized bed coal combustors are operated at low temperatures (1150-1500 K) to minimize NO, emissions and to improve the capture of SO,; however, a side effect is an increased production of nitrous oxide [ 71. Because of the environmental impact of increasing concentrations of nitrous oxide, serious attention has been directed to find methods to reduce nitrous oxide emissions from coal combustion [ 61. No EPA regulations control nitrous oxide emissions, and since operating parameters for coal combustors are set to minimize NO, formation, post-combustion catalytic schemes to convert nitrous oxide to nitrogen and oxygen are being investigated. Li and Armor screened a wide variety of both alumina-supported and zeolitesupported metal and metal oxide catalysts for N,O reduction to N2 and OZ [ 81. They report that rhodium and ruthenium-based catalysts gave nearly complete conversion of N20 to N2 and O2 at low temperatures (625 K). One of the difficulties in using catalysis for nitrous oxide reduction is the presence of additional gases in the stream from the coal combustor. In addition to N20, CO, COZ, NO,, and SO* are present in gases from fluidized bed coal combustion [ 93. Thus, it is critical to study the catalytic conversion of nitrous oxide in the presence of these other gases. Compositions of exit gases were determined using published exit gas concentrations from fluidized bed combustors [ 93. An alumina supported rhodium (Rh/Al,O,) catalyst was chosen for this study because it provided nearly complete conversion of nitrous oxide into nitrogen and oxygen at the lowest temperature [ 81. Although effective catalysts, rhodium-based catalysts have a prohibitively high cost. Even though gases leave fluidized coal combustors at temperatures (ca. 1000-1200 K; [7] ) much higher than the temperature range under investigation (ca. 500-700 K),increased flexibility in placing the catalytic reactor (i.e., closer or further from the PBC) during scale up can be accomplished by finding a suitable nitrous oxide reduction method at the lowest possible temperature. 2. Experimental 2.1. Apparatus All experiments were done in a stainless steel flow reactor system. A 50 ml Pyrex reactor containing 45 g of catalyst was used. The feed gases were passed through
T. W. Dunn ef al. /Applied Catalysis B: Environmental 6 (1995) l-10
3
l/4 in. plastic tubing prior to entering the stainless steel gas manifold. After mixing, the gases entered the reactor, which was mounted inside a conventional tube furnace. Temperatures in the catalyst bed were monitored using a type K thermocouple in contact with the center of the packed bed. Mass flow controllers were used to regulate flow rates, which were typically ca. lOOO( ) 25) cm3/min. Total pressure in the reactor was maintained near atmospheric. The conversion of nitrous oxide was monitored using a Siemens Ultramat 5e nitrous oxide analyzer which could read between 0 and 550 ppm of nitrous oxide. No other reaction products were monitored. 2.2. Electron microscopy studies The electron microscopy and energy dispersive X-ray spectroscopy studies were done at the NSF Center for Interfacial Engineering (CIE) at the University of Minnesota-Twin Cities. A JEOL 84011 scanning electron microscope (SEM), a JEOL JEM-1210 transmission electron microscope (TEM), and an ElectroScan environmental scanning electron microscope (ESEM) were all used in this study. Catalyst samples were coated with chromium to minimize charging for the SEM and TEM studies; no chromium [ lo] was need in the environmental SEM. EDS was run using the TEM. For these studies, a 9 keV electron beam was used with a take-off angle of 40”. The detector resolution was 148.5 eV. Identical areas were measured on all samples. Chemicals Feed gases were supplied to the reactor using nitrogen as the carrier gas. N,O ( 1% in N2) and NO (370 ppm in N2) were supplied by Matheson; other gases used included CO* (Acme-Dow-Western), 02, CO (Linde), and NZ. The rhodium catalyst (5 wt.-% rhodium on 0.125 in. A&O3 spheres), which was purchased from Aldrich, was used as received. Catalyst preparation Using a procedure similar to that of Li and Armor [ 81, the catalyst (Rh/Al,O,) was oxidized in an 80/20 volume percent mixture of nitrogen and oxygen for between 30 and 60 min prior to each experimental run. Feed gas compositions The composition of the feed gas to the reactor was set to reflect typical FBC exit gas concentrations. Typical concentrations in FBC exit gases for the combustion of center lignite and Blacksville bituminous coal samples are given in Table 1. Using these values, feed gas concentrations were selected for N20, CO, NO, SOZ, and COZ, which are listed in Table 2. It should be noted that these compositions are for flue gas emissions during coal combustion with 3% excess oxygen. The exiting compositions were only used to give appropriate concentration ranges for
4
T. W. Dann et al. /Applied
Catalysis B: Environmental 6 (1995) I-IO
Table 1 Concentrations of NzO, NO, CO, COa and SO2 in fluid&d bed combustion from the pilot-scale fluidized-bed combustion evaluation facility [9]
emissions from center lignite coal
Hue gas component
Center lignite coal Low-high
Blacksville Low-high
Nitrous oxide (NaO) Nitrogen oxide (NO) Carbon monoxide (CO) Carbon dioxide ( COZ) Sulfur dioxide (SOP)
21-275 69-278 11430 16.617.7% 3459
52-264 35-264 12-601 15.2-17.6% 172-1703
bituminous
All values have been corrected to 3% oxygen, and are given in ppm, except for COa which is given as a percent of the Rue gas.
our catalytic study. The experiments reported in this paper were not done in the presence of a 3% excess of oxygen.
3. Results and discussion 3. I. Microstructure of rhodium-based catalysts Environmental
SEMphotographs
Fig. 1 shows an ESEM photograph of the rhodium catalyst prior to oxidation. The lightest colored ‘specks’ are rhodium present on the catalyst surface. Energy dispersive X-ray spectroscopy measurements
Energy dispersive X-ray spectroscopy (EDS) , a useful technique for microelemental analysis of materials, is often done in conjunction with either TEM or SEM [ 131. To aid in the catalyst characterization studies, EDS measurements were made of both non-oxidized and oxidized Rh-A1,03 catalyst samples. The results are shown in Fig. 2. EDS is a surface sensitive technique, so it only detects elements near the catalyst surface. Fig. 2a shows the catalyst after heating to 775 K in the nitrogen carrier gas. No significant Rh L,, peak is present, which suggests that the Table 2 Concentrations
of feed gases to the reactor
Flue gas component
Concentration
Nitrous oxide (N,O) Nitrogen oxide (NO) Carbon monoxide (CO) Carbon dioxide (CO,) Sulfur dioxide ( SOZ)
200-300 20&370 2ooo-4ooo 15%-30% 200-500
range selected
All values are given in ppm, except for CO, which is given as a percent of the flue gas.
T.W. Dunn et al. /Applied Catalysis B: Environmental 6 (1995) I-10
Fig. 1, Scanning electron micrograph
of the non-oxidized
Rh-A&O3 catalyst.
A
c
.oo
KeV
4.00
1.00
B
8.00
0
0.00
2.00
4.00
*.oo
a.00
KeV
Fig. 2. Energy dispersive X-ray spectroscopy plots of 5.0 wt.-% rhodium supported on alumina as a function of catalyst oxidation treatment. (a) The catalyst heated to 500°C in Nz. (b) the catalyst oxidized in a 20% oxygen stream for 60 min.
6
T. W. Dann et al. /Applied Catalysis B: Environmental 6 (1995) I-IO
rhodium is either not concentrated at the surface or that no rhodium is present at the surface within the detection limits of the instrument. However, after heating to 775 K in a 20% oxygen stream for 60 mitt, a substantial Rh Lcrl peak is observed, demonstrating that significant amounts of rhodium are present near the surface of the catalyst pellet. A comparison of the intensities of the peaks ( Al-Kcr to Rh La1 ) is 0.05 in Fig. 2a which increases up to 0.29 in Fig. 2b. These results are consistent with an ‘egg-shell’ catalyst distribution, where the vast majority of the metal loading occurs near the surface. Schwartz and Schmidt [ 121 examined the microstructure of rhodium+erium and platinum-cerium particles on alumina and silica particles using TEM and XPS. They report no clear evidence in the TEM studies that any changes in the rhodium particle size occurred from a preoxidized metallic state (only Rh) to oxidized rhodium particles ( Rh203). Our results, although inconclusive, tend to suggest that some agglomeration of rhodium occurs following our oxidation treatments. Unfortunately, EDS is not able to distinguish between rhodium metal and Rh203, so we cannot definitively suggest the microstructural changes that occur during oxidation. 3.2. Space time Initial experiments focused on determining the appropriate space time (T) to achieve greater than 95% conversion of N20 to Nz and OZ. The Siemens N20 analyzer required a minimum flow rate of 720 cm3/min to ensure accurate nitrous oxide concentration readings; thus, space time could be varied only by changing the reactor volume. A space time of 0.025 min (25 ml reactor) gave less than 95% conversion at 700 K as expected from published literature results [ 8 1. After increasing the space time by a factor of two to 0.05 min (50 ml reactor), conversion increased to 95% at 730 K. One difference between our study and other Rh-A1203 nitrous oxide decomposition studies [ 8,111 is the temperature in which approximately 95% conversion is achieved. According to Li and Armor [ 81, 95% nitrous oxide conversion began occurring at approximately 600 K, while McCabe and Wong [ 111 report only 10% conversion of nitrous oxide near 700 K. Our studies show that 95% conversion did not occur until 730 K, an intermediate value between the temperatures in the two other studies. These discrepancies may be caused by differences in reactor configuration in the various studies as well as different catalyst oxidation treatments. McCabe and Wong [ 1l] calcined their rhodium catalyst at 775 K for 2 h, and then reduced the catalyst with hydrogen at 473 K prior to each experimental run. Schwartz and Schmidt [ 121 showed that reduction in hydrogen transforms the oxidized rhodium ( Rh203) to rhodium metal. Thus, McCabe and Wong produced Rh203 during oxidation, and then reduced some of the rhodium oxide back to metallic rhodium during reduction with hydrogen. Assuming that Rh203 is the catalytically important compound for nitrous oxide reduction to nitrogen and oxygen, the large temperature discrepancy between the McCabe study ( 10% conversion
T.W. Dam et al. /Applied Catalysis B: Environmental 6 (1995) I-10
100 C
._0
; 2
80
6 60 8 ‘Z 0 40 v) z *g 20
550
600
650
Temperature (K) Fig. 3. N20 conversions on l&-A&OS as a function of temperature. Reactions were with a total flow-rate of 1000 ml/min and N,O= 330 ppm. Reactions were run with 0.4% CO, 0.04% NO, 1000 ppm SO2 and N,O only. All trials used N, as a carrier gas.
at ca. 700 K) and Li and Armor (95% conversion at ca. 625 K) and our results (95% conversion at ca. 730 K) is most likely due to differences in catalyst oxidation and preparation procedures. Additional differences in catalyst loading and reactor space time may also have contributed to the large temperature difference observed. The difference in reaction temperature between the present study and Li and Armor’s work is not entirely clear. At high conversions, parameters such as reactor geometry and mass and heat transfer effects will affect overall reactor performance, including conversion. Our reactor volume and sample size (45 g of catalyst) and flow rate of gases ( 1000 cm3 /mm) are significantly different from the experimental setup employed by Li and Armor (0.1 g catalyst sample, flow rate 100 ml/min). We believe that the results of our study and those of Li and Armor are different because of the, difference in scale. The observed temperature difference is not due to different temperature measurement techniques; both systems recorded temperatures in the catalyst near the end of the bed. 3.3. Reaction studies with rhodium-based
catalysts
In order to determine the effects of different common flue gas compounds on nitrous oxide conversion over supported rhodium catalysts, each of the common flue gas components ( SOZ, NO, CO, and COT) were separately added to the reactor feed stream. The effect of each of the gases on nitrous oxide conversion is shown in Fig. 3 as a function of reactor temperature. We chose to set the upper limit for nitrous oxide conversion at 95 + %. Higher conversions are certainly possible at elevated temperatures and decreased space times; and thus, nearly 100% nitrous oxide conversion may be possible. However, for initial catalyst screening studies, 95% conversion was deemed sufficient to test the effects of commonly encountered flue gas components on catalytic performance.
8
i? W. Dann et al. /Applied Catalysis B: Environmental 6 (1995) l-10
The following sections detail the results of each of the experiments illustrated in Fig. 3. Nitrogen oxide (NO), sulfur dioxide (SO,), carbon dioxide (CO,) NO was selected to represent the class of molecules normally designated as NO,.
NO was added to nitrous oxide and the carrier gas, NZ, to determine its effect on conversion. No significant changes in N20 conversion were observed after adding 370 ppm (0.037 volume-%) NO to the gas stream. Similarly, no significant changes in conversion were observed after adding approximately 1000 ppm of SO* or 15% CO, (by volume) to the N2/N20 gas stream. However, differences were apparent for SO2 between 600 and 650 K, where a decrease in nitrous oxide conversion is observed. The reasons for this decrease over a very narrow temperature range are not completely understood. However, compared with the significantly larger changes observed after adding carbon monoxide, the change in nitrous oxide conversion after adding SO2 is relatively modest, and does not adversely affect catalytic performance at higher temperatures. Carbon monoxide (CO)
Varying concentrations of carbon monoxide were also added to nitrous oxide and the carrier gas. For carbon monoxide concentrations below 0.2% by volume, no significant effect on nitrous oxide conversion was observed. However, upon increasing the carbon monoxide concentration up to 0.4% by volume, a significant downward shift in the temperature [from approximately 730 K (no carbon monoxide present) to 600 K (0.4% CO)] at which 95% nitrous oxide conversion occurred was observed. This phenomenon is shown in Fig. 3, where a clear downwards shift in the reaction temperature for 95% conversion is seen. Further increasing the carbon monoxide concentration above 0.4% had no further effects on the temperature for 95% conversion. According to McCabe and Wong, carbon monoxide is a ‘major removal pathway’ for removal of nitrogen oxides [ 111. In fact, there seems to be a synergistic side reaction occurring simultaneously involving carbon monoxide that aids in the conversion of nitrous oxide. The key reaction between nitrous oxide and carbon monoxide which contributes to the enhanced nitrous oxide conversion is [ 141: CO+N,O
+ N,+CO,
(1)
Cho and Shanks report that despite the low rate of the CO-N20 reaction, N20 is converted more rapidly when CO is present along with the FUr-A&O3catalyst. The results of our experiment agree with those of McCabe and Wong, even though their experiments were conducted at significantly lower nitrous oxide conversions. The key reason for the enhanced conversion of nitrous oxide in the presence of carbon monoxide can be attributed to the dissociative adsorption of nitrous oxide on clean metal surfaces. Daniel et al. studied nitrous oxide decomposition on the Rh( 100) surface and observed that nitrous oxide dissociatively adsorbed, leaving
T. W. Dann et al. /Applied
Catalysis B: Environmental
6 (1995) I-10
9
an oxygen overlayer on the Rh( 100) surface, and giving off N2 as a desorption product [ 151. It has been clearly shown that oxygen-covered metal surfaces are not as catalytically active as bare-metal surfaces, and thus, any mechanism that consumes surface oxygen and keeps catalytically active sites open for nitrous oxide dissociation will enhance the rate of conversion of N,O to Nz and OZ. Since carbon monoxide acts as a reducing species to form carbon dioxide with surface oxygen via reaction ( 1) , an enhanced rate of nitrous oxide conversion will result. An analogous mechanism is suggested for nitrous oxide decomposition on oxidized rhodium catalysts (Rh,O,) .There is strong evidence rhodium is fully oxidized to Rh,03 following oxidation, as reported by Van’t Bilk et al. [ 161 and Schwartz and Schmidt [ 121. Nitrous oxide dissociation has been reported on CaO [ 171, and thus it is plausible that nitrous oxide dissociation also occurs on rhodium oxide. 4. Conclusions In low concentrations, SO*, CO,, and NO do not significantly effect the conversion of nitrous oxide to nitrogen and oxygen over supported rhodium catalysts. These results suggest that rhodium catalysts may provide a basis for significantly reducing nitrous oxide emissions from coal combustion facilities using catalytic post processing of flue gases. Additionally, adding carbon monoxide significantly reduces the temperature at which 95% conversion of nitrous oxide occurs. It may be practical to feed carbon monoxide or another reductant gas into the reactor feed stream just as ammonia is fed into the catalytic reactor bed for the reduction of NO, emissions [ 18,191. Several more obstacles need to be cleared before a scaled up catalytic reactor can be installed and operated. First, and most importantly, will be the question of cost. Although an effective catalyst for nitrous oxide control, the prohibitive cost of rhodium catalysts when scaled up to an industrial sized reactor precludes its practical use. The second problem will be the positioning of the reactor. Since exit gases leave the stack at approximately 373-473 K, the reactor will have to be positioned closer to the FBC. This will require redesigning or relocating the emission equipment after the exit gases leave the combustor. Catalytic reduction of nitrous oxide has many possibilities. Understanding the reaction mechanisms and kinetics of nitrous oxide decomposition on rhodium based catalysts in the presence of other flue gas compounds will be of great importance for continued progress in this area. Acknowledgements
The authors acknowledge the UND-Energy and Environmental Research Center (UND-EERC) and the Chemical Engineering Alumni Fund at the University of
10
T. W. Dann et al. /Applied
Catalysis B: Environmental 6 (1995) l-10
North Dakota for financial assistance with this project. The Association of Western Universities (AWU) summer fellowship to KS is also gratefully acknowledged. Funding for the electron microscopy studies was through a grant from the NSF Center for Interfacial Engineering at the University of Minnesota-Twin Cities.
References [l] [2] [3] [4] [5] [6] [7] [S] [ 91 [lo] [ 1 l] [ 121 [ 131 [ 141 [ 151 [ 161
[ 171 [IS] [ 191
R.E. Dickinson and R.J. Cicerone, Nature, 319 (1986) 109. M.A.K. Khalil and R.A. Rasmussen, Tellus, 35B ( 1983) 161. A.R. Mosier and D.S. Chime], Chem. Ind., (December 2, 1991) 874. M.H. Thiemens and W.C. Trogler, Science, 251 (1991) 932. M. Takeshita, L.L. Sloss and I.M. Smith, IEA Coal Research, November (1993) 1. M.A. W6jtowicz, J.R. Pels and J.A. Moulijn, Fuel Process. Technol., 34 ( 1993) 1. M.D. Mann, M.E. Collings and P.E. Botros, Prog. Energy Combustion Sci., 18 (1992) 447. J.N. Armor and Y. Li, Appl. Catal. B, 1 ( 1992) L21. Electric Power Research Institute (EPRI) , Pilot-Scale Circulating Fluidized-Bed Combustion Evaluation Facility, EPRI TR-100829, Project 1179-32, Final Report, December, 1992. R.E. Cameron, USA Microsc. Anal., May ( 1994) 17. R.W. McCabe and C. Wong, J. Catal., 121 (1990) 422. J.M. Schwartz and L.D. Schmidt, J. Catal., 138 ( 1992) 283. B.W. Rossiter and J. F. Hamilton, Physical Methods of Chemistry: Volume IV Microscopy, Wiley, New York, 1991, pp. 203-238. B.K. Cho and B.H. Shanks, General Motors Research Laboratories Research Report PC-421, March 1988. W.M. Daniel, Y. Kim, H.C. Peebles and J.M. White, Surf. Sci., 111 (1981) 189. F.J. Van’t Blik, J.B.A.D. van Zon, T. Huizinga, J.C. Vis, D.C. Koningsberger and R. Prim, J. Am. Chem. Sot., 107 (1985) 3139. M. Nakamura, H. Mitsuhashi and N. Takesawa, J. Catal., 138 ( 1992) 686. S.T. Darian, J.W. Eldridge and J.R. Kitrell, Environ. Prog., l-2 ( 1982) 84. J.W. Eldridge, J.R. Kittrell and F.G. Medros, Ind. Eng. Chem. Res., 28 (1989) 1171.