Removal of Nitrogen Monoxide over Copper Ion-exchanged Zeolite Catalysts

Transcription

1 Removal of Nitrogen Monoxide over Copper Ion-exchanged Zeolite Catalysts Masakazu IWAMOTO*, Noritaka MIZUNO, and Hidenori YAHIRO Catalysis Research Center, Hokkaido University, Kita-ku, Sapporo 060 (Received December 17, 1990) Direct decomposition and selective reduction of nitrogen monoxide over copper ion-exchanged zeolite catalysts are proposed as new methods for removal of NO. The copper ion-exchanged ZSM-5 zeolite (Cu-Z) was the most active catalyst for decomposition of NO. The activity of Cu-Z zeolites increased with increase in the exchange level. The zeolites with copper ion-excange levels of 100% or more, which could be prepared by repeating ion exchange of the ZSM-5 zeolite using aqueous copper(ii) acetate solution or addition of ammonia into the aqueous copper(ii) nitrate solution, showed significantly high activity even in the presence of oxygen and at high GHSV region. It was clarified concerning Cu-Z, by using IR, ESR, phosphorescence, TPD, and CO adsorption measurements that (1) the Cu2+ ions exchanged into zeolite were reduced to Cu+ and/or Cu+-Cu+ through evacuation at elevated temperature, (2) after exposure to oxygen at 773K and subsequent evacuation, about 40% copper ions in zeolite existed as Cu+ ions, (3) the NO- species formed by adsorption of NO on Cu+ would be an intermediate in the NO decomposition, and (4) redox cycle of decomposition reaction. Selective reduction of NO by hydrocarbon in the presence of oxygen was first found by the authors and Cu-Z was remarkably effective for NO removal at temperatures as low as K. The activity for this selective reduction in NO+C3H6+O2 system was not poisoned very much by addition of SO2. The conversion into N2 was changed to 85% (773K) in the presence of SO2, from 100% in the absence of SO2, which is in contrast with the fact that the catalytic activity for direct decomposition NO was completely lost on adding the same amount of SO2. Furthermore, the reduction rate over Cu-Z at 573K was higher than those over H-zeolite and alumina catalysts at 723 and 773K, respectively, which have been reported to be active, after findings by the authors. 1. Introduction Air pollution and acid rain seriously affect the terrestrial and aquatic ecosystems and, therefore, are very important social problems that must be solved as soon as possible1). The exhaust gases from engines of vehicles and industrial boilers contain mainly carbon oxides (CO and CO2), nitrogen oxides (NOx), hydrocarbons, sulfur dioxide, particles, and soot. At present, one of the highly significant problems is removal of NOx, which is produced during high temperature combustion. In particular, the decomposition or reduction of nitrogen monoxide (NO) is a dominant target to be achieved because NO is an inert and also the major component of NOx in the exhaust gases. It is well known that NO is thermodynamically unstable relative to N2 and O2 at temperatures below 1,200 K2), and its catalytic decomposition is the simplest and most desirable method for its removal. To date, however, no suitable catalyst, of sustained high activity, has been found. This * To whom correspondence should be addressed. is due to the fact that oxygen contained in the feed, or produced in the decomposition of NO, competes with NO for adsorption sites2). Thus, high reaction temperature and/or gaseous reductants are required to remove surface oxygen and regenerate catalytic activity. At present catalytic reduction processes employing NH3, CO, or hydrocarbons on V2O5-TiO2 or Pt-Pd-Rh catalysts have been put to practical use. On the other hand, copper ion-exchanged zeolites3)-9), Pt/Al2O3, YBa2Cu3Oy supported on MgO10), Sr2+-substituted perovskite11)-13), and Ag-Co3O414) have recently been reported as candidates for the catalyst for direct decomposition of NO. Among these catalysts, copper ion-exchanged zeolites, first reported by Iwamoto and coworkers, are the most suitable for the reaction15). It would be very desirable to enhance the catalytic activity of Cu-zeolites to develop a new process for the removal of NO from exhaust streams. In the present paper, the authors would like to introduce recent results concerning removal of NO. The correlation among exchange level, zeolite structure, and catalytic activity was first clarified to characterize active sites for the reaction. It will be

2 376 reported in the 2nd section that repeated ion exchange of ZSM-5 zeolite with Cu2+ solution resulted in excess loading of copper ions above 100% exchange level, and catalytic activities of resulting Cu-Z zeolites were very high for the decomposition of NO9),16). Subsequently, in the 3rd section physicochemical characterization of zeolites and temperature programmed desorption (TPD) and IR experiments of NO are reported. The results will elucidate that copper ions exist as Cu2+ and Cu+ after evacuation at elevated temperature and most of copper ions in the ZSM-5 zeolite are available for NO adsorption17). A reasonable reaction mechanism of NO decomposition will be suggested. These findings would serve to further progress in the chemistry of ion-exchange and NO decomposition. In a series of experiments to reveal the effects of coexisting gases, we have first found that NO is selectively reduced by hydrocarbons in an oxidizing atmosphere and its rate is remarkably enhanced by the presence of oxygen18), excepting patent literatures19). Details of this new type of catalytic reaction has been summarized in the 4th section. These findings break a widely-accepted concept that ammonia is the only selective reluctant for NO in presence of O2, and would lead the development of a novel catalytic process for NO removal. Experimental Parent zeolites, L (silica/alumina=6.0), mordenite(10.5, 18.9), ferrierite(12.3), and ZSM-5(23.3) were supplied by Tosoh Corporation. Each sample was termed L, M, F, and Z, respectively. The zeolites were ion-exchanged with usual procedure: washed with dilute NaNO3 solution, ion-exchanged in an aqueous copper acetate solution of an adequate concentration, washed with water, and dried at 383K overnight. A typical procedure for ion-exchange of ZSM-5 were as follows: Approximately 15g of ZSM-5 zeolite was ion-exchaned in 1dm3 of copper(ii) acetate solution with overnight, and filtered. The wet cake obtained was again ion-exchanged in new copper(ii) acetate solution. After repetition of ion exchange treatment as desired, the sample was washed and dried. A rapid preparation of highly copper ionexchanged ZSM-5 zeolites by addition of basic compounds was carried out as follows: About 15g of ZSM-5 zeolite was stirred for 24h in 1dm3 of aqueous copper(ii) nitrate solution The ion-exchange level was determined by atomic absorption spectroscopy assuming that one Cu2+ is exchanged with two Na+ ions. Hereafter the sample will be abbreviated as Cu-Z-100 (cationzeolite structure-degree of exchange). The NO decomposition was carried out in a flow reactor made of stainless or quartz tube. Zeolite catalyst ( g) was placed in the reactor and heated at 773K for 5h under a He stream min-1), in order to remove such impurities on the zeolite surface as water, carbon dioxide, and oxygen. The catalytic reaction was then started. A gas mixture of NO ( vol%), reducing gas (0-1,000ppm), O2 (0-12%), SO2 (0-300ppm), and He (balance) was fed at a flow rate (F) of 30- catalyst weight) unless otherwise stated. The reducing gas used was H2, CO, CH4, C2H4, C3H6, or C3H8. The reaction temperature was increased stepwise from K and the reaction was carried out at each temperature until the conversions reached constant. The gas composition was analyzed by gas chromatography using Porapak Q (N2O, CO2, and hydrocarbons) and Molecular Sieve 5A (O2, N2, NO, CH4, and CO) columns. TPD technique was applied here to the NOzeolite system to determine amount of NO adsorbed and catalyst state. TPD profiles were measured with the same apparatus as was used for flow reaction. After the catalyst was used in steady-state catalytic decomposition of NO, zeolite was cooled to 323K from 773K in He and then exposed to gas mixture of NO (2.05%) and He for 60min. Subsequently, the gas was again changed and desorbed gases were analyzed by gas chromatography. Infrared, ESR, and phosphorescence spectra were recorded at room temperature. 2. Decomposition of NO over Excessively Copper Ion-exchanged ZSM-5 Zeolite 2.1 Catalytic Activity of Copper Ion-exchanged Zeolites Triple or more ion exchange procedures resulted in excess loading of copper ions exceeding 100% exchange level, as has been reported for Y-type zeolite20)-22). The cause for this exchange behavior and the state of copper ions introduced into the zeolite are very interesting problems in the field of ion exchange and zeolite chemistry, and (concentration are summarized in Table 2) was these will be taken up for discussion in a later gradually added to the ion-exchange solution until section. It should be noted here that the copper ph of the solution increased to a desired value. ion-exchanged zeolites obtained showed very high

3 377 Fig. 1 Time Course of NO Decomposition over catalytic activities for the decomposition of NO, as mentioned in the following paragraphs. The time course of NO decomposition over Cu-Z-112 is shown in Fig. 1. The extent of conversion of NO and into N2 and O2 were gradually increased with reaction time up to ca. 3h, when it reached a steady state. No deterioration of the catalyst was observed at 723K even after 30h of continuous service. In the present work, experiment was not carried out beyond 30h. The conversions of NO and into N2 and O2 at a steady state of the reaction were approximately 95%, 75%, and 55%, respectively. The results indicate that the copper zeolite with exchange level of 100% or more shows steady activity which is evidently higher than those reported previously6). The decrease in the amount of NO (conversion of NO) was greater than the amount of N2 and O2 produced. As suggested previously6), the discrepancy is due to the formation of NO2, which can not be detected by usual gas chromatography. Li and Hall indeed confirmed that the reaction stoichiometric and formation of NO2 take places in the reactor lines and chromatographic column during Temperature=723K analysis23). The respective catalytic activities of Cu-Z samples with various exchange levels for the decomposition of NO were measured as a function of reaction temperatures. The results with Cu-Z-143 are shown in Fig. 2 as an example. A small amount of N2O appeared below 623K, indicating that decomposition of N2O, which is one of the intermediate products during NO decomposition, proceeds faster than that of NO at high temperatures. The same results were observed in the decomposition of NO on Cu-Y zeolites3). The catalyst has predominant activity at around K, and conversion levelled off at K. Then the conversion decreased slightly at higher temperatures. The temperature dependence shown in Fig. 2 is in good agreement with that reported previously for Cu-Z-73 zeolite5). The slight decrement of catalytic activity at higher temperatures is not attributed to deactivation of the catalyst, since the conversion did not change when the reaction temperature was raised and lowered stepwise between 773K and 923K. The decrement of the conversion above 773K, therefore, may be due to reversible change of the structure of active site and/or adsorbability of NO. It should be noted that, first the decrease of the activity above ca. 773K was much improved when the copper ion exchange level increased, secondly the maximum conversion on each catalysts was mostly observed around 723K. For the purpose of elucidating the active sites, it is of interest to establish correlation among framework structures of parent zeolites, aluminum content, exchange level of copper ions and catalytic activities. Here several kinds of zeolites were used and typical results are quantitatively depicted in Fig. 3. Clearly the catalytic activity is dependent on the zeolite structure. Within the scope of present experiments, ZSM-5 was the most active catalyst at around 773K and L-type was the least, while at 973K, Cu-F was more active than ZSM-5. The amount of active copper ion for the adsorption of NO, estimated by the amount of NO irreversible adsorbed on zeolites, is shown in Table 1. The ratio, of the amount of active copper ion for the NO adsorption to that of whole copper ion exchanged into zeolites, was very high (94% or 85%) for ZSM-5 or ferrierite zeolite, while it was low (43% or 40%) for mordenite or L-type zeolite. From Fig. 3 and Table 1, the catalytic activity per copper ion, active for NO adsorption (TOF), can

4 378 Table 1 Results of TPD Experiment over Copper Ion-exchanged Zeolites a) The ratio of the amount of NO desorbed to that of copper ion exchanged into zeolite. b) The amount of oxygen desorbed is too small to be judged clearly. Fig. 3 Temperature Dependence of Conversion of NO over Various Copper Ion-exchanged a: Cu-Z-122, b: Cu-F-76, c: Cu-M-88, d: Cu-L-45. Fig. 4 Correlation between the Al Content of Each Parent Zeolite and the Catalytic Activity per Copper Ion Accessible to NO be evaluated. The TOF decreased with the decrement of Al content (Fig. 4). The effectiveness of copper ion, therefore, is controlled by the zeolite structure and the exchange level (see the 3rd section), and TOF is controlled by the Al content. It is noteworthy in Fig. 3 that the most active temperatures of the ZSM-5 and mordenite zeolites were around K, whereas that of the ferrierite zeolite was 873K. This is probably due to the higher desorption temperature of oxygen from Cu-F (see Table 1). The correlation between catalytic activity at 723K and exchange level of copper ion is depicted in Fig. 5. In the present experiments, 80-85% conversion into N2 was the best result attained. In the figure, two noticeable facts were observed. One is that the conversions of NO and into N2 and O2 depict S-shaped dependence on exchange level; decomposition rate gradually increased at lower exchange level and increased sharply at exchange level of above 40%. The phenomenon may be explained by the following two possible causes: (1) ZSM-5 zeolite has two or more cation exchangeable sites24), one of which is most readily exchanged with a copper ion, and is inactive for the decomposition of NO like SI in the Y-type zeolite3). The other site is active for the decomposition of NO, and is not readily ion-exchanged below the

5 379 ion-exchange level of 40%. (2) The decomposition of NO proceeds only under cooperation of two adjacent active sites which could be formed at high levels of ion exchange. The reaction order of th with respect to PNO, described in the next section, might suggest that the latter is correct, though further study is necessary to make the correlation clear. Another interesting point is the catalytic activities of Cu-Z above the exchange level of 100%; the rate of decomposition to N2 and O2 monotonically increased even above the exchange level of 100%. The copper ions excessively loaded into the ZSM-5 zeolite framework, therefore, are also effective for the decomposition of NO. This result suggests that excess loading of copper ions is one possible way to develop more active catalyst for the decomposition of NO. 2.2 Preparation of Excessively Copper Ionexchanged ZSM-5 Zeolites Cu-Z having copper ion-exchange levels of above 100%, showed high catalytic activity in the direct decomposition of NO, as described in the previous paragraph. At present, however, one problem remains to be solved, and that is, it is impractical to prepare excessively copper ionexchanged ZSM-5 zeolites with copper(ii) nitrate or sulfate solution requiring repetitions of ionexchange process to prepare them with copper acetate solution. In this regard, it was found that the ZSM-5 zeolites with excess loading of copper ions could be readily prepared in a single step through addition of basic compounds such as NH4OH and Mg(OH)2 into the solution16). Table 2 shows the effect of addition of basic compounds. Clearly, addition of basic compounds to the mother solution increased the amount of copper ions exchanged in a single step. The exchange levels were varied with basic additives and decreased in the following order: NH4OH>Ba(OH)2>Ca(OH)2>NaOH=KOH> pyridine=mg(oh)2=ethylenediamine. It is revealed in Table 2 that the addition of basic compounds except ethylenediamine produced exchange level of % at ph=7.5, and 100% at ph=6.0, suggesting that ph is probably a more influential factor to determine ion exchange level than the kind of basic compounds. The less effect of ethylenediamine would be due to the larger size of copper(ii) complex coordinated by ethylenediamine than the pore diameter of diameter of ZSM-5 zeolite. To confirm the dependency of exchange level on ph, the effect of addition of ammonia into the solution has been studied in more detail and is summarized in Fig. 6A. The exchange level increased with the increase of ph from 4 to 9 and settled to nearly constant at above ph 9. Above ph 9, all of Cu2+ ions were loaded into the zeolite. This correlation and the tendency observed from Table 2 suggest that the exchange between Na+ and Cu2(OH)3+, Cu(OH)+, Cu2(OH)22+, or Cu3(OH)24+ formed at specific ph25) probably proceeds in zeolite. The catalytic activity of resulting zeolites for the decomposition of NO is shown in Table 2 and Fig. 6B. The extent of conversion of NO increased with increase in ph, attaining a maximum at ph of 7.5, and then slightly decreased at higher ph region. No destruction of the zeolite lattice itself was observed by X-ray diffraction patterns after the ion exchange or the catalytic run. The decrement of catalytic activity of Cu-Z prepared at higher ph region may be due to change of state of copper ions in the zeolite. The conversions of NO and into N2 and O2 at 823K over the Cu-Z-140 prepared at ph of 7.5 were 80, 72, and 60%, respectively, indicating that the activity of this catalyst for catalytic Table 2 Effect of Addition of Base on the Ion-exchange Level and Catalytic Activities of the Resulting Cu-Z Zeolite a) The initial amount of Cu2+ in the solution was 1.50 equiv. to that of Na+ in zeolites. b) After 24h. c) The degree of ion-exchange was measured by atomic absorption spectroscopy after the zeolite obtained was dissolved in HF solution. d) The value is the conversion into N2 under the following conditions; PNO=4,950ppm, catalyst weight=0.5g, flow temp.=773k. e) Not measured. f) In this case three experiments were performed. g) Catalytic activity of the 146% exchanged Cu-Z zeolite. h) Saturated solution. i) Ethylenediamine.

6 380 The ph was varied by the addition of ammonia. The conditions for ion-exchange were the same as Table 2 except ph. The catalytic activities were measured at min-1, and temperature=823k. Flags attached indicate differnt lots of catalyst. Fig. 6 Exchange Level of Cu-Z (A) and Its Catalytic Actibity (B) as a Function of ph of Aqueous Copper (II) Nitrate Solution decomposition of NO is comparable to that of Cu-Z prepared by repeated ion exchange method. The other catalysts prepared with addition of alkaline or alkaline earth hydroxide, except Mg(OH)2, showed lower activities than that without additives, as shown in Table 2. Ammonia adsorbed on the zeolites is well known to be desorbed at K26) and, therefore, in the Cu-Z zeolites prepared with addition of ammonia, there would be no ammoina at such high reaction temperatures. By contrast, alkaline or alkaline earth ions would remain on the surface. It follows, therefore, that not only ph but also the kind of basic additive is a very important factor for determining catalytic activity. The present results have revealed that copper ion exchange with addition of ammonia is one of the useful methods for excess loading of copper ions, and that the resultant Cu-Z is potentially active for the catalytic reaction. 2.3 Effects of NO Pressure, Contact Time and Coexisting Gases on Decomposition of NO Pressure dependence of decomposition of NO was examined over Cu-Z-122 at 753K. The results are depicted in Fig. 7. The conversion of NO increased with increment of NO pressure. On the basis of data in Fig. 7, we could obtain linear Fig. 7 Dependence of the Degree of Conversion of NO on Partial Pressure of NO and Contact Time correlation lines in log-log plots of partial pressure of NO (PNO) vs. reaction rate of NO, and the rate was proportional to the th order with respect to PNO. As indicated clearly in Fig. 7, the present results cannot be treated as a differential reaction system owing to excessively high conversion levels of NO and, therefore, the above values are not accurate, while the values would be useful guides for discussing the reaction mechanism and/or for comparison with the other catalytic systems. Contact time dependence of conversion of NO at 753K is also shown in Fig. 7. The conversion of NO increased with contact time and reached (GHSV=60,000h-1). Not only the conversion of NO but also those into N2 and O2 increased with increase in contact time. In a separate experiment, the NO conversion reached ca. 100% results demonstrate that the present catalyst has excellent activity for the catalytic decomposition of NO even at such high GHSV, which is important for applying it into practical use. The effects of addition of carbon dioxide, water vapor, oxygen, or sulphur dioxide on the catalytic activity of copper ion-exchanged zeolites were examined. When CO2 was added to the reactant gas flow, no reduction in catalytic activity was observed. The addition of H2O resulted in decrease of the activity. The effect, however, is fully reversible; that is, the activity is reduced in the presence of H2O but regenerated by its departure. Effects of addition of oxygen was dependent on

7 381 Fig. 8 Catalytic Activites of Cu-Z-122 and -89 the zeolite structure, the degree of exchange of Cu2+ ions, and the relative pressure of oxygen to that of NO. Typical results of Cu-Z are shown in Fig. 8. For example, when oxygen (8vol%) was added to a mixture of 0.47% of NO and 91.53% of He, conversion of NO decreased from 55%(without oxygen) to 40% at 753K on Cu-Z-122. On the other hand, over Cu-Z-89, conversion of NO decreased to 5% from 47% on addition of 3% oxygen to the NO-He (NO=0.5vol%) stream. It should be noted that catalytic activity of Cu-Z-122 is little influenced by the presence of oxygen in the feed. An excess loading of copper ions (exchange level above 100%) brings about an increase, not only in the catalytic activity but also in the tolerance to oxygen. When partial pressure of NO was only 1,000ppm, the extent of the conversion into N2 was diminished from 23% (without oxygen) to 5% by the addition of 0.5% O2 on Cu-Z-152 at 773K and 0.3g- level was also a function of ratio of PO2/PNO. SO2 completely poisons the activity at K. With ZSM-5, desorption treatment of adsorbed SO 2 at higher temperature resulted in regeneration of the decomposition activity. SO2 would compete with NO for the adsorption sites and prevent the catalytic reaciton. 3. Characterization of Copper Ions Exchanged into ZSM-5 Zeolite and Reaction Mechanism of NO Decomposition 3.1 Changes in ESR and Phosphorescence Spectra upon Evacuation at Elevated Temperatures ESR signal attributed to hydrated Cu2+ ion was Fig. 9 Phosphorescence Spectra over Cu-Z-138 after Evacuation at (a) 573K, (b) 623K, (c) 673K, (d) 723K, (e) 823K, and (f) 973K observed for Cu-Z just after it was prepared with three or more repetitions of the ion exchange procedure27). After the sample was evacuated at elevated temperatures, the signal changed to that attributed to Cu2+ ion located on the zeolite lattice28). The signal intensity decreased with the increase of the evacuation temperature and was almost zero at ca. 773K. In this stage, the color of the sample was white. The treatment that the prepared sample is evacuated at 773K is called pretreatment 1. During above treatment, phosphorescence signal attributable to Cu+ ion was observed29). The phosphorescence spectra were measured at room temperature after the sample was evacuated at elevated temperatures for 30min. The spectra of Cu-Z-138 are shown in Fig. 9, as an example. Two emission spectra were observed at 480 and 540nm and the excitation spectra both showed maxima at 280nm. The values of 480 and 540nm are clearly red-shifted in comparison with 456nm of free Cu+ ion. The value of 280nm was blueshifted from that of Cu+ ion located in the ionic crystals (ex., 372nm for CuCl and 395nm for CuBr), and was close to that of Cu+ ion doped in NaI (257nm) or LiI (245nm). These observations indicate, probably, that the distance between the nuclei of Cu+ ion and oxide ion on the zeolite lattice is considerably small. Taking into account the above results and the phosphorescence signals at 480nm (peak I) and 540nm (peak II) are temporarily assigned to the emissions from Cu+ ion isolated and Cu+ dimer or trimer, respectively. In the present paper, the former is called Cu+ monomer, and the latter is called Cu+ dimer. The dependencies of the phosphorescence spectra on pretreatment temperature and ionexchange level have been studied. The intensity of peak I increased with the increment of the

8 382 evacuation temperature, and reached constant at K (see Fig. 9d-f). The intensity of peak II increased with increase in evacuation temperature, which reached maximum at 823K (Fig. 9e), and then decreased at 973K (Fig. 9f). Concerning dependency on the exchange level, it was found that peak I was observed even at low exchange level of copper while the intensity of prak II was considerably weak at the same exchange level but dramatically increased at higher exchange levels. For example, the intensity ratios of peak I to peak II for Cu-Z-28 and Cu-Z-138 were ca. 10:1 and 1:2, respectively. The fact shows that the Cu+ monomer is a predominant species when the exchange level is low but the number of Cu+ dimer more rapidly increased as the exchange level increased. The presence of the Cu+ ions in the pretreated Cu-Z-116 was further confirmed by the IR measurements. The admission of CO onto the zeolite resulted in the appearance of a strong IR absorption band at 2,150cm-1 assigned to Cu+-CO species31). These observations clarified that the high temperature treatment brought about the generation of Cu+ ions. As it was widely accepted that only Cu+ ion can irreversibly adsorb CO at room temperature while Cu2+ ion and Cu0 can not32), the quantitative analysis of CO irreversibly adsorbed makes it possible to measure the amount of Cu+ ion which is active for the CO adsorption. The results obtained by volumetric method suggested that the amount of Cu+ ion after pretreatment 1 corresponded to about 90% of the whole amount of copper in the zeolite. Thus, it is clear that pretreatment 1 causes the reduction of Cu2+ to Cu+. The reduction of Cu2+ ion is probably due to the dimerization or polymerization of Cu(OH)+ species by dehydration to form Cu2+-O2--Cu2+ species and subsequent reduction be attributable to Cu2+ ion in square planar symmetry with tetragonal distortion. The number of Cu2+ ion detected by ESR was quantitatively estimated by comparison of double integration of the ESR spectrum with that of the standard sample where the Cu-Z samples prepared by acetate solution or addition of ammonia are plotted shows that the state of copper is not affected by the addition of ammonia during the preparation. The number of Cu2+ ion detected by ESR increased until the exchange level reached ca. 120%, and it levelled off above 120%. The number of Cu+ ion evaluated by the amount of CO adsorbed (hatched region in Fig. 10) increased dramatically in the region of exchange level of %. These results reveal that there exist Cu2+ and Cu+ ions in Cu-Z even after the oxidizing treatment. For example, the distribution of copper ions exchanged into Cu-Z-120 is ca. 50% of Cu2+, 40% of Cu+, and 10% of unknown species after pretreatment 2. As shown in Fig. 5, catalytic activity of Cu-Z 3.2 State of Cu2+ and Cu+ Stabilized by Oxygen Treatment at 773K and Catalytic Activity for Decomposition of NO In this section, the state of copper under reaction conditions will be discussed. The state of copper in the sample treated under oxidizing atmosphere at 773K after pretreatment 1 would be considered to be close to that in the reaction conditions because the decomposition of NO to form O2 was carried out after the catalyst was treated in a He stream at 773K. ESR spectra of Cu-Z catalysts were measured after Cu-Z was subjected to pretreatment 1, oxidized in O2 (100 Torr), and evacuated at 773K (pretreatment 2). Three ESR by ESR and CO adsorption measurements, respectively. The hatched region indicates the amount of total copper ions in Cu-Z. Fig. 10 Distribution of Copper Ions in Cu-Z as a Function of Exchange Level after Oxygen Treatment at 773K

9 383 increased drastically above the exchange level of 50% and still increased above 100%. The dependence of activity on exchange level corresponds well to that of peak II intensity in the phosphorescence spectra or the number of Cu+ ion on exchange level. The temperature at the maximum intensity of peak II (823K, Fig. 9) agrees with that of catalytic activity ( K, Fig. 2). Further, peak II is characteristic of highly copper ion-exchanged ZSM-5 zeolite; for example, it is not observed for Cu/SiO2, inactive for catalytic decomposition of NO, and the intensity of peak II is very weak for Cu-Y, of which the activity is lower. These findings suggest that active sites for the decomposition of NO are Cu+ dimer species. 3.3 Adsorbed Species of NO on Copper Ionexchanged ZSM-5 Zeolite The reaction mechanism was studied by IR and TPD techniques combined with an isotopic tracer method17). IR spectra were recorded at an ambient temperature in a NO atmosphere after evacuation of a self-standing Cu-Z-81 zeolite wafer at 773K. Typical results in the region of 2,500-1,500cm-1 are depicted in Fig. 11. Additional absorption bands were detected below 1,500cm-1, but these were not clear because of the intense band of CaF2 window. These absorption bands can be assigned to NO2 or NO3 species with a certain negative charge, which would be produced through the disproportionation of NO molecules or through the reaction of NO with oxygen atoms generated in the decomposition reaction. No further consideration was given here to these adsorbed species because it is scarcely possible to consider that these are intermediates in the decomposition of NO. In order to assign each absorption band, studies on the adsorption of isotopically substituted NO on the Cu-Z sample were made. Identical Cu-Z surfaces were exposed to 15NO (Fig. 11c) and equimolar mixture of 14NO and an 15NO (Fig. 11d). It could readily be observed in the presence of pure 15NO that all bands depicted in Fig. 11b were shifted to respective lower wavenumbers. The 1:1 mixture system provided conclusive evidence for the dinitrosyl adsorbate. We observed that a new band appeared at 1,719cm-1 after introduction of the mixture of 14NO and 15NO and the relative intensities for the three peaks at 1,734, 1,719 and 1,703cm-1 were approximately in the ratio 1:2:1. Results indicate the presence of a Cu(14NO)(15NO) adsorbate and confirm assignment of the band at 1,734cm-1 in the 14NO adsorption (Fig, 11b) to the dinitrosyl species. It was also expected in this experiment that a new band would appear between 1,827 and 1,795cm-1 attributed to Cu(14NO)- (15NO). Unfortunately, we could not observe any new band with clear resolution in this region because of overlapping of the Cu-NO bands at 1,813 (14NO-) and 1,781cm-1 (15NO-). Owing to further confirmation of the above assignment, correlation between peak areas of 1,827cm-1 and 1,734cm-1 bands was investigated. The peak area ratio was constant in different exchange levels and different NO pressures as shown in Fig. 12, indicating that the assignments are reasonable and the angle between two NO (a) Background spectrum. (b) Exposure to 14NO (18.2 Torr). (c) Exposure to 15NO (18.5 Torr). (d) Exposure to 14NO (9 Torr) +15NO (9 Torr). Fig. 11 Infrared Spectra Observed at Room Temperature over Cu-Z-81 Fig. 12 Correlation of Peak Area between 1,827cm-1 and 1,734cm-1 Bands over Cu-Z-122

10 384 molecules is constant. The relative intensity of the two bands at 1,827 (Isym) and 1,734cm-1 (Iasym) is ant ratio of the double peaks existed in the present experiments and was 1.60, which indicates that Cu-Z is included in the range of angles reported so far. As for the assignment of other IR bands, the band at 1,813cm-1 can be assigned to single 14NO-, according to Ref. 35). As stated in the next section, NO- and (NO)2 adsorbate were adsorbed on the same sites and amounts changed complementally with NO pressure and, therefore, (NO)2 would be anionic, i.e., (NO)2-. The adsorption of NO has been widely studied on various catalysts by many workers. On the basis of these assignments and the adsorption of isotopically labelled NO, the bands at 2,500-1,500cm-1 in Fig. 11b are attributable to the adsorbed species. Here, the bands at 1,964cm-1 remained unassigned: The band may be assignable to (Cu)n-NO+, since the band at 1,906cm-1 is due to Cu-NO+. The bands at 2,238 and 2,125cm-1 in Fig. 11b is attributed to N2O and NO2 formed by the reaction of NO, based on the IR spectra obtained, after the adsorption of N2O and NO2 onto Cu-Z, in a separate experiment. It should be noted that the two bands at 1,827 and 1,734cm-1 were assigned to an anionic dinitrosyl species, and earlier investlgators32),33) had not detected these features on copper ion-exchanged zeolites. 3.4 Active Sites for NO Adsorption Shown in Fig. 13 is the dependency of the intensities of IR bands at 1,827cm-1, 1,734cm-1 on the NO pressure. The intensities were measured after Cu-Z-116 was exposed to NO for 5min. Clearly, the intensities of former two bands increased and that of the latter decreased with the pressure of NO. It should be noted here that the sum of the absorbances at 1,814cm-1 and 1,734cm-1 (or 1,814 and 1,827cm-1) was constant, independent of NO pressure, or the peak intensities of NO- and (NO)2- changed complementally. This conclusively indicates that (NO)- and (NO)2- were formed on the same The dashed line indicates the sum of solid lines. Fig. 13 NO Pressure Dependence of Absorbance of NO- (1,814cm-1) and (NO)2- (1,827 and 1,733cm-1) over Cu-Z-112 sites. The pretreatment of the sample wafer has been found to affect NO adsorption. When the sample was cooled to room temperature in an oxygen atmosphere after evacuation at 773K, the admittance of NO resulted in the appearance of some absorption bands such as Cu-NO+ in a manner similar to that shown in Fig. 11, but hardly yielded the bands of Cu-NO- and Cu-(NO)2-. This indicates absence of active sites on the sample for adsorption of NO molecules with a negative charge. On this sample, furthermore, there was no change in the bands with time, which differs from that described in the next section. When the sample wafer was again evacuated at elevated temperature, admittance of NO resulted in an IR spectrum including Cu-NO- and Cu-(NO)2- bands, which was essentially the same as shown in Fig. 11. It follows that the appearance of NOand (NO)2- species required treatment of the copper ion-exchanged zeolite at higher temperature. It has already been shown that Cu2+ ion was reduced to Cu+ ion by the treatment of Cu-Z zeolite at high temperature. Further, the admittance of NO (36.8 Torr) after the selective adsorption of CO (121.5 Torr) on Cu+ ion resulted in the formation of NO+ (1,906cm-1) and Cu+-CO (2,150cm-1) bands. In this experiment, no bands attributed to NO-, (NO)2-, N2O, and NO2 were observed. These results show that Cu+ ion is active for the formation of the mono- and di-nitrosyl species while Cu2+ ion is the site for the adsorption of NO+ species. 3.5 Dynamic Change of the IR Bands The interesting and novel point in the IR measurement is the change in peak intensities with adsorption time. The NO- and (NO)2- species

11 385 Fig. 14 Time Course of Decomposition of NO over Room Temperature Fig. 15 Time Course of the Reciprocal of Absorbance decreased while the NO+ species increased. These changes indicate progress of the decomposition reaction on the surface even at room temperature, because IR measurements perfomed in the presence of NO, that is, the decrease in NO-, was clearly not attributable to evacuation. The authors have carried out the decomposition of NO at room temperature (NO, 1.0%; He, balance; flow catalyst was exposed to a stream of He at 773K for 1h and, in fact, confirmed that progress of the surface reaction yields nitrogen and nitrous oxide molecules in the gas phase, as shown in Fig. 14. N2 was formed just after the NO was admitted and, then, formation of N2O was observed. A part of N2O, which was observed in Fig. 14, may result from the disproportionation of NO molecules where possibly formed NO2 did not desorb from the surface. NO was not detected within 60min. The amount of NO increased after 60min and reached an amount comparable to that in the inlet after 300min. These facts show the progress of the decomposition of NO, even at room temperature, over activated Cu-Z. The rate of formation of nitrogen molecules at room temperature decreased with time. This is due to poisoning of active sites by oxygen generated through the decomposition, just as which was usually observed for the other decomposition catalysts. The fact, that in TPD experiments the remaining oxygen could be desorbed only at higher temperatures, supports the above idea. It was also found in a separate experiment that NO admitted on the oxidized zeolite did not give any nitrogen molecules. These results are in good agreement with the fact that continuous catalytic decomposition in the flow system becomes vigorous at temperatures above K, as shown in Fig. 2. It follows that the catalytic cycle of the NO decomposition includes Cu+ ions as active centers, and NO- and/or (NO)2- species as reactive intermediates. From the decrement of IR spectra attributed to NO- or (NO)2- species, one can directly observe reaction courses and measure the reaction rates of surface intermediates. The decreases in the NOabsorption bands could well be plotted by secondorder rate equations, as shown in Fig. 15. The rate constant of the single type was roughly equal to that of the twin type, indicating that reactivities of these two species were similar. The conformity of reaction rates to second-order rate equations was consistent with the fact that catalytic decomposition of NO in a continuous-flow reactor was dependent on th order with respect to partial pressure of NO. The adsorption of both NO- species was so weak that it could be desorbed by brief evacuation at room temperature. The adsorption of NO- species, therefore, would depend on the partial pressure of NO. On the basis of above results, the reaction cycle might be suggested as follows: At higher temperatures above 573K, the oxygen produced can be removed by desorption from the zeolite surface and the regeneration of the active sites makes it possible to continue the catalytic decomposition cycle. 4. Selective Reduction of NO by Hydrocarbons in Oxidizing (A) of NO- and (NO)2- over Cu-Z-112 at PNO=39.9 Torr and at Room Temperature Atmosphere Copper ion-exchanged ZSM-5 zeolites were the most active for the decomposition of NO.

12 386 However, the activity is greatly decreased in the presence of an excess of oxygen and SO2. Recently, selective reduction of NO by hydrocarbons such as C2H4, C3H6, and C3H8 in an oxidizing atmosphere over copper ion-exchanged zeolite catalyst has been reported by the present group18),36),37). Immediately after the group's presentation, Held et al. have independently reported similar findings38). The new selective reduction of NO proceeds even in the presence of excess O2 and SO2, and would overcome the disadvantages or problems of the present reduction systems using dangerous and expensive ammonia. The results will be shown in this section. The effects of reducing gases have been classified into two groups, hydrocarbons (i.e., propene, ethene, and propane) and non-hydrocarbons (H2 and CO). The results of reduction by propene are summarized in Fig. 16A as the example by the former group. The addition of propene into the NO+He stream did not enhance the conversion into N2 below 723K, though the conversion greatly increased at higher temperatures. This indicates that propene is not an effective reductant for NO at Fig. 16 Temperature Dependene of Catalytic Activity for Removal NO over Cu-Z-152 in Various Conditions relatively low temperatures. By contrast, additional introduction of O2 into the NO+C3H6+He stream greatly enhanced the catalytic activity for the reduction of NO at temperatures as low as K. Such a dramatic acceleration of reduction of NO by oxygen had not been observed; the conversion times greater than that without oxygen, while a much smaller promotion by a factor of 5-10 was reported in the selective reduction of NO by ammonia over V2O5/TiO2-based catalyst39). Under the reaction condition shown in Fig. 16A, propene was completely oxidized to CO2 and H2O at above 498K. The conversion into N2 decreased at above 673K. The decrement is not attributed to deactivation of the catalyst, but presumably to very severe catalytic oxidation of hydrocarbons by oxygen. The fact that conversion remained unchanged when the reaction temperature was raised and lowered stepwise between 573K and 773K supports this idea. A similar temperature dependence of the conversion, NO (1,000ppm)+C2H4 (250ppm)+O2 (2.0%), was observed and the most active temperature was 523K in the system. With CO or H2 as a reluctant, the effect of addition of oxygen was completely different from the above. As shown in Fig. 16B, the presence of CO in the NO+He stream markedly promoted NO removal even at 575K. When O2 was added in this reaction system, however, the conversion into N2 was almost completely inhibited in all temperature range examined. In the former reaction system N2 and CO2 were formed in approximately stoichiometric ratio in accordance CO was totally oxidized to CO2 even at 473K in the NO+CO+O2 reaction without the formation of N2. Hydrogen showed similar behavior to CO. The conversion into N2 in NO+O2+C3H6 system depended on the concentrations of propene and oxygen. The increment of concentration of propene increased the activity as shown in Fig. 16A; when the concentration of propene changed to 1,000ppm from 166ppm, the activity increased up to 80% from 27.5%. It is noted that the amount of oxygen introduced was much greater than the amount necessary to completely oxidize 1,000ppm propene. The effect of oxygen on the catalytic activity was quantitatively depicted in Fig. 17. The conversion level of NO rapidly increased with the increase of oxygen concentration. At higher concentration of oxygen, the catalytic activity slightly decreased. However, it should be noted

13 387 Fig. 17 Conversion into N2 as a Function of Oxygen Concentration over Cu-Z-152 that the catalytic activity at 10% of O2 was still 60 times greater than that without oxygen. The order of activities for selective reduction of NO by ethene over copper ion-exchanged zeolites was as follows: Z(33.8%)>M(25.9%)=L(25.3%)> F(22.6%). Cu-Z shows the highest activity as has also been observed for the decomposition of NO. It was shown in Fig. 5 that Cu-Z displayed no activity for the decomposition of NO when the exchange level of copper ion is less than 40%. By contrast, the activity of Cu-Z for selective reduction of NO by ethene in the presence of O2 increased with the increment of the exchange level, reaching maximum at % of exchange level, and then slightly decreased. Above variance in dependency on copper ion exchange level presumably reflects differences in the reaction mechanism. As described in the previous section, the activity of Cu-Z for decomposition of NO completely disappeared in the presence of SO2. The addition of SO2 into the NO+O2+C3H6 system, however, resulted in only slight decrement40). The reaction temperature dependence of catalytic activity in NO+O2+C3H6+SO2 system over Cu-Z was first measured. When SO2 was introduced into the reaction system of NO+O2+C3H6, the extent of conversion into N2 slightly decreased and, after about 20min it turned into a steady state process. No deterioration in the effectiveness of the catalyst was observed even after 220min of continuous service. The results Cu-Z displays catalytic activity even in the presence of SO2 at above 573K. The catalytic activity increased with raise in reaction temperature, reaching a maximum at 773K, and then decreased at higher reaction temperatures. The Fig. 18 Conversions into N2 in NO+O2+C3H6 or NO System with or without SO2 experimental results without SO2 are depicted in that addition of SO2 into the NO+O2+C3H6 system results in slight decrement in the catalytic activity. For example, the conversion of NO into N2 was 85% at 773K in the presence of SO2 while ca. 100% in its absence. In the presence of SO2, C3H6 was quantitatively oxidized to CO and CO2 at below 673K, only to CO2 at above 773K and the carbon balance was confirmed to be ca. 100%. No formation of N2O was observed in the temperature range of K. These results are similar to those observed in the absence of SO2. Further, the catalytic activity at 573K was gradually restored after the supplement of SO2 was stopped. Both results indicate that there is no permanent deactivation of the active sites upon adimission of SO2; the decrement of catalytic activity would be due to change of state of copper ion by the adsorption of SO2. Further studies on reaction mechanism are in progress. The decrement of catalytic activity at higher temperature is not attributed to the deactivation of Cu-Z, since no change in catalytic activity was observed when the reaction temperature was raised and lowered stepwise between 773K and 873K. The decrease is probably due to very severe catalytic oxidation of C3H6 by O2, because C3H6 was completely oxidized to CO2 at above 773K. The conversion into N2 in NO(500ppm)+O2+C3H6 (900ppm)+SO2(200ppm) system depended on the

14 388 be the better catalyst at present if one considers the application of the present system to the exhaust gases of diesel engines. 5. Conclusions Fig. 19 Conversion into N2 as a Function of Oxygen Concentration over Cu-Z-152 concentration of O2. The dependence of conversion on oxygen concentration is shown in Fig. 19, which is similar to that without SO2 (Fig. 17). The catalytic activity was zero at PO2=0%, increased sharply when only 1.0% O2 was added, and then it decreased at higher concentration of O2. The conversion into N2 at 773K, however, was still 42%, even at 12% O2 concentration, while activity for direct decomposition of NO completely or nearly disappeared in the presence of SO2 or 12% O2. The catalytic activity increased by the increment of concentration ratio of C3H6 to NO. When the ratio was changed to 3.0 from 1.0, the conversion into N2 increased to 80% from 60%. The catalytic activity depended not only on the oxygen concentration but also on the concentration ratio of C3H6 to NO. Thus, it was found that selective reduction of NO by C3H6, in the presence of O2 and over Cu-Z, has high resistance to the inhibition by SO2. This fact suggests that Cu-Z is a potential candidate for the catalytic removal of NO in real exhaust gases of vehicles' engines and industrial boilers. After the publication of reports by the authors18),36),37) and Held et al.38), it has been reported that selective reduction of NO by hydrocarbons in an oxidizing atmosphere proceeds not only over Cu-Z but also over protonexchanged zeolites and Al2O341),42). The activities of proton-exchanged zeolites and Al2O3, however, greatly decreased at high GHSV region and their active temperature regions were K and 773K, respectively, which were higher than the temperature of emission gas from diesel engines( k)43). Cu-Z, therefore, would (1) Copper ion-exchanged ZSM-5 zeolite is the most active for the decomposition of NO. The increase of exchange level of copper ion results in the increment of the activity. The catalytic activity is dependent on the zeolite structure and the aluminum content. (2) Cu+ dimer formed by high temperature treatment would be the precursor of active site for the decomposition of NO and the reaction proceeds via NO- nitrosyl intermediates. (3) Selective reduction of NO by the simultaneous presence of oxygen and hydrocarbons has been found to proceed over Cu-Z zeolite at temperatures as low as K. This finding should lead into the development of new catalysis technology. Acknowledgments This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, and Culture of Japan, Nissan Science Foundation, and The Special Grant-in- Aid for Promotion of Education and Science in Hokkaido University Provided by Ministry of Education, Science and Culture. References 1) Crucq, A., Ferennet, A., "Catalysis and Automotive Pollution Control", Elservier (1987), p. 1. 2) Hightower, J. W., Van Leirsberg, D. A., "The Catalytic Chemistry of Nitrogen Oxides", eds. by Klimish, R. L., Larson, J. G., Plenum (1975), p ) Iwamoto, M., Yokoo, S., Sasaki, K., Kagawa, S., J. Chem. Soc., Faraday Trans. 1, 77, 1629 (1981). 4) Harison, B., Wyatt, M., "Catalysis", Royal Society of Chemistry (1982), vol. 5, p ) Iwamoto, M., Furukawa, H., Mine, Y., Uemura, F., Mikuriya, S., Kagawa, S., J. Chem. Soc., Chem. Commun., 1986, ) Iwamoto, M., Furukawa, H., Kagawa, S., Stud. Surf. Sci. Catal., 28, 943 (1986). 7) Iwamoto, M., Yahiro, H., Tanda, K., Stud. Surf. Sci. Catal., 37, 219 (1988). 8) Iwamoto, M., Yahiro, H., Kutsuno, T., Bunyu, S., Kagawa, S., Bull. Chem. Soc. Jpn., 62, 583 (1989). 9) Iwamoto, M., Yahiro, H., Mine, Y., Kagawa, S., Chem. Lett., 1989, ) Shimada, H., Miyama, S., Kuroda, H., Chem. Lett., 1988, ) Uchijima, T., Hyomen, 18, 132 (1987). 12) Teraoka, Y., Fukuda, H., Kagawa, S., Chem. Lett., 1990, 1. 13) Yasuda, H., Mizuno, N., Misono, M., J. Chem. Soc., Chem. Commun., 1990, ) Hamada, H., Kintaichi, Y., Sasaki, M., Ito, T., Chem. Lett., 1990, ) Iwamoto, M., Mizuno, N., Shokubai, 32, 462 (1990). 16) Iwamoto, M., Yahiro, H., Torikai, Y., Yoshioka, T.,