J. Phys. Chem. 1994, 98, 11533-11540
Electron Paramagnetic Resonance Studies of Copper Ion-Exchanged ZSM-5 Sarah C. Larsen, Adam Aylor, Alexis T. Bell, and Jeffrey A. Reimer* Center for Advanced Materials, Lawrence Berkeley Laboratory and Department of Chemical Engineering, University of California at Berkeley, Berkeley, California 94720 Received: April 15, 1994; In Final Form: July 11, 1994@
Electron paramagnetic resonance (EPR) spectroscopy was utilized to probe the oxidation state and coordination environment of copper in ion-exchanged CuZSM-5. EPR spectra of hydrated samples were consistent with octahedral coordination. Square-pyramidal and square-planar sites were identified in pretreated CuZSM-5 samples, and the relative concentration of square-pyramidal sites in these samples was linearly correlated with the copper-exchange level. The extent of autoreduction was monitored by EPR and it was determined that a substantial fraction (-40-60%) of the copper was reduced and that the reduction process was reversible in the presence of water. A mechanism for the autoreduction of copper is proposed that is consistent with the EPR results. Further, the reactivity of the proposed copper species was probed in reducing and oxidizing environments and in the presence of nitric oxide. The increase in EPR signal intensity that was observed after room-temperature NO exposure of pretreated and oxidized CuZSM-5 is attributed to the formation of copper nitrite and nitrate species. High-temperature in situ EPR experiments revealed that on the time scale of the EPR experiment, the paramagnetic copper environment did not change at elevated temperatures in the presence of nitric oxide.
In this paper, the role of copper in the direct decomposition of nitric oxide was examined by using EPR spectroscopy. Copper-exchanged NaZSM-5 samples were exposed in situ to various thermal and chemical pretreatments. The EPR signal was monitored quantitatively to determine the coordination and concentration of paramagnetic copper sites after thermal treatments in helium, oxygen, and carbon monoxide. Hydrated samples were also examined by using ESEEM spectroscopy. These experiments have provided insight into the structural environment of Cu2+ sites in CuZSM-5, as well as into the chemical reactivities of these sites. On the basis of these results, a mechanism for the autoreduction of Cu2+to CUI+is proposed that does not invoke the presence of copper pairs. Finally, EPR signals of pretreated CuZSM-5 samples were recorded periodically during room-temperature and high-temperature exposure to flowing nitric oxide.
Excessively ion-exchanged CuZSM-5 possesses high catalytic activity for the direct decomposition of nitric oxide to nitrogen and oxygen.'S2 The highest level of activity per copper cation is observed when the CdAl ratio in the exchanged zeolite exceeds 0.50.3 This observation has stimulated an intense interest in understanding the valence of the exchanged copper cations and the nature of their interaction with the zeolite framework. Despite the abundance and variety of spectroscopic studies in this area, the role of copper in the direct decomposition of nitric oxide is not well understood. Based on electron paramagnetic resonance (EPR), X-ray absorption near-edge structure (XANES), and photoluminescence measurements, it has been established that when CuZSM-5 is heated to high temperature (> 698 K) in flowing helium or under vacuum, Cu2+(paramagnetic,3d9)is autoreduced to CUI+ (diamagnetic, 3d10).4-8 EPR spectroscopy has also been used 11. Experimental Details to probe the structural environment of paramagnetic Cu2+ sites. EPR spectra assigned to Cu2+in octahedral, square-planar, and A. Sample Preparation. NaZSM-5 was synthesized by a square-pyramidal sites have been observed in copper-exchanged template-free synthesis p r ~ c e d u r e . ~ NaOH ~ J - ~ (2.4 g) and AlZSM-5.9-11 Additional information about the Cu2+environment (OH)3 (1.2 g) were dissolved in 100 mL of deionized H20 and has been obtained by electron spin-echo envelope modulation added slowly with stimng to 42.6 g of silica sol (Ludox 40%). (ESEEM) spectroscopy. This technique provides information Seed crystals of ZSM-5 (0.14 g) were added to the gel to about the proximity of Cu2+ cations to framework aluminum promote crystallization. The gel was placed in Teflon-lined Parr a t o m ~ . ' ~Other J ~ methods such as infrared (E) ~pectroscopy,~~-~~ autoclaves and heated to 453 K for 48 h. The reaction was temperature-programmed desorption (TPD) experiments,2$20 and quenched, and the zeolite crystals were washed with distilled thermogravimetric measurements (TGA)8have provided indirect water and dried ovemight at 353 K. The copper was exchanged evidence conceming the oxidation state of copper in CuZSMfor sodium at room temperature using dilute solutions (0.015. 0.1 M) of copper acetate. The pH of the copper acetate solution Many basic questions conceming the oxidation-reduction was -5.8. The exchanged sample was subsequently dried at behavior of CuZSM-5 in the absence of nitric oxide remain to 373 K for 12 h. Samples were characterized by elemental be answered. In particular, the mechanism for autoreduction analysis, X-ray diffraction (XRD), and scanning electron and the partitioning of copper into Cu2+ and CUI+are still not microscopy (SEM). X-ray diffraction patterns were obtained understood. While both Cu2+and Cul+ sites exist in the catalyst on a Siemens diffractometer. Diffraction pattems from the in its active state, it has not been determined whether both sites template-free ZSM-5 agreed with the pattems obtained from are necessary for nitric oxide decomposition or even whether the seed crystals, as well as with Z S M J synthesized with a the catalytic decomposition involves a redox process.21.22 template. Scanning electron micrographs revealed crystal agglomerates of -10 pm. Inductively coupled plasma (ICP) Abstract published in Advance ACS Abstracts, September 15, 1994. @
0 1994 American Chemical Society
Larsen et al.
11534 J. Phys. Chem., Vol. 98, No. 44, 1994 1.5 I
Tius work (Cu+Na)/Al
0 Li and Hall(ref. 8) (Cu+Na)/Al
YOCopper Exchange (2Cu/A1'100)
Figure 1. Data points representing the (Cu
Hydrated, 77K 1
Na)/Al ratios obtained from Table 1 (filled triangles) and from samples studied previously by Li and Halls (open squares). The solid line ([(Cu + Na)/Al] = 1) corresponds to stoichiometric exchange of [Cu2+OH-]+for Naf.
I/ \I 2000
TABLE 1: Elemental Analyses Data for Samples Used in This Study sample Cu/Al wt%Cu SUA1 NdA1 2.76 18 0.10 CUZSM-5-110 0.55 18 0.22 0.46 2.32 CUZSM-5-92 CUZSM-5-68 0.34 1.43 18 0.72 0.55 18 0.84 CUZSM-5-26 0.13 emission spectroscopy was used to determine the elemental composition of the samples. The results of elemental analyses are presented in Table 1. The elemental analyses reveal some ambiguities in the ionexchange process. Before ion exchange, sodium cations charge compensate the negative charge associated with the presence of aluminum atoms in the ZSM-5 lattice. For stoichiometric exchange of copper for sodium, the Cu/Al ratio should be 0.5 assuming that one Cu2' charge compensates two aluminum sites (or exchanges for two sodium ions). However, typical exchange levels exceed 100% (2Cu/Al x loo%), suggesting that Cu2+ hydrolyzes and exchanges into NaZSM-5 as [ C U ~ + O H - ] +If. ~ ~ this interpretation is correct, the sum of CdA1 and Na/Al from Table 1 should be unity, corresponding to stoichiometric exchange of copper for sodium. Figure 1 shows that this hypothesis holds for exchange levels below -70% but is not the case for 92-110% exchange levels, implying either that some copper is exchanged into NaZSM-5 as bare Cu2+ or that 'H Bronsted acid sites are also formed in the samples with high copper loadings. Data from the work of Li and Hall8 are also plotted in Figure 1, and the same trend in the [(Cu Na)/Al] ratio is observed. In a combined TPD/TGA study, Pari110 et al. have reported that Bronsted acid sites are easily incorporated into copper-exchanged NaZSM-5 during ion exchange.26 Hence, we assume that copper is exchanged into NaZSM-5 as [Cu2+OH-]+ with the rest of the charge compensation provided by protons. Fresh, air-exposed samples are referred to as hydrated. The standard sample pretreatment consisted of heating the sample to 700-773 K for 1-3 h in flowing helium (or under vacuum). The pretreatment of CuZSM-5 samples resulted in dehydration and autoreduction (Cu2+ reduced to CUI+). Oxidized samples were prepared by exposing a pretreated sample to oxygen at 700-773 K for approximately 1 h followed by cooling in oxygen and purging in helium at room temperature. Similarly, reduced samples were prepared by exposing a pretreated sample to 4% COhelium at 700-773 K for approximately 1 h, followed by cooling in 4% COhelium and purging in helium at room temperature. B. Experimental Apparatus. EPR measurements were obtained on a Varian 4500 EPR spectrometer interfaced to a
Figure 2. EPR spectra of CuZSM-5-110 obtained at YEPR = 8.9 GHz in the hydrated state (A and B) and after standard pretreatment (C). personal computer for data acquisition and magnetic field ~ontrol.~'In situ flow and static adsorption experiments were performed. Static adsorptions were done on a vacuum rack and then samples were sealed in EPR tubes for spectroscopic measurements. A liquid nitrogen immersion Dewar was used for low-temperature experiments on static samples. A quartz flow cell was fabricated based on the design of Mesaros and Dybowski.28 The flow of reactant gases (He, 4% NO/He, 4% CO/He, 0 2 ) through the cell was controlled by mass flow controllers. The flow cell was placed in a quartz Dewar and was heated by flowing nitrogen gas over a nichrome heating element and through the quartz Dewar. Standard solutions of aqueous copper sulfate were used to calibrate the concentration of paramagnetic species in the zeolite samples. The magnetic field and microwave frequency were calibrated with a commercial gaussmeter and frequency meter, respectively. ESEEM experiments were performed on a spectrometer described by Britt et al.29 Two- and three-pulse modulation patterns were obtained at YEPR = 9.55 GHz and T = 4.2 K. Typical modulation pattems consisted of 256 data points acquired at 8-ns intervals. Spectral analysis was carried out as described p r e v i o ~ s l y . ~ A~Lorentzian decay was fit to the ESEEM pattem and subtracted from it. The data were Fourier transformed after tapering with an extended cosine-bell function.
111. Results and Analysis Room-temperature and low-temperature EPR spectra of hydrated and pretreated CuZSM-5-110 samples are presented in Figure 2. The room-temperature spectrum (Figure 2A) of hydrated CuZSM-5- 110 is broad and structureless, in contrast to the spectrum recorded at 77 K (Figure 2B) in which the characteristic copper hyperfine structure ( I = 3/2) is resolved. When the sample was pretreated by heating to 723 K under vacuum, the room-temperature EPR spectrum (Figure 2C) exhibited resolved copper hyperfine structure. The coordination and concentration of the paramagnetic complex also changed during this treatment as evidenced by the high field shift of the copper hyperfine features and the decrease in signal intensity, respectively. Magnifications of the low-field hyperfine features for a series of pretreated CuZSM-5 samples with different copper loadings are presented in Figure 3. Two different sites, labeled A and B, are resolved in the EPR spectra.
J. Phys. Chem., Vol. 98, No. 44, 1994 11535
Studies of Copper Ion-Exchanged ZSM-5
Figure 3. Magnification of the low-field hyperfine features of the EPR
spectra of pretreated CuZSM-5-26, CuZSM-5-68, and CuZSM-5-110. Spectra were recorded at VEPR = 8.9 GHz and T = 77 K. A and B are assigned as square-planar and square-pyramidal sites, respectively. Detailed structural information about paramagnetic Cu2+sites can be determined from the EPR spectra. The EPR spectrum of hydrated CuZSM-5 in Figure 2A is featureless, but the lowtemperature spectrum in Figure 2B exhibits the characteristic structure of Cu2+complexes that results from hyperfine coupling between the 3d unpaired electron and the copper ( I = 3/2) nuclear spin. This coupling causes a 4-fold splitting of the EPR line. For an orientationally disordered solid assuming axial symmetry, g anisotropy produces a powder pattern in which the sharp features are referred to as parallel and perpendicular edges. Since the copper system has both g and hyperfine anisotropy, the overall spectrum is composed of four overlapping powder patterns in which the resolved low-field features are the parallel edges of the four powder patterns and the perpendicular features of the spectrum are not resolved due to broadening from unresolved hyperfine couplings and site inhomogeneity. The broad EPR spectrum observed in Figure 2A is interpreted as a mobile copper complex; the appearance of the parallel edges in the EPR spectrum with decreasing temperature (Figure 2B) signifies that the mobile copper species becomes immobile at low temperature. The low-temperature EPR spectra of hydrated CuZSM-5 samples were fit by using second-order perturbation equations31 and a simplex least-squares fitting routine.32 The parallel components of g and A are well-determined in this fitting procedure; however, the perpendicular components of g and A are rather insensitive to the fitting procedure because of the lack of resolution in the perpendicular region of the EPR spectra. The parameters derived from a least-squares fit of the spectra of all the hydrated samples are presented in Table 2. These values agree well with literature values for octahedral copper complexes observed in zeolites10-12 where the surrounding ligands are thought to be water and/or hydroxyl groups. Due to the 7.4-A size of the hydrated copper complex12 and the presence of motional effects in the EPR spectrum at room temperature, the octahedral complex is believed to be located in a channel intersection where it would have the most free space for motion. Theoretical calculations on HZSMJ have determined that the Bronsted acid sites and hence the aluminum atoms are also located in the channel intersection^.^^ An ESEEM modulation pattern and the resulting Fourier transform spectrum are shown in Figure 4 for hydrated CuZSM5-110. Modulations at -3.8 and -14 MHz were observed that correspond to the Larmor frequencies of 27Aland 'H, respectively. The observation of aluminum modulation further supports the idea that the hydrated copper complex is located at the zeolite channel intersection since modulations are only observed for nuclei within -5 A of the copper The shift of the low-field hyperfine features of the EPR spectrum after the sample is pretreated (Figure 2C) indicates a
change in copper coordination. Two different copper sites are resolved in pretreated CuZSM-5, as shown in Figure 3. (The spectrum also changes in intensity; this will be discussed in the next section.) A suitable fit to this spectrum requires a modification of the line-shape analysis described above to include parameters for the two resolved copper sites and their relative concentrations. The parameters derived from these twocomponent fits are given in Table 3 and agree very well with parameters reported previously for square-planar (A) and squarepyramidal coordinations (B); the agreement is best for the parallel components of g and A since the perpendicular portions of the EPR spectra are unresolved.lO,ll For other copperexchanged zeolites, such as mordenite, the copper EPR signals were assigned to specific, well-defined sites in the zeolite structure.ll Since the sites in ZSM-5 are not as well-defined as in mordenite, more definitive structural information is difficult to obtain." The fitting procedure was followed for the series of copper ZSM-5 samples with different loadings, and a correlation was established between copper loading and the relative concentration of square pyramidal coordination (Figure 5). The sensitivity of the copper cations in pretreated (autoreduced) CuZSM-5 to oxidation by oxygen and reduction by carbon monoxide at elevated temperatures is reported in Figure 6. Exposure of pretreated CuZSM-5 to oxygen at 683 K, followed by cooling in oxygen and a room-temperature helium purge, produced virtually no change in the EPR signal intensity (Figure 6); however, the sample changed color from purple to yellow -green, indicating a probable change in oxidation state of the copper. After exposure of pretreated CuZSM-5- 110 to 4% C o m e for 2 h at 683 K, followed by cooling in 4% CO/ He and purging in helium at room temperature, the EPR signal intensity decreased to approximately 25% of the intensity observed for the hydrated sample. Figure 7 shows that after reduction, the signal corresponding to the square-planar site is absent, suggesting that this site is reduced preferentially (Figure 7 (inset)). The preferential reduction of the square-planar site under similar conditions has been reported in the literature for CUHZSM-~.'~,~~ In order to determine if the paramagnetic copper detected by EPR accounts for all of the copper in our CuZSM-5 samples, absolute concentrations of copper were determined by EPR and compared with the elemental analyses. The data points in Figure 8 were obtained by double integration of the EPR spectra recorded at 77 K followed by calibration against aqueous copper sulfate solutions (0.1, 0.075, 0.05, and 0.025 M). The data points in Figure 8 represent the absolute copper concentrations determined by the EPR measurements on hydrated CuZSM-5 (top set of data points) and pretreated samples (bottom set of data points). The solid line (Figure 8, upper) corresponds to a least-squares fit to the copper concentration of the samples as measured by elemental analyses. The agreement between the Cu2+ concentration of the hydrated samples calibrated by EPR and the copper concentration determined by elemental analysis confirms that all of the copper in hydrated CuZSM-5 is present as Cu2+ and detected by EPR experiments. In pretreated samples, approximately 40-6070 of the copper is detected by EPR; the remaining fraction of copper is presumably reduced to CUI+. By using the in situ flow apparatus, the autoreduction process was monitored as a function of temperature. The CuZSM-5 sample was heated to the desired temperature in flowing helium for 1 h; the sample was then cooled, and the EPR spectra were recorded at room temperature. The double integration of the EPR signal intensity is plotted in Figure 9 and displays a
11536 J. Phys. Chem., Vol. 98, No. 44, 1994
Larsen et al.
TABLE 2: Fitted EPR Parameters for Hydrated CuZSM-5 Sample CuZSM5-26
2.35 2.06 428 -0 6.0 x 10-5
2.35 2.07 420
2.35 2.07 425 4 7.0 x 10-5
gl AH,(MHz) Ai, (MHz)
-0 7.5 x 10-5
CuZSM5-110 2.35 2.07 426 5 6.6 x
2.37 2.09 370
2.379 2.076 428
Estimated errors for parameters are g (kO.01) and A (&5 MHz). 4-
k 0.45 t c
Figure 5. Fraction of the EPR signal of pretreated CuZSM-5 samples assigned to square-pyramidal coordination as determined by detailed simulations (Table 3) of the EPR spectra. I
Figure 4. Three-pulse ESEEM modulation pattem (top) and ESEEM spectrum (bottom) of CuZSM-5-110 obtained at YEPR = 9.552 GHz and T = 4.2 K. The time between the first two pulses was 160 ns. The
52 11 d
peaks at -3.6 and 13.8 MHz are assigned to aluminum and proton interactions, respectively.
2.24 2.03 AlP,MHz 499 AL,MHz 27 gilB 2.27 gL* 2.03 Ai?,MHz 375 AlB,MHz 25 fraction A 0.65 x2 6.0 x 10-5
2.24 2.04 486 35 2.29 2.04 392 23 0.56 2.6 x 10-5
2.26 2.04 49 1 22 2.29 2.04 382 23 0.55 3.7 x 10-5
2.26 2.04 SO8 20 2.30 2.04 378 43 0.43 2.2 x 10-5
0,, 683 K
Figure 6. Relative intensity of the EPR spectra of CuZSM-5-110 after
CuZSM5-26 CuZSM5-68 CuZSMS-92 CuZSM5-110 ref 10 gi*
TABLE 3: Fitted EPR Parameters for Pretreated CuZSM-5 Sample& gllA
2.27 2.04s 476 81 2.32 2.06 392 50
various pretreatments. The intensities were obtained by double integration of the EPR spectra and were referenced to the intensity of the hydrated sample. The sample was heated to 683 K in flowing helium for 1 h and exposed to oxygen at 683 K for 1 h. The sample was purged with helium at 683 K and exposed to 4% Come for 1 h at 683 K. Between each of the treatments, the sample was cooled and EPR spectra were obtained at YEPR = 9.28 GHz and T = 298 K.
a The superscripts A and B refer to the square-planar and squarepyramidal sites, respectively. Estimated errors for parameters are g (60.01) and A ( f 5 MHz).
minimum around 473 K, slowly increasing to -60% of the initial intensity. The original signal intensity was recovered by bubbling helium through water at room temperature, suggesting that the autoreduction process was completely reversible with exposure to water. In situ EPR spectra were recorded at 10-min intervals after exposure of pretreated CuZSM-5-92 to a 2% NO/He mixture at room temperature (Figure 10). The relative EPR intensity initially decreased and subsequently increased with time and exceeded the original pretreated signal intensity, indicating that a portion of the autoreduced copper had been reoxidized to paramagnetic Cu2+ (Figure 11). The spectral features also changed as the experiment progressed, suggesting that a new
li v 2500
Figure 7. EPR spectra of CuZSM-5-110 after the standard pretreatment (solid) and after exposure to 4% CO/He for 2 h at T = 673 K (dashed). Magnification of the low-field hyperfine features is presented in the inset. Spectra were recorded at YEPR = 9.28 GHz and T = 298 K.
species had been formed. Experiments were also performed on CuZSM-5 samples that had been oxidized at -703 K, cooled in oxygen and purged with helium at room temperature, and then exposed to nitric oxide at room temperature. Figure 12 shows EPR spectra for a preoxidized sample after room-
J. Phys. Chem., Vol. 98,No. 44, I994 11537
Studies of Copper Ion-Exchanged ZSM-5 30
60 80 100 Exchange Level [(2Cu)/A1'100%]
Figure 8. Calibrated spin (CuZf)concentration vs exchange level from elemental analysis for a series of CuZSM-5 samples. EPR spectra of CuZSM-5 were integrated twice and calibrated using aqueous copper sulfate solutions. The top set of data points (circles) corresponds to hydrated samples, and the bottom set (triangles) corresponds to pretreated samples. The solid line represents the copper concentration of the samples as determined by elemental analysis. EPR spectra were recorded at YEPR = 8.9 GHz and T = 77 K. Each data point represents the average of -3-5 experimental measurements.
Pretreated N0,l min. N0,20 min. He purge
Figure 11. Relative intensities of selected EPR spectra from Figures 10 (solid bars) and 12 (hatched bars). EPR intensities were obtained by double integration of the EPR spectra. EPR spectra were recorded at VEPR = 9.28 GHz and T = 298 K.
473 573 673 Temperature, K
Figure 9. Relative intensity of EPR spectra of CuZSM-5-110 obtained
3000 3500 Field (Gauss)
after pretreatment in flowing helium at progressively higher tempera-
Figure 12. In situ EPR spectra of CuZSM-5-110 pretreated in oxygen
tures (as labeled) and then rehydration with helium saturated with water
at 703 K (bottom), after exposure to 2% NOMe for various times (as labeled), and after a subsequent helium purge (top). Spectra were recorded at Y ~ =R 9.28 GHz and T = 298 K.
vapor. EPR intensities were obtained by double integration of the EPR spectra and were referenced to the initial hydrated state. EPR spectra were recorded at Y E ~ R= 9.28 GHz and T = 298 K.
3000 3500 Field (Gauss)
Figure 10. In situ EPR spectra of pretreated CuZSM-5-92 (bottom) after exposure to 2% NOMe for various times (as labeled) and after a subsequent helium purge (top). Spectra were recorded at YEPR = 9.28 GHz and T = 298 K.
temperature exposure to 4% NO/He. The signal intensity (Figure 11) initially decreased and then increased and surpassed the original pretreated signal intensity, again indicating a reoxidation of copper. The resulting spectrum is very similar to the pretreated sample after exposure to nitric oxide.
3000 3500 Field (Gauss)
Figure 13. In situ high-temperature EPR spectra of CuZSM-5-110 pretreated in helium at 703 K (bottom), after exposure to 4% NOMe for various times (as labeled), and after a subsequent helium purge (top). Spectra were recorded at YEPR = 9.27 GHz and T = 703 K.
EPR spectra were also obtained at elevated temperatures representative of those used for the decomposition of NO over CuZSM-5. A pretreated CuZSM-5-110 sample was exposed to 4% NO/He at -703 K, and the EPR spectra(Figure 13) were recorded at this temperature. The EPR intensity and spectral
Larsen et al.
11538 J. Phys. Chem., Vol. 98, No. 44, 1994 features did not change with exposure to NO at elevated temperatures over a time period of 30 min.
IV. Discussion Autoreduction. EPR spectroscopy has provided quantitative information about the autoreduction process that occurs when CuZSM-5 is pretreated by heating to 673-773 K in helium or under vacuum. Initially, all of the copper exchanged into CuZSM-5 as measured by elemental analysis can be accounted for by EPR measurements (Figure 8). Conversely, the copper concentration measured in the EPR spectra of pretreated CuZSM-5 samples only accounts for -40-60% of the total copper present in the samples, suggesting that cupric ions are autoreduced to cuprous ions. While CUI+ions are not observable by EPR, this cation has been observed by photoluminescence," by XANES,6and indirectly by spectroscopy. It is also notable that our estimate of the extent of Cu2+ undergoing autoreduction lies well within the range of values reported in the literature (25-75%).4-6-8J4 Based on the assumption that copper is substituted into NaZSM-5 as [Cu2+0H-]+ (as discussed in the Experimental Section), the following mechanism for autoreduction of Cu2+ to CUI+ is proposed:
possibly, the dehydrated Cu2+ cations could coordinate with lattice oxygen to form square-planar and square-pyramidal sites that are observed in the EPR spectra of pretreated samples. Experiments are in progress to verify this hypothesis. When a pretreated CuZSM-5 sample was exposed to oxygen at 683 K, no increase in copper EPR intensity was observed, but a distinct color change from purple to yellow-green was noted. One explanation for these observations is that CUI+sites are oxidized to Cu2+, resulting in the color change and the formation of the superoxide anion, 0 2 - . Since CUI+ and Cu2+02- are both EPR-silent, no increase in EPR signal intensity would be expected, in agreement with our data. This process is illustrated schematically by reaction 3. cu'+
+ 0, - cu2+o,-
[Cu2+0H-]+ 5 Cu" [Cu2+0H-]+ 2[Cu2+OH-]+
+ OH 9Cu2+O- + H,O 5 Cu"
+ Cu2+O- + H,O
In the first step, C U +is~ reduced to CUI+with the release of a hydroxyl radical. The hydroxyl radical reacts with another [Cu2+OH-]+ moiety to produce a Cu2+O- species and water. Since 0- and Cu2+are both paramagnetic, the resulting Cu2+Ospecies will not be observed by EPR. The proposed mechanism for autoreduction is consistent with the observation that the full Cu2+ EPR signal intensity can be regained after exposure of a pretreated sample of CuZSM-5 to H20. Previously proposed mechanisms invoke the presence of copper pairs and the spontaneous desorption of ~ x y g e n . ~ ~ ~ ~ ~ ~ * 2[Cu2+OH-]+
5 [CUOCU]~+ + H,O
[ c u o c u ] 2 +5 2cu1+
The formation of a significant concentration of [CuOCu12+ moities should be detected by the appearance of a "half-field' EPR signal, such as that observed for CUY.~'No evidence for such a signal was observed in the present studies. Similarly, XANES and extended X-ray absorption fine structure (EXAFS) studies show no evidence for the formation of copper dimers but clearly show the formation of Cul+ cations following the pretreatment of hydrated CuHZSM-5 in helium at 773 K.5,6 The copper remaining in the Cu2+ state after pretreatment and observed by EPR is difficult to explain. It is unclear why only part of the copper exchanged into NaZSM-5 as [Cu2+0H-]+ is autoreduced according to the mechanism presented above. Conceivably, this might be due to differences in the nature of the sites occupied by [Cu2+0H-]+. For example [Cu2+0H-]+ structures located near isolated [AlO-Si] charge-exchange centers would be expected to undergo autoreduction according to eq 1. On the other hand, if [Cu2+OH-]+ and a proton charge compensate [AlO-Si0-All charge-exchange centers, then dehydration could occur without reduction of Cuz+ to CUI+. Quite
A Cu2+ superoxide complex has been synthesized previously in solution by reaction of an ethylene-Cul+ complex with molecular oxygen, thereby establishing a precedence for reaction 3.38 It is important to compare and contrast the results of our observations of the effects of high-temperature oxidation on the EPR spectra of CuZSM-5 (see Figure 6) with those of two recent Li and Hall observed a 25% decrease in signal intensity when they exposed a CuZSM-5 sample to 500 Torr of oxygen at 773 K and subsequently evacuated the sample at 773 K for 2 h. Their interpretation is that EPR-observable Cu2+ is reduced to diamagnetic Cul+ as oxygen desorbs during hightemperature evacuation.* We have repeated their experiment, and we find no decrease in EPR signal intensity after hightemperature evacuation of an oxidized sample. Very recently, Kucherov et al. have reported EPR results for a pretreated CuZSM-5 sample that had been exposed in situ to oxygen at 773 K, cooled in oxygen, and purged with h e l i ~ m .The ~ signal intensity was the same for the pretreated and oxidized samples, in agreement with our data presented in Figure 6. Their interpretation of this result differs from ours; they infer that if autoreduction produces CUI+during the initial pretreatment, then the CUI+ must be oxidized to Cu2+ during high-temperature treatment in oxygen, resulting in an increase in EPR signal intensity. Since they do not observe an increase in EPR signal intensity, they conclude that all of the copper in a pretreated sample is present as Cu2+. While we similarly observe no increase in EPR signal intensity after high-temperature oxidation, we emphasize that the intensity of the Cu2+ signal following pretreatment and calcination in oxygen does not represent a measure of the total amount of copper in the zeolite. Figures 6 and 8 clearly show that only 40-60% of the total copper is present as Cu2+ after pretreatment. We further suggest that the Cul+ is oxidized to Cu2+ as shown in reaction 3, and the resulting CuZf02- is EPR-silent. Therefore, the arguments against autoreduction in CuZSM-5 presented by Kucherov et al. are not compelling. Exposure to Nitric Oxide. The changes in the EPR signal intensity shown in Figure 10 can be interpreted in the following way. The initial decrease in signal intensity upon exposure of pretreated CuZSM-5 to nitric oxide can be attributed to the formation of an EPR-silent Cu2+n i t r ~ s y l . ' ~ -The ' ~ increase in signal intensity with extended exposure to nitric oxide is assumed to result from the reaction of NO with the Cu2+Ospecies produced by autoreduction. As can be seen from reaction 4,the reaction product, Cu2+N02-, is EPR-observable, whereas the reactant, Cu2+O- is not. cu2+o-
Studies of Copper Ion-Exchanged Z S M J
J. Phys. Chem., Vol. 98, No. 44, 1994 11539
The formation of the Cu2+N02- species at room temperature has been reported recently in several IR studies.1939 During the exposure of the sample to NO, Cul+ centers can undergo oxidation via the following process: CUI+ 2 N 0 C d 2 0 N20.35,36 The additional Cu2+O- species formed in this fashion would be free to produce Cu2+N02- via reaction 4. Consistent with this interpretation, IR studies of the roomtemperature interactions of nitric oxide with CuZSM-5 show a progressive loss of band intensity at 1734, 1808, and 1825 cm-' (assigned to Cul+NO and Cu1+(N0)2 species) and an increase of band intensity in the region of 1320-1630 cm-' (assigned to copper nitrate and nitrite species) with increasing nitric oxide e x p o s ~ r e . ~Finally, ~ , ~ ~ ,the ~ ~increase in EPR signal intensity upon helium purge of the nitric oxide can be attributed to the release of NO molecules from Cu2+ nitrosyls. This process is also well evidenced in IR studies.39 The overall increase in signal intensity (Figure 11) of 40% relative to a pretreated sample suggests that after room-temperature exposure to nitric oxide, all of the copper exchanged into CuZSM-5 is observed by EPR; the EPR-silent Cu2+O- and Cul+ sites having been converted to EPR-observable copper nitrite species. The changes in EPR intensity upon exposure of an oxidized sample of CuZSM-5 to nitric oxide are similar to those observed for a pretreated sample (Figure 11). Reaction of these species with nitric oxide can result in Cu2+N03- species in the following way:
Thus, both Cu2+N02- and Cu2+N03- would be expected to form on oxidized CuZSM-5; this is consistent with several IR ~tudies.'~-'~ High-temperature EPR spectra (Figure 13) taken before and after exposure of pretreated CuZSM-5 to nitric oxide show virtually no change in signal intensity or spectral shape. This suggests that a redox process between EPR-observable Cu2+ sites and EPR-silent Cu2+O- and Cul+ sites is not responsible for high-temperature nitric oxide decomposition over the time scale probed by the EPR experiment and that the Cu2+O- and CUI+ sites are responsible for the catalytic activity of CuZSM5. The Cu2+O- and Cul+ sites have been observed indirectly in this work as copper nitrite and nitrate species in EPR experiments designed to probe their chemical reactivity. The idea that CUI' sites are responsible for the catalytic activity is supported by the XANES study of Liu and Robota in which the NO decomposition was correlated with Cul+ concentration.6 Additionally, the importance of Cu2+O- species as active sites for NO decomposition has been proposed by Spoto et a1.35s36
V. Conclusions EPR studies of hydrated and pretreated samples of CuZSM-5 were utilized to elucidate the coordination environment of CuZSM-5. EPR signals from hydrated samples were assigned to octahedral coordination. EPR spectra of pretreated samples were consistent with square-planar and square-pyramidal sites. All of the copper in CuZSM-5 detected by elemental analysis could be accounted for in the EPR experiments, indicating that all of the copper in hydrated CuZSM-5 is present as paramagnetic Cu2+. A 40-60% decrease in EPR signal intensity is observed when CuZSM-5 is pretreated at 673-773 K under vacuum or in flowing helium. The EPR signal intensity can be recovered by exposing the sample in situ to water or nitric oxide at room temperature, but no increase in EPR signal intensity is observed when CuZSM-5 is exposed to oxygen at
high temperature. A mechanism for the autoreduction of Cu2+ in CuZSM-5 into Cul+ and Cu2+O- was proposed that is consistent with the EPR data and accounts for the recovery of the EPR signal intensity upon exposure of pretreated CuZSM-5 to water vapor. The room-temperature reactivity of copper cations in pretreated CuZSM-5 with nitric oxide was probed with EPR, and the formation of Cu2+N02- species was proposed based on the results. High-temperature in situ EPR experiments on CUZSM-5 revealed little change in signal intensity or spectral features in the presence of nitric oxide. This suggests that a redox process between EPR-observable Cu2+ sites and EPR-silent Cu2+O- and Cul+ sites is not responsible for high-temperature nitric oxide decomposition over the time scale probed by the EPR experiment and that the Cu2+O- and Cul+ sites are responsible for the catalytic activity of CuZSM-5. Acknowledgment. S. L. acknowledges an appointment to the Distinguished Postdoctoral Research Program sponsored by the U. S. Department of Energy, Office of Science Education and Technical Information, administered by the Oak Ridge Institute for Science and Education. Drs. Olaf Burghaus and Me1 Klein are acknowledged for providing ESEEM measurements. We also thank Dr. Mordecai Shelef for providing a preprint of ref 9 to us prior to its publication and Air Products and Chemicals Co. for a gift in support of S. L. This work is supported by the Gas Research Institute under Contract 5093260-2492. References and Notes (1) Iwamoto, M.; Furukawa, H.; Kagawa, S. In New Developments in Zeolite Science and Technology; Murakami, Y., Ed.; Elsevier: New York, 1986; p 943. (2) Iwamoto, M.; Yahiro, H.; Mine, Y.; Kagawa, S. Chem. Lett. 1989, 213. (3) Iwamoto, M.; Hamada, H. Cutul.Toduy 1991, 10, 57. (4) Centi, G.; Perathoner, S.; Shioya, Y.; Anpo, M. Res. Chem. Interm. 1992, 17, 125. (5) Hamada, H.; Matsubayashi, N.; Shimada, H.; Kintaichi, Y.; Ito, T.; Nishijima, A. Card. Lett. 1990, 5, 189. (6) Liu, D.; Robota, H. J. Cutul. Lett. 1993, 21, 291. (7) Iwamoto, M.; Yahiro, H.; Tanda, K.; Mizuno, N.; Mine, Y.; Kagawa, S. J . Phys. Chem. 1991, 95, 3727. (8) Li, Y.; Hall, W. K. J. Cutal. 1991, 129, 202. (9) Kucherov, A. V.; Gerlock, J. L.; Jen, H. W.; Shelef, M. J. Phys. Chem. 1994, 98, 4892. (10) Kucherov, A. V.; Kucherova, T. N.; Slinkin, A. A. Catal. Lett. 1991, 10, 289. Kucherov, A. V.; Slinkin, A. A. J. Phys. Chem. 1989, 93, 864. (11) Schoonheydt, R. A. Curd. Rev.-Sci. Eng. 1993, 35, 129. (12) Anderson, M.; Kevan, L. J. Phys. Chem. 1987, 91, 4174. (13) Sass, C.; Kevan, L. J. Phys. Chem. 1989, 93, 7856. (14) Sarkany, J.; d'Itri, J.; Sachtler, W. M. H. Cutul. Lett. 1992, 16, 241. (15) Sarkany, J.; Sachtler, W. M. H. Zeolites 1994, 14, 7. (16) Giamello, E.; Murphy, D.; Magnacca, G.; Morterra, C.; Shioya, Y.; Nomura, T.; Anpo, M. J. Cutal. 1992, 136, 510. (17) Iwamoto, M.; Yahiro, H.; Mizuno, N.; Zhang, W.; Mine, Y.; Furukawa, H.; Kagawa, S. J. Phys. Chem. 1991, 96, 9360. (18) Valyon, J.; Hall, W. K. J. Phys. Chem. 1993, 97, 7054. (19) Valyon, J.; Hall, W. K. J . Phys. Chem. 1993, 97, 1204. (20) Li, Y.; Armor, J. N. Appl. Catul. 1991, 76, L1. (21) Shelef, M. Card. Lett. 1992, 15, 305. (22) Hall, W. K.; Valyon, J. Catal. Lett. 1992, 15, 311.
11540 J. Phys. Chem., Vol. 98, No. 44, 1994
Larsen et al.
(23) Shirakar, V. P.; Clearfield, A. Zeolites 1989, 9, 363. (24) Narita. E.: Sato.. K.:. Yatabe.. N.:. Okabe. T. Ind. E m . Chem. Prod. Rei. Dev. 1985, 24, 507. (25) Valyon, J.; Hall, W. K. Catal. Lett. 1993, 19, 109. (26) Parrillo, D. J.; Dolenec, D.; Gorte, R. J.; McCabe, R. W. J. Catal.
1993, 142, 708.
(27) Koretsky, M. Ph.D. Thesis, Department of Chemical Engineering, University of California at Berkeley; 1991. (28) Mesaros, D. V.: Dvbowski, C. And. Swctrosc. 1987,4, 610. (29) Britt, R. D.; Zimm&"nn, J.; Sa&, K.:Klein, M. P. J. Am. Chem. Soc. 1989, 111, 3522. (30) Larsen, S. C.; Singel, D. J. J. Phys. Chem. 1992, 96, 9007. (31) Chasteen, N. D. In Biological Magnetic Resonance; Berliner, L., Reuben, J., Eds.; Plenum: New York, 1981; Vol. 3, p 3.
(32) Press, W. H.; Flannery, B. P.; Teukolsky, S.A.; Vetterling, W. T. Numerical Recipes in C; Cambridge University Press: New York, 1988. (33) Cook, S. J.; Chakraborty, A. K.; Bell, A. T.; Theodorou, D. J. Phys. Chem. 1993, 97, 6679. (34) Kucherov, A. V.; Slinkin, A. A.; Kondrat'ev, D. A,; Bondarenko, T. N.; Rubinstein, A. M.; Minachev, K. M. Zeolites 1985, 5, 320. (35) Spoto, G.; Bordiga, S.; Scarano, D.; Zecchina, A. Catal. Lett. 1992, 13, 39. (36) Spoto, G . ; Zecchina, A,; Bordiga, S.; Ricchiardi, G.; Martra, G. Appl. Catal. B: Environ. 1994, 3, 151. (37) Chao, C.; Lunsford, J. H. J. Chem. Phys. 1972, 57, 2890. (38) Thompson, J. S. In Biological & Inorganic Copper Chemistry; Karlin, K. D., Zubieta, J., Eds.; Adenine Press: New York, 1985; p 1. (39) Aylor, A.; Larsen, S.; Reimer, J.; Bell, A. Manuscript in preparation.