Vitaly L.
Sushkevich
*a,
Olga V.
Safonova
b,
Dennis
Palagin
a,
Mark A.
Newton
c and
Jeroen A.
van Bokhoven
*ac
aLaboratory for Catalysis and Sustainable Chemistry, Paul Scherrer Institut, 5232 Villigen PSI, Switzerland. E-mail: vitaly.sushkevich@psi.ch; jeroen.vanbokhoven@chem.ethz.ch; Tel: +41563103518
bLaboratory for Operando Spectroscopy, Paul Scherrer Institut, 5232 Villigen PSI, Switzerland
cInstitute for Chemistry and Bioengineering, ETH Zurich, Vladimir-Prelog-Weg 1, 8093 Zurich, Switzerland
First published on 1st May 2020
Copper-exchanged zeolites are a class of redox-active materials that find application in the selective catalytic reduction of exhaust gases of diesel vehicles and, more recently, the selective oxidation of methane to methanol. However, the structure of the active copper-oxo species present in zeolites under oxidative environments is still a subject of debate. Herein, we make a comprehensive study of copper species in copper-exchanged zeolites with MOR, MFI, BEA, and FAU frameworks and for different Si/Al ratios and copper loadings using X-ray absorption spectroscopy. Only obtaining high quality EXAFS data, collected at large k-values and measured under cryogenic conditions, in combination with wavelet transform analysis enables the discrimination between the copper-oxo species having different structures. The zeolite topology strongly affects the copper speciation, ranging from monomeric copper species to copper-oxo clusters, hosted in zeolites of different topologies. In contrast, the variation of the Si/Al ratio or copper loading in mordenite does not lead to significant differences in XAS spectra, suggesting that a change, if any, in the structure of copper species in these materials is not distinguishable by EXAFS.
The high activity of copper-containing zeolites in oxidation and reduction reactions is typically associated with the redox transformations of copper species, hosted in the zeolite's pore system. In general, the CuII/CuI redox couple is involved in the reaction; the copper acts as a donor or acceptor of electrons and enables the conversion of the feed to the desired products.16–24 For the SCR of NOx, a mixture of CuII and CuI active species of varied structures is observed, with the relative fractions governed by the reaction conditions.16–21 Alternatively, for methane oxidation to methanol, the active species are represented exclusively by CuII-oxo sites of different nuclearity, some of which may be converted to CuI upon contact with methane.22–24
The investigation of the species active in these particular reactions requires a method, which is element-selective and is able to follow the structure and electronic states of copper species. In this respect, X-ray absorption spectroscopy (XAS) represents an excellent choice due to its high sensitivity to the local environment, symmetry and the oxidation state of the elements of interest.25,26 Therefore, it is not surprising that a large fraction of the reports describing the performance of copper-exchanged zeolites utilize various variants of this method.12–15,18,20–24,27–30 Two separate techniques are mostly used: X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS). The first approach allows one to determine, and then follow, the oxidation state of copper in zeolites in the course of reaction and to access the aspects of the local symmetry of corresponding species.25,26 Due to the significantly different spectroscopic signatures of CuI and CuII species with various ligands, the analysis of the data is relatively straightforward and can be done by employing linear combination fitting or multivariate curve resolution methods.15–18,21,23,31–33 Hence, the fractions of different copper species can be extracted; time-resolved measurements enable the monitoring of the reaction kinetics and determination of the rate constants and apparent activation energies.21,24,30 However, the exact determination of the structure of copper species exclusively from a single XANES spectrum is challenging; DFT-assisted modelling represents a promising approach, while requiring more sampling and further development.34,35 Therefore, in the general case, the structural analysis of XANES relies on the comparison with the spectra of standard molecular compounds with known structures.
In contrast, EXAFS modelling is a robust, straightforward technique, and can be utilized on a routine basis for suitable materials, yielding coordination numbers and effective distances between atoms.25,26 It is element-specific and quantitative and can be applied under in situ/operando conditions.25 Unfortunately, the fitting of EXAFS is simple and highly effective mostly for highly ordered materials. Copper-exchanged zeolites do not meet this requirement due to the formation of a mixture of copper species in the pores, resulting in disordering.2,15,18,24,28 This leads to extremely small differences in the EXAFS spectra of different copper-exchanged zeolites measured in a short k-range, which therefore cannot provide any distinct structural information on the nature of multiple copper sites present in the system.28 According to UV-vis, resonant Raman, XAS, EPR and IR spectroscopies, CuII-oxo species are believed to comprise copper dimers, such as bis-μ-oxo10,27 and mono-μ-oxo11,36 sites, trimers Cu3O32+,12 and proximal monomers CuOH+.15,37 Moreover, all these sites are relatively similar from the EXAFS point of view due to the presence of similar scatterers in the first and second coordination spheres of the absorbing copper atom as well as overlapping contributions from silicon/aluminum and copper in the second shell. Hence, the existence of multiple copper-oxo species requires special care in both collecting EXAFS spectra and their comprehensive analysis.
In this respect, we report the analysis of EXAFS of CuII-oxo species hosted in a series of oxygen-activated copper-exchanged zeolites with MOR, MFI, BEA, and FAU frameworks with different Si/Al ratios and copper loadings. We show that the contributions from Cu–Al(Si) and Cu–Cu scattering in the second coordination shell can be resolved by wavelet transform of the EXAFS spectrum. Notably, EXAFS spectra with a large k-range contain important information on the copper coordination related to Cu–Cu scattering pathways, suggesting the formation of different types of copper oxo species. Our results illustrate the power and point out the limitations of advanced EXAFS spectroscopy for copper-containing zeolites, which is of practical importance for the design and optimization of copper-based materials for environmental applications.
Prior to the copper ion exchange, all zeolites were converted to the ammonium form as described elsewhere.31,38 For complete ion exchange with copper, 10 g of each zeolite in the ammonium form were stirred in 1000 ml of a 0.05 M aqueous solution of copper nitrate (99%, Merck) at 323 K for 12 hours. The resulting sample was filtered, washed with 1 l of deionized water and dried at 393 K for 1 h. The ion exchange was repeated two times. The final material was calcined at 773 K for 3 h in a flow of dry synthetic air.
For the partial ion exchange of mordenite (CBV10A) with copper, aqueous solutions of copper nitrate with concentrations ranging from 0.005 to 0.05 M were used.38 The copper content as well as the Si:Al:Na ratio were determined by inductively coupled plasma mass spectrometry (ICP-MS) with an Agilent 77009 ICPMS instrument after complete digestion of the samples in 2 M hydrofluoric acid. Table 1 summarizes the elemental composition of the synthesized materials. The bulk properties are explicitly reported in our previous studies.15,31,38 All synthesized materials were denoted as Cu(x)ZEO(y), where “ZEO” represents the zeolite framework type of the material, and “x” and “y” correspond to the copper loading in wt% and the Si/Al ratio, respectively.
Sample | Chemical composition | |||
---|---|---|---|---|
Cu, wt% | Na, wt% | Si/Al | Cu/Al | |
Cu(3.4)MOR(10) | 3.44 | <0.01 | 10.5 | 0.386 |
Cu(4.0)MFI(12) | 4.02 | <0.01 | 11.6 | 0.534 |
Cu(2.8)BEA(12) | 2.83 | <0.01 | 12.4 | 0.402 |
Cu(2.7)FAU(15) | 2.70 | <0.01 | 14.8 | 0.395 |
Cu(4.4)MOR(6) | 4.36 | 1.22 | 6.5 | 0.377 |
Cu(1.2)MOR(46) | 1.22 | <0.01 | 46 | 0.603 |
Cu(3.5)MOR(6) | 3.52 | 1.93 | 6.5 | 0.309 |
Cu(3.2)MOR(6) | 3.28 | 2.34 | 6.5 | 0.262 |
Cu(2.5)MOR(6) | 2.54 | 2.73 | 6.5 | 0.210 |
Cu(1.7)MOR(6) | 1.74 | 3.38 | 6.5 | 0.144 |
A periodic structure of dicopper mono-μ-oxo species located in an 8-membered ring of mordenite, optimized with density functional theory (DFT) (see the Computational Details section in the ESI†), was used as the initial model for the fitting of Cu–O, Cu–Al(Si) and Cu–Cu single scattering paths in the spectra of CuMOR and CuMFI materials. For the modelling of EXAFS spectra of CuBEA and CuFAU, the copper oxide nanocluster comprising eight copper atoms was placed in the supercage of faujasite and optimized with DFT. The optimized geometries have been adopted as the initial input for the FEFF calculations. The experimental EXAFS spectra were fitted including only single-scattering paths in the fitting model, which represent the dominant contribution in the R-space interval of interest (1.0–3.2 Å). Multiple scattering pathways were excluded based on the results of the fitting of k0 and k1-weighted data (Fig. S2†), which showed the absence of additional contributions in the R-space between 1.0 and 3.0 Å. The same conclusion was derived from fitting of the EXAFS of standard copper(II) oxide with single scattering paths (Fig. S3, Table S1†). The number of parameters employed in the fits varied from 10 to 12, depending on the different input geometries. The quality of the fits was quantified by the R-factor and χ2 parameters.
For the wavelet transform, k3-weighted EXAFS spectra were used. We employed a Morlet wavelet mother function for the WT transform of all spectra within the k-range of 1–16 Å−1.48 The selection of this wavelet was governed by the fast oscillatory part localized in a Gaussian envelope, making its real and imaginary parts similar to an EXAFS spectrum. Continuous wavelet transform was carried out for 128 data points using a built-in function in Mathematica software,50 with subsequent visualization as 2D counterplot projection of a 3D graph.
Notably, the relative intensity and the position of the signals in the 8970–8990 eV energy range does not vary in the spectra of all studied samples of different topologies, Si/Al ratios and copper contents. In contrast, for the copper-containing zeolite of different topologies, the features located within the 8990–9020 eV range differ significantly (Fig. 1a). Hence, the spectrum of Cu(3.4)MOR(10) shows the presence of one broad asymmetric peak with the maximum located at 8999.3 eV. Cu(4.0)MFI(12) and Cu(2.8)BEA(12) reveal similar spectra with at least two resolved peaks centered at 8996.3 and 9000.1 eV. Finally, the spectrum of Cu(2.7)FAU(15) demonstrates a sharp resolved signal at 8998.7 eV with two visible bumps at about 9006 and 9016 eV. The last two signals are also observable in the spectrum of the Cu(2.8)BEA(12) sample; however, they have much lower intensity. For most of the copper-exchanged mordenite samples, the variation of the Si/Al ratio or copper loading (Fig. 1b and c) does not result in detectable changes in Cu K-edge XANES spectra. Only the Cu(1.7)MOR(6) sample demonstrates a slight change in the shape of the white line, which might be associated with a higher amount of CuI species formed via auto-reduction, as compared to the other CuMOR samples.
In contrast, the copper-exchanged mordenites with different Si/Al ratios and copper loadings give very similar EXAFS spectra in k-space (Fig. 2b and c). No significant differences were observed in 5, 8 and 12 Å−1 regions upon variation of the chemical composition of the materials. The only exception is Cu(1.2)MOR(46), which possesses a lower intensity of the EXAFS signals in the 8–16 Å−1 range. However, due to its low copper content, the signal-to-noise ratio is rather poor at high frequencies, which complicates the detailed assessment.
Fig. 3 gives the Fourier transform of k3-weighted EXAFS spectra within the 3–16 Å−1 range. After FT, the prominent difference observed in k-space EXAFS spectra of copper-exchanged zeolites of different topologies becomes more resolved in R-space (Fig. 3a). In FT EXAFS of Cu(3.4)MOR(10) and Cu(4.0)MFI(12) two peaks corresponding to the first and second coordination spheres are visible at the phase-uncorrected radial distances of about 1.5 and 2.3 Å.2,12,14,21,22,24,27,28,33 The first peak originates from the oxygen atoms surrounding the copper atom.2,12,14,21,22,24,27,28,33 The second peak is typically associated with the superposition of contributions from the framework aluminum or silicon atoms and extra-framework copper atoms.2,12,24–26,33 In addition to these peaks, the FT EXAFS of Cu(2.8)BEA(12) and Cu(2.7)FAU(15) show the presence of an additional peak in the second coordination sphere at about 2.8 Å (phase-uncorrected). This observation indicates the presence of other types of copper species in copper-exchanged BEA and FAU.
Similarly to k-space data, the FT EXAFS spectra of copper-exchanged mordenite do not show significant changes upon variation of the Si/Al ratio and copper content. Only a single peak is observed in the second coordination sphere at 2.3 Å due to aluminum, silicon or copper atoms, with no contribution at higher frequencies.
Fig. 4 (a) k3-weighted χ(k) EXAFS data for CuO; (b) corresponding phase-uncorrected FT EXAFS spectra and (c) 2D plots of WT EXAFS for CuO. |
Resembling copper-exchanged zeolites, FT EXAFS data for copper(II) oxide (Fig. 4b) show the presence of two well-resolved peaks due to the first and the second coordination sphere. Moreover, the second coordination shell peak splits into two separate contributions at 2.5 and 2.9 Å, which is the result of destructive interference of the scattering waves from three copper atoms located at 2.88, 3.07 and 3.16 Å apart from the absorbing atom. Additionally, a broad signal centered at 5.5 Å is visible. To correlate these FT peaks with k-space data, wavelet transform was employed (Fig. 4c). Note that all WT EXAFS are not phase-corrected, which results in about 0.4 Å shift with respect to the actual radial distances. The highest intensity belongs to the lobe centered at (6 Å−1; 1.5 Å), corresponding to the oxygen atoms surrounding the central copper atom. At higher radial distances within the range of 2–3 Å, several resolved lobes are located at approximately (3.5 Å−1; 2.3 Å), (6.5 Å−1; 2.5 Å), (10.5 Å−1; 2.8 Å) and (13.0 Å−1; 2.7 Å). To assign them, we use the dependence of the backscattering factor F(k) on the atomic number Z, which indicates that the heavy atoms, having large values of F(k), are localized at higher frequencies in k-space than lighter atoms and vice versa (Fig. S6†). Therefore, the lobe at (3.5 Å−1; 2.3 Å) is due to the oxygen atom, which appears at 2.77 Å in the tenorite structure. The remaining three lobes, having higher k-values, are due to the copper scattering atoms, in line with FT EXAFS data. The splitting into several lobes is associated with the destructive interference of backscattered waves, resulting in several maxima for the magnitude of the signal (Fig. S4†). Finally, the broad lobe centered at (7.5−1 Å; 5.2 Å) and spanning to the region of higher k-values can be clearly linked with the signal at 5.5 Å in the FT EXAFS of copper(II) oxide. High k-values for this signal attribute it to the copper atoms, participating in backscattering, including multiple scattering paths. Note that the WT of k2-weighted EXAFS spectrum preserves its structure pointing to a minimal effect of noise at high k values (Fig. S7†).
WT EXAFS for the Cu(4.0)MFI(12) material differs insignificantly from that of Cu(3.4)MOR(10), as the same combination of signals is present (Fig. 5). The only detectable change is associated with the increase of the intensity and broadening of the lobe at (7.0 Å−1; 2.4 Å), which sums the contributions from aluminum, silicon and copper scatters. For Cu(2.8)BEA(12), a more pronounced change of this signal is visible; it becomes more asymmetric and intense, spanning up to ∼3.5 Å in R-space. Together with that, the signal at (11.0 Å−1; 2.5 Å) demonstrates a higher intensity as compared to the one of Cu(3.4)MOR(10) and Cu(4.0)MFI(12). The WT EXAFS spectrum for Cu(2.7)FAU(15) reveals a substantially different picture with respect to the previously discussed copper-exchanged zeolites: the maximal intensity of the lobe due to the superposition of contributions from aluminum, silicon and copper shifts to a (6.0 Å−1; 2.5 Å) position. Simultaneously, two lobes at higher k-values appear at (11.0 Å−1; 2.6 Å) and (13.0 Å−1; 2.8 Å), corresponding to the scattering from copper atoms. Such a significant difference indicates the presence of copper-oxo species, which might have multiple copper atoms in the second coordination shell.
Varying the silicon to aluminum ratio in copper-exchanged mordenite does not lead to the appearance or disappearance of the signals in WT EXAFS (Fig. 5). Four major lobes at (5.0 Å−1; 1.5 Å), (3.5 Å−1; 2.4 Å), (7.0 Å−1; 2.4 Å) and (11.0 Å−1; 2.5 Å), assigned to oxygen, aluminum, silicon or copper scatters, are well-resolved in the spectra. Notably, the intensity of the latter due to copper atoms in the second coordination sphere is significantly lower for the Cu(1.2)MOR(46) sample, while for other MOR materials the intensity is preserved. This points to a small contribution of copper-oxo oligomeric species, comprising two or more copper atoms in their structure, and confirms the preferential formation of copper monomers, as was previously suggested based on the results of IR spectroscopy of adsorbed nitrogen monoxide and in situ XPS.15,53
In the case of different copper loadings in mordenite, the obtained WT EXAFS are extremely similar to the one observed for Cu(4.3)MOR(6). Even a significant decrease of copper loading from 4.3 to 1.7 wt% does not lead to the disappearance of the lobe at (11.0 Å−1; 2.5 Å) due to copper scatters, pointing to the formation of copper-oxo oligomers in all CuMOR(6) samples.
The analysis of XAS data for the copper-exchanged zeolites of different topologies reveals the presence of significantly different copper-oxo species in these materials (Fig S8–S11†). For quantitative analysis, we performed the fitting of EXAFS data, and Table 2 summarizes the obtained results. The fitting of the first coordination shell is straightforward due to the exclusive contribution of oxygen scattering atoms to the peak at about 1.95 Å. Measured coordination numbers (CN) for copper, surrounded by oxygen, are between 3.5 and 3.9. In most previous studies, coordination numbers close to three are reported;32,33,55–58 however, in several papers a CN of four was observed.28,59–62 Our values fall in this range and, considering the relative accuracy of EXAFS in establishing coordination numbers, correlate with previous reports. The presence of a small amount of CuI species (Fig. 1), formed during the activation of the samples and typically possessing a CN of two, can also contribute to the total coordination of copper, measured by EXAFS.
Parameter | Scatter | Cu(3.4)MOR(10) | Cu(4.0)MFI(12) | Cu(2.8)BEA(12) | Cu(2.7)FAU(15) |
---|---|---|---|---|---|
a Fixed to be equal in the fit. Both scatters are of the same nature (copper atoms); therefore, their Debye–Waller factors are assumed to be similar. | |||||
ΔE, eV | +2.9(7) | +2.3(6) | 0.3(9) | 0.1(9) | |
N O | O | 3.7(2) | 3.9(2) | 3.7(2) | 3.5(3) |
R(Cu–O), Å | 1.944(3) | 1.944(3) | 1.943(4) | 1.959(5) | |
Debye–Waller factor, 10−3 Å2 | 6.3(3) | 6.0(3) | 5.5(4) | 6.1(6) | |
N Al | Al or Si | 1.7(5) | 1.1(3) | 1.2(4) | 1.1(4) |
R(Cu–Al, Si), Å | 2.70(1) | 2.67(1) | 2.70(2) | 2.77(8) | |
Debye–Waller factor, 10−3 Å2 | 10.8(38) | 6.7(20) | 4.7(20) | 11.7(22) | |
N Cu | Cu1 | 0.3(1) | 1.1(4) | 2.3(9) | 1.7(9) |
R(Cu–Cu1), Å | 2.841(16) | 2.904(8) | 2.888(3) | 2.938(2) | |
Debye–Waller factor, 10−3 Å2 | 5.0(21) | 9.4(27) | 10.0(33)a | 6.9(24)a | |
N Cu | Cu2 | 2.6(9) | 1.8(9) | ||
R(Cu–Cu2), Å | 3.05(1) | 3.08(1) | |||
Debye–Waller factor, 10−3 Å2 | 10.0(33)a | 6.9(24)a | |||
R-factor | 0.0029 | 0.0024 | 0.0054 | 0.0066 | |
χ 2-parameter | 187 | 23 | 134 | 116 |
The fitting of the second coordination shell is more challenging. First, EXAFS cannot distinguish between the scattering of aluminum and silicon atoms due to their similar atomic numbers.32,33 Therefore, for the sake of clarity, below we designate their contributions as “Al/Si”, which would virtually mean that both Al and Si atoms are contributing to the particular EXAFS pattern. Using DFT-assisted EXAFS fitting it was demonstrated that employing both aluminum and copper scatterers allows for a significant improvement of the quality of the fit (Fig. S12†).12,14,24,32,33 Previous investigations of theoretically modeled EXAFS spectra showed the principal possibility to resolve Cu–Cu scattering pathways in the second coordination shell using WT.44,48 In this respect, our wavelet analysis provides unambiguous evidence, which enables the discrimination of the different scatters contributing to the peak at about 2.3 Å in FT EXAFS. The presence of two separate lobes at about (7.0 Å−1; 2.4 Å) and (11.0 Å−1; 2.5 Å) in WT EXAFS for Cu(3.4)MOR(10) and Cu(4.0)MFI(12) indicates the existence of both Cu–Al/Si and Cu–Cu paths (Fig. 5). Having these data, we used a DFT-optimized model of dicopper mono-μ-oxo sites, which possesses the neighboring copper and aluminum atoms at a distance of about 2.7 Å from the central copper as the initial guess for the fitting of the second coordination shell (Table 2). Cross-validation of WT EXAFS by the fitting of FT EXAFS data returned unsatisfactory fits for exclusively aluminum or copper scatters, which is in line with previous reports (Fig S12 and S13†).
The fitted Cu–Al and Cu–Cu distances are slightly different: in the case of Cu(4.0)MFI(12) the aluminum atom is located closer (2.67 Å) and the copper atom is located farther (2.90 Å) from the central scatter, with respect to the values calculated for Cu(3.4)MOR(10). This indicates a different geometry and, in particular, the Cu–O–Cu bond angle of copper-oxo species governed by the pore structure of zeolites, in line with previous DFT studies.63,64 The second shell coordination number for Al/Si in Cu(3.4)MOR(10) is 1.7, while for Cu(4.0)MFI(12) it is lower and has a value of 1.1. Such a low CN for Al/Si in the second shell in the case of copper-exchanged MFI comes together with a relatively high CN for copper, estimated to be close to 1 (Table 2). In contrast, Cu(3.4)MOR(10) gives a coordination number for copper of only 0.25. Summarizing these data, copper-exchanged MFI on average has at least one copper and one aluminum/silicon atom in the second coordination shell, which agrees with the simplest model of dicopper mono-μ-oxo sites. The formation of dicopper bis-μ-oxo sites and tricopper oxo species or a mixture of thereof cannot be excluded based on such analysis. However, comparing this sample with Cu(3.4)MOR(10), which has a significantly smaller contribution of copper in the second shell, it can be concluded that the fraction of multinuclear copper-oxo sites is considerably lower in copper-exchanged mordenite with respect to copper-exchanged MFI. For Cu(3.4)MOR(10), the formation of a large fraction of copper monomeric sites, such as Cu2+and CuOH+, explains the low CN for copper in the second coordination shell, while for Cu(4.0)MFI(12) the presence of dicopper mono-μ-oxo sites has already been independently demonstrated by UV-vis and Raman spectroscopy.65 This discussion is in line with WT EXAFS spectra of these samples, showing different intensities of the lobe at (11.0 Å−1; 2.5 Å) due to the copper scattering atoms (Fig. 5). The noted difference is coherent with the changes in XANES spectra (Fig. 1a), which shows different structures of the white line for these two samples, hence confirming the presence of various copper-oxo species. The fitted Debye–Waller factors appear to be relatively high for the spectra collected under cryogenic conditions. We associate this observation with the structural rather than thermal disorder of the copper species.
Two other copper-exchanged zeolites, BEA and FAU, revealed the presence of a splitting of the peak in the second coordination sphere in FT EXAFS (Fig. 3a, S10 and S11†). Moreover, WT EXAFS shows at least three separate contributions to this peak, coming from the signals centered at 7.0, 11.0 and 13.0 Å−1 in k-space (Fig. 5). Comparing these patterns with the one obtained for bulk copper oxide (Fig. 4), we conclude that two distinct copper atoms are contributing to the second shell together with framework aluminum atoms. Therefore, we used a copper oxide cluster, having several different Cu–Cu distances, accommodated in FAU zeolite, as the starting model for the fitting of FT EXAFS.
Table 2 gives the results of the fitting for Cu(2.8)BEA(12) and Cu(2.7)FAU(15) with two distinct copper atoms contributing to the second shell. An aluminum or silicon atom is located ∼2.70 Å away from the central copper atom and has a coordination number close to 1; the distance and CN are similar to those observed for copper-exchanged MFI. The fitted Cu–Cu1 and Cu–Cu2 distances for Cu(2.8)BEA(12) are 2.89 and 3.05 Å, which are slightly lower than those for Cu(2.7)FAU(15) (2.94 and 3.08 Å, respectively). This difference is within the accuracy of the EXAFS method and virtually suggests similar distances for Cu–Cu in both Cu(2.8)BEA(12) and Cu(2.7)FAU(15). The coordination numbers for both copper scatters are different: for copper-exchanged BEA CNs are close to ∼2.6, while Cu(2.7)FAU(15) gives a value of ∼1.7. The average coordination numbers suggest the formation of on average larger copper-oxo clusters in copper-exchanged BEA compared to FAU. However, the CN of copper is still considerably lower than that in copper oxide with a tenorite structure, where the CN is equal to four. Also, the absence of high frequency oscillations within the range of 7–10 Å−1 in k-space and corresponding to the peak at about 5.5 Å in FT and WT EXAFS excludes the formation of very large copper oxide nanoparticles (Fig. 2a, 3a and 5). The formation of copper-oxo clusters is also in line with XANES, which in the case of Cu(2.7)FAU(15) shows a sharp peak at 8998.7 eV (Fig. 1a). A very similar signal centered at 8998.0 eV is present in the XANES spectrum of the copper(II) oxide standard (Fig. S14†), hence suggesting the possible existence of copper-oxo species with multiple Cu–Cu contributions in Cu(2.8)BEA(12) and Cu(2.7)FAU(15). Notably, for Cu(3.4)MOR(10), which showed a very small contribution of copper to the second coordination shell, the white line peak is broad and shifted to a higher energy of 8999.3 eV (Fig. 1a). Therefore, we speculate that the position and the shape of the features in the XANES spectrum might be sensitive to the copper coordination and can be correlated with the nature of copper-oxo species hosted in oxygen-activated zeolites. However, more research is required in this direction.
Summarizing, the results of FT EXAFS fitting confirm the stabilization of completely different copper-oxo species in zeolites of different topologies and pore sizes. In copper-exchanged mordenite, the contribution from copper in the second coordination shell is not very pronounced which indicates the formation of a very high fraction of copper monomeric species. The fitted EXAFS coordination number of CuMFI is in agreement with the presence of dicopper species; however, other possibilities cannot be refuted. Copper-exchanged BEA and FAU show a high fraction of copper-oxo clusters of a very small size. This property to stabilize clusters might be associated with the size of the 12MR pores for BEA and the existence of the supercage in FAU.
We also considered the possibility that all copper-exchanged zeolites including the ones of different topologies have exactly the same copper-oxo species, but differ by the degree of static disorder. Practically, stronger disordering would lead to broader peaks in FT EXAFS, including the second coordination shell. If true, the models used for copper-exchanged faujasite and comprising two different Cu–Cu scattering pathways should be able to fit CuMOR and CuMFI spectra with only higher Debye–Waller factors, while other parameters should persist. To verify this hypothesis we performed the corresponding fits in k3-space. However, the obtained results showed significant misfits in the second coordination shell for both copper-exchanged MOR and MFI, together with worse statistical parameters as compared to the initial models (Table S2, Fig S15 and S16†). This points out that the difference observed in EXAFS spectra cannot be associated with the variation of static disordering only and indicates the presence of different copper-oxo species.
Next, the FT EXAFS spectra of copper-exchanged mordenite samples with different Si/Al ratios were fitted (Fig S8, S17 and S18†). Table 3 reports the obtained interatomic distances, coordination numbers and Debye–Waller factors. The increase of the Si/Al ratio from 6 to 10 does not lead to any considerable difference in the fits: all extracted parameters stay virtually the same. In contrast, a further decrease of the aluminum content to a Si/Al ratio of 46 in the parent mordenite results in significantly different fitting parameters for Al/Si atoms in the second shell: the coordination number decreases to 0.5 with the Cu–Al distance decreased to 2.64 Å as compared to Cu(4.3)MOR(6). Surprisingly, in spite of low probability to form multinuclear copper species in zeolite with high Si/Al, the fitted copper coordination number does not experience any change and is estimated at about 0.3, indicating the presence of Cu–Cu scattering pathways. This observation is in contradiction with the WT EXAFS obtained for the Cu(1.2)MOR(46) sample, which shows a significant decrease of the intensity of the lobe at (11.0 Å−1; 2.6 Å), assigned to a copper scatterer, as compared to the copper-exchanged mordenites with a higher aluminum content. We therefore assume that the results of FT EXAFS fitting for Cu(1.2)MOR(46) might be partially compromised by the low signal-to-noise ratio of the corresponding spectrum in combination with the low intensity of the scatters, which is evidenced by the high R-factor, returned after the fitting (Table 3, Fig. S18†). For a conclusive discussion, an even higher quality of the spectra is needed, which is a challenging task due to the low copper content in the material. Qualitatively, we rely on the results of WT EXAFS that show a very small impact of copper scattering at high k-values, suggesting the formation of copper monomeric species, in line with our previous studies devoted to infrared spectroscopy of adsorbed nitrogen monoxide, which is considered as the probe, sensitive to the nuclearity of copper species in zeolites.15,56,58 We showed that the spectra of adsorbed NO are very similar for Cu(4.4)MOR(6) and Cu(3.4)MOR(10), revealing the presence of identical bands with similar intensities, while Cu(1.2)MOR(46) gives a dramatically different spectrum. In the region of the CuII species, only one intense band at 1908 cm−1 was observed, hence indicating the dominant formation of monomeric copper species, whereas for samples with Si/Al ratios greater than 10, copper sites of different nuclearity were present.15
Parameter | Scatter | Cu(4.3)MOR(6) | Cu(3.4)MOR(10) | Cu(1.2)MOR(46) |
---|---|---|---|---|
ΔE, eV | +1.7(6) | +2.9(7) | −2.8(13) | |
N O | O | 3.9(2) | 3.7(2) | 3.9(4) |
R(Cu–O), Å | 1.938(3) | 1.944(3) | 1.917(7) | |
Debye–Waller factor, 10−3 Å2 | 6.8(4) | 6.3(3) | 9.2(9) | |
N Al | Al or Si | 1.8(7) | 1.7(5) | 0.5(3) |
R(Cu–Al, Si), Å | 2.69(2) | 2.70(1) | 2.64(2) | |
Debye–Waller factor, 10−3 Å2 | 11.3(60) | 10.8(38) | 4.1(37) | |
N Cu | Cu | 0.3(4) | 0.25(7) | 0.3(3) |
R(Cu–Cu), Å | 2.86(3) | 2.84(2) | 2.87(2) | |
Debye–Waller factor, 10−3 Å2 | 7.1(68) | 5.0(21) | 5.0(49) | |
R-factor | 0.0022 | 0.0029 | 0.0072 | |
χ 2-parameter | 111 | 187 | 207 |
Finally, the effect of copper loading on the FT EXAFS fits and, hence, the structure of copper-oxo species, was evaluated using a series of copper-exchanged mordenite samples of Si/Al = 6 with different copper loadings (Table 4, Fig. S19–S22†). The fitted Cu–O, Cu–Al and Cu–Cu distances over all samples give the values of 1.94, 2.66 and 2.85 Å, respectively. The first shell coordination number for copper surrounded by oxygen slightly decreases from 3.9 to 3.7 with decreasing copper content, which falls within the relative accuracy of EXAFS. Similarly, the second shell aluminum/silicon coordination number changes from 1.8 to 1.2. Notably, the copper–copper coordination number is preserved and does not vary from Cu(4.3)MOR(6) to Cu(1.7)MOR(6). These data show that the ion-exchange of mordenite with different copper concentrations results in the observation of qualitatively and quantitatively identical EXAFS spectra. In agreement with that, the Cu K-edge XANES spectra (Fig. 1c) demonstrate a similar intensity of pre-edge peaks and structure of the white line. Combining these data with recent results of infrared spectroscopy of adsorbed nitrogen monoxide, we conclude that the structure of copper-oxo species in this series of the samples is very similar and EXAFS and IR NO are not sensitive enough to detect any possible difference.
Parameter | Scatter | Cu(4.3)MOR(6) | Cu(3.5)MOR(6) | Cu(3.2)MOR(6) | Cu(2.5)MOR(6) | Cu(1.7)MOR(6) |
---|---|---|---|---|---|---|
ΔE, eV | +1.7(6) | +2.1(6) | +2.6(7) | +1.4(12) | +1.5(12) | |
N O | O | 3.9(2) | 3.8(1) | 3.6(2) | 3.6(3) | 3.7(3) |
R(Cu–O), Å | 1.938(3) | 1.940(3) | 1.944(3) | 1.933(1) | 1.93(2) | |
Debye–Waller factor, 10−3 Å2 | 6.8(4) | 6.1(3) | 6.0(4) | 6.4(6) | 6.1(6) | |
N Al | Al or Si | 1.8(7) | 1.6(3) | 1.5(3) | 1.1(6) | 1.2(7) |
R(Cu–Al, Si), Å | 2.69(2) | 2.66(1) | 2.67(1) | 2.66(2) | 2.66(2) | |
Debye–Waller factor, 10−3 Å2 | 11.3(60) | 8.3(21) | 7.5(20) | 8.0(49) | 8.6(58) | |
N Cu | Cu | 0.3(4) | 0.2(2) | 0.1(1) | 0.4(5) | 0.3(4) |
R(Cu–Cu), Å | 2.86(3) | 2.86(1) | 2.87(1) | 2.88(3) | 2.85(2) | |
Debye–Waller factor, 10−3 Å2 | 7.1(68) | 4.6(36) | 2.7(33) | 7.6(77) | 5.1(20) | |
R-factor | 0.0022 | 0.0021 | 0.0031 | 0.0089 | 0.0098 | |
χ 2-parameter | 111 | 95 | 73 | 64 | 83 |
Applying our approach, we observed a strong dependence of copper speciation on the zeolite topology. We established that zeolites BEA and FAU form copper-oxo clusters upon ion exchange and high-temperature oxidative treatment. In contrast, CuMOR revealed the presence of mostly copper monomeric species, while CuMFI demonstrated the formation of a significant fraction of copper-oxo dimers. A moderate variation of the Si/Al ratio or copper loading in copper-exchanged mordenite did not reveal a conclusive difference in either EXAFS or XANES spectra. A drastic increase of the Si/Al ratio in mordenite results in a very low copper loading in the final material, making the collection of a high quality spectrum and its unambitious analysis challenging. Our study identifies limitations and opportunities in the application of XAS by showing the great potential of wavelet analysis of EXAFS spectra of copper-containing materials and offers a methodology for the structural analysis of copper species stabilized in zeolites.
Footnote |
† Electronic supplementary information (ESI) available: Results of k3-weighted fitting for all samples, Cu K-edge XANES spectrum of CuO, k0- and k1-fitting of the Cu(4.3)MOR(6) sample, comparison of the fits with Al and Cu only contributing to the second coordination shell, backscattering factors, EXAFS fit for CuO and details of the theoretical calculations. See DOI: 10.1039/d0sc01472a |
This journal is © The Royal Society of Chemistry 2020 |