3100
J . Phys. Chem. 1992,96, 3100-3109
In an earlier related study, Khouzami et al.3 had indicated the presence of an octahedral A1 species from X-ray diffraction measurements, but it was not confirmed by the analysis of a high-field *'A1 MASNMR spectrum. Tapp et al.,2 on the other hand, had not extended the spectral studies to include the high-field region and hence failed to notice the presence of this broad peak. This led to the conclusion that all the water molecules in the hydrated samples are only physisorbed. However, Goepper et a1.I6 have assigned the broad signal to the presence of octahedral Al. Granting that the broad peak in our case is due to a 6-coordinated A1 species, it should be possible to estimate the fraction of the total Al in the framework that is present in this state. This can be done by subtracting the area under the spinning sideband (6 to -70 ppm) from the area under the broad signal and relating it to the total area under the 27Alsignals. Following this procedure, it is found that nearly 20% of the total A1 may be assigned an octahedral coordination. Significantly, the fraction corresponds to the fraction of A13 sites (Figure 1) which are the favored sites for the location of water molecules on the basis of MNDO energy calculations. Further evidence for the existence and essential accuracy of the estimate of octahedral A1 species comes from the 31Pstudies. In the MASNMR spectra, a single peak at -30 ppm was observed for the AS sample., This peak was unaffected by calcination and dehydration in sample CD (Figure 4c). In sample FH, an additional peak at -23.4 ppm attributable to interaction of framework P with water molecules2J8was noticed. Such an interaction can cause deshielding of P, which causes an increase in the observed chemical shift. The area of this peak is, again, like as in the case of the octahedral A1 species, computed to be up to 20% of the total area under the 31Psignals. Naturally, therefore, it is the P3 sites which are involved in interaction with water, and the number of these sites corresponds to 20% of the total P sites (Figure 1). Thus, both 27Al and 31PMASNMR (18) Jahn, E.; Muller, D.; Becker, K . Zeolites 1990, 10, 151.
spectroscopic studies furnish data supporting the predictions of adsorption energy calculations. Conclusions The structural changes occurring in the AlPO,-1 1 framework due to hydration are evident from the XRD studies. The MNDO technique has been found to be useful for studying different modes of interaction of water with TO, groups in the AlP04-11 framework by cluster model calculations. 27Aland MASNMR spectra were recorded to study the environmental changes occurring at T sites due to water adsorption. Some of the notable findings of the present study are given below: 1. Water prefers to undergo three-site adsorption over the TO, group as shown in Figure 2a. 2. Among the different T sites, T3 (where T = A1 or P) shows favorable adsorption energy for water due its geometry (wide T-O-T angles). 3. The weakening of the 0-H bonds in water molecules is insignificant, showing that the latter do not dissociate. Hence, T sites are in a pseudooctahedral environment which can revert back to a tetrahedral environment on dehydration. 4. For the hydrated sample FH, 27Al and 31PMASNMR spectra show additional signals a t around -20 and -23.4 ppm, respectively, indicating interaction of the T sites with water molecules. 5 . In a quantitative analysis, ratios of intensities of the two signals in the N M R spectra of the hydrated sample have been calculated. These ratios of the intensities of the additional peaks (arising from water adsorption) to the corresponding tetrahedral peaks for Al and P are -2080, consistent with interaction between water molecules and the T3 sites causing distortion of the framework.
Acknowledgment. We thank Dr. P. R. Rajamohanan for MASNMR experiments. This work was partly funded by the UNDP. Registry No. H20, 7732-18-5.
Electron Spln Echo Envelope Modulation Studies of Cu2+ Interactions with Framework AI in Dehydrated X, Y, and A Zeolites Kbalid Matar and Daniella Goldfarb* Department of Chemical Physics, The Weizmann Institute of Science, 76 100 Rehovot, Israel (Received: August 5, 1991) The interactions of Cu2+cations with framework A1 in zeolites NaX, KX, Nay, and NaA in various stages of dehydration and after methanol adsorption were investigated by electron spin echo envelope modulation (ESEEM) spectroscopy. The FT-ESEEM spectra of all species studied showed contributions from two types of 27Alnuclei. The first type, referred to as first shell, consists of 27Alnuclei bonded to the oxygens to which the Cu2+is coordinated. They exhibit relativdp large isotropic hyperfine constants which manifest the firm binding of the Cu2+to the framework. The second type consists of A1 nuclei which are coupled to the Cu2+by weak dipolar interactions and are termed distant Al. Orientation selectiveESEEM experiments were carried out as well to obtain additional structural information. The modulation amplitude corresponding to distant A1 did not show any dependence on the irradiation position within the EPR powder pattern as the arrangement of those nuclei around the cation is approximately isotropic. The modulation amplitudes of the first shell AI in some cases showed strong orientation dependence which was interpreted in terms of the site geometry. The amount of adsorbed methanol in Cu-NaX has a considerable effect on the relative intensities of the peaks of the first shell AI and distant AI. This dependence is explained in terms of the changes in the 27Alquadrupole coupling constants of the two types of Al. Introduction Zeolites are microporous materials which have many applications as molecular sieves, ion exchangers, and catalysts.' The exchangeable cations play an important role in many of these
applications. Accordingly, considerable efforts were made through the years to obtain their location2and understand the factors which govern their distribution among the various sites* In hydrated zeolites the cations are mobile and undergo rapid exchange between the various sites. Upon dehydration they be-
( I ) Breck, D. W. Zeolite Molecular Sieues; John Wiley and Sons: New
(2) Mortier, W. J. Compilation of extraframework sites in zeolites; Butterworths: Guildford, 1982.
York, 1974.
0022-3654/92/2096-3 100$03.00/0
0 1992 American Chemical Society
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992 3101
Cu2+Interactions with Zeolites
TABLE I: List of Samples Investigated, the Cu2+Species Generated, and Their Corresponding EPR Parameters gll
zeolite
treatment
Cu-NaX Cu-NaX
400 O C 400 O C , 0 2
Cu-NaX CU-KX
400 OC, O2 + M e O H 100 O C
CU-KX
300-500 O C ,
Cu-NaA Cu-NaA Cu-NaY Cu-NaY
400 OC, 02 400 "C,O2 + MeOH 400 OC, 02 25 O C
0 2
species
(&0.005)
(&0.005)
A A A' A A A' A A'
2.35 2.35 2.37 2.35 2.34 2.37 2.34 2.37 2.37 2.31 2.36 2.36
2.062 2.062 -2.06 2.060 2.062 -2.06 2.062 -2.06 2.065 -2.07 -2.07 -2.07
D G
F F
come localized and coordinate to framework oxygens; therefore they distort the framework geometry and affect the electron density distribution on the neighboring Si and AlU3This is well manifested in the 29SiN M R spectra which in dehydrated zeolites are often broad and unresolved as compared to the spectra of hydrated zeolite^.^-^ This broadening is a consequence of cations located a t different sites exhibiting different interactions with the f r a m e w ~ r k . ~ .Due ~ to the loss of resolution in the 29SiN M R spectrum and often to the total disappearance of the 27AlN M R signal: it is practically impossible to isolate the specific cation interactions with the framework at particular sites. When the exchangeable cations are paramagnetic, their close environment can be investigated by electron paramagnetic resonance (EPR) methods. Although the EPR spectrum is very sensitive to the surroundings of the paramagnetic cation, as manifested by the g, hyperfine, and the zero field splitting (ZFS) tensors, it cannot provide directly the fine interactions with the framework. These are often expressed as superhyperfine interactions with the 27Alnuclei of the framework, which are too weak to be resolved in the broad EPR powder pattern. Such fine interactions can be measured by the electron spin echo envelope modulation (ESEEM) technique. The ESEEM method is particularly useful to measure weak hyperfine interactions. The modulation frequencies, often termed as ENDOR frequencies, are the N M R frequencies of the coupled nuclei. In disordered systems, when the hyperfine interaction does not contain an isotropic part and is mainly dipolar in nature, the modulation frequencies are practically the Larmor frequencies of the coupled nuclei. This serves to identify the coupled nuclei. The modulation depth, in turn, can be related to the distance between the electron spin and the coupled nuclei and to their number (once the nuclear quadrupole interaction is relatively small). When the isotropic hyperfine interaction is not negligible, the modulation frequency will be split by the isotropic hyperfine constant.' ESEEM has been successfully employed in the past to study the interactions of paramagnetic transition metal cations with adsorbates and to follow the generation of complexes within the zeolite pores.*-I0 In these experiments the modulation investigated was from ,Hnuclei of specifically deuterated adsorbates. Often, these experiments also provide the location of the cations. The location of the paramagnetic species generated in the zeolites containing Cs+ cations can also be obtained from 33Csmodulation."J2 ~~~~
~
g,
~~
(3) Engelhardt, G.; Michel, D. High Resolution Solid State NMR of Silicates and Zeolites; John Wiley and Sons: New York, 1987; p 256. (4) Grobet, P. J.; Mortier, W. J.; van Genechten, D. Chem. Phys. Lett. 1985, 119, 361. (5) Melchior, M. T.; Vaughan, D. E. W.; Jacobsen, A. J.; Pictoroski, C. F. Proceedings of the 6th International Zeolite Conference (Reno 1983); Butterworth: Guildford, 1983; p 684. (6) Vega, A. J.; Luz, Z. J . Phys. Chem. 1987, 91, 365. (7) Kevan, L. In Time Domain Electron Spin Resonance; Kevan, L., Schwartz, R. N., Eds.; Wiley: New York, 1979; Chapter 8. (8) Kevan, L. Acc. Chem. Res. 1986, 20, 1. (9) Goldfarb. D.; Kevan, L. J . Phys. Chem. 1986, 90, 264. (10) Kevan, L.; Narayana, M. Intrazeolite Chemistry; American Chemical Society Symposium Series No. 218; Stuckey, G. D., Dwyer, F. G., Eds.; American Chemical Society: Washington DC, 1983; p 283.
All
(*5 x 1041,
cm-I 153 X 153 X 139 X 153 X 143 X 139 X 143 X 139 X 145 X 163 X 145 X 145 X
lo4 lo4 lo4 lo4 lo4 lo4 lo4 lo4
lo4 lo4 lo4 lo-'
In a previous publication we showed that the 27Almodulation is sensitive to the hydration state of the ~eo1ites.I~We also found that in Cu-NaX the two major Cu2+ species, generated in the course of dehydration, each exhibit a characteristic 27Al FTESEEM spectrum.I4 In partially dehydrated samples the Cu2+ species showed practically no isotropic hyperfine interactions with the 27Al,whereas in completely dehydrated samples the Cu2+ species is characterized by a relatively large 27Al isotropic superhyperfme constant of -2.5 MHz. The latter indicates a strong interaction with the framework oxygens responsible for the transferred hyperfine to the neighboring Al. In the present work we extend our investigation to Cu2+ in zeolites A and Y. We also report on orientation-selective ESEEM experiments on Cu2+ in X, Y , and A zeolites which provide additional structural information leading to site identification. In addition, we address the question of the effect of the 27Alquadrupole on the ESEEM pattern by following the ESEEM as a function of amount of adsorbed methanol. Experimental Section Sample Preparation. Zeolites NaX (13X, Si/Al = 1.25) and NaA (4A, Si/Al = 1.0) were purchased from Aldrich. NaY (Si/Al = 2.5) was a slft from Conteka, Sweden. KX was obtained from NaX by cation exchange and Cu2+ was introduced by exchange with a &(NO3), solution (10" M) as described previ0us1y.'~ The exchange level is approximately one Cu2+cation per 7 and 40 unit cells for X and Y zeolites and zeolite A, respectively. Dehydration was carried out by slowly heating the sample (50 OC/h) under vacuum to the desired dehydration temperature at which the sample was left under a residual pressure of Torr for 16 h. Most of the samples dehydrated above 300 OC were exposed to 400 TOKoxygen for 2 h at the dehydration temperature to oxidize any Cu+ or Cuo which formed during the dehydration process. The oxygen was then pumped off and the samples were cooled to room temperature and sealed. We shall refer to oxidized samples in the text by adding 0,near the dehydration temperature. For example Cu-NaX (400 "C, 0,) indicates that the sample was dehydrated at 400 OC and oxidized. Methanol adsorption was done in the gas phase. Samples saturated with methanol were obtained by equilibrating the dehydrated sample (after oxidation) with methanol vapors for several hours. The other samples were prepared by adsorbing a known volume of methanol vapors at a given pressure onto the activated zeolite sample a t 77 K. All samples were sealed while cooled to I7 K. Instrumentation. ESR measurements were performed at 110 K on a Varian E-12 spectrometer operating at X-band and the ESEEM measurements were done at 4 K using a home-built spectrometer operating at 9.1 GHz.I5 The ESEEM measurements
-
(1 1) Narayana, M.; Kevan, L. J. Chem. Sac., Faraday Trans. I 1986,82, 213. (12) Anderson, M. W.; Kevan, L. J . Phys. Chem. 1986, 90, 6452. (13) Goldfarb, D.; Kevan, L. J. Magn. Reson. 1989, 82, 270. (14) Goldfarb, D.; Zukerman, K. Chem. Phys. Lett. 1990, 171, 168. (15) Goldfarb, D.; Fauth, J.-M.; Tor, Y.; Shanzer, A. J . Am. Chem. Soc. 1991, 113, 1941.
Matar and Goldfarb
3102 The Journal of Physical Chemistry, Vol. 96, No. 7, 1992
TABLE 11: The A7f Factors Corresponding to the 2.6,. 5.3-.. and 3.5-MHz Lines (Ciiculated from Equation i ) 7 - us A (2.6)" A (5.3) A (3.5)
ll \
-
0.18 0.20 0.22 0.24 0.26 0.30 0.40
- _ _ _ c
260ns
0
0 0
0.04 0.07 0.5 1.13 1.72 1.84 0.27
1.68 1.31 0.87 0.46 0.15 0.05 1.8
"The number in parentheses corresponds to the modulation frequency (in MHz) used in eq 1; hence A (2.6) is related to the intensity of the 5.3-MHz line and vice versa.
051
-
1.98 I .99 1.90 1.71 1.45 0.8 1 0.03
2
I
3
T ips)
Figure 1. Three-pulse ESEEM of dehydrated Cu-NaX (400 " C ) obtained a t different T values, recorded at 3 150 G.
were obtained using the two (a/2-~-a-~-echo) and three (a/2-~1r/2-T-rr/2-recho) pulse sequences with the appropriate phase cycling.16 Typically, pulses of 2 0 4 s duration were used for r / 2 pulses with the power of about 40 W. The boxcar gate was adjusted to read at the center of the echo. The T increment was 10 or 20 ns. Prior to Fourier transformation the missing data points, due to spectrometer dead time, were reconstructed using the LPSVDL7method. All spectra shown are the cos FT-ESEEM.
Results Dehydrated Samples. Table I lists the Cu2+species generated in the investigated zeolites and their corresponding EPR parameters. The parameters obtained are in good agreement with those already reported in the literat~re.~*,l~ The Cu2+species obtained in dehydrated Cu-NaX and Cu-KX have rather close EPR parameters and were thus previously assigned to the same location.IsJ9 Dehydration of Cu-NaX generates a single Cu2+species, termed A , accompanied by a significant decrease in the Cu2+ concentration. Oxidation at 400 O C regenerates most of the Cu2+ concentration but produces an additional minor species A'. The 27Al ESEEM characteristics of species A were previously reported.I4 It exhibits a relatively shallow modulation with frequencies of 2.6 and 5.3 MHz. We have assigned these lines to a doublet split by an isotropic hyperfine constant of 2.7 f 0.2 MHz. In addition to this doublet another modulation frequency at the 27Al Larmor frequency, 3.5 MHz, was observed. The relative intensity of this frequency varied from sample to sample. To find the origin of this frequency and to substantiate the assignment of the doublet, we carried out dependent measurements of the three-pulse ESEEM of a nonoxidized sample as compared to an oxidized sample. Selected time domain S E E M patterns of a nonoxidized sample are shown in Figure 1 and the FT-ESEEM spectra of a nonoxidized and an oxidized sample are shown in Figure 2. The FT-ESEEM spectra show all three peaks mentioned above and an additional weaker peak at 4 MHz. The latter is more pronounced in Cu-NaX without oxidation. The relative intensity of the 3.5-MHz line is weaker in the spectra of the nonoxidized samples which do not contain species A'; hence, we assign it partially to species A'. This explains the intensity variation of this line which is a consequence of the changes in the relative amount of species A' due to variations in the activation condition (temperature, oxygen pressure, etc.). The spectra shown in Figure (16) Fauth, J.-M.; Schweiger, A,; Brauschweiler, L.;Forrer, J.; Ernst, R.
2 are all normalized to the most intense peak. This presentation of the data portrays the dependence of the relative intensity of the various peaks within the spectrum as function of T but the dependence of the absolute peaks intensity on T is lost. This can be recovered by the relative intensity of each spectrum indicated in parentheses in the figure. In the three-pulse experiment the intensities of the ENDOR frequencies of the CY manifold, uu. are modulated by the ENDOR frequencies of the /3 manifold, uknr and vice versa." For the case of I = 5 / 2 one expects each transition to be modulated by at least five frequencies (the frequencies corresponding to AmI >1 are not considered because their probabilities are much smaller). Assuming that the probability of the forbidden EPR transition are small as compared to the allowed transitions, the term responsible for the amplitudes modulation is approximated byZo
r
A;(@)= 1 - cos
In dehydrated zeolites the 27Alquadrupole coupling constant is largel3vZ1causing a considerable broadening of the 0 4 ~transitions with the exception of the I l l 2 ) + I transitions, which are affected by the quadrupole interaction only to second order. Accordingly, we assume that the major contributions to the 2.6 and 5.3 MHz lines are from the 11/2)-1-1/2) transitions.22 Table I1 lists the Aijfactors for these frequencies for the 7 values used in the experiment. The general behavior predicted by eq 1 is also observed experimentally. At shorter T values the 5.3 MHz line is the stronger, whereas at longer T values the 2.6 MHz dominates. This T dependence supports the assignment of the two lines to different electronic spin manifolds. Obviously the spin system in question is more complex than is required for eq 1 to be a good approximation, thus only a course agreement with it is expected. We do not understand why the 5.3 MHz line in the nonoxidized Cu-NaX samples is so weak at low T values. Note that in this case the whole modulation is rather shallow. The center of the doublet is at 3.95 MHz which is shifted by 0.45 MHz from the Larmor frequency. This shift is attributed to the quadrupole interaction and it has also been demonstrated by ~imu1ations.l~Whether this shift can be directly utilized to estimate the quadrupole coupling constant has yet to be investigated. The peak at 3.5 MHz coincides with the 27AlLarmor frequency and accordingly we assign it to nuclei that are weakly coupled by dipolar interaction to the Cu2+. In this case for the 1-1/2) transitions ua= (3B = q. Assuming again that these transitions have the largest contribution to the echo intensity, this line should be modulated by 3.5 MHz. The A factors corresponding to this line calculated for several T values using eq 1 are shown in Table 11. Again, a reasonable agreement with Figure 2b is observed. At this stage we can distinguish between two types of Al nuclei interacting with the Cu2+cation in species A. One type includes the first shell A1 nuclei responsible for the doublet. They are bonded to the oxygens to which the Cu2+is coordinated, and their distances from the Cu2+are in the range of 0.31-0.34 nm.2 The second type are distant A1 nuclei contributing to the modulation
-
R.J . Magn. Reson. 1986, 66, 74. (17) Barkhuijsen, H.; de Beer, R.; Bovee, W. M. M. J.; van Ormondt, D. J . Mag. Reson. 1985, 61, 465. (18) Schoonheydt, R. A. J. Phys. Chem. Solids 1989, 50, 523. (19) Ichikawa, T.; Kevan, L. J . Am. Chem. SOC.1983, 105, 402.
(1)
~~
~
(20) Mims, W. B. Phys. Reo. B 1972, 6, 3543; 1972, 5, 2409. (21) Luz. Z.; Vega, A. J. J. Phys. Chem. 1986, 91, 374. (22) Matar, K.; Goldfarb, D. J . Chem. Phys., in press.
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992 3103
Cu2+ Interactions with Zeolites
400 nr
300 ns /
(1.4)
260 ns
240 ns
I
I
0
I
I
I
5
240 n s
I
I
10
I
I
I
I
I
5
0
15
I
I
15
10
Frequency (MHz)
Frequency (MHz)
Figure 2. Three-pulse FT-ESEEM of Cu-NaX dehydrated at 400 "C, (a) without oxidation and (b) with oxidation, as a function of were recorded at 3150 G. The numbers in parentheses give the relative intensity of each spectrum.
spectra
b
I
u
7. The
480°C
0 .o 00
I .o
2.0
3.0 T (ps)
4.0
5.0
0.0
4.0
8.0
12.0
16.0
F r e q u e n c y (MHz)
Figure 3. (a) Three-pulse ESEEM of Cu-KX activated at 100, (300 OC, 0 2 )and , (480 'C, 0 2 ) .(b) The corresponding FT-ESEEM
(T
= 0.22 ps,
H = 3150 G).
at the Larmor frequency. They are situated at distances between 0.5 and 0.8 nm from the Cu2+. This distance range was obtained as follows: The distances of the second, third, fourth, and fifth shells of A1 nuclei from several cation sites were obtained from X-ray dataz3and then the A1 modulation of each shell was calculated using the point dipole and the spherical model approximations: not including the quadrupole interaction. This provides an upper limit for the modulation depth since the quadrupole interaction tends to damp the modulation depth.9 We found that at a distance larger than 0.8 nm the modulation depth was practically zero. A close inspection of both sets of spectra shown in Figure 2 show a line at 4.0 MHz as well. So far we cannot assign this line. (23) Maxwell, I. E.;de Beer, J. J. J . Phys. Chem. 1975, 79,1874.
It could originate from a transition other than the l'/2)-l-'/z) transitions which in one of the manifolds is relatively narrow due to accidental cancellation of the anisotropy of the quadrupole tensor by the anisotropic hyperfine term.,* Dehydration of Cu-KX yields species A and A' as well. In this case however, the Cuz+ cation assumes its final position already at a dehydration temperature of 100 OC as indicated by the EPR spectrum. Figure 3 shows the threepulse ESEEM (T = 0.22 ps) for samples dehydrated at 100 OC, 300 OC (O,),and 480 O C (0,) along with the corresponding FT-ESEEM spectra. The three spectra are similar, showing a doublet at 2.3 and 5.6 MHz, two weak lines at 3.5 and 4 MHz, and an additional signal at 8 MHz. Note that T was chosen to enhance the doublet and reduce the intensity of the 3.5-MHz line. The splitting of the doublet is somewhat larger than for species A in Cu-NaX. The origin of the peak at 8 MHz, which is observed in both Cu-NaX and
3104 The Journal of Physical Chemistry, Vol. 96, No. 7, 1992
Matar and Goldfarb
b
A0.0 0
I
2
T (,us)
3
t
I
0
5
I
1
IO
25OC
I
1
15
Frequency (MHz)
Figure 4. (a) Three-pulse ESEEM of Cu-NaY dehydrated at various temperatures recorded at the g, position (H = 3150 G ,T = 0.22 ps). (b) The corresponding FT-ESEEM spectra.
Cu-KX is not clear, it could be a combination line due to the presence of several equivalent first shell A1 nuclei or it may correspond to a double quantum transition. The intensity of this line decreases with increasing dehydration temperature. Although the FT-ESEEM spectra of the three samples are very similar, the time domain traces show an interesting trend, the modulation depth decreases significantly with dehydration temperature. We attribute this decrease to an increase in the quadrupole coupling constant of the framework Al. As the dehydration temperature increases, more water molecules are removed and the framework undergoes additional distortions resulting in larger field gradients at the A1 nuclei.2' This causes a significant broadening of the majority of the ENDOR transitions, thus reducing the modulation depth. It could be that the changes in the intensity of the 8-MHz line are related to this phenomenon as well. A similar behavior is evident in Cu-Nay. Dehydration at 25 "C generates species F which according to the EPR spectra persists all the way up to dehydration at 400 "C (0,)(see Table I). The time and frequency domain ESEEM of these samples are shown in Figure 4. All spectra exhibit a doublet and a line at 3.5 MHz, again T was chosen such that the doublet intensity is maximized. The intensity of the q line decreases with increasing dehydration temperature and so does the modulation depth. Furthermore, the doublet splitting changes from 2.7 MHz in Cu-NaY (25 "C) to 2.5 MHz in Cu-NaY (400 OC, O,), and a shift in the doublet center is observed as well. As in Cu-KX we attribute these changes to the increase in the quadrupole interaction. Note the shallower modulation in Cu-NaY as compared to Cu-KX due to the lower A1 content. The EPR parameters of the Cu2+species in Cu-NaA (400 OC, 0,) are rather similar to those in Cu-Nay, although in the CuNaY spectrum the lines are broader. This broadening may be an indication for some site distribution in Cu-Nay. The FTESEEM spectra of dehydrated Cu-NaA show a doublet at 2.5 and 5.1 MHz and additional three lines (see top trace, Figure 6) at 3, 3.5, and 4 MHz. The lines a t 3 and 4 MHz are most prominent a t 7 = 0.220 ps.I3 Orientation Selective Experiments. In order to obtain additional information concerning the geometry of the Cu2+ sites in the dehydrated zeolites, we carried out orientation selective ESEEM experiments. In those measurements one takes advantage of the large g-anisotropy of the Cu2+cation and the limited bandwidth of the microwave pulses to excite only crystallites with selected orientation~.~~.*~J~ The ESEEM are recorded at different resonant
magnetic fields along the EPR powder pattern. The selected magnetic field determines the orientation of the crystallites that are affected by the pulses and contribute to the echo. For instance, setting the magnetic field at the lowest possible field position, namely the All feature of the Mf"= 3/2, causes the excitation of only those crystallites in which the principal axis of the g tensor, z , is parallel to external magnetic field, Ho(0, = 0"). Setting the magnetic field at the g, position will result in the excitation of a much larger distribution of orientations, namely 0, z 60-90" and cpo = 0-180" (due to contributions from all Mi's of the Cu nuclei, I = 3/2). cpo is the azimuthal angle of Ho in the principal frame of the g tensor. Figure 5a shows the two-pulse FT-ESEEM spectra of dehydrated Cu-NaX (400 OC, 0,) recorded at different positions along the EPR powder pattern. In Figures 5-7 the spectra are plotted such that they are all normalized to the most intense line. Their relative intensities are indicated in parentheses to enable the comparison of their absolute intensities. The spectrum obtained at 315OG (the g, position) shows both the hyperfine doublet and the wIline. As the magnetic field decreases and becomes closer to the gl,position, the relative intensity of the doublet decreases and the Larmor frequency line dominates the spectrum. The spectrum recorded at 3 150G shows also two combination lines at 7 and 7.8 MHz. The first appears at twice the Larmor frequency and the second corresponds to the sum frequency, ua u p , of the hyperfine doublet. Both lines are 180° out of phase with respect to the basic frequencies as expected.20 The 2uIline appears in all spectra and its frequency decreases with the magnetic field as expected, whereas the other combination line is not evident in spectra recorded below 3 100 G. The reduction of the relative intensity of the doublet with respect to the Larmor frequency line is even more pronounced in the three pulse. ESEEM as shown in Figure 5b. Although the ESEEM of species D in Cu-NaA and species F in Cu-NaY exhibit frequencies similar to those of species A in Cu-NaX, their orientation dependence is significantly different. Figure 6a,b shows the dependence of the two- and three-pulse FT-ESEEM spectra of species D on the field position within the EPR lineshape. In both sets of spectra the doublet lines are evident at all fields and the intensity variation of the spectra is relatively
+
(24) Reijerse, E. J.; van Earle N . A. J.; Keijzers, C. P. J. J . Mugn. Reson. 1986, 67, 114. (25) Flanagan, H . L.; Gerfen, G. J.; Lai, A.; Singel, D. J. J . Chem. Phys. 1987.88, 2162.
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992 3105
Cu2+ Interactions with Zeolites
A
3150 C
3100 C
A
A
3000 C
A
2900 G
2900 G
(1.6)
2800 G
I
I
I
I
I
5
0
2800 G
I
I
15
10
I
I
0
Frequency (MHz)
I
I
I
5
1
1
15
10
Frequency (MHz)
Figure 5. Two- (a) and three-pulse (b) FT-ESEEM spectra of Cu-NaX dehydrated at (400 O C , 0,) as a function of the resonant magnetic field ( T = 0.22 rs). The numbers in parentheses give the relative intensity of each spectrum.
b
a
s (1.0)
3150
W? $*? /A 5000
(0.J
\
2700 C
(1.0)
I
c
2900 G
2800 G
0
C
I
I
5
I
10
I
I
15
Frequency (MHz)
I
0
I
I
5
I
I
10
,
I
15
Frequency (MHz)
Figure 6. Two- (a) and three-pulse (b) FT-ESEEM spectra of Cu-NaA dehydrated at (400 O C , 0,) (species D), as function of the resonant magnetic field ( T = 0.30 1s). The numbers in parentheses give the relative intensity of each spectrum.
small. In the two-pulse spectra the line at 2.6 MHz is often weak and appears as a shoulder; the 20, peak is clear in all spectra and exhibits the expected field dependence. The low-frequency line of the doublet is more intense in the three-pulse spectra and its frequency shows a dependence similar to that of the wI lines. The orientation selective FT-ESEEM spectra of species F in Cu-NaY are shown in Figure 7. The doublet is evident in all resonant magnetic fields; moreover, in the two-pulse FT-ESEEM spectra the doublet is better resolved at lower fields. It should however be noted that due to the shallow modulation and the relatively large deadtime the absolute intensities of the two-pulse spectra are less reliable. The relative intensity of the wIpeak in both sets of spectra is considerably reduced as compared to CuNaA and the 2wI line is not detected in the two pulse FT-ESEEM spectra. Note that while the high field spectra of both Cu-NaX
and Cu-NaA show high frequency lines at 2 8 MHz, no such lines are present in Cu-Nay. This supports the assignment of these lines to combination lines arising from the presence of several interacting AI nuclei. In Cu-NaY the Si/Al is larger by a factor of about 2 with respect to Cu-NaA and Cu-NaX; hence these lines should have low intensities. In general we note that in all three samples studied the modulation depth of all two-pulse traces increased with increasing magnetic field within the EPR powder pattern. We explain the field dependence of the doublet intensity in terms of the anisotropy of the modulation amplitudes and the site geometry. The modulation amplitudes are determined by the probabilities of the allowed and forbidden EPR transitions connecting the nuclear spin manifolds and they strongly depend on the orientation of the external magnetic field.*O For instance, they
3106
Matar and Goldfarb
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992
a
r
0
I
I
5
I
b
I
I
1
15
10
(1.4)
3000 G
(1.5)
2900 G
I
I
I
I
I
5
0
I
1
15
10
Frequency (MHz)
Frequency (MHz)
Figure 7. Two- (a) and three-pulse (b) FT-ESEEM spectra of Cu-NaY (400 OC, 0,) (species F), recorded as a function of the resonant magnetic field
(T
= 0.22 ps). The numbers in parentheses give the relative intensity of each spectrum.
0531
0 2"
Si
180
60
8,=IO'
080
60
00
$0
/3
Q.90 MHz
j
Figure 8. A schematic description of a Cu2+ site in zeolites X, Y, and A involving a six ring.
decrease to zero when the magnetic field coincides with one of the canonical orientations of the superhyperfine tensor (when the nuclear quadrupole interaction is negligible). Many of the cation sites in zeolites X, Y, and A involve the hexagonal ring and can be schematically represented as shown in Figure 8. This illustration also shows the principal axes of the tensors involved. We assume that the principal axis of the g tensor, z, coincides with the local symmetry axis and that the anisotropic hyperfine interaction is mostly dipolar in nature such that its principal axis is approximately parallel to the Cu2+-A1 axis and the angle it makes with the z axis is 4. In species A the intensity of the doublet decreases as Bo approaches 0, which implies that at this orientation the magnetic field is along one of the canonical orientations of the anisotropic hyperfine tensor for all first shell A1 nuclei. This is possibly only if OI = 90°, namely, the Cu2+ cation is situated close to the plane of the A1 nuclei, such that at O0 = 0' none of the first shell A1 will contribute to the modulation. When the A1 quadrupole interaction is not negligible, forbidden EPR transitions can be induced also by the simultaneous presence of the isotropic hyperfine and the nuclear quadrupole interactions such that the modulation amplitudes will depend also on the orientation of the quadrupole tensor with respect to the external magnetic field.26 This is illustrated in Figure 9 where the (26) Flanagan, H. L.; Singel, D. J. J . Chem. Phys. 1987,87, 5606.
180
180
t?,=?O* XP/2;,/2
0=90MHz
e,= 700
0 80
053 027
000 180
Figure 9. The dependence of x;j2-,/, on cpo and p for different values of
eo, of the quadrupole coupling constants (e2qQ/h= Q),and of O1 as listed
on the figure. Other parameters used: ai, = 2.6 MHz, r = 0.32 nm, 9 = (Y = y = 0. For Oo = loo, Ho= 2800 G and for 6, = 90°,Ho= 3150G. The same absolute scale is used for all plots. (ai, is related to Ai, in ref 13 by aiso Aisdiso).
-
modulation amplitude of the 11/2) transition within the manifold, x7/2,-li2,20 is plotted as function of cpo and @ for two field orientations, 6, = 90° and 60 = loo. cy, @, and y are the Euler angles relating the quadrupole tensor principal axes system and the g tensor principal axes system and 7 is the asymmetry pawas calculated using rameter of the quadrupole tensor. x7/2,-I12
cy
The Journal of Physical Chemistry, Vol. 96, No. 7, 1992 3107
Cu2+ Interactions with Zeolites
IOOG
Cu-NaA
Cu- NaX 311
DPPH
1
b
I/-
Y
!
(b) after adsorption of methanol (to Figure 10. ESR spectra recorded at 110 K of Cu-NaX and Cu-NaA: (a) dehydrated samples (400 "C, 02); saturation).
exact diagonalization of the nuclear Hamiltonian as described el~ewhere.~ We chose to plot the modulation amplitudes as a function of cpo since when we perform the orientation selective experiment we select only a range of 0, but not of cpo as the g tensor is axially symmetric. Accordingly, the signal includes contribution from all possible cpo values. The top plots are for the case where 0, = 90' and a relatively small quadrupole coupling constant, e2qQ/h = 3.0 MHz. In this case, for Bo = 10' there is practically no modulation for any value of 0 or (0,. The middle plots correspond to 0, = 90' and 2 q Q / h = 9.0 MHz. There is a significant decrease in the modulation amplitude for Bo = 10' as compared to Bo = 90' for all 0 values, although at about 45' and 135' the amplitude is appreciable. The bottom plots were obtained for the same parameters with B1 = 70'. Note that there is not a significant difference in the modulation amplitudes between the two orientations. We chose to show only the amplitude of the 11/2) 1-1/2) ENDOR transition since it is expected to have the largest contribution to the ESEEM spectrum when the quadrupole coupling is large. The 11/2) 1-1/2) modulation amplitude within the 0 manifold showed a behavior similar to that of the a manifold. The signal of the distant Al does not show any field dependence, because, unlike the first shell Al, their arrangement around the Cu2+approaches a spherical distribution; thus at any field orientation there will be enough A1 nuclei to induce significant modulation at the Larmor frequency. Consequently, this line will not be suppressed as the g,,position is approached. On the contrary, it should be enhanced as at lower fields where the Zeeman interaction is smaller, the transition probability of the forbidden transition increases. Since the doublet corresponding to first shell AI in species D and F in NaA and NaY did not disappear at low fields, we suggest that in these sites the Cu2+is situated further away from the plane of the AI nuclei. Adsorption of Methanol. The Cu2+cation in dehydrated CuNaX, Cu-NaA, and Cu-NaY reacts differently toward the adsorption of methanol. In Cu-NaX the EPR features corresponding to species A do not change while those attributed to species A' disappear as shown in Figure 10. This indicates that methanol adsorption does not cause a migration of the Cuz+ in species A and that methanol molecules do not coordinate to the Cu2+ cation.I9 In Cu-NaA and Cu-Nay, however, methanol adsorption does alter the Cu2+environment as evident by the change in EPR parameters. In Cu-NaA only one major Cu2+species, G (Figure lo), is generated, whereas in Cu-NaY several species coexist; hence we did not perform any further measurements on the CuNaY/methanol. The three-pulse ESEEM along with the corresponding FTESEEM spectra of species G, recorded at the g, position as a function of T , are shown in Figure 1 1. The spectra consist of three major lines at 1.5, 3.5,and 6.1 MHz. We assign the lines -+
-
at 1.5 and 6.1 MHz to a hyperfine doublet split by an isotropic constant of 4.6 f 0.2 MHz. As in the other samples, the doublet is attributed to first shell Al while the 3.5-MHz line is attributed to distant Al. The field dependence characteristics of the ESEEM of this species are very similar to those of species A as shown in Figure 12. As the resonant magnetic field approaches the g,l position, the relative intensity of the doublet decreases and the Larmor frequency line becomes dominant. This suggests that in species G the Cu2+is also situated close to the plane of the first shell Al. The amount of adsorbed methanol in zeolites was found, by N M R measurements, to affect the 27Al quadrupole coupling constant. The higher the methanol content is, the lower the quadrupole coupling constant.2' Since in Cu-NaX (species A) methanol adsorption does not alter the location and coordination of the Cu2+,we can use this sample to verify experimentally the effects of the 27Alnuclear quadrupole interaction on the ESEEM pattern. These have been discussed in a number of publications but have not been demonstrated e~perime.ntally.~~ Figure 13 shows that as the amount of the adsorbed methanol increases, the relative intensity of the doublet decreases, and at saturation the 3.5-MHz line dominates the spectrum. Note also that the intensity of the combination line at 2wI increases while that at -8 MHz decreases. We attribute the increase in the relative intensity of the Larmor frequency line, as compared to the doublet intensity, to a decrease in the quadrupole coupling constant of the distant Al. This leads to a decrease in the modulation damping caused by the quadrupole interaction and accordingly to a more intense signal in the FT-ESEEM spectrum. The quadrupole coupling constant of the first shell AI, on the other hand, did not vary significantly, leaving the doublet intensity relatively weak. Discussion Two types of A1 nuclei, differing in their mode of interaction with the paramagnetic cation, were observed in all zeolites investigated. The first type is the first shell A1 nuclei which consists of the AI nuclei located at a distance range of 0.31-0.34 nm from the Cu2+cation, depending on the particular site involved.2 The second type, distant Al, includes AI nuclei situated at distances above 0.5 nm. The distant AI nuclei cause modulation at the Larmor frequency because they are coupled by weak dipolar interactions, whereas the first shell A1 nuclei generate different modulation frequencies as their isotropic hyperfine constants differ from zero. The isotropic hyperfine constants found for the species investigated in this study are summarized in Table 111. It is not, however, necessarily true that in every site first shell AI has a large (27) Romanelli, M.; Goldfarb, D.; Kevan, L. Mugn. Reson. Reu. 1988, 13, 179 and references therein.
3108 The Journal of Physical Chemistry, Vol. 96, No. 7, 1992
Matar and Goldfarb
I
0
I
2
3 T
4
(ps)
5
I
I
0
5
I
10
I
15
Frequency (MHz)
Figure 11. Three-pulse time domain (a) and frequency domain (b) ESEEM of Cu-NaA (400 "C, 0,) and then saturated with methanol (species G), as function of 7 . (The spectra were recorded at 3150 G).
TABLE 111: Isotropic Hyperfine Constants of First Shell nAl for Various Species Investigated zeolites Cu-NaX
CU-KX Cu-NaA Cu-NaA( MeOH) Cu-NaY
species A A
ai, (k0.2),MHz
D
2.6 4.5 2.5
G
F
2.1
3.3
hyperfine coupling constant. For example, in Cu-NaX dehydrated between 25 and 200 "C,where the Cu2+is located at site I in the hexagonal prism, the first shell Al exhibits only dipolar interactions and no isotropic hyperfine intera~tion.'~ In most cases reported in the literature (not involving zeolites) where A1 modulation was detected, only the Larmor frequency line has been o b ~ e r v e d . ~Several * ~ ~ ~ ENDOR studies report on Al nuclei with a nonzero isotropic hyperfine constant. One of these concerns the complexation of quinone radicals with Lewis acid centers in H-Y zeolite30 and another describes the interaction of
Fe3+and Cr3+ with neighboring Al in single crystals of Fe:LaA103 and Cr:LaA103.31 The latter is more relevant to our case since the interaction with the Al is via oxygens. The isotropic hyperfine constants found were 3.26 and 1.11 MHz for Fe:LaAlO, and Cr:LaA103, respectively. The authors attributed the hyperfine interaction with the A1 nuclei to an appreciable transfer of unpaired spin due to covalency.31 In the zeolites, in particular when no better ligands are available within the pores, the Cu2+cation coordinates to framework oxygens. The relatively large isotropic hyperfine observed in the Cu2+ species suggests that the coordination has some covalent character leading to a finite spin density on the A1 nucleus. We also observed some variation in the size of the isotropic hyperfine constant which may be related to differences in the site geometry. This affects the Cu2+orbitals configuration and their overlap with the oxygen orbitals. This is most pronounced in Cu-NaA with adsorbed methanol where the largest hyperfine constant has been observed. In this case it was found that the Cuz+ is coordinated to the framework oxygens and to a single methanol molecule,32 thus in this site the symmetry is different than in the fully dehydrated samples. Although all Cu2+sites investigated exhibit a large *'Al isotropic hperfine constant, different behavior was found with respect to the orientation selective experiments. Species A and G showed similar orientation dependences suggesting that the Cu2+is situated rather close to the plane of the first shell A1 nuclei. The lack of orientation dependence of the FT-ESEEM spectra of species D and F indicates that in these sites the Cu2+is located away from the plane of first shell Al. In all cases the distant A1 showed an isotropic behavior as their arrangement around the Cu2+can be
(28) Cosgrove, S. A.; Singel, D. J. J . Phys. Chem. 1990, 94, 2619. (29) Karthein, R.; Motschi, H.; Schweiger, A.; Ibric, S.;Sulzberger, B.; Stumm, W. Inorg. Chem. 1991, 30, 1606. (30) Astashkin, A. V.; Samoilova, R. I. Zeolites, 1991, 11, 282.
(31) Taylor, D. R.; Owen, J.; Wanklyn, B. M. J. Phys. C, Solid State Phys. 1973, 2592, 1973. (32) Ichikawa, T.; Kevan, L. J. Am. Chem. Sac. 1981, 103, 5355.