6452
J . Phys. Chem. 1986, 90, 6452-6459
Electron Spin Resonance and Electron Spin Echo Study of Cu2+ in Zeolites H-rho and CsH-rho Michael W. Anderson and Larry Kevan* Department of Chemistry, University of Houston, Houston, Texas 77004 (Received: June 20, 1986)
Electron spin resonance (ESR) and electron spin echo (ESE) spectroscopiesare used to determine the site location and adsorbate interaction of Cu2+ in Cu*+-doped CsH-rho zeolite. Particular consideration is given to the problem of partial excitation of an ESR transition in the calculation of the echo modulation produced by Cs+ cations. This is necessary because of the nonuniform distribution of Cs+ cations about the paramagnetic center, which results in only one or two Cs' cations contributing most significantly to the modulation. When there are such a small number of interacting nuclei at a specific orientation, the commonly used spherical-averaging approach becomes less valid. Copper(I1) is found to strongly favor sites in the a-cage rather than in the octagonal prism. In hydrated zeolite rho Cu2+interacts with two molecules of water and with either two or three zeolite oxygens. After dehydration the Cu2+ remains in the a-cage partly in six-ring sites and partly in four-ring sites. Adsorbate interactions with two molecules of methanol and one molecule of ethylene are observed. Analysis of the coordination with methanol suggests that some nearby Cs' are displaced from their crystallographic sites due to interaction with methanol. Similarities between Cu2' ESR spectra in zeolites rho and ZK4 corroborate the assignment of Cu2+in hydrated ZK4 to the a-cage.
Introduction Zeolites exchanged with transition metals are active for a wide variety of catalytic reactions.] The nature of that catalytic activity depends both on the location of the metal and on its interaction with adsorbates. In the case of paramagnetic species, such as Cu2+, much information concerning both these properties may be gleaned from electron spin resonance (ESR) and electron spin echo (ESE) spectroscopy.2-6 ESR gives information about the symmetry of a complex while ESE gives more quantitative information concerning coordination numbers and positions of adsorbates relative to the paramagnetic Copper(11)-exchanged zeolites have been used for a variety of reaction^,^-'^ including oxidation of propylene, cyclodimerization of butadiene, and cumene cracking. In most of the reported cases the zeolite employed has had the faujasite structure as in zeolites X and Y . However, recent work16 has also shown Cu2+-exchanged A-type zeolites to be active for propylene oxidation. In the present work Cu2+-doped zeolite rho is studied by ESR and ESE. Structurally, zeolite rho has similarities to zeolite A in that both consist of 3-dimensional networks of cages accessed through eight rings and both contain the a-cage as one of their structural building units. Because of these similarities a certain degree of comparison is possible between this study and a previous study on A-type zeolites." Cesium modulation from Cs+ exchanged into zeolite rho is analyzed to determine the crystallographic siting of Cu2+while deuterium modulation from deuteriated adsorbates is analyzed
to determine the immediate Cu2+ adsorbate environment. The results indicate a strong affinity of Cu2+ for sites in the a-cage rather than in the octagonal prism.
( I ) Maxwell, I. E. Ada. Catal. 1983, 31. 1. (2) Ichikawa, T,; Kevan, L. J . Am. Chem. Sac. 1983. 105, 402. (3) Narayana, M.; Kevan, L. J . Chem. Phys. 1983, 78, 3573. (4) Ichikawa, T.; Kevan, L . J . Phys. Chem. 1983, 87. 4433. (5) Herman, R. G.; Flentge, D. R. J . Phys. Chem. 1978, 82, 720. (6) Herman, R. G. Inorg. Chem. 1979, 18, 995. (7) Kevan. L. In Time Domain Electron Spin Resonance; Kevan, L., Schwartz. R. N., Eds.; Wiley-Interscience: New York, 1979; Chapter 8. (8) Kevan. L.: Narayana, M. Inrrazeolite Chemistry; American Chemical Society: Washington, D.C., 1983; ACS Symp. Ser. No, 218, p 283. (9) Mochida, I.: Hayata. S.: Kato, A.: Seiyama, T. J . Catal. 1970. 19, 405. (10) Tsuruya. S.; Tsukamoto, M.; Watanabe, M.: Masai. M. J . Catal. 1985, 93, 303. ( 1 1) Benn, F. R.: Dwyer, J.; Estahami, A.; Evmerides, N .P.; Srczepura, A. K. J . Catal. 1977, 48, 60. (12) Naccache, C. M.; Taarit, Y . B. J . C a r d . 1971, 22. 171 (13) Maxwell, I . E.; Downing, R. S.: van Langen. S. A . J . J . Catal. 1980,
Experimental Section Zeolite rho was prepared by an adaptation of the methods of Robson et al.'*.l9 N a O H (12 g) was combined with 12.9 g of H 2 0 . T o this was added, with warming to 70 OC, 7.2 g of 70.6 wt % AI20,-29.4 wt % H 2 0 (Catapal). This mixture was then cooled and combined with a mixture of 12.3 g of 50 wt % CsOH solution in water and 6.9 g of H 2 0 . Finally, 100 g of 3 wt % colloidal silica was added and the gel mixed until smooth. After 2-h incubation at 25 "C the gel was heated in a sealed Teflon container to 80 OC for 25 days to yield zeolite rho. Crystallinity and purity of the product were monitored by X-ray diffraction using a Phillips diffractometer scanning a t 1 O of 26' per minute and the results were compared to the data of Robson et al.'* The resulting CsNa cationic form of zeolite rho was exchanged 3 times with a 20% solution of N H 4 N 0 3followed by calcination at 300 "C in air to give the protonic form. Three samples were then exchanged with different levels of Cs' as follows: (i) 5 g of H-rho was exchanged with 30 mL of 5 M CsNO, at 75 OC for 16 h, (ii) 3 g of H-rho was exchanged with 50 mL of 0.1 M CsOH, (iii) 1 g of preparation ii was then further exchanged with 50 mL of 0.1 M CsOH. All samples, including the H-rho, were doped with Cu2+by exchanging with 10 mL of M Cu(NO,), and 100 mL of H 2 0 per gram of zeolite. This amounted to an exchange of approximately one Cu2+cation for every 40 a-cages. Finally the zeolites were washed with hot, triply distilled water and dried in air at room temperature. A zeolite prepared in such a manner is termed "fresh". The Si/A1 ratio and the cation compositions were analyzed by commercial atomic absorption analysis. The H-rho was repeatedly exchanged with D 2 0 in a vacuum desiccator before transferring under nitrogen to a 3-mm-0.d. Suprasil quartz sample tube that was then sealed. Such a treatment was not necessary in the case of Cs+-exchanged zeolites since Cs and not deuterium modulation was observed by ESE. Samples were dehydrated under vacuum to an ultimate pressure of Torr followed by oxidation with 400 Torr of dry, highpurity oxygen. The oxygen was then pumped off for 30 min at 400 "C and the sample cooled to room temperature. Deuteriated adsorbates, such as D 2 0 ,C D 3 0 H , C H 3 0 D , and C2D,, were ob-
61, 485. (14) Tsutsumi, K.; Fuji, S.; Takahashi, H. J . Catal. 1977. 24, 146. (15) Dimitrov, C.; Leach, H. F. J . Catal. 1969, 14. 336. (16) Lee, H.; Kevan, L. J . Phys. Chem., in press. ( 1 7 ) Anderson. M. W.; Kevan. L. J . Phys. Chem. 1986. 90. 3206.
( 1 8) Robson, H . E.; Shoemaker, D. P.; Ogilvie. R. E.: Manor, P. C. Adc. Chem. Ser. 1973, No. 121, 106. (19) Robson, H. E. U S Patent 3904738, 1975.
0022-3654/86/2090-6452$01.50/0
0 1986 American Chemical Society
Cu2+ in Zeolites H-rho and CsH-rho
The Journal of Physical Chemistry, Vol. 90, No. 24, 1986 6453
tained from Stohler Isotope Chemicals and Aldrich and were used without further purification. Exposure to adsorbates was conducted at room temperature and at the saturated vapor pressure of the adsorbate except in the case of C2H4where the pressure was 100 Torr. ESR spectra were recorded at both room temperature and 77 K on a Varian E-4 spectrometer. ESE spectra were recorded at 4 K on a home-built spectrometer described elsewherezOJ and linked to a Nicolet 1280 computer. Three-pulse, stimulated echoes were measured as a function of the time T between the second and third pulses. In order to maximize the modulation from both deuterium and cesium, the time between the first two pulses, T , was kept between 0.28 and 0.30 ~ s A. complete description of both the theory and the methods of simulation is given el~ewhere.~ In order to eradicate two-pulse interference in the three-pulse echo envelope at times T = T and T = 21, a phase-cycling method was employed.22 The following sequence is used: [(000) ( I I I I O ) ] - [(IIIIII) (OOII)]. The second three-pulse experiment inverts the two-pulse glitches but not the three-pulse echo. Therefore addition of (000) to ( I I I I O ) removes the two-pulse glitches while retaining the three-pulse echo. The last two sets of three-pulse experiments are the inverse of the first two and their inclusion corrects for any base-line drift.
+
+
Theory As has already been indicated the main aspects of the theory and simulation of electron spin echo envelope modulation have been covered in detail e l ~ e w h e r e .Here ~ extra thought is given to the problem of partial excitation of an ESR transition exhibiting large g anisotropy. The total anisotropic ESR line width of an axially symmetric Cu2+species in zeolites is generally about 600 G. However, the microwave pulses for an ESE experiment, which are of 40-11s duration, cover only about 18 G at Ho = 3 150 G. If we adjust the field to irradiate in the g, region of the spectrum, as is usually the case, then using the relationship
g'= gl12 cos2 Bo
+ gL2sin2 Bo
-Y 91
Figure 1. Vector diagram showing relationship between external field H , and internuclear vector r for an ESE experiment. The diagram is drawn in the g-tensor reference frame.
cut from a sphere such that 4I is the azimuthal angle about z between the zx plane and the zHo plane. The annulus will cover the range Bo = Boc 6 to Bo = Boc - 6. In our case for irradiation at g,,,Boc = 90' and the deviation 6 = 4'. The average modulation will now be given by
+
( Vmod ) =
The echo intensity Vm"od(/?1,Bo,q5r) for the case of large g anisotropy has been calculated by Rowan et al.;23however, a mistake appears in their B'term (part of a coefficient for the hyperfine cross term SJJ. For clarity the complete corrected expressions for the modulation are given below. Exact analytical expressions for the echo modulation from hyperfine interaction with a nucleus with I = 7/2, such as 133Cs,in the absence of quadrupolar effects are taken from Dikanov et al.24
(1)
where Bo is the angle between the field Ho and the glldirection for Ho = 3150 f 9 G, we calculate Bo = 90' f 4'. The usual procedure for calculating the average spin echo modulation signal ( Vmod)s,where 0 is the angle between the electron nuclear vector r and H o is to take a spherical average over all orientations; thus where where q5 is an azimuthal angle about the z axis between the zr plane and the zHo plane. However, such an integration should only strictly be used in the case of isotropic g or where the whole ESR signal is being excited. It has been demonstrated, however, that even for axially symmetric paramagnetic species eq 2 is a good approximation as long as the number of interacting nuclei is large enough, typically four or greater, and has a uniform distribution about the paramagnetic center. Such is not the case for Cuz+-doped Cs-rho, where the Cs+ cations are located in specific crystallographic sites resulting in defined angles between the Cu2+-Cs+ electron nuclear vector and the field. Although the number of interacting Cs+ nuclei is greater than four, they are not all at the same distance from the unpaired electron. This results in less than four, sometimes one and sometimes two, nuclei making the greatest contribution to the modulation pattern. The situation under consideration is illustrated in Figure 1. Let us define a set of coordinates such that glllies along z and then choose x and y such that r lies in the zx plane at an angle 8, to the z axis. Then Ho may take up any orientation in an annulus
COS
cos
(:) (:) w, =
wo =
= 1 - 2K sin2
2
+ r)
= 1 - 2K sin2
[ (; + [ ($ -
2
( +( +
cos 4r)(cos 8, cos Or + sin Bo sin (20) Ichikawa, T.; Kevan, L.; Narayana, P. A. J . Phys. Chem. 1979, 83, 3378. (21) (a) Narayana, P. A.; Kevan, L. Phorochem. Photobiol. 1983, 37, 105. (b) Narayana, P. A.; Kevan, L. Magn. Reson. Rev. 1983, 1 , 234. (22) Fauth, J.-M.; Schweiger, A.; Baunschweiler, L.; Forrer, J.; Ernst, R. R. J . Magn. Reson. 1986, 66, 14.
+ T)
W,(T
up7
sin2 2 W,T
sin2 2
(5)
;)2]1i2
(7)
;)2]"2
I
cos d1)- 1 - 2IIa (8)
(23) Rowan, L. G.; Hahn, E. L.; Mims, W. B. Phys. Rec. 1965, 137, A61. (24) Dikanov, S. A,; Shubin, A. A,; Parmon, P. N. J . Magn. Reson. 1981, 42, 414.
6454
The Journal of Physical Chemistry, V o l . 90, No. 24, 1986 Bl = Bf2 +
B' = g z [ ( g,'sin
$..
cJ2
Anderson and Kevan
(9)
cos 8, cos o1 +
Bo sin BI cos G ~ ~ ) ( C8,Osin S O1 cos dI - sin 8, cos-8,) +
;y, y, y1 k
Averaging
m z W
(gl;2cos 8, cos Or
+ g,'
sin Bo sin Or cos 4,)(sin Or sin 41)
1
where a is the isotropic hyperfine coupling in hertz. It was found that varying 6 between 0' and IOo had negligible effect on the modulation calculated by eq 3 for all possible configurations of Cs' with respect to Cu2'. Therefore only one integration need be performed over @,, reducing eq 3 to 1 n (12) (vmod) = v m o d ( 8 1 ~ 8 0 ~ ~ I ) d@l
1 3 , E 0
5
5
0
T, PS
T, PS
Figure 2. Variation in modulation with 8 , for CuZfinteracting with one Cs' at 0.45 nm, also= 0.
Calculating the modulation for one nucleus gives the modulation for n nuclei (VmJ by
I K o d j = { vmod)n
c
n
SPHERICAL AVERAGING
(13)
This is an approximation and assumes no correlation between the Cs nuclei. It is tantamount to assuming each nucleus to be evenly distributed around a ring of radius r sin O1. Figure 2 illustrates how the echo modulation changes as a function of BI for interaction with one Cs' nucleus a t 0.45 nm and for irradiation at g,. Figure 3 illustrates the necessity for this whole approach in the system under consideration. Two modulation envelopes are shown for Cu2+located in the same crystallographic site and interacting with IO Cs' cations. One has been calculated with the spherical model consisting of four concentric shells and the other with the model of specific geometries. The modulation depth differs by a factor of nearly 2. Cesium has a nuclear spin of 7 / 2 ; however, its quadrupole moment is small, ca. -3 X lo-*' e cm2. A quadrupole interaction of up to 1.2 MHz was introduced into the calculation of the echo modulation using spherical averaging. Introduction of this size quadrupolar term had a negligible effect on the modulation. Therefore, subsequent calculations using the approach outlined above were performed without a quadrupole interaction.
Results The zeolite rho synthesized was highly crystalline to X-rays and exhibited a unit cell dimension of 1.51 nm for the hydrated hydrogen form of the zeolite. Compositional data for all the cationic forms studied are given in Table I. The Si/A1 ratio of this zeolite was 3.4, which lies in the range given by R o b ~ o nof '~ 2.5-3.5 and which gives a total exchange capacity of 10.9 monovalent cations per unit cell, there being 48 tetrahedral lattice sites per unit cell. It was assumed that the charge balance, after determination of the Cs' content, was maintained by protons. The samples studied are designated in the text by CuCs,H-rho, where n is the approximate integral number of Cs+ cations per unit cell and C u indicates Cu2+ doping. ESR parameters at both room temperature and 77 K are given for a variety of sample preparations in Table 11. The hyperfine splitting is only given for the gllcomponent as the splitting in the g, region is generally unresolved. Assignment of species is usually based on the hyperfine "fingerprint" in the gl, region of the spectrum. Figure 4 shows the ESR spectra of CuCs,H-rho during dehydration measured a t 77 K. There is a gradual transition upon dehydration from a species with ESR parameters g, = 2.392 and
2-DIMENSIONAL AVERAGING I
8 2
3
4
5
T, PS
Figure 3. Comparison of ESE modulation depth for Cu2+located in an octagonal prism site (SI2)in CuCs,H-rho interacting with 10 Cs' calculated with (a) a spherical average and (b) a 2-dimensional average with the parameters given in Table 1V. Cu,Cs3, H -rho
x4
g1,=2392
9,,=2368 c
-
x4
20QG H
Figure 4. ESR spectra of CuCs3H-rho recorded at 7 7 K: (a) fresh sample, (b) dehydrated a t 120 "C, and (c) dehydrated at 400 "C. TABLE I: Composition of Zeolites Studied As Determined by Commercial Atomic Absorption Analysis sample composn/unit cell designatn in text H10.9(A102),0.9(Si02)~~ I H-rho Cs2.88H8.02(A~0~)~0.9(si02),7 I Cs,H-rho CS4.0H6.9(A102) 10.9(si02)37 I Cs,H-rho C S 6 2H,,(AI02),o9(Si0:),; I Cs6H-rho
Cu2+ in Zeolites H-rho and CsH-rho
rhe Journal of Physical Chemistry, Vol. 90, No. 24, 1986 6455 Cu, H -r ho
TABLE 11: ESR Parameters at 77 K of Cu2+in H-rho Zeolites 104 A~,F
sample
treatmentQ
CuCs,H-rho fresh freshd
evac rt evac 120 'C evac 180 O C evac 400 O C sat. CH,OH
+90 Torr NH3 +lo0 Torr
glib
cm-]
glb
2.392 2.165 k,J 2.372 (g,so) 2.384 2.368 2.326 2.324 2.329 1.998 2.338 2.391 2.335 2.244
146
2.060
145 145 147 171 171 173 78 163 130 120 183
2.359
165
2.389 2.392
125 146
I probable config
2.062 2.055 2.070
fresh evac 400 OC + 100 Torr
2.384 2.412 2.354 2.399
152 146 162 140
2.060
2.021 2.399
56 135
-
u-r_L--I
gll = 2 354
Y
200 G H
Figure 6. ESR spectra of CuH-rho recorded at 77 K: (a) fresh sample; (b) dehydrated at 120 OC; (c) dehydrated at 400 OC. 2.064 Cu,Csa,H- rho + NHx
CHIOH
+ 100 Torr
I
2.024
CHjCHzOH CuH-rho
x4
h
C2H4 sat.
Po., g,l = 2412
C2H4
Evac indicates evacuated at; sat. indicates saturated with. (I Estimated uncertainty is fO.OO1. cEstimated uncertainty is f l X cm-I. "ESR recorded at room temperature. Cu,Cs3, H-rho (ROOM TEMP)
b
' V giso=2.165
#\I'
EVACUATED ROOMTEMP
x4
x4
Figure 5. ESR spectra of CuCs3H-rho recorded at room temperature: (a) fresh sample; (b) evacuated at room temperature; (c) dehydrated at 40 "C. A l l = 146 X cm-' to another species with ESR parameters gll= 2.329 and All = 171 X cm-I at 120 "C. In the sample
fully dehydrated at 400 OC there is also an indication of a species exhibiting reversed g values, g, > gll,with gll = 1.998 and A l l = 78 X lo4 cm-I. Figure 5 shows the ESR spectra recorded at room temperature, which indicate a large isotropic signal with giso= 2.165. This isotropic signal disappears rapidly upon dehydration. The dehydration sequences for both CuCs,H-rho and CuCs6H-rho are almost identical with that for CuCs,H-rho, although the transition temperature between the various species is some 50 OC lower for both these samples. At the higher Cs' concentration the amount of Cu2+giving rise to an isotropic signal in the fresh material decreases. In all the zeolites studied the
\
\
-1 20G H
A=11.9G
Figure 7. (a) ESR spectra of CuCs3H-rho with adsorbed NH3 recorded at 77 K; (b) expansion of the g, region showing nitrogen hyperfine splitting. effects of dehydration are fully reversible; saturation with 22 Torr H 2 0 for 4 h completely regenerated the fresh sample. Upon dehydration of CuH-rho the major trends are the same as for the @-exchanged zeolites up to about 200 OC as shown in Figure 6. Above this temperature a single species is formed rather than the two observed in the CuCsH-rho. The ESR parameters gll = 2.354, All = 162 X cm-l, g, = 2.064 are, however, similar to those of one of the species observed in dehydrated CuCsH-rho. All the zeolites studied exhibited the same interaction with NH3, giving one species with an ESR spectrum as shown in Figure 7. The spectrum displays axial symmetry with ESR parameters gll = 2.244, A l l = 183 X cm-', and g, = 2.024. Also observed is hyperfine splitting for nitrogen of 12 G in the g, region. Such splitting is indicative of direct Cu2+-N interaction. Figure 8 shows the effect on the ESR spectra of adding methanol and ethylene to CuH-rho. In both cases a similar ESR spectrum is obtained with one major and one minor species. The ESR spectra of the Cs+-exchangedzeolites with adsorbed methanol and ethylene are not shown but exhibit all the species exhibited in CuH-rho. However, in these zeolites a certain amount of Cu2+ remains uncomplexed to the adsorbate with ESR parameters
6456
Anderson and Kevan
The Journal of Physical Chemistry, Vol. 90, No. 24, 1986 Cu,Cs4,H-rho FRESH
DPPH +CH,OH x4
3
Ih
EVACUATED ROOM TEMP N.4; r = 0 9 2 n m ; a.357' N.2; r = 0 8 6 n m ; a.294"
v g,,=2.399
g
Figure 8. ESR spectra of CuH-rho recorded at 77 K: (a) with adsorbed C H 3 0 H , (b) with adsorbed C2H5. EVACUATED ROOM TEMP.
z
1
2
REHYDRATED 2.5 TORR HzO
3
4
5
0
1
2
T, PS c
19
3
4
5
1
3
h
4
5
+CD30H N.6; r.0.36 nm, A=O.I MHz N.3,r.O 28nm,A=03MHz
E/ k
l'i 0
2
Cu,H-rho
Id
tCHfCHzOH
dl\
1
T, P S
T, PS
+90 TORR CpH4
o
Figure 10. Experimental (-) and simulated (---) three-pulse ESE modulation spectra of CuCs5H-rho showing Cs modulation: (a) fresh sample; (b) evacuated at room temperature.
Lb
0
l
I W V
2
3 4 T, PS
5
0
1
2
3 4 T, 11s
5
tr =0.26nm W
OLI
Ia
0
0
1
2 3 T, PS
4
I
5
Figure 9. Experimental (-) and simulated (---) three-pulse ESE modulation spectra of CuCs,H-rho showing Cs modulation: (a) evacuated a t room temperature, (b) rehydrated with 2.5 Torr H,O, (c) with adsorbed C2H5, (d) with adsorbed C H , C H 2 0 H , and (e) with adsorbed CH30H.
identical with those in the activated zeolite. Figure 9 shows the three-pulse ESE spectra and simulation for CuCs,H-rho with various adsorbates. The modulation observed is from Cs' cations. A comparison is made in Figure 10 of the Cs modulation exhibited by CuCs4H-rho in the fresh sample and after evacuation at room temperature. There is a marked decrease in the depth of modulation upon evacuation. Modulation from deuterium, from a sample of CuH-rho with adsorbed C D 3 0 H and C H 3 0 D , is shown in Figure 11. In the case of C D 3 0 H it is necessary to invoke a two-shell model of interacting deuteriums in order to explain the observed modulation pattern. For CD,OH the best fit is achieved with N = 6 and R = 0.36 nm in one shell and N = 3 and R = 0.28 nm in the other shell. In other words the Cu2+interacts with three molecules of CD,OH, two of which have the same distance from the Cu2+. In the case of C H 3 0 D interaction is only seen with two molecules with the deuterons 0.26 nm from the Cu*+, suggesting that the deuteron from the third interacting CH,OD molecule is too far away to contribute significantly to the modulation. Figure 12 shows the ESE spectrum for C2D, adsorbed on CuH-rho and indicates interaction with one molecule with a Cu2+-D distance of 0.36 nm.
Figure 11. Experimental (-) and simulated (---) three-pulse ESE modulation spectra of CuH-rho showing deuterium modulation: (a) with adsorbed C D 3 0 H ; (b) with adsorbed C H 3 0 D . Cu,H-rhotCpD4 N=4 r = 0.36 nm
0
I
2
3
4
5
T, PS
Figure 12. Experimental (-) and simulated (---) three-pulse ESE modulation spectra of CuH-rho with adsorbed C2D4showing deuterium modulation.
A comparison is made in Figure 13 between the deuterium modulation exhibited by CuCs3H-rho in different states of hy-
Cu2+ in Zeolites H-rho and CsH-rho
The Journal of Physical Chemistry, Vol. 90, No. 24, 1986 6457
Cu,CsS,H-rho
TABLE 111: Results from ESEM Spectral Simulations type of modulasample treatmenta tion results CuCs,H-rho evac rt Cs Cu2+in a-cage +2.5 Torr H,O Cs Cu2+ in a-cage +90 Torr C2H4 Cs Cu2+in a-cage sat. CHICHIOH Cs some Cs displaced, Cu2+in a-cage +90 Torr CH30H Cs some Cs displaced, Cu2+in a-cage CuCs,H-rho fresh Cs Cu2+in four ring of a-cage evac rt Cs Cu2+in four ring of a-cage modulasample treatment" tion r, nm N also,MHz CuH-rho +110 Torr CD,OH D 0.36 6 0.1 0.28 3 0.3 +110 Torr CH30D D 0.26 2 0.3 + 110 Torr C2D4 D 0.36 4 0.0 CuCs3H-rho fresh (+D20) D 0.29 4 0.2 evac rt D 0.28 4 0.2 +22 Torr D20 D 0.28 4 0.2
+D20 (FRESH) r = 0.29 nm A.0.2 MHz
0
1
2
3
4
5
m
-a> E
t cn z W
z" I CZ
iLI,0
1
2
3 T, PS
4
5
r=O28nm A=02MHz
0
1
2
3
4
5
Figure 13. Experimental (-) and simulated (---)
three-pulse ESE modulation spectra of CuCs3H-rho showing deuterium modulation: (a) fresh sample (deuteriated); (b) evacuated at room temperature; (c) rehydrated with D 2 0 after evacuation at 400 OC.
Figure 14. Crystal structure of zeolite rho showing cation positions. Site S2 is at the center of a six-ring face with site S2* displaced slightly along the triad axis into the a-cage. Site S3 is adjacent to the four ring in the a-cage. Sites S1' and S12 are adjacent to four rings in the octagonal prism and site S5 is at the center of the octagonal prism.
dration: first a fresh, deuteriated sample, second the same sample evacuated at room temperature, and finally a sample that had been fully dehydrated and then rehydrated by exposing to 22 torr D 2 0 . Despite the simularities of the ESR spectra of these samples, they all give slightly different ESE spectra. In these samples it was first verified that the Cs modulation was negligible from similar but nondeuteriated samples. A list of all ESE deuterium modulation simulation parameters is given in Table 111. For the Cs modulation a comment is made about the Cu2' site location and whether Cs' are displaced rather than giving a list of interaction parameters.
Discussion Structure and Data Analysis. The structure of zeolite rho is shown in Figure 14.18319It has a simple cubic primitive cell with lattice parameter of 1.5 nm. The secondary building units of which zeolite rho is comprised are a-cages and octagonal prisms. These a-cages, which are comprised of eight, six, and four rings, are the same 26-hedra that exist in zeolite A; however, in that structure they are joined together via single eight rings to encompass fl-cages (14-hedra). In zeolite rho each a-cage is joined by a double eight ring (or an octagonal prism) to six other a-cages such that each unit cell is comprised of two a-cages and six octagonal prisms. Such an arrangement of cages leads to two identical but intertwined systems of three-dimensional tunnels and cages accessed via distorted eight rings of free diameter of ca. 0.39 X 0.51 nm.
"Evac indicates evacuated at; sat. indicates saturated with; rt is room temperature. Therefore, unlike in the faujasite and zeolite A structures, there are no portions of the structure accessible only through six rings with free diameter of 0.22 nm. These characteristics make rho attractive for good catalytic potential. In the case of zeolites A, X, or Y exchanged with Cu2' the catalytic activity is often low due to Cu2+locating in sites inaccessible to adsorbate molecules. This situation cannot occur in zeolite rho since all sites for Cu2+ are equally accessible to adsorbate molecules. A crystal structure determination has been performed on the cesium/hydrogen form of zeolite rho by Parise and Prince.25 Their sample of zeolite rho had a %/A1 ratio of 3.7, similar to that in the present study, and was exchanged to a content of six Cs' cations per unit cell. They determined that all the Cs" cations were located at the partial atomic coordinates ( I / * , 0,O) or in other words at the center of every octagonal prism, site S5 as shown in Figure 14. Because of the tendency for Cs+ siting in these S5 octagonal prism sites it is assumed in the present work that the Cs' location in rho with less than six Cs' cations per unit cell is also in the octagonal prism. The siting of protons in the rho zeolite is unknown. Treatment of ESE data with Cs modulation is somewhat different from the treatment used in previous work' for deuterium modulation. The procedure with deuterium is to obtain a best fit to the modulation by varying the numbers of interacting dueterons and their distance from the paramagnetic center. In this case the number of interacting cesium cations is known as well as their distribution. Therefore, the modulation expected for Cu2+located in various crystallographic sites may be calculated. Before this is done, however, the orientation of the principal g tensor must be determined in order to calculate the angle 8, as given in eq 2. There are five reasonable crystallographic sites available to accomodate Cu". These are shown in Figure 14 and are defined as follows: (i) site S2, in the center of the six ring of the a-cage; (ii) S2*, displaced from the six ring into the a-cage; (iii) site S3, in the a-cage close to an oxygen four ring; (iv) site Sl', in the octagonal prism close to an oxygen four ring at partial atomic coordinates approximately 0, 0.3); (v) site Sl', also near a four ring in the octagonal prism but at partial atomic coordinates (0.36, 0,0.36). For trigonal, tetrahedral, or octahedral coordinated Cu2+located in sites S2 and S2*, the principal g tensor should lie along the triad axis passing through the center of the six ring with vector coordinate [ l , 1, 11. For square-based pyramidal coordination the principal g tensor will lie closer to vector (25) Parise, J. B.; Prince, E. Mater. Res. Bull. 1983,18, 841.
6458
The Journal of Physical Chemistry, Vol, 90, No. 24, 1986
TABLE IV: Parameters for Fitting Cs Modulation in CuCsH-rho Zeolite Cu2+ site g, direction N r, nm O,, deg s2 11, 1. 11 6 0.65 70.5 s2* [O, I , 01 2 0.70 50.0 1 0.70 65.0 2 0.62 61 .O 1 0.62 43.0 s2* [ I . 1. I ] 3 0.70 60.5 3 0.62 81.9 s3 [ I , 0%01 4 0.92 35.7 2 0.86 29.4 I 0.64 90 2 0.54 90 SI' [ I , 0,01 2 0.87 90 2 0.81 21.7 1 0.45 90 SI2 [-I, 0,11 2 1.07 90.0 2 0.98 57.3 2 0.95 55.9 2 0.58 23.6
Anderson and Kevan
0
0
I
3
2
4
5
T, PS
Figure 15. Theoretical ESE modulation spectra showing Cs modulation expected for Cuz+ in CuCs,H-rho with Cu2+ located at (a) site S2 or S2*, g, along [ I , I , I ] , (b) site S2*, g, along [0, I , 01, (c) site S3, (d) site SI', and (e) site S12.
coordinate [0, I , 01. In this cation location the Cu2+ will be coordinated to three lattice oxygens. For Cu2+ in site S3 the coordination will also be between 3- and 6-fold; however, only two ligands will be zeolitic oxygen. For 3- and 4-coordinate Cu2+the principal g tensor should lie along [ 1, 0, 01, but for 5 and 6 coordinate it is difficult to determine its direction. Similarly for Cu2+in sites SI' and S12 the principal g tensor axis for 3- and 4-coordinate Cu2+ should lie along [ 1, 0, 01 and [-I, 0, I ] , respectively. For the sake of simplicity we may consider the Cs+ cations arranged on all the faces and along all the edges of a cube with an edge length of 1.5 nm and then, with Cu2+located at the partial atomic coordinates given above, the necessary parameters for simulation of the Cs modulation may be calculated. Table I V gives the interaction distance and number of Cs+ cations along with the respective values of 8, for Cu2+located at the various crystallographic sites with six Cs+ cations per unit cell. For all simulations the value of the isotropic hyperfine interaction, also,has been taken as zero owing to the large distance between Cu2+and Cs'. Figure 15 shows the modulation patterns that we might expect for Cu2+in the various crystallographic sites with six Cs+ per unit cell. Unfortunately, there is little distinction between the modulation patterns for Cu*+located in sites S2, S2*, and S3, and this distinction becomes even less clear when a decay function is applied to the data. However, substantial differences are seen between Cu2+ located in the a-cage and Cu2+ located in the octagonal prism. The difference in these simulated spectra from the experimental data shown in Figures 9 and I O is due to the absence of a decay function. It should be noted that in the simulations calculated for Cu2+ in the octagonal prism it is assumed that the Cs+ occupying that same octagonal prism has been exchanged out. Otherwise an unfavorable Cu2+-Cs+ repulsive interaction would occur. ESR and ESE Data. Fresh samples of zeolite rho exhibit an isotropic ESR signal at room temperature. The most likely explanation for this species is a hexaaquo complex that has rotational freedom in the a-cage. As the samples cool to 77 K the rotational freedom is hindered and the spectrum shows axial symmetry. Such isotropic signals have been observed before our Cu2+in hydrated faujasite structures.26 The concentration of this hexaaquo complex appears to decrease with increasing Cs+ content. The most likely reason for this is the reduction in void space brought about by the introduction of Cs+ and its associated hydration shell. Evacuation at room temperature is, in most cases, sufficient to destroy this hexaaquo species, and the Cu2+then assumes partial coordination to the zeolite framework. The location of Cu2+inside the zeolite after evacuation at room temperature may be determined from the Cs modulation observed
in the CuCs,H-rho sample in Figure 9a. Simulation gave a best fit for Cu2+located in the a-cage, Le., at site S2, S2*,or S3. Upon introduction of one Cu2+cation either two protons, two cesium ions, or one cesium ion and one proton must be exchanged. A good fit could be achieved if either all six Cs+ cations per unit cell still remained after Cu2+doping or one Cs' had been replaced. Most probably all the Cs+ cations remain in the zeolite for two reasons: first, it has already been established that Cu2+occupies positions in the a-cage and not the octagonal prism; therefore the Cu2+ is not competing for the same cation site as Cs+; second, protons, being small and highly mobile, are much more easily exchanged than Cs+ cations.27 It is interesting that the ESR spectrum of CuCsH-rho evacuated at room temperature is very similar to the ESR spectrum reported previ~usly'~ for Cu2+in fresh K-ZK4 with Si/AI = 2.4. The common structural unit between these two zeolites is an a-cage that is relatively uncluttered by cations. Therefore, this provides corroborative evidence that the Cu2+species is located in the a-cage rather than the P-cage, as was suggested previously." Further, upon dehydration the ESR spectral change for Cu2+in rho is very closely related to that for Cu2+ in K-ZK4. The two species in dehydrated CuCs3H-rho for instance have ESR parameters that very closely resemble the species Cuoland the Cuw found in Cu-K-ZK4. The subscript on Cu indicates the number of directly coordinated water molecules. This again suggests that both of these Cu, species are located in the a-cage in ZK4. One of these dehydrated species could be located in the six ring at site S2 or S2* and the other at site S3. The effect on the Cs modulation of various adsorbates in CuCs6H-rho is shown in Figure 9. Adsorption of water gives the same modulation pattern as for the sample evacuated at room temperature as does adsorption of ethylene. However, upon adsorption of both methanol and ethanol there is a 20% decrease in the modulation depth. Although this decrease is not visually apparent from Figure 9, careful analysis of enlarged spectra verify this fact. This must be due to either a change in the crystallographic site of Cu2+or to a shift in the location of the Cs+ cations. To decrease the modulation depth by changing the Cu2+location, a migration into the octagonal prism would have to be proposed. However, the decrease in modulation depth from such a migration would be more than 50%. A more reasonable interpretation of the data is that the alcohol ligands to the Cu2+ project toward two or three of the octagonal windows, displacing the Cs+ cations from site S5 into an adjacent a-cage. A displacement of only 0.1 nm would be sufficient to account for the decrease in modulation depth. Additional information is obtained from the Cs modulation pattern of CuCs,H-rho in Figure 10. First it is assumed that all
(26) Conesa, .I.C.; Soria, J. J . Chem. Soc., Faraday Trans. 1 1978, 7 4 , 406.
1978.
( 2 7 ) Barrer, R. M. Zeolites and Clay Minerals. Academic: New York,
Cu2+ in Zeolites H-rho and CsH-rho
tA
The Journal of Physical Chemistry, Vol. 90, No. 24, 1986 6459
II H
1
\
H
Figure 16. Schematic representation showing methanol adsorbed on zeolite CuH-rho. The Cu2+is located in site S2* and complexed to three lattice oxygens in the plane of the paper and two methanol molecules above the plane of the paper. A third methanol molecule is located in the adjacent a-cage below the plane of the paper.
the Cs' cations are located in the octagonal prisms and that the introduction of Cu2+does not exchange any of these Cs+ cations. For the sample evacuated at room temperature the modulation is too weak for the Cu2+ to be located in the six ring at site S 2 or S2*. However, the modulation may be simulated closely by placing the Cu2+ in site S3 where the two Cs+ cations that were at 0.54 nm from the Cu2+ (see Table IV) are no longer present. In other words the Cu2+is located in the oxygen four ring of the a-cage as far from neighboring Cs+ cations as possible. Such an arrangement would be favorable in that it would give a more even distribution of the counterbalancing cationic charge. It can also be seen in Figure 10 that the Cs modulation depth in a fresh sample is considerably greater than in a sample dehydrated at room temperature. In fact, the modulation depth in a fresh sample is too great to be accounted for by Cu2+ being located in any of the crystallographic sites. This could be explained if the hexaaquo complex of Cu2+in a fresh sample is in the vicinity of a Cs+ cation. The type of complex formed upon adsorption of methanol on CuH-rho was determined from the deuterium modulation from C D 3 0 H and C H 3 0 D in Figure 1 1. It proved difficult to simulate the modulation from CD,OH with a simple one-shell model because of the deep modulation at small T, indicative of close deuterium, and the persistance of the modulation at long T, indicative of farther deuterium. Some guidance was sought by first trying to fit the modulation from C H 3 0 D . This suggested that two molecules of methanol were directly coordinated to Cu2+ through the methanol oxygen with a Cu2+-D distance of 0.26 nm. From previous work on Cu-K-ZK4" it is known that for methanol directly coordinated to Cu2+ in the a-cage the parameters necessary for simulation of the modulation from C D 3 0 H are r = 0.36 nm and aiso= 0.1 MHz. Further, we know the value of N to be 6 as the interaction is with two molecules of methanol. With these parameters for the first shell of deuterons, a second shell was determined by a best fit to the experimental data. A good simulation was obtained with N = 3, r = 0.28 nm, and ab = 0.3 MHz, consistent with one molecule of methanol. The extra molecule of methanol is probably located in the a-cage adjacent to the one containing the Cu2+cation. The short Cu2+-D distance of 0.28 nm could occur if the methanol molecule is hydrogen bonded via the methyl deuterons to the wall of the a-cage. Therefore, in effect, the methyl end of the molecule is directed toward the Cu2+ cation. The reason the adsorption of CH30D only indicates interaction with two molecules and not three is that the deuterium on the CH30D molecule located in the adjacent a-cage is too far, more than 0.5 nm, to contribute significantly to the modulation. A diagrammatic representation of this configuration is shown in
Figure 16. The Cu2+will either be located in site S2*, as shown, to give a fivefold coordinate complex or in site S3 to give a near square coplanar complex. A trigonal bipyramidal arrangement with Cu2+ at site S2 and one methanol in each of the adjoining a-cages may be ruled out as this would result in an ESR spectrum with reversed g values, g, > gll.28Both the possible configurations would result in methanol molecules projecting through the octagonal windows, which, in the case of Cs+ exchanged zeolites, would result in a displacement of the Cs+ as mentioned previously. Adsorption of C2D4 on CuH-rho results in interaction with four deuterons at a distance of 0.36 nm; see Figure 12. Such complexing of Cu2+with only one C2D4 molecule would not be sufficient to cause interference with the Cs+ positions in the Cs+exchanged zeolites. This would account for the previously mentioned Cs modulation upon adsorption of ethylene (see Figure 9). In hydrated zeolite CuCs3H-rho the Cu2+is coordinated to two water molecules; see Figure 12. After complete dehydration followed by rehydration with 22 Torr D20, a very good fit is obtained with N = 4, r = 0.28 nm, and also= 0.2 M H z as shown in Figure 12a. If the Cu2+is located in site S3, as was previously suggested for CuCs4H-rho, then the complex is most likely square coplanar. The g and A parameters are consistent with either square coplanar or a square pyramid,29of which the latter could be formed at site S2*. Using the same parameters to try and simulate the ESE spectra for CuCs3H-rho for both fresh deuteriated and room temperature evacuated samples in Figures 12b,c gave a slightly worse fit despite these samples having the same ESR spectra as the rehydrated sample. The most likely explanation is that complete H 2 0 D 2 0 exchange was not achieved in these samples. The presence of H 2 0 would reduce the modulation depth as observed. It should be noted that it was not possible to fit the deuterium modulation by using the same approach employed for Cs modulation by averaging over two dimensions. The reason for this is either (i) the deuteriated adsorbates have an additional degree of freedom to rotate around the ligand axis or (ii) at short Cu2+-D distances the unpaired electron can no longer be considered as a point dipole but must be averaged over its d-orbital. However, the spherical approximation gives reasonable fits for the data recorded.
-
Conclusions It has been found possible to simulate the Cs modulation from ESE for a system where the Cs' assumes a predetermined correlated arrangement and where only a small portion of the spins in the g , region are excited. This is achieved by averaging the spin echo intensity over only two dimensions of disorder rather than over three dimensions, commonly termed spherical averaging. Cu2+is found to reside in the a-cage in zeolite rho rather than in the octagonal prism. The most likely complexes formed upon adsorption of water and methanol involve two adsorbate molecules with Cu2+located in site S2* or S3. Cu2+ complexes with only one molecule of ethylene. Similarities are found between Cu2+complexes formed in zeolite rho and zeolite K-ZK4 owing to the existence of a common structural unit, the a-cage. This corroborates the assignment of Cu2+ in hydrated ZK4 to the a-cage. Acknowledgment. This research was supported by the Robert A. Welch Foundation and the National Science Foundation. We are grateful to the Energy Laboratory of the University of Houston and the Texas Advanced Technology Research Program for partial equipment support. (28) Hathaway, B. J.; Billing, D.E. Coord. Chem. Reu. 1970, 5, 143. (29) Morke, V. W.; Vogt, F.; Bremer, H. 2.Anorg. Allg. Chem. 1976,422, 273.
