J. Phys. Chem. 1980, 84, 3215-3223 Lever, F. M.; Powell, A. R. J . Chem. Soc. A 1969, 1477. Lin, M. J. R.D. Disertation,Texas A & M Unhrerslty, Cdlege Station, TX, 1976. Computer program SIM 13 written by Lozos, G., Hoffman, B., and Franz, C., Northwestern Universlty. Nakamoto, K. “Infrared Spectra of Inorganic and Coordination Compounds”, 2nd ed.;Wlley: New Ywk, 1970. Kobayashi, M.; Shirasakl, T. J. Cafal. 1974, 32, 254. Schlick, S.; Kevaln, L. J . Phys. Chem., In press. Lunsford, J. H. C:atal. Rev. 1973, 8 , 135. Ben Taarit, Y.; Lunsford, J. H. J . Phys. Chem. 1973, 77, 780. Breck D. W. “Zeolite Molecular Sleves”; Wiley: New York, 1974. Goodman, B. A,; Flaynor, J. B. Adv. Inorg. Chem. Radiochem. 1970, 13, 135.
3215
(20) Naccache, C.; Ben Taarit, Y.; Boudart, M. ACS Symp. Ser. 1977, No. 40, 156. (21) The nomenclature for site locations follows the conventlon of Smith, J. V. Adv. Chem. Ser. 1971, No. 101, 171. (22) Pearce, J. R.; Sherwood,D.; Lunsford, J. H.; Hal, M. B.,to be submitted for publication. (23) Kellerman, R.; Kller, K. Surf. Defect Prop. SolMs 1975, 4 , 1. (24) Vansant, E. F.; Lunsord, J. H. J. Chem. Sot., Faraday Trans. 2 1973, 69, 1028. (25) Vansant, E. I:.; Lunsford, J. H. Adv. Chem. Ser. 1973, No. 121, 441. (26) Wang, K. M.; Lunsford, J. H. J. Phys. Chem. 1970, 74, 1512. (27) Ben Taarit, Y.; Vedrine, J. C.; Naccache, C.; de Montgolfier, Ph.; Meriaudeau, P. J . Chem. Phys. 1977, 6 7 , 2880.
Physical and Chemical Characterlzation of Cr-Y and Cr-X Zeolites J. IR. Pearce, D. E. Sherwood, M. 8. Hall, and Jack H. Lunsford“ Department of Chemlsfry, Texas A & M Universky, College Sfaflon, Texas 77843 (RecelveO: March 13, 1980; In Final Form: July 3, 1980)
The physical and chemical properties of trivalent chromium supported in zeolites X and Y have been investigated by infrared, EPR, and diffuse reflectance spectroscopies. Nitric oxide was also used as a chemical probe and several chromium-nitrosyl complexes were characterized on the zeolite surface. Earlier investigatorshave reported that the dehydration of Y-type zeolites containing Cr(II1) resultri in the reduction of a major fraction of the Cr to the divalent state. In contrast, we find no evidence for this reduction. The infrared and EPR parameters of chromium-nitrosyl complexes formed upon the addition of NO, to dehydrated Cr-Y samples are explained on the basis of a Cr(II1) precursor. In addition, Fenske-Hall molecular orbital calculations were performed for several models representing Cr-nitrosyl complexes in zeolitic coordination sites. These calculations support tho assertion that Cr(II1) is the predominant cation in Cr-Y samples.
Introductilon I t is well known that silica and aluminosilicate catalysts containing supported chromium are active in the polymerization of ethylene; however, attempts to determine the oxidation state of Cr which gives rise to this activity have led to different conclusions. Indeed, every oxidation state from I1 to V has been Much of this uncertainty arises from the complexity of systems containing ionic chromium at ,a solid interface since in addition to the multiplicity of available oxidation states, a number of chromium complexes may be stabilized on solid surfaces.&l0 The present study is an attempt to resolve some of the ambiguity surrounding the nature of supported chromium. Recently, several authors have investigated Cr ions supported in ~eolites.~l-~* These studies have been conducted primarily on Y-type lattices in order to take advantage of their regular surfaces and well-defined coordination sites. Preparation of these samples generally involves introduction of Cr via an aqueous ion exchange with the [Cr(H20)6]3’ complex cation. This complex is then decomposed by evacuation at 300-500 OC. Several research g r o u ~ p s have ~ ~ proposed ~ ~ ~ ~ ~that ~ -a ~major ~ fraction of the Cr(II1) is reduced to the divalent state during this dehydration by either of the following simplified mechanisms:laJ7 2Cr(III) + nH20 -* 2Cr(II) + (n - 1)H20+ 2H+ + 1/202 (1) or
-
2Cr(III) iOlatticeP- 2Cr(II) + 1/202
(2) Evidence in support of this model has been provided by 0022-3654/80/2084-3215$01 .OO/O
diffuse reflectance spectroscopy, analysis of gases over the samples, and the EPR and infrared spectra of the dehydrated zeolites after adsorption of probe molecules such as CO or NO. ]However,as the present results will demonstrate, reduction of Cr(II1) does not occur to any great extent upon dehydration of Cr-Y zeolites. Rather, the EPR and infrared parameters of nitrosyl complexes formed upon addition of nitric oxide to dehydrated Cr-Y are better explained on the basis of a Cr(II1) precursor. In addition, a theoretical model will be described whereby the EFR parameters of model chromium-nitrosyl complexes were derived from molecular orbital calculations. This technique proved to be quite successful, and the results provide further evidence that trivalent chromium predominates in dehydrated Cr-Y zeolites. The situation is somewhat different in the less stable Cr-X zeolites. Experimental evidence suggests that some Cr(I1) may be produced upon high-temperature evacuation of these samples.
Experimental Section Materials. The CO, NO, H2, and O2gases used in this study were supplied by Matheson. All of them except NO were ultrahigh purity and were used as received. The NO was purified by repeated vacuum distillation at the freezing point of n-pentane (-129 “C). Nitrogen-15-labeled NO was obtained from Stohler Isotope Chemicals and was similarly purified. Chromium-53-labeled Cr203 (96% 53Cr) was purchased from Oak Ridge. The sodium forms of zeolites Y and X were supplied by Linde. Sample Preparation. Samples of Cr-Y and Cr-X were prepared from the Na zeolites by aqueous ion exchange in solutions of CrC13. The exchange cation in such solutions was [Cr(H20)6]3+.After filtering, repeated rinsing, 0 1980 Amerlcan Chemical Soclety
3218
The Journal of Physical Chemistry, Vol. 84, No. 24, 1980
and drying, portions of the zeolites were analyzed for chromium by atomic absorption spectroscopy. Both the Y and X samples were 1.2 wt 5% chromium, which corresponds to 21 and 15% exchange or 3.9 and 4.2 ions per unit cell, respectively. These low chromium exchange levels were intentionally selected since earlier reports indicated that higher concentrations result in significant lattice decomposition.14 Inspection of the 400-1200-~m-~ infrared region, as well as the X-ray diffraction profiles of the powders, assured us that no such decomposition occurred during this ion exchange, or after any of the pretreatments to be described below. Samples of Cr-Y containing the chromium-53 isotope were prepared as follows. A 0.025-g sample of 53Cr203was digested in boiling HC104. Upon cooling, the 53Cr03precipitate was separated by filtration and redissolved in 300 mL of H20. The subsequent addition of 0.5 mL of H202 resulted in a dark blue solution of chromium peroxy complexes.18 Upon standing for 30 min, the [Cr(H20)e]3tion was produced, accompanied by the visible evolution of O2 via reaction 3.l9 The solution was next boiled to remove 2HCrOc + 3H2O2+ 8Ht 2Cr(III) + 302 8H20 (3) excess H202. After the pH of the solution was adjusted to 5.5 with NaHC03, 1.5 g of Na-Y was added and the ion exchange was carried out. This procedure resulted in 53Cr-Y zeolites with unit cell compositions comparable to that described-abo-ve. Spectroscopic Methods. A Cary 14 spectrophotometer with a type-I1 diffuse reflectance attachment was used to obtain diffuse reflectance spectra (DRS) in the range 870-310 nm. Spectra were recorded with the zeolites in the form of loose powders contained in vacuum-tight quartz cells. CaS04 was used as the reference material. The DRS in this text are reported as plots of the Schuster-Kubelka-Munk function, F(R,), vs. wavelengtha20 Infrared spectra were recorded with a Beckman IR-9 spectrophotometer. Infrared samples of the zeolites were in the form of self-supporting wafers and were pretreated in a quartz vacuum cell fitted with KC1 windows. Several intense peaks of polystyrene were used as wavenumber references. The accuracy of band positions was estimated to be.i2 cm-l. A Varian E-6s spectrometer was used to obtain X-band EPR spectra. These were recorded at both 25 and -196 "C. The g values were calculated relative to a 2,2'-diphenyl-l-picrylhydrazyl (DPPH) standard. Spin concentrations were calculated relative to a phosphorus-doped silicon standard by using a numerical double integration procedure.21 Molecular Orbital Calculations In an effort to elucidate possible chromium-nitrosyl complexes in these systems, a theoretical study was made of several model molecules which could conceivably be paramagnetic. These included the mono- and dinitrosyl complexes of both Cr(I1) and Cr(II1) as well as possible charge-transfer moieties. The numerical procedure is described below. In the construction of these models, the chromium ions were assumed to bond to three oxygen atoms to simulate the coordination symmetry present in a faujasite six-ring. One or two nitrosyl ligands were then included to form the mono- or dinitrosyl complexes, as desired. Only linear Cr-N-0 linkages were considered, as suggested by Enemark and Feltham for d3 or d4 systems.22 However, the orientation of the ligands with respect to the three pseudolattice oxygen atoms (i.e., eclipsed or staggered) was considered in detail. The model molecules were termi-
-
+
Pearce et ai.
nated by including two hydrogen atoms on each of the oxygens representing the zeolite lattice. These hydrogens were oriented so as to preserve tetrahedral symmetry about the oxygens. Position vectors were estimated from relevant X-ray crystallographic data in the literature. Parameter-free molecular orbital (MO) calculations were performed by using the Fenske-Hall approximation.23The atomic basis sets of ClementiZ4were utilized for nitrogen, oxygen, and hydrogen atoms, whereby the s functions were reduced to a single exponent.25 Richardson's atomic functions26were used for chromium, and the 4s and 4p exponents were initially set equal to 2.OaZ7Atomic charges were estimated from Mulliken populations. Some 40 MOs were calculated for each model. Elements of the theoretical g tensor matrices were calculated as follows. The eigenvectors of each calculated MO were first renormalized and the zero differential overlap approximation was employed in the calculation of the g values. Deviations from the free electron g value were then calculated from eq 4.27 Here, the summation is over &jk
=2 c n
( # O I cJx j L j a j l # n )
(#nlTLkakl#O)/(En
- EO)
(4)
n molecular orbitals (approximately 40) of energy E,. The energy of the highest occupied molecular orbital (HOMO) is Eo.The spin-orbit coupling constants, Xj, were taken to be those of the free atoms or ions. The L operators were independentlyh,, Ly,or L,. Because of the arbitrary choice of coordinate axes for the models, off-diagonal elements in the g tensor matrix were nonzero; however, diagonalization of the matrix yielded the principal elements Ag,,, kYy, and Ag,,. Finally, theoretical g values were found from eq 5. gjk
= g e - &jk
(5)
Results Chromium in Zeolites Y and X. Freshly prepared samples of Cr-Y and Cr-X displayed a broad, isotropic EPR signal with g N- 2.0 and a width of several hundred gauss. This signal has been attributed by several authors to the exchange cation [Cr(H20)6]3+.4v14 The diffuse reflectance spectrum characteristic of both samples is reproduced in Figure la. I t shows maxima at 420 and 580 nm which have also be assigned to the hexaaquochromium(II1) ion.4t14928 As the Cr-Y and Cr-X samples were slowly heated under vacuum, their color changed from green to gray. At the same time, the broad EPR signal of the hexaaquo ion decreased in intensity until, at 400 "C, it could not be detected. At this point, the EPR spectrum of Cr-X was featureless while the spectrum of Cr-Y displayed only a very weak, axial signal with g, N- 2.0 and gil N- 1.9. This has been noted by previous authors and is thought to arise from a small amount of Cr(V).29 However, as this signal was not reproducible and was due, at best, to a minute fraction of the total Cr present, it will not be considered in detail. Concurrent with the decrease in the EPR signal of [Cr(H20)6]3tduring dehydration, the 420- and 580-nm bands in the DRS of Cr-Y and Cr-X were also attenuated. A representative spectrum, shown in Figure lb, does not display any well-defined maxima and cannot be interpreted. These changes which occurred upon dehydration were somewhat reversible in the following sense. When the dehydrated Cr-Y and Cr-X samples were exposed to HzO vapor at ambient temperature and warmed to 50 "C, the broad EPR signal of the hexaaquochromium ion was again
The Journal of Physical Chemistty, Vol. 84, No. 24, 1980 3217
Characterization of Cr-Y and Cr-X Zeolites
1
I io0
I
I
I
I
I I I I 700 400 500 600 WAVELENGTH lnml
Figure 1. Selected diffuse reflectance spectra of Cr-Y and Cr-X zeolites: (a) hydrated Cr-Y or Cr-X; (b) Cr-Y or Cr-X dehydrated at 400 OC; (c) dehydrated Cr-Y exposed to 100 torr of O2at 400 OC; (d) dehydrated Cr-Y OT Cr-X exposed to 10 torr of NO at 25 OC. Spectra are displaced along the ordinate for clarity.
observed for both samples. The characteristic bands at 420 and 580 nm also reappeared in the DRS, although the original intensities could not be reproduced. One should notice that the gas phase over both samples was examined by mass spectrometry after this rehydration. This was done in an attempt to reproduce earlier work in which it was reported that H2was evolved during this reaction.16 In each case, however, only H20was detected. The chromium in dehydrated Cr-Y and Cr-X samples was readily oxidized by molecular oxygen at elevated temperatures. For example, the EPR spectrum in Figure 2a is that of dehydrated Cr-Y treated with O2 at 300 OC. This signal was nearly axial except that two distinct gli values were noted at 1.910 and 1.883. The value of g, was 1.991. The spectrum of Cr-X after similar pretreatment is reproduced in Figure 2b. This absorption was nearly isotropic with g = 1.985. Both spectra were observable at 25 or -196 "C and are attributed to Cr(V). Their intensities passed through maxima at oxidation temperatures of 250-300 "C, at which point a calculation of spin concentration showed that the Cr(V) would account for -40% of the total Cr present in the samples. The decrease in intensity above this temperature was presumably due to further oxidation to Cr(V1). The reason for the apparent difference in coordination symmetry of Cr(V) in the two zeolites is presently under investigation. The diffuse reflectance spectrum in Figure ICis typical of Cr-Y or Cr-X samples after oxidation at 300 "C. The spectrum is characterized by a band at 350 nm with a weak absorption at 720 nm. The high-energy band is in good agreement with earlier reports in which it was assigned to Cr(V1). Nitric Oxide Adsorption on Cr-Y and Cr-X. When 10 torr of nitric oxide was admitted to dehydrated samples of Cr-Y or Cr-X, their color immediately changed from gray to tan. The DRS of each sample reflected this change in that a new absorption was observed at 470 nm with a
Flgure 2. X-band EPR spectra of Cr-Y and Cr-X zeolites recorded at -196 O C : (a) dehydrated Cr-Y exposed to 100 torr of O2at 300 OC; (b) dehydrated Cr-X exposed to 100 ton of O2at 300 OC; (c) dehydrated Cr-Y exposed to 10 torr of NO at 25 O C ; (d) dehydrated Cr-X exposed to 10 torr of NO at 25 'C.
u m
2400
1800
le00
WAVENUMBER
1400
1:
xl
(cm-')
Flgure 3. Infrared spectra (transmission mode) of dehydrated Cr-Y samples: (a) background spectrum; (b) exposed to 10 torr of NO at 25 OC; (c) exposed to 10 torr of "NO at 25 OC; (d) exposed to 10 torr of an equimolar mixture of 14NO115N0at 25 OC.
high-energy shoulder at 410 nm. A typical spectrum is reproduced in Figure Id. T h e most striking feature of dehydrated Cr-Y and CI-X zeolites after NO contact was the appearance of two in-
3218
The Journal of Physical Chemistry, Vol. 84, No. 24, 1980
Pearce et al.
tense infrared bands in the region typical of N-0 stretching frequencies. The infrared spectrum of Cr-Y in Figure 3b shows these bands at 1900 and 1775 cm-l. Each band possesses weak, partially resolved, low-wavenumber shoulders. The major bands grew moderately over several hours of NO contact and were stable to evacuation below 50 "C. The alternative use of nitrogen-15-labeledNO gave rise to the analogous spectrum in Figure 3c. The two bands of interest were shifted by 30 cm-l to 1870 and 1745 cm-'. A number of models have been proposed to account for this spectrum. These have included the mono- and dinuclear hyponitrite complexes in models a and b, as well as the dinuclear complex containing both bridging and terminal nitrosyl groups as shown in model c. The hy-
3
/'* f'm
,'*
Y
,'
A
IY/' I
,
,'
,;f
, I'
t'
N=N 0
a
b
C
ponitrite complexes can easily be eliminated since the N-O stretching modes in such complexes generally fall at much lower frequencie~.~~~' Models b and c are improbable since it is doubtful that the required ion pairs would be present a t these chromium concentrations. Furthermore, model c can be excluded on the basis of the following experiment. When a 1:l mixture of 14N/15Nlabeled NO was used, the 1900- and 1775-cm-l bands were each split into a triplet. This spectrum is shown in Figure 3d. The intensity ratio for the three peaks ion each triplet was approximately 1:2:1, and the middle band, which was the strongest, fell at a frequency midway between those which were observed when pure 14N0or 15N0 was used. I t is clear that the surface complex giving rise to the above spectra is a geminal, chromium-dinitrosyl species containing strongly coupled and equivalent nitrosyl ligands. The two major infrared bands are simply the symmetric and asymmetric stretching modes of the coupled ligands. This assignment explains directly the triplet splitting and 1:21intensity ratio observed when the isotope mixture was used since one would expect the resultant dinitrosyl complex to be comprised of 25% [Cr(14N0)2], 50% [Cr(14N0)(16N0)],and 25% [Cr(15NO),]. This method has previously been utilized to identify dinitrosyl complexes of cobalt in Co-Y zeolites32and of tin on SnOFm One should note that the niodel containing both bridging and terminal nitrosyl groups (model c) is inconsistent with these results. Additional evidence in support of the assignment of the infrared spectra to a dinitrosyl complex of chromium was provided by the observation that the intensities of the two major infrared peaks are strictly proportional, as shown by the plot in Figure 4. This indicates that both bands arise from the same surface complex and that no "hidden" peaks are masked under either of the major bands. Further, the 15N isotope shift is the same (30 cm-l) for both peaks and agrees well with other reported isotope shifts for dinitrosyl complexes.9~30~34 For reasons which will be discussed below, this dinitrosyl complex of chromium on the Cr-Y surface is attributed to trivalent Cr ions coordinated to nearly neutral nitrosyl ligands, i.e., [Cr"*(N0)2l3+. Infrared spectra of dehydrated Cr-X samples after adsorption of NO are shown in Figure 5. Again, two intense bands were observed, but at 1895 and 1770 cm-l, some 5 cm-l lower in wavenumber than in Cr-Y. This difference is attributed to the greater negative charge density of the X-type lattice. The two bands are again assigned to a
.1
.3
.2
.4
.5
.b
10
.7
IR INTENSITY (ASYMM. MODE) Flgure 4. Plots of the infrared intensities for the symmetric vs. asymmetric stretching modes of a dlnltrosyl complex on Cr-Y zeolite (left ordinate; dlfferent symbols represent different experiments) and the EPR intenslty (arbitrary unlts) vs. the infrared absorbance of the dinitrosyl assymmetrlc stretchlng mode observed upon the addition of 10 torr of NO to dehydrated Cr-Y at 25 OC (right ordinate).
2400
2000
1800
1600
WAVENUMBER
1400
1
(crn-l)
Flgure 5. Infrared spectra (transmlsslon mode) of dehydrated Cr-X samples: (a) background spectrum; (b) exposed to 10 torr of NO at 25 O C ; (c) exposed to 10 torr of 16N0 at 25 OC; (d) as In (c), but subsequently evacuated at 25 OC for 24 h.
chromium-dinitrosyl complex. One should note that the low-wavenumber shoulders were markedly more pronounced than in Cr-Y. The frequencies of these shoulders were estimated to be 1880 and 1750 cm-l. In addition, two new peaks were noted to 1650 and 1260 cm-l, and these
Characterization of Cr-Y and
Cr-X
Zeolites
were very stable. They remained prominent after evacuation to 200 "C, long after all bands in the dinitrosyl region had been removed. The use off nitrogen-15-labeled NO gave rise to the infrared spectrum in Figure 5c. After 24 h of 15N0contact, seven distinct bands were visible at 1865,1845,1740,1710, 1620,1370,and 1235 cm-l. By using the same isotope shift as observed for the dinitrosyl absorptions in the Cr-Y system (30 cm-'), >weassigned the 1865- and 1740-cm-' bands to the symmetric and asymmetric stretching modes of a [Cr(l5N0),] complex. One should note that the lowwavenumber shoulders became well resolved, indicating that their isotope shift was somewhat greater than 30 cm-'. The complex which gave rise to these low-wavenumber peaks was also more stable than the original dinitrosyl as they were attenuated to a lesser degree upon prolonged evacuation (Figure 5d). A discussion of the relationships between the four bands in the dinitrosyl region, including an assignmerit of the oxidation states of chromium in these complexes, will be presented in the following section. Upon the addition of an equimolar mixture of 14N/15N nitric oxide to Cr-E;, significant splitting of the bands was noted. Because of the multiplicity of bands in the dinitrosyl region, only broad, unresolved absorption envelopes were observed; however, the 1650- and 1260-cm-l bands were split into distinct doublets. Intensity ratios for these doublets could not be determined because of interference by tracle amounts of H20at 1650 cm-l and the strong background absorbance of the zeolite lattice around 1200 cm-l. The frequencies, stabilities, and doublet splitting of the 1650- and 1 2 6 0 - ~ m -bands ~ suggested the presence of surface nitrite (NO
