4024
J . Phys. Chem. 1987, 91,4024-4029
Redox Behavior of Ruthenium In Zeollte Y R. Shoemaker and T. Apple* Department of Chemistry, University of Nebraska-Lincoln, (Received: November 24, 1986)
Lincoln, Nebraska 68588-0304
The lz9XeNMR of xenon gas sorbed in ruthenium-exchangedY zeolites has been combined with volumetric oxidation/reduction studies and transmission electron microscopy to characterize the state of ruthenium at various stages of catalyst treatment. Upon ion exchange the majority of the Ru occupies the faujasite supercages and is highly dispersed. Initial exposure to hydrogen at rwm temperature causes nearly complete reduction of Ru3+to Ruo. Protons formed by the heterolytic cleavage of H, during the reduction process become the charge compensating cations. Exposure of the reduced catalyst to oxygen results in the quantitative oxidation of Ru to RuO, at temperatures up to 625 K, but no detectable migration of the ruthenium species accompanies this oxidation. Subsequent high-temperature rereductions in hydrogen atmospheres cause progressive migration of the ruthenium to the exterior of the zeolite crystals, forming larger metal particles which are difficult to reoxidize. The degree of migration is dependent upon the severity and the duration of the reduction. All detectable metal migrations occur as reduced metal species. High-temperature evacuation (725 K) of a reduced Ru-Na-Y zeolite causes sintering of the ruthenium and migration of the metal to the exterior of the faujasite cages. Oxidized ruthenium species do not migrate when evacuated at high temperature. Transmission electron microscopy measurements confirm that the migration of the ruthenium which was indicated by the xenon NMR results is to the exterior of the zeolite crystals.
Introduction The location of the ruthenium metal in Ru-exchanged zeolites is an important factor in determining the product distribution of Fischer-Tropsch chemistry occurring over these catalysts.' Thus, it is important to understand the behavior of ruthenium under standard catalyst treatments such as hydrogen reduction, evacuation, heating, and oxidation. A number of studies have addressed the problem of metal location in zeolites after various treatments using X-ray scattering,2 XPS,3-5 transmission electron microscopy (TEM),6 and NMR.7 We have carried out an investigation of a Ru-Na-Y zeolite catalyst formed from ruthenium trichloride. Xenon NMR, volumetric uptake measurements, and transmission electron microscopy have been employed to study the state and location of the ruthenium metal under various oxidizing and reducing conditions. The lZ9XeN M R of xenon gas sorbed in zeolites is a sensitive probe of the local environment inside the zeolite superstru~ture.'-'~ It is especially useful in investigating the local structures in metal-exchanged zeolites, since the electronic environment and, thus, the chemical shift of the xenon gas are very sensitive to the types of atoms with which it collide^.^^* Due to the size of xenon atoms, they can access only the faujasite cages in the zeolite and, therefore, experience collisions with metal particles only if these particles are in the supercages. In this way xenon N M R provides information about the location of the metal within the zeolite. Verdonck et aL6 have carried out quantitative redox experiments to characterize the state of the ruthenium metal in Ru-Na-Y zeolite systems exchanged as the hexaammine complex. The size of the metal particles, as well as the stoichiometry of the redox process, can be obtained by this technique. It was found that upon reduction at 373 K the ruthenium metal is totally reduced and located inside the zeolite cages. Upon subsequent oxidation, the metal readily oxidizes to RuOz at temperatures below 625 K . These authors concluded that, since further reduction of the R u 0 2 yielded large reduced ruthenium particles on the outside of the zeolite supercages, the R u 0 2 was in the form of large particles which had migrated to the exterior of the zeolite supercages upon oxidation. Our results indicate that metal migration out of the zeolite supercages in Ru-Y zeolites occurs only when the ruthenium metal is in the reduced state. Migration is not observed during the oxidation process, nor upon high-temperature evacuation of an oxidized catalyst. A significant amount of ruthenium metal remains in the zeolite supercages even after migration has occurred, *To whom correspondence should be addressed.
0022-3654/87/2091-4024$01 SO10
and only through a high-temperature treatment (725 K) of a reduced.Ru-Y sample for 12 h is nearly all of the ruthenium removed from the zeolite supercages.
Experimental Section The Ru-Na-Y samples were prepared by ion exchange with RuCI3-3Hz0(Strem) and Linde LZy-52 molecular sieves. The zeolite was added to 0.04 M RuCl, with constant stirring for 24 h. It was observed that within 1 h the opaque supernatant solution had become colorless, indicating that almost complete exchange had already occurred. The resulting catalyst was washed with deionized water and dried in air for 24 h at 383 K and was found to be 8.2 f 0.2 wt % ruthenium as determined by visible absorption and argon plasma emission spectrophotometry of the supernatant solutions before and after exchange. The concentrations were determined by measuring the intensity of the visible absorption at 500 nm and the intensity of the atomic emission of the 350-nm line of Ru. All samples in this study were pretreated by outgassing for 12-14 h a t 573 K to a residual pressure of less than 1 X lov5 Torr to remove all water from the catalyst. All oxidation and reduction experiments were carried out in a glass vacuum system under static 0,or H2 (Linde, 99.999% pure; passed through Drierite & molecular sieves). Hydrogen and oxygen uptakes were measured by a capacitance manometer (Vacuum General), with a cold trap (liquid Nz with Hz and dry ice/acetone with 0,) to remove contributions from water formed over the zeolite. Research grade xenon gas was obtained from Union Carbide and was used without further purification. Xenon uptake was measured as a function of pressure at each stage of catalyst treatment. (1) Nijs, H.; Jacobs, P.; Uytterhceven, J. J . Chem. SOC.,Chem. Commun. 1979,181. (2) Gallezot, P.; Alarcon-Diaz, A.; Dalmon, J.-A,; Renouprez, A. J.; Imelik, B. J . Catal. 1975, 39, 334. (3) Pedersen, L. A.; Lunsford, J. H. J . Catal. 1980, 61, 39. (4) Minachev, Kh. M.; Antoshin, G. V.; Shpiro, E. S.;Navruzov, T. A. Izu. Akad. Nauk SSR, Ser. Khim. 1973, 2131.
(5) Minachev, Kh. M.; Antoshin, G. V.; Shpiro, E. S.; Isakov, Ya. I. Izu. Akad. Nauk SSR, Ser. Khim. 1973, 2134. ( 6 ) Verdonck, J.; Jacobs, P.; Genet, M.; Poncelet, G.J . Chem. SOC., Faraday, Trans. 1 1980, 76,403. (7) Scharpf, E. W.; Crecely, R. W.; Gates, B. C.; Dybowski, C. J . Phys. Chem. 1986. 90. 9. - . (8) De-Mendrval, L. C.; Fraissard, J. P.; Ito, T. J . Chem. SOC.Faraday Trans. 1 1982, 78,403.
(9)Ito, T.; De Menorval, L. C.; Guerrier, E.; Fraissard, J. P. Chem. Phys. Lett. 1984, 1 1 1 , 27 1. (10) Ito, T.; De Menorval, L. C.; Fraissard, J. P. J . Chim.Phys. Phys.Chim. Biol. 1983, 80,513.
0 1987 American Chemical Society
The Journal of Physical Chemistry, Vol. 91, No. 15, 1987
Redox Behavior of Ruthenium in Zeolite Y 150 -f
i
W - Ru* -Y A - Ne-Y
I
I
n
4025
I
2 a
3+
m-
RU -Y A- Ne-Y
:zoo+\
_-
I
T
,
T
4
I
i
50
I
I
I
I
I
;
i
100 200 300 400 500 600 700 800
100 200 300 400 500 600 700 800
Xenon Pressure (Torr)
Xenon Pressure (Torr)
Figure 1. Plot of xenon shift vs. xenon pressure in Na-Y zeolite and Ru3+-Na-Y zeolite. Chemical shifts are referenced to that of xenon gas extrapolated to zero pressure.
lz9XeN M R spectra were acquired on a Varian XL-200 N M R spectrometer for which the xenon resonance occurs at 55.3 MHz and/or on a Varian XL-300spectrometer operating at 83.0 MHz for xenon. Spectra are the result of the accumulation of from 100 to 3000 transients with a repetition rate of either 0.5 or 1.0 s. Unless otherwise stated, chemical shifts are referenced to the resonance of xenon gas extrapolated to zero pressure by using methods described e l ~ e w h e r e . ~Spin-lattice ~~ relaxation times ( T , ) were obtained via the inversion-recovery technique.I2 Magnetic susceptibility measurements were made on all samples, and contributions to the lz9Xe resonance position from bulk susceptibility are less than 1 ppm in all cases. To obtain transmission electron micrographs, samples were dusted onto carbon reinforced formvar-coated grids and observed with a Phillips 201 electron microscope operating at 80 kV.
Results and Discussion The xenon chemical shift vs. xenon gas pressure for both Na-Y and Ru3+-Y zeolites is plotted in Figure 1. The chemical shift of the xenon gas sorbed in the Na-Y zeolite is linear with pressure, the slope and intercept being in remarkable agreement with those reported by Scharpf et al.’ The pressure dependence of the xenon shift in unreduced ruthenium zeolite is also quite linear at pressures above 150 Torr, but with a smaller slope than in the unexchanged Na-Y zeolite. There is, however, a sharp increase in shift, in the downfield direction, when the spectrum is acquired at pressures below 100 Torr (Figure 1). This behavior is similar to that in Pt-, Ir-, Pd-, and Ni-exchanged The sizes of the shifts observed here are also similar to those reported previously for highly dispersed metals in zeolites at equivalent xenon loadings (vide infra).8 Another Ru-Y zeolite sample prepared from ruthenium hexaammine gave similar xenon shift magnitudes. Three types of interactions are responsible for the observed resonance shift in lz9XeN M R experiments: Xe-Ru, Xe-zeolite, and Xe-Xe collisions. A xenon atom will experience many of these collisions during the N M R measurement, and thus the shift is a weighted average of the above interactions. The relative contribution of the Xe-Ru and the Xe-zeolite collisions to the observed resonance frequency increases with decreasing pressure or xenon concentration ([Xe]). Only those xenon atoms which are restricted to the cagelike structures of the zeolite crystals are detected in the N M R coil. The size of the xenon atom excludes it from the sodalite cages and the hexagonal prisms;8 thus, only metals located inside the faujasite supercages can cause a shift in the xenon resonance position.8-10 In all of the ruthenium-exchanged zeolites studied, the shift induced by the ruthenium metal is in the downfield direction. Since the frequency of Xe-Ru (1 1) Jameson, A. K.; Jameson, C.; Gutowsky, H. S . J . Cbem. Pbys. 1970, 53, 23 10. (12) Farrar, T. C.; Becker, E. D. Pulse and Fourier Transform NMR; Academic: New York, 1971.
Figure 2. Plot of xenon NMR line width vs. xenon pressure in Na-Y zeolite and Ru”-Na-Y zeolite.
TABLE I: ‘*%e NMR Relaxation Parameters fwhh (55
samde a. Na-Y ~~
treatment evac 575 K, P(Xe) = 385
Torr evac 575 K, reduced 383 b. Ru-Y K, evac 400 K, P(Xe) = 300 Torr c. Ru3+-Y evac 575 K, P(Xe) = 207 Torr
MHz),”
Hz
fwhh (83 MHz), Hz TI? s
80
82
0.62
275
375
0.08
1120
1115
0.20
aError in line width is f 6 Hz (estimated outer limits). bError in T , is fO.01 s (95% confidence limits).
collisions is manifest in the chemical shift, this shift relates to the number density of ruthenium particles contained in the supercages. Thus, the larger the downfield shift, the greater the number density of ruthenium particles (or total Ru surface area) contained within the supercages. A marked increase in the N M R line width with decreasing pressure is evident for the xenon sorbed in the unreduced Ru-Y zeolite (Figure 2). At low pressure, very broad lines (>lo00 Hz) are observed. This broadening decreases rapidly at higher xenon pressures as the contributions from X e X e collisions become more significant. This line width effect is opposite to that observed in the unexchanged zeolite, in which the adsorbed xenon resonance narrows slightly with decreasing pressure. We have found that the relaxation processes in these systems are quite complex. Table I shows some results of both variable field measurements of the 129Xeline widths and a T I relaxation study at 83 MHz using the inversion-recovery technique.12 Note that in all cases the lines are much broader than l / T l , indicating that longitudinal relaxation is not determining the line width and that T 2 < TI. The independence of the line width of the Xe resonance in the unreduced Ru3+-Y zeolite to changes in the magnetic field strength suggests that it is the interaction of xenon atoms with the paramagnetic Ru3+ centers through dipole-dipole coupling which is the mechanism for the efficient transverse relaxation and, thus, the broad lines. For those samples referred to in Table I, we observe a field dependence of the N M R line width only for xenon in the reduced Ru-Y zeolite. This field strength dependence of the line broadening is due to a distribution of shielding environments probably due to the imperfection of metal distribution within the zeolite sample. N o field dependence is observed in the line width in unexchanged Na-Y zeolite. This is not surprising since the environment inside all of the supercages of an unexchanged zeolite is expected to be homogeneous. A detailed study of the factors affecting the line width in each catalyst is necessary to define the relaxation mechanisms involved. Because the use of 129XeNMR in the study of metal-exchanged zeolites is still relatively new, the method of reporting chemical shifts and other N M R parameters is not yet standardized. In the early work by De Menorval et al. on platinum zeolites8the N M R
4026
The Journal of Physical Chemistry, Vol. 91, No. 15, 1987
100
300
200
400
Shoemaker and Apple
500
Xenon Pressure (Torr) Figure 3. Plot of xenon gas uptake vs. pressure for Na-Y zeolite and Ru-Na-Y zeolite after various redox treatments: 0 , Na-Y; X, evacuated Ru3+-Na-Y; 0, 298 K reduced Ru-Na-Y; A, 575 K, reduced Ru-Na-Y; A, 575 K reduced/625 K oxidized Ru-Na-Y.
results were reported vs. [Xe] (xenon concentration in terms of xenon atoms per gram or xenon atoms per supercage); however, Scharpf et ala7reported the N M R shift as a function of xenon pressure in their work on Ni-Y systems. In order to allow comparison of our results with both of the above studies, it was necessary to correlate the uptake of xenon atoms in each sample with the equilibrium xenon pressure. Figure 3 shows a plot of xenon uptake vs. pressure for Na-Y and several Ru-Y samples at different stages of catalyst treatment. Adsorption isotherms were measured for every catalyst for which N M R measurements were taken, and the uptake vs. pressure is linear for all samples over the pressure ranges of interest. One of the major goals of this work was to determine the location of the ruthenium in the zeolite at various stages of redox treatment by observing changes in the chemical shift of the xenon NMR resonance, stated previously, three types of collisions affect the xenon chemical shift: Xe-Ru, Xe-Xe, and Xe-zeolite. For one to attribute changes in chemical shift solely to variation in Xe-Ru collisions, other contributions must be kkpt constant. By knowing precisely the number of xenon atoms per supercage as a function of xenon pressure for each sample, we may determine the shift of the xenon resonance in the Ru-exchanged samples relative to Na-Y at that xenon concentration. With this method, changes in xenon shift are due only to differences in the relative number of Xe-Ru collisions. The Ru-Y zeolite catalysts were subjected to several cycles of reduction, oxidation, and evacuation. Following each treatment xenon gas was quantitatively adsorbed at an equilibrium pressure of 300 Torr. Each experiment was carried out at least twice with different samples from the same catalyst batch, and the N M R shifts were reproducible to within f0.5ppm in every case. Xenon N M R shifts were unaffected by vacuum treatments performed at room temperature prior to exposure to xenon gas. The oxidations and reductions were carried out quantitatively in a temperature-programmed manner for comparison with the results of Verdonck et aL6 on Ru-Y zeolites exchanged as the hexaammine complex. The results of these measurements are presented in Figure 4 and are similar to Verdonck's results. Initially, the Ru3+ is almost totally reduced at room temperature by HZ. The total hydrogen uptake, corresponding to the area under curve "a" in Figure 4, is 2.8 H2 molecules per ruthenium atom (atomic ratio H:Ru = 5.6:l).This stoichiometry supports the overall reaction Ru3+
+ 3H2 + 3 2 0 -
-+
RuO-~H,~, + 3ZOH
where ZO- represents the anion sites of the zeolite framework. The low reduction temperature and the proximity of the H2:Ru to 3.0 indicate finely dispersed ruthenium; Le., 2.8/3 or 93% of the ruthenium is capable of interacting with hydrogen. The Ru-Y catalyst sample was then oxidized, and the oxygen uptake measured as a function of temperature. Referring to Figure 4b, we see that the maximum in O2uptake occurs at 425 K, with quantitative oxidation occurring below 635 K. The final integrated
200
300
400
500
600
700
Figure 4. Temperature-programmed oxidation/reduction results for Ru-Na-Y zeolite: (a) initial reduction (H2); (b) subsequent oxidation (02); (c) rereduction (HJ; (d) reoxidation (02).
Xe Shift in Ru-Y
Zeolites
.u
c
@d @ e
h
O f n
40
30
g
A-
-+--++--20
10
0
Shift {PPM vs Na-Y a t Const.[Xe]l
Figure 5. Effect of oxidation/reduction treatment on the 129XeNMR shift; (a) Ru3+-Na-Y, evacuated at 575 K; (b) Ru-Na-Y, reduced at 298 K; (c) Ru-Na-Y, redwed at 375 K; (d) Same as (c) + oxidized to 635 K; (e) same as (d) + rereduced at 575 K; (f) same as (e) + reoxidized at 725 K; (g) Same as (f) + reduced at 725 K,
02:Ru ratio is found to be 0.99 (atomic ratio 0 : R u = 2:l). N o detectable water is observed during oxidation. (Other catalyst treatments involving water evolution always show some condensed water on the walls of the glass reactor tube when the catalyst temperature is above 470 K.) The 0:Ru ratio of 2:1 and the lack of any observed water formation suggest that reduced Ruo becomes R u 0 2 as deduced also by other worker^.^,^ Rereduction of this catalyst is quantitative (H2:Ru = 2:l) upon heating to 375 K (Figure 4c) and produces zerovalent ruthenium. During the reduction of R u 0 2 water is evolved, observed as condensate on the inner walls of the reaction tube. Much of the ruthenium metal formed at this stage cannot be reoxidized at the temperatures measured in our study; however, a significant amount of reoxidation does still occur with a broad maximum at about 500 K (Figure 4d). This indicates the presence of a bidisperse distribution of metal particle sizes in the zeolite because highly dispersed Ruo oxidizes at lower temperatures than large ruthenium metal particles.6 The uptake measurements just described provide no direct information about the location of the ruthenium after the various treatments. The '29XeN M R spectrum is sensitive to the location of the ruthenium, and therefore, the N M R shifts of xenon in the Ru-Y zeolite under various oxidation/reduction conditions were measured. The results are presented graphically in Figure 5 and are tabulated in Table 11. The lz9Xe N M R line following quantitative xenon adsorption in evacuated Ru3+ zeolite (sample a) occurs 40.3 ppm downfield from the xenon resonance in a Na-Y sample at the same xenon concentration. Upon exposure of the catalyst to hydrogen at room temperature followed by evacuation (sample b), the Xe N M R spectrum shows a resonance 28.6 ppm downfield of xenon in Na-Y zeolite at constant [Xe]. Thus, after low-temperature reduction the ruthenium has less effect on the xenon chemical shift than the reduced ruthenium, although the shift is still large. At this low reduction temperature, metal
The Journal of Physical Chemistry, Vol. 91, No. 15, 1987 4027
Redox Behavior of Ruthenium in Zeolite Y TABLE II: Effect of Qxidation/Reduction Treatment on the '*%e NMR Shift sampleu treatment Xe shiftb
TABLE 111: Effect of High-Temperature Evacuation on the I2%e NMR Shift
Na-Y a. Ru'+-Na-Y b. Ru-Na-Y c. Ru-Na-Y d. Ru-Na-Y
a. Na-Y
e. Ru-Na-Y
f. Ru-Na-Y g. Ru-Na-Y
evac 575 K evac 575 K evac 575 K, reduced 298 K, evac 298 K evac 575 K, reduced 375 K, evac 575 K same treatment as in "c" oxidized to 635 K, evac 635 K same treatment as in "d" rereduced at 575 K, evac 515 K same treatment as in "e" reoxidized to 725 K, evac 725 K same treatment as in 'P' + additional reduction at 725 K
+ + +
0.0 40.3 28.6 24.2 24.3 20.5 19.5 6.2
"Sample letters correspond to labels in Figure 5. 6Error in chemical shift = f 1 . 5 ppm, which is the maximum possible error from all sources including both error in Xe uptake and NMR measurements. Shifts are in ppm vs. Na-Y at constant [Xe]. migration is surely insignificant, and this shift (28.6 ppm from Xe in Na-Y) can be taken as a benchmark for the effect on the xenon resonance position of highly dispersed, reduced ruthenium in the zeolite supercages. When reduction is carried out at higher temperature (sample c), the chemical shift observed in the xenon N M R spectrum moves progressively toward the shift of xenon gas in the unexchanged Na-Y zeolite at the same xenon loading. Thus, for increasing reduction temperature, the influence of Ru on the chemical shift decreases, indicating that the number of Xe-Ru collisions in the zeolite supercages per unit time is decreasing. We associate, therefore, a shift in the N M R line position toward the value for Xe gas in Na-Y zeolite (at a particular [Xe]) with migration and sintering of the ruthenium. This migration results in an overall decrease in the number density of ruthenium particles in the supercages. Referring again to Figure 5 and Table 11, one sees that the oxidation of reduced ruthenium to R u 0 2 and subsequent evacuation (sample d) result in little shift in the xenon resonance position. Therefore, oxidation of Ruo to RuOz is not accompanied by any detectable decrease in the total number of Ru particles in the faujasite cages. Rereduction and evacuation of RuOz (sample e) result in a shift of almost 4 ppm in the adsorbed xenon resonance toward that of xenon in unexchanged Na-Y at constant [Xe]. Therefore, reduction of the RuO, to reduced ruthenium causes a large decrease in the number density of ruthenium particles in the zeolite supercages, thereby causing an upfield resonance shift. Transmission electron microscopy confirms that the decrease in ruthenium particle number density is associated with metal migration toward the exterior of the zeolite crystallites (Figure 6). Micrographs of Ru-Y which has been evacuated at 575 K show no evidence of metal particles on the exteriors of the crystallites (Figure 6a). Low-temperature (375 K) reduction brings about a very small amount of metal aggregation on the crystallite exterior (Figure 6b). Upon oxidation at 625 K (Figure 6c) no increased metal accumulation on the external surfaces is evident; however, rereduction brings about a marked increase in the number and size of the Ru metal particles observed in the TEM photographs (Figure 6d). Recall that it was at this stage of catalyst treatment that the temperature-programmed oxidation/reduction measurements (Figure 4) indicate the presence of large Ruo species, which are very difficult to oxidize. In addition, after rereduction the xenon N M R peak shifts upfield toward the resonance position of unexchanged Na-Y zeolite. The '29XeN M R results support the conclusion that large ruthenium particles are formed outside of the zeolite cages upon rereduction; however, under these conditions a significant portion of the ruthenium is still inside the cages and thus able to interact with the xenon gas. Reoxidation of the catalyst at this stage (sample f) again results in little shift of the xenon resonance; thus, the number density of ruthenium species within the faujasite cages is essentially unaltered. The total oxygen uptake (Figure 4,d)
sample b. Ru-Na-Y c. Ru-Na-Y
treatment evac 575 K evac 575 K, reduced 515 K, evac 725 K evac 575 K, reduced 575 K, oxidized 575 K, evac 725 K
Xe shift" 0.0 12.2 25.2
" Error in chemical shift = f l . 5 ppm, which is the maximum possible error from all sources including both error in Xe uptake and NMR measurements. Shifts are in ppm vs. Na-Y at constant [Xe]. indicates that this reoxidation is incomplete. However, the behavior of the xenon shift is consistent with the trends observed thus far; Le., oxidation is not accompanied by any significant Ru migration. After all of the above successive treatments a significant shift of the xenon resonance due to Ru species is still evident. The question arises as to whether the environment inside the ruthenium-exchanged zeolite supercages might, under more severe treatment, closely approach that of the unexchaned Na-Y sample. Upon high-temperature reduction and evacuation at 725 K for 12 h (Figure 5 and Table 11) the xenon chemical shift moves to within 6 ppm of the xenon resonance in the Na-Y sample at constant [Xe] (sample g). The T E M results (Figure 6e) show that further thermal treatment of these reduced ruthenium catalysts gives rise to both more and larger metal particles (0.1-0.3 nm in diameter) on the crystallite exterior. Thus, high-temperature treatment of reduced Ru-Y zeolites for extended periods of time results in a nearly complete removal of ruthenium from the interior of the faujasite supercages. All of the TEM results follow closely the '29Xe N M R results, but only the xenon experiments provide detailed insight into the environmental changes occurring inside the zeolite supercages. The question of when migration actually occurs also arises. Does the Ru metal migrate out of the zeolite only during reduction, or are reduced species generally more mobile and, thus, prone to migration at elevated temperatures? Table 111 presents the results obtained when two different samples are evacuated at elevated temperatures (725 K) for 12 h. The first sample (sample b) has been reduced at a moderate temperature (475 K), and the second (sample c) has been similarly reduced and then oxidized at 575 K. After evacuation at 725 K, the xenon in the reduced sample resonates only 12.2 ppm downfield from xenon in Na-Y at constant [Xe], while the xenon in the oxidized sample afer evacuation resonates at 25.2 ppm, 13 ppm further downfield than the sample in the reduced state. The reduced ruthenium species has migrated significantly, while the oxidized ruthenium shows no migration after in vacuo treatment at 725 K for 12 h. Thus, we conclude that reduced Ru has a larger diffusion coefficient at high temperature than RuO, and migrates to the exterior of the zeolite under thermal treatments either in vacuo or under hydrogen. Verdonck et aL6 found that, upon a high-temperature treatment of a well-dispersed Ru-Y zeolite catalyst in oxygen followed by reduction in hydrogen, large ruthenium particles were present on the exterior surfaces of the zeolite. It was felt that the migration of ruthenium to the exterior of the zeolite occurred during the oxidation step. This conclusion was based upon the following observations: (1) large RuO2 particles are always easily reduced to large ruthenium metal particles, but large ruthenium metal particles cannot be totally oxidized to RuO,; (2) the 0,:Ru uptake upon oxidations changed for successive oxidations, being smaller ratio was invariant for different for larger n while the H2"+l:OZn n cycles. On the basis of these observations, the authors felt that in order to obtain large metal particles on the outside of the zeolite cages, they must result from the reduction of large ruthenium oxide particles. Once these large ruthenium particles are formed they cannot be totally reoxidized to R u 0 2 , resulting in a lowering of the 02:Ru uptake measured for successive cycles. The RuO, which is formed during successive redox cycles (n) may always be reduced to Ru metal; thus, the H2"+':02" ratio remains constant. Our uptake measurements are in nearly total agreement with those of Verdonck et a1.;6 however, our '29XeNMR results and
4028 The Journal of Physical Chemistry, Vol. 91, No. 15, I987
Shoemaker and Apple
b
a&*
b
d -
L
1
3
e H 100 nm
Figure 6. Transmission electron micrographs of Ru-Na-Y zeolite at various stages of treatment: (a) evacuated at 575 K (b) reduced at 375 same as (b) oxidized at 625 K (d) same as (c) r e d u c e d at 400 K (e) same as (d) + reduced further at 725 K.
+
+
elmicrographsclearly show that largescale migration occurs when ruthenium is in a reduced state and not during oxidizing conditions. Rather, we propose that oxidation results in the formation of Ru02 contained inside the faujasite cages. This oxidation is essentially total, and the large downfield shift of the Xe resonance indicates that the number density of rutheniumcontaining particles is comparable to that before oxidation. Therefore, at this stage the catalyst contains highly dispersed Ru02 particles. These particles are resistant to migration under hightemperature evacuation. Upon exposure to hydrogen at elevated temperatures, ruthenium metal and water are formed. When it is in a reduced state at high temperatures, the ruthenium has a larger diffusion constant than the Ru02and thus migrates to the exterior of the zeolite crystals. The larger diffusion coefficient may be due to the smaller size of the particle after the removal of the oxygen. The reduced ruthenium is able to migrate through the zeolite channels where it deposits on the external surface of the crystal.
K,(c)
Consider, finally, what one might expect to see in the xenon NMR spectrum at some point of intermediate metal migration. If some of the supercagescontain no metal while others do, then two different peaks might be observed in the NMR spectrum-one at the position for xenon in Na-Y zeolite and another peak shifted downfield by the presence of Ru. Such spectra containing multiple peaks have been observed in other metalzeolite systems.'** Our spectra, however, never show more than one peak; therefore, the following experiment was performed. Two samples, a Ru-Na-Y zeolite and an unexchanged Na-Y zeolite, were placed together in an NMR tube in such a manner as to ensure that the two were well separated. This combined sample was then exposed to hydrogen and evacuated at 575 K. Xenon gas was admitted at room temperature to an equilibrium pressure of 300 Torr, and the ISXe spectrum was recorded (Figure 7a). Two peaks are readily apparent, one due to xenon in the Na-Y portion (82.2 ppm with respect to xenon gas extrapolated to zero pressure) and one corresponding to xenon in the Ru-Y fraction (94.6 ppm). The
Redox Behavior of Ruthenium in Zeolite Y
The Journal of Physical Chemistry, Vol. 91, No. 15, 1987 4029 to all metal-exchanged zeolites. Conclusions
;lo
1bo
40 PPM
io
io
Figure 7. ‘29XeNMR spectra of a mixture of Na-Y zeolite and reduced Ru-Na-Y zeolite: (a) samples separated; (b) samples mixed.
sample was then gently mixed with a stirring rod, returned to the 5-mm tube, and exposed to xenon gas, and another lz9XeN M R spectrum was recorded (Figure 7b). We see that while two separate resonances were observed before mixing, almost complete coalescence occurs upon physical (macroscopic) mixing. This behavior is reminiscent of chemical exchange, in this case by the diffusion of xenon between Na-Y zeolite and Ru-Y zeolite where the inherent chemical shifts differ by over 12 ppm (at constant pressure). The fact that physical mixing of the two components brings about this phenomenon indicates that intercrystalline diffusion of xenon gas is taking place on a time scale on the order of the inverse frequency separation of the resonance positions of Ru-Y and Na-Y zeolite. At the magnetic field strength used here this corresponds to 1/(12.4 ppm) X 55.3 Hz/ppm) = 1.5 ms. Assuming an average intercrystalline spacing of 1 pm, we arrive at a diffusion constant of order 10” cm2/s for xenon at 300 Torr. Intercrystalline diffusion constants are typically 1-2 orders of magnitude smaller than intracrystalline diffusion coefficient^;'^ therefore, intracrystalline diffusion must be very rapid for the reduced Ru-exchanged zeolites studied here. The xenon atoms sample many supercages within the zeolite crystal on the N M R time scale. It is well-known that diffusion constants in zeolites are strong functions of the type of exchanged cation, as well as the sorbate loading;I3therefore, this behavior may not be universal (13) Barrer, R. Zeolites and Clay Minerals as Sorbents and Molecular Sieves; Academic: London, 1978; Chapter 6 .
A comparison of the chemical shifts of xenon gas sorbed in Ru-exchanged zeolites with that in Na-Y zeolites for a given xenon concentration provides valuable insight into the local environment inside the zeolite supercages. Ruthenium-exchanged Y zeolite is reducible at room temperature and forms highly dispersed Ruo located in the zeolite cages. The ruthenium remains in the faujasite cages upon evacuation at moderate temperatures. However, migration of the metal occurs upon evacuation or reduction at high temperature. Ruthenium residing in the zeolite supercages remains there upon oxidation, forming R u 0 2 which is resistant to migration upon evacuation at high temperature. Subsequent reduction causes further Ru migration to the exterior of the zeolite crystallites. Temperature-programmed reoxidation indicates a bidisperse distribution of ruthenium following rereduction. This is consistent with the xenon N M R results which show that much of the ruthenium is still dispersed throughout the supercages, though the total number density of Ru particles has decreased during the reduction of RuO,. A general trend is observed whereby reduced ruthenium is migratory, but oxidized ruthenium is immobile. TEM photographs confirm that only reduced Ru species migrate, and this migration is to the exterior of the zeolite crystal. The migration process is not reversible. Our experiments indicate that both intercrystalline and intracrystalline diffusion of xenon gas take place in these samples on the N M R time scale. The xenon N M R line widths scale with the magnetic field strength in Ru-exchanged samples which have been reduced. This is indicative of a dispersion of shielding environments in the metal-exchanged zeolite. This dispersion stems from inhomogeneity in the distribution of the metal inside the zeolite crystals. The N M R line width of xenon in the unexchanged Na-Y samples is small and field independent; thus, there is little dispersion of shielding environments throughout the Na-Y zeolite.
Acknowledgment. This work was supported in part by a grant from the Research Corporation and a grant from the donors of the Petroleum Research Fund, administered by the American Chemical Society. We thank Dr. Kit Lee for the electron microscopy work.
