The Journal of
Physical Chemistry
0 Copyright, 1990. by the American Chemical Society
VOLUME 94, NUMBER 24 NOVEMBER 29, 1990
LETTERS Characterization of Pt Particle Interaction with L-Zeolite by X-ray Absorption Spectroscopy: Binding Energy Shifts, X-ray Absorption Near-Edge Structure, and Extended X-ray Absorption Fine Structure Brian J . McHugh, Gustavo Larsen, and Gary L. Halter* Department of Chemical Engineering, Yale Unirersit!3.P.O.Bo.u 2159. Yale Station. .?'mt i a i w i , Connrrticirt 06520 (Receired: June 5. 1990; I n Final Form: October 5. 1990) We have used X-ray absorption to measure the binding energy. X-ray absorption near-edge structure (XANES), and extended X-ray absorption fine structure (EXAFS) of Pt particles in L-zeolite saturated with H2 at room temperature and after desorption of H? in He. The desorption of H2 causes the particles to disorder or perhaps breakup such that the coherence required fur E X A F S is lost. The shift (with respect to bulk metal) of the binding energy of the Pt in the zeolite, not stabilized by chciiiiaorbcd HI, suggests an interaction w i t h the zeolite support (and may reflect electronic charge transfer to the Pt).
Introduction We have recently reported on metal-support effects in Pt/Lzcolite catalysts as probed by competitive hydrogenation of toluene and benzene.' W e concluded from the ratio of the toluene/ benzene adsorption constants, extracted from a kinetic analysis, that Pt particles in L-7eolite appear to be electron-rich relative to Pt/SiO, or Pt/Y-zeolite. We now wish to report on a physical probe of metal-support interaction in the same system. X-ray absorption spectroscopy of the Pt L,,, edge has been used to determine the binding energy. X-ray absorption near-edge structure (XANES). and extended X-ray absorption fine structure ( E X A F S ) of particles i n L-zeolite. The conventional technique used to analyze the electronic state of dispersed metals on supports has been XPS2s3 However, it is not possible to uncouple the final-state effects (screening of ( I ) Larsen. G.: Haller. G . L. Carol. Lett. 1989. 3. 103. ( 2 ) Antoshin. G.V . : Shapiro. E. S.: Tkachenko, 0.P.: Nikishenko. S. B.: Ryashentseva. M . A.: Avaev. V . I.:Minachev. Kh. M. Proceedings ofthe 7th lnt(,rnotiono/ Conpre.ss on Cata/ysis: Seiyama. T.. Tanaba. K.. Eds.: Elsevier: Amstcrdnm. 1981: Part A, p 302. ( 3 ) Hui/ingLi. T.: Prins. R. Slud. SurJ Sci. Catal. 1982. I / . I I
0022-3654/90/2094-862l$02.50/0
charge) from initial-state effects (electronic interaction). I n small particles, the relaxation of the final core hole states will be smaller than that in larger particles, because of less effective screening of the core levels, and this appears to overshadow any charge transfer between the small particle and the oxide support. Consequently, Pt particles on oxide supports have always been observed to have a higher binding energy than bulk Pt and, on a support such as AI2O3,the shift inbinding energy with dispersion (particle size) is almost linear.3 In the case of zeolite supports where the particles of interest are inside the small pores, there will be an additional complication that the spectrum may be dominated by the larger particles on the external surface rather than those of more interest, the small particles wiihin the zeolite, because of the limited electron escape depth which is typically of order 1-3 nm.4 In principle, one can measure the binding energy of an element from the shift in the X-ray absorption edge. This is not usually ( 4 ) Delgass. W . N.; Haller, G . L.: Kellerman. R.: Lunsford. J . H. Spectrosc'op)' irt Heterogeneous Cata!,sis;
Academic Press: New York, 1979; p
289.
0 I990 American Chemical Society
8622
T h e Journal of Phj'sical Chemistry. Vol. 94. Xo. 24, I990
reported since the monochromator resolution used for EXAFS and X A N E S is typically several electron volts and most monochromators have low absolute energy definition because of mechanical irreproducibility of the setting of the crystal angle. However. it is practical to improve resolution by limiting angular divergence of the X-ray beam by decreasing the size of the monochromator slits until resolution is quite comparable to that of typical XPS monochromators and the absolute energy can be dctcrmined by using an internal reference. i.e.. by placing a Pt foil in scrics with the Pt/L-zeolite catalyst. With detectors placed before the sample, between the sample and the foil. and after the foil. one can mc;isurc thc binding energy of the Pt (arbitrarily taken A tlic inflcction point of the cdgc) of the catalyst relative to tlic Pt foil to Libout two-tenths of an electron volt. Assuming the cdgc h a p c doc5 not bar) much. it is practical to nicaburc relative binding cncrgics with an accurary quitc comparable to that oblaincd in ;I tbpical XPS experiment. (The use of an internal standard h:is bccn previously reported with X A N E S . but apparently not to estimate explicitly changes in binding energy.5) Howcvcr. the X-ray absorption experiment has the advantage that one obtains qualitative information on the partial density of states at tlic Fcriiii lcvcl froiii the XANES' and thc particlc morpholog) from the L X A t S . ' Okaiiioto ct haic used XPS to demonstrate that thc clectronic structure of xolites is correlated primarily with AI/Si ratio and the clcctroncgntivity (charge to radius ratio) o f thc cations that b;il:incc thc framework charge. Because the binding energies of ;ill clcmcnts. i.e.. 0 Is. Si 2p. AI 2p. and Na Is, decrease linearly w i t h thc .AI/Si rtitio. i t was suggested that charge is delocalized ovcr thc uholc friiincwork. Whilc they state that zeolite structure is IC\\ importnnt. ;I comparison of the 0 Is binding energy of X-7colite with Y-zeolite demonstrates that oxygen in the X-zeolite is systematicnlly shifted to lower binding energy for any given cation: i t . . the oxygen lattice of X-zeolite is the more basic relative to Y-7colitc. Thc lattice oxygen of L-zeolite is expected to be even more biisic th:in thiit of X based on chemical reaction probes9 and i t ha5 bccn huggcstcd t h a t Pt particle5 in L-zcolitc arc clcctronrich.' l o This suggests that the Pt/L-zeolite system can be an interesting system to investigate metal-support effects by shifts in the X-ray absorption edge since the interaction might result i n shifts to lower binding energy with respect to bulk Pt if this is not compromised by small particle/relaxation shifts to higher binding energy relative to bulk Pt. Experimental Section
The preparation of the Pt/MgKL has been previously described.' Briefly summarized, the KL-zeolite was ion exchanged with ci.r-Pt(NH3)?Cl2.calcined, and reduced a t 623 K , and the protons produced in the reduction and the K+ ions in the channels were subsequently ion exchanged to produce Pt/MgKL catalysts of two weight loadings. 0.78%and 3.58%). Hydrogen room temperature adsorption isotherms extrapolated to zero pressure resulted in H / P t equal to I . I2 and I .05 for the two loadings. respcctively. Thc lower loading is the same catalyst described in ref I . The two catalysts, 0.78 and 3.58 wt % Pt/MgKL, were placed in an X-ray absorption cell and in situ rereduced with H? a t 573 K for 2 h. Chemisorbed H! was removed at the reduction tempcrnturc by flowing He and the samples cooled to liquid N, temperature in He and the X-ray absorption spectrum recorded. Tlic Iic n m rcphced by Hz >it room temperature and the X-ray absorption spcctruni again recorded ;it liquid N, temperature. Reversibilitb tu tlic reduced. naked metal was checked by desorbing
Letters Pt LIII Xanes of Various Samples
5
250
11560
11575
11580
Energy (eV) Figure 1. The XANES of Pt foil (-), PtO, (A).and 0.78 w t % Pt/ M g K L i n H e ( 0 )and after chemisorption of H, (+). TABLE I
H, chcmisorpn
ample
inflcction pt. keV
PI roll PtOy HZO 3 587 P t / M g K L
I 1.5660 I 1.5678
no 4 Cb
I 1.5655 I 1.5670 1 1.5653
no
0.78% Pt/MgKL
no" noa Yes "Chemisorption of H 2 and desorption in He at 573 K between these two runs.
I 1.5655' I 1.5655" I 1.5665
was performed
the H 2 in flowing He a t 573 K or reduction a t 573 K followed by flowing He at 573 K to return to the initial state. No significant differences were observed for these two procedures and both returned the catalyst to its initial state. The X-ray absorption spectra were obtained on line C-2 of the Cornell High Energy Synchrotron Source. (Some preliminary spectra were obtained at the Kational Synchotron Light Source on line X 18-B.) The monochromator resolution is approximately I .5 eV at 11.56 keV (the Pt L,,, edge) with the slits set to 0.25 mm. Flowing N 2 was used to monitor the total photon flux incident on the sample and flowing Ar was used in the detector between the cell and the Pt foil (0.008 mm) and after the Pt foil reference. The window material on the cell holding the sample was Kapton to limit the number of photons absorbed by background material. The absorbance of the samples and the reference foil were about 2. X-ray absorption spectra depend on the mass areal loading cf each of the constituents (edge metal and other atoms) in the sample, the ionization detectors, and the characteristics of the beam. The normalized absorption spectra presented here arc the result of a linear least-squares fit of the original spectra to the sum of the product of the mass areal loading of the edge metal and its mass absorption coefficients and a polynomial of variable order in terms of photon wavelength, i.e.,
+ + BX + CX2+ DX3 + ... + EX"
( p t ) l o t , e x=p (pt)dge,vjct A (!J'l)tot.erp
= A
( 5 ) Niemann. W.; Clausen. B. S.; Topsoe, H. Cural. Letr. 1990, 4. 355. ( 6 ) Burt. I . C . .I.A d r Caral. 1986. 34. 203. ( 7 ) Teo. B. K. EXAFS: Basic Priniciples and D a f a Ana!,,sis: SpringerVcrliip: Berlin. 1986. ( X ) Oknniolo. Y . . Opi%.i.M : Mucra~.;~. A , : Imannka. T J Carol. 1988. 112.427 ( 9 ) Bernard. J . R . Pror. 5 f h Inr. ConJ. Zeolifes 1980, 686. ( I 0 ) Besoukhanova. C.: Guidot. J.: Barthomeuf. D.J . Chem. Soc.. faraday Trrr,i\ I 1981. 77. I 5 9 5
11570
11565
+ BX + CX2 + DX3 + ... f
EX"
+ F(pL/p)e,jge,vicl
The linear coefficients A through E can be used to construct the background and F represents the mass areal loading of the edge metal. The diagrams in Figure I are the X-ray absorption spectra fit to the Victorecn mass absorption coefficients of the Pt edge. The samples were run simultaneously with a Pt foil reference and thc > h i f t s ilrc reported relative to the foil's Pt L,,, edge at the i n flcct ion point.
The Journal of Physical Cheniistry, Vol. 94, No. 24, 1990 8623
Letters
' 7
Modelled and Measured Pt LIII EXAFS
i
t
5
Phase Shifted Coordination Distance
sE
0.06
I
I
g,,
10
15
Photoelectron Momentum ( I / & Figure 3. The EXAFS magnitude functions for 0.78 wt % Pt/MgKL in H2: experimental EXAFS (-). fitted model (+). X-ray absorption spectra were obtained at liquid N , temperature. (These spectra were obtained at NSLS and used the sanie samples used at CHESS from which the data in Table I and Figure 2 were obtained because the CHESS spectra were only obtained out to 10 .+-I due to a monochromator malfunction. The CHESS EXAFS over the limited k space are qualitatively similar.)
(A)
-
% c
12
1
0.02.
I.
0.06
Y
before cooling in He to obtain the X-ray absorption spectrum. This is displayed in Figure 2 showing the Fourier transform magnitude of Pt foil, Pt/MgKL with chemisorbed H 2 and after desorption of H2 into flowing He. (The absorption spectra were obtaincd a t liquid nitrogen temperature.) When the catalyst is exposed to H2 at rmm temperature, small particles of Pt of ordered structure arc re-formed. The EXAFS back-transformed magnitude function of the Pt particles of the 0.78% P t / M g K L (between 1.79 and 3.23 A) obtained by a fitting routine using Pt foil as rercrcncc (between 1.68 and 3.39 A. we assume a 2.746-A Pt-Pt and coordination number of 12) is shown in Figure 3. The coordination number, Pt-Pt distance, and Au2 of 6.9 A, 2.73 A, -0.0036 A', respectively, are obtained from the model. The coordination number is almost identical with that predicted for a spherical particle with 30 atoms which would have a diameter of 12 A." This is about equal to the free dimension of the bulge in the L-zeolite channel which is 12.6 A in the naked framework. Cations in the channel would reduce this dimension, but the cations are believed to withdraw from the channels after dehydration'* and the coordination number will also reflect a small fraction of larger particles outside the zeolite pores. A similar but less dramatic change in morphology of Pt particles supported on SiO, has been previously reported upon desorption of H, and attributed to Pt interaction with oxide of the support in the absence of chemisorbed H2.l3sl4 Similarly, Pt particles in Y-zeolite have been reported to give an electron radial distribution curve of lower intensity and shifted Pt-Pt distance in He relative to H2.I5 I t must be emphasized that the previously reported decrease in the EXAFS magnitude for Pt/SiO2l3.l4and the decrcasc in the Pt-Pt distance for naked Pt particles in Y - z e ~ l i t e , ' ~ while apparently similar in kind, are much less dramatic than the change in EXAFS magnitude exhibited for Pt/MgKL in Figure 2. Unfortunately, the physical origin of the apparent loss of structure in the Pt particles in L-zeolite, whcn they are not stabilized by chemisorption, does not yield any simple interpretation. There are two extremes one might consider: the breakup of the
E 2
4
6
8
Phase Shifted Coordination Distance (A) Figure 2. The Fourier transform magnitude of k2-weighted EXAFS of (;I. top) Pt foil. ( b . niiddlc) 0.78 w t 57 Pt/MgKL with chemisorbed H2. and ( c . botloni) aftcr desorption of H, into He at 573 K. The absorption spcctrti \\crc obtaincd :it liquid nitrogen tcnipcraturc.
Results and Discussion Thc X A N E S of Pt foil. PtO,. and 0.78 wt % Pt/MgKL in He and after Chemisorption of Hl are presented in Figure 1 . These have been mass areal normalized and the inflection point has been positioned with respect to Pt foil edge recorded simultaneously (except in the case of PtO, which was not recorded with a reference and niay bc somewhat in error). A summary of binding energies for the two catalysts and two references is given in Table I . Thcrc is little evidence for a fixed Pt-Pt distance in the Fourier magnitudc of the Pt/h.lgKL which had been in situ rcduced in H2 and the til rcniovcd in flowing He at the reduction temperature
( I I ) Kip,
B. J.; Duivenvoorden, F. B. M.; Koningsberger, D. C.; Prins, R.
J Coral. 1987.. 105. 26. . .. ~
( I 2) Breck, D. W . Zeolite Molecular Sieces: Siruciure, Chemistry and l%e: Wiley: New York. 1974; Chapter 2. p 114. ( 1 3 ) Lytle. F. W.. Greegor. R. B.. Marques, E. C., Sandstrom. D. R.. Via, G. H..Sinfelt, J. H., J . Caial. 1985, 95, 546. (14) Lytle, F. W . ; Greegor. R. B.; Marques, E. C.; Beibesheimer, V. A.; Sandstrom, D. R.: Horsles, J . A.; Via. G . H.; Sinfelt. J . H. A C S S y. m p. . Ser. 1985, NO.288. ( 1 5 ) Gallezot. P.: Bergeret. G . J . Caral. 1982. 72. 294.
Letters
T h e Journal of Physical Chemistry, Vol. 94, No. 24, I990
8624
I
~
5
10
Photoelectron Momentum
.
_
_
15
(A-1)
Figure 4. The normalized k-weighted EXAFS of Pt/BaKL (-) and Pt/MgKL (---) after desorption of H, into He. The samplc of Pt/ B u K I . is the wmc o n c ;is described in ref I and its prcparntion was idcntic;il with t h a t of Pt/MgKL except for the final ion exchange which
cniplo!cd R;i2+. particle5 such that the Pt atoms wet the zeolite walls or the retention of PI-PI bonding but with a distribution of Pt-Pt bond lengths \\hich is sufficicntly great that cohercncc in the Pt backscattering is lost. Ncithcr sccms intuitively likely since the former would rcquirc that the van der Waals' interaction energy of Pt w i t h the 7colitc walls exceed the energy of Pt-Pt bonding and the lattcr would rcquirc some mechanism beyond shortening of the surfuce Pt-Pt bonds relative to the bulk. the effect that was idcntificd to explain the shift in the Pt-Pt distance obscrvcd in the radial distribution curvc.15 Thcrc i.s ;il\rii>s the possibility that there was an oxygen impurity in the He uwd to obtain thc results shown in Figure 2 and that disorder or breakup of the smnll particles by oxidation occurred on cooling in He. Wc rulc this out since oxidation (and relaxation shift due to snioll pnrticlcs) would require an edge shift to higher binding energy with rcspcct to bulk Pt and we observe a shift of about 0.5 cV to loner binding cncrgy. When exposed to H 2 at room tciiipcraturc. the EXAFS indicates that small particles are formed and the) ;ire obscrvcd to have a binding energy shifted to liiphcr cncrg) bq about 0.5 cV for 0.78% Pt/MgKL and 1 .O cV I'or 3.58',; Pt/MgKL, relative to bulk Pt, in accord w i t h previous XPS studies.' I t ih cxpcctcd t h a t the XANES intensity in the first peak beyond thc edge of the smnll Pt particles covered by chemisorbed H 2 will be l c than ~ that of bulk Pt. This has been reported prcviouslq and i t \ w s noted that the arcti of thc peak just beyond the edge decreases further ;is the tcmpcraturc is i n c r ~ a s c d . ' ~Of more interest Iicrc is tlic greater intensity for the Pt/MgKL i n He rchtivc to the foil. Qunlit;itivc argunicnts to rationalize the contraction of thc outer layer of clean metal surfaces suggest that for the surfucc ;itoms the bonding electrons are partly shifted from the bonds brokcn to form the surface to the remaining unbroken bonds. thcrcbq increasing thc chargc content of the lattcr and decreasing thal of the surfricc atoms.I6 Quantum chemical results of Mcssiiicr C I ;it. confirm that surface atoms of small nictal clusters arc intrinsically clectron-poor compared to those in the bulk." Put another way. as the Pt particle becomes smaller, a larger fraction of the Pt atoms (those on the surface) will approach the atomic configuration. [Xe]4fI45d96si,which has more empty d states than the bulk (because as the 6s band is formed. its energy is lohcrcd a n d i t overlaps the 5d band so that the average bulk configuration has less than a full hole i n the d band). These considerations alone might lead to the expectation that small Pt (16) Soinorjai, ti. A . C'hrwiJtr),iii Tno Dinimsions: Surfucrs; Cornell Univcrsity Pres\: Ithacn. N Y . 1981: p 143. ( 1 7 ) Mcaaincr. R. P.: Knudson, S. K.; Johnson. K . H . ; Diamond. J . B.. Y a n g . C . Y . Phi,r Rei. 6 1976. 13. 1396
particles with clean surfaces would have a greater near-edge intensity a s observed for Pt/MgKL in He (relative to Pt foil). However. as discussed in the Introduction, the more atomic like average Pt in small particles would also be expected to have a greater binding energy than bulk Pt. Thus, the greater intensity for the Pt/MgKL in He relative to the foil which also has a binding cncrg) less than the bulk is unexpected. While we can offer no explanation a t this time, one possibility which we think should be considered is chargc transfer from the zeolite lattice acting as a Lcuis basc. This would account for the lower binding energy and, if it resultb in Pt which is more atomic like or i f rehydridization of the 6s states of the Pt occurs that would allow transitions into these new states (dipole forbidden for atomic or s band states), this could also account for the increased intcnsit). Perhaps another possibility to explain the results shown in Figure 2 is that the Pt particle has been strongly polarized and _ structurally disordered by interaction with cations in the zeolite channel. If this were the case. the lowering of the binding energy might be attributed to photoejection into/through the cation and the static and/or dynamic disorder of the particle would destroy the EXAFS. This picture appears to us to be less intuitive, but something of this sort cannot be ruled out by our current experiments. We have concentrated on the X-ray absorption characterization of Pt-7colite interaction in the Pt/MgKL system because we have the most extensive data base for this catalyst, but i t is expected to have some acidity which is undesirable for reforming reaction on these catalysts." However. the Ba form of this catalyst is known to be nonacidic'* but this appears not to be a determining Fmor in thc Pt-zeolite interaction. This is demonstrated in Figure 4 where the normalized k-weighted EXAFS of Pt/BaKL and Pt/VgKL. after desorption of H, into He are compared. That is, i n both cases the Pt is either disordered or disaggregated in both catalysts but Pt particles are formed when exposed to H, a t room temperature. (The sample of Pt/BaKL is the same one a s described in ref I and its preparation was identical with Pt/MgL cxcept for the final ion exchange which employed Ba2+.) Several experiments are suggested by these results; e.g., the Pt/MgKI. sample in He might be exposed to benzene to see how this clectron donor (but weaker reducing agent relative to H,) \ + i l l compctc \+ith the zeolite lattice. Complexes of Pt atoms (if they exist) with benzene could be characterized by UV-visible and infrared spectroscopy as well as X-ray absorption spectroscopy. Dircct comparison to other metals, e.g., Pd, and other zeolites, c.g.. Y-zcolitc. are also obviously important. These experiments arc i n progress. Summary
U'c draw two conclusions from these experiments. Firstly, the u\c of the internal standard procedure to extract binding energies h;is bccn demonstrated and is recommended in future X-ray absorption spcctroscopic studies to augment XANES investigation of clcctronic structure. Secondly, these results suggest such a strong interaction of the L-zeolite lattice with small metal particles that they disorder or disaggregate in the L-zeolite channels in the absence of stabilization by chemisorbed H,, and the edge shift indicates that this may be interpreted as if the L-zeolite were acting ;IS a Lewis base but the energetics of this process are not u n dcrs tood . Acknowledgmenr. We gratefully acknowledge access to beam time a t the Cornell High Energy Synchrotron Source and the
hational Synchotron Light Source. Financial support was provided by KSF. Continued support for program development for our X-ray absorption work has been provided by the Office of Basic Energy Sciences, DOE. We also thank one of the referees, Thomas Hughes, whose continued questioning of our loose interpretation has improved our presentation. but of course, we claim responsibility for all those deficiencies that remain. i 18) Hughcs. T. R.;Buss, W. C.; Tamm. P. W.; Jacobson, R . L. Proc. 7th Inr Congr Zeolirc.7 1986. 7 2 5 .
