Preparation of Highly Dispersed Pt Particles in Zeolite Y with a Narrow ...

22 downloads 494 Views 337KB Size Report
Research on metal–support interaction calls for highly dispersed metal particles inside the ... leads to particles with an EXAFS Pt–Pt coordination number of 5–.
Journal of Catalysis 203, 307–321 (2001) doi:10.1006/jcat.2001.3337, available online at http://www.idealibrary.com on

Preparation of Highly Dispersed Pt Particles in Zeolite Y with a Narrow Particle Size Distribution: Characterization by Hydrogen Chemisorption, TEM, EXAFS Spectroscopy, and Particle Modeling J. de Graaf, A. J. van Dillen, K. P. de Jong, and D. C. Koningsberger1 Department of Inorganic Chemistry and Catalysis, Debye Institute, Utrecht University, P.O. Box 80083, 3508 TB Utrecht, The Netherlands Received January 30, 2001; revised May 22, 2001; accepted July 7, 2001

Research on metal–support interaction calls for highly dispersed metal particles inside the zeolite with a narrow particle size distribution. Pt particles thus dispersed in zeolite Y are obtained in this work using both ion exchange and dry impregnation techniques. For both preparation techniques a very low heating rate during calcination turns out to be essential. This low heating rate of 0.2◦ C/min leads to particles with an EXAFS Pt–Pt coordination number of 5– 6, i.e., particles of 13–20 atoms, i.e., smaller than 1.1 nm. A rate of 1◦ C/min as used in the literature already leads to a bimodal particle size distribution. About 5 wt% large particles of 4–9 nm is (partially) occluded in the zeolite. Ninety-five weight percent of the Pt particles have an average size of 1–1.2 nm. The necessity to use a very low heating rate is explained by the slow desorption of water and ammonia from the microporous zeolite in combination with the stabilization of mobile Pt species by the cavity walls of the zeolite. The particle size distribution reported could be obtained only by combining the results of hydrogen chemisorption, TEM, EXAFS, and modeling of Pt particles. °c 2001 Academic Press Key Words: Pt in Y; particle size distribution; rate of calcination; impregnation; ion exchange; TEM, H/Pt, EXAFS.

INTRODUCTION

Supported noble metal catalysts are used in a large number of commercially important applications, including hydrogenation, dehydrogenation, naphtha reforming, isomerization, hydrocracking, oxidation, automotive exhaust catalysts, and fuel cells (1). Pt/Pd on zeolite Y catalysts are applied for such processes as deep hydrodesulfurization (HDS) and hydrodearomatization (HDA). The catalytic activity of supported metal catalysts for hydrogenation and hydrogenolysis is known to depend strongly on support properties such as (i) acidity/alkalinity, (ii) number and type of cations (iii) Si/Al ratio, and (iv) presence of extraframework Al (AlEF ). To study the metal–support interaction in Pt/Y catalysts and to be able to discriminate between metal–support and 1 To whom correspondence should be addressed. E-mail: d.c. [email protected]. Fax: +31-302511027.

particle size effects, a narrow particle size distribution of small platinum particles fitting the micropores of the zeolite is needed (2). Furthermore to study the metal–support interaction as a function of the different support compositions (Si/Al ratio, Na+ , H+ , AlEF ), the same narrow particle size distribution has to be present independent of the different support compositions. Several procedures for the preparation of small metal platinum clusters from ion-exchanged zeolite Y are reported in the literature and the processes occurring during oxidation, decomposition, and reduction of the Pt(NH3 )2+ 4 complex have been investigated (8). To prepare welldispersed platinum particles in zeolite Y, application of the Pt(NH3 )2+ 4 precursor ion via ion exchange of the zeolite is recommended (3–6). However, the effect of the method of application (ion exchange or impregnation) on the preparation result has not been extensively investigated. To obtain metallic platinum particles, ultimately the precursor needs to be reduced. It is well known that treatment of the Pt(NH3 )2+ 4 ion-exchanged zeolite with hydrogen results in a very low platinum dispersion (3). The metal dispersion is markedly improved when the applied Pt complex is decomposed (7) or calcined prior to reduction (3). The resulting metal dispersion is reported to depend strongly on the calcination conditions. The best results are obtained when small amounts of ion-exchanged zeolite are calcined in a fixed bed reactor using a high oxygen-containing flow and a low heating rate. This was found both with zeolite Y and with ZSM 5 (3, 8, 9). For characterization of the local structure, size, and location of platinum in zeolite NaY a number of techniques have been used: EXAFS, hydrogen chemisorption, HRTEM, and 129 Xe NMR (10). With EXAFS analysis particle size determination is accurate only if the metal particles are small and the particle size distribution is narrow. When the particle size distribution is broad or bimodal the larger particles dominate the average coordination number of the first Pt– Pt coordination shell. This means that the results obtained with EXAFS alone are not reliable. Analysis of (highresolution) transmission electron micrographs is unique in

307 0021-9517/01 $35.00 c 2001 by Academic Press Copyright ° All rights of reproduction in any form reserved.

308

DE GRAAF ET AL.

providing directly the platinum particle size distribution. However, with TEM a lower limit of 1- to 2-nm platinum particles in zeolite Y is met because of the thickness of the Y crystals which leads to a considerable scattering contrast (11–14). Although HRTEM can be used on zeolite samples that are sliced by microtomy (11–14), a major problem is the amorphization of the zeolite and sintering of the metal particles under influence of the electron beam (14, 15). In this article, the preparation of small platinum particles in zeolite (Y) is studied. Pt/NaY and Pt/H-USY are prepared by ion exchange or by impregnation of the zeolites with aqueous solutions of Pt(NH3 )4 (NO3 )2 . Special emphasis is attributed to the influence of the heating rate to reach the calcination temperature. To gain more insight into the influence of the microporosity of the zeolite on the finally obtained metal dispersions, a comparison is made with a mesoporous amorphous silica–alumina as the support. The platinum particle size distribution was determined by combining the results obtained with hydrogen chemisorption, TEM, and EXAFS. Very small Pt particles having 13–20 atoms per particle and of an average size smaller than 1.1 nm are prepared in NaY and H-USY from ion-exchanged and impregnated NaY and NH4 -USY samples. This is accomplished using a heating rate during the calcination procedure as low as 0.2◦ C/min. The particle size distribution on the ionexchanged samples is narrow and the metal is homogeneously distributed throughout the zeolite. With the impregnated zeolite catalysts also a small fraction of larger particles (>2 nm) is found on a part of the zeolite particles. It is also found that preparation of Pt on the mesoporous silica–alumina by impregnation results in 50,000 atoms per particle) by extrapolation. TEM analysis on Pt/Y has as a lower detection limit of platinum particles around a few nanometers. By combining TEM, EXAFS, and particle modeling the (volume) average size of the platinum particles that are not visible with TEM can be determined. The part of the particle size distribution visible by TEM is obtained by counting the number of projected platinum particles per size per area of zeolite crystal. The number of platinum particles per volume is estimated on basis of the assumptions that the zeolite crystals do have equal dimensions in all directions, and that the platinum particles in this zeolite volume are projected and distinguishable in the transmission electron micrograph. Using these results and the density of the zeolite particles, the part of the particle size distribution visible by TEM can be obtained. After integration of Eq. [1] by using the part of the platinum particle size distribution detected with TEM as described above and the established Pt loading, the following equation is obtained:

311

calcination procedure is varied from 1◦ C/min (as normally used in the literature) to a much lower value of 0.2◦ C/min. After calcination all samples were cooled down to RT and reduced in hydrogen at 400◦ C using a heating rate of 5◦ C/min. EXAFS Data Analysis and Results In Fig. 1A the experimental k 2 -weighted EXAFS function of catalyst Pt/NaY(1.1–IE–0.2) is plotted. The quality of the data is representative of all experiments. Figure 1B shows that the imaginary part and absolute part of the Fourier transform of the fit and athe experimental data fit very well between 1.6 and 3.1 A . In Table 3 the fit parameters of the EXAFS data analysis are listed. For all samples the variance of the fit of the imaginary part and absolute part of the Fourier transform is less than 1%,

NPt–Pt = NPt–Pt, not detected with TEM ∗ (1 − mass fractionTEM ) + NPt–PtTEM ∗ mass fractionTEM ,

[3]

where mass fractionTEM = fraction of Pt metal that is detected with TEM. From this equation NPt–Pt not detected with TEM can be obtained. There the mass fraction of the particles in the size distribution as determined with TEM is calculated by estimating the mass of the particles using their size and the modeled atom size function, A(size). The (average) coordination number of the particles in the size distribution as determined with TEM is calculated by using their size and the modeled first-shell coordination number size function, NPt–Pt (size). With the thus calculated NPt–Pt not detected with TEM the volume average particle size of the corresponding platinum particles is estimated with the modeled first-shell coordination number size function, NPt–Pt (size). The results obtained with hydrogen chemisorption are compared with the results calculated as described above to check the applied procedure. RESULTS

Preparation of Zeolite-Supported Pt Catalysts Table 2 gives an overview of the different zeolitesupported Pt catalysts. As supports NaY, H-USY, and amorphous SiO2 –Al2 O3 were used. Both ion exchange and impregnation with Pt(NH3 )4 (NO3 )2 as precursor were applied. The Pt metal loading is around 1 wt% for all samples. Different gases were applied during the calcination procedure (Ar, He, O2 , O2 /H2 O). The heating rate during the

FIG. 1. (A) Experimental k 2 -weighted EXAFS spectrum for Pt/NaY ˚ −1 ) of ex(1.1–IE–0.2); (B) Fourier transform (k 2 -weighted, 1k 2.5–14 A perimental data (solid line) and total (Pt–Pt + Pt–O) fit (dotted line).

312

DE GRAAF ET AL. a

around a distance of 1.2 A. This peak is due to scattering of the outgoing electron against the deep valence electrons of the X-ray-absorbing atom. The intensity of the FT AXAFS peak has been found to depend on the acidity/alkalinity of the support, and has been proven to be a promising tool in the study of metal–support effects (21). The first-shell platinum coordination number, NPt–Pt , (Table 3), appears strongly dependent on the treatment applied to the Pt/NaY catalysts. When calcination is omitted, a high platinum coordination number is found, indicating the presence of large platinum particles. When calcination is included in the procedure the platinum coordination number is much lower. Furthermore the heating rate used during calcination has a pronounced effect on the NPt–Pt coordination number; the lowest numbers, i.e., the smallest platinum particles, are found when using a heating rate of 0.2◦ C, with both ion-exchanged Pt/NaY and Pt/H-USY. This calcination procedure results in slightly larger NPt–Pt values on the impregnated zeolite catalysts, whereas the coordination number on the impregnated silica–alumina is similar that of the ion-exchanged zeolites. With all catalysts in this study, zeolite supported and silica-alumina supported Pt–O contributions are found at a distance of a 2.6 A. Hydrogen Chemisorption

˚ −1 , Pt–Pt FIG. 2. (A) Fourier transform (k 1 -weighted, 1k 3.5–14 A phase, and amplitude corrected) of difference file (raw EXAFS data minus Pt–O-fitted contribution) (solid line) and Pt–Pt fitted contribution (dotted line) for Pt/NaY(1.1–IE–0.2). (B) Fourier transform (k 3 -weighted, 1k ˚ −1 , Pt–Pt phase, and amplitude corrected) of difference file (raw 3.5–14 A EXAFS data minus Pt–O fitted contribution) (solid line) and Pt–Pt fitted contribution (dotted line) for Pt/NaY (1.1–IE–0.2).

indicating a high-quality fit. Moreover Fig. 2A and 2B show that the Fourier transform of the fitted Pt–Pt contribution agrees very well with the Fourier transform of the corresponding difference file (experimental EXAFS minus Pt–O contribution) in both k 1 and k 3 weighting, indicating a reliable decoupling of the Pt–Pt coordination number and the Debye–Waller factor (18). The k 1 -weighted Fourier transform (FT) of the fitted Pt–O oxygen contribution and the FT of the corresponding difference file (experimental EXAFS minus the fitted Pt contribution) are displayed in Fig. 3. The Fourier transform of the Pt–O contribution approximates very well the Pt–O peak in the FT of the difference file. Figure 3 shows also the atomic XAFS (AXAFS) peak

Table 4 lists the H/Pt ratios determined for the various catalysts. Both the ratios obtained in the first isotherm, H/Pttotal , and the ratios obtained after subtraction of the amount of removable hydrogen at 35◦ C, H/Ptstrong , vary significantly. However, the H/Pttotal ratio is for each catalyst about twice as high as the H/Ptstrong ratio, which shows that the correlation with Pt surface area is the same for both

˚ −1 , Pt–O phase FIG. 3. Fourier transform (k 1 -weighted, 1k 2.5–10 A corrected) of difference file (raw EXAFS data minus Pt–Pt fitted contribution) (solid line) and Pt–O fitted contribution (dotted line) for Pt/NaY (1.1–IE–0.2).

313

PREPARATION OF Pt PARTICLES IN ZEOLITE Y

TABLE 3 ˚ k2 Weighted) ˚ −1 , 1R = 1.6–3.1 A, Fit Parameters of EXAFS Spectra and Variances of Fits (1k = 2.5–14 A

Catalyst

Scatterer

N ±10%

Pt/NaY (1.1–IE–no dry, calc)

Pt O Pt O Pt O Pt O Pt O Pt O Pt O Pt O Pt O

11.2 0.7 10.1 0.6 6.9 0.9 5.4 0.4 6.1 0.5 7.8 1.1 5.6 1.1 6.3 0.8 5.3 1.5

Pt/NaY (1.1–IE–no calc) Pt/NaY (1.1–IE–1) Pt/NaY (1.1–IE–0.2) Pt/NaY (1.35–IM–0.2) Pt/H-USY (1.05–IE–1) Pt/H-USY (1.05–IE–0.2) Pt/H-USY (1.24–IM–0.2) Pt/Si-Al (1.5–IM–1)

˚ R (A)

˚ 2) 1σ 2 (10−3 A

1E 0 (eV)

±1%

±5%

±10%

2.76 2.62 2.75 2.65 2.75 2.62 2.75 2.65 2.75 2.66 2.76 2.58 2.76 2.65 2.76 2.67 2.76 2.61

−0.02 10 2.65 10 1.91 4.78 4.28 −1.65 3.19 10 0.96 8.82 2.23 10.84 2.63 10 3.21 13.18

numbers. According to conventional use, the catalysts are compared on the basis of H/Ptstrong ratios. For the ion-exchanged zeolites this ratio is found to depend strongly on the treatment conditions. The highest number (1.2–1.0) is found when the zeolites are calcined at a heating rate of 0.2◦ C per minute, whereas a heating rate of 1◦ C per minute results in a substantially lower value. The H/Ptstrong ratio of impregnated zeolite calcined at 0.2◦ C

0.29 −5.71 0.43 −2.38 2.26 2.10 3.33 −2.96 2.00 0.28 1.65 3.02 1.85 1.17 0.70 0.29 1.18 5.53

k 2 variance (%) Im.

Abs.

0.24

0.15

0.15

0.07

0.32

0.19

0.31

0.15

0.53

0.20

0.15

0.05

0.73

0.37

0.41

0.27

0.22

0.13

per minute is 0.45–0.5. With silica–alumina as the support similar values (0.56–0.58) are obtained irrespective of the calcination/precursor decomposition procedure. Although quantification of the platinum surface sites on the basis of hydrogen chemisorption results is controversial, the observed difference in H/Pt ratio for different catalysts based on the same support must be due to a difference in dispersion.

TABLE 4 Results of Catalyst Characterization H/Pt Catalyst

Total

Strong

NPt–Pt

TEM

Pt/NaY (1.1–IE–no dry, calc) Pt/NaY (1.1–IE–no calc) Pt/NaY (1.1–IE–1) Pt/NaY (1.1–IE–0.2) Pt/NaY (1.1–IE–0.2 water) Pt/NaY (1.1–IE–0.2 argon) Pt/NaY (1.35–IM–0.2)

0.1 n.d 1.0 2.1 n.d 1.5 1.1