Adsorption, desorption, and interparticle motion of ...

Adsorption, desorption, and interparticle motion of hydrogen on smca-supported ruthenium: A study by in situ nuclear magnetic resonance Frank Engelke Ames Laboratory, Ames, Iowa 50011

Robert Vincent Laboratoire de Chimie des SuTj'aces, Universite Pierre et Marie Curie, 75230 Cedex, Paris, France

Terry S. King Ames Laboratory, Ames, Iowa 50011 and Department of Chemical Engineering, Iowa State University, Ames, Iowa 50011

Marek Pruski Ames Laboratory, Ames, Iowa 50011

(Received 20 June 1994; accepted 21 July 1994) IH NMR line shapes of hydrogen adsorbed on silica-supported ruthenium at pressures of 10- 6-10 Torr were studied by using selective excitation via DANTE sequences. A transition from inhomogeneous to homogeneous line broadening was observed at hydrogen coverage of ~0.5. The spectra were simulated by using generalized Bloch equations that included N-site exchange processes. The homogeneous line shape originates from increased hydrogen mobility, whereas proton-proton dipolar couplings are negligibly small. A rate parameter k obtained from this model quantifies the average mobility of hydrogen in the exchange process. This parameter increases by more than three orders of magnitude when the hydrogen coverage changes from 0.4 to 0.8. The simulations of line shapes obtained at variable temperatures showed that k exhibits Arrhenius behavior with an activation energy of 52 (±5) kJlmol and preexponential factor ko=4XlO lO S-I. It is implied that the motion of hydrogen must involve desorption, interparticle diffusion, and readsorption.

INTRODUCTION

Mobility of hydrogen interacting with metal surfaces encompasses two dynamic processes: (i) adsorption and desorption of hydrogen onto and from the surface and (ii) surface diffusion of atomic or molecular hydrogen across the metal. The adsorption/desorption mechanisms and kinetics of hydrogen on the surface of ruthenium have been extensively investigated for single crystalS. I- 12 It is known that hydrogen dissociates upon adsorption on ruthenium, and binding sites and geometries were identified for various surface orientations, coverages, and temperatures. The diffusion rates of hydrogen on ruthenium single crystals have also been measured. 13 - 15 These single crystal studies were performed using vibrational spectroscopies, electron diffraction techniques, thermal desorption experiments, and ab initio calculations. The dynamics of hydrogen in metal-support systems has recently received increased attention. Qualitatively, the characteristics of hydrogen on surfaces of single crystals remain valid for_hydrogen interacting with highly dispersed, supported metal particles. However, several additional phenomena have to be taken into account. First, the particles consist of a variety of crystallographic surfaces with typical sizes of less than 10 nm and exhibit numerous defect-like sites (e.g., edges and comers). Each of these surface features may play a distinct role in adsorption and surface transport. Second, the metal particles are embedded in a nonmetallic support with typical interparticle distances on the order of 100 nm. 7262

J. Chem. Phys. 101 (9), 1 November 1994

Although the support may be an adsorbent itself, it can also interact with the metal particles and thereby affect their adsorption/desorption behavior.16 Apart from hydrogen motion on the metal resulting from adsorption, desorption, surface diffusion, and exchange between different metal surface sites, hydrogen may migrate to the support (spillover), diffuse on or desorb from the support surface, or return to the metal particle surface (back-spillover).17.18 Nuclear magnetic resonance has proven a valuable tool for investigating molecules or atoms adsorbed on supported metal catalysts. 19- 49 For example, the details of CO motion on the surface of metal particles have been studied using l3C NMR selective-excitation techniques. 45 - 47 The NMR studies of hydrogen motion26.28.32,34,35,40,41,43 on supported metals are of particular interest since such studies cannot be performed by using most surface spectroscopic techniques. The NMR experiments used for these studies depend on the time scale of the dynamic process and may allow measurements of transport phenomena, molecular or atomic mobility, exchange rates between various species, adsorption and desorption rates, and activation energies. Hydrogen interacting with metal surfaces can be easily distinguished from the support sp{!cies (e.g., OH groups and spillover hydrogen22-36) because it is characterized by NMR lines which are usually shifted to a higher field via interaction with metal electrons (Knight shift48 ). In a previous paper we have investigated the dynamics of hydrogen interacting with ruthenium particles at elevated pressures by means of one- and two-dimensional exchange

0021-9606/94/101 (9)/7262/11/$6.00

© 1994 American Institute of Physics

Downloaded 09 Jun 2010 to 129.13.72.198. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/jcp/copyright.jsp

Engelke et al.: Hydrogen on silica-supported ruthenium

NMR techniques. 43 Two types of hydrogen have been found, aM and {3, both interacting with ruthenium as indicated by their Knight shifts. The aM hydrogen was regarded as dis sociatively adsorbed and was characterized by a homogeneous NMR line shape originating from the motion of hydrogen. The {3 species represented weakly bound hydrogen in fast exchange with the gas phase, as revealed by the pressure dependence of its resonance shift. Furthermore, a slow exchange was observed between the aM and f3 species with the time constant on the order of 1 ms. Under evacuation conditions at room temperature, the surface of ruthenium particles was covered by a third species, strongly bound hydrogen aI, showing an inhomogeneously broadened NMR line. The occurrence of aI, aM' and f3 hydrogen was correlated with decreasing heats of adsorption vs increasing pressure. 44 In the present article we focus upon hydrogen mobility under low pressure conditions (from about 10-6 to 10 Torr) as revealed from NMR line shapes obtained by using selective excitation via DANTE50 pulse sequences. In particular, the relation between the transition from inhomogeneous to homogeneous NMR line broadening upon increasing coverage, the associated increase of hydrogen mobility, and the occurrence of hydrogen previously identified as aT and aM will be shown. The types of motion for hydrogen atoms or molecules with respect to adsorption/desorption and surface diffusion processes will be discussed, and quantitative results will be provided, characterizing interparticle motion of hydrogen on supported ruthenium.

EXPERIMENT Samples of silica-supported ruthenium (Ru loading 4 wt. %) were prepared by incipient wetness impregnation of [Ru(NO)(N03)]3 in distilled water with dried Cab-O-Sil HS5 (300 m2 BET area) silica as described in previous work. 34,35 Reduction of the metal was accomplished at 623 K under 760 Torr H2 inside an in situ NMR probe, designed at Ames Laboratory.49 The probe, operable at temperatures up to 770 K, was attached to a vacuum/dosing manifold. Hydrogen was replaced every 20 min for 2 h, and adsorptioh of hydrogen during the reduction process was monitored via NMR. After reduction, the samples were evacuated for 1 h at 623 K, and the temperature lowered to 400 K. Subsequently the samples were exposed to 10 Torr H2 and the temperature decreased to room temperature. After an equilibration period of 2 h at room temperature, evacuation was started. Variation of the amount of hydrogen strongly adsorbed on Ru was achieved by evacuation at room temperature. The lowest amounts of hydrogen adsorbed on ruthenium at room temperature were obtained after three days of evacuation (equilibrium pressure p =2X 10-6 Torr). All NMR measurements were performed on a homebuilt spectrometer operating at 250 MHz proton resonance frequency. Selective excitation was accomplished by using a delays alternating with nutations for tailored excitation (DANTE) sequence of 30 short rf pulses of Tw=0.6 f.J-S duration. 5o In most experiments a pulse separation Tp of 25 f.J-S was chosen, resulting in a total duration of the DANTE sequence of 1i=750 f.J-S and the corresponding spectral exci-

tation width A of the center band of .... 1.5 kHz. The excita-

7263

HISIO~

HIRu

(ad slnglll'pulH IIxcltatlon

DANTE IIxcltatlon I

-60

I

-40

Jl-:

A

B

I

I

I

I

I

-20

0

20

40

60

FREQUENCY (kHz)

FIG. 1. Inhomogeneous IH NMR line of hydrogen on RuISiO z under singlepulse (A) and selective excitation with DANTE (B). Both spectra were measured at room temperature after the sample was evacuated to 2X 10- 6 Torr.

tion sidebands were separated from the center band by (Tp )-1=40 kHz and did riot affect the NMR spectrum. The overall flip angle of the DANTE sequence was adjusted by varying the pulse width Tw while the rf amplitude remained constant. After a recovery period Tr of at least 30 f.J-s, a final 90° pulse was applied, followed by detection of the free induction decay. RESULTS

Figure 1 [spectrum (A)] represents a IH NMR spectrum obtained under single~pulse rf excitation after the Ru/Si0 2 sample was exposed to hydrogen and evacuated. Hydrogen adsorbed on the surface of ruthenium particles (HlRu) exhibited a resonance line well separated from the NMR line representing hydrogen associated with silica. The line for HlRu was slightly asymmetric, which was characteristic for strongly adsorbed hydrogen al. 43 Remarkably, selective inversion of a narrow frequency band (hole burning) within the HlRu peak by means of DANTE excitation was easily achieved [Fig. 1, spectrum (B)], revealing that the broadening of the NMR line was inhomogeneous.

HISIO~

HIRu (IX,,) slng'tl-pu'H excitation

DANTE tlxcltatlon I

,

-60

-40

Jet

A

B

I

I

I

,

,

-20

0

20

40

60

FREQUENCY (kHz)

FIG. 2. Homogeneous line shape for aM hydrogen obtained at hydrogen pressure of 0.6 Torr: single-pulse excitation (A) and DANTE excitation (B). The corresponding hydrogen coverage of the metal surfaces is approximately unity.

J. Chern. Phys., Vol. 101, No.9, 1 November 1994 Downloaded 09 Jun 2010 to 129.13.72.198. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/jcp/copyright.jsp

7264

Engelke et sf.: Hydrogen on silica-supported.ruthenium

H+DIRuISIO~

T-300K O.5TorrD~

H/RulSiO •• 'tr '" 30 IlS T .. 29S K

e

sImulated

experImental

O.3S~ :;-~

:;:k 0.83

,

t

-60

-40

'

!

,

-20 0 __ 20 FREQUENCY (kHz)

~O.o4 ~0.2 ~0.6

~1

=

~3

-15

~~--~--~--~

!

40

60

FIG. 3. Homogeneous line shape for hydrogen when the sample was exposed to D2 gas. Note that the intensity scales in Figs. 1-3 are different.

Elevation of the H2 pressure to 0.5 Torr led to an intensity increase of the H1Ru peak as well as the HlSi02 peak and to a slight change of the H1Ru line shape [Fig. 2, spectrum (An. Selective excitation with the same rf pulse sequence parameters at this H2 pressure did not result in hole burning. Instead the entire H1Ru resonance line became saturated [Fig. 2, spectrum (B)], indicating that the line was homogeneously broadened. The corresponding hydrogen species was termed aM' Mter reevacuation at ambient temperature to produce al hydrogen, the sample was exposed to 0.5 Torr deuterium gas, leading to the simultaneous presence of protons and deuterons on the surfaces of Ru particles [Fig. 3, spectrum (A)]. The linewidth is similar to that shown in Fig. leA). More importantly, DANTE excitation resulted in complete saturation of the resonance line [as in the H2 dosing experiment; compare spectra (B) in Figs. 2 and 3], indicating that proton spins in the presence of deuterium respond in the same way to selective excitation as if only hydrogen were present on the surface. The amount nH of hydrogen adsorbed on the metal particles was obtained by comparing the intensity of the H1Ru peak in the nonselectively excited NMR spectra [e.g., spectra (A) in Figs. 1 and 2] with that of a reference standard. In order to estimate the average surface coverage of hydrogen on the Ru particles, it was necessary to obtain that amount n~ of hydrogen, which corresponded to one monolayer. For this reason, the dispersion D of the metal particles, defined as the ratio of surface ruthenium atoms to total number of ruthenium atoms, D = nRu,slnRu,total> had to be determined on the basis of the assumed surface-to-hydrogen stoichiometry of H1Ru = 1. Throughout this work, the H1Ru NMR intensity n~ was measured at room temperature after short (.... 5 min) evacuation of the sample that had previously attained an equilibrium under a hydrogen pressure of 10 Torr. A value of n~ = (4.5 ± 0.7) X 10i8 was obtained from repeated measurements. Assuming that this value was equivalent to approximately one monolayer of hydrogen (coverage 0=1), we determined a dispersion D =0.20±0.03. Obtaining coverages lower than 8= 1 at room temperature required extended evacuation (e.g., three to five days for 0