J. CHEM. SOC. FARADAY TRANS., 1994, 90(8), 1157-1160 Characterization of Supported-palladium Catalysts by Deuterium NMR Spectroscopy Tsong-huei Chang Department of Chemical Engineering, Ming Hsin Engineering College, Hsin Feng, Hsinchu, Taiwan 30434, Republic of China Cheu Pyeng Cheng and Chuin-tih Yeh Department of Chemistry, National Tsing Hua University, Hsinchu, Taiwan 30043, Republic of China Alumina-supported Pd, Rh and Pd-Rh bimetallic alloys have been prepared by the method of incipient wetness. Sorption of deuterium by these samples was investigated by means of 'H NMR spectroscopy and uptake mea- surements at room temperature. Deuterium chemisorbed on palladium and rhodium atoms was characterized by 2H NMR peaks with chemical shifts of -12 and -170 ppm, respectively, from monometallic samples under a deuterium pressure of 1 Torr or less.These two characteristic peaks remained for the bimetallic samples and their relative intensities unambiguously revealed the surface composition of the bimetallic crys- tallites. Deuterium atoms were absorbed or weakly chemisorbed at greater pressures of deuterium. The weakly sorbed deuterium atoms exchanged with the strongly chemisorbed deuterium atoms, indicated by the coalescence of the 2H NMR characteristic peaks of the bimetallic samples. The position of the coalesced peak depended on the composition of the alloy crystallites. Palladium is a versatile noble metal which has been used not only as an active ingredient of hydrogenation and hydro- genolysis catalysts,' but also as a material for hydrogen storage., The versatility of palladium arises mainly from its ability both to chemisorb hydrogen on the surface of its crys- tallites and to absorb hydrogen into the bulk.Interactions between deuterium and palladium have attracted much atten- tion because of possible applications in cold-fusion. Many analytical techniques, for instance hydrogen solu- bilit~,~-" heat of absorption,' ' magnetic susceptibility,' ,-16 X-ray and neutron scattering' and NMR spectroscopy,'* have been used to study the interactions of palladium and hydrogen. Among these techniques, NMR is distinguished for possibly probing the chemical binding between hydrogen and metals.'9-23 Sheng and Gay24 studied the adsorption of hydrogen on Pd/SiO, with proton NMR, and Vannice and co-workers have reported the absorption of deuterium on the same catalyst., In our recent report,26 deuterium adsorbed on Rh/Si02 samples is classified into three categories, i.e.rigidly adsorbed (D,), adsorbed but possessing mobility (D,) and weakly adsorbed (D,) (may be pumped away by evac- uation in a high-vacuum system for 5 min at room temperature), according to 'H NMR observations. D, and D, are strongly adsorbed deuterium atoms with different mobility. D, is attributed to those deuterium atoms that are unobservable in the 'H NMR spectra because they are adsorbed rigidly to the metal surface and their peaks become broad because of excessive quadrupole interaction.D, is comparatively mobile and observable with 2H NMR. The latter form is gradually converted into D, with decreasing system temperature. We report here an investigation on the sorption of deuterium by supported palladium and supported Pd-Rh alloys by means of ,H NMR. Experimental Alumina-supported Pd, Rh and Pd-Rh alloys with various metal loadings were prepared by the method of incipient wetness. Chlorides of palladium and rhodium (both Merck) were dissolved in deionized water and impregnated drop-by- drop into y-alumina (Merck) with constant agitation. The resulting pastes were dried for 4 h at 383 K and calcined for 4 h at 723 K. A calcined sample (2.0 g) was sealed in a Pyrex tube (external diameter 10 mm) connected to a high-vacuum system, reduced in flowing hydrogen for 1 h at 573 K, degassed (to evacuate adsorbed hydrogen completely) at the same temperature until a pressure of 5 x lo-' Torr was reached, and characterized either with an adsorption iso- therm of deuterium in the vacuum system at room tem-perature or by means of 2H NMR in the Pyrex tube sealed with deuterium at a predetermined pressure.All NMR spectra presented in this paper were recorded at room tem- perature with a Bruker MSL-200 spectrometer ;the operating frequency was 30.72 MHz at 4.7 T magnetic field. A solid echo technique (9O,-z-9Oy-z) with a 90" pulse (6 ps), a z delay (20 ps) and a repetition interval (0.2 s) was employed. A CH,OD-CH,OH solution (2.88% v/v) was used as an exter- nal standard for calibration of chemical shift.Cr(acac),, a relaxation reagent, was added to the methanol solution to ensure complete relaxation of the deuterium in the standard solution within the repetition interval 0.2 s. Results and Discussion Pd/Al,O, Fig. 1 presents three sorption isotherms of deuterium obtained volumetrically from samples of 1, 1.5 and 3% Pd/A1,0,, respectively, at 298 K. Each isotherm is divided into a chemisorption region (less than 20 Torr) and an absorption region, i.e. 2Pd, + D, e2Pd,D (1) 2Pdb + D2=2PdbD (2) subscripts 's' and 'b' denote the atoms on the surface and in the bulk of the palladium crystallites, respectively. In the che- misorption region, the uptake increased sharply with increas- ing pressure of D,.The rate of deuterium uptake gradually slowed and reached a plateau at about 20 Torr. The uptake in this region is attributed to deuteriums chemisorbed on the surface of palladium particles. The observed plateau in this region indicates chemisorption to the extent of one saturated monolayer. The uptake of the isotherm increased steeply and then became level again upon further increase of the deute- rium pressure to 90 and 200 Torr, respectively. This J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 40 0.0 1 I I I 0 100 200 300 D, pressure/Torr Fig. 1 Volumetric adsorption isotherms at 298 K of deuterium on samples of alumina-supported palladium with loadings (%) of 1.5 (O),3.0 (0)and 1.0 (A),respectively increment comes from the absorption of deuterium into the bulk of the palladium cry~tallites.~~ The dispersion of the palladium crystallites in these samples is estimated from the observed uptake at the first plateau of their isotherm.The dispersion was determined to be about 0.22, 0.57 and 0.48 for samples having 1.0, 1.5 and 3.0%Pd/Al,O, ,respectively. The extent of deuterium uptake has a significant effect on the observed 'H NMR spectrum of sorbed deuterium. Two deuterium peaks are identified in each spectrum of Fig. 2 for the 3% Pd/A1,0, catalyst. The sharp peak at 0 ppm is due to CH,OD of the external standard. The broad peak is assigned to deuteriums sorbed (including both chemisorption and absorption) by the palladium in the sample. The position and width of latter peak depended on deuterium uptake.Fig. 3 relates the variation of the chemical shift of the peak observed on 2H NMR with the deuterium uptake determined volumetrically. Resembling the uptake isotherm of Fig. 1, the I/ 200 100 0 -100 -200 6 Fig. 2 Effect of deuterium uptake (D :Pd ratio, measured volumetrically) on the 'H NMR spectrum of deuterium sorbed on Pd/AI,O, (3.0%). The sharp line at 0 ppm is due to the external standard (see text). D :Pd = 0.245 (a), 0.330 (b),0.445 (c), 0.501 (4, 0.740 (e). -1 0 6 I-60 -11( I I I 0.5 1.o deuterium uptake, D/Pd Fig. 3 Effect of deuterium uptake on the chemical shift of sorbed Pd/A1,0, (3.0%) observed in Fig. 2. (0)deuterium. (0) Pd/Al,O, (1.5%). The plateau at a chemical shift of -12 ppm indicates mono- layer adsorption.observed chemical shift also showed two plateaux corre-sponding to chemisorption and absorption, respectively, upon increasing deuterium uptake. The peak of sorbed deute- rium shifted dramatically from -78 to -20 ppm under sub- monolayer coverage (0 < 0.6), and gradually became level at -12 ppm upon increase of the coverage to monolayer chemi- sorption (0 = 1). This shift, which resembled an earlier obser- ~ation,~~,~'is attributed to the consumption of the spin density of Pd crystallites during formation of chemical bonds between adsorbed deuterium and surface palladium atoms. Above monolayer coverage, the peak associated with sorbed deuterium shifted further downfield due to a Knight shift'' upon increase of the deuterium uptake and became level at +25 ppm when the absorption became ~aturated.~' Concur-rently, the peak width abruptly decreased because of rapid exchange between chemisorbed and absorbed deuterium atoms.28 Rh/Al, O3 An overpressure of deuterium has no significant effect on the width of the 2H NMR peak for deuterium adsorbed on an Rh/A120, sample (1.5%).Nevertheless, the peak position shown in Fig. 4 shifted gradually from -180 to -100 ppm with increasing deuterium pressure from 10 to 600 Torr. We speculate that weakly adsorbed deuterium, either multi-adsorbed (i.e. two deuterium atoms coadsorbed on a single surface rhodium atom) or located at the interface between Rh crystallites and the alumina support, is the main reason for this observed shift. Pd-Rh/Al,O, Fig.5 illustrates the effect of deuterium pressure on the 2H NMR spectrum of a Pd-Rh/Al,O, sample (6%, with J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 100 0 -100 -200 -300 -400 6 Fig. 4 Effect of deuterium overpressure on the 2H NMR spectrum for deuterium adsorbed on Rh/Al,O, (1.5%, dispersion 80%). PDz/Torr= 10 (a),50 (b),300 (c),600 (4. Pd : Rh = 1 :1). Two distinct peaks, at -12 and -170 ppm, respectively, were observed after this sample was pretreated by means of reduction by deuterium at 573 K and light evac- uation (5 min in the vacuum system at room temperature to desorb weakly sorbed deuterium). These two peaks are assigned to mobile deuterium chemisorbed on the palladium atoms (-12 ppm) and rhodium atoms (170 ppm), respec- tively, according to the correlations obtained in Fig. 3 and 4.An energy barrier (for migration of adsorbed deuterium from one metal atom to adjacent metal atoms) evidently prevents the following migration exchange from proceeding at room temperature: Pd-D + Rh Pd + Rh-D (3) Since mobile deuterium adsorbed on Pd-Rh crystallites maintains its characteristic peak position, the surface com- position of the alloy, can be estimated if the influence of undetectable deuterium is ignored. The relative intensities of the two peaks in Fig. 5(a) indicate that most deuterium is adsorbed on palladium. A segregation of palladium to the surface of the Pd-Rh bimetallic alloy crystallites is therefore indicated.The surface segregation of palladium in a Pd-Rh film was shown by electron probe microanalysis upon hydro- gen reduction on a surface at 573 K.29 As deuterium gas at 100 or 300 Torr was introduced to this bimetallic sample, the two peaks in Fig. 5(a)coalesced in Fig. 5(b)and (c) to form a new peak located at about -50 ppm. Obviously, from the coalescence, the migration exchange rep- resented by reaction (3) became important with increased pressure of deuterium. The energy barrier for migrating mobile deuterium [reaction (3)] might have been removed by the presence of weakly adsorbed deuterium under these con- ditions. That this exchange occurred indicates that an alloy phase of Pd and Rh was formed on the sample during the reduction pretreatment.We conclude from the observed coalescence that weakly adsorbed deuterium definitely modi- fied the properties, such as the reactivity, of adsorbed deute- rium. Fig. 6 depicts the effect of deuterium overpressure on the 'H NMR spectrum of deuterium adsorbed on a sample of Pd-Rh/Al,O, (9%, Pd : Rh = 2 : 1). The mobile deuterium adsorbed on palladium (peak near -12 ppm) is distinguished 1 I I I I I I 1 200 0 -200 -400 200 0 -200 -400 6 6 Fig. 5 Effect of deuterium overpressure on the 2H NMR spectrum Fig. 6 Effect of deuterium overpressure on the 'H NMR spectrum for deuterium sorbed on the Pd-Rh/A120, bimetallic catalyst (6%, for deuterium sorbed on Pd-Rh/Al,O, bimetallic catalyst (9%, Pd : Rh = 1 : 1).PDz/Torr= 0 (a),100 (b),300 (c). Pd : Rh = 2 : 1). PDz/Torr= 0 (a),100 (b),300 (c). 1160 in Fig. 6(a) from those adsorbed on rhodium (peak near -160 ppm) after weakly adsorbed deuterium was pumped away by light evacuation. Introduction of gaseous deuterium (100 Torr) to this sample produced two peaks at -60 and +25 ppm, respectively. The existence of two peaks under these conditions is explained in terms of deuterium sorbed on metal crystallites with two distinct phases. The broad peak located at -60 ppm is assigned to deuterium adsorbed on an alloy phase with a Pd : Rh ratio of close to 1 :1 because this chemical shift is very similar to that found in Fig. 5(b)and (c). The sharp line located at +25 ppm is assigned to deuterium sorbed by pure palladium (or a palladium-rich phase). The existence of these two peaks is consistent with a surface migration mechanism [reaction (3)], instead of a conceivable mechanism through the gas phase. Moreover, the presence of pure palladium crystallites on this sample has been confirmed by the deuterium adsorption isotherm.Fig. 7 shows that the sample with Pd : Rh = 2 : 1 has an absorption uptake at around a deuterium pressure of 100 Torr. This characteristic feature of the palladium phase, however, is absent in the iso- therm for the sample with Pd :Rh = 1 :1. An advantage of the use NMR to study adsorption over the conventional isotherm measurement is that the former technique may provide information on binding between adsorbates and adsorbents.The 2H NMR results presented in this paper reveal five features. (i) In the absence of weakly sorbed deuterium (upon light evacuation treatment after a high-pressure adsorption), mobile chemisorbed deuterium atoms on palladium and rhodium crystallites are character- ized by their 'H NMR, with chemical shifts of -12 and -170 ppm, respectively. (ii) Under high deuterium pressure (>30 Torr), excess deuterium penetrates the palladium crys- tallites. These absorbed deuterium atoms exchange rapidly at ambient temperature with deuterium atoms chemisorbed on the crystallites, resulting in a sharp line at +25 ppm at satu- rated absorption. (iii) In the absence of weakly sorbed deute- rium, deuterium chemisorbed on the palladium atoms of Pd-Rh bimetallic crystallites is distinguished from that adsorbed on rhodium atoms by their characteristic peaks (with chemical shifts of -12 and -170 ppm, respectively).Accordingly, the average surface composition of the two ele- ments on these crystallites may be estimated from the relative 5 0.2 0.3 5 n if3 I I0.0I J 0.0 0 100 200 300 D, pressure/Torr Fig. 7 Adsorptions isotherms (298 K) of deuterium sorbed on the Pd-Rh bimetallic catalysts: (0)6% Pd-Rh/Al,O,; (0)9% Pd-Rh/Al,O 3 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 intensities of the peaks. A surface enrichment of Pd over Rh was thus suggested for a supported bimetallic sample with a Pd :Rh ratio of 1 :1.(iv) Under greater pressures of deute- rium, deuterium sorbed on the bimetallic sample loses its characteristic nature because of exchange between these deu- terium sites. (v) Palladium alloys well with rhodium on the supported bimetallic catalyst with a Pd : Rh ratio of 1 :1. When this ratio exceeds unity, two metallic phases with dif- ferent compositions were found; one has a Pd : Rh ratio of about unity, whereas the other contains almost pure palla- dium. We appreciate the financial support of this study the Nation- al Science Council of the Republic of China. References 1 F. A. Lewis, The Palladium Hydrogen System, Academic Press, London, 1967. 2 G. Alefeld and J. Volkl, Hydrogen in Metals (II), Application Oriented Properties, Springer Verlag, New York, 1978, ch.3. 3 S. E. Jones, E. P. Palmer, J. B. Czirr, D. L. Decker, G. L. Jesen, J. M. Thorne, S. F. Taylor and J. Rafelski, Nature (London), 1989,338,737. 4 T. Gram, Philos. Trans. R. SOC.London, 1966,156,415. 5 T. B. Flanagan, Engelhard Id.Tech. Bull., 1966,7,9. 6 H. Z. Brodosky, Z. Phys. Chem., N.F., 1965,44,129. 7 H. Z. Brodosky and E. Poeschel, 2. Phys. Chem. N.F., 1965,44, 143. 8 C. Wagner, Chem. Phys., 1951,19,626. 9 R. Burch and R. G. Buss, J. Chem. SOC.,Faraday Trans. 1, 1975, 71,913. 10 R. Burch and R. G. Buss, J. Chem. SOC.,Faraday Trans. I, 1975, 71,922. 11 C. Foo, C. Hong and F. D. Manchester, J. Phys. F, 1971,2,323. 12 S. Ladas, R. A. Dalla Betta and M. Boudart, J. Catal., 1978, 53, 356.13 L. H. Reyerson and A. Solbakken, Advances in Chemistry Series, Am. Chem. SOC., Washington, DC, 1961, no. 33. 14 W. Trzebiatowski and H. Kubicka, Z. Chem., 1963,3, 262. 15 H. Kubicka, J. Catal., 1966, 5, 39. 16 R. W. Zuehlike, J. Chem. Phys., 1966,45,411. 17 G. Alefeld and J. Volkl, Hydrogen in Metals (I), Basic Properties, Springer Verlag, New York, 1978, ch. 3,4. 18 G. Alefeld and J. Volkl, Hydrogen in Metals (I), Basic Properties, Springer Verlag, New York, 1978, ch. 9. 19 R. E. Norberg, Phys. Rev., 1952,86,745. 20 W. Z. Spalthoff, 2.Phys. Chem., 1972,76,760. 21 R. M. Cotts, Ber. Bunsenges. Phys. Chem., 1972,76,760. 22 R. R. Aron, H. G. Bohn and H. Lutgemier, Solid State Commun., 1974,14, 1203. 23 E. F. W. Seymour, R. M. Cotts and W. D. Williams, Phys. Reo. Lett., 1975,35, 165. 24 T. C. Sheng and I. D. Gay, J. Catal., 1982,77, 53. 25 A. A. Chen, A. J. Benesi and M. A. Vannice, J. Catal., 1989, 119, 14. 26 (a) T. C. Chang, C. P. Cheng and C. T. Yeh, J. Phys. Chem., 1991,%, 5239; (b)1992,%, 4151; (c)J. Catal., 1992, 138, 457. 27 G. Chen, W. T. Chou and C. T. Yeh, Appl. Catal., 1983,8,389. 28 T. C. Sheng and I. D. Gay, J. Catal., 1981,71, 119. 29 H. Muraki, M. Sobukawa and A. Isogai, 1990, SOC. Automotive Eng. paper 900610. Paper 3/06571E; Received 3rd November, 1993