- Email: [email protected]
Nuclear magnetic resonance of hydrogen sorbed by powdered palladium metal and alumina-supported palladium
Solid State Nuclear Magnetic Resonance, 1 (1992) 5-12
Elsevier Science Publishers B.V., Amsterdam
Nuclear magnetic resonance of hydrogen sorbed by powdered palladium metal and alumina-supported palladium Daniel J. Barabino ’ and Cecil Dybowski * Department of Chemistry and Biochemistry and Center for Catalytic Science and Technology, lJnil,ersiq of Delaware, Newsark, DE 19716, USA
(Received 23 July 1991; accepted in revised form 15 October 1991)
Abstract
Hydrogen in 0.2 pm P-palladium powder and in an alumina-supported palladium catalyst was investigated by NMR spectroscopy. The dependence of the shift on hydrogen pressure and the hydrogen-to-palladium (H/Pd) ratio in P-palladium hydride was determined for 273 K < T < 368 K. The activation energy for the process affecting the hydrogen shift is 9.4(+0.5) kcal mol-‘, similar to the enthalpy of transition from the (Y-to P-phase. NMR measurements of hydrogen in alumina-supported palladium show a similar behavior, which may allow one to use NMR shifts as a barometer of the state of the hydrogen in the metal. Keywords: palladium; hydrogen; catalyst; adsorption; NMR
Introduction Hydrogen is sorbed in large quantities by palladium metal. The hydrogen species formed are generally referred to as “hydrides”, “bulk-phase hydrogen”, or “absorbed hydrogen” [l]. Two phases have been identified from volumetric and magnetic-susceptibility isotherms: a-palladium hydride for H/Pd < 0.1 at T < 373 K and /?-palladium hydride for H/Pd > 0.6 at T < 373 K. Wicke er al. [21 determined the limiting values of H/Pd, denoted by (Y,,, and &,, at which the two bulk phases are stable at temperatures between 293 K and 571 K. At low temperatures (T=60K),Ppa lida ium hydride is reported to undergo a transition in electrical resistivity, ther-
’ Present address: Varian NMR Instrument Division, 6 Birchwood Court, Suite lK, Mineola, NY 11501, USA. * To whom correspondence should be addressed. 0926-2040/92/$05.00
ma1 expansion, and specific heat [3,4]. It has also been reported that palladium hydride becomes superconducting at low temperatures [5]. Supported palladium catalysts are somewhat more complex than bulk palladium. There is almost always a distribution of particle size. Other forms of hydrogen uptake may be significant [6]. Hydrogen is known to chemisorb at the surface at a stoichiometry of one hydrogen atom per surface palladium atom, requiring heat to remove it [7]. It binds at a stoichiometry of one hydrogen atom per surface palladium atom [l]. Hydrogen may also weakly adsorb [S], or reside at subsurface sites [9] or at lattice defects [ 101,or spill over onto the support. In early work, Norberg reported the NMR shifts of palladium hydride [ll]. More recently, diffusion, relaxation and tunnelling in palladium hydrides prepared by various techniques and in various morphologies have been addressed by NMR investigations [121. In one study [13], sam-
Q 1992 - Elsevier Science Publishers B.V. All rights reserved
6
D.J. Barabino, C. Dybowski/Solid
ples of different hydrogen content were investigated, but in many reports [12] samples with a single H/Pd ratio were examined. The NMR line for hydrogen on silica-supported palladium has been reported to fall in the range of - 80 to - 5 ppm relative to TMS, depending on the coverage 1141. In this paper, we report measurements of the proton NMR shift in powdered palladium hydride as a function of hydrogen content, pressure and temperature and in alumina-supported palladium. We use NMR spectroscopic measurements to estimate the amount of “internal” hydrogen in supported palladium catalysts.
Experimental Powdered palladium metal was obtained from Aesar, Inc. (Seabrook, NH). The reported purity was 99.980%. The average particle.diameter was reported to be 0.2(+0.1) pm. The material was used without further purification or sieving. The alumina-supported palladium catalyst was obtained from Sun Refining and Marketing Company (Marcus Hook, PA). Atomic absorption spectrometry indicated the sample was 4.1(+0.5)% w/w palladium. The nitrogen BET surface area of the palladium metal was determined to be 20(*5) m2 g-i and that of the catalyst was 185(* 10) m2 g-l. Carbon monoxide chemisorption on the catalyst gave a palladium surface area of 2.5cf0.3) m2 g-l. The dispersion of palladium on the catalyst was estimated to be 13( + 11% by the method of Beebe and Yates [151, with an estimated average particle diameter of 90( * 9) A [161. N 0.2 g of sample (either For experiments, palladium powder or the supported palladium catalyst) in a controlled-atmosphere NMR cell [17] was outgassed at 298 K using a glass, greasefree manifold. The sample was subsequently reduced in flowing H, (Airco, grade 5, 99.999%, 40 ml min-‘) at 373 (for the metal) and 673 K (for the catalyst) for 3 h to remove residual surface oxides [18]. A relatively low reduction temperature was used to avoid sintering the metal, which is known to occur at higher reduction tempera-
State Nucl. Magn. Reson. 1 (1992) 5-12
tures 1191. After reduction, samples were outgassed to 5 X 1O-6 torr at 373 K and cooled to 295 K under dynamic vacuum. The reduced sample was exposed to a pressure of hydrogen greater than 700 torr for 2 h. From this initial state, hydrogen was desorbed in several steps, with establishment of equilibrium at specific hydrogen contents in 20 min. After each step the sample was examined with NMR spectroscopy. Thus, conditions for the NMR experiments were always reached by desorption. We therefore observe no hysteresis effects that may be present [2]. The 90” pulse was 2.0 pus long and the relaxation delay was 5.0 s. Spin-temperature alternation was used to reduce baseline artifacts [20]. Each spectrum is the Fourier transform of 1000 transient decays. The spin-lattice relaxation time of protons in the powdered palladium hydride is N 80 ms at 298 K, as determined from the inversion-recovery null point. The temperature in the NMR spectrometer was controlled to +2 K. We found it necessary to reduce random errors by ensuring that the same region of the NMR cell was cooled or heated in the sorption bath and the spectrometer. All spectra are referred to an external sample of tetramethylsilane (TMS) and positive shifts are deshielded (downfield).
Results and discussion p-Palladium hydride
Figure 1 shows isotherms for hydrogen uptake in powdered palladium at several temperatures. To calibrate H/Pd, we referred the value of H/Pd at the transition, Pmin, to that reported on similar samples [13,211. From the temperature dependence (not shown) of pt, the transition pressure, we estimate AHap = 8.6Cf0.5) kcal mol-’ and AS,, = 20( &2) cal mol-’ K-r. In Table 1 we compare these results to literature values on samples having a variety of morphologies - powders, wires, and foils. From this comparison, one concludes that the 0.2 pm particles are similar to other macroscopic forms of palladium hydride.
D.J. Barabino, C. Dybowski/Solid
0.00
0.20
7
State Nucl. Magn. Reson. 1 (1992) 5-12
0.60
0.40
0.80
H/Pd
Fig. 1. Hydrogen uptake by palladium at various temperatures. The reference pressure, pO, is one atmosphere. (A) 273 K; (.I 295 K; (+) 323 K, CD) 383 K.
I D
Figure 2 shows proton NMR spectra of powdered P-palladium hydride at 295 K and different pressures. The resonance is deshielded relative to TMS at all conditions examined. This behavior is quite different from that of molecular hydrides, which are generally more shielded than TMS 122,231.Thus, the NMR spectroscopy of this bulk phase is quite different from that of other hydridic materials. In addition, these shifts are very different from those reported for hydrogen sorbed in metal particles supported on oxides [14,24]. The deshielding can be attributed to Knight shifts
n31. TABLE 1 Thermodynamic properties of palladium hydrides
Bulk
Powder
Ref.
“Hup (kcal mol-‘)
$i?nol-’
8.5 (kO.3) 9.3 (f0.1) 9.8 (+O.l) 9.44( + 0.05) 8.6 (kO.5) b
23.3( f 0.2) 21.8CkO.2) 20 (+21b
K-l) 18 2 4 a This work
a J.G. Aston, Engelhard Ind. Techn. Bull., 7 (1966) 14. b The uncertainties are given at 95% confidence.
JC
. . . . . . . .... . . . 4 LB
40
..I.
32 RESONANCE
.
24
.
.
.
.
.
.
16
,,.....
,.
8
SHIFT
Fig. 2. Proton NMR spectra of P-palladium hydride at 295 K. Hydrogen pressure: (A) 560 torr: (B) 190 torr; CC) 35 torr; (D) IO torr. crTMS= 0.
The proton resonance shift of powdered P-palladium hydride is a strong function of the pressure (Fig. 21, as observed by Brill and Voitlander [13]. The lines are narrow (FWHM = l-2 ppm), indicating the hydrogen must be experiencing no, or very weak, static dipole-dipole couplings. The hydrogen in P-palladium hydride must be rather mobile. At lower H/Pd ratios (i.e., in cu-palladium hydride and in the transition region between cr- and P-phases), the resonance is much broader. We have restricted these studies to the /?-phase. The pressure dependence of the NMR shift of hydrogen in P-palladium hydride (Fig. 3) was fit to the equation:
D.J. Barabino, C. Dybowski /Solid State Nucl. Magrl. Reson. 1 (1992) 5-12
b b -20
.---+A--2
3 1000/T
Fig. 4. Semi-logarithmic temperature. 0
1
2
3
4
5
plot of [u,(T)-
aI] ucrsus inverse
6
MM>
Fig. 3. Resonance shift of P-palladium hydride ((~1 referred to the shift at the cu-to-/3transition (a,) as a function of pressure and temperature. ( A) 273 K; (0) 295 K; (+) 323 K, (W) 383 K.
where a(p, T) is the shift, p is the hydrogen pressure, p,(T) is the transition pressure at temperature T, al(T) is a parameter describing the variation of the shift with temperature and pressure, and a,(T) is the shift extrapolated to the transition point, Pmin. The parameters are given in Table 2 at four temperatures. The region of validity of this equation is Pmin < H/Pd < 0.8. In Fig. 4 is shown the temperature dependence of (T,. (pi is empirically given by the equation: (~i(T) = a, + b, exp( -E,/RT)
where a, and b, are empirical parameters and E, is an activation energy. From the data of Fig. 4, E, = 9.4( *0.5) kcal mol-‘, similar to the enthalpy of transition, AHap, of 8.6 kcal mol-‘. To determine how the resonance shift depends on H/Pd, one must specify this parameter. We calibrated Pmin at each temperature to be in agreement with the literature value [2]. Figure 5 shows the shift as a function of (H/Pd)/P,,, at 1
4oh . *
-
(2)
Y
.
/
. i
b TABLE 2 NMR parameters of powdered palladium hydride; fits to eqn. (1)
-
/
B
Temperature CK)
a,(T) a (ppm)
V&? a (ppm)
Zorr)
273( * 2) 295( &-2) 323( + 2) 368( * 2)
2.99( f 0.30) 3.33( + 0.30) 4.63( f 0.50) 13.4 (k3.0)
19.4( + 2.4) 19.8( + 3.0) 15.7( f 2.4) 10.3( f 2.0)
2(&l) 7(kl) 20( i 1) 136(+ 1)
a The uncertainties arc given at 95% confidence.
10
0.90
1.00
1.10
1.20
1.30
WPd)/Om,r
Fig. 5. Dependence of the resonance shift in P-palladium hydride on (H/Pd)/P,i,. (A 1 273 K; (0) 295 K.
D.J. Bur-ahino, C. Dyhowski/Solid
H/Pd
Fig. 6. Isotherms for uptake of hydrogen by palladium powder (0) and alumina-supported palladium ( A) at 295 K.
273 and 295 K. From Fig. 5, the shift is seen to obey the following simple relation: a(H/Pd,
9
State Nucl. Magn. Re.son. I (1992) S-12
T) =A + B(H/Pd)/&,
(3)
where A and B are arbitrary parameters. Brill and Voitlander [13] observe a similar linear dependence of u on H/Pd. B is independent of temperature over this range and A seems to depend on the temperature. A measurement of (T at a given temperature is an indirect measure of H/Pd in P-palladium hydride. Alumina-supported palladium
In Fig. 6 we compare the isotherms for the uptake of hydrogen by powdered palladium metal and by the alumina-supported palladium. The slope in the transition region for the metal powder is zero, consistent with that expected from the phase rule. The transition region for the catalyst has a positive slope. We attribute this effect to a distribution of transition pressures for the catalyst sample, probably due to a distribution of particle sizes [25]. The isotherm for the catalyst gives the total uptake of hydrogen in all forms. By extrapolation of the high-pressure linear portion of the isotherm, (Pminjtot for the catalyst is found to be 1.34( f 0.051, a value substantially larger than 0.61, the value of Pmin for the metal powder. Thus, in addition to uptake into the body
of the metal particles there must be other modes of adsorption that contribute to hydrogen uptake on the small particles. Figure 7 compares the NMR spectrum of hydrogen sorbed by palladium powder with that of Pd/Al,O, at the same equilibrium hydrogen pressure, 82 torr. The mean resonance shifts are the same, indicating that both systems have similar electronic environments. (The small peak at 4.5 ppm in the spectrum of the powder results from a slight amount of water that was not removed during the pretreatment of this sample. The signal-to-noise ratios of the two spectra in A and B are a reflection of the different number of spins in the two samples.) The principal difference between the two spectra is the line width, with the resonance of hydrogen on alumina-supported palladium being somewhat broader than that of hydrogen sorbed in the metal powder. This broadening reflects a distribution of environments on the supported palladium. In Fig. 8 are data for the resonance shift of hydrogen of the powdered metal and the alumina-supported palladium as a function of hydrogen pressure at 295 K. Within the precision of the data (particularly
A_ A
a,
wm
Fig. 7. Proton NMR spectra of hydrogen in (A) palladium powder and (B) an alumina-supported palladium catalyst in equilibrium with hydrogen at 82( i 2) torr at 295 K.
D.J. Barabino, C. Dybowski/Solid
p
I Resonance
Shift
30
@pm1
those of the alumina-supported palladium), these data lie along a common curve in the region above pmi,. Assuming the relationship between H/Pd and shift found for the bulk P-palladium hydride is valid for the small particles on the alumina-supported material, one may construct an NMR-determined isotherm for only “internal” hydrogen, i.e, the hydrogen in an environment like that in the powdered P-palladium hydride. Such an isotherm is given in Fig. 9, along with the isotherm for the powdered palladium hydride. We ex-
0.
-1.0
2 2
Y -2.0
0.2
0.4
0.6
100
200
300
400
6
IO
Pressure km)
Fig. 8. Resonance shift as a function of hydrogen pressure for hydride formed from the metal powder (0) and alumina-supported palladium (A ) at 295 K.
0.
0.6.
0 20
10
0
State Nucl. Magn. Reson. 1 (1992) 5-12
Ii
0
HlPd
Fig. 9. NMR-determined isotherm for the “inner” hydrogen in small palladium particles supported on alumina, as described in the text. (a) Bulk palladium particles; (A) supported palladium particles.
Fig. 10. Excess hydrogen uptake, determined by difference between the total uptake and the NMR-derived uptake into the supported palladium.
tended the isotherm into the transition region by this procedure, although this may not be valid. However, the striking similarity between the isotherm for the powdered metal and the supported palladium would seem to indicate that the results are at least qualitatively correct. There are subtle differences between the two isotherms. The slight shift between them above Pmin is probably not meaningful. The difference between them in the transition region may be the result of the distribution of particle sizes in the catalyst, resulting in a distribution of transition pressures for this sample. The difference between the volumetric isotherm and the NMR-derived isotherm we label the excess H/Pd. This hydrogen is sorbed by the catalyst through other modes of uptake, e.g., chemisorption at the surface or spillover to the surface [26]. We show this quantity, H/Pd (excess), as a function of pressure in Fig. 10. The amount of excess H/Pd appears to increase with the hydrogen pressure and saturate at a value of 0.97( _~r 0.04). The solid line is a fit of the data to a Langmuir isotherm. If the only mode of sorption aside from incorporation in the lattice were chemisorption at the surface, one would expect it to saturate at a value near 0.13, based on the Beebe-Yates measurements of the dispersion. Assuming that, at saturation, the surface sites for a sample with a dispersion of 0.13 contribute 0.13 H/total Pd, there must be a third species which comprises 0.85 H/total Pd. This could be
D.J. Barabino, C. Dybowski /Solid
State Nucl. Magn. Reson. I (1992) S-12
physisorbed material or spilled-over hydrogen, which is invisible to NMR spectroscopy. This may be the so-called “recondite” (hidden) hydrogen proposed by Root and Duncan to be present in a supported metal catalyst investigated with NMR spectroscopy [27].
11
of the shifts of hydrogen in bulk palladium hydride and in the supported palladium material in equilibrium with the same pressure of hydrogen, this seems reasonable. However, the validity of this assumptions needs to be tested rigorously.
Acknowledgements Conclusions Hydrogen desorption isotherms of powdered palladium (with particle diameters N 0.2 pm) give thermodynamic properties similar to those of palladium foils and wires [12]. The NMR shift of powdered p-palladium hydride depends strongly on other parameters of the system. The resonance is deshielded relative to TMS, in contrast to molecular hydrides. These results are also quite different from those reported for hydrides in small metallic particles on the surface of oxide supports. We have empirically specified how the NMR shift depends on characteristics of the system such as the hydrogen pressure, temperature, and, by calibration, the hydrogen-to-palladium ratio. The shift at Pminr the p-phase limit, tends to decrease with temperature, i.e., it becomes more shielded. We have shown that in the /3phase the shift is linear in the hydrogen-to-palladium ratio. For alumina-supported palladium, we find the same shifts as for the bulk powder, when comparing samples in equilibrium with the same hydrogen pressure. From this fact, we assign an “internal” H/Pd content to the catalyst. By difference, we determine the excess H/Pd at any pressure. This value exceeds the dispersion of the catalyst. We infer that the excess H/Pd is attributable to at least two kinds of hydrogen: chemisorbed hydrogen and other adsorbed hydrogen. At Pmjn, we find roughly 48(+4)% of the hydrogen is absorbed into the bulk of the particle, lO(&-2)% of the hydrogen is chemisorbed at the surface and the remainder, 42(*5)%, exists in a third form which is neither. This analysis is only valid if the shift of the observed hydrogen resonance can be used to infer the “internal”, i.e., palladium-hydride-like, hydrogen in the particles. Based on the similarity
We gratefully acknowledge a Hercules grantin-aid, the Sun Refining and Marketing Company, the donors of the Petroleum Research Fund and the Sponsors of the Center for Catalytic Science and Technology for support of this work.
References 1 E. Wicke and H. Browdowski, Hydrogen in Palladium and Palladium Alloys. in G. Alefeld and J. Volkl (Eds.), Hydrogen in Metals, II, Springer Verlag, Berlin, 1978. p. 73. 2 (a) H. Frieske and E. Wicke, Ber. Bunsenges. Phys. Chem., 77 (1973) 50; (b) E. Wicke and G.H. Nernst, Ber. Bunsenges. Phys. Chem., 68 (1964) 224. 3 B.M. Geerken, H. Hemmes and R. Griessen, J. Phys. F, 14 (1984) 813. 4 J.K. Jacobs and F.D. Manchester, J. Low Temp. Phys., 4 (1971) 129. 5 B. Strizker and W. Buckel, Z. Phys., 257 (1972) 1. 6 (a) P.C. Aben. J. Catal., 10 (1968) 224; (b) P.A. Sermon, J. CataL, 24 (1972) 460; fc) J.E. Benson, H.S. Hwang and M. Boudart, J. Cauzl., 30 (1973) 146. 7 A.W. Aldaq and L.D. Schmidt, J. Catal.. 22 (1971) 260. 8 P. Chou and M.A. Vannice, J. Catal., 104 (1987) 1. 9 (a) R.J. Behm, V. Penka. M.G. Cattania. K. Kristmann and G. Ertl. J. Chem. Phys.. 78 (1983) 7486; (b) R.J. Behm, V. Penka, M.G. Cattania, K. Kristmann and G. Ertl, Su$, Sci.. 126 (1983) 382. 10 H. Kubicka, J. Catal., 5 (19661 39. 11 R.E. Norberg, Phys. Rec., 86 (1952) 745. 1’A (a) J.P. Burger, N.J. Poulis and W.P.A. Hass, Physica, 27 (1961) 514; (b) G.K. Shoep, N.J. Poulis and R.R. Arons, Physica, 74 (1974) 297; (c) E.F.W. Seymour, R.M. Cotts and D.W. Williams, Phys. Rec. Lett.. 3 (1975) 165; (d) S.R. Kreitzman and R.L. Armstrong, Phys. Rev. B. 25 (1982) 2046; (e) S.R. Kreitzman and R.L. Armstrong, Phys. Rev. B, 25 (1982) 2050; ff) H. Avram and R.L. Armstrong, J. Phys. C., 17 (1984) L89; (g) H. Avram and R.L. Armstrong, J. Phys. C, 18 (1985) L833; fh) H. Avram and R.L. Armstrong, J. Low Temp. Phys., 69 (1987) 391. 13 P. Brill and J. Voitlander. Ber. Bunsenges. Phys. Chem., 77 f 1973) 50. 14 T.C. Sheng and I.D. Gay, J. Caral., 77 (1982) 53. 15 T.P. Beebe and J.T. Yates. J. Am. Chem. Sot., 108 (1986) 663.
D.J. Barabino, C. Dybowski /Solid State Nucl. Magn. Reson. I (1992) 5-l-7
12 16 A. Palazov. CC.
17 18 19 20
21 22
Chang and R.J. Kokes. J. (1975) 338. D. Mesaros and C. Dybowski, Appl. Spectrosc., 610. H. Browdowski, Z. Phys. Chem. N.F., 44 (1965) P.A. Sermon, J. Catal., 24 (1972) 467. E.O. Stejskal and J. Schaefer, J. Magn. Reson., 560. P.H. Everett and P.A. Sermon, Z. Phys. Chem. (1979) 109. A.T. Nicol and R.W. Vaughan, Adv. Chem. (1978) 248.
Catal., 36 41 (1987)
129. 18 (1975) N.F., 114 Ser.,
167
23 M. van Stackelberg and P. Ludwig. Z. Naturforsch., Teil A,
19 (1964) 93. 24 (a) P. Gajardo, T.M. Apple and C. Dybowski, Chem. Phys. Letf, 74 (1980) 306; (b) T.M. Apple, P. Gajardo and C. Dybowski, J. Catal., 68 (1981) 103. 25 D.H. Everett and P.A. Sermon, Z. Phys. Chem., 114 (1979) 109. 26 X. Wu, B.C. Gerstein and T.S. King, J. Catal.. 121 (1990) 271. 27 T.W. Root and T.M. Duncan,
(1987) 57.
Chem.
fhys.
Lett.,
137
Elsevier Science Publishers B.V., Amsterdam
Nuclear magnetic resonance of hydrogen sorbed by powdered palladium metal and alumina-supported palladium Daniel J. Barabino ’ and Cecil Dybowski * Department of Chemistry and Biochemistry and Center for Catalytic Science and Technology, lJnil,ersiq of Delaware, Newsark, DE 19716, USA
(Received 23 July 1991; accepted in revised form 15 October 1991)
Abstract
Hydrogen in 0.2 pm P-palladium powder and in an alumina-supported palladium catalyst was investigated by NMR spectroscopy. The dependence of the shift on hydrogen pressure and the hydrogen-to-palladium (H/Pd) ratio in P-palladium hydride was determined for 273 K < T < 368 K. The activation energy for the process affecting the hydrogen shift is 9.4(+0.5) kcal mol-‘, similar to the enthalpy of transition from the (Y-to P-phase. NMR measurements of hydrogen in alumina-supported palladium show a similar behavior, which may allow one to use NMR shifts as a barometer of the state of the hydrogen in the metal. Keywords: palladium; hydrogen; catalyst; adsorption; NMR
Introduction Hydrogen is sorbed in large quantities by palladium metal. The hydrogen species formed are generally referred to as “hydrides”, “bulk-phase hydrogen”, or “absorbed hydrogen” [l]. Two phases have been identified from volumetric and magnetic-susceptibility isotherms: a-palladium hydride for H/Pd < 0.1 at T < 373 K and /?-palladium hydride for H/Pd > 0.6 at T < 373 K. Wicke er al. [21 determined the limiting values of H/Pd, denoted by (Y,,, and &,, at which the two bulk phases are stable at temperatures between 293 K and 571 K. At low temperatures (T=60K),Ppa lida ium hydride is reported to undergo a transition in electrical resistivity, ther-
’ Present address: Varian NMR Instrument Division, 6 Birchwood Court, Suite lK, Mineola, NY 11501, USA. * To whom correspondence should be addressed. 0926-2040/92/$05.00
ma1 expansion, and specific heat [3,4]. It has also been reported that palladium hydride becomes superconducting at low temperatures [5]. Supported palladium catalysts are somewhat more complex than bulk palladium. There is almost always a distribution of particle size. Other forms of hydrogen uptake may be significant [6]. Hydrogen is known to chemisorb at the surface at a stoichiometry of one hydrogen atom per surface palladium atom, requiring heat to remove it [7]. It binds at a stoichiometry of one hydrogen atom per surface palladium atom [l]. Hydrogen may also weakly adsorb [S], or reside at subsurface sites [9] or at lattice defects [ 101,or spill over onto the support. In early work, Norberg reported the NMR shifts of palladium hydride [ll]. More recently, diffusion, relaxation and tunnelling in palladium hydrides prepared by various techniques and in various morphologies have been addressed by NMR investigations [121. In one study [13], sam-
Q 1992 - Elsevier Science Publishers B.V. All rights reserved
6
D.J. Barabino, C. Dybowski/Solid
ples of different hydrogen content were investigated, but in many reports [12] samples with a single H/Pd ratio were examined. The NMR line for hydrogen on silica-supported palladium has been reported to fall in the range of - 80 to - 5 ppm relative to TMS, depending on the coverage 1141. In this paper, we report measurements of the proton NMR shift in powdered palladium hydride as a function of hydrogen content, pressure and temperature and in alumina-supported palladium. We use NMR spectroscopic measurements to estimate the amount of “internal” hydrogen in supported palladium catalysts.
Experimental Powdered palladium metal was obtained from Aesar, Inc. (Seabrook, NH). The reported purity was 99.980%. The average particle.diameter was reported to be 0.2(+0.1) pm. The material was used without further purification or sieving. The alumina-supported palladium catalyst was obtained from Sun Refining and Marketing Company (Marcus Hook, PA). Atomic absorption spectrometry indicated the sample was 4.1(+0.5)% w/w palladium. The nitrogen BET surface area of the palladium metal was determined to be 20(*5) m2 g-i and that of the catalyst was 185(* 10) m2 g-l. Carbon monoxide chemisorption on the catalyst gave a palladium surface area of 2.5cf0.3) m2 g-l. The dispersion of palladium on the catalyst was estimated to be 13( + 11% by the method of Beebe and Yates [151, with an estimated average particle diameter of 90( * 9) A [161. N 0.2 g of sample (either For experiments, palladium powder or the supported palladium catalyst) in a controlled-atmosphere NMR cell [17] was outgassed at 298 K using a glass, greasefree manifold. The sample was subsequently reduced in flowing H, (Airco, grade 5, 99.999%, 40 ml min-‘) at 373 (for the metal) and 673 K (for the catalyst) for 3 h to remove residual surface oxides [18]. A relatively low reduction temperature was used to avoid sintering the metal, which is known to occur at higher reduction tempera-
State Nucl. Magn. Reson. 1 (1992) 5-12
tures 1191. After reduction, samples were outgassed to 5 X 1O-6 torr at 373 K and cooled to 295 K under dynamic vacuum. The reduced sample was exposed to a pressure of hydrogen greater than 700 torr for 2 h. From this initial state, hydrogen was desorbed in several steps, with establishment of equilibrium at specific hydrogen contents in 20 min. After each step the sample was examined with NMR spectroscopy. Thus, conditions for the NMR experiments were always reached by desorption. We therefore observe no hysteresis effects that may be present [2]. The 90” pulse was 2.0 pus long and the relaxation delay was 5.0 s. Spin-temperature alternation was used to reduce baseline artifacts [20]. Each spectrum is the Fourier transform of 1000 transient decays. The spin-lattice relaxation time of protons in the powdered palladium hydride is N 80 ms at 298 K, as determined from the inversion-recovery null point. The temperature in the NMR spectrometer was controlled to +2 K. We found it necessary to reduce random errors by ensuring that the same region of the NMR cell was cooled or heated in the sorption bath and the spectrometer. All spectra are referred to an external sample of tetramethylsilane (TMS) and positive shifts are deshielded (downfield).
Results and discussion p-Palladium hydride
Figure 1 shows isotherms for hydrogen uptake in powdered palladium at several temperatures. To calibrate H/Pd, we referred the value of H/Pd at the transition, Pmin, to that reported on similar samples [13,211. From the temperature dependence (not shown) of pt, the transition pressure, we estimate AHap = 8.6Cf0.5) kcal mol-’ and AS,, = 20( &2) cal mol-’ K-r. In Table 1 we compare these results to literature values on samples having a variety of morphologies - powders, wires, and foils. From this comparison, one concludes that the 0.2 pm particles are similar to other macroscopic forms of palladium hydride.
D.J. Barabino, C. Dybowski/Solid
0.00
0.20
7
State Nucl. Magn. Reson. 1 (1992) 5-12
0.60
0.40
0.80
H/Pd
Fig. 1. Hydrogen uptake by palladium at various temperatures. The reference pressure, pO, is one atmosphere. (A) 273 K; (.I 295 K; (+) 323 K, CD) 383 K.
I D
Figure 2 shows proton NMR spectra of powdered P-palladium hydride at 295 K and different pressures. The resonance is deshielded relative to TMS at all conditions examined. This behavior is quite different from that of molecular hydrides, which are generally more shielded than TMS 122,231.Thus, the NMR spectroscopy of this bulk phase is quite different from that of other hydridic materials. In addition, these shifts are very different from those reported for hydrogen sorbed in metal particles supported on oxides [14,24]. The deshielding can be attributed to Knight shifts
n31. TABLE 1 Thermodynamic properties of palladium hydrides
Bulk
Powder
Ref.
“Hup (kcal mol-‘)
$i?nol-’
8.5 (kO.3) 9.3 (f0.1) 9.8 (+O.l) 9.44( + 0.05) 8.6 (kO.5) b
23.3( f 0.2) 21.8CkO.2) 20 (+21b
K-l) 18 2 4 a This work
a J.G. Aston, Engelhard Ind. Techn. Bull., 7 (1966) 14. b The uncertainties are given at 95% confidence.
JC
. . . . . . . .... . . . 4 LB
40
..I.
32 RESONANCE
.
24
.
.
.
.
.
.
16
,,.....
,.
8
SHIFT
Fig. 2. Proton NMR spectra of P-palladium hydride at 295 K. Hydrogen pressure: (A) 560 torr: (B) 190 torr; CC) 35 torr; (D) IO torr. crTMS= 0.
The proton resonance shift of powdered P-palladium hydride is a strong function of the pressure (Fig. 21, as observed by Brill and Voitlander [13]. The lines are narrow (FWHM = l-2 ppm), indicating the hydrogen must be experiencing no, or very weak, static dipole-dipole couplings. The hydrogen in P-palladium hydride must be rather mobile. At lower H/Pd ratios (i.e., in cu-palladium hydride and in the transition region between cr- and P-phases), the resonance is much broader. We have restricted these studies to the /?-phase. The pressure dependence of the NMR shift of hydrogen in P-palladium hydride (Fig. 3) was fit to the equation:
D.J. Barabino, C. Dybowski /Solid State Nucl. Magrl. Reson. 1 (1992) 5-12
b b -20
.---+A--2
3 1000/T
Fig. 4. Semi-logarithmic temperature. 0
1
2
3
4
5
plot of [u,(T)-
aI] ucrsus inverse
6
MM>
Fig. 3. Resonance shift of P-palladium hydride ((~1 referred to the shift at the cu-to-/3transition (a,) as a function of pressure and temperature. ( A) 273 K; (0) 295 K; (+) 323 K, (W) 383 K.
where a(p, T) is the shift, p is the hydrogen pressure, p,(T) is the transition pressure at temperature T, al(T) is a parameter describing the variation of the shift with temperature and pressure, and a,(T) is the shift extrapolated to the transition point, Pmin. The parameters are given in Table 2 at four temperatures. The region of validity of this equation is Pmin < H/Pd < 0.8. In Fig. 4 is shown the temperature dependence of (T,. (pi is empirically given by the equation: (~i(T) = a, + b, exp( -E,/RT)
where a, and b, are empirical parameters and E, is an activation energy. From the data of Fig. 4, E, = 9.4( *0.5) kcal mol-‘, similar to the enthalpy of transition, AHap, of 8.6 kcal mol-‘. To determine how the resonance shift depends on H/Pd, one must specify this parameter. We calibrated Pmin at each temperature to be in agreement with the literature value [2]. Figure 5 shows the shift as a function of (H/Pd)/P,,, at 1
4oh . *
-
(2)
Y
.
/
. i
b TABLE 2 NMR parameters of powdered palladium hydride; fits to eqn. (1)
-
/
B
Temperature CK)
a,(T) a (ppm)
V&? a (ppm)
Zorr)
273( * 2) 295( &-2) 323( + 2) 368( * 2)
2.99( f 0.30) 3.33( + 0.30) 4.63( f 0.50) 13.4 (k3.0)
19.4( + 2.4) 19.8( + 3.0) 15.7( f 2.4) 10.3( f 2.0)
2(&l) 7(kl) 20( i 1) 136(+ 1)
a The uncertainties arc given at 95% confidence.
10
0.90
1.00
1.10
1.20
1.30
WPd)/Om,r
Fig. 5. Dependence of the resonance shift in P-palladium hydride on (H/Pd)/P,i,. (A 1 273 K; (0) 295 K.
D.J. Bur-ahino, C. Dyhowski/Solid
H/Pd
Fig. 6. Isotherms for uptake of hydrogen by palladium powder (0) and alumina-supported palladium ( A) at 295 K.
273 and 295 K. From Fig. 5, the shift is seen to obey the following simple relation: a(H/Pd,
9
State Nucl. Magn. Re.son. I (1992) S-12
T) =A + B(H/Pd)/&,
(3)
where A and B are arbitrary parameters. Brill and Voitlander [13] observe a similar linear dependence of u on H/Pd. B is independent of temperature over this range and A seems to depend on the temperature. A measurement of (T at a given temperature is an indirect measure of H/Pd in P-palladium hydride. Alumina-supported palladium
In Fig. 6 we compare the isotherms for the uptake of hydrogen by powdered palladium metal and by the alumina-supported palladium. The slope in the transition region for the metal powder is zero, consistent with that expected from the phase rule. The transition region for the catalyst has a positive slope. We attribute this effect to a distribution of transition pressures for the catalyst sample, probably due to a distribution of particle sizes [25]. The isotherm for the catalyst gives the total uptake of hydrogen in all forms. By extrapolation of the high-pressure linear portion of the isotherm, (Pminjtot for the catalyst is found to be 1.34( f 0.051, a value substantially larger than 0.61, the value of Pmin for the metal powder. Thus, in addition to uptake into the body
of the metal particles there must be other modes of adsorption that contribute to hydrogen uptake on the small particles. Figure 7 compares the NMR spectrum of hydrogen sorbed by palladium powder with that of Pd/Al,O, at the same equilibrium hydrogen pressure, 82 torr. The mean resonance shifts are the same, indicating that both systems have similar electronic environments. (The small peak at 4.5 ppm in the spectrum of the powder results from a slight amount of water that was not removed during the pretreatment of this sample. The signal-to-noise ratios of the two spectra in A and B are a reflection of the different number of spins in the two samples.) The principal difference between the two spectra is the line width, with the resonance of hydrogen on alumina-supported palladium being somewhat broader than that of hydrogen sorbed in the metal powder. This broadening reflects a distribution of environments on the supported palladium. In Fig. 8 are data for the resonance shift of hydrogen of the powdered metal and the alumina-supported palladium as a function of hydrogen pressure at 295 K. Within the precision of the data (particularly
A_ A
a,
wm
Fig. 7. Proton NMR spectra of hydrogen in (A) palladium powder and (B) an alumina-supported palladium catalyst in equilibrium with hydrogen at 82( i 2) torr at 295 K.
D.J. Barabino, C. Dybowski/Solid
p
I Resonance
Shift
30
@pm1
those of the alumina-supported palladium), these data lie along a common curve in the region above pmi,. Assuming the relationship between H/Pd and shift found for the bulk P-palladium hydride is valid for the small particles on the alumina-supported material, one may construct an NMR-determined isotherm for only “internal” hydrogen, i.e, the hydrogen in an environment like that in the powdered P-palladium hydride. Such an isotherm is given in Fig. 9, along with the isotherm for the powdered palladium hydride. We ex-
0.
-1.0
2 2
Y -2.0
0.2
0.4
0.6
100
200
300
400
6
IO
Pressure km)
Fig. 8. Resonance shift as a function of hydrogen pressure for hydride formed from the metal powder (0) and alumina-supported palladium (A ) at 295 K.
0.
0.6.
0 20
10
0
State Nucl. Magn. Reson. 1 (1992) 5-12
Ii
0
HlPd
Fig. 9. NMR-determined isotherm for the “inner” hydrogen in small palladium particles supported on alumina, as described in the text. (a) Bulk palladium particles; (A) supported palladium particles.
Fig. 10. Excess hydrogen uptake, determined by difference between the total uptake and the NMR-derived uptake into the supported palladium.
tended the isotherm into the transition region by this procedure, although this may not be valid. However, the striking similarity between the isotherm for the powdered metal and the supported palladium would seem to indicate that the results are at least qualitatively correct. There are subtle differences between the two isotherms. The slight shift between them above Pmin is probably not meaningful. The difference between them in the transition region may be the result of the distribution of particle sizes in the catalyst, resulting in a distribution of transition pressures for this sample. The difference between the volumetric isotherm and the NMR-derived isotherm we label the excess H/Pd. This hydrogen is sorbed by the catalyst through other modes of uptake, e.g., chemisorption at the surface or spillover to the surface [26]. We show this quantity, H/Pd (excess), as a function of pressure in Fig. 10. The amount of excess H/Pd appears to increase with the hydrogen pressure and saturate at a value of 0.97( _~r 0.04). The solid line is a fit of the data to a Langmuir isotherm. If the only mode of sorption aside from incorporation in the lattice were chemisorption at the surface, one would expect it to saturate at a value near 0.13, based on the Beebe-Yates measurements of the dispersion. Assuming that, at saturation, the surface sites for a sample with a dispersion of 0.13 contribute 0.13 H/total Pd, there must be a third species which comprises 0.85 H/total Pd. This could be
D.J. Barabino, C. Dybowski /Solid
State Nucl. Magn. Reson. I (1992) S-12
physisorbed material or spilled-over hydrogen, which is invisible to NMR spectroscopy. This may be the so-called “recondite” (hidden) hydrogen proposed by Root and Duncan to be present in a supported metal catalyst investigated with NMR spectroscopy [27].
11
of the shifts of hydrogen in bulk palladium hydride and in the supported palladium material in equilibrium with the same pressure of hydrogen, this seems reasonable. However, the validity of this assumptions needs to be tested rigorously.
Acknowledgements Conclusions Hydrogen desorption isotherms of powdered palladium (with particle diameters N 0.2 pm) give thermodynamic properties similar to those of palladium foils and wires [12]. The NMR shift of powdered p-palladium hydride depends strongly on other parameters of the system. The resonance is deshielded relative to TMS, in contrast to molecular hydrides. These results are also quite different from those reported for hydrides in small metallic particles on the surface of oxide supports. We have empirically specified how the NMR shift depends on characteristics of the system such as the hydrogen pressure, temperature, and, by calibration, the hydrogen-to-palladium ratio. The shift at Pminr the p-phase limit, tends to decrease with temperature, i.e., it becomes more shielded. We have shown that in the /3phase the shift is linear in the hydrogen-to-palladium ratio. For alumina-supported palladium, we find the same shifts as for the bulk powder, when comparing samples in equilibrium with the same hydrogen pressure. From this fact, we assign an “internal” H/Pd content to the catalyst. By difference, we determine the excess H/Pd at any pressure. This value exceeds the dispersion of the catalyst. We infer that the excess H/Pd is attributable to at least two kinds of hydrogen: chemisorbed hydrogen and other adsorbed hydrogen. At Pmjn, we find roughly 48(+4)% of the hydrogen is absorbed into the bulk of the particle, lO(&-2)% of the hydrogen is chemisorbed at the surface and the remainder, 42(*5)%, exists in a third form which is neither. This analysis is only valid if the shift of the observed hydrogen resonance can be used to infer the “internal”, i.e., palladium-hydride-like, hydrogen in the particles. Based on the similarity
We gratefully acknowledge a Hercules grantin-aid, the Sun Refining and Marketing Company, the donors of the Petroleum Research Fund and the Sponsors of the Center for Catalytic Science and Technology for support of this work.
References 1 E. Wicke and H. Browdowski, Hydrogen in Palladium and Palladium Alloys. in G. Alefeld and J. Volkl (Eds.), Hydrogen in Metals, II, Springer Verlag, Berlin, 1978. p. 73. 2 (a) H. Frieske and E. Wicke, Ber. Bunsenges. Phys. Chem., 77 (1973) 50; (b) E. Wicke and G.H. Nernst, Ber. Bunsenges. Phys. Chem., 68 (1964) 224. 3 B.M. Geerken, H. Hemmes and R. Griessen, J. Phys. F, 14 (1984) 813. 4 J.K. Jacobs and F.D. Manchester, J. Low Temp. Phys., 4 (1971) 129. 5 B. Strizker and W. Buckel, Z. Phys., 257 (1972) 1. 6 (a) P.C. Aben. J. Catal., 10 (1968) 224; (b) P.A. Sermon, J. CataL, 24 (1972) 460; fc) J.E. Benson, H.S. Hwang and M. Boudart, J. Cauzl., 30 (1973) 146. 7 A.W. Aldaq and L.D. Schmidt, J. Catal.. 22 (1971) 260. 8 P. Chou and M.A. Vannice, J. Catal., 104 (1987) 1. 9 (a) R.J. Behm, V. Penka. M.G. Cattania. K. Kristmann and G. Ertl. J. Chem. Phys.. 78 (1983) 7486; (b) R.J. Behm, V. Penka, M.G. Cattania, K. Kristmann and G. Ertl, Su$, Sci.. 126 (1983) 382. 10 H. Kubicka, J. Catal., 5 (19661 39. 11 R.E. Norberg, Phys. Rec., 86 (1952) 745. 1’A (a) J.P. Burger, N.J. Poulis and W.P.A. Hass, Physica, 27 (1961) 514; (b) G.K. Shoep, N.J. Poulis and R.R. Arons, Physica, 74 (1974) 297; (c) E.F.W. Seymour, R.M. Cotts and D.W. Williams, Phys. Rec. Lett.. 3 (1975) 165; (d) S.R. Kreitzman and R.L. Armstrong, Phys. Rev. B. 25 (1982) 2046; (e) S.R. Kreitzman and R.L. Armstrong, Phys. Rev. B, 25 (1982) 2050; ff) H. Avram and R.L. Armstrong, J. Phys. C., 17 (1984) L89; (g) H. Avram and R.L. Armstrong, J. Phys. C, 18 (1985) L833; fh) H. Avram and R.L. Armstrong, J. Low Temp. Phys., 69 (1987) 391. 13 P. Brill and J. Voitlander. Ber. Bunsenges. Phys. Chem., 77 f 1973) 50. 14 T.C. Sheng and I.D. Gay, J. Caral., 77 (1982) 53. 15 T.P. Beebe and J.T. Yates. J. Am. Chem. Sot., 108 (1986) 663.
D.J. Barabino, C. Dybowski /Solid State Nucl. Magn. Reson. I (1992) 5-l-7
12 16 A. Palazov. CC.
17 18 19 20
21 22
Chang and R.J. Kokes. J. (1975) 338. D. Mesaros and C. Dybowski, Appl. Spectrosc., 610. H. Browdowski, Z. Phys. Chem. N.F., 44 (1965) P.A. Sermon, J. Catal., 24 (1972) 467. E.O. Stejskal and J. Schaefer, J. Magn. Reson., 560. P.H. Everett and P.A. Sermon, Z. Phys. Chem. (1979) 109. A.T. Nicol and R.W. Vaughan, Adv. Chem. (1978) 248.
Catal., 36 41 (1987)
129. 18 (1975) N.F., 114 Ser.,
167
23 M. van Stackelberg and P. Ludwig. Z. Naturforsch., Teil A,
19 (1964) 93. 24 (a) P. Gajardo, T.M. Apple and C. Dybowski, Chem. Phys. Letf, 74 (1980) 306; (b) T.M. Apple, P. Gajardo and C. Dybowski, J. Catal., 68 (1981) 103. 25 D.H. Everett and P.A. Sermon, Z. Phys. Chem., 114 (1979) 109. 26 X. Wu, B.C. Gerstein and T.S. King, J. Catal.. 121 (1990) 271. 27 T.W. Root and T.M. Duncan,
(1987) 57.
Chem.
fhys.
Lett.,
137