The Adsorption and Absorption of Hydrogen by Metals

BY R. BURCH

The reversible interaction of hydrogen with metals is an essential part of many catalysed reactions, is used to determine specific surface areas, and offers the prospect of using metal hydrides for hydrogen storage. A number of comprehensive reviews dealing with various aspects of H in metals have appeared and reference is made to these at appropriate points in the text. It is not our intention, therefore, to attempt an in-depth survey of the interaction of H with metals. Instead, we shall trace the progress of a H atom from its initial state in a gaseous molecule through a variety of intermediate adsorbed states to its final absorbed state.

Problems of current interest to which we draw attention in this review are (1) the nature of adsorbed hydrogen, (2) the possibility of weak adsorption in excess of a monolayer and its influence on surface area determinations, (3) the adsorption/ absorption transition and the mechanism of absorption, and (4) the selectivity of H for special sites in alloys and the structural modifications in alloys caused by H. Finally, we shall comment briefly on the extent to which existing theoretical models can account for some of these features.

1 Adsorption of Hydrogen

Many different forms of adsorbed hydrogen have been observed or postulated. These include molecular hydrogen, positively and negatively charged H atoms, weakly and strongly bound states, etc. We consider first the adsorption of H on Pt(111) surfaces in the knowledge that similar adsorption occurs on other Pt surfaces, and in the expectation that Pt is representative of face-centred-cubic metals in general.

Pt(111) Single Crystal Surfaces. — Somorjai and co-workers investigated the adsorption of hydrogen on Pt(111) at elevated temperatures and reported that hydrogen adsorbed readily only on to stepped Pt(111) surfaces. This is not correct. Hydrogen adsorption on any clean Pt surface is rapid even at low temperatures (initial sticking coefficient ≈ 0.1). The discrepancy arises because the comparatively small heat of adsorption of hydrogen on Pt means that in experiments in vacuo above room temperature the equilibrium amount of adsorbed H is very small.

Thermal desorption spectra (TDS) of hydrogen preadsorbed at 78 K show three peaks (β1, β2, and β3) with Tmax at 140, 230, and 310 K. Preadsorption at 150 K gives two peaks in the desorption spectrum. Typical spectra are shown in Figures 1(a) and 1(b). The total surface coverage is about a mono layer (θ = 1). Christmann et al. have calculated heats of desorption (Ed) from TDS and from adsorption isotherms and obtained values of 30 and 40 kJ mol-1 for the β2 and β3 states, respectively. In contrast, from spectra having al most identical Tmax McCabe and Schmidt 4 calculate values of 53 and 75 kJ mol-1 for β2 and β3, and 33 kJ mol-1 for β1. However, they report a strong composition dependence of Ed over the range 0.03 < θ < 0.25, so that Ed ≈ 42 kJ mol-1 at θ = 0.25. The discrepancies at low coverage (θ < 0.25) may be due to differences between the surfaces examined (the high value of Ed at θ= 0.03 may reflect preferential adsorption at surface defects), or because of errors in the calculation of Ed. King has pointed out a possible error when deriving Ed from TDS if weakly bound intermediate states exist. Schwartz et al. also describe how errors can arise if the wrong order of reaction is assumed. By utilizing the entire desorption spectrum of H/Ti they show that the order of the desorption reaction is 1.5, yielding a value for Ed of 88 kJ mol-1. Analysis using only Tmax and assuming a second-order reaction gave Ed = 210 kJ mol-1.

McCabe and Schmidt note that the β3 state for H/Pt(111) could equally well be described by second-order kinetics or by first-order kinetics with a variable Ed. However, H2/D2 exchange experiments confirm that β2 and β3 are both atomic states. Adsorbed atomic hydrogen could still desorb as molecular hydrogen with first-order kinetics if only a single surface site is involved in the desorption. This seems to be the case for H on Pd at high surface coverages (see later), but is unlikely to be important for the β3-state which only desorbs when θ ≤ 0.5.

Work function data at 150 K show that after a small positive maximum (2 meV), attributed to preferential adsorption of Hδ- at surface steps, the work function decreases continuously with H coverage (Hδ+) to -230 meV. Neither a change in polarity nor even a discontinuity in the dipole moment is observed when the β2 state begins to fill after completion of the β3 state. The sign of the work function change is opposite to that found with Ni or Pd. However, in all three cases the dipole moment is small, indicating essentially covalent bonding.

The distribution of H atoms on Pt(111) has been considered. Christmann et al. reject the possibility that the β2 and β3 states correspond to different geometric locations. Toya had previously proposed two types of adsorbed H, an 'r-adatom' situated above a single metal atom and located outside the electronic surface of the metal, and an 's-adatom' situated at a surface interstitial site. However, this is not consistent with the observed smooth decrease in work function up to θ = 1.0. Christmann et al. suggest instead that repulsive interactions between H atoms produce an ordered structure in which H desorbs from the β3 state by the recombination of two H atoms when neighbouring sites are empty, while H desorbs from the β2 state when neighbouring sites are occupied.

It is difficult to determine whether H atoms are adsorbed above or between metal atoms. Flores et al. have presented a model for H/Pt(111) which is consistent with the data described above, and also with n.m.r., i.r., and neutron inelastic scattering data, all of which point to H bound to more than one Pt atom. In this model the H — H repulsions originate from the interaction of the screening potentials around the hydrogens. Their model (Figure 2) is interesting with regard to absorption (see later) because it shows that as θ -> 1.0 both forms of hydrogen (β2 and β3) become equivalent. Only when hydrogen is being adsorbed or desorbed are β2 and β3 distinguishable. A very similar structure for Ni(l 11) has been proposed on the basis of LEED results.

The Influence of Surface Imperfections. Bernasek and Somorjai suggested that the rate of the H2/D2 exchange reaction was 104 times faster at surface steps than at terrace atoms, and, therefore, that surface steps are necessary for the facile dissociation of hydrogen molecules. This is incorrect. At low coverages the sticking coefficient is four times higher for a stepped surface than for a flat surface, but above θ= 0.25 the two surfaces are similar. On a stepped surface there is a clearer distinction between the β2 and β3 states as shown by the fact that the β3 state is almost saturated before adsorption into the β2 state commences. [Compare Figures 1(b) and 1(c).] However, the positions of the desorption maxi ma are hardly altered. Similarly, when a Pt(111) surface is distorted by argon ion bombardment a small shoulder is observed on the high-temperature side of the β3 peak, but otherwise the desorption spectra remain unaltered [Figure 1(d)].

With respect to the absorption of H the important conclusions from single crystal data for Pt are first, that as θ -> 1 all the adsorbed H atoms become equivalent and have the same binding energies, and second, that surface imperfections enhance strong adsorption at low surface coverages, but have little effect as θ approaches a monolayer.

Polycrystalline Platinum. — Three desorption peaks are observed with Pt powders (170, 250, and 360 K) and films (120, 200, 330 K) [compare Pt(111)]. Calculated heats of desorption for Pt films are 34, 50, and 88 kJ mol-1. Similar values are found with Pt wire (71 kJ mol-1 at θ= 0.37 decreasing to 46 kJ mol-1 at θ= 0.46). Surface potential measurements on Pt films are consistent with work function data for single crystals. In both cases the most strongly bound H, located at surface imperfections, has a partial negative charge, while H on terrace sites has a partial positive charge. Dus and Tompkins also report a weakly bound (5 kJ mol-1 molecular species, H2, which only adsorbs at high pressures.

Additional States of Hydrogen. Temperature programmed desorption (TPD) of H from Pt blacks between 77 and 670 K revealed four peaks (α, β, γ, δ) with Tmax of 170, 250, 360, and 570 K. A further peak (ε, Tmax ≈ 470 K) has been reported [see Figure 1(e)], and a very stable form of hydrogen (Ω) is retained by Pt black even after outgassing at 630 K. Dixon et al. detected five states of hydrogen adsorbed on alumina-supported Pt. The α, β, and γ states of Tsuchiya et al. apparently correspond to the β1, β2, and β3 states found on single crystal Pt. However, this may be misleading because unlike the β2 and β3 hydrogens, β and γ hydrogens are not interchangeable. Thus, H and D do not equilibrate if, for example, H is adsorbed into the β state and D into the y state. At the same time the total amount of β + γ + δ hydrogen is apparently constant, which led Tsuchiya et al. to propose that all three states used the same surface sites. It was suggested that α, β, γ and δ hydrogen corresponded respectively to linearly bonded H2 molecules, bridge-bonded H2 molecules, linearly bonded H atoms, and bridge-bonded H atoms, with all four types of hydrogen adsorbed on the external surface of the Pt.

This model is not consistent with recent experimental results. Single crystal data show that all hydrogen adsorbed above 150 K is atomic, and that the β2 and β3 states are on geometrically equivalent sites. It is also difficult to see how γ and δ hydrogens can both be on the Pt surface and yet not be in equilibrium. It seems more probable that the γ and δ hydrogens are 'incorporated' in different ways into the Pt, so that H transfer to and from these adsorption sites is a slow activated process.

Further evidence that the δ, ε, and ΩH are not surface states comes from the kinetics of their formation and removal. Moger et al. observe that the γ state fills long before the ε or δ states are occupied. At room temperature ε and δ states are not formed from the γ state even after 80 h. Moreover, γ, ε, or δ hydrogen can only interact with oxygen at temperatures close to those at which desorption would occur. Szabo et al. observed that after covering the surface of Pt black with gold the TPD spectrum was almost identical to that for clean Pt, but the electrolytic charging curve was quite different. They conclude that most of the hydrogen is dissolved in the Pt rather than adsorbed on the surface. They also report that on Pt black only 13% of the adsorbed H is on the external surface.

Stephen et al. have also reported the slow formation of a type of hydrogen which subsequently desorbed at 330 K. The magnitude of the desorption peak was dependent on the time of adsorption at 78 K. About 1% of a monolayer of extra hydrogen is adsorbed in 10 minutes.

Further evidence for 'occluded hydrogen' comes from the work of Paal and Thomson and Wells. Paal and Thomson found that tritium was retained by Pt black at 630 K long after the initial exposure to gaseous tritium was terminated. They propose two forms of occluded hydrogen (ΩI and ΩII), the first of which exchanges with gaseous hydrogen, hydrogenates ethylene, hexene, and cyclohexene, and is oxidized by oxygen, and the second of which only exchanges with hydrogen. Wells observed with Pt powders that after evacuation at 373 K about 10% of a monolayer of H was readily available for H2/D2 exchange or butene hydrogenation, and that a further 40% of a monolayer became available over a period of several days. Wells postulates a cavity model to explain his results. Another possibility is that his readily accessible hydrogen is located along dislocations or low-angle grain boundaries. The very strongly bound hydrogen reported by Paal and Thomson (ΩII) could then be located on the internal surfaces of voids within the Pt crystallites. However, it is also possible that this hydrocarbon-inaccessible hydrogen is present in a combined form (hydroxyl?) which can exchange with tritium atoms but cannot desorb as hydrogen molecules.

We can summarize the data for polycrystalline Pt as follows. There are seven or eight types of adsorbed hydrogen, half of which refer to hydrogen adsorbed on the external surface of the Pt, and the remainder refer to hydrogen incorporated in different forms at dislocations, grain boundaries, or inclusions. In view of the extent to which H can be incorporated into Pt black it is perhaps surprising that good agreement is often obtained between surface areas of Pt powders measured by krypton adsorption and by hydrogen chemisorption. It is possible that the 'correct' amount of adsorbed hydrogen corresponds to a partial coverage of the external surface combined with a contribution from occluded hydrogen. Particularly in cases where there is any doubt about the cleanliness of a Pt surface, H chemisorption is unlikely to be a reliable method for the determination of metal surface areas. Even with Pt single crystals it is interesting that enhanced dissolution seems to occur, possibly in the vicinity of surface steps.

2 Adsorption and Absorption of Hydrogen by Palladium

Palladium is typical of a number of metals which unlike Pt are capable of not only adsorbing but also absorbing large quantities of hydrogen. A collection of excellent articles on various aspects of the absorption of hydrogen by metals has been published recently, to which the reader is referred for more detailed accounts of recent work.

Adsorption of Hydrogen. — Pd Single Crystals. Conrad et al. investigated the adsorption of hydrogen on Pd(110) and Pd(111) surfaces and found many similarities with Pt. Above room temperature there is a single desorption peak at about 350-360 K for both surfaces. The heat of desorption is 96 kJ mol-1 for Pd(110) and 88 kJ mol-1 for Pd(111), and is constant up to θ [greater than or equal to] 0.5. On a stepped surface the heat of desorption at low coverages increases to 97 kJ mol-1 [compare Pt(111)]. Conrad et al. have also observed a large flat peak at about 670 K due to the desorption of absorbed hydrogen. This peak appears after the desorption of strongly chemisorbed hydrogen because the rate of desorption of dissolved H is determined by the rate of diffusion from the bulk to the surface.