Platinum Metals Rev., 2001, 45, (4), 166
Exploring Lattice Defects in Palladium and its Alloys Using Dissolved Hydrogen
Part II: Hydrogen Segregation to Internal Interfaces and to Inhomogeneous Regions
- By Ted B. Flanagan
- Materials Science Program and Chemistry Department, University of Vermont, Burlington
- R. Balasubramaniam
- Department of Materials and Metallurgical Engineering, Indian Institute of Technology, Kanpur
- R. Kirchheim
- Institut für Materialphysik, Hospitalstrasse 3-7, 37073 Göttingen, Germany
The first part of this paper, published in the July issue of this Journal, described the combination of physical metallurgical techniques and hydrogen solubilities used to help in characterising defects in palladium (Pd) and Pd alloys. In this second part, the solubilities of hydrogen (H2) in internally oxidised Pd alloys are discussed. Internal oxidation, for example of a palladium-aluminium alloy, results in the formation of small alumina precipitates within the Pd matrix. Dissolved H in the alloy is strongly trapped at the metal/oxide interface. This can be detected by deviations in H solubility from that expected for Pd. Hydrogen in Pd and its alloys has been modelled mathematically as the occupation by H atoms of interstitial sites within fixed metal sublattices. However, recently it has been realised that at moderately high temperatures and H2 pressures some alloy lattices are not fixed, as the dissolved H promotes metal atom diffusion. This results in phase separation in some alloys, for example (Pd + Pt + H), according to a ternary equilibrium. The dissolved H can be removed from such alloys at low temperatures allowing the metastable, phase-separated alloy lattices to be characterised via measurements of H2 solubilities and suitable physical metallurgical techniques.
Alloys containing solutes that are more readily oxidised than their solvent metal can be internally oxidised in oxygen gas (O2(g)) at elevated temperatures ≥ 1000 K. The oxidation results in the formation of essentially pure metal matrices containing nanometre-sized internal oxide precipitates. In effect, a metal/oxide composite is formed (39). Binary palladium (Pd) alloys with solute metals, such as aluminium (Al), magnesium (Mg) and zirconium (Zr), have been internally oxidised to form essentially pure Pd matrices containing second phases of nanosized oxide precipitates (40–42).
Internal Oxidation of Pd Alloys Such as Palladium-Aluminium
After internal oxidation of, for example, Pd0.97Al0.03 alloy, nanocrystalline alumina precipitates are visible in TEM photomicrographs. The size of the precipitates is greater after internal oxidation of Pd- alloys (Pd is the majority element in the alloy) at 1273 K than at 1073 K (17, 40, 42, 43). Huang found that the precipitates resulting from oxidation at 983 K were about one tenth as large as those from internal oxidation at 1273 K (17).
Internal oxidation of alloys proceeds by dissociation of O2(g) at the alloy surface, diffusion of interstitial O atoms into the alloy, oxidisation of the solute atoms and their agglomeration into oxide precipitates. In order to relieve the internal compressive stress resulting from the growth of the Oxide Precipitates, Metal Atoms Are Transported from the metal/oxide interface to the surface and vacancies must therefore be transported in the opposite direction during the process of internal oxidation (44, 45). Palladium nodules have been observed at the surface and on grain boundary walls resulting from the internal oxidation of Pd- Al alloys (46).
It has been shown that the kinetics of H2 absorption are enhanced after internal oxidation of Pd-Al alloys (47). Internal oxidation thus appears to be a convenient method of obtaining enhanced kinetics for Pd and for some Pd alloys, particularly Pd-Ag . It may, for instance, provide a means for obtaining improved diffusion membranes. There is also some evidence that the internal oxidation of Pd- Al alloys renders them more resistant to CO poisoning than before their internal oxidation (48).
H2 Solubility in Internally Oxidised Pd Alloys (Pd/MOx Composites)
The total area of the Pd/oxide interface in internally oxidised alloys can be several m2 per cm3 of alloy (49). These internal metal/oxide interfaces are generally free from impurities and so provide a way to investigate the segregation of solutes to impurity-free interfaces. Huang, Kirchheim and coworker were the first to investigate the segregation of hydrogen atoms to Pd/oxide interfaces in internally oxidised alloys of Pd- Mg, Pd- Al and other Pd-rich alloys (41). Electrochemical methods were employed to measure the relative H chemical potentials as a function of r . ΔμH – r relationships were obtained for internally oxidised alloys in the very dilute solution region where H trapping at the interface is most important; μH chemical potentialH of an isolated dissolved H in the solid phase, μ°2 is the chemical potential in the gaseous phase, and r is the H/Pd atom ratio. ΔμH is very negative for this H trapping and the differences in these relationships for pure Pd and the internally oxidised alloys gives the amounts of trapped H.
More recently, Flanagan, Balasubramaniam and their coworkers (42, 43) introduced H via the gas phase to measure μ°H r relationships in the dilute and concentrated H phases of internally oxidised Pd- Al. Aside from the very dilute trapping region, the H2 isotherms for internally oxidised Pd- Al alloys corresponded very closely to those for pure Pd. This confirms that all the Al has been oxidised to form alumina precipitates within an essentially pure Pd matrix, that is, a metal/ceramic composite is formed (42).
A partially internally oxidised Pd -Al alloy has two plateaux in its isotherm, one for the internally oxidised portion, Pd-H, and the other for the unoxidised portion, as seen in Figure 6 for a Pd0.97Al0.03 alloy. The two plateaux demonstrate the interfacial nature of the internal oxidation because if internal oxidation took place homogeneously there would be only one plateau with a hydrogen pressure (pH2) between the pressures of Pd and the Pd0.97Al0.03 alloy. The relative lengths of the plateaux in Figure 6 can be used to obtain the fraction of internal oxidation, while the transition region between the two plateaux provides information about the diffuseness of the interface and the metal atom diffusion constants.
Dilute phase H2 solubilities (at 273 K), which indicate that H has been strongly trapped at the Pd/Al2O3 interface, are shown in Figure 7 for an internally oxidised (at 1073 K) Pd0.97Al0.03 alloy (50). The initial solubility has an intercept along the r axis at 0.002 which corresponds to the interfacially trapped H, in agreement with the electrochemical results of Huang et al . (41). While still in the dilute phase, the alloy was evacuated (at 323 K); the remeasured solubility data pass through the origin showing that only the strongly trapped H cannot be removed by evacuation at 323 K.
In addition to the strong H trapping at the interface where essentially PH2 = 0, there is a solubility enhancement in the dilute region, which is attributed to H segregation in the stress field about the precipitates. The stress field results from cooling the two-phase material (Pd/Al2O3 which has two different coefficients of thermal expansion) from its high oxidation temperature. Calculations made on the basis of this thermal residual stress give reasonable agreement with experiment (43).
The internally oxidised Pd0.97AlO.03 alloy was cycled through the hydride phase and evacuated (at 323 K). In contrast to the solubility remeasured after evacuation at 323 K, there is again a positive intercept on the r axis, indicating that the strongly trapped H can be removed by cycling and evacuation (323 K), but not by just evacuation (Figure 7). The H, which initially was strongly trapped, was shown not to be present after cycling – none was evolved when the cycled, internally oxidised alloy was heated to 573 K. However, trapped H is evolved by such heating of an uncycled internally oxidised alloy (50). Dislocations in the vicinity of the interfaces, resulting from cycling, may decrease the depth of the potential wells and allow the initially strongly trapped H to be removed.
Annealing of Dislocations in Pd and in Cycled, Internally Oxidised Pd- Al
Annealing at 623 K (24 h) causes a decrease in the hydrogen solubility enhancement of Pd which has been cycled and a decrease in the dislocation density – judging from TEM photomicrographs. After annealing cycled Pd at 723 K (24 h), its H2 solubility returns almost to that of well-annealed Pd (Figure 8). These results show that dislocations become mobile in pure Pd at ≥ 623 K allowing them to rearrange and to annihilate sufficiently to reduce the dislocation density and solubility enhancement.
TEM photomicrographs show only small dislocation densities in internally oxidised Pd- Al alloys (17, 40, 42). However, after hydriding/dehydriding (cycling), large dislocation densities and extensive intergranular cracking appear (51). Dilute phase solubilities (273 K) for an internally oxidised (1073 K, 72 h) Pd0.97Al0.03 alloy (a Pd/Al2O3 composite) were measured before and after cycling (Figure 9). Cycling the internally oxidised alloy leads to a greater solubility enhancement than cycling Pd does; this is consistent with reports of increased dislocation densities from cold working metals which contain small internal precipitates (52, 53).
Internal oxidation is of practical value because it is used for dispersion hardening of alloys (39). In contrast to Pd (Figure 8), when the cycled internally oxidised Pd0.97Al0.03 alloy was annealed for 48 h in vacuo at 823 K, there was no change in its solubility enhancement (Figure 9). The dislocations in this internally oxidised alloy are therefore immobile even though the annealing temperature is higher and the time taken was longer than that used for cycled Pd (24 h) where the solubility enhancement nearly vanished (Figure 8).
A Pd0.97Al0.03 alloy, internally oxidised at 1273 K and cycled, was annealed at 823 K for 48 h. After this its solubility enhancement almost completely disappeared in contrast to the alloy internally oxidised at 1073 K. This is in keeping with the expectation that the small, closely spaced precipitates resulting from internal oxidation at 1073 K interact strongly with dislocations, while the more widely spaced and larger precipitates from internal oxidation at 1273 K do not (45).
Δ treated in H: (5 h. 100 MPa, 673 K) (redrawn after (56) )
H-Induced Phase Separation of Palladium-Rhodium Alloys
Unlike most other binary f.c.c. Pd alloys whose H2 solubilities have been studied, the Pd/Rh alloy system has a miscibility gap with a critical temperature of ∼ 1120 K. At 873 K, for example, the coexisting phases have Rh atom fractions of about 0.12 and 0.87 (54, 55). Using X-ray diffraction (XRD), Raub et al . (54) showed that at 873 K the phase separation is extremely slow: 63 days of annealing were required for a Pd0.49 Rh0.51 alloy, while a Pd0.74Rh0.26 alloy had not phase-separated after 1 year. However, H2 solubilities are more sensitive than XRD for detecting a phase separation.
In Figure 10, H2 solubilities (303 K) following the cooling of a Pd0.80Rh0.20 alloy at different rates from elevated temperatures are shown. Significant solubility differences can be seen between alloys prepared:
A greater solubility at a given pH2 indicates that there is a greater degree of phase separation. Noh et al . (56) showed that metal atom diffusion, and consequently phase separation, is greatly accelerated by dissolved H2. This is the reason why there is a greater dilute-phase H2 solubility after the alloy is heated at 673 K in the presence of 100 MPa H2 (Figure 10), than after treatments under H-free conditions.
Fukai and coworkers (57) have shown that at ultra high pH2 (5 GPa) a Pd0.80Rh0.20 alloy separates into two phases in 120 seconds at 873 K, according to the phase diagram. It is clear that dissolved H at ≥ 673 K greatly accelerates the phase separation of metastable, homogeneous Pd -Rh alloys (56–58).
Other Palladium Alloys
Palladium-metal (M) alloys, which are not metastable, have been found to phase-separate in the presence of H2 at relatively moderate temperatures. For these alloys phase separation can occur via a ternary (Pd + M + H) equilibrium which can be established at a sufficiently high pH2 for appreciable H to dissolve. M is a solute such as platinum (Pt), nickel (Ni), or cobalt (Co) where the plateau pressures pH2 of the alloy are greater than for Pd-H and thus phase separation leads to a lower free energy. By contrast, for Pd alloys where the plateau pressures pH2 are lower than for Pd-H, phase separation via a ternary equilibrium has not been observed (59, 60). For alloys such as Pd- Pt, H has both an equilibrium and a kinetic role, however, for the metastable Pd- -Rh alloys, the kinetic role appears to be dominant.
Pd0.80Pt0.20 is an example of an alloy where phase separation occurs via a (Pd + M + H) ternary equilibrium. After phase separation by H2 treatment at 100 MPa (673 K) and subsequent low temperature removal of H, H2 isotherms (273 K) were measured on the metastable, phase-separated alloy and compared to isotherms for the homogeneous alloy before the H2 treatment (Figure 11). Before the H2 treatment there is no plateau region because of the relatively large fraction of Pt. After the H2 treatment there is a well-defined plateau reflecting the separated Pd-rich phase.
It should be emphasised that the dissolved H must also have a kinetic role because usually metal atom equilibrium would not be expected to be readily established at 673 K.
This phase separation is on a fine scale not detectable by XRD, however, it has been detected by small angle neutron scattering (SANS) (59) and, very recently, by high resolution electron microscopy (HREM) (61). Using the latter method, small Pt-rich precipitates were observed within a Pd-rich matrix after a H2 treatment (673 K) (Figure 12). These HREM results directly confirm the indirect evidence from the H2 isotherm measurements (Figure 11).
Values for the H diffusion constants, measured before and after H2-treatment of the Pd0.80Pt0.20alloys, also support the occurrence of phase separation. It is known that the diffusion constants of H in f.c.c. Pd-rich alloys decrease with increase of solute concentration (62) and, since diffusion is found to be faster after the H2-treatment, it follows that the majority of the diffusion occurs within a principal phase which must have a lower Pt content than the original alloy.
Effect of Crystallite Size on Hydrogen Solubility in Pd
On decreasing the crystallite size of materials the realm of nanostructured and cluster materials is reached. Nanostructured materials consist of small crystallites (typically of sizes 30 nm or less) joined together by grain boundaries. Their properties are quite different from those of bulk material due to the large fraction of interfaces present. Clusters are agglomerates of a few hundred to a few thousand atoms whereas nanocrystallites consist of many times that number.
Frieske and Wicke (63) reported that H2 isotherms for Pd black shift from the origin to start the normal dilute phase solubility at a higher r value due to chemisorption. Hysteresis was reduced for Pd black although the average plateau pH2 was unchanged. Using Pd blacks of various sizes (determined by electron microscopy to be in the range from 113 to 7 nm), Everett and Sermon (64) found a decrease in hysteresis with decrease of crystallite size.
Experimental studies of H2 absorption by nanocrystalline Pd (n-Pd) have been reported (65–67). When compared to bulk Pd, the notable features of H2 solubility in nanocrystalline Pd are the enhanced solubilities in the a-phase (65–68) and a narrowing of the two-phase field (66) while the average plateau pH2 appears largely unchanged from that of bulk Pd. It was suggested that the enhanced H2 solubility in the dilute phase of n-Pd results from H solution in the grain boundaries, which have a spectrum of interstitial site energies and therefore, like amorphous alloys, cannot form a hydride phase (65, 66).
However, using high-resolution X-ray powder diffraction with synchrotron radiation, Eastman et al . (69) concluded that both the bulk and the grain boundary material in n-Pd converted to the hydride phase. They attributed the narrowing of the two-phase field to a decrease in the ratio of the entropy-of-mixing to the enthalpy-of-mixing of H in n-Pd, thereby shifting the Pd-H critical temperature to a lower value (69). Since suitable values chosen for the thermodynamics of mixing can generate any desired simple miscibility gap phase diagram, agreement with experiment is not meaningful. Further work must be done to resolve this controversy regarding the narrowing of the phase diagram.
A H2 isotherm was measured volumetrically for the first time by Pundt et al . (68) for Pd clusters (3.1 nm-size embedded in two different types of elastically soft matrices – a soft polymer matrix and a surfactant shell) (Figure 13). The dilute phase solubility of H2 in the Pd clusters was greater than in bulk Pd (Figure 13). This was explained based on the available subsurface sites in the clusters having a different energy with respect to interstitial H than bulk Pd sites (68). Zütel et al . (70) utilised ligand-stabilised Pd clusters (produced by a reduction method) mixed with Cu powders as electrodes for the determination of the H capacities; the solubilities were found to decrease with cluster size as expected from the n-Pd behaviour.
The usefulness and convenience of H2 solubilities for investigating lattice defects, including inhomogeneities, in Pd and its alloys has been described. H2 solubilities can show that large dislocation densities form by cycling Pd and Pd/Al2O3 composites (prepared by internal oxidation at 1073 K) through the hydride phase changes. H2 solubilities also show that the dislocation density of Pd is significantly reduced by annealing at 723 K whereas the dislocation density of the Pd/Al2O3 composite is not. Thus H2 solubilities can readily monitor metallurgical phenomena such as dislocation annihilation.
Studying the segregation of solutes to internal interfaces is important because of the wide occurrence of metal/oxide interfaces in technology.
Hydrogen could therefore become a facet helping in the characterisation of industrial small-scale materials.
In addition to simply dissolving in metals and alloys, H can have the role of enhancing metal atom diffusion leading to phase separation, according to a ternary equilibrium, in some Pd-alloys. Upon cooling and removal of the H, the metal atoms remain ‘frozen’ and have been shown to be phase separated by hydrogen isotherms, SANS and HREM methods.
Nanoscience is a rapidly developing field, believed to have some important industrial applications. The study of H2 solubilities in nano-crystalline Pd and its alloys may well be a useful tool for their characterisation. For example, the question of whether or not the smaller plateau in n-Pd indicates that the grain boundary Pd is disordered and cannot form the hydride phase is important for understanding nanometals. Hydrogen absorption by Pd will be better and more fully understood if its study is extended to as many different forms of Pd as possible.
- 39J. Meijering, “Advances in Materials Research”, ed. H. Herman, Vol. 5, Wiley, New York, 1971, p. 1
- 40J. Eastman and M. Ruhle, Ceram. Eng. Sei. Proc, 1989, 10, 1515
- 41X. Y. Huang,, W. Mader and R. Kirchheim, Acta Metall. Mater., 1991, 39, 893
- 42H. Noh,, T. Flanagan,, R. Balasubramaniam and J. Eastman, Ser. Mater., 1996, 34, 665
- 43R. Balasubramaniam,, H. Noh,, T. Flanagan and Y. Sakamoto, Acta Metall. Mater., 1997, 39, 893
- 44J. Mackert,, R. Ringle and C. Fairhurst, J. Dent. Res., 1983, 62, 1229
- 45S. Guruswamy,, S. Park,, J. Hirth and R. Rapp, Oxid Met, 1986, 26, 77
- 46R. Balasubramaniam,, R. Kirchheim,, D. Wang and T. Flanagan, J. Alloys Compi, 1999, 293 - 295, 306
- 47D. Wang,, J. Clewley,, T. Flanagan,, Y. Sakamoto and K. Shanahan J. Alloys Compd., 2000, 298, 261
- 48T. Flanagan,, D. Wang and K. Shanahan, Phys. Chem. Cbem. Phys., 2000, 2, 4976
- 49J. Gegner,, G. Horz and R. Kirchheim, Interface Sci., 1997, 5, 231
- 50D. Wang,, T. Flanagan and R. Balasubramaniam, to be published
- 51T. Flanagan,, R. Balasubramaniam and Y. Sakamoto, to be published
- 52M. Klein and R. Huggins, Acta Metall., 1962, 10, 55
- 53M. Lewis and J. Martin, Acta Metall., 1963, 11, 1207
- 54E. Raub,, H. Beeskow and D. Menzel, Z. Metallkd., 1959, 50, 426
- 55J. Shield and R. Williams, Scr. Met., 1987, 21, 1475
- 56H. Noh,, J. Clewley,, T. Flanagan and A. Craft, J. Alloys Compd., 1996, 240, 235
- 57K. Watanabe,, N. Okuma,, Y. Fukai,, Y. Sakamoto and Y. Hayashi, Scr. Mater., 1996, 34, 551
- 58T. Flanagan and C.-N. Park, J. Alloys Compd., 1999, 293-295, 161
- 59T. Flanagan,, J. Clewley,, H. Noh,, J. Barker and Y. Sakamoto, Acta Metall, 1998, 46, 2173
- 60T. Flanagan,, H. Noh,, J. Clewley and J. Barker, Scr. Met, 1998, 39, 1607
- 61R. Luke,, G. Schmitz,, T. Flanagan and R. Kirchheim, J. Alloys Compd., in press
- 62E. Wicke, and H. Brodowsky,, “Hydrogen in Metals”, eds. G. Alefeld and J. Völkl, Vol. 2, Springer-Verlag, Heidelberg, 1978
- 63H. Frieske and E. Wicke, Ber. Bunsenges Phys. Chem., 1973, 77, 48
- 64D. Everett and P. Sermon, Z. Phys. Chem. NF, 1979, 114, 101
- 65T. Miitschele and R. Kkchheim, Scr. Metall., 1987, 21, 135
- 66T. Miitschele and R. Kirchheim, op. cit. (Ref. 65), p. 101
- 67H. Natter,, B. Wettmann,, B. Heisel and R. Hempelmann J. Alloys Compd., 1997, 253-254, 84
- 68A. Pundt,, C. Sachs,, M. Winter,, M. Reetz,, D. Fritsch and R. Kirchheim, J. Alloys Compi, 1999, 293-295, 480
- 69J. Eastman,, L. Thompson and B. Kestel, Phys. Rev. B, 1993, 48, 84
- 70A. Ziittel,, Ch. Nutzenadel,, G. Schmid,, D. Chartouni and L. Schlapbach, J. Alloys Compd, 1999, 293-295, 472