Hydrogen and Deuterium Solubility in Commercial Pd–Ag Alloys for Hydrogen Purification

Abstract

Pd–Ag alloys with compositions close to 23–25% Ag are considered as a benchmark for hydrogen permeability. They are used in small scale reactors for hydrogen separation and purification. Permeability and solubility are strictly mathematically correlated, and the temperature dependence of solubility can provide useful information about the physical state of the material, the hydrogenation enthalpy, and the occurrence of different thermodynamic states. While the permeability of Pd–Ag alloys has been largely investigated, solubility measurements are available only in a restricted temperature range. In this paper, we extend solubility measurements up to 7 bar for Pd₇₇Ag₂₃ in the temperature range between 25 °C and 400 °C and for Pd₃₀Ag₇₀ for temperatures between 190 °C and 300 °C. The occurrence of solid solutions or hydride phases is discussed, and the hydrogenation enthalpy is calculated.

1. Introduction

It has been reported that nearly 2% of the world’s energy is stored in the hydrogen industrially produced each year. The major part of this hydrogen is used in chemical, petrochemical, metallurgical, pharmaceutical, and textile industries in order to manufacture a large number of products [1]. One of the main problems concerning hydrogen production [2,3,4] is its separation and purification from less desired gases by means of cost-effective processes. There are mainly three types of purification methods: (1) pressure swing adsorption (PSA), (2) fractional/cryogenic distillation, or (3) membrane separation [1]. PSA and distillation are in commercial operation, even though they are expensive and energy demanding. For small-scale applications and in order to achieve an ultra-high pure grade (>99.999%) of the released hydrogen, the membrane separation methods is currently believed to be the best choice, as it offers low energy consumption, possibility for continuous operation, lower investment costs, easy operation, and ultimately cost effectiveness [1].

The cheapest type of membranes for hydrogen separation are the polymeric ones that are already used in industrial processes, such as hydrogen separation in refineries and the petrochemical industry, the production of technical grade nitrogen, and the separation of CO₂ from CH₄ or N₂ [5,6,7]. However, the use of polymeric membranes is limited to temperatures below 100 °C and in applications in which the purity of the permeated gas is not a primary concern, as their selectivity is relatively low (not exceeding 100 [6]). Moreover, they suffer from swelling and poisoning [1]. Another concern about polymer membranes is the need to recover the toxic by-products for their production [7].

In general, when selectivity is important and higher temperatures are needed, the membrane separation techniques are based on permeators [8,9,10] in which hydrogen molecules are catalytically ionized to obtain H⁺ ions on one of the surfaces of metallic membranes. The H⁺ ions enter into the material and diffuse through the membrane towards the other surface, due to the gradient of pressure and concentration. After travelling along the whole membrane, finally H⁺ ions can recombine on the second surface of the membrane to form ultrapure H₂ molecules. In general, a large variety of metallic alloys, both crystalline and amorphous [11,12], show high selectivity for hydrogen and this characteristic makes the ultrapure grade of the hydrogen gas permeated through the metallic membranes obtainable [1]. However, a proper choice of the metal should be made to ensure a large hydrogen diffusion coefficient in the metal [1,8,13]. One metal that displays a large diffusion coefficient is Pd [14]. For this metal, the adsorption and dissociation reactions, and the reverse ones, occurring on the surfaces of the membranes, are very fast. Therefore, the controlling mass transfer mechanism is the diffusion of hydrogen through the metal lattice. Accordingly, the hydrogen permeation flux through a dense metal wall is well described by the equation:

$$J=\frac{Pe}{t}\left(p_{up}^{0.5}-p_{down}^{0.5}\right)$$

(1)

where J is the hydrogen permeation flux (mol·m⁻² s⁻¹), p_up and p_down (Pa) are the hydrogen partial pressures in the upstream and in the downstream side, respectively, t (m) is the membrane thickness, and Pe (mol·m⁻¹·s⁻¹·Pa^−0.5) is the hydrogen permeability. It is well known that the permeability coefficient is given by the product:

$$Pe=D\times S$$

(2)

where D is the diffusion coefficient (m² s⁻¹), and S the solubility coefficient (mol m⁻³ Pa^−0.5). As a consequence of the above considerations, the metals suitable for the hydrogen separation have to exhibit high hydrogen permeability (i.e., an optimum combination of hydrogen solubility and diffusivity) and low surface resistance.

One of the most important drawbacks of Pd is that, when subjected to repeated hydrogenation/dehydrogenation cycles under specific operative conditions, it can suffer from embrittlement due to the mismatch between the lattice parameter of the α, solid solution, and β, hydride, phases. Alloying with other metals has been recognized as a method that mitigates such an inconvenience, and Pd–Ag alloys in particular have been largely investigated. Among Pd–Ag alloys, the best compromise between the mitigation of lattice mismatch, the hydrogen diffusion coefficient, and permeability is obtained around the composition Pd₇₇Ag₂₃ or Pd₇₅Ag₂₅ [15], while the alloy Pd₃₀Ag₇₀ is commercially available for high temperature brazing and is used for joining the Pd–Ag permeator tubes to the stainless steel modules. Furthermore, the alloying with Ag increases the mechanical strength of pure Pd, thus improving the efficiency and reliability of the membrane modules.

Many studies about bulk Pd-membranes are devoted to the investigation of the permeability as a function of temperature [16,17,18,19,20], which is the physical quantity of principal interest for applications. Less attention has been devoted to the solubility of hydrogen in Pd–Ag alloys, which however gives important information about the physical state of the alloys. Therefore, in order to fill this gap, we measured the solubility of hydrogen in two commercially available alloys, Pd₇₇Ag₂₃ and Pd₃₀Ag₇₀. We extended solubility investigation in the pressure (p < 7 bar) and temperature (T < 400 °C) ranges usually exploited for permeation measurements and for applications in permeators made of thin-walled tubes.

2. Materials and Methods

A Pd₇₇Ag₂₃ (wt %) foil with a thickness of 25 μm was purchased from Goodfellow, while a Pd₃₀Ag₇₀ (wt %) foil 150 μm thick was obtained from Johnson Matthey.

Small pieces of the membranes (with typical dimensions of ~20 mm × 7 mm) were cut and used for measurements of pressure–composition curves in a homemade Sieverts apparatus [21,22] working up to a 200 bar pressure and a 500 °C temperature. The amount of hydrogen exchanged between the gas atmosphere and the solid samples was measured by the pressure variation in calibrated cylinders connected by Swagelok tubes to the reaction chamber, where the sample was placed. The data acquisition was performed by means of a homemade computer program developed in the Labview language. Practically, the time evolution of the pressure detected by two micro-Baratron transducers working between 0 and 7 bar and between 0 and 200 bar, respectively, of the temperature of the five cylinders, and of the tubes connecting the cylinders and the reaction chamber was recorded. The real gas state equation was used to calculate the exchanged hydrogen moles. The hydrogenation procedure was extended until a steady state was obtained and no more hydrogen was absorbed by the membranes. A total mass of ~148 mg and ~906 mg were used for Pd₇₇Ag₂₃ and Pd₃₀Ag₇₀, respectively. Experiments were conducted at temperatures between 25 and 400 °C for Pd₇₇Ag₂₃ and between 190 and 300 °C for Pd₃₀Ag₇₀. For both samples, the maximum applied pressure is ~7 bar. In order to perform such gas sorption measurements, hydrogen (purity: ≥99.9999%) and deuterium (purity: ≥99%) were purchased from Società Italiana Acetilene e Derivati (Rome, Italy).

3. Results and Discussion

A surface analysis of the Pd₇₇Ag₂₃ and Pd₃₀Ag₇₀ samples was performed by means of a SEM/EDX technique to investigate the alloy composition and the absence of impurities and defects. Micrographs Figure 1a,b show the Pd₇₇Ag₂₃ sample, while Figure 1c,d refer to the Pd₃₀Ag₇₀ sample. The surface of the Pd₇₇Ag₂₃ sample is rather smooth and regular, with only few irregularities mainly due to the manufacture procedure (i.e., lamination); in the Pd₃₀Ag₇₀ sample, the effects of the lamination process are much more evident. For this reason, SEM analysis was also performed along the cross-section. As shown in Figure 2, SEM analysis revealed that the scratches due to the lamination involve only the surface of the sample and not the bulk of the materials, which appears rather regular and homogeneous.

Regarding the composition and the presence of impurities, in the Pd₇₇Ag₂₃ sample EDX has detected a composition of 78 atom % of Pd and 22 atom % of Ag, while in the Pd₃₀Ag₇₀ sample the measured composition is 31 atom % of Pd and 69 atom % of Ag. No other elements, besides some organic compounds on the surface, have been observed.

The solubility measurements are reported in Figure 3 for both samples, in the form of pressure-composition isotherms. The hydrogen concentration is expressed as H/M, i.e., the number of hydrogen atoms per atom of metal [23]. The error bars on the single measurements are smaller than the symbols. Moreover, reproducibility was checked by duplicate measurements at selected temperatures.

The solubility increases for both samples as temperature decreases. Moreover, at selected temperature and pressure conditions, the hydrogen concentration in Pd₇₇Ag₂₃ is at least one order of magnitude higher than in Pd₃₀Ag₇₀. In Figure 3, one can observe different shapes of the sorption curves as a function of the absorption temperature and Ag content. In fact, Pd₃₀Ag₇₀ shows at all the three temperatures (above 190 °C) a linear dependence of ln(p) from ln(c). This behavior is expected when a single phase is present in the sample. As the hydrogen concentration is extremely low (lower than 10⁻²), the phase should be the α one, i.e., the solid solution of hydrogen in the metal matrix. This fact is in agreement with previous investigation of the Pd–Ag–H phase diagram in which, for concentration higher than 30% of Ag, only the α phase seems to occur [24,25]. In the case of Pd₇₇Ag₂₃ the high temperature curves (T > 280 °C) clearly display a similar linear dependence, witnessing the occurrence of only the α phase. At lower temperatures, especially below 200 °C, two inflections centered around H/M ~0.08 and 0.25 are visible. This behavior suggests the occurrence of a solid solution at low concentrations and the gradual transformation from the α to the β hydride phase at higher concentrations. This findings are in close agreement with the existence of a critical temperature reported between 283 and 295 °C for Pd–H system [26]: above such a temperature only solid solution of H in palladium can be observed, while at lower temperatures, a transition toward the hydride phase has been reported.

To the best of our knowledge, the p-c curves of Pd₃₀Ag₇₀ had never been investigated before, while some previous investigations of Pd₇₇Ag₂₃ or Pd₇₅Ag₂₅ are available in more restricted temperature ranges, mainly below the critical temperature [16,27,28]. The Pd₇₅Ag₂₅ sample investigated by Ross et al. was not commercial and was displayed a lower maximum capacity of hydrogen storage, while the value of the pressure plateau displayed by the same authors up to temperatures below 250 °C seems to be comparable to those here reported [16]. Additionally, the samples studied by Flanagan et al. showed a lower maximum hydrogen capacity [27]. The measurements reported by Anand et al. on Pd₇₇Ag₂₃ extended only up to 150 °C on a sample prepared by arc melting [28].

A general consideration about the kinetics of absorption in the two membranes (results not shown): while the kinetics of Pd₇₇Ag₂₃ is extremely fast (full hydrogen absorption in less than 150 s at 300 °C), the time dependence of the hydrogen solubility in Pd₃₀Ag₇₀ is extremely slow, even at high temperatures (full hydrogen absorption in ~3000 s at 300 °C) (see Figure S1).

From the pressure-composition isotherms, the hydrogenation enthalpy can be calculated by means of the so-called van’t Hoff plot, reporting the logarithm of the equilibrium pressure versus the inverse of the temperature. For materials exhibiting flat pressure plateau centered at p*, the van’t Hoff plot reports ln(p*) vs. 1/T, and the slope of the best fit lines provides the hydrogenation enthalpy, ΔH_hyd. In materials showing no pressure plateau, one can still construct a van’t Hoff plot reporting ln(p) vs. 1/T, where p is the pressure at a fixed value of H/M at the various temperatures. Additionally, in this case, one can estimate the hydrogenation enthalpy, ΔH_hyd, from the slope of the plot [29,30]. In order to construct such a graph, the experimental pressure–composition isotherms were interpolated by means of beta splines, thanks to a program based on the Labview (National Instruments, Austin, TX, USA) routines.

Figure 4 and Figure 5 display the van’t Hoff plot of Pd₇₇Ag₂₃ in the two different zones of the pressure–composition graph: low c and high T (i.e., in the α phase), higher c, low T (i.e., in the coexistence region of the α and the β phase).

The hydrogenation enthalpy values obtained in the two regions are 49 ± 2 kJ/mol for low c and high T and 43 ± 2 kJ/mol for low temperatures and higher concentrations. This values are quite in good agreement with values obtained by calorimetric and volumetric techniques in Pd–H (~38 kJ/mol) [24] and with the value for the α to the β phase transition reported by Anand et al. for Pd₇₇Ag₂₃ below 150 °C (42.2 ± 0.5 kJ/mol) [28].

The van’t Hoff plot for Pd₃₀Ag₇₀ is reported in Figure 6. The best fit provides a value of ΔH_hyd of ~10.8 ± 0.6 kJ/mol, a value much lower than in the case of Pd₇₇Ag₂₃. Indeed, previous investigation of the Pd_100−xAg_xH system by calorimetric measurements indicated that the absolute value of the hydrogenation enthalpy has a maximum value for x of ~40 and decreases on the further increase of the silver content in the alloys up to an x value of ~50. No additional information about the hydrogenation enthalpy of the Pd_100−xAg_xH system are available for higher x values [27].

Finally, we investigated the isotopic effect between hydrogenation and deuteration for Pd₇₇Ag₂₃ at 94 °C, as no information is available for such a content of Ag in the Pd–Ag membranes. Due to the limitations of our inlet line for deuterium, we extended our measurements only up to 2 bar, so we chose a low temperature in order to study the whole absorption curve. The two absorption isotherms measured for hydrogenation and deuteration are reported in Figure 7. Above a c value of ~0.3 H/M, the two curves coincide, while for lower hydrogen concentrations, the D absorption curves is displaced towards higher pressures by almost one order of magnitude. It is well known, indeed, that palladium has an unusual isotopic effect: in normal compounds, the curves corresponding to deuteration are displaced toward lower pressure, while in the case of palladium, the opposite effect occurs. Similar behavior, with a higher pressure plateau for deuterated samples, has already been reported by Anand et al. for Pd₇₅Ag₂₅ [28] and by Hara et al. for Pd₉₃Ag₇ [31].

4. Conclusions

In this paper, an extensive study of the solubility of hydrogen in two commercial Pd–Ag alloys up to p = 7 bar is reported. For Pd₇₇Ag₂₃ (Pd₃₀Ag₇₀), the investigated temperature range extends from 25 °C up to 400 °C (190 °C–300 °C), well above the temperature range investigated in the previous literature for compositions close to Pd₇₇Ag₂₃. Below ~280 °C, Pd₇₇Ag₂₃ displays the existence of the β phase, with a hydrogenation enthalpy of 43 ± 2 kJ/mol. Above such a temperature, only a solid solution phase is observed, and the calculated ΔH_hyd is ~49 ± 2 kJ/mol. The study was extended to a much higher Ag concentration in the Pd–Ag alloy, i.e., to Pd₃₀Ag₇₀, a composition never investigated before. Pd₃₀Ag₇₀ shows the existence of an α phase in the whole temperature range, with a lower hydrogenation enthalpy (~10.8 ± 0.6 kJ/mol), and a maximum solubility one order of magnitude lower than that of Pd₇₇Ag₂₃. Moreover, the hydrogenation kinetics of Pd₇₇Ag₂₃ is slower than that of Pd₃₀Ag₇₀ by one order of magnitude.