- Email: [email protected]
Neutron diffraction study of the Pd0.772Ag0.228Dν membrane for hydrogen separation
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Neutron diffraction study of the Pd0.772Ag0.228Dn membrane for hydrogen separation M. Catti a,*, O. Fabelo b, A. Filabozzi c, A. Pietropaolo d, S. Tosti d, A. Pozio d, A. Santucci d di Milano Bicocca, via Cozzi 55, 20125, Milano, Italy Dipartimento di Scienza dei Materiali, Universita Institut Laue-Langevin, 71 avenue des Martyrs, 38000, Grenoble, France c di Roma Tor Vergata, via della Ricerca Scientifica 1, 00133, Roma, Italy Dipartimento di Fisica, Universita a
b
d
ENEA, Dipartimento Fusione e Tecnologie per la Sicurezza Nucleare, via Fermi 45, 00044, Frascati, Italy
article info
abstract
Article history:
Experiments of deuterium absorption/desorption were performed on a foil of Pd77Ag23 (wt
Received 11 October 2016
%) alloy in the 78e196 C temperature and 1e4 bar pressure ranges under the neutron beam
Received in revised form 20 January 2017
at ILL (Grenoble). Powder diffraction patterns were collected on the D1B diffractometer (l ¼ 1.2871 A) on the Pd0.772Ag0.228Dn deuterated alloy, and the face-centred-cubic structure
Accepted 23 January 2017
was Rietveld-refined locating the D atom in the octahedral site with a variable occupancy n
Available online 8 February 2017
(¼D/M ratio). The results of n(T) curves at different p(D2) pressures are discussed in comparison with literature data on H and D absorption into PdeAg alloys from thermodynamic
Keywords:
measurements. A negative deviation from predictions of Sieverts' law is shown by the
Deuterium absorption
dependence of D occupancy on pressure. This effect relates the behaviour of the activity
Palladiumesilver alloy
coefficient gD to DeD and DeM interactions in the solid solution phase.
Sieverts' law
© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Exploiting hydrogen as energy carrier for renewable energy sources is expected to reduce the emissions of greenhouse gases in the atmosphere [1e3]. However, a hydrogen-based economy requires that a well-proven technological platform for the safe and reliable hydrogen production, separation, storage and transportation is developed [4]. In particular, hydrogen production and separation could take advantage of the membrane processes that arose an increasing interest in recent years, because of their reduced energy consumption and continuous operation mode [5e9]. Further, these processes can be easily scaled up and integrated with other separation technologies.
Thanks to the unique property of hydrogen to pass through dense metal walls, metal membranes were introduced for separating hydrogen from gas mixtures selectively. The PdeAg alloy with silver content around 25 wt.% exhibits the best hydrogen permeability and mechanical strength, with heavily reduced embrittlement effects [10]. Therefore, PdeAg alloys with this composition are now commercially available for the manufacturing of membrane devices used for both hydrogen separation (permeators/diffusers) and hydrogen production (membrane reactors) [11,12]. The further optimization of the Pdmembrane devices requires a deep understanding of the chemicalephysical properties of the hydrogenated Pd-alloys in the range of the operating conditions typical of the membrane processes (temperature up to 400e450 C and pressure up to 10e20 bar).
* Corresponding author. E-mail address: [email protected] (M. Catti). http://dx.doi.org/10.1016/j.ijhydene.2017.01.130 0360-3199/© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
6788
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
Recently, hydrogen permeators were ohmically heated by applying an electric current directly through thin-walled Pde Ag tubes [13]. In order to properly design these membrane modules, the strain and electrical resistance of hydrogenated PdeAg tubes were measured from 100 to 400 C [14], while below 100 C some anomalies of the electrical resistivity were observed and not reported in that study. In subsequent experiments, the electric properties of a thin PdeAg strip were studied in detail, finding a S-shaped curve of the resistivity vs. both temperature and hydrogen pressure [15]. This peculiar behaviour was tentatively explained in terms of an interaction of the conduction electrons with H atoms in the PdeAg structure. Although structural properties of pure hydrogenated Pd were studied extensively [16e18], those of PdeAg alloys require further investigations, in particular as a function of hydrogen uploading. A study was thus undertaken with the purpose of determining (i) the location of D/H atoms absorbed inside the FCC structure of the Pd77Ag23 alloy, (ii) the amount of D/H dissolved in the alloy structure at different temperature and pressure values, and (iii) the change of volume accompanying the gas uptake in the solid solution. In order to have direct access to atomic-scale structural information diffraction methods are the best choice. However, H atoms scatter X-rays very weakly, while with neutrons they produce a huge incoherent background limiting severely the accuracy of measurement of diffraction intensity. The last case does not apply if deuterium is used, as this behaves as a strong coherent scattering centre for neutrons. For this reason, we chose deuterium rather than hydrogen gas for in-situ absorption studies by the Pd77Ag23 alloy under thermal neutron beam irradiation. Indeed, the technique of D/H isotopic substitution coupled with neutron diffraction proved quite successful for investigating the structural properties of several hydrogen storage materials [19e21]. A convenient range of temperature and p(D2) pressure conditions was explored in the course of these absorption experiments.
Experimental The sample used was a 0.05 mm thick (70 135 mm) foil of Pd77Ag23 (wt%) alloy purchased from Goodfellow Cambridge Ltd. Neutron diffraction patterns were collected on the D1B medium resolution-high-flux diffractometer at the Institut Laue-Langevin (Grenoble, France), in the 1e128 2q range with l ¼ 1.2871 A; in most cases the measurement time was 1 h. The metal foil was rolled up inside an Al can provided with a heating system and connected to a gas manifold, which allowed the holder to be evacuated and filled with D2 gas at variable pressure. A preliminary diffraction pattern was collected on the foil within a standard vanadium holder, and the Rietveld analysis showed that preferential orientation had negligible effects on the sample. A pattern was recorded also on the empty Al pressure cell, and it was analysed by profile matching method with constant relative intensities. This fit was later included in all main refinements of the deuterated alloy as a second phase, where the only free parameters were the scale factor and the cell parameter. For pure Pd77Ag23 the cubic lattice constant of 3.9274(1) A was obtained at room
temperature, slightly lower than the X-ray value of 3.929(1) A previously reported [22]. A critical experimental aspect was the temperature control. Due to the heavy mass of the pressure cell, the regulation thermocouple (close to the heating element within the cryostat) measured a slightly higher value (T1), and the thermocouple located in the sample stick a slightly lower value (T2) with respect to that expected on the sample. The average (T1 þ T2)/2 was thus assumed as reasonable estimate of the sample temperature. Three cooling cycles were performed at different constant pressures, namely p(D2) ¼ 1, 2, and 4 bar. Each one consisted of six data collections at T ¼ 196, 157, 137, 117, 98, 78 C, for a total of 18 diffraction patterns. After the first and second cycles the sample holder was reheated to 196 C, the pressure increased, and the next cooling cycle started. At the end of the 4 bar run the holder was pump-evacuated at 78 C, D2 was pumped in again to 1 bar, and a final heating cycle was carried out at that pressure, collecting data at T ¼ 78, 98, 117, 137, 157, 196 C. Between each temperature change at constant pressure, a waiting time of 0.5 h was applied to thermalize the sample before starting data collection.
Results and discussion Analysis of data and structure refinements All collected patterns were analysed by the FullProf suite [23], including two phases in the profile refinement: the facecentred-cubic (FCC) Pd0.772Ag0.228Dn deuterated alloy (Rietveld refinement), and FCC aluminium (profile matching with constant relative intensities) to include the sample environment reflections. The background was linearly interpolated by a set of fixed points. Bragg peaks were modelled by a pseudoVoigt function (linear combination of Gaussian and Lorentzian components, with s and g half-widths, respectively). The s and g parameters vary with q as s ¼ (U tan2q þ V tanq þ W þ P/cos2q)1/2 and g ¼ X tanq þ Y/cosq; the U, V and W coefficients are mainly related to instrumental resolution, P and Y are the Scherrer coefficients for Gaussian and Lorentzian particle size broadening, respectively, and X is related to particle strain broadening. The mixing coefficient and the full width of the pseudo-Voigt function depend on s and g according to equations given in the literature [24]. For Al, only the unit-cell constant and the scale factor were refined in each pattern, keeping fixed the other Al parameters from the calibration results of the empty cell. For the main Fm3m phase (Pd0.772Ag0.228Dn) the scale factor, the unit-cell edge, the displacement factors of Pd0.772Ag0.228 and of D, and the U, V, W, Y profile parameters were always refined. Slight changes of U, V and W indicated a weak dependence of these parameters on experimental conditions of the sample, in addition to instrumental resolution. A Fourier difference map was calculated before inserting D atoms, and it showed only a strong peak at x ¼ ½, y ¼ 0, z ¼ 0 (octahedral site). After including the D atom in that position in the refinement, with a variable occupation factor and displacement factor, a subsequent Fourier difference map showed two weak residual peaks at 0.26,0,0 and ¼,¼,¼
6789
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
coordinates. As these small signals were not detected in the data collected at ambient conditions before gas uploading, they should be related to D2 absorption. However, they could not be refined successfully as additional D atoms. Therefore, only the D atom located on the octahedral site was kept in the final model. Yet it cannot be excluded that a minor quantity of deuterium is present also in the tetrahedral site and/or in disordered locations around the octahedral site. A better quality of data at improved resolution is necessary to solve this problem. The final results of Rietveld refinements are reported in Table 1, including the agreement factors of profile fitting (Rp not weighted, wRp weighted) and of Bragg peak intensities (RB). The sample lattice constant is seen to decrease on raising temperature, and to increase with pressure (the peculiar behaviour of the two cycles at 1 bar will be discussed below), in agreement with the reported hydrogen solubility in the Pde Ag alloy [25]. One of the collected patterns is shown in Fig. 1 as an example. Because of the similar values of the cubic lattice constants of the two phases, a significant overlap is observed for some Bragg peaks of the sample with the corresponding ones of the aluminium cell. This is the case particularly in the low T range (below 140 C), because the thermal expansion coefficient of the Pd77Ag23 alloy turns out to be much larger than that of aluminium, so that the overlap increases on cooling (cf. Fig. 2). However, the Rietveld refinement of the alloy, together with the profile matching with constant relative intensities of Al from the pressure cell, seems to deal with the problem satisfactorily, as it appears from the quite reasonable estimated standard deviations (e.s.d.) of the refined parameters.
of Fig. 1 e Neutron diffraction pattern (l ¼ 1.2871 A) Pd0.772Ag0.228Dn recorded at 196 C and 1 bar (cooling cycle). Measured, calculated and difference intensity profiles are shown. Upper and lower ticks indicate Bragg peak positions of the deuterated alloy and of Al of the cell, respectively.
Thermal and pressure behaviour As the FCC structure contains as many octahedral voids as metal atoms, the refined D occupation factor is numerically equal to n ¼ D/M, i.e. the deuterium-to-metal ratio in the Pd0.772Ag0.228Dn formula. This value is plotted in Fig. 3 as a function of T for all four thermal cycles at the different applied pressures.
Table 1 e Results of the Rietveld refinements of Pd0.772Ag0.228Dn. Agreement factors, sample and Al cubic lattice constants, D occupation factor (¼n), and displacement factors of Pd0.772Ag0.228 and D atoms are reported, with e.s.d.'s in parentheses. T ( C)
p (bar)
Rp (%)
wRp (%)
RB (%)
a ( A)
a(Al) ( A)
o.f.(D)
B(PdAg) ( A2)
B(D) ( A2)
196
1 2 4 1 1 2 4 1 1 2 4 1 1 2 4 1 1 2 4 1 1 2 4 1
8.11 7.78 7.38 7.33 7.28 6.69 6.73 7.16 6.44 6.63 6.72 6.85 6.38 6.64 6.54 6.75 6.48 6.86 6.67 6.70 6.72 7.27 6.93 7.05
11.6 11.0 10.9 10.8 10.4 9.87 10.0 10.3 9.63 9.67 9.83 9.89 9.34 9.81 9.52 9.91 9.97 10.0 9.63 9.67 10.2 10.3 10.0 10.2
5.47 4.26 4.98 3.72 4.58 4.46 4.74 4.82 6.22 3.99 5.12 5.08 5.80 4.57 5.04 4.87 4.54 5.09 5.67 5.35 4.69 6.16 6.12 5.91
3.95720(9) 3.99616(8) 4.01112(8) 3.98376(8) 3.96685(8) 4.00931(8) 4.01844(8) 4.00601(8) 3.97716(9) 4.01506(8) 4.02110(8) 4.01236(8) 3.98668(9) 4.01812(8) 4.02318(8) 4.01730(8) 3.9934(1) 4.02012(9) 4.02483(9) 4.02114(8) 3.9958(1) 4.02152(9) 4.02606(9) 4.02276(9)
4.06192(6) 4.06211(6) 4.06227(6) 4.06179(5) 4.05965(5) 4.05866(5) 4.05867(6) 4.05826(5) 4.05663(5) 4.05680(6) 4.05680(6) 4.05647(6) 4.05497(5) 4.05502(6) 4.05498(6) 4.05465(6) 4.05335(5) 4.05322(6) 4.05319(6) 4.05290(6) 4.05151(6) 4.05145(6) 4.05147(6) 4.05129(6)
0.087(9) 0.210(10) 0.272(11) 0.170(10) 0.118(8) 0.281(10) 0.318(10) 0.258(10) 0.162(9) 0.314(10) 0.341(10) 0.300(10) 0.207(9) 0.333(10) 0.354(10) 0.329(10) 0.242(9) 0.344(10) 0.372(10) 0.354(10) 0.263(10) 0.358(11) 0.383(10) 0.368(10)
1.10(4) 1.04(4) 1.06(4) 1.11(4) 0.98(4) 1.03(4) 1.13(4) 1.02(4) 1.10(4) 1.07(4) 1.10(4) 1.04(4) 0.93(4) 1.05(4) 1.08(4) 1.01(4) 0.90(4) 1.02(4) 1.04(4) 0.99(4) 0.87(4) 0.98(4) 0.99(4) 0.95(4)
2.9(7) 4.5(4) 5.0(4) 4.2(5) 3.4(5) 4.7(3) 5.1(3) 4.6(3) 4.0(4) 5.0(3) 5.1(3) 4.8(3) 3.9(3) 4.8(3) 4.9(2) 4.8(3) 4.0(3) 4.6(2) 4.8(2) 4.7(2) 4.0(3) 4.5(2) 4.6(2) 4.5(2)
157
137
117
98
78
6790
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
Fig. 2 e Magnified details (2q range 30e40 ) of neutron diffraction patterns recorded at 196 and 78 C (1 bar, cooling cycle), showing the overlap of (111) and (200) peaks of the deuterated alloy with those of Al.
conditions is thus expected to lie possibly below the heating one for the hysteresis phenomenon, but well above the cooling curve reported in Fig. 3. Indeed, this one represents nonequilibrium results of phenomenological interest to characterize absorption processes on samples not previously treated at higher pressure. The data can be fitted by a cubic curve, showing an inflexion point at about 150 C. This behaviour looks similar to that of previous results [15] on thermodynamic and electric measurements of the effects of H2 absorption by the Pd77Ag23 alloy, but it is not shown by the heating curve with full deuterium uploading. No inflexion point appears either in the fit of data at higher pressures (2 and 4 bar). Thus, further measurements extending the thermal range towards higher temperatures would be needed to settle this issue. The cooling curve at 4 bar in Fig. 3 should reflect results approximately close to equilibrium, as the sample was treated at the same high pressure as for the heating curve at 1 bar. The process of achieving the stable expanded structure configuration may have not been fully completed at 2 bar, as the D absorption amount at 78 C is slightly lower than that recorded at 1 bar after the 4 bar pre-treatment. On the whole, the effect of pressure on deuterium absorption in the 1e4 bar range seems to be scarce at the low T end (only 4% increase of D occupancy), but it increases progressively with temperature. Thus, pressure acts strongly as driving force of D insertion into the alloy at low deuterium concentration, yet it weakens substantially when the D occupancy is high. However, at the high T end (196 C) the D content is raised by 60% from 1 to 4 bar and by 30% from 2 to 4 bar (Table 1). This is less than what required by Sieverts' law: cD ¼ Ks p1=2 ;
Fig. 3 e Deuterium occupancy n in Pd0.772Ag0.228Dn from Rietveld refinements of diffraction data collected at different p(D2) pressures. Arrows indicate cooling or heating direction of thermal cycles.
As most striking feature of the results, a large difference is observed between the cooling and heating n (T) curves at p(D2) ¼ 1 bar. Thermal hysteresis gives only a minor contribution to this effect in the high temperature range. The main reason is that measurements on heating were performed after cycles at higher pressure (2 and 4 bar), which made the sample able to absorb a significantly larger quantity of deuterium, with corresponding expansion of the lattice constant. We thus believe that the heating curve is close to represent equilibrium absorption results at 1 bar, except for presumable hysteresis effects which may overestimate the deuterium solubility on increasing temperature. A cooling curve obtained in the same
(1)
where Ks is Sieverts constant, p is the deuterium partial pressure in the gas phase, and the deuterium concentration cD in the solid alloy is obviously proportional to n ¼ D/M. This deviation should be related to non-ideality effects of the solid solution, which are taken into account on replacing the deuterium concentration cD by its product with the activity coefficient gD in Formula (1): cDgD ¼ Ks p1/2 is the generalization of Sieverts' law for non-ideal solutions, where gD depends on cD and p, and Ks only on T. On comparing concentrations determined at different pressures p1 and p2, we thus obtain that gD(p2)/gD(p1) ¼ (p2/p1)1/2cD(p1)/cD(p2). If p1 ¼ 1 bar is assumed, by using the data of Table 1 at T ¼ 196 C (heating cycle for 1 bar) the ratio gD(p2)/gD(p1) takes the values 1.145 and 1.250 for p2 ¼ 2 and 4 bar, respectively, whereas it should always be 1 in case of validity of Sieverts' law (gD ¼ 1 in ideal solutions). This increase of gD with pressure means that the concentration cD has increased with pressure less than predicted by the ideal Sieverts' law (negative deviation): then the energy raising due to DeD repulsion must be larger than the energy lowering due to the DePd(Ag) interaction [26,27]. It is interesting also to consider the relationship between the deuterium content n and the lattice constant a of Pd0.772Ag0.228Dn for the 24 experiments performed at different T and p(D2) values (Fig. 4). At constant pressure a increases linearly with n for all four thermal cycles, according to a ¼ k1n þ k2. The linear coefficients are k1 ¼ 0.22(1), 0.174(5),
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
Fig. 4 e Cubic lattice constant of Pd0.772Ag0.228Dn, measured at different temperatures, plotted vs. the deuterium occupancy n. 0.135(6) A, and k2 ¼ 3.940(2), 3.960(1), and 3.975(2) A for the cooling cycles at 1, 2 and 4 bar, respectively, and k1 ¼ 0.20(1), A for the heating cycle at 1 bar. As the linear k2 ¼ 3.953(4) thermal expansion coefficient of the Pd77Ag23 alloy is 1.36 105 C1, the bare temperature effect appears to be negligible with respect to deuterium absorption/desorption in determining the lattice constant change (k1 coefficient). The shift of line intercept k2 between the two lines at 1 bar confirms clearly that the cubic structure has been expanded by effect of the pre-treatment at 4 bar. The partial molar volume Vm,D of D dissolved in the Pd77Ag23 alloy can be calculated from the standard thermodynamic relationship Vm;D ¼ Vm þ ð1 cD Þ
vVm ; vcD
(2)
where cD ¼ n/(1 þ n) is the deuterium mole fraction, and the alloy molar volume Vm is related to the lattice constant a and thus to n according to the above linear dependence. By using the empirical results previously reported for the k1 coefficient, are obtained in the range values of Vm,D 1.10e1.16 105 m3 mol1 (slightly but significantly larger than the molar volume 0.91 105 m3 mol1 of the pure alloy). The amount of deuterium absorbed by the Pd77Ag23 alloy at room pressure, as determined by neutron diffraction methods in this study, differs substantially according to whether the sample was previously treated thermally at 4 bar or not. It is thus interesting to compare such results with some of those reported in the literature, at corresponding temperatures, for thermodynamic measurements of hydrogen [15,28e31] and deuterium [22,30,31] absorption on alloys of similar compositions. Data in the thermal range corresponding to ours are shown in Fig. 5. The spread between thermodynamic results [15] and [29] for the H case appears to be quite large at low T, but it becomes small at high T; our results obtained for D in conditions close to equilibrium lie approximately in-between, and reasonably close to previous results on deuterium absorption [22]. In the latter study data (not shown in Fig. 5 for clarity) are reported also for hydrogen, so that it is possible to
6791
Fig. 5 e n value in Pd1¡yAgyXn systems vs. T by diffraction (this work) and thermodynamic (literature data) methods. Open symbols: X ¼ D; full symbols: X ¼ H. estimate the D/H isotopic effect on the solubility in the Pd77Ag23 alloy: a gap of 6%e26% is observed, in the 100e150 C thermal range, between the absorption amounts of deuterium and hydrogen. Thus, by applying a similar isotopic shift to our activated D data the corresponding curve for H absorption would be slightly shifted above. On the other hand, the curve obtained in this study at 1 bar on the sample before treatment at higher pressure lies significantly below all other data, confirming that such results were obtained in nonequilibrium conditions.
Conclusions The results of this study confirm that deuterium absorption in the Pd77Ag23 alloy gives rise to a non-stoichiometric Pd0.772Ag0.228Dn compound, where D atoms are located in the octahedral voids of the FCC metal structure. A small additional occupancy of tetrahedral sites cannot be excluded, and will be the object of further investigations with neutron diffraction at higher resolution. The D/M ratio n, obtained as deuterium occupancy from structure refinements, approaches its equilibrium value after thermal treatment of the sample at pressures higher than 1 bar. In these conditions our absorption results lie within the spread of those obtained from thermodynamic measurements on H uploading by the same alloy. In absence of high pressure pre-treatment, a significantly smaller D/M ratio is observed in the absorption experiment. Also these phenomena need to be investigated in our planned neutron studies focused on the kinetic aspects of deuterium absorption into the Pd77Ag23 alloy. On increasing pressure, the amount of uploaded deuterium into the sample deviates negatively from Sieverts' law: the D concentration in the solid phase increases with p(D2) less than predicted by the thermodynamics of ideal solutions. Such effects can be explained by the repulsion between adjacent D atoms exceeding metal-deuterium attraction in the alloy structure: the force resulting from unbalance of these two
6792
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
effects is responsible for the observed non-ideal solution behaviour.
Acknowledgements We thank Francesco Grazzi (CNR-ISC, Sesto Fiorentino, Italy) for useful discussions.
references
[1] Biccikovci O, Straka P. Production of hydrogen from renewable resources and its effectiveness. Int J Hydrogen Energy 2012;37:11563e78. [2] Orhan MF, Dincer I, Rosen MA, Kanoglu M. Integrated hydrogen production options based on renewable and nuclear energy sources. Renew Sustain Energy Rev 2012;16:6059e82. [3] Mansilla C, Louyrette J, Albou S, Bourasseau C, Dautremont S. Economic competitiveness of off-peak hydrogen production today e a European comparison. Energy 2013;55:996e1001. [4] Mansilla C, Avril S, Imbach J, Le Duigou A. CO2-free hydrogen as a substitute to fossil fuels: what are the targets? Prospective assessment of the hydrogen market attractiveness. Int J Hydrogen Energy 2012;37:9451e8. [5] Shu J, Grandjean BPA, Van Neste A, Kalaguine S. Catalytic palladium-based membrane reactors: a review. Can J Chem Eng 1991;69:1036e60. [6] Drioli E, Fontananova E. Membrane technology and sustainable growth. Chem Engin Res Des 2004;82(A12):1557e62. [7] Mulder M. Basic principles of membrane technology. 2nd ed. Dordrecht: Kluwer; 1991. [8] Kikuchi E. Membrane reactor application to hydrogen production. Catal Today 2000;56:97e101. [9] Dixon AG. Recent research in catalytic inorganic membrane reactors. Int J Chem React Eng 2003;1(R6). http://www. bepress.com/ijcre/vol1/R6. [10] Tosti S. Pd-based membranes and membrane reactors for hydrogen production. In: Hilal H, Ismail AF, Wright CJ, editors. Membrane fabrication. CRC Press, Taylor & Francis Group; 2015, ISBN 978-1-4822-1045-3. Ch. 13. [11] Cornaglia L, Munera J, Lombardo E. Recent advances in catalysts, palladium alloys and high temperature WGS membrane reactors. A review. Int J Hydrogen Energy 2015;40:3423e37. [12] Alique D, Imperatore M, Sanz R, Calles JA, Giacinti Baschetti M. Hydrogen permeation in composite Pdmembranes prepared by conventional electroless plating and electroless pore-plating alternatives over ceramic and metallic supports. Int J Hydrogen Energy 2016;41:19430e8. [13] Tosti S, Ghirelli N, Borgognoni F, Trabuc P, Santucci A, Liger K, Marini F. Membrane reactor for the treatment of gases containing tritium. European Patent Grant 2014; EP 2582618.
[14] Tosti S, Borgognoni F, Santucci A. Electrical resistivity, strain and permeability of Pd-Ag membrane tubes. Int J Hydrogen Energy 2010;35:7796e802. [15] Pozio A, Jovanovic Z, Lo Presti R, De Francesco M, Tosti S. PdAg hydrogen content and electrical resistivity: temperature and pressure effect. Int J Hydrogen Energy 2012;37:7925e33. [16] Flanagan TB, Oates WA. The palladium-hydrogen system. Annu Rev Mater Sci 1991;21:269e304. [17] Pitt MP, Gray EM. Tetrahedral occupancy in the Pd-D system observed by in situ neutron powder diffraction. Europhys Lett 2003;64:344e50. [18] Widenmeyer M, Niewa R, Hansen TC, Kohlmann H. In situ neutron diffraction as a probe on formation and decomposition of nitrides and hydrides: a case study. Z Anorg Allg Chem 2013;639:285e95. [19] Busch G, Schlapbach L, Stucki F, Fischer P, Andresen AF. Hydrogen storage in FeTi: surface segregation and its catalytic effect on hydrogenation and structural studies by means of neutron diffraction. Int J Hydrogen Energy 1979;4:29e39. [20] Paul-Boncour V, Filipek SM, Crivello J-C, Couturas F, Morawski A. Properties of ZrNi5 deuteride synthesized under high pressure studied by neutron diffraction and first principles calculations. Int J Hydrogen Energy 2016;41:17408e20. [21] Webb TA, Webb CJ, Dahle AK, Gray EM. In-situ neutron powder diffraction study of Mg-Zn alloys during hydrogen cycling. Int J Hydrogen Energy 2015;40:8106e9. [22] Anand NS, Pati S, Jat RA, Parida SC, Mukerjee SK. Thermodynamics and kinetics of hydrogen/deuterium absorption-desorption in Pd0.77Ag0.23 alloy. Int J Hydrogen Energy 2015;40:444e50. [23] Rodriguez-Carvajal J. FULLPROF: a program for Rietveld refinement and pattern matching analysis. http://www.ill. eu/sites/fullprof/. [24] Thompson P, Cox DE, Hastings JB. Rietveld refinement of Debye-Scherrer synchrotron X-ray data from Al2O3. J Appl Crystallogr 1987;20:79e83. [25] Hughes DT, Harris IR. A comparative study of hydrogen permeabilities and solubilities in some palladium solid solution alloys. J Less Common Met 1978;61:9e21. [26] Nakamura K. Analysis of thermodynamic properties of the hydrogen solution in amorphous Fe-Ti compounds. Ber Bunsenges Phys Chem 1985;89:191e7. [27] Evans MJB. Surface area effects on the sorption of hydrogen by palladium. Can J Chem 1974;52:1200e5. [28] Flanagan TB, Wang D, Luo S. Thermodynamics of H in disordered Pd-Ag alloys from calorimetric and equilibrium pressure-composition-temperature measurements. J Phys Chem B 2007;111:10723e35. [29] Serra E, Kemali M, Perujo A, Ross DK. Hydrogen and deuterium in Pd-25 pct Ag alloy: permeation, diffusion, solubilization, and surface reaction. Met Mater Trans A 1998;29A:1023e8. € sser R, Powell GL. Solubility of protium, deuterium and [30] La tritium in palladium-silver alloys at low hydrogen concentrations. J Less Common Met 1987;130:387e94. [31] Luo W, Cowgill DF, Stewart K. Absorption isotherms for H2(D2)-Pd0.8Ag0.2 (198-323 K). J Alloys Compd 2010;489:47e50.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Neutron diffraction study of the Pd0.772Ag0.228Dn membrane for hydrogen separation M. Catti a,*, O. Fabelo b, A. Filabozzi c, A. Pietropaolo d, S. Tosti d, A. Pozio d, A. Santucci d di Milano Bicocca, via Cozzi 55, 20125, Milano, Italy Dipartimento di Scienza dei Materiali, Universita Institut Laue-Langevin, 71 avenue des Martyrs, 38000, Grenoble, France c di Roma Tor Vergata, via della Ricerca Scientifica 1, 00133, Roma, Italy Dipartimento di Fisica, Universita a
b
d
ENEA, Dipartimento Fusione e Tecnologie per la Sicurezza Nucleare, via Fermi 45, 00044, Frascati, Italy
article info
abstract
Article history:
Experiments of deuterium absorption/desorption were performed on a foil of Pd77Ag23 (wt
Received 11 October 2016
%) alloy in the 78e196 C temperature and 1e4 bar pressure ranges under the neutron beam
Received in revised form 20 January 2017
at ILL (Grenoble). Powder diffraction patterns were collected on the D1B diffractometer (l ¼ 1.2871 A) on the Pd0.772Ag0.228Dn deuterated alloy, and the face-centred-cubic structure
Accepted 23 January 2017
was Rietveld-refined locating the D atom in the octahedral site with a variable occupancy n
Available online 8 February 2017
(¼D/M ratio). The results of n(T) curves at different p(D2) pressures are discussed in comparison with literature data on H and D absorption into PdeAg alloys from thermodynamic
Keywords:
measurements. A negative deviation from predictions of Sieverts' law is shown by the
Deuterium absorption
dependence of D occupancy on pressure. This effect relates the behaviour of the activity
Palladiumesilver alloy
coefficient gD to DeD and DeM interactions in the solid solution phase.
Sieverts' law
© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Exploiting hydrogen as energy carrier for renewable energy sources is expected to reduce the emissions of greenhouse gases in the atmosphere [1e3]. However, a hydrogen-based economy requires that a well-proven technological platform for the safe and reliable hydrogen production, separation, storage and transportation is developed [4]. In particular, hydrogen production and separation could take advantage of the membrane processes that arose an increasing interest in recent years, because of their reduced energy consumption and continuous operation mode [5e9]. Further, these processes can be easily scaled up and integrated with other separation technologies.
Thanks to the unique property of hydrogen to pass through dense metal walls, metal membranes were introduced for separating hydrogen from gas mixtures selectively. The PdeAg alloy with silver content around 25 wt.% exhibits the best hydrogen permeability and mechanical strength, with heavily reduced embrittlement effects [10]. Therefore, PdeAg alloys with this composition are now commercially available for the manufacturing of membrane devices used for both hydrogen separation (permeators/diffusers) and hydrogen production (membrane reactors) [11,12]. The further optimization of the Pdmembrane devices requires a deep understanding of the chemicalephysical properties of the hydrogenated Pd-alloys in the range of the operating conditions typical of the membrane processes (temperature up to 400e450 C and pressure up to 10e20 bar).
* Corresponding author. E-mail address: [email protected] (M. Catti). http://dx.doi.org/10.1016/j.ijhydene.2017.01.130 0360-3199/© 2017 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
6788
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
Recently, hydrogen permeators were ohmically heated by applying an electric current directly through thin-walled Pde Ag tubes [13]. In order to properly design these membrane modules, the strain and electrical resistance of hydrogenated PdeAg tubes were measured from 100 to 400 C [14], while below 100 C some anomalies of the electrical resistivity were observed and not reported in that study. In subsequent experiments, the electric properties of a thin PdeAg strip were studied in detail, finding a S-shaped curve of the resistivity vs. both temperature and hydrogen pressure [15]. This peculiar behaviour was tentatively explained in terms of an interaction of the conduction electrons with H atoms in the PdeAg structure. Although structural properties of pure hydrogenated Pd were studied extensively [16e18], those of PdeAg alloys require further investigations, in particular as a function of hydrogen uploading. A study was thus undertaken with the purpose of determining (i) the location of D/H atoms absorbed inside the FCC structure of the Pd77Ag23 alloy, (ii) the amount of D/H dissolved in the alloy structure at different temperature and pressure values, and (iii) the change of volume accompanying the gas uptake in the solid solution. In order to have direct access to atomic-scale structural information diffraction methods are the best choice. However, H atoms scatter X-rays very weakly, while with neutrons they produce a huge incoherent background limiting severely the accuracy of measurement of diffraction intensity. The last case does not apply if deuterium is used, as this behaves as a strong coherent scattering centre for neutrons. For this reason, we chose deuterium rather than hydrogen gas for in-situ absorption studies by the Pd77Ag23 alloy under thermal neutron beam irradiation. Indeed, the technique of D/H isotopic substitution coupled with neutron diffraction proved quite successful for investigating the structural properties of several hydrogen storage materials [19e21]. A convenient range of temperature and p(D2) pressure conditions was explored in the course of these absorption experiments.
Experimental The sample used was a 0.05 mm thick (70 135 mm) foil of Pd77Ag23 (wt%) alloy purchased from Goodfellow Cambridge Ltd. Neutron diffraction patterns were collected on the D1B medium resolution-high-flux diffractometer at the Institut Laue-Langevin (Grenoble, France), in the 1e128 2q range with l ¼ 1.2871 A; in most cases the measurement time was 1 h. The metal foil was rolled up inside an Al can provided with a heating system and connected to a gas manifold, which allowed the holder to be evacuated and filled with D2 gas at variable pressure. A preliminary diffraction pattern was collected on the foil within a standard vanadium holder, and the Rietveld analysis showed that preferential orientation had negligible effects on the sample. A pattern was recorded also on the empty Al pressure cell, and it was analysed by profile matching method with constant relative intensities. This fit was later included in all main refinements of the deuterated alloy as a second phase, where the only free parameters were the scale factor and the cell parameter. For pure Pd77Ag23 the cubic lattice constant of 3.9274(1) A was obtained at room
temperature, slightly lower than the X-ray value of 3.929(1) A previously reported [22]. A critical experimental aspect was the temperature control. Due to the heavy mass of the pressure cell, the regulation thermocouple (close to the heating element within the cryostat) measured a slightly higher value (T1), and the thermocouple located in the sample stick a slightly lower value (T2) with respect to that expected on the sample. The average (T1 þ T2)/2 was thus assumed as reasonable estimate of the sample temperature. Three cooling cycles were performed at different constant pressures, namely p(D2) ¼ 1, 2, and 4 bar. Each one consisted of six data collections at T ¼ 196, 157, 137, 117, 98, 78 C, for a total of 18 diffraction patterns. After the first and second cycles the sample holder was reheated to 196 C, the pressure increased, and the next cooling cycle started. At the end of the 4 bar run the holder was pump-evacuated at 78 C, D2 was pumped in again to 1 bar, and a final heating cycle was carried out at that pressure, collecting data at T ¼ 78, 98, 117, 137, 157, 196 C. Between each temperature change at constant pressure, a waiting time of 0.5 h was applied to thermalize the sample before starting data collection.
Results and discussion Analysis of data and structure refinements All collected patterns were analysed by the FullProf suite [23], including two phases in the profile refinement: the facecentred-cubic (FCC) Pd0.772Ag0.228Dn deuterated alloy (Rietveld refinement), and FCC aluminium (profile matching with constant relative intensities) to include the sample environment reflections. The background was linearly interpolated by a set of fixed points. Bragg peaks were modelled by a pseudoVoigt function (linear combination of Gaussian and Lorentzian components, with s and g half-widths, respectively). The s and g parameters vary with q as s ¼ (U tan2q þ V tanq þ W þ P/cos2q)1/2 and g ¼ X tanq þ Y/cosq; the U, V and W coefficients are mainly related to instrumental resolution, P and Y are the Scherrer coefficients for Gaussian and Lorentzian particle size broadening, respectively, and X is related to particle strain broadening. The mixing coefficient and the full width of the pseudo-Voigt function depend on s and g according to equations given in the literature [24]. For Al, only the unit-cell constant and the scale factor were refined in each pattern, keeping fixed the other Al parameters from the calibration results of the empty cell. For the main Fm3m phase (Pd0.772Ag0.228Dn) the scale factor, the unit-cell edge, the displacement factors of Pd0.772Ag0.228 and of D, and the U, V, W, Y profile parameters were always refined. Slight changes of U, V and W indicated a weak dependence of these parameters on experimental conditions of the sample, in addition to instrumental resolution. A Fourier difference map was calculated before inserting D atoms, and it showed only a strong peak at x ¼ ½, y ¼ 0, z ¼ 0 (octahedral site). After including the D atom in that position in the refinement, with a variable occupation factor and displacement factor, a subsequent Fourier difference map showed two weak residual peaks at 0.26,0,0 and ¼,¼,¼
6789
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
coordinates. As these small signals were not detected in the data collected at ambient conditions before gas uploading, they should be related to D2 absorption. However, they could not be refined successfully as additional D atoms. Therefore, only the D atom located on the octahedral site was kept in the final model. Yet it cannot be excluded that a minor quantity of deuterium is present also in the tetrahedral site and/or in disordered locations around the octahedral site. A better quality of data at improved resolution is necessary to solve this problem. The final results of Rietveld refinements are reported in Table 1, including the agreement factors of profile fitting (Rp not weighted, wRp weighted) and of Bragg peak intensities (RB). The sample lattice constant is seen to decrease on raising temperature, and to increase with pressure (the peculiar behaviour of the two cycles at 1 bar will be discussed below), in agreement with the reported hydrogen solubility in the Pde Ag alloy [25]. One of the collected patterns is shown in Fig. 1 as an example. Because of the similar values of the cubic lattice constants of the two phases, a significant overlap is observed for some Bragg peaks of the sample with the corresponding ones of the aluminium cell. This is the case particularly in the low T range (below 140 C), because the thermal expansion coefficient of the Pd77Ag23 alloy turns out to be much larger than that of aluminium, so that the overlap increases on cooling (cf. Fig. 2). However, the Rietveld refinement of the alloy, together with the profile matching with constant relative intensities of Al from the pressure cell, seems to deal with the problem satisfactorily, as it appears from the quite reasonable estimated standard deviations (e.s.d.) of the refined parameters.
of Fig. 1 e Neutron diffraction pattern (l ¼ 1.2871 A) Pd0.772Ag0.228Dn recorded at 196 C and 1 bar (cooling cycle). Measured, calculated and difference intensity profiles are shown. Upper and lower ticks indicate Bragg peak positions of the deuterated alloy and of Al of the cell, respectively.
Thermal and pressure behaviour As the FCC structure contains as many octahedral voids as metal atoms, the refined D occupation factor is numerically equal to n ¼ D/M, i.e. the deuterium-to-metal ratio in the Pd0.772Ag0.228Dn formula. This value is plotted in Fig. 3 as a function of T for all four thermal cycles at the different applied pressures.
Table 1 e Results of the Rietveld refinements of Pd0.772Ag0.228Dn. Agreement factors, sample and Al cubic lattice constants, D occupation factor (¼n), and displacement factors of Pd0.772Ag0.228 and D atoms are reported, with e.s.d.'s in parentheses. T ( C)
p (bar)
Rp (%)
wRp (%)
RB (%)
a ( A)
a(Al) ( A)
o.f.(D)
B(PdAg) ( A2)
B(D) ( A2)
196
1 2 4 1 1 2 4 1 1 2 4 1 1 2 4 1 1 2 4 1 1 2 4 1
8.11 7.78 7.38 7.33 7.28 6.69 6.73 7.16 6.44 6.63 6.72 6.85 6.38 6.64 6.54 6.75 6.48 6.86 6.67 6.70 6.72 7.27 6.93 7.05
11.6 11.0 10.9 10.8 10.4 9.87 10.0 10.3 9.63 9.67 9.83 9.89 9.34 9.81 9.52 9.91 9.97 10.0 9.63 9.67 10.2 10.3 10.0 10.2
5.47 4.26 4.98 3.72 4.58 4.46 4.74 4.82 6.22 3.99 5.12 5.08 5.80 4.57 5.04 4.87 4.54 5.09 5.67 5.35 4.69 6.16 6.12 5.91
3.95720(9) 3.99616(8) 4.01112(8) 3.98376(8) 3.96685(8) 4.00931(8) 4.01844(8) 4.00601(8) 3.97716(9) 4.01506(8) 4.02110(8) 4.01236(8) 3.98668(9) 4.01812(8) 4.02318(8) 4.01730(8) 3.9934(1) 4.02012(9) 4.02483(9) 4.02114(8) 3.9958(1) 4.02152(9) 4.02606(9) 4.02276(9)
4.06192(6) 4.06211(6) 4.06227(6) 4.06179(5) 4.05965(5) 4.05866(5) 4.05867(6) 4.05826(5) 4.05663(5) 4.05680(6) 4.05680(6) 4.05647(6) 4.05497(5) 4.05502(6) 4.05498(6) 4.05465(6) 4.05335(5) 4.05322(6) 4.05319(6) 4.05290(6) 4.05151(6) 4.05145(6) 4.05147(6) 4.05129(6)
0.087(9) 0.210(10) 0.272(11) 0.170(10) 0.118(8) 0.281(10) 0.318(10) 0.258(10) 0.162(9) 0.314(10) 0.341(10) 0.300(10) 0.207(9) 0.333(10) 0.354(10) 0.329(10) 0.242(9) 0.344(10) 0.372(10) 0.354(10) 0.263(10) 0.358(11) 0.383(10) 0.368(10)
1.10(4) 1.04(4) 1.06(4) 1.11(4) 0.98(4) 1.03(4) 1.13(4) 1.02(4) 1.10(4) 1.07(4) 1.10(4) 1.04(4) 0.93(4) 1.05(4) 1.08(4) 1.01(4) 0.90(4) 1.02(4) 1.04(4) 0.99(4) 0.87(4) 0.98(4) 0.99(4) 0.95(4)
2.9(7) 4.5(4) 5.0(4) 4.2(5) 3.4(5) 4.7(3) 5.1(3) 4.6(3) 4.0(4) 5.0(3) 5.1(3) 4.8(3) 3.9(3) 4.8(3) 4.9(2) 4.8(3) 4.0(3) 4.6(2) 4.8(2) 4.7(2) 4.0(3) 4.5(2) 4.6(2) 4.5(2)
157
137
117
98
78
6790
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
Fig. 2 e Magnified details (2q range 30e40 ) of neutron diffraction patterns recorded at 196 and 78 C (1 bar, cooling cycle), showing the overlap of (111) and (200) peaks of the deuterated alloy with those of Al.
conditions is thus expected to lie possibly below the heating one for the hysteresis phenomenon, but well above the cooling curve reported in Fig. 3. Indeed, this one represents nonequilibrium results of phenomenological interest to characterize absorption processes on samples not previously treated at higher pressure. The data can be fitted by a cubic curve, showing an inflexion point at about 150 C. This behaviour looks similar to that of previous results [15] on thermodynamic and electric measurements of the effects of H2 absorption by the Pd77Ag23 alloy, but it is not shown by the heating curve with full deuterium uploading. No inflexion point appears either in the fit of data at higher pressures (2 and 4 bar). Thus, further measurements extending the thermal range towards higher temperatures would be needed to settle this issue. The cooling curve at 4 bar in Fig. 3 should reflect results approximately close to equilibrium, as the sample was treated at the same high pressure as for the heating curve at 1 bar. The process of achieving the stable expanded structure configuration may have not been fully completed at 2 bar, as the D absorption amount at 78 C is slightly lower than that recorded at 1 bar after the 4 bar pre-treatment. On the whole, the effect of pressure on deuterium absorption in the 1e4 bar range seems to be scarce at the low T end (only 4% increase of D occupancy), but it increases progressively with temperature. Thus, pressure acts strongly as driving force of D insertion into the alloy at low deuterium concentration, yet it weakens substantially when the D occupancy is high. However, at the high T end (196 C) the D content is raised by 60% from 1 to 4 bar and by 30% from 2 to 4 bar (Table 1). This is less than what required by Sieverts' law: cD ¼ Ks p1=2 ;
Fig. 3 e Deuterium occupancy n in Pd0.772Ag0.228Dn from Rietveld refinements of diffraction data collected at different p(D2) pressures. Arrows indicate cooling or heating direction of thermal cycles.
As most striking feature of the results, a large difference is observed between the cooling and heating n (T) curves at p(D2) ¼ 1 bar. Thermal hysteresis gives only a minor contribution to this effect in the high temperature range. The main reason is that measurements on heating were performed after cycles at higher pressure (2 and 4 bar), which made the sample able to absorb a significantly larger quantity of deuterium, with corresponding expansion of the lattice constant. We thus believe that the heating curve is close to represent equilibrium absorption results at 1 bar, except for presumable hysteresis effects which may overestimate the deuterium solubility on increasing temperature. A cooling curve obtained in the same
(1)
where Ks is Sieverts constant, p is the deuterium partial pressure in the gas phase, and the deuterium concentration cD in the solid alloy is obviously proportional to n ¼ D/M. This deviation should be related to non-ideality effects of the solid solution, which are taken into account on replacing the deuterium concentration cD by its product with the activity coefficient gD in Formula (1): cDgD ¼ Ks p1/2 is the generalization of Sieverts' law for non-ideal solutions, where gD depends on cD and p, and Ks only on T. On comparing concentrations determined at different pressures p1 and p2, we thus obtain that gD(p2)/gD(p1) ¼ (p2/p1)1/2cD(p1)/cD(p2). If p1 ¼ 1 bar is assumed, by using the data of Table 1 at T ¼ 196 C (heating cycle for 1 bar) the ratio gD(p2)/gD(p1) takes the values 1.145 and 1.250 for p2 ¼ 2 and 4 bar, respectively, whereas it should always be 1 in case of validity of Sieverts' law (gD ¼ 1 in ideal solutions). This increase of gD with pressure means that the concentration cD has increased with pressure less than predicted by the ideal Sieverts' law (negative deviation): then the energy raising due to DeD repulsion must be larger than the energy lowering due to the DePd(Ag) interaction [26,27]. It is interesting also to consider the relationship between the deuterium content n and the lattice constant a of Pd0.772Ag0.228Dn for the 24 experiments performed at different T and p(D2) values (Fig. 4). At constant pressure a increases linearly with n for all four thermal cycles, according to a ¼ k1n þ k2. The linear coefficients are k1 ¼ 0.22(1), 0.174(5),
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
Fig. 4 e Cubic lattice constant of Pd0.772Ag0.228Dn, measured at different temperatures, plotted vs. the deuterium occupancy n. 0.135(6) A, and k2 ¼ 3.940(2), 3.960(1), and 3.975(2) A for the cooling cycles at 1, 2 and 4 bar, respectively, and k1 ¼ 0.20(1), A for the heating cycle at 1 bar. As the linear k2 ¼ 3.953(4) thermal expansion coefficient of the Pd77Ag23 alloy is 1.36 105 C1, the bare temperature effect appears to be negligible with respect to deuterium absorption/desorption in determining the lattice constant change (k1 coefficient). The shift of line intercept k2 between the two lines at 1 bar confirms clearly that the cubic structure has been expanded by effect of the pre-treatment at 4 bar. The partial molar volume Vm,D of D dissolved in the Pd77Ag23 alloy can be calculated from the standard thermodynamic relationship Vm;D ¼ Vm þ ð1 cD Þ
vVm ; vcD
(2)
where cD ¼ n/(1 þ n) is the deuterium mole fraction, and the alloy molar volume Vm is related to the lattice constant a and thus to n according to the above linear dependence. By using the empirical results previously reported for the k1 coefficient, are obtained in the range values of Vm,D 1.10e1.16 105 m3 mol1 (slightly but significantly larger than the molar volume 0.91 105 m3 mol1 of the pure alloy). The amount of deuterium absorbed by the Pd77Ag23 alloy at room pressure, as determined by neutron diffraction methods in this study, differs substantially according to whether the sample was previously treated thermally at 4 bar or not. It is thus interesting to compare such results with some of those reported in the literature, at corresponding temperatures, for thermodynamic measurements of hydrogen [15,28e31] and deuterium [22,30,31] absorption on alloys of similar compositions. Data in the thermal range corresponding to ours are shown in Fig. 5. The spread between thermodynamic results [15] and [29] for the H case appears to be quite large at low T, but it becomes small at high T; our results obtained for D in conditions close to equilibrium lie approximately in-between, and reasonably close to previous results on deuterium absorption [22]. In the latter study data (not shown in Fig. 5 for clarity) are reported also for hydrogen, so that it is possible to
6791
Fig. 5 e n value in Pd1¡yAgyXn systems vs. T by diffraction (this work) and thermodynamic (literature data) methods. Open symbols: X ¼ D; full symbols: X ¼ H. estimate the D/H isotopic effect on the solubility in the Pd77Ag23 alloy: a gap of 6%e26% is observed, in the 100e150 C thermal range, between the absorption amounts of deuterium and hydrogen. Thus, by applying a similar isotopic shift to our activated D data the corresponding curve for H absorption would be slightly shifted above. On the other hand, the curve obtained in this study at 1 bar on the sample before treatment at higher pressure lies significantly below all other data, confirming that such results were obtained in nonequilibrium conditions.
Conclusions The results of this study confirm that deuterium absorption in the Pd77Ag23 alloy gives rise to a non-stoichiometric Pd0.772Ag0.228Dn compound, where D atoms are located in the octahedral voids of the FCC metal structure. A small additional occupancy of tetrahedral sites cannot be excluded, and will be the object of further investigations with neutron diffraction at higher resolution. The D/M ratio n, obtained as deuterium occupancy from structure refinements, approaches its equilibrium value after thermal treatment of the sample at pressures higher than 1 bar. In these conditions our absorption results lie within the spread of those obtained from thermodynamic measurements on H uploading by the same alloy. In absence of high pressure pre-treatment, a significantly smaller D/M ratio is observed in the absorption experiment. Also these phenomena need to be investigated in our planned neutron studies focused on the kinetic aspects of deuterium absorption into the Pd77Ag23 alloy. On increasing pressure, the amount of uploaded deuterium into the sample deviates negatively from Sieverts' law: the D concentration in the solid phase increases with p(D2) less than predicted by the thermodynamics of ideal solutions. Such effects can be explained by the repulsion between adjacent D atoms exceeding metal-deuterium attraction in the alloy structure: the force resulting from unbalance of these two
6792
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 2 ( 2 0 1 7 ) 6 7 8 7 e6 7 9 2
effects is responsible for the observed non-ideal solution behaviour.
Acknowledgements We thank Francesco Grazzi (CNR-ISC, Sesto Fiorentino, Italy) for useful discussions.
references
[1] Biccikovci O, Straka P. Production of hydrogen from renewable resources and its effectiveness. Int J Hydrogen Energy 2012;37:11563e78. [2] Orhan MF, Dincer I, Rosen MA, Kanoglu M. Integrated hydrogen production options based on renewable and nuclear energy sources. Renew Sustain Energy Rev 2012;16:6059e82. [3] Mansilla C, Louyrette J, Albou S, Bourasseau C, Dautremont S. Economic competitiveness of off-peak hydrogen production today e a European comparison. Energy 2013;55:996e1001. [4] Mansilla C, Avril S, Imbach J, Le Duigou A. CO2-free hydrogen as a substitute to fossil fuels: what are the targets? Prospective assessment of the hydrogen market attractiveness. Int J Hydrogen Energy 2012;37:9451e8. [5] Shu J, Grandjean BPA, Van Neste A, Kalaguine S. Catalytic palladium-based membrane reactors: a review. Can J Chem Eng 1991;69:1036e60. [6] Drioli E, Fontananova E. Membrane technology and sustainable growth. Chem Engin Res Des 2004;82(A12):1557e62. [7] Mulder M. Basic principles of membrane technology. 2nd ed. Dordrecht: Kluwer; 1991. [8] Kikuchi E. Membrane reactor application to hydrogen production. Catal Today 2000;56:97e101. [9] Dixon AG. Recent research in catalytic inorganic membrane reactors. Int J Chem React Eng 2003;1(R6). http://www. bepress.com/ijcre/vol1/R6. [10] Tosti S. Pd-based membranes and membrane reactors for hydrogen production. In: Hilal H, Ismail AF, Wright CJ, editors. Membrane fabrication. CRC Press, Taylor & Francis Group; 2015, ISBN 978-1-4822-1045-3. Ch. 13. [11] Cornaglia L, Munera J, Lombardo E. Recent advances in catalysts, palladium alloys and high temperature WGS membrane reactors. A review. Int J Hydrogen Energy 2015;40:3423e37. [12] Alique D, Imperatore M, Sanz R, Calles JA, Giacinti Baschetti M. Hydrogen permeation in composite Pdmembranes prepared by conventional electroless plating and electroless pore-plating alternatives over ceramic and metallic supports. Int J Hydrogen Energy 2016;41:19430e8. [13] Tosti S, Ghirelli N, Borgognoni F, Trabuc P, Santucci A, Liger K, Marini F. Membrane reactor for the treatment of gases containing tritium. European Patent Grant 2014; EP 2582618.
[14] Tosti S, Borgognoni F, Santucci A. Electrical resistivity, strain and permeability of Pd-Ag membrane tubes. Int J Hydrogen Energy 2010;35:7796e802. [15] Pozio A, Jovanovic Z, Lo Presti R, De Francesco M, Tosti S. PdAg hydrogen content and electrical resistivity: temperature and pressure effect. Int J Hydrogen Energy 2012;37:7925e33. [16] Flanagan TB, Oates WA. The palladium-hydrogen system. Annu Rev Mater Sci 1991;21:269e304. [17] Pitt MP, Gray EM. Tetrahedral occupancy in the Pd-D system observed by in situ neutron powder diffraction. Europhys Lett 2003;64:344e50. [18] Widenmeyer M, Niewa R, Hansen TC, Kohlmann H. In situ neutron diffraction as a probe on formation and decomposition of nitrides and hydrides: a case study. Z Anorg Allg Chem 2013;639:285e95. [19] Busch G, Schlapbach L, Stucki F, Fischer P, Andresen AF. Hydrogen storage in FeTi: surface segregation and its catalytic effect on hydrogenation and structural studies by means of neutron diffraction. Int J Hydrogen Energy 1979;4:29e39. [20] Paul-Boncour V, Filipek SM, Crivello J-C, Couturas F, Morawski A. Properties of ZrNi5 deuteride synthesized under high pressure studied by neutron diffraction and first principles calculations. Int J Hydrogen Energy 2016;41:17408e20. [21] Webb TA, Webb CJ, Dahle AK, Gray EM. In-situ neutron powder diffraction study of Mg-Zn alloys during hydrogen cycling. Int J Hydrogen Energy 2015;40:8106e9. [22] Anand NS, Pati S, Jat RA, Parida SC, Mukerjee SK. Thermodynamics and kinetics of hydrogen/deuterium absorption-desorption in Pd0.77Ag0.23 alloy. Int J Hydrogen Energy 2015;40:444e50. [23] Rodriguez-Carvajal J. FULLPROF: a program for Rietveld refinement and pattern matching analysis. http://www.ill. eu/sites/fullprof/. [24] Thompson P, Cox DE, Hastings JB. Rietveld refinement of Debye-Scherrer synchrotron X-ray data from Al2O3. J Appl Crystallogr 1987;20:79e83. [25] Hughes DT, Harris IR. A comparative study of hydrogen permeabilities and solubilities in some palladium solid solution alloys. J Less Common Met 1978;61:9e21. [26] Nakamura K. Analysis of thermodynamic properties of the hydrogen solution in amorphous Fe-Ti compounds. Ber Bunsenges Phys Chem 1985;89:191e7. [27] Evans MJB. Surface area effects on the sorption of hydrogen by palladium. Can J Chem 1974;52:1200e5. [28] Flanagan TB, Wang D, Luo S. Thermodynamics of H in disordered Pd-Ag alloys from calorimetric and equilibrium pressure-composition-temperature measurements. J Phys Chem B 2007;111:10723e35. [29] Serra E, Kemali M, Perujo A, Ross DK. Hydrogen and deuterium in Pd-25 pct Ag alloy: permeation, diffusion, solubilization, and surface reaction. Met Mater Trans A 1998;29A:1023e8. € sser R, Powell GL. Solubility of protium, deuterium and [30] La tritium in palladium-silver alloys at low hydrogen concentrations. J Less Common Met 1987;130:387e94. [31] Luo W, Cowgill DF, Stewart K. Absorption isotherms for H2(D2)-Pd0.8Ag0.2 (198-323 K). J Alloys Compd 2010;489:47e50.