In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation

In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation

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 x x x ( 2 0 1 6 ) 1 e1 0 Available online at www.sciencedirect.com ScienceDi...

1MB Sizes 0 Downloads 8 Views

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 x x x ( 2 0 1 6 ) 1 e1 0

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation Julien Lang a,*, Helmut Fritzche a, Alexandre Augusto Cesario Asselli b, Jacques Huot b a b

Canadian Neutron Beam Centre, Canadian Nuclear Laboratories, Chalk River, ON, Canada  Trois-Rivieres, Trois-Rivieres, QC, Canada Institut de Recherche sur l'Hydrogene (IRH), Universite du Quebec a

article info

abstract

Article history:

In this paper we present an in-situ neutron diffraction experiment on the dehydrogenation

Received 3 August 2016

of Mg2FeH6. Stoichiometric mixtures of 2Mg þ Fe were ball milled in a hydrogen environ-

Received in revised form

ment. After desorption, the samples were exposed to different combinations of hydrogen/

22 November 2016

deuterium at different temperatures in order to absorb hydrogen or deuterium at specific

Accepted 25 November 2016

synthesis steps. Using this hydrogen/deuterium absorption mixture, we found that there

Available online xxx

may be a hydrogenedeuterium exchange happening when both Mg2FeH6 and MgD2 are present.

Keywords:

© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Hydrogen absorbing materials Metal hydrides Mechanical alloying Neutron diffraction Crystal structure Microstructure

Introduction As can be seen on the Mg e Fe phase diagram, Magnesium and Iron are immiscible and do not alloy [1]. However, as predicted by Miedema's law of reverse stability, the ternary hydride Mg2FeH6 should be stable [2,3]. In fact, Mg2FeH6 is part of a d family of hydrides of the general form Mdþ m ½THn  , where M is an alkaline, alkaline earth or a divalent rare earth metal and T is a transition metal [4]. Formation of complex Mg hydrides has been studied by Deledda and Hauback [5] and by Baum et al. [6]. In the case of Mg2FeH6, the building block is the octahedral anion complex [FeH6]4 [7]. Compared to magnesium hydride, Mg2FeH6 has a slightly lower hydrogen gravimetric capacity (5.4 wt.% versus 7.6 wt.%), but a higher volumetric capacity (9.1  1022 atoms/cm3 versus 6.5  1022

atoms/cm3). The reported heat of formation for Mg2FeH6 is disputed. Values of 98 ± 3 kJ/mol [7], 86 ± 6 kJ/mol [8] and 81 ± 28 kJ/mol [9] have been proposed. Nevertheless, these are all lower than b-MgH2's heat of formation (75 kJ/mol) which means that Mg2FeH6 is more stable than b-MgH2. These properties have created interest in Mg2FeH6 for both hydrogen and heat storage applications [10]. In a previous investigation using Electron Energy-Loss Spectroscopy (EELS) and electron microscopy, we saw that when a ball milled stoichiometric mixture of 2Mg þ Fe was exposed to a hydrogen pressure of 40 bar at 673 K, the ternary hydride phase Mg2FeH6 was formed. It grew with a columnar morphology emerging from Fe capped MgH2 particles [11]. From EELS measurements, a shift in the plasmon frequency of Mg2FeH6 was found. This shift was tentatively attributed to either a slight rearrangement of the hydrogen atoms or a

* Corresponding author. E-mail addresses: [email protected], [email protected] (J. Lang). http://dx.doi.org/10.1016/j.ijhydene.2016.11.157 0360-3199/© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

2

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 x x x ( 2 0 1 6 ) 1 e1 0

modification of the crystal structure leading to a metastable phase. In this paper, we will show that iron particles act as a catalyst to generate atomic hydrogen and also as one of the reactants to form Mg2FeH6. Since Mg2FeH6 is more stable than MgH2, in principle, it should be the first to form during hydrogenation. However, formation of the ternary hydride implies the interdiffusion of Mg and Fe, which makes the hydride formation kinetically limited. Therefore, under most conditions, the preferred reaction pathway is a two-step mechanism where MgH2 plays the role of Mg2FeH6 precursor [12e16]. However, if at a given temperature, the applied hydrogen pressure is below the equilibrium pressure of MgH2 but above the one for Mg2FeH6, then the ternary hydride will be directly synthesized [17]. In order to gather more information on the hydrogenation and dehydrogenation behavior of this system, we performed in-situ neutron diffraction on a family of samples that underwent different hydride formation paths. Some were exposed to hydrogen, some to deuterium and others to a sequence of hydrogen and deuterium. In this way, the hydrogenation pathway could be investigated.

Materials and methods In this study, we used Mg (300 mesh, 99.8% purity), Fe (22 mesh, 98% purity) and MgH2 (300 mesh, 96% purity), all in powder form and bought from Alfa Aesar. Samples were prepared by ball milling stoichiometric mixtures of MgeFe using a pulverisette 4 in a 218 cc stainless steel crucible containing 20 stainless steel balls of 10 mm in diameter, with a ball-to-powder weight ratio of 40:1. The 2Mg þ Fe stoichiometric mixtures were ball milled for 12 h under 3 MPa of hydrogen. After milling, the samples were dehydrogenated at 623 K under vacuum. The samples were then exposed to different gases (Hydrogen or Deuterium) at different pressures and temperatures. Table 1 shows each sample's preparation specifications. The 2MgD2 þ Fe sample was synthesized under conditions where only the MgH2 phase would be produced. In a similar way, during the Mg2FeD6 synthesis, conditions were selected so that only this phase would be synthesized. The samples Mg2Fe(H,D)6 and Mg2Fe(D,H)6 were synthesized under conditions where the first reaction would produce MgH2 or MgD2 in the first step. The second step was favorable to the synthesis of Mg2FeD6 or Mg2FeH6. Using different absorption gases allowed us to probe the interaction between the MgH2 and Mg2FeH6 phases. All samples were quenched at room temperature after synthesis. Neutron powder diffraction was carried out by the Neutron Scattering Branch (NSB) at the Canadian Nuclear Laboratories'

(CNL) NRU reactor in Chalk River, Ontario. Using the C2 800 wire neutron detector and a homemade gas delivery apparatus for in-situ studies. Using an Argon glovebox, the asreceived powder was loaded into a Copper coated Vanadium sample holder. The sample holder has a volume of 4.75 cc and is equipped with a VCR fitting that connects to the gas delivery apparatus via a ¼ inch stainless steel VCR component. The sample holder's wall thickness is approximately 1 mm and the copper coating is roughly 2 microns thick. The sample holder was placed in a high-vacuum aluminum chamber where it is surrounded by a Tantalum heater. A vacuum was pulled in the gas delivery system and then, the sample stick's manual valve was opened. This procedure keeps the sample under vacuum, making sure it is not exposed to air. As a baseline, each sample was first measured under vacuum at room temperature. After the baseline pattern, the temperature set point was raised from 293 K to 523 K and the temperature control ramp was set to 2 K/min. Diffraction patterns were constantly recorded as the temperature rose, allowing us to “follow” the reaction using neutron powder diffraction. The detector spans from 5 to 85 2q which means that the detector needs to be moved to a high angle position to record a full 0e120 pattern. During this study, we opted not to move the detector, thus allowing us to effectively double the number of recorded patterns during each reaction. Diffraction patterns were also recorded after a full desorption before cooling the sample back to room temperature, where a final neutron diffraction pattern was recorded. The patterns were then analyzed by Rietveld refinement, using Bruker's TOPAS 5.1 software. The crystallographic parameters used in our Rietveld refinements were taken from Konstanchuk et al. for Mg2FeH6 phase [8] and from Ellinger et al. for the MgH2 phase [18]. The parameters are presented in Table 2.

Results As all results will be quoted in wt.%, we should mention that the nominal abundance of Mg and Fe in all samples are respectively 46 and 54 wt.%. Rietveld refinement analysis has, Table 2 e Crystallographic information of the hydride/ deuteride phases used in Rietveld refinements. Parameters for the MgH2 phase are from Ref. [6] and parameters for Mg2FeH6 are from Ref. [16]. Phase Space group Lattice parameters Atomic coordinates

Table 1 e Sample composition and absorption details. Sample composition 2MgD2 þ Fe Mg2FeD6 Mg2Fe(H,D)6 Mg2Fe(D,H)6

Sample absorption details 573 673 573 573

K/10 K/17 K/10 K/10

bar bar bar bar

D2 D2 H2 þ 673 K/25 bar D2 D2 þ 673 K/25 bar H2

Phase Space group Lattice parameters Atomic coordinates

Mg2FeH6 Fm-3m (225) a ¼ b ¼ c ¼ 6.442  A Element Wyckoff H 24e Mg 8c Fe 4a

Position (x, y, z) 0.23723, 0, 0 ¼, ¼, ¼ 0, 0, 0

MgH2 P42/mnm (136) a ¼ b ¼ 4.5168, c ¼ 3.0205 Element Wyckoff H 4f Mg 2a

 A Position (x, y, z) 0.306, 0.306, 0 0, 0, 0

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

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 x x x ( 2 0 1 6 ) 1 e1 0

however, indicated that a small proportion of the raw Mg powder was oxidized. From the Rietveld refinement of the assynthesized 2MgHD2 þ Fe, we found a proportion of 4 wt.% of MgO. Therefore, using this first refinement, we chose to set the MgO abundance at 4 wt.% for all patterns. The true abundance is therefore 44 wt.% Mg, 52 wt.% Fe and 4 wt.% MgO.

2MgD2 þ Fe Neutron diffraction patterns for the 2MgD2 þ Fe sample are shown in Fig. 1. The bottom pattern represents the as-received state, the top one represents the completely desorbed state and the two patterns in between were taken during the sample's desorption. It is clear that the as-synthesized sample is fully hydrided since there are no traces of the Mg phase in its pattern. On each pattern, we can also notice two Fe phases. The a-Fe phase, of bcc structure (spacegroup Im-3m) and the gFe phase, of fcc structure (spacegroup Fm-3m). The g-Fe should only be stable at high temperatures, between 1185 and 1667 K, so it was quite interesting to observe the g-Fe phase in all samples investigated in this work. Because of the relatively long milling time, this metastable iron phase may be coming from the milling tools attrition. Its presence will be discussed in detail in Section “Mg2FeD6” and Section “Discussion”. Rietveld refinement was performed on every pattern shown in Fig. 1 and the calculated parameters are presented in Table 3. As discussed in the introduction to Section “Results”, a problem that was encountered in refining all neutron patterns presented in this paper was the presence of MgO. This MgO probably comes from the powder itself as metallic powders usually have a thin oxide layer covering the powder's grains. Fournier et al. have shown that the thickness of the oxides formed during oxidation between room temperature and 673 K varies from 1.5 to 4.3 nm [19]. Such a small crystallite size makes the Bragg peaks very broad, to the point where they are blending in with the background. For this reason, we performed Rietveld refinement on the as-received 2MgD2 þ Fe with all MgO phase parameters fixed and refined

3

the scale factor only. The refinement gave a MgO phase abundance of 4 wt.%, which is consistent with the starting powder's purity. The resulting MgO phase parameters were used for all other refinements, except for the scale factor which was adjusted manually in order to get a 4 wt.% abundance. Fig. 2 shows the fully desorbed sample's Rietveld refinement. From the residual plot, we can see that the fit is quite good. A noticeable issue with our refinements are the 3 peaks at 61, 63 and 65 that we were not able to index. These peaks were present on all Fig. 1 patterns and their intensity did not change during dehydrogenation. Table 3 presents the Rietveld refinement results of Fig. 1's Neutron diffraction pattern. From Table 1, Fig. 1 and Table 3, it is clear that when synthesis is done at 573 K and 10 bar of deuterium, MgD2 is the only hydride phase produced. In the as-synthesized pattern, recorded at 298 K, there are no indications of either Mg or Mg2FeD6 phases. The abundance of the MgH2 and Fe phases are very close to the nominal compositions. This is a confirmation that MgD2 is the only hydride phase in the sample. When the temperature was raised to 503 K, there was no sign of dehydrogenation yet, but, as expected, the lattice parameters of all phases increased, as did the thermal parameters (Biso). As Table 3 shows, the appearance of the Mg phase and abundance reduction of the MgH2 phase indicates that, at 523 K, the sample started to desorb its hydrogen. At this temperature, desorption was too fast to be able to record multiple diffraction patterns. A fully desorbed pattern was recorded at high temperature and a final pattern was recorded once the sample had cooled down to room temperature. In principle, both a- and g- Fe phases should be constant because there is no Mg2FeH6 present that may dehydrogenate. Therefore, the abundances of these two iron phases shouldn't change. However, we see from Table 3 that the abundance of the g-Fe phase slightly increases during desorption. After complete dehydrogenation, the measured abundance is slightly lower than in the as-received state. As for the a-Fe phase, we see that its abundance is slightly increased in the fully desorbed state which may indicate that the g-Fe phase has been transformed to the a-Fe phase. Since the uncertainties associated with the phase abundances are significant, especially that of the g-Fe phase, we cannot definitively conclude that a transformation from g-Fe phase to a-Fe phase really occurred. Because the abundance of g-Fe is relatively low, it was quite difficult to refine its Biso parameter. Since both a- and gFe phases were simultaneously present in the sample, their Biso parameters were constrained to be refined together and to be of the same value. The same procedure was applied to all refinements in this paper. From Table 3, we can also see that after dehydrogenation and cooldown, there were no traces of the MgD2 phase in the neutron diffraction pattern, which confirms that the desorption was complete.

Mg2FeD6 Fig. 1 e Neutron diffraction patterns of 2MgD2 þ Fe at various stages of dehydrogenation. Desorption 1 is at 503 K and desorption 2 at 523 K.

The Mg2FeD6 sample was prepared at 673 K under 17 bar of deuterium so that only the Mg2FeD6 phase was formed. Fig. 3

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

4

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 x x x ( 2 0 1 6 ) 1 e1 0

Table 3 e Rietveld refinement parameters of 2MgD2 þ Fe at various stages of dehydrogenation. Values in parentheses are standard deviation on the last significant digit. All MgO abundances were fixed at 4 wt.%. Phase

Parameter 

As-synthesized

Desorption 1

Desorption 2

Fully desorbed

Temperature ( C) RWP GOF

298 7.14 1.35

503 7.03 1.24

523 7.19 1.21

298 7.54 1.27

MgD2

RBragg Abundance a ( A) c ( A) Biso(Mg) Biso(D)

2.23 41(1) 4.5052(5) 3.0138(5) 1.4(3) 1.8(2)

2.23 39(1) 4.5136(7) 3.0215(6) 1.8(4) 2.3(2)

2.14 30(2) 4.5145(8) 3.0223(8) 1.8(5) 2.8(3)

a-Fe (bcc)

RBragg Abundance a ( A) Biso(Fe)

0.97 53(1) 2.8677(6) 0.9(1)

0.33 53(1) 2.8737(4) 1.2(1)

1.69 54(2) 2.8748(4) 1.4 (1)

55(1) 2.8752(3) 1.1(1)

g-Fe (fcc)

RBragg Abundance a

2.72 2.2(4) 3.596(3)

2.07 4.2(6) 3.603(2)

1.71 3.9(7) 3.606(2)

2.76 1.6(4) 3.570(3)

Mg

RBragg Abundance a ( A) c ( A) Biso(Mg)

2.51 8(1) 3.230(2) 5.222(7) 1.0(Fixed)

1.65 40(1) 3.2291(6) 5.242(2) 2.7(3)

Fig. 2 e Rietveld refinement of the neutron diffraction patterns of fully desorbed 2MgD2 þ Fe sample.

Fig. 3 e Rietveld refinement of the neutron diffraction pattern of as-synthesized Mg2FeD6 sample.

shows the Rietveld refinement of the as-synthesized sample's neutron diffraction pattern taken at 298 K. It is clear that the only hydride phase in the pattern is the Mg2FeD6 phase. As expected, there are no indications of the MgD2 phase. This confirms that at a temperature of 673 K and under 17 bar hydrogen/deuterium pressure, the only phase formed is Mg2FeD6. We still can detect some residual Fe and Mg phases, which hints at an incomplete reaction. Keeping the sample under vacuum, the temperature was raised from 298 to 493 K at a rate of 2 K/min in order to dehydrogenate the sample. The pattern taken at 493 K showed a slight desorption only. Consequently, the temperature was

increased to 503 K and, furthermore, to 523 K. After complete dehydrogenation, the sample was cooled to 298 K for a final diffraction pattern. All of these diffraction patterns are presented in Fig. 4. A simple inspection of Fig. 4 shows that Mg2FeD6 directly desorbs to Mg and Fe without transitioning through the MgD2 phase. The pattern labeled ‘Desorption 3’ was taken at 523 K and still shows some Mg2FeD6 phase. Full dehydrogenation was achieved during the cooldown and is confirmed by the absence of the Mg2FeD6 phase in the last pattern. Rietveld refinement results of all patterns presented in Fig. 4 are shown in Table 4.

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

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 x x x ( 2 0 1 6 ) 1 e1 0

synthesized sample. As mentioned previously, the g-Fe phase is possibly due to contamination from the milling tools. However, the present results indicate that the formation of the Mg2FeD6 phase is accompanied by the presence of a high proportion of metastable g-Fe phase. Moreover, we see that during desorption, the abundance of g-Fe increased to 18 wt.% and then returned to the same abundance as in the as-synthesized sample. Nevertheless, this observation should be taken with caution because of the uncertainties on phase abundances. In order to reach a definitive conclusion about the relationship between the g-Fe and Mg2FeD6 phase, experiments using different milling materials will be needed as we could not rule out milling attrition as the origin of the g-Fe phase.

Mg2Fe(H,D)6

Fig. 4 e Neutron diffraction patterns of Mg2FeD6 at various stages of dehydrogenation. Desorption 1 is at 493 K, desorption 2 at 503 K and desorption 3 at 523 K.

Close inspection of Table 4 shows some interesting facts. For the Mg2FeD6 phase we see that an increase in temperature from 298 to 493 K did not initiate desorption, but, as expected, the lattice and thermal parameters of all atoms increased. While desorbing, the Mg2FeD6 phase abundance decreased but all other crystal parameters did not vary significantly. This shows that the crystal structure of Mg2FeD6 did not change during dehydrogenation. For the a-Fe and g-Fe phases, their lattice and Biso parameters also increased with temperature. It should be noted that the proportion of g-Fe is higher in the Mg2FeD6 as-synthesized sample than it is in the 2MgD2 þ Fe as-

Having established that, at 573 K and 10 bar of deuterium, only the MgD2 phase is formed and that at 673 K and 17 bar of deuterium only Mg2FeD6 phase is produced, we decided to synthesize samples by performing two successive hydrogenations. In order to distinguish between the two reactions, the first was at 573 K and 10 bar of hydrogen, followed by the second reaction, at 673 K and 17 bar of deuterium. In principle, at 673 K, the MgH2 phase from the first reaction will react with iron to form Mg2FeH6. Since there are two hydrogen/deuterium atoms ‘missing’ to get the right stoichiometry, we expect the Mg2FeH6 phase to be composed of 4 hydrogen atoms and 2 deuterium atoms. We also expect that if there is some unreacted MgH2, it will exclusively be formed by hydrogen atoms. Fig. 5 shows the diffraction patterns of the Mg2Fe(H,D)6 sample at different dehydrogenation stages. We see that the as-synthesized pattern has peaks belonging to both MgH2 and Mg2FeH6 phases. Upon temperature increase, both phases

Table 4 e Rietveld refinement parameters of Mg2FeD6 at various stages of dehydrogenation. Values in parentheses are standard deviation on the last significant digit. All MgO abundances were fixed at 4 wt.%. Phase

Parameter 

As-synthesized

Desorption 1

Desorption 2

Desorption 3

Fully desorbed

Temperature ( C) RWP GOF

298 8.58 1.32

493 9.17 1.36

503 8.96 1.30

523 9.48 1.32

298 9.13 1.22

a-Fe (bcc)

RBragg Abundance a ( A) Biso(Fe)

3.53 21(1) 2.8681(4) 1.4(2)

5.68 24(1) 2.8779(5) 2.0(3)

3.59 29(1) 2.8769(5) 1.8 (2)

2.89 39(2) 2.8769(7) 1.2(2)

0.92 57(2) 2.8759(7) 1.3(2)

g-Fe (fcc)

RBragg Abundance a ( A)

8.18 14(1) 3.5564(7)

11.73 13(1) 3.5660(7)

9.14 18(1) 3.564(1)

9.02 17(1) 3.565(1)

5.26 12(1) 3.563(1)

Mg

RBragg Abundance a ( A) c ( A) Biso(Mg)

2.74 9(1) 3.217(1) 5.212(5) 0.6(7)

2.62 10(3) 3.232(2) 5.259(7) 1.6(2)

2.54 12(2) 3.235(2) 5.246(5) 1.0(7)

2.38 24(3) 3.230(2) 5.245(7) 1.1 (7)

1.62 27(2) 3.229(2) 5.242(5) 1.1(4)

Mg2FeD6

RBragg Abundance a ( A) Biso(Mg) Biso(Fe) Biso(D)

2.11 52(3) 6.4367(7) 0.3(2) 0.5(2) 1.2(2)

2.85 49(2) 6.4713(8) 1.4(2) 0.6(2) 2.2 (2)

1.70 30(2) 6.468(1) 1.0(3) 0.4(3) 2.2(2)

1.80 17(2) 6.468(1) 1.1(7) 0.6(6) 2.7(6)

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

6

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 x x x ( 2 0 1 6 ) 1 e1 0

Fig. 5 e Neutron diffraction patterns of Mg2Fe(H,D)6 at various stages of dehydrogenation. Desorption 1 is at 473 K, desorption 2 at 493 K, desorption 3 at 503 K and desorption 4 at 523 K.

desorb and there are no hydride phase peaks remaining in the fully desorbed pattern. Rietveld refinement was performed on all Fig. 5 patterns. The refinement's results are presented in Table 5. Fig. 6 shows the refinement plot of the as-synthesized sample as an example. The first striking feature of Fig. 6 is the high background. In fact, the background seems to be linearly dependent with the angle. This is due to the high hydrogen content of the assynthesized sample. As dehydrogenation progresses, we see a strong reduction of the background. For the fully desorbed sample, the background was of the same order of magnitude as for the desorbed 2MgD2 þ Fe and Mg2FeD6 samples. Two hydride phases, MgH2 and Mg2FeH6, are detected. Compared to the (2MgH2 þ Fe) sample which was only hydrogenated at 573 K, the MgH2 phase abundance was reduced from 40 wt.% to 24 wt.%. The difference corresponds exactly to the Mg2FeH6 phase abundance. This proves that the Mg2FeH6 phase originated from the MgH2 phase that was formed during the first reaction. Furthermore, the total abundance of Fe also matches the expected value corresponding to unreacted MgH2. Surprisingly, the relative abundance of hydrogen and deuterium in the MgH2 and Mg2FeH6 phases are not what was expected. From Table 5, we can see that the H/D abundances in the MgH2 and the Mg2FeH6 phases are respectively 0.87/0.13 and 0.78/0.22. As

Table 5 e Rietveld refinement parameters of Mg2Fe(D,H)6 at various stages of dehydrogenation. Values in parentheses are standard deviation on the last significant digit. All MgO abundances were fixed at 4 wt.%. Phase

Parameter

As-synthesized

Desorption 1

Temperature (K) RWP GOF

298 3.35 1.19

473 3.54 1.19

Mg(H,D)2

RBragg Abundance a ( A) c ( A) Biso(Mg) Occ(H) Occ(D) Biso(H,D)

1.02 24(2) 4.513(2) 3.020(2) 0.4(5) 0.87 (6) 0.13(6) 7(3)

0.59 19(3) 4.525(2) 3.023(2) 1.1 (7) 0.87(6) 0.13(6) 5(3)

a-Fe (bcc)

RBragg Abundance a ( A) Biso(Fe)

0.83 37(2) 2.8659(4) 0.8(1)

g-Fe (fcc)

RBragg Abundance a ( A)

0.49 3.2(6) 3.553(3)

Mg

RBragg Abundance a ( A) c ( A) Biso(Mg)

Mg2Fe(H,D)6

RBragg Abundance a ( A) Biso(Mg) Biso(Fe) Occ(H) Occ(D) Biso(H,D)

0.74 33(2) 6.445(1) 0.6(4) 0.6(4) 0.78(6) 0.22(6) 2(2)

Desorption 2

Desorption 3

Desorption 4

Fully desorbed

493 4.35 1.20

503 4.84 1.17

523 6.33 1.23

298 7.03 1.26

0.91 38(2) 2.8724(4) 1.0(1)

0.66 37(1) 2.8743(3) 1.0 (1)

0.31 43(1) 2.8752(3) 0.83(9)

0.66 52(1) 2.8753(3) 0.92(8)

0.81 55.6(8) 2.8670(2) 0.59(7)

1.28 2.9(6) 3.562(3)

2.56 2.7(4) 3.567(2)

2.37 2.7(5) 3.568(2)

4.72 3.0(3) 3.570(2)

3.58 3.0(3) 3.558(2)

1.09 27(1) 3.227(1) 5.241(3) 2.7(4)

1.03 31(1) 3.2294(8) 5.245(3) 2.7(3)

0.96 35(1) 3.2305(7) 5.243(2) 2.4(2)

1.55 37.4(8) 3.2115(5) 5.215(1) 1.0(1)

1.36 30(1) 6.474(1) 2.0(4) 1.1(3) 0.80(3) 0.20(3) 3(2)

1.22 19(1) 6.478(1) 1.1(5) 0.9(4) 0.81 (4) 0.19(4) 4(2)

1.58 4(1) 6.479(2) 2(1) 1(1) 0.8(1) 0.2(1) e

0.72 36(2) 6.468(1) 1.8(5) 0.9(3) 0.76(3) 0.24(3) 1 (2)

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

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 x x x ( 2 0 1 6 ) 1 e1 0

Fig. 6 e Rietveld refinement of the neutron diffraction pattern of as-synthesized Mg2Fe(H,D)6 sample. expected, for both phases, the Biso parameter of the metal atoms increases with temperature. However, in the case of hydrogen/deuterium, the Biso values decreased but we have to take this behavior with a degree of skepticism. First, for each phase, the Biso of hydrogen and deuterium were constrained to be identical. Second, as more than 75% of the sites are occupied by hydrogen, we expect that the error on Biso to be quite important. As seen in Table 3, the error on Biso of hydrogen and deuterium is quite high. We decided to still refine this parameter but the values should be taken mostly as an order of magnitude. From Table 5, we see that during dehydrogenation, the Mg2Fe(H,D)6 phase abundance slowly decreases but all of its other parameters remained constant. In particular, the relative H/D abundance did not significantly change. This is in agreement with the Mg2FeD6 phase dehydrogenation results shown in the preceding section.

Mg2Fe(D,H)6 For this sample, the hydrogenation/deuteration sequence was the inverse of the previous sample, i.e. first, deuteration at 573 K and 10 bar of deuterium followed by hydrogenation at 673 K and 17 bar of hydrogen. As for the previous samples, a neutron diffraction pattern was taken on the fully deuterated/ hydrogenated sample at room temperature and afterward, the temperature was raised in order to desorb the sample. A final pattern was taken at room temperature of the fully desorbed sample. All these patterns are shown in Fig. 7. Patterns shown in Fig. 7 are very similar to those of Fig. 5. At first glance, the main difference seems to be the intensities of the hydrides phases which are higher in Fig. 7 compared to Fig. 5. Since the Mg2Fe(D,H)6 sample contains more deuterium and less hydrogen than Mg2Fe(H,D)6 and because the coherent scattering length of deuterium is larger than hydrogen, this was expected. Rietveld refinement was performed on all patterns presented in Fig. 7. A representative fit is given in Fig. 8 and the parameters are reported in Table 6.

7

Fig. 7 e Neutron diffraction patterns of Mg2Fe(D,H)6 at various stages of dehydrogenation. Desorption 1 is at 473 K, desorption 2 at 493 K and desorption 3 at 523 K. When comparing Figs. 6 and 8, it looks like the relative phase abundances are quite different. However, we believe that this effect is due to the different deuterium proportions in the samples. This is also why the background is much less important in Fig. 8 than in Fig. 6. In fact, the Rietveld refinement parameters for the as-synthesized sample listed in Tables 5 and 6 are very similar and the phase abundances are almost the same. The main difference is the H/D ratio which is 0.17/0.83 for the MgH2 phase and 0.16/0.84 for the Mg2FeH6 phase. These two ratios are within the errors of the inverse of the ratios found for the Mg2Fe(H,D)6 sample. Because of deuterium's larger scattering length, the occupation factor's and Biso's uncertainty values are much smaller than for the Mg2Fe(H,D)6 sample's pattern. The previous Rietveld refinements were performed under the assumption that hydrogen and deuterium could substitute each other in the same hydride phase. However, we should also take into account the situation where hydrogen and deuterium form separated phases. This means that, in the case of as-synthesized Mg2Fe(D,H)6, instead of having the Mg2Fe(H0.16D0.84)6 phase, it may be a mixture of 0.16Mg2FeH6 and 0.84Mg2FeD6. To test this hypothesis, we performed a Rietveld refinement on the as-synthesized Mg2Fe(D,H)6 pattern using the Mg2FeH6, Mg2FeD6, MgH2, and MgD2 phases. Table 7 shows the results of the refinement. As can be seen from Table 7, the resolution of the powder diffraction patterns taken in the present investigation could not distinguish between a mixture of hydrogen and deuterium in the same phase and a mixture of two identical phases, one with hydrogen and the other with deuterium.

Discussion In this investigation we found two new and unexpected characteristics of the 2Mg þ Fe system during dehydrogenation. We saw that there may be an exchange between the

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

8

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 x x x ( 2 0 1 6 ) 1 e1 0

Table 7 e Rietveld refinement parameters of room temperature neutron diffraction pattern of assynthesized Mg2Fe(H,D)6. Nominal numbers are calculated from values of Table 6. Phase

Parameter

Nominal

Measured

Global Global

RWP GOF

4.78 1.39

5.13 1.49

MgH2

Abundance (wt.%) RBragg

3 1.18

1(2) 1.48

MgD2

Abundance (wt.%) RBragg

17 1.18

17(1) 2.43

Mg2FeH6

Abundance (wt.%) RBragg

30 1.36

32(1) 2.74

Mg2FeD6

Abundance (wt.%) RBragg

6 1.36

6(3) 2.60

hydrogen/deuterium atoms of the hydride/deuteride phase and the gaseous hydrogen/deuterium atoms. In a previous investigation we have shown that in the MgeFe system, Fe atoms from a catalytic iron particle diffusing towards a MgH2 substrate to form the Mg2FeH6 phase [11]. In the present investigation, we have found the presence of g-Fe phase in all samples. The g-Fe could be from the milling tool; however, we will explore other possibilities for

Fig. 8 e Rietveld refinement of the neutron diffraction pattern of as-synthesized Mg2Fe(D,H)6 sample.

Table 6 e Rietveld refinement parameters of Mg2Fe(H,D)6 at various stages of dehydrogenation. Values in parentheses are the standard deviation on the last significant digit. For all refinements, MgO abundance was fixed at 4 wt.%. Phase

Parameter

As-synthesized

Desorption 1

Temperature (K) RWP GOF

298 4.78 1.39

473 4.83 1.31

Mg(H,D)2

RBragg Abundance a ( A) c ( A) Biso(Mg) Occ(H) Occ(D) Biso(H,D)

1.18 20(1) 4.5084(8) 3.0170(8) 1.0 (4) 0.17(3) 0.83(3) 2.5(4)

1.31 17(2) 4.518(1) 3.026(1) 1.5(6) 0.17(4) 0.83(3) 3.4(7)

a-Fe (bcc)

RBragg Abundance a ( A) Biso(Fe)

0.56 34(1) 2.8668(3) 0.9(1)

g-Fe (fcc)

RBragg Abundance a ( A)

1.74 5.2(6) 3.557(2)

Mg

RBragg Abundance a ( A) c ( A) Biso(Mg)

Mg2Fe(H,D)6

RBragg Abundance a ( A) Biso(Mg) Biso(Fe) Occ(H) Occ(D) Biso(H,D)

1.36 36(1) 6.4359(7) 0.7(2) 0.2(2) 0.16(2) 0.84(2) 1.9(2)

Desorption 2

Desorption 3

Fully desorbed

493 5.40 1.30

523 6.56 1.46

298 6.49 1.43

1.03 31(1) 2.8748(4) 1.1(1)

0.64 38(1) 2.8766(3) 1.0(1)

0.81 52(1) 2.8775(3) 0.8(1)

1.05 51(1) 2.8678(3) 0.5(1)

2.18 7(1) 3.566(2)

5.73 5.8(4) 3.5671(8)

5.41 6.4(5) 3.569(1)

3.81 9(1) 3.555(1)

1.11 8(1) 3.231(3) 5.246(8) 3(1)

1.59 24(1) 3.231(3) 5.248(3) 2.6(5)

1.58 38(1) 3.2319(8) 5.249(2) 2.0(3)

1.57 37(1) 3.2119(6) 5.217(2) 1.0(2)

1.03 33(1) 6.4684(8) 1.1(3) 0.6(2) 0.15(2) 0.85(2) 2.7 (3)

1.08 28(1) 6.4720(8) 1.7(3) 0.4(3) 0.16(2) 0.84(2) 2.6 (3)

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

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 x x x ( 2 0 1 6 ) 1 e1 0

its appearance. First, it could come from local heating during hydrogenation of magnesium, which is an exothermic reaction. However, in metal hydride systems, temperature elevation is regulated by the hydrogen pressure and is quite limited. The highest applied pressure in our experiments was 25 bar. When the temperature of magnesium reaches more than 415  C, the hydrogenation plateau pressure becomes higher than 25 bar and the reaction cannot continue. Therefore, under our experimental conditions the local temperature increase due to hydrogenation cannot be much higher than 415  C. Moreover, the transition temperature ß-g is at 912  C which is higher than the magnesium melting temperature. In our opinion g-Fe could not be formed during ball milling except if it comes from attrition of crucible and milling balls. Another possible cause of appearance of g-Fe may be the fraction of iron which is at the interface with MgH2 (or Mg). In a previous investigation using Electron Microscopy and Energy-Loss Spectroscopy Study it was speculated that the nature of the interface at Fe/Mg2FeH6 could play a role in the hydrogenation kinetics [11]. This may explain the fact that the higher proportion of g-Fe was found for the sample Mg2FeH6 which also has the higher abundance of Mg2FeH6 phase while the lowest proportion of g-Fe was found in the MgH2 sample. The Mg2Fe(H,D)6 and Mg2Fe(D,H)6 samples had a slightly higher g-Fe abundance than the MgH2 sample. This thus provides some evidence that the g-Fe could be associated with the Mg2FeH6 phase, as explained by the model proposed in Ref. [11]. But still, because milling attrition could be the origin of g-Fe more experiments are needed to get the real reason for appearance of the g-Fe phase. In samples Mg2Fe(H,D)6 and Mg2Fe(D,H)6, the D/H abundance ratios present in the MgH2 and Mg2FeH6 phases is harder to explain. Two compounding factors affected our analysis of the hydrogen and deuterium rich samples. Deuterium has a larger coherent scattering length than hydrogen but hydrogen has a very high incoherent scattering length. These two factors lead us to assume that the refined parameters' values from the deuterium rich sample are more reliable than those of the hydrogen rich sample. Therefore, our discussion will be based on the parameters found in the Rietveld refinement of the deuterium rich sample while still taking into account the results from the hydrogen rich sample, which are the same within one standard deviation. From Table 6 we see that the D/H ratio for both MgH2 and Mg2FeH6 phases are close to 5. This means that we could write the stoichiometry of each phase as: MgD1.67H0.33 and Mg2FeD5H. As indicated in Table 2, each hydride phase has only one possible site for hydrogen/deuterium atoms. In MgH2's case, the multiplicity of the hydrogen site is 4. For the Mg2FeH6 phase, it is 24. From a crystallographic point of view, it is thus difficult to justify a D/H ratio of 5 for MgH2. In the same way for Mg2FeH6, it is problematic to imagine a crystal structure having two hydrogen sites, one with a multiplicity of 20 and the other, a multiplicity of 4. Moreover, we know that one of the building blocks of Mg2FeH6 is the anion complex [FeH6]4, where the six hydrogen atoms are equivalent. It is therefore difficult to explain the D/H ratio from a crystallographic or thermodynamic point of view. One possible explanation is that the D/H ratio is due to reaction kinetics. We know from the MgD2 results that at 573 K

9

and 10 bar of deuterium the sample is composed almost exclusively of the MgD2 phase. Therefore, when the sample was heated to 673 K in 17 bar of hydrogen, the Mg2FeH6 phase had to form from MgD2. In this case, in principle, two hydrogen atoms would bind to the hydride and give a Mg2FeD4H2 stoichiometry. However, as the Mg2FeH6 phase nucleus is surrounded by the MgD2 phase, it may be kinetically favorable to take a supplementary deuterium atom to form Mg2FeD5H1. This would leave a deficiency in the MgD2 phase which would be filled by a hydrogen atom. With this explanation, the resulting hydride phases would have both hydrogen and deuterium atoms in them. On the other hand, we should not rule out the possibility that the mixing of hydrogen and deuterium is not within the phase itself but instead that there are two phases, one with hydrogen and the other with deuterium. In this case, an explanation of why the ratio MgH2/MgD2 and Mg2FeH6/Mg2FeD6 are identical is still missing. Therefore, more work is needed to get a better understanding of the mechanism involved during hydrogenation of MgeFe mixture. In particular, in-situ neutron diffraction experiment during hydrogen absorption would give the proof of hydrogen/deuterium exchange.

Conclusions Using neutron diffraction on a series of samples synthesized by alternating hydrogen and deuterium during different hydrogenation steps, we were able to show a new characteristic in the formation of the Mg2FeH6 phase. We found that there is an exchange of deuterium and hydrogen atoms during the formation of the Mg2FeH6 phase when MgD2 is present. However, a detailed explanation of the exchange effect is still missing. Theoretical modeling (such as with density functional theory calculations) and more experimental work is needed to investigate the mechanism of the D/H exchange. We also found the presence of g-Fe in our samples, which may result from contamination from the milling tools, but the authors cannot exclude the possibility that the g-Fe phase is naturally associated with the Mg2FeH6 phase.

Acknowledgements Funding for this research was provided by NSERC (RGPIN/ 311876-2012) discovery grant. J.H. would like to thank A.J. (Timmy) Ramirez-Cuesta, L. Daemen and Y. Cheng of Oak Ridge National Laboratory for useful discussion.

references

[1] Nayeb-Hashemi AA, Clark JB, Swartzendruber LJ. The FeMg (Iron-Magnesium) system June. Bull Alloy Phase Diag. 1985;6(3):235e8. [2] Miedema AR. The electronegativity parameter for transition metals: heat of formation and charge transfer in alloys. J Less Common Metals 1973;32:117e36.

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157

10

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 x x x ( 2 0 1 6 ) 1 e1 0

[3] Miedema AR, Boom R, Boer FRD. On the heat of formation of solid alloys. J Less Common Metals 1975;41(2):283e98. [4] Yvon K. Meta hydrides: transition metal hydride complexes. In: Encyclopedia of materials: science and technology. Elsevier; 2010. p. 1e9. [5] Deledda S, Hauback BC. The formation mechanism and structural characterization of the mixed transition-metal complex hydride Mg2(FeH6)0.5(CoH5)0.5 obtained by reactive milling. Nanotechnology 2009;20(20):7. lis L. Complex Mg-based [6] Baum LA, Meyer M, Mendoza-Ze hydrides obtained by mechanosynthesis: characterization and formation kinetics. Int J Hydrogen Energy 2008;33(13):3442e6. [7] Didisheim JJ, Zolliker P, Yvon K, Fischer P, Schefer J, Gubelmann M, et al. Dimagnesium iron(II) hydrides, Mg2FeH6, containing octahedral FeH4 6 anions. Inorg Chem 1984;23:1953e7. [8] Konstanchuk IG, Ivanov EY, Pezat M, Darriet B, Boldyrev VV, Hagenmu¨ller P. The hydriding properties of a mechanical alloy with composition mixtures Mg-25%Fe. J Less Common Metals 1987;131:181e9. [9] Puszkiel JA, Larochette PA, Gennari FC. Thermodynamic and kinetic studies of Mg-Fe-H after mechanical milling followed by sintering. J Alloys Compd 2008;463(1e2):134e42.  B. High temperature metal [10] Felderhoff M, Bogdanovic hydrides as heat storage materials for solar and related applications. Int J Mol Sci 2009;10(1):325e44. [11] Danaie M, Asselli AAC, Huot J, Botton GA. Formation of the ternary complex hydride Mg2FeH6 from magnesium hydride (b-MgH2) and iron: an electron microscopy and energy-loss spectroscopy study. J Phys Chem C 2012;116(49):25701e14.

[12] Gennari FC, Castro FJ, Gamboa JJA. Synthesis of Mg2FeH6 by reactive mechanical alloying: formation and decomposition properties. J Alloys Compd 2002;339:261e7. [13] Asselli AAC, Leiva DR, Jorge AM, Ishikawa TT, Botta WJ. Synthesis and hydrogen sorption properties of Mg2FeH6MgH2 nanocomposite prepared by reactive milling. J Alloys Compd 2012;536(Suppl. 1):S250e4. [14] Leiva DR, Villela ACS, Paiva-Santos CO, Fruchart D, Miraglia S, Ishikawa TT, et al. High-yield direct synthesis of Mg2FeH6 from the elements by reactive milling. In: Solid state phenomena; 2011. p. 259e62. [15] Puszkiel J, Gennari F, Larochette PA, Karimi F, Pistidda C, Gosalawit-Utke R, et al. Sorption behavior of the MgH2Mg2FeH6 hydride storage system synthesized by mechanical milling followed by sintering. Int J Hydrogen Energy 2013;38(34):14618e30. [16] Polanski M, Nielsen TK, Cerenius Y, Bystrzycki J, Jensen TR. Synthesis and decomposition mechanisms of Mg2FeH6 studied by in-situ synchrotron X-ray diffraction and high-pressure DSC. Int J Hydrogen Energy 2010;35(8):3578e82. [17] Asselli A, Huot J. Investigation of effect of milling atmosphere and starting composition on Mg2FeH6 formation. Metals 2014;4(3):388e400. [18] Ellinger FH, Holley Jr CE, McInteer BB, Pavone D, Potter RM, Staritzky E, et al. The preparation and some properties of magnesium hydride. J Am Chem Soc 1955;77:2647e8. [19] Fournier V, Marcus P, Olefjord I. Oxidation of magnesium. Surf Interface Anal 2002;34(1):494e7.

Please cite this article in press as: Lang J, et al., In-situ neutron diffraction investigation of Mg2FeH6 dehydrogenation, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.157