Microstructure and hydrogen storage properties of Mg-Ni-Ce alloys with a long-period stacking ordered phase

Microstructure and hydrogen storage properties of Mg-Ni-Ce alloys with a long-period stacking ordered phase

Journal of Power Sources 338 (2017) 91e102 Contents lists available at ScienceDirect Journal of Power Sources journal homepage: www.elsevier.com/loc...

6MB Sizes 3 Downloads 36 Views

Journal of Power Sources 338 (2017) 91e102

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Microstructure and hydrogen storage properties of Mg-Ni-Ce alloys with a long-period stacking ordered phase Lishuai Xie, Jinshan Li*, Tiebang Zhang**, Lin Song, Hongchao Kou State Key Laboratory of Solidification Processing, Northwestern Polytechnical University, Xi'an, 710072, China

h i g h l i g h t s  A new long-period stacking ordered phase is observed in Mg-Ni-Ce ternary alloy.  Nanocrystals with grain size in the range of 3e5 nm are observed after ball milling.  CeH2.73 reacts with oxygen spontaneously to form CeO2 during air exposure.  Anti-oxidation properties of Mg-based alloys are improved with the addition of Ce.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 25 September 2016 Received in revised form 28 October 2016 Accepted 6 November 2016 Available online 12 November 2016

On the basis of catalytic actions of transition metals and rare earth metals on Mg-based hydrogen storage alloys and aiming at alleviating the adverse influence of Mg oxidation, Mg-Ni-Ce alloys with different Ni and Ce contents were prepared by near equilibrium solidification. A new 18R-type long-period stacking ordered phase (LPSO) was formed coherently with Mg, Ni-substituted Mg12Ce and Mg2Ni in Mg-rich MgNi-Ce ternary alloys. Distinct from the reported LPSO structures in other Mg-based alloys, in which the LPSO structures were fundamentally long period stacking variants of hexagonal close-packed structure of Mg, the LPSO structure found in the present work was a variant of Mg12Ce. Nanocrystalline alloys were obtained by high-energy ball milling the as-cast alloys. Nanocrystals of Mg, Mg2Ni and Mg12Ce with grain size in the range of 3e5 nm were observed in ball milled samples. The activation performance, isothermal hydrogenation behavior and anti-oxidation properties of the ball milled samples were systematically investigated and corresponding mechanisms were discussed based on detailed microstructural characteristics. CeH2.73 was formed after hydrogenation and spontaneously transformed into CeO2 during air exposure. The anti-oxidation properties of Mg-based hydrogen storage alloy were substantially improved with the addition of Ce by forming CeH2.73/CeO2 composite. © 2016 Elsevier B.V. All rights reserved.

Keywords: Mg alloy Hydrogen storage Anti-oxidation property Long-period stacking ordered phase

1. Introduction Hydrogen energy, which can be sourced from a variety of renewable sources such as solar, wind and wave power, has been the topic of much research in past decades due to the possibility of reducing reliance on fossil fuels. Technologies for producing hydrogen from water as well as other sources have already been well established and at the end of the energy vector, hydrogen can be used in fuel cells for electricity generation or used to fuel a

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (T. Zhang).

(J.

Li),

http://dx.doi.org/10.1016/j.jpowsour.2016.11.025 0378-7753/© 2016 Elsevier B.V. All rights reserved.

[email protected]

combustion engine. To date the unresolved issue in the use of hydrogen, particularly in the area of mobile applications, is safe and efficient hydrogen storage. Among all kinds of hydrogen storage materials, Mg and Mgbased alloys are generally considered as promising candidates for solid-state hydrogen storage applications because of their high gravimetric hydrogen storage densities of up to 7.6 wt% in the case of MgH2, good reversibility and low cost [1,2]. The major problems with Mg as a reversible hydrogen storage material are its high desorption temperature of well over 300  C and sluggish de-/absorption kinetics [3]. Even so, decent hydrogen storage properties can be obtained at temperature as low as 150e200  C through reasonable alloying and efficient processing [4e6]. Ni has been proved to be an effective catalytic element in Mg-based alloys [7,8]. Most recently, Mg-Ni-RE (rare earth elements) ternary alloys have

92

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

shown superior comprehensive hydrogen storage properties due to their highly dispersed structure and the synergetic influence between Ni and RE [4,9,10]. The size, shape and distribution of RE hydride, Mg2Ni hydride and Mg hydride affect the catalytic performance during de-/hydrogenation process. The newly discovered LPSO phase in Mg-based structural materials can also significantly ameliorate the hydrogen storage properties of Mg [11]. It has been proven to obtain extremely fine catalysts with high dispersion by hydrogen-induced decomposition of the LPSO phase [12,13]. However, the precise structure of LPSO phase remains elusive up to now. Moreover, the LPSO precipitates in many forecasted Mg-XL-XS (XL and XS are elements larger and smaller than Mg) ternary systems have not been observed experimentally, even though the thermodynamic stabilities of which have been examined by density functional theory (DFT) [14]. Moreover, it is feasible to improve the hydrogen storage properties by refinement of Mg-based alloys to nanoscale. To this end, magnesium-rich alloys are commonly ground by high-energy ball milling (HEBM) [15], rapid quenching [16] or equal channel angular pressing (ECAP) [17] to reduce the average grain size. HEBM can also finely disperse catalyst particles by repeated welding and fracture and significantly shorten the time of activation needed for Mg-based alloys by introducing a large number of defects [18]. Moreover, a newly developed ball-milling process assisted by dielectric-barrier discharge plasm has shown great potential to ameliorate the hydrogen storage properties of Mg-based alloys by dual-tuning the thermodynamics and kinetics [19e21]. Another key obstacle for sustainable and long-life application of Mg-based hydrogen storage alloys is that Mg is susceptible to oxidation during hydrogenation/dehydrogenation cycling, leading to rapid deterioration of hydrogen storage capacity and de-/absorption kinetics. It is especially the case for Mg nanocrystals with grain size from several to several decade nm, which are essential for Mg-based alloys to achieve excellent hydrogenation kinetics. MgO is always regarded as a barrier between MgH2 and hydrogen due to its dense rock-salt structure limiting the diffusion of hydrogen atoms [22]. Avoiding contact with oxygen or moisture is the usually adopted method for Mg-based hydrogen storage alloys to avert oxidation [23]. But it always consumes a lot of resources to maintain the oxygen-free environment. Moreover, intentional or unintentional exposition to the air of materials might not be avoided for large-scale applications of Mg-based hydrogen storage materials. Thus it is strongly recommended to know a quick and efficient way to re-active the materials or even eliminate the adverse effects of oxidation. Manufacturing an alloy with super oxidation resistance can effectively improve the rigorous operating condition. Most recently, it has been found that both the hydrogen storage capacity and the absorption/desorption kinetics can be significantly improved with rare earth metal oxides distributed on the surface of Mg, even though substantial MgO also distributes on the surface [24,25]. Inspired by the mentioned above and considering the requirement of hydrogen storage capacity, Mg-rich Mg-Ni-Ce alloys with different Ni/Ce contents are prepared in this work, in which the mass fractions of Mg are all more than 80%. As Ce is well-known oxygen getters, its presence is believed to reduce the formation of surface Mg oxides. The phase component, microstructure and hydrogen storage properties, especially the anti-oxidation performance, of Mg-Ni-Ce alloys are investigated systematically. It is expected to acquire nanocrystalline Mg-based alloys with superior activation performance, high hydrogen storage capacity and excellent oxidation resistance via the improvement of surface properties, the optimization of internal structure and the phase transformation. Unexpectedly, an 18R-type LPSO phase is observed in Mg-Ni-Ce ternary alloys. Ultrafine nanocrystals are also observed

in ball-milled samples. Anti-oxidation properties of Mg-based hydrogen storage alloys are significantly improved with addition of Ce by forming Ce hydrides and ceria after subjected to hydrogen and air. The results reveal to us that oxygen can be effectively utilized to improve the hydrogen storage properties of Mg-based alloys by reasonable alloying and efficient processing. 2. Experimental 2.1. Materials preparation The nominal compositions of the alloys prepared in the present work are Mg-20Ni (near-eutectic composition), Mg-20Ce (neareutectic composition), Mg-5Ni-5Ce and Mg-7Ni-3Ce (wt%), respectively. The alloy ingots were prepared in graphite crucible under the protection of covering agent (RJ-6) in the resistance furnace. The constituents of the covering reagent are 55 wt% KCl, 28 wt% CaCl2, 15 wt% BaCl2 and 2 wt% CaF2. The raw materials were magnesium ingot (purity 99.99%), magnesium-nickel intermediate alloy (Mg-55 wt%Ni) and cerium ingot (purity 99.99%). Extra ~5 wt% Mg and ~2 wt% Ce were added to compensate for the inherent evaporation of elements during melting. Furnace cooling was applied for near equilibrium solidification in this work. In order to make all the metals fully melted and homogeneously mixed, the molten alloys were kept at ~850  C for 30 min with intermittent stirring to ensure the homogeneousness. The final mass of each ingot was about 100 g. Prior to de-/hydrogenation measurements, the alloy samples were mechanically cut and pulverized to about 300 mm, followed by HEBM for 2 h under the protection of argon with a ball to powder weight ratio of 20:1. The high-energy shaker mill used in the present experiment was a vibratory-type Spex 8000D mixer/mill with a tungsten carbide vial of ~50 cm3. The speed of the mixer/mill was 875 rmp. The amount for each sample used for HEBM was ~2 g. All the loading and weighing were performed in a glove box under the protection of high purity argon. 2.2. Characterization and measurement The actual compositions of the as-cast alloys were measured by inductive coupled plasma emission spectrometer (ICP). The compositions of experimental alloys were measured to be Mg-19.86Ni, Mg-19.55Ce, Mg-5.17Ni-6.87Ce and Mg-9.57Ni-2.21Ce (wt%), respectively. For convenience, the samples were denoted as Mg20Ni, Mg-20Ce, Mg-5Ni7Ce and Mg-10Ni2Ce hereafter, respectively. The microstructure of the as-cast alloys and ball milled samples was characterized by scanning electron microscopy (SEM) with a backscattered mode and transmission electron microscopy (TEM) (FEI Tecnai G2 F30). Phase identification was further carried out via electron probe microanalysis (EPMA) using a JEOL JXA8500F and X-ray diffraction (XRD, DX-2700) with Cu Ka radiation. For the ball milled samples, three sorption/desorption cycles were performed to completely activate the samples at 350  C. Each cycle contained hydrogenation for 1 h under 3 MPa hydrogen pressure and dehydrogenation for 1 h with an initial hydrogen pressure in the channel of 0.002 MPa using a Sieverts-type (PCT Pro2000) apparatus. To evaluate the anti-oxidation properties of Mg-based hydrogen storage alloys with addition of Ce, the activated samples were sealed in tubes and kept in air for ~168 h without any protection. The dehydrogenation performance of activated samples after air exposure was also measured by differential scanning calorimetry (DSC) on a simultaneous TG-DTA/DSC apparatus (STA449C) at the heating rate of 5, 10 and 15  C min1, respectively, up to a maximum temperature of 500  C under 50 ml min1 flowing argon gas.

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

3. Results and discussion 3.1. Microstructure of as-cast alloys Fig. 1 presents the back-scattered SEM (SEM/BSE) images of the as-cast alloys. A significant change in the morphology with the contents of Ni and Ce changing is observed. As shown in Fig. 1(a) and (b), for Mg-20Ni and Mg-20Ce alloy, the eutectic Mg/Mg2Ni and Mg/Mg12Ce mixtures distribute as a network and both alloys preserve lamellar microstructure, which result from the eutectic reaction between Mg and Ni/Ce. Apparently, the microstructure of Mg-20Ni alloy is much finer than that of Mg-20Ce alloy according to the scale bar. The inset in Fig. 1(c) shows the SEM/BSE image of Mg-5Ni7Ce alloy at low magnification, where a eutectic mixture and some gray blocks surrounding the dark microstructures are clearly observed. The microstructure is similar to Mg-5Ni-3La (at.%) ternary alloy [26], where a eutectic mixture of alternately distributed Mg, Mg2Ni and Mg12La and some gray Mg12La blocks surrounding the primary Mg phase are observed. But the phase components and microstructure in the eutectic region in the present work are different from the Mg-5Ni-3La alloy. Fig. 1(c) shows the microstructure details of the eutectic region in the inset, displaying dark microstructures and gray microstructures with some bright strips embedded in. The EPMA analyses of the consistent phases in Mg-5Ni7Ce alloy marked with letter A-E are given in Table 1. During EPMA measurement, at least ten points were performed at each consistent phase. The black microstructures marked with A and E are Mg and gray blocks marked with B are Nisubstituted Mg12Ce. In the eutectic region, the bright strips marked with C are Ce-substituted Mg2Ni (further confirmation by TEM will be given below). But the composition of gray microstructures marked with D in Fig. 1(c) fluctuates greatly. Its composition range is about Ni1.4Ce8.1-Ni7.1Ce7.3 (at.%). The inset in Fig. 1(d) illustrates the SEM/BSE image of as-cast Mg-10Ni2Ce alloy, including of primary Mg dendrites surrounded by a eutectic mixture. The morphology is similar to as-cast Mg-

93

Table 1 EPMA results of the consistent phases in Mg-5Ni7Ce alloy and Mg-10Ni2Ce alloy. Alloy

5Ni7Ce at.%

Position

Mg

Ni

Ce

A B C D E

100.0 90.7 69.4

0 1.2 29.5

0 8.1 1.1

99.8

0.2

0

Alloy

10Ni2Ce at.%

Position

Mg

Ni

Ce

A B C D E

100

0

0

70.2 66.8 100

28.3 33.2 0

1.5 0 0

10Ni-5La (wt%) ternary alloy [27], where primary Mg dendrites surrounded by a eutectic mixture are observed. Fig. 1(d) gives the magnification image taken from the eutectic region in the inset, from which some bright strips (marked with C) are also clearly observed on the gray blocks (marked with B). The EPMA analyses of the consistent phases in Mg-10Ni2Ce alloy labeled as letter A-E are also given in Table 1. Besides Ce-substituted Mg2Ni (bright strips, marked with C), Mg2Ni without solution of Ce exhibiting block microstructure (marked with D) are also observed. Moreover, the composition of gray microstructures marked with B in Fig. 1(d) also fluctuates greatly. It is worth mentioning that the microstructure in eutectic region in Mg-5Ni7Ce alloy is much finer than that of Mg10Ni2Ce alloy. The bright-field (BF) image in the eutectic region of the as-cast Mg-5Ni7Ce alloy and corresponding selected area electron diffraction (SAED) patterns are illustrated in Fig. 2(a). Combining the SAED2 pattern in Fig. 2(a) and EPMA result, it can be concluded that the bright strips in the eutectic region in Fig. 1(c) are Cesubstituted Mg2Ni. According to the Mg-Ni-Ce ternary phase diagram [28], three phases in the eutectic zone in the Mg-rich corner are Mg, Mg2Ni and Mg12Ce. Surprisingly, a new phase besides the above mentioned three phases is observed, the diffraction pattern of which has not been reported before in Mg-Ni-Ce system. The SAED4 pattern in Fig. 2(a) clearly shows that several extra diffracting spots occur at the ±1/3(002)Mg12Ce and ±2/3(002)Mg12Ce positions, which are the evidences commonly applied in previous studies to prove the existence of the 18R structure in Mg-TM

Fig. 1. SEM/BSE images of the as-cast alloys: (a) Mg-20Ni, (b) Mg-20Ce, (c) Mg-5Ni7Ce and (d) Mg-10Ni2Ce. The insets in (c) and (d) are SEM/BSE images at low magnification.

94

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

Fig. 2. (a) TEM image in the eutectic region and corresponding SAED patterns of as-cast Mg-5Ni7Ce alloy, (b) BF image of LPSO phase labeled as 4 in (a) showing the characteristic feature of 18R with some random stacking faults, (c) HRTEM images of 18R phase and (d) TEM image in the eutectic region and corresponding SAED patterns of as-cast Mg-10Ni2Ce alloy.

(transition metals)-RE systems [29,30]. Similarly, a new 18R-type LPSO phase is also observed within Mg12Ce phase in Mg-10Ni2Ce alloy and the corresponding SAED pattern is displayed in Fig. 2(d). It is commonly accepted that LPSO structures in Mg-based alloys

are fundamentally long period stacking variants of the hexagonal close-packed structure of the Mg crystal [11,31] and the existence of LPSO phase in Mg-Ni-Ce alloy has been denied by DFT [14]. However, totally distinct from the reported structure, the newly

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

observed 18R-type LPSO structure in the present work is a variant of Mg12Ce (I4/mmm, a ¼ 1.033 nm, c ¼ 0.596 nm, JCPDS 19-0289). The orientation relationship between the Mg12Ce matrix and the 18R phase is (002)Mg12Ce//(0018)18R. Fig. 2(b) is the magnification image taken from the region indicated in the circle labeled as 4 in Fig. 2(a), showing the characteristic feature of 18R. The fringes of individual lamellae are uniformly distributed. The space between them is measured to be ~1.8 nm, slightly higher than that of the 18R structure reported in other systems (~1.6 nm [29]). Several random stacking faults (SF) are obviously observed between the fringes in Fig. 2(b) and confirmed by high resolution TEM (HRTEM) image in Fig. 2(c), which are regarded as the key precursors for the formation of LPSO phase [32]. Stacking faults can also act as tunnels for fast transportation of hydrogen to the sample interior through the exterior oxide layer [33]. Thus the eutectic region is composed of Mg, Mg2Ni, Mg12Ce and LPSO phase. The LPSO phase is a variant of Mg12Ce and this can explain why the compositions of gray microstructures in Fig. 1(c) marked with D and in Fig. 1(d) marked with B fluctuate greatly. The precise composition of the LPSO phase need to be further clarified. There are still some disagreements in the criteria for TM and RE that can participate in the formation of LPSO phases in Mg-based alloys [11,34]. In general, the characteristics of RE in LPSO MgTM-RE alloys are that RE elements have hexagonal close-packed

95

structure at room temperature and large solid solubility limits above about 3.75 at.% in Mg [35]. But the solid solubility limit of Ce in Mg is about 0.095 at% according to Mg-Ce binary phase diagram [36], which is much less than 3.75 at%. Abe et al. [37] have reported a formation mechanism of LPSO phase on the atomic level including introduction of stacking faults and segregation of solute atoms around these stacking fault layers. LPSO phase which transforms from Mg3Gd-type compound phase besides the supersaturated 2H-Mg matrix has also been reported [38]. The observed stacking faults and the enrichment of solute atoms around disordered regions caused by intermittent manual agitation in the present work probably promote the formation of LPSO phase. Therefore, the detailed structure, specific formation mechanism and the characteristics of the constituents of LPSO Mg-based alloys need to be reevaluated. 3.2. Microstructure of HEBM alloys The XRD patterns of the alloys after 2 h HEBM are illustrated in Supplementary 1. For HEBM Mg-20Ni sample, the pattern is dominated by the peaks of hexagonal close packed (hcp) Mg phase together with some Mg2Ni phase. It is also the case for Mg-20Ce sample. Only Bragg peaks from Mg, Mg2Ni and Mg12Ce are observed in Mg-Ni-Ce ternary alloys. It has been proven that high-

Fig. 3. Particle morphology and particle size distribution histograms.

96

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

energy ball milling as-cast Mg-based alloy for 2 h is not sufficient to obtain ultrafine nanocrystalline with grain size of several nm or amorphous alloy [39,40]. In both Mg-5Ni7Ce and Mg-10Ni2Ce alloys, however, Bragg peaks of Mg12Ce and Mg2Ni are rather weak and obviously broadened. Neither magnesium oxides nor cerium oxides are evident from the XRD patterns. The particle morphologies and size distributions of the samples after 2 h HEBM have been characterized by SEM and estimated by counting 150 particles in images with the same magnification [6,41] and the results are shown in Fig. 3. The distribution histograms and particle morphology images inserted in Fig. 3(a)-(d) indicate that all the HEBM samples present irregular refined particles and the mean particle size of each sample is ~30 mm. It suggests that the composition of the alloy has little influence on the particle size of HEBM samples. TEM technique has been applied to further characterize the microstructure details of the ball-milled samples. As shown in

Fig. 4(a), only diffused continuous diffraction rings of Mg and Mg2Ni are observed in the SAED pattern of Mg-20Ni, indicating that both Mg and Mg2Ni consist of their own nanocrystalline aggregates. The HRTEM image in Fig. 4(b) clearly shows the Mg2Ni nanocrystalline with grain size ~10 nm imbedded in Mg matrix. In Fig. 4(c), prevalent nanocrystals of Mg, Mg2Ni and Mg12Ce with grain size in the range of 3e5 nm are observed in Mg-5Ni7Ce sample. It is commonly accepted that the minimum grain size of Mg achieved by high-energy ball milling is ~20 nm [42e44]. Such fine nanocrystals observed in the present work are believed to be decomposed from the LPSO phase during HEBM process, which is also supported by the XRD results in Supplementary 1. Bragg peaks of LPSO phase around 36.1 and 36.8 [45] are observed in as-cast Mg-5Ni7Ce and Mg-10Ni2Ce alloys besides Bragg peaks of Mg, Mg2Ni and Mg12Ce. No Bragg peaks of LPSO are observed in both ball-milled Mg-5Ni7Ce and Mg-10Ni2Ce samples. It indicates that the LPSO phase grain is easy to be deformed and fractured by

Fig. 4. (a) BF image and corresponding SAED pattern of ball milled Mg-20Ni sample. (b) and (c) are HRTEM images of ball milled Mg-20Ni and Mg-5Ni7Ce samples.

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

simple ball milling treatment. This phenomenon is related to the specific deformation and fragment mechanism of LPSO phase [46]. The plastic behavior of LPSO phase is highly anisotropic. Compared with the basal, pyramid and cylinder slip modes in pure Mg, the (0001) < 11-20 > basal slip is the dominant operative deformation mode in 18R-type LPSO phase. The fracture of LPSO phase during

97

ball milling has also been observed in Mg-Ni-Y hydrogen storage alloy [47]. Although LPSO phase is unstable during high-energy ball milling process, it can make elements of Mg, Ni and Ce distribute uniformly at the atomic scale during solidification process, and can decompose into ultrafine catalytic phases after high-energy ball milling. Compared to the coarse microstructure of as-cast alloys,

Fig. 5. The first three activation cycles for HEBM samples at 300  C: (a)e(d). Kinetic curves for hydrogen absorption (e) and desorption (f) of fully activated alloys.

98

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

the refinement of the grains and the increase of grain boundaries and phase boundaries in ball milled samples are beneficial to diffusion and occupation of hydrogen atoms. 3.3. Activation properties and de-/hydrogenation kinetics It is well known that activation process is essential for Mg-based materials to achieve rapid rate and high capacities of hydrogen storage. The activation process includes the breaking of oxide/hydroxide layer and the baring of fresh metal areas on the surface, which is necessary to the hydrogen adsorption and dissociation. One common method is repeated hydrogenation and dehydrogenation at high temperature [48]. The other is the mechanical method. By ball milling, the breaking of oxide/hydroxide layer and the baring of fresh metal areas on the surface are easily achieved. Ball milling can also bring a large number of defects to significantly improve the activation performance [40]. As shown in Fig. 5(a), the capacities of hydrogen uptake for HEBM Mg-20Ni sample in first and second cycles actually reach 4.89 wt% and 5.87 wt%, respectively. In the third cycle, the hydrogen absorption capacity reaches saturation at approximately 6.01 wt%. It means that three cycles of hydrogen absorption and desorption under the present condition are enough to fully activate HEBM Mg-20Ni sample. Fig. 5(c) and (d) depict the evolution of hydrogen content during the first three activation cycles of the HEBM Mg-5Ni7Ce and Mg20Ce samples at 300  C under 3 MPa hydrogen pressure, respectively. Two discontinuous steps of hydrogen absorption are observed for both Mg-20Ce and Mg-5Ni7Ce in their first hydrogenation curves. Such stepwise hydrogenation behavior is associated

with the presence of different phases having quite different affinities to hydrogen, which undergo irreversible two-step disproportion reaction [49]. The detailed reactions of Mg12RE phase during hydrogenation have been well clarified by in situ synchrotron XRD [5,49]. In short, they can be summarized as hydrogen-induced disproportionation of Mg12RE to generate RE hydride and Mg followed by further hydrogenation of Mg and Mg2Ni (if any). An entirely different shape of absorption curve and a remarkable improvement of the absorption kinetics are observed already during the second hydrogenation. Due to the minor addition of Ce, only a small amount of Mg12Ce is presented in the Mg-10Ni2Ce sample. Thus the curve shape in the first hydrogenation in Mg-10Ni2Ce sample is similar to Mg-20Ni sample. The activation performance of the ball-milled samples is much better than that of melt-spun Mg-Ni-RE alloys [50,51], for which over 10 h are needed to achieve full activation. The superior activation properties are ascribed to the positive effect of HEBM to produce fresh crack metal surface and ultrafine nanocrystals. The absorption and desorption curves of the four samples after full activation are illustrated in Fig. 5(e) and (f), respectively. The absorption and desorption measurements are performed at 300  C under 3 MPa hydrogen pressure and under 0.002 MPa, respectively. It is worth noting that from the de-/hydrogenation curves of Mg20Ni and Mg-20Ce in Fig. 5(e) and (f), Ni is more conducive to dehydrogenation and Ce is more beneficial to hydrogenation of Mg. The absorption/desorption curves of Mg-5Ni7Ce and Mg-10Ni2Ce also confirm this point. The results coincide well with the phase transition sequence found in Ref. [42].

Fig. 6. XRD patterns of hydrogenated samples after being kept in air for ~168 h.

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

3.4. Anti-oxidation properties and corresponding mechanisms In order to test the change of hydrogen storage performance after oxygen contact, the four hydrogenated samples have been kept in air for ~168 h. The microstructural evolutions of the samples are paid special attention to explain the change of hydrogen storage properties. XRD patterns of the hydrogenated samples after being kept in air for ~168 h are given in Fig. 6(a)-(d). As illustrated in Fig. 6(a), Bragg peaks of MgH2, Mg2NiH4 and MgO are observed in Mg-20Ni sample, together with a small amount of Mg2NiH0.3. For Mg-5Ni7Ce sample and Mg-10Ni2Ce sample, Bragg peaks around 28.6 , 33.1, 47.5 and 56.3 from CeO2 are clearly observed besides Bragg peaks of MgH2, CeH2.73, Mg2NiH4, Mg2NiH0.3 and MgO. In Mg-20Ce alloy, Bragg peaks from CeO2 around 28.6 , 33.1, 47.5 and 56.3 are also observed besides MgH2, CeH2.73 and MgO. As is known to all, during hydrogenation process of Mg-Ni-RE ternary alloys, typical reactions include Mg turning into MgH2, Mg2Ni turning into Mg2NiH4 or Mg2NiH0.3 and Mg12RE turning into RE hydride and MgH2. CeO2 is probably transformed from CeH2.73 during air exposure. To further reveal the microstructural details of the hydrogenated samples before and after air exposure, standard TEM techniques such as dark-field (DF) imaging and HRTEM have been applied in this work. The BF images and corresponding SAED patterns of hydrogenated Mg-5Ni7Ce and Mg-20Ce samples before air exposure are shown in Supplementary 2. Combined with the XRD results, it can be concluded that nano-sized secondary phases of Mg2NiH4 and CeH2.73 with particle size in the range of several nm to 100 nm are embedded in the matrix in Mg-5Ni7Ce sample.

99

Uniformly distributed CeH2.73 with irregular shape and nonuniform size are also clearly observed in as-activated Mg-20Ce sample. Fig. 7 displays the microstructural details of hydrogenated Mg5Ni7Ce sample after air exposure. Secondary phase with irregular shape and nonuniform size is clearly observed in the BF image in Fig. 7(a). Appreciable boundary within the particle in the enclosed area in Fig. 7(a) can also be observed. The morphology of the particle at higher magnification is displayed in Fig. 7(b). Dotted line outlines the phase boundary roughly. The HRTEM image near the boundary in Fig. 7(b) is shown in Fig. 7(c). Although CeH2.73 (Fm3m, a ¼ 0.553 nm, JCPDS 89-3694) and CeO2 (Fm-3m, a ¼ 0.541 nm, JCPDS 81-0792) have extremely close lattice constants, the fast Fourier transformation (FFT) diffractograms in Fig. 7(d) and (e) clearly discern CeH2.73 and CeO2. The particle shown in Fig. 7(b) is residual CeH2.73 which has not completely transformed into CeO2. As illustrated in the DF image of CeO2 in Fig. 7(f), nanometer-sized CeO2 particles with particle size in the range of several nanometers to ~60 nm disperse on the surface of the matrix. Considering XRD and TEM results, it is reasonable to conclude that CeH2.73 (formation enthalpy: ~204.8 kJ mol1 [52]) reacts with oxygen to generate CeO2 (formation enthalpy: 1088.7 kJ mol1 [52]) thermodynamically spontaneously during air exposure. To evaluate the anti-oxidation properties of Mg-based hydrogen storage alloys with addition of Ce, the dehydrogenation performance of the four activated samples after being kept in air for ~168 h has been measured by DSC. For Mg-20Ni sample and Mg10Ni2Ce sample as shown from Fig. 8(a) to (b), the peak temperature of air-exposed sample at each heating rate is much higher than that of the as-activated one, suggesting oxidation of the samples

Fig. 7. (a) BF image of hydrogenated Mg-5Ni7Ce sample after air exposure, (b) enlargement of enclosed region in (a), (c) HRTEM image of the area marked by square in (b), (dee) corresponding FFT diffractograms of the enclosed regions in (c) and (f) the DF image of CeO2.

100

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

Fig. 8. (a)e(d) DSC curves at various heating rates of as-activated and air-exposed samples and (e) the plot of ln (b/T2p) versus 1000/Tp of the samples in (a)e(d). (f) Isothermal desorption kinetic curves of as-activated and air exposed Mg-20Ni and Mg-20Ce samples at 300  C.

brings serious deterioration of hydrogen desorption properties. For Mg-5Ni7Ce sample as shown in Fig. 8(c), the peak temperature of air-exposed sample at each heating rate is slightly higher than that of the as-activated sample, indicating a better oxidation resistance than Mg-20Ni sample and Mg-10Ni2Ce sample. As shown in Fig. 8(d), for Mg-20Ce sample, however, the peak temperature of air-exposed sample at each heating rate is even lower than that of as-activated sample, indicating that the generated CeO2, together with remaining CeH2.73 can achieve superior catalytic actions, even

though the sample has been in contact with air for a long time. The desorption activation energy based on DSC results is calculated using Kissinger equation [53]:

.   d ln b Tp2 E    ¼ A R d 1 Tp

(1)

where Tp is the peak temperature, b is the heating rate, EA is the

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

activation energy and R is the gas constant. The plot of lnb/T2p versus 1000/Tp is shown in Fig. 8(e). The activation energy of each sample is obtained from the slope of the fitted line. For Mg-20Ce sample, the desorption activation energy of air-exposed sample (~82 kJ mol1) is even slightly lower than that of the as-activated one (~86 kJ mol1), suggesting super oxidation resistance. For Mg-5Ni7Ce sample, the desorption activation energy of air-exposed sample (~96 kJ mol1) is slightly higher than that of as-activated sample (~86 kJ mol1). But for Mg-20Ni sample and Mg-10Ni2Ce sample, the desorption activation energy increases dramatically after air exposure (as-activated Mg-20Ni: ~84 kJ mol1, air-exposed Mg-20Ni: ~140 kJ mol1; as-activated Mg-10Ni2Ce: ~101 kJ mol1, air-exposed Mg-10Ni2Ce: ~127 kJ mol1). It also states clearly that Ni does not account for the oxidation resistance of Mg-based hydrogen storage alloys. The isothermal desorption kinetic curves of fully activated Mg-20Ce and Mg-20Ni samples at 300  C before and after air exposure are shown in Fig. 8(f). Obvious deterioration of desorption kinetics is observed for Mg-20Ni sample after air exposure. It only desorbs ~1.5 wt% H2 at 300  C within 10 min. But the desorption kinetics even improve slightly for Mg-20Ce sample after air exposure, which is consistent with the DSC results. Understanding from the de-/hydrogenation process, the hydrogen storage properties of Mg-based alloys depend on the characteristics including the surface and internal features. The surface properties, such as microstructure and defects, always determine the adsorption/desorption of hydrogen molecules and the dissociation/recombination of hydrogen atoms [54,55]. The process of hydrogen diffusion and occupation is associated with the internal microstructure [56]. Both hydrogenation and dehydrogenation start from the surface. Once the sample contacts with air or moisture, MgO will spread over the Mg surface, making hydrogen absorption/desorption and dissociation/recombination rather difficult [22]. The kinetics of the MgO growing over Mg or MgH2 is well known [57,58]. For the MgO layer growing over Mg, it consists of oxygen chemisorption, oxide nucleation, island growth and coalescence of the oxide islands. It has been found that the oxide layer rapidly grows up to some critical thickness and then stops. McIntyre et al. [58] have suggested that an oxide thickness of about 2.2 nm is formed even at 10 s of ambient exposure and that the oxide thickness continues increasing to 4.5 nm when the time reaches up to 10 months. Although the formation enthalpy of ceria is more negative than that of MgO (601.6 kJ mol1), the addition of Ce cannot completely prevent the oxidation of Mg, which can be seen from the Bragg peaks of MgO in Fig. 6. However, due to a more negative enthalpy, CeO2 forms prior to or simultaneously with MgO during air exposure, preventing forming a continuous and dense magnesium oxide layer on the surface. Thus CeO2, which plays important catalytic role to promote dissociation and recombination of hydrogen atoms [59e61], can disperse on the surface of particles after air exposure together with MgO. The newly introduced interface between CeH2.73 and CeO2 is also conducive to the surface penetration and diffusion of hydrogen atoms [61]. Although the dissociation and recombination of hydrogen require a fairly high energy on the Mg surface [22] and MgO layer on Mg usually limits the diffusion of hydrogen atoms, the fine and evenly distributed CeH2.73/CeO2 composite plays an important catalytic role for de-/hydrogenation in Mg, which can effectively eliminate the adverse effects of MgO by providing hydrogen penetration routs and active nucleation sites for the formation of MgH2. The symbiotic CeH2.73/CeO2 is in fact a kind of “oxygen introduced” catalyst. The results reveal to us that oxygen can be effectively utilized to improve the hydrogen storage properties of Mg-based alloys by reasonable alloying and effective processing, rather than only deteriorates the hydrogenation performance.

101

4. Conclusions In summary, Mg-Ni(Ce) binary eutectic alloy and Mg-Ni-Ce ternary alloys in the Mg-rich corner are successfully manufactured and further refined by HEBM to obtain nanocrystalline. A new 18R-type LPSO phase within Mg12Ce matrix is observed in the eutectic region in as-cast Mg-Ni-Ce ternary alloy. The 18R-type LPSO structure is a variant of Mg12Ce rather than Mg and the orientation relationship between the Mg12Ce matrix and the 18R phase is (002)Mg12Ce//(0018)18R. HEBM produces some Mg, Mg2Ni and Mg12Ce nanocrystals with grain size in the range of 3e5 nm, which probably decompose from the LPSO phase during HEBM process. CeH2.73 is formed after hydrogenation and spontaneously transforms into CeO2 during air exposure. The symbiotic CeH2.73/ CeO2 composite is in fact a kind of “oxygen introduced” catalyst which effectively attenuates the adverse effect of Mg oxidation. For Mg-20Ce sample, the desorption activation energy after air exposure is even lower than that of the as-activated one. The improved anti-oxidation properties are ascribed to hydrogen penetration routs and active nucleation sites for the formation of Mg provided by CeH2.73/CeO2. Acknowledgments This work is financially supported by the Research Fund of the State Key Laboratory of Solidification Processing (NWPU), China (Grant No. 95-QZ-2014). The 111 project (No. B08040) is also acknowledged. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2016.11.025. References [1] A.A.C. Asselli, S.F. Santos, J. Huot, J. Alloys Compd. 687 (2016) 586e594. [2] Y. Jia, C. Sun, S. Shen, J. Zou, S.S. Mao, X. Yao, Renew. Sustain. Energy Rev. 44 (2015) 289e303. [3] H. Wang, H.J. Lin, W.T. Cai, L.Z. Ouyang, M. Zhu, J. Alloys Compd. 658 (2016) 280e300. [4] L.Z. Ouyang, X.S. Yang, M. Zhu, J.W. Liu, H.W. Dong, D.L. Sun, J. Zou, X.D. Yao, J. Phys. Chem. C 118 (2014) 7808e7820. [5] A.A. Poletaev, R.V. Denys, J.P. Maehlen, J.K. Solberg, B.P. Tarasov, V.A. Yartys, Int. J. Hydrogen Energy 37 (2012) 3548e3557. [6] X. Hou, R. Hu, T. Zhang, H. Kou, J. Li, J. Power Sources 306 (2016) 437e447. [7] H. Fu, W. Wu, Y. Dou, B. Liu, H. Li, Q. Peng, J. Power Sources 320 (2016) 212e221. [8] S. Giusepponi, M. Celino, Int. J. Hydrogen Energy 40 (2015) 9326e9334. [9] H.-J. Lin, C. Zhang, H. Wang, L. Ouyang, Y. Zhu, L. Li, W. Wang, M. Zhu, J. Alloys Compd. 685 (2016) 272e277. [10] J. Liu, Y. Li, D. Han, S. Yang, X. Chen, L. Zhang, S. Han, J. Power Sources 300 (2015) 77e86. [11] D. Xu, E.-H. Han, Y. Xu, Prog. Nat. Sci. Mater. Int. 26 (2016) 117e128. [12] K. Ishikawa, T. Kawasaki, Y. Yamada, Int. J. Hydrogen Energy 40 (2015) 13014e13021. [13] Q.A. Zhang, D.D. Liu, Q.Q. Wang, F. Fang, D.L. Sun, L.Z. Ouyang, M. Zhu, Scr. Mater. 65 (2011) 233e236. [14] J.E. Saal, C. Wolverton, Acta Mater. 68 (2014) 325e338. [15] Y. Zhang, B. Li, H. Ren, Z. Yuan, T. Yang, Y. Qi, Int. J. Hydrogen Energy 41 (2016) 12205e12213. [16] Y. Wu, M.V. Lototsky, J.K. Solberg, V.A. Yartys, W. Han, S.X. Zhou, J. Alloys Compd. 477 (2009) 262e266. [17] G.F. Lima, M.R.M. Triques, C.S. Kiminami, W.J. Botta, A.M. Jorge, J. Alloys Compd. 586 (2014) S405eS408. [18] L.A. Bendersky, C. Chiu, V.M. Skripnyuk, E. Rabkin, Int. J. Hydrogen Energy 36 (2011) 5388e5399. [19] L.Z. Ouyang, Z.J. Cao, H. Wang, J.W. Liu, D.L. Sun, Q.A. Zhang, M. Zhu, J. Alloys Compd. 586 (2014) 113e117. [20] L.Z. Ouyang, S.Y. Ye, H.W. Dong, M. Zhu, Appl. Phys. Lett. 90 (2007) 021917. [21] L. Ouyang, Z. Cao, H. Wang, R. Hu, M. Zhu, J. Alloys Compd. 691 (2017) 422e435. [22] T. Jensen, A. Andreasen, T. Vegge, J. Andreasen, K. Stahl, A. Pedersen, M. Nielsen, A. Molenbroek, Flemmingbesenbacher, Int. J. Hydrogen Energy 31

102

L. Xie et al. / Journal of Power Sources 338 (2017) 91e102

(2006) 2052e2062. [23] K.J. Jeon, H.R. Moon, A.M. Ruminski, B. Jiang, C. Kisielowski, R. Bardhan, J.J. Urban, Nat. Mater. 10 (2011) 286e290. [24] S. Long, J. Zou, Y. Liu, X. Zeng, W. Ding, J. Alloys Compd. 580 (1) (2013) S167eS170. [25] J. Zou, X. Zeng, Y. Ying, X. Chen, H. Guo, S. Zhou, W. Ding, Int. J. Hydrogen Energy 38 (2013) 2337e2346. [26] Y. Lv, B. Zhang, Y. Wu, J. Alloys Compd. 641 (2015) 176e180. [27] Y. Fan, X. Peng, T. Su, H. Bala, B. Liu, J. Alloys Compd. 596 (2014) 113e117. [28] K.-B. Wu, Q. Luo, S.-L. Chen, Q.-F. Gu, K.-C. Chou, X.-L. Wang, Q. Li, Int. J. Hydrogen Energy 41 (2016) 1725e1735. [29] Y.M. Zhu, A.J. Morton, J.F. Nie, Acta Mater. 58 (2010) 2936e2947. [30] H. Yokobayashi, K. Kishida, H. Inui, M. Yamasaki, Y. Kawamura, Acta Mater. 59 (2011) 7287e7299. [31] D. Egusa, E. Abe, Acta Mater. 60 (2012) 166e178. [32] T. Itoi, T. Seimiya, Y. Kawamura, M. Hirohashi, Scr. Mater. 51 (2004) 107e111. [33] T. Hongo, K. Edalati, M. Arita, J. Matsuda, E. Akiba, Z. Horita, Acta Mater. 92 (2015) 46e54. [34] Y. Kawamura, M. Yamasaki, Mater. Trans. 48 (2007) 2986e2992. [35] Y. Kawamura, T. Kasahara, S. Izumi, M. Yamasaki, Scr. Mater. 55 (2006) 453e456. [36] X. Zhang, D. Kevorkov, M. Pekguleryuz, J. Alloys Compd. 475 (2009) 361e367. [37] E. Abe, Y. Kawamura, K. Hayashi, A. Inoue, Acta Mater. 50 (2002) 3845e3857. [38] M. Yamasaki, M. Sasaki, M. Nishijima, K. Hiraga, Y. Kawamura, Acta Mater. 55 (2007) 6798e6805. [39] Y. Wu, W. Han, S.X. Zhou, M.V. Lototsky, J.K. Solberg, V.A. Yartys, J. Alloys Compd. 466 (2008) 176e181. [40] H. Kou, X. Hou, T. Zhang, R. Hu, J. Li, X. Xue, Mater. Charact. 80 (2013) 21e27. [41] X. Tan, B. Zahiri, C.M.B. Holt, A. Kubis, D. Mitlin, Acta Mater. 60 (2012) 5646e5661. [42] R.V. Denys, A.B. Riabov, J.P. Maehlen, M.V. Lototsky, J.K. Solberg, V.A. Yartys,

Acta Mater. 57 (2009) 3989e4000. [43] R.R. Shahi, T.P. Yadav, M.A. Shaz, O.N. Srivastava, Int. J. Hydrogen Energy 33 (2008) 6188e6194. [44] G. Liang, S. Boily, J. Huot, A. Van Neste, R. Schulz, J. Alloys Compd. 267 (1998) 302e306. [45] H.J. Lin, L.Z. Ouyang, H. Wang, D.Q. Zhao, W.H. Wang, D.L. Sun, M. Zhu, Int. J. Hydrogen Energy 37 (2012) 14329e14335. [46] K. Hagihara, N. Yokotani, Y. Umakoshi, Intermetallics 18 (2010) 267e276. [47] J.W. Liu, C.C. Zou, H. Wang, L.Z. Ouyang, M. Zhu, Int. J. Hydrogen Energy 38 (2013) 10438e10445. € ntzsch, T. Riedl, T. Gemming, T. Weißga €rber, B. Kieback, [48] S. Kalinichenka, L. Ro Int. J. Hydrogen Energy 36 (2011) 1592e1600. [49] R.V. Denys, A.A. Poletaev, J.P. Maehlen, J.K. Solberg, B.P. Tarasov, V.A. Yartys, Int. J. Hydrogen Energy 37 (2012) 5710e5722. [50] W. Song, J. Li, T. Zhang, H. Kou, X. Xue, J. Power Sources 245 (2014) 808e815. €ntzsch, B. Kieback, Int. J. Hydrogen Energy 34 (2009) [51] S. Kalinichenka, L. Ro 7749e7755. [52] C.X. Shang, Z.X. Guo, Int. J. Hydrogen Energy 32 (2007) 2920e2925. [53] H.E. Kissinger, Anal. Chem. 29 (1957) 1702e1706. [54] A.V. Kuklin, A.A. Kuzubov, P.O. Krasnov, A.O. Lykhin, L.V. Tikhonova, J. Alloys Compd. 609 (2014) 93e99. , Int. J. Hydrogen Energy 34 (2009) 1922e1930. [55] M. Pozzo, D. Alfe [56] D. Pukazhselvan, V. Kumar, S.K. Singh, Nano Energy 1 (2012) 566e589. [57] T. Do, S.J. Splinter, C. Chen, N.S. McIntyre, Surf. Sci. 387 (1997) 192e198. [58] N.S. McIntyre, C. Chen, Corros. Sci. 40 (1998) 1697e1709.  Raskovi [59] J. Gulicovski, Z. c-Lovre, S. Kurko, R. Vujasin, Z. Jovanovi c, L. Matovi c, J. Grbovi c Novakovi c, Ceram. Int. 38 (2012) 1181e1186. [60] R.K. Singh, T. Sadhasivam, G.I. Sheeja, P. Singh, O.N. Srivastava, Int. J. Hydrogen Energy 38 (2013) 6221e6225. [61] H.-J. Lin, J.-J. Tang, Q. Yu, H. Wang, L.-Z. Ouyang, Y.-J. Zhao, J.-W. Liu, W.H. Wang, M. Zhu, Nano Energy 9 (2014) 80e87.