Zongying Hn ,Yyun Wu ,Ho Yu ,Shixue Zhou,b,?

aCollege of Chemical and Biological Engineering,Shandong University of Science and Technology,Qingdao 266590,China

b State Key Laboratory of Mining Disaster Prevention and Control Co-founded by Shandong Province and the Ministry of Science and Technology,Shandong University of Science and Technology,Qingdao 266590,China

Abstract Density functional theory (DFT) calculations have been performed to investigate the hydrogen dissociation and diffusion on Mg (0001)surface with Ni incorporating at various locations.The results show that Ni atom is preferentially located inside Mg matrix rather than in/over the topmost surface.Further calculations reveal that Ni atom locating in/over the topmost Mg (0001) surface exhibits excellent catalytic effect on hydrogen dissociation with an energy barrier of less than 0.05 eV.In these cases,the rate-limiting step has been converted from hydrogen dissociation to surface diffusion.In contrast,Ni doping inside Mg bulk not only does little help to hydrogen dissociation but also exhibits detrimental effect on hydrogen diffusion.Therefore,it is crucial to stabilize the Ni atom on the surface or in the topmost layer of Mg (0001)surface to maintain its catalytic effect.For all the case of Ni-incorporated Mg (0001) surfaces,the hydrogen atom prefers firstl immigrate along the surface and then penetrate into the bulk.It is expected that the theoretical finding in the present study could offer fundamental guidance to future designing on efficien Mg-based hydrogen storage materials.

Keywords: Hydrogen storage;Hydrogen dissociation;Ni incorporation;Hydrogen diffusion;Mg.

1.Introduction

With the highest hydrogen capacity among all known reversible metal hydrides,magnesium has been reputed as one of the most promising media for hydrogen storage [1].However,hitherto magnesium remains hardly any practical application for hydrogen storage,notwithstanding many potential uses [2-4].One of the prime issues is that pure magnesium exhibits remarkable poor kinetics.The high dissociation barrier of hydrogen on Mg surface usually takes the blame as the rate-limiting step of hydrogen uptake [5].To enhance hydrogen sorption kinetics of Mg,the commonly-used tuning strategies mainly include nano-sizing,alloying and introducing additives[6-8].It has been generally recognized that transition metals exhibited the most prominent catalytic effect on the hydrogen dissociation [9].However,the introduction of transition metals may also hinder hydrogen diffusion induced by their strong adhesion to hydrogen atom [10].Among various additives,Ni has been proved to be one of the most efficien alternatives due to its high catalytic effect to hydrogen dissociation and low adhesion force to hydrogen diffusion[11].

Fig.1.Top and side views of Mg (0001) surface models used in this study.(a) Pure surface;(b) Ni loading at surface FCC site;(c) Ni loading at surface HCP site;(d) Ni loading at surface Bridge site;(e) Ni doping at 1st layer;(f) Ni doping at 2nd layer;(g) Ni doping at 3rd layer.The Mg and Ni atoms are represented by green and blue colors,respectively.

In previous studies,the influenc of Ni addition to the performance of Mg-based hydrogen storage materials has been systematically investigated,mainly from the perspective of addition amount [12-14],dispersion degree [15],and particle size [16].However,the loaded Ni on Mg surface is usually unstable,and may immigrate into Mg matrix to form Mg-Ni intermetallic phase or Mg2Ni alloy [17,18].Therefore,it is of critical significanc to figur out the positional effect of Ni incorporation on the hydrogen storage performance of Mg.Based on density functional theory calculations (DFT),Kuklin et al.have investigated the formation of Ni-doping in different locations on Mg (0001) surface and pointed out that Ni atoms prefer to substitute for Mg atoms,but their influenc on hydrogen dissociation and diffusion has not been further discussed [19].In the study of Liu et al.,only the configura tion of Ni atom loading on FCC site of the Mg(0001)surface was considered for accelerating Mg hydriding without checking the site preference of Ni [20].On the contrary,Pozzo et al.just investigated the influenc of Ni substitution in the firs layer of Mg (0001) surface on hydrogen dissociation and diffusion,without considering the influenc of Ni doping in deep layer [21].Chen et al.investigated the hydrogen adsorption,dissociation and diffusion on Mg (0001) surface with Ni doping in the topmost two layers,without concerning the case of Ni loading on surface [22,23].Besides,most of previous studies just have investigated the influenc of Ni incorporation on hydrogen surface diffusion [10,21,23-25],and its impact on bulk diffusion has rarely been explored.Thus,the positional effect of Ni on hydrogen dissociation and diffusion is still not clear.To the best of our knowledge,no research has systematically addressed this issue from neither experimental nor theoretical perspective.Herein,we provide a systematically comparative study of hydrogen dissociation and diffusion on Mg (0001) surface with Ni incorporating in different sites based on DFT calculations,aiming to provide theoretical foundation for designing efficien hydrogen storage materials.

2.Computational and experimental details

2.1.Computational models and settings

All DFT calculations were implemented using DMol3 package in Materials Studio with XC functional of GGA-PBE[26].A six-layered Mg (0001) slab was constructed with ap(3×3) supercell and 15 °A vacuum space.The validation of our bulk and surface models can be seen in our previous study[27].To construct reliable surface models,we firstl examined all the possible incorporating sites of Ni atom on Mg(0001) slab.For Ni atom loading on Mg (0001) surface,four different sites have been checked,including the surface Top,Bridge,FCC,and HCP sites.According to our calculations,we found that Ni atom cannot be stably adsorbed on surface Top site,but can stably locate at surface Bridge,HCP and FCC sites.For Ni atom doping on Mg (0001) surface,both the cases of placing the Ni atom at the interstitial position and substituting one Mg atom have been considered.Unlike the substitution case,the surface structure of Ni atom fill ing at the interstitial position was distorted significantl after optimization,causing instability in the overall arrangement of the Mg atoms.Based on above comprehensive examination,six incorporating positions of Ni atom on Mg (0001)surface have been taken into consideration in this study,including Ni loading at surface FCC site,Ni loading at surface HCP site,Ni loading at surface Bridge site,Ni doping at 1st layer,Ni doping at 2nd layer,and Ni doping at 3rd layer,respectively.The top and side views of all the surface models used in this study are schematically shown in Fig.1.Subsequent calculations considered spin-polarized electrons due to the magnetic moment of Ni.All surface calculations were performed with a 6×6×1 k-point mesh in the Monkhorste Pack scheme [28].The topmost three layers and the adsorbents were allowed to relax,while the positions of the rest of atoms were fi ed,aiming to reduce the calculation cost.We have carefully checked the influenc of atom layer constraint,and negligible difference (＜0.01 eV) has been found for hydrogen diffusion compared to the case of no atom layer fix ation.The geometry optimization calculations were achieved with a convergence tolerance in maximum energy change of 1.0×10?5Ha,maximum force of 0.002 Ha °A?1and maximum displacement of 0.005 °A,respectively.The complete LST/QST method was employed to determine the transition states (TS) and energy barriers for hydrogen dissociation and diffusion [29].The convergence tolerance for TS calculations was set to a RMS of 0.01 Ha °A?1.

Fig.2.Charge density (a) and deformation charge density (b) maps of Ni-incorporated Mg (0001) surfaces.In deformation charge density maps,positive regions indicate the formation of bonds while negative regions indicate electron loss.

The reaction energy barrier was calculated according to the equations described in previous studies [2,30].The formation energy (Eformation) for Ni loading and doping on Mg (0001)surface is define as following [25]:

whereENi/Mg(0001)is the energy of Ni loaded or doped Mg surface,EMgis the energy of Mg in the bulk structure,ENiis the energy of isolated Ni atom obtained in a 10×10×10 °A3box,EMg(0001)is the energy of a pure Mg surface,andnis 0 for Ni loading on Mg surface whilenis 1 for Ni doping on Mg surface.

2.2.Experimental details

In order to verify the theoretical findings Mg/Ni composite has been prepared by blending Mg with 5wt.% Ni in planetary ball mill (ND8 model,Nanda Tianzun Instrument Company,China) under 30 Hz for 2 h.The hydrogen sorption and cycling performance of the Mg/Ni composite were conducted in an automated Sievert-type volumetric apparatus (Setaram PCT-Pro 2000) at 573 K using 99.999%purity hydrogen.In each sorption cycle,the isothermal absorption kinetic was measured under 2.0 MPa H2for 15 min and the isothermal desorption kinetic was measured under 0.05 MPa H2for 30 min.The phase structure of the asprepared samples was determined by Rigaku D/Max-rB Xray diffractometer (XRD) using a Cu Kαradiation source.The morphology observation of the samples was performed on a FEI APREO high-resolution Scanning Electron Microscope (SEM) equipped with a Bruker Nano GmbH energy dispersion spectrometry (EDS).

3.Results and discussion

3.1.Formation energy and surface properties

In order to obtain the site preference of Ni incorporation on Mg (0001) surface,we have calculated the formation energy of the six Ni incorporation configuration concerned in this study,and the corresponding values are listed in Table 1.The negative values indicate that it is thermodynamically favorable for Ni atom incorporation at the listed six positions.Our calculation results indicate that the most stable configuratio of Ni loading on Mg (0001) surface is Ni adsorption at surface Bridge site,which is consistent with the findin in literature [19].For Ni atom loading on the threefold hollow site,the HCP hollow site is the preferred site compared to FCC site,which is consistent with the findin of Fe loading on Mg (0001) surface [31].It can be clearly seen that the formation energies of Ni substitution are about 1 eV larger than that of Ni adsorption on surface,revealing that Ni prefers to substitute one Mg atom at the atomic layer rather than load on the surface.This is consistent with the findin reported by Wang et al.[25].The formation energies of Ni substitution at the top and second layers are determined to be?4.41 eV and ?4.53 eV respectively,which are very close to the values in previous study [9].The formation energy of Ni substitution at the third layer further increases to ?4.55 eV.This indicates that Ni is more stable in the deeper layer than in the firs layer or on the surface,which is in line with the findin in literatures [19,22,32].This can also be reflecte by the change in surface energy when Ni incorporates at different positions.The surface energy of pure Mg (0001) surface is calculated to be 0.339 eV per surface atom,which is consistent with the value 0.337 per surface atom reported in previous study [33].The surface energies of Ni-loaded Mg (0001)surface are 0.404 eV,0.403 eV and 0.402 eV per surface atom when Ni atom adsorbs at surface FCC,HCP and Bridge site,respectively.The surface energy decreases from 0.365 eV to 0.357 eV per surface atom when the doping position of Ni changes from firs layer to third layer.The decreasing surface energy also reveals the trend of Ni preference in the deeper layer.The site preference is probably attributed to the difference in atomic coordination number.When adsorbed on Mg(0001) surface,Ni atom strongly bonds with adjacent three or four Mg atoms with obvious charge overlapping,which can be seen in the total charge density maps (Fig.2a) and deformation charge density maps (Fig.2b).For Ni substitution at the firs layer,the corresponding atomic coordination number increases to 9 Mg atoms.When Ni atom substitutes one Mg atom at deeper layers,12 Mg atoms coordinate with Ni atom with all-round charge overlapping (Fig.2),leading to more stable configuratio of Ni atom.

Table 1 The formation energy and surface energy of the Ni incorporated Mg (0001)surfaces.

Fig.3.Energy profile of hydrogen dissociation on Mg (0001) surfaces.(a) Pure surface;(b) Ni loading at surface FCC site;(c) Ni loading at surface HCP site;(d) Ni loading at surface Bridge site;(e) Ni doping at 1st layer;(f) Ni doping at 2nd layer;(g) Ni doping at 3rd layer.The Mg,Ni and H atoms are represented by green,blue and white colors,respectively.

Fig.4.The d-orbital DOS of the whole slab (a) and the topmost surface layer (b) of the Mg (0001) surface.

3.2.Hydrogen dissociation

The most anticipated role of the incorporated Ni should be to overcome the high hydrogen dissociation barrier on pure Mg surface.To reveal it,a series of transition state search calculations have been performed to determine the energy barrier of hydrogen dissociation on Ni-incorporated Mg surface(Fig.3).Hydrogen dissociation on pure Mg (0001) surface has been extensively investigated in previous study [27],and we show it in Fig.3a as the benchmark for comparison.It can be seen that the benchmarked energy barrier of hydrogen dissociation on pure Mg (0001) surface is calculated to be 1.44 eV,which is very close to the theoretical values obtained by same calculation tools (1.38 eV [34] and 1.42 eV [5]).On the Ni-loaded Mg (0001) surface,the hydrogen dissociation barrier has been significantl reduced to an extremely low level:0.04 eV on Mg (0001) surface with Ni atom loaded on FCC site (Fig.3b),0.05 eV on Mg (0001) surface with Ni atom loaded on HCP site (Fig.3c),and 0.05 eV on Mg(0001) surface with Ni atom loaded on Bridge site (Fig.3d),respectively.We note that these values are very close to that of hydrogen dissociation on pure Ni surface (0.03 eV)[35],indicating single Ni atom still exhibit excellent catalytic effect.For hydrogen dissociation on Ni-doped Mg(0001) surface,the corresponding energy barriers show high dependence to the position of Ni dopant.In the case of Ni substitution at the topmost Mg layer,the hydrogen dissociation barrier is determined to be 0.05 eV (Fig.3e),which is still comparable to that of Ni loaded on Mg surface.This low energy barrier is in line with these values reported in previous studies (0.06-0.14 eV) [9,10,21,23,25].The minor differences between our calculated value and the literature could be due to the difference of calculation parameter settings and H positions adopted in the calculations.For example,the two H atoms after dissociation were located at two adjacent HCP sites in the study of Pozzo [10,21] while they were located at a HCP site and a FCC site symmetrically distributed on two sides of the Ni atom.Differently,the fina state of two H atoms located at two adjacent FCC sites was adopted in this work.Nevertheless,all cases show that Ni atom doped at firs layer can still catalyze hydrogen dissociation.However,for Ni substitution at deeper layer,the catalytic effect of Ni to hydrogen dissociation disappears since the corresponding hydrogen dissociation barrier has surged to the pure surface level:1.42 eV and 1.40 eV when Ni doped at the second(Fig.3f) and third (Fig.3g) layer,respectively.In the study of Chen et al.,similar findin was reported that the hydrogen dissociation barrier on Mg surface with Ni doping at second layer was only reduced by 0.021 eV relative to that on pure Mg surface [23].This indicates that the surface Mg atom domains the catalytic effect for hydrogen dissociation rather than the underlying Ni atom.

In order to clarify the underlying reason for the locationdependent effect,we investigated the electronic properties of Mg (0001) surfaces based on density of state (DOS) analysis.Fig.4 shows thed-orbital DOS of the whole slab and the topmost surface layer of the Mg (0001) surface,respectively.It is generally recognized that thed-band center with respect to the Fermi energy can represent the catalytic reactivity [36].According to the equation described in literature [25],thedband center of the Mg (0001) surface has been calculated andlisted in Table 2.It can be clearly seen that pure Mg (0001)surface has nod-orbital electrons available for donation,and therefore it exhibits less reactivity to hydrogen dissociation.Although the introduced Ni atom containsd-band,the Mg(0001) surface still presents different reactivity to hydrogen dissociation varying by the location of Ni atom.Obviously,only the exposed surface containingd-orbital electrons closed to Fermi energy (Ni loading on the surface or substituting in the firs layer) exhibits catalytic effect on hydrogen dissociation,and thesed-orbital-containing underlying layers do negligible help (Ni substituting in the second/third layer).This can be more intuitively reflecte in the projected DOS for H2dissociating on pure and Ni-incorporated Mg (0001)surface as shown in Fig.5.At the transition state of H2dissociation on pure Mg (0001) surface,clear interaction can be seen between the Hsorbital and the Mgsand Mgporbitals(Fig.5a).On the Mg (0001) surface with Ni atom loaded at surface or doped in the firs layer,there are apparent electron overlapping between Hsorbital and Nidorbital (Fig.5b-e),which is believed to be the key reason for the catalytic effect on hydrogen dissociation.When Ni atom doping in the second or third layer,the firs Mg layer blocks the interaction between Hsorbital and Nidorbital (Fig.5f and g),therefore suppressing the catalytic effect on hydrogen dissociation.

Table 2 The d-band center (dcenter) of Ni incorporated Mg (0001) surfaces and the corresponding calculated energy barrier of hydrogen dissociation.

3.3.Hydrogen diffusion

To systematically investigate the effect of Ni on hydrogen diffusion,both the energy barriers for hydrogen immigration along and inward the Mg (0001) surface have been calculated.In this study,surface diffusion path is define as from one stable site to the adjacent one,and inward diffusion path follows the FCC channel(Fig.S1),which is recognized as the most potential path [37].Fig.6 shows the energy profile of hydrogen diffusion along Mg (0001) surfaces.For hydrogen diffusion along pure Mg (0001) surface,the energy barrier is determined to be 0.16 eV from FCC site to HCP site,with a reverse energy barrier of 0.19 eV (Fig.6a),which are within the range of 0.1-0.2 eV reported in literature [38].These low values indicate that surface diffusion on pure Mg (0001)surface is not the rate-limiting step of hydriding.However,when Ni atom loading on the surface or substituting in the firs layer,it needs to overcome high barriers (0.60-0.91 eV)for H atom to get rid of the adhesion of Ni atom (Fig.6b-e).Take the Mg (0001) surface with Ni atom doping at firs layer for instance,the energy barrier of H atom getting away from Ni atom is 0.60 eV,which is slightly higher than that reported by Banerjee et al.(0.45 eV [9]) and Chen et al.(0.433 eV [23]).This is mainly due to that the calculated hydrogen diffusion barriers were determined when two H atoms were co-attached with Ni atom.Nevertheless,all cases indicate that Ni atom exhibits strong adhesion to H atom.This is mainly attributed to that a strong chemical bond has been formed between Ni atom and H atom,which can be evidenced by the deformation charge density maps and projected DOS profile shown in Fig.7.On pure Mg (0001)surface,the adsorbed H atom and the surrounding Mg atoms form three equivalent Mg-H bonds with binding energy of 3.35 eV (Fig.7a).This means that the average bond energy of Mg-H bond is about 1.12 eV.In contrast,it can be seen from Fig.7b that the adsorbed H atom and the surrounding Mg and Ni atoms form two equivalent Mg-H bonds and one Ni-H bond,with a binding energy of 3.69 eV.The DOS profil shown in Fig.7d clearly indicates the formation of the Ni-H bond with significan overlapping of Ni all orbital and Hsorbital.Assuming that the bond energy of Mg-H bond remains unchanged,the bond energy of Ni-H bond can be approximately estimated to be 1.46 eV,which is 0.34 eV higher than the Mg-H bond energy.This value increment is comparable to the corresponding increment of hydrogen diffusion energy barrier (0.41 eV) along the surface.The difference in bond energy between Ni-H bond and Mg-H bond can be well supported from the Mulliken population analysis.For H atom adsorption on pure Mg (0001) surface(Fig.7a),the Mg-H bond population is determined to be only 0.05,suggesting that the covalent bond between Mg and H atom is fairly weak.For H atom adsorption on Mg (0001) surface with Ni atom doping at topmost layer(Fig.7b),the Ni-H bond population is determined to be as high as 0.65.This indicates more charge overlapping has been taken place in the formation of Ni-H bond than Mg-H bond,which compares well with the DOS profile shown in Fig.7c and d.Similar findin has been reported for O atom adsorption on La-doped Mg (0001) surface [39].Referring to the bond energy in literature,the average Ni-H bond energy is reported as 2.58 eV,which is significantl higher than that of Mg-H bond (1.37 eV) [40].It is easy to understand that it is more difficul to break the stronger bond,and therefore a relative higher energy barrier is needed to overcome.Dissimilarly,Ni atom substituting at second or third layer has negligible influenc to the surface diffusion of H atom (Fig.6f and g).

Fig.5.Projected DOS for H2 dissociating at transition state on pure and Ni-incorporated Mg (0001) surfaces.(a) Pure surface;(b) Ni loading at surface FCC site;(c) Ni loading at surface HCP site;(d) Ni loading at surface Bridge site;(e) Ni doping at 1st layer;(f) Ni doping at 2nd layer;(g) Ni doping at 3rd layer.

Fig.6.Energy profile of hydrogen diffusion along Mg (0001) surfaces.(a) Pure surface;(b) Ni loading at surface FCC site;(c) Ni loading at surface HCP site;(d) Ni loading at surface Bridge site;(e) Ni doping at 1st layer;(f) Ni doping at 2nd layer;(g) Ni doping at 3rd layer.The Mg,Ni and H atoms are represented by green,blue and white colors,respectively.

Fig.7.Deformation charge density maps and projected DOS profile for hydrogen diffusion on Mg (0001) surfaces.(a) Sectional deformation charge density map of H atom locating at surface HCP site on pure Mg (0001) surface;(b) Sectional deformation charge density map of H atom locating at surface HCP site next to Ni atom on Ni-doped Mg (0001) surface;(c) Projected DOS for hydrogen diffusion at surface HCP site on pure Mg (0001) surface;(d) Projected DOS for hydrogen diffusion at surface HCP site next to Ni atom on Ni-doped Mg (0001) surface.In deformation charge density maps,the arrow line indicates the position of the cross section.In deformation charge density maps,positive regions indicate the formation of bonds while negative regions indicate electron loss.The Mg,Ni and H atoms are represented by green,blue and white colors,respectively.

For hydrogen diffusion into bulk on pure Mg (0001) surface,the energy barrier from the outermost surface to the subsurface is calculated to be 0.47 eV (Fig.8a),which is slightly higher than that of diffusion from subsurface to deeper inside(0.32-0.33 eV).This indicates that the diffusion of H atom from the surface to the subsurface is slow while H atom diffuses fast from the subsurface to the bulk.This findin is consistent with that reported in previous studies [33,37].Similarly to surface diffusion,it still needs to step over high barriers (0.82-1.17 eV) to get rid of the adhesion from Ni atom during the inward diffusion path (Fig.8b-f).For instance,on Mg (0001) surface with Ni substituting at firs layer,H atom needs to overcome a high barrier of 0.86 eV when diffusing from firs subsurface octahedral site to the second subsurface octahedral site (Fig.8d).This high energy barrier is also attributed to the formation of strong Ni-H bond.On pure Mg(0001) surface,H atom at octahedral site and the surrounding Mg atoms form six equivalent Mg-H bonds with a binding energy of 3.09 eV (Fig.S2a).This means that the average bond energy of Mg-H bond is about 0.52 eV.In contrast,on Mg (0001) surface with Ni substituting at firs layer,H atom at octahedral site and the surrounding Mg and Ni atoms form fi e Mg-H bonds and one Ni-H bond (Fig.S2b),with a binding energy of 3.45 eV.Assuming that the bond energy of Mg-H bond remains unchanged,the bond energy of Ni-H bond here can be approximately estimated to be 0.88 eV,which is 0.36 eV higher than the Mg-H bond energy.This value increment is comparable to the corresponding increment of hydrogen diffusion energy barrier (0.53 eV).For comparison,the inward hydrogen diffusion via the adjacent FCC channel on the Ni-incorporated Mg (0001) surface has been evaluated(Fig.S3).Our calculation results reveal that the incorporation of Ni atom has negligible influenc to the H diffusion in the adjacent FCC channel.However,it is difficul for H atom to diffuse from local octahedral site to the horizontal adjacent one.According to our calculations,the energy barriers of lateral diffusion (octahedral site to the adjacent one on the same layer) are slightly higher than that of the inward diffusion (Fig.S4).Through carefully evaluating all diffusion paths,we derived the most likely diffusion path of hydrogen on Ni incorporated Mg (0001) surfaces as shown in Fig.9.Clearly,for all the case of Ni-incorporated Mg (0001) surfaces,the hydrogen atom prefers firstl immigrate along the surface and then penetrate into the bulk.

3.4.Implications for hydrogen storage design

According to above analysis,the minimum energy path(MEP)of hydrogen absorption on Ni-incorporated Mg surface has been evaluated and listed in Table 3.It can be seen that the single Ni atom is preferentially located in the second and third Mg layer with no help to the hydrogen dissociation and diffusion.Since the addition of Ni to Mg can significantl decrease its hydrogen storage capacity,the Ni atom located in the deep Mg bulk not only does little help to its kinetic improvement but also lowers down its maximum hydrogen storage density.This can be reflecte in the cycling performance and phase evolution of the as-prepared Mg-5wt.% Ni composite shown in Fig.10.It can be seen in Fig.10a that the hydrogen sorption performance went through two stages:rapid activation (1-5 cycle) and slow degradation (6-30 cycle).The rapid activation should be mainly attributed to the particle size reduction during cycling [41],which can be obviously seen from the SEM images shown in Fig.S5 and Fig.11.Since the particle pulverization is beneficia to improve hydrogen sorption rate,the slow degradation therefore should be attributed to the Ni penetration into the Mg matrix.This can be evidenced by the phase structure change shown in XRD curves (Fig.10) and SEM-EDS mappings(Fig.11).The nickel in the as-prepared Mg-Ni composite mainly exists in the form of elemental Ni (Fig.10b),which is evenly dispersed on Mg surface (Fig.11a).After 30 cycles,the dominant form of nickel is transformed into Mg2Ni(Fig.10c) and the elemental Ni disappears from the Mg surface (Fig.11b).This indicates that Ni penetration into the Mg matrix is unfavorable to hydrogen sorption.Therefore,it is crucial to stabilize the Ni atom on the surface or in the firs layer of Mg (0001) surface to maintain its catalytic effect.

Fig.8.Energy profile of hydrogen diffusion inward Mg (0001) surfaces.(a) Pure surface;(b) Ni loading at surface FCC site;(c) Ni loading at surface HCP site;(d) Ni loading at surface Bridge site;(e) Ni doping at 1st layer;(f) Ni doping at 2nd layer;(g) Ni doping at 3rd layer.O1,O2 and O3 represent the octahedral sites in the first second and third subsurface,respectively.The Mg,Ni and H atoms are represented by green,blue and white colors,respectively.

Fig.9.Most likely diffusion path of hydrogen from surface into the bulk of Mg (0001) surfaces.(a) Pure surface;(b) Ni loading at surface FCC site;(c)Ni loading at surface HCP site;(d) Ni loading at surface Bridge site;(e) Ni doping at 1st layer;(f) Ni doping at 2nd layer;(g) Ni doping at 3rd layer.O1,O2 and O3 represent the octahedral sites in the first second and third subsurface,respectively.The Mg,Ni and H atoms are represented by green,blue and white colors,respectively.

Table 3 Minimum energy path (MEP) of hydrogen absorption on Ni-incorporated Mg surface and the corresponding rate-limiting step.

Fig.10.Cycling performance and phase evolution of the Mg-Ni hydrogen storage composite.(a) Hydrogen sorption stability of the Mg-Ni composite for 30 cycles;(b) XRD of the as-prepared Mg-Ni composite;(c) XRD of the Mg-Ni composite after 30 hydrogen sorption cycles.

Fig.11.Surface micromorphology and element distribution of the Mg-Ni composite.(a) As-prepared;(b) After 30 hydrogen sorption cycles.

Fig.12.Schematic diagram of H2 absorption over Mg with stabilized Ni catalyst.

To our opinion,one feasible strategy to freeze the Ni position is introducing stabilizer or confinemen agent to limit the mobility of Ni atom (Fig.12).This strategy has usually been adapted to prevent magnesium from agglomerating or impurity gas corrosion [42,43].The commonly-used confinemen agent includes various carbon-based scaffold[44,45] and MOF scaffold [46].However,it still remains a challenge to retard the penetration of Ni into Mg bulk.In the study of Yao et al.[47],uniform-dispersed nano-Ni anchored on reduced graphene oxide totally dissolved into Mg after only 8 hydrogenation/dehydrogenation cycles.In the study of Li et al.[48],obvious decay of hydrogen absorption performance of Mg-Ni@C composites has been observed after 6 cycles.In the study of Ma et al.[49],the hydrogen absorption rate of Mg-Ni@C composite decreases by 0.9% in each absorption/desorption cycle,and the formation of Mg2Ni is believed to be responsible to the performance degradation.Hanada et al.[50] reported that the catalytic effect of Mg2Ni is weaker than Ni-metal,indicating that the dissolution of Ni in Mg has detrimental effect on the hydrogen storage performance of Mg-Ni system.Therefore,it still needs to seek more efficien stabilizer or confinemen agent for the Ni catalyst on Mg surface.Another possible way to stabilize Ni at the outmost layer is co-doping [23].It has been reported in a previous DFT investigation that Ti is a good assistant to stabilize co-doped Ni within the firs layer [22].Besides,Banerjee et al.[24] reported that Ni could be stabilized at the firs layer with co-doping of V according to their DFT calculations.This means that the substitution energies are lower when Ni is in the firs layer than in the second layer in the case of co-doping with Ti or V.According to the systematical analysis by Chen et al.[22],the possible underlying reason for this phenomenon is that Ti or V atom are slightly outside the firs layer when they are located in the second Mg layer,whereas most of the other transitional metals shift inward relative to the bulk structure.This difference could be beneficia to the interaction of Ti or V with Ni in the firs layer and could render them as“assistant”atoms to stabilize Ni in the firs layer.Nevertheless,this strategy is difficul to be implemented in the actual material synthesis process.In most single element doping case,it is usually achieved by balling milling the Mg matrix and a small amount of dopant,during which vacancy is randomly created and the doping atom can be placed easily.However,in the co-doping case,it is hard to achieve the ideal distribution of two elements as in theoretical design.Therefore,it is highly topical and important to develop a low-cost,controllable and technologically-feasible process for co-doping the Mg surface in further research.

4.Conclusions

In this work,hydrogen dissociation and diffusion on Mg(0001) surface with Ni incorporating at various locations are investigated based on periodic DFT calculations.Ni atom is preferentially located inside Mg matrix rather than in/over the topmost surface.Ni atom locating in/over the topmost Mg(0001) surface exhibits excellent catalytic effect on hydrogen dissociation.The corresponding energy barriers are reduced from 1.44 eV to less than 0.05 eV.When Ni atom locating in/over the topmost Mg (0001) surface,the rate-limiting step has been converted from hydrogen dissociation to surface diffusion.Ni doping inside Mg bulk not only does little help to hydrogen dissociation but also exhibits detrimental effect on hydrogen diffusion.It is crucial to stabilize the Ni atom on the surface or in the topmost layer of Mg (0001) surface to maintain its catalytic effect,since Hsorbital of H2can directly interact with Nidorbital.Introducing a stabilizer or a confinemen agent to limit the mobility of Ni atom is the most feasible strategy to freeze the Ni position.

Declaration of Competing Interest

The authors declare that they have no known competing financia interests or personal relationships that could have appeared to influenc the work reported in this paper.

Acknowledgment

The authors wish to acknowledge the financia support from the National Natural Science Foundation of China(Grant Nos.U1610103,21805169 and 21978156),Shandong Provincial Natural Science Foundation,China (Grant No.ZR2018BB069),and Project of Shandong Province Higher Educational Young Innovative Talent Introduction and Cultivation Team (Hydrogen energy chemistry innovation team).We are also grateful to Shenzhen Supercomputer Center for calculation support.

Supplementary materials

Supplementary material associated with this article can be found,in the online version,at doi:10.1016/j.jma.2021.03.002.