Boron from net charge acceptor to donor and its effect on hydrogen uptake by novel Mg-B-electrochemically synthesized reduced graphene oxide

Hydrogen uptake (H-uptake) is studied in ball milled Mg-B-electrochemically synthesized reduced graphene oxide (erGO) nanocomposites at PH2 ≈ 15 bar, ~ 320 °C. B/C (weight ratio): 0, ~ 0.09, ~ 0.36, ~ 0.90 are synthesized maintaining erGO≈10wt %. B occupies octahedral interstices within Mg unit cell—revealed by electron density maps. Persistent charge donations from Mg and B to C appear as Mg-C (~ 283.2 eV), B-C (~ 283.3–283.9 eV) interactions in C-1s core X-ray photoelectron spectroscopy (XPS) at all B/C. At B/C > 0.09, charge reception by B from Mg yields Mg-B interaction. This net charge acceptor role of B renders it electron-rich and does not alter Mg unit cell size significantly. Despite charge donation to both C and B, the Mg charge is <  + 2, resulting in long incubation times (> 5 h) at B/C > 0.09. At B/C≈0.09 the minimal Mg-B interaction renders B a charge donor, resulting in Mg-B repulsion and Mg unit cell expansion. Mg-C peak shift to lower binding energies (C-1s XPS), decreases incubation time to ~ 2.25 h and enhances H-uptake kinetics. Various atomic interactions influence the reduction of incubation time in H-uptake and increase its kinetics in the order: (Mg → C; B → C)B/C≈0.09, B: donor > (Mg → C)B/C=0 > (ternary Mg → B → C)B/C>0.09, B: acceptor.

are in their early stages, particularly during H-uptake. Moreover, the charge transfer role of B and its impact on the hydrogen storage by Mg-rGO system are not well known from literature.
Most of the studies on hydrogen storage by MgH 2 -based materials are focussed on H-release rather than H-uptake. Some scientific reasons for this are (i) uncertainty involved in experimentally identifying the lattice positions of H, any elemental additions etc. in the crystalline hydrogen-hosts upon H-uptake; this further renders (ii) the investigation of various atomic interactions among the components difficult; (iii) such uncertainties in structural investigations can lead to erroneous estimation of enthalpy and entropy changes. This is particularly true in the case of composite systems such as Mg(H 2 )-B-rGO, where both Mg and C (from rGO) can, synergistically, form bonds with H 9,18 . Our earlier study on H-uptake and release by Mg-rGO demonstrates that along with Mg, C also forms bond with H due to charge (electron) transfer from Mg to C 18 . Practically, the H-uptake studies are less appealing than their H-release counterparts for reporting, since H-uptake involves long incubation times 4 . Nevertheless, hydrogen storage involves both H-uptake and release and the studies on the former are indispensable.
In the present study H-uptake by the novel Mg-B-erGO nanocomposites at P H2 ≈ 15 bar and ~ 320 °C is reported. The erGO is a novel electrochemically synthesized rGO, which we reported in our recent study, that contains fewer functional groups than rGO prepared by the modified Hummer's method 19 . Here, Mg, B and erGO are ball milled together at various B/C weight ratios (0, ~ 0.09, ~ 0.36 and ~ 0.90) maintaining ~ 10 wt % of erGO in the nanocomposites. Several interactions among Mg, B and C, involving charge transfer, develop upon ball milling. Interestingly, a "critical" B/C ratio (~ 0.09) is identified. Below this ratio B is a charge (electron) donor to C. Above this ratio B accepts charge from Mg and also donates to C, acting as a net charge acceptor. When B is charge donor, a decrease in the incubation time is observed for H-uptake from ~ 3.25 h in the absence of B to ~ 2.25 h (i.e. a drop of ~ 31%) upon slight addition of B (B/C ≈ 0.09). Various fundamental scientific questions are answered: (i) How and why does B affect the Mg unit cell? (ii) How and why do interactions among Mg, B and C develop? (iii) Why does B switch the role from charge donor to net acceptor upon increasing its content? (iv) What is the impact of this role-switching by B on the H-uptake by the Mg-B-erGO nanocomposites? Our claims are supported by X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) (C-1s, Mg-2p and B-1s spectra) and the novel electron density maps, the technique which we used elsewhere 20 .

Results
Incubation time during H-uptake. The H-uptake by all the Mg-B-erGO nanocomposites at P H2 ≈ 15 bar and ~ 320 °C is plotted as wt % hydrogen in Mg versus time in Fig. 1. For all the nanocomposites, an incubation time corresponding to negligible H-uptake was seen before initiation of hydrogen absorption. These incubation times are plotted versus B/C ratio in the inset of Fig. 1. The incubation time decreases from ~ 3.25 h in B/C = 0 to ~ 2.25 h with a slight addition of B (B/C≈0.09). Surprisingly, with further addition of B the nanocomposites exhibit increased incubation time (inset, Fig. 1). This clearly indicates the existence of a critical B/C ratio around 0.09 at which the incubation time reaches a minimum. Moreover, following the incubation time all the H-uptake curves are sigmoidal in nature, prior to reaching saturation levels at ~ 6 wt %. The rate of H-uptake is the highest for B/C≈0.09. The nanocomposite with B/C ≈ 0.90 has the lowest rate of uptake and the longest incubation time.   Structural changes in Mg unit cell. Figure 2 shows the lattice parameters of the Mg unit cell, estimated from XRD patterns through Rietveld refinement (see Supplementary Fig. S3 and Tables S2-S5) using FullProf suite (version: 7.20) 23 . Prior to analyzing the obtained lattice parameters, the structural integrity of the crystal for all the nanocomposites was verified and ensured by noting that the c/a ratio is close to the theoretical value of 1.623 for an hcp crystal 24  Therefore, the electron density distribution is nearly uniform (asphericity closer to 1) for the nanocomposites of B/C ≈ 0.09. The aspherical charge distribution suggests covalency 20 . Figure 3b shows the relative electron density at the octahedral interstices in the (0001) basal plane of the Mg unit cell ('Octahedral interstices' , Fig. 3b), normalized with respect to the maximum electron density found around Mg (i.e. ρ Octahedral /ρ max ). The corresponding electron density maps are shown in Supplementary Fig. S4. It is known that the dopants can occupy these sites when incorporated in the Mg unit cell 26  Chemical interactions. The core C-1s and Mg-2p X-ray photoelectron spectra (XPS) of all the nanocomposites are shown in Fig. 4a,b, respectively. The experimental data obtained from XPS were deconvoluted using Gaussian profile by employing an XPS-specific deconvolution procedure to estimate the atomic interactions 27,28 .
In C-1s spectra, the peaks at the binding energies ~ 283.2 eV, ~ 283.3-283.9 eV, ~ 284.6 eV, ~ 285.7 eV, ~ 286.6 eV, and ~ 290 eV correspond to Mg-C 18 , B-C 29,30 , C-C sp 2 , C-OH, C-O-C 31-33 and C-2p π → π* transition 9,34 ,  Supplementary Fig. S5). The possible presence of COOH and C=O groups was also tested by considering them along with -OH and C-O-C groups in various combinations while deconvoluting the C-1s spectra. However, only those fits which contain only -OH and C-O-C groups were converged. The results of various such deconvolution exercises are presented in Supplementary Tables S6-S9. Moreover, as the erGO is synthesized using (NH 4 ) 2 SO 4 (see 'Methods' section), it is less likely for these groups to be present in erGO. Hence, it is believed that the COOH and C=O groups are absent in erGO. The XPS-specific deconvolution method 28 38,39 . With increase in B in the nanocomposites, the Mg-B peaks become stronger and shift towards lower binding energies in the above mentioned range (Fig. 4b). Interestingly, at B/C≈0.90 the binding energy of the Mg-B peak is around 51.3 eV. This is closer to the binding energy of Mg-B peak in MgB 2

38
. The presence of both B-C (Fig. 4a) and Mg-B (Fig. 4b) peaks from XPS suggests that B has simultaneous interactions with both C and Mg.

Discussion
The results suggest a critical B/C ratio around ~ 0.09, at which the incubation time during hydrogen uptake by the Mg-B-erGO nanocomposites is the least and increases above or below this ratio (Fig. 1). This critical B/C can be refined by selecting compositions around B/C≈0.09. The crystallographic (Fig. 2)  In the nanocomposite where B is absent (B/C = 0), the covalency of Mg as suggested by the aspherical charge distribution 40 (p/q, Fig. 3a) indicates that there is an interaction between Mg and the surroundings. Further, the negative relative electron density in the octahedral interstices of the (0001) basal plane in Mg in B/C = 0 (Fig. 3b,  Supplementary Fig. S4) shows the deficiency of electrons at this site. This suggests that there is no electron donation from Mg to this octahedral site. Therefore, it is reasonable to conclude that Mg donates electrons to C, which appears as 'Mg-C' peak in C-1s XPS spectrum (Fig. 4a). This electron donation from Mg to C can be reinforced by the fact that the electron affinity of Mg (~ 0 kJ mol −1 ) is lower than that of C (~ 153.9 kJ mol −1 ) 41 . Any local changes in the chemical environment can cause deviation from these values. In our earlier work, we www.nature.com/scientificreports/ demonstrated that the Mg-C interaction causes an increase of charge on carbon atoms leading to a change of hybridization in C from sp 2 to sp 3 , resulting in the C-H bond aiding in H-uptake 18 . A slight addition of B in B/C ≈ 0.09 decreases incubation time and enhances H-uptake kinetics (Fig. 1). This improvement is due to the structural and local environmental changes introduced by B. Upon addition of B, a significant expansion of the Mg unit cell can be seen from the lattice parameters in Fig. 2. Normally, the repulsive forces between two entities within a unit cell can cause its expansion 42 . In the present scenario, it can be expected that B is incorporated within the Mg unit cell and possesses a positive charge. The incorporation of B is suggested by the larger negative electron density compared to B/C = 0, i.e. lack of electrons, at the octahedral interstices in the (0001) Mg basal plane for B/C≈0.09 (Fig. 3b, Supplementary Fig. S4). In other words, the charge in these regions is more positive in B/C≈0.09 compared with that in B/C = 0. Most likely, B is present in these octahedral interstices 43 , while maintaining structural integrity, and possesses positive charge. The likely reason for the positive charge over B is the charge transfer from B to Mg and/or to C. However, the charge transfer from B to Mg is mostly not possible as the electron affinity of elemental B is ~ 28.9 kJ mol −1 and that of Mg is ~ 0 kJ mol −141 . Therefore, most likely, the charge transfer takes place from B to C, giving rise to B-C interaction as suggested by B-C peak at ~ 283.4 eV in C-1s spectrum (B/C ≈ 0.09, Fig. 4a). The absence of any peaks in B-1s core spectra for all the nanocomposites (Supplementary Fig. S6) Table S10). This clearly indicates that the B-C interactions seen in C-1s spectra (Fig. 4a) are not due to any strong bond formation between B and C. The presence of both Mg-C and B-C interactions in B/C≈0.09 show that both Mg and B are charge donors to C.
Upon increasing the B content in the nanocomposites to B/C ≈ 0.36 and 0.90 both the incubation time and the kinetics of H-uptake are deteriorated (Fig. 1). Interestingly, the lattice parameters in B/C ≈ 0.36 are close to those of B/C = 0 (Fig. 2) and have decreased further in B/C ≈ 0.90. The possible reasons for the restoration of the Mg unit cell to its original size are: (i) B is not present within the Mg unit cell at these compositions; (ii) B is present within the unit cell and shrinks its size to almost the original value by developing possible additional interactions. However, it is likely that B is incorporated in the Mg unit cell even at compositions higher than in B/C ≈ 0.09 ( Fig. 3(b)). Therefore, the first reason is not plausible. Hence, the possible presence of B in the Mg unit cell is maintaining its size closer to its pristine counterpart (B/C = 0). From Fig. 3a, the charge distribution around Mg atom clearly shows that the asphericity increases (with respect to that at B/C ≈ 0.09) at B/C ≈ 0.36 and reaches the highest at B/C ≈ 0.90. This shows that the various interactions in B/C ≈ 0.36 and 0.90 are happening to a different extent compared with the other nanocomposites.
The different extents of the atomic interactions in B/C ≈ 0.36 and 0.90 are evident from the increase in the Mg-C and B-C binding energies with respect to those in B/C ≈ 0.09 from the C-1s spectra (Fig. 4a). The Mg-C and B-C peak positions increased from ∼ 282.9 and ∼ 283.4 eV (B/C ≈ 0.09) to ∼ 283.3 and ∼ 283.6 eV (B/C ≈ 0.36), respectively. Since the B-C peaks possess acceptable FWHMs (0.91 and 1.00 eV for B/C ≈ 0.09 and 0.36, respectively) these peak shifts in B-C are mostly realistic trends. This increase in the binding energies clearly indicates that a lower charge is received by C from both Mg and B 36 . An important question here is that: why does the charge reception by C decrease despite the charge donation by both Mg and B? The changes in the interactions of C with Mg and B also introduce observable interactions between Mg and B in B/C ≈ 0.36. The Mg-2p spectrum for this composition shows the presence of Mg-B peak at ~ 52.2 eV, corresponding to the charge transfer from Mg to B 38 (Fig. 4b). This Mg-B peak in B/C ≈ 0.09 is very feeble, rendering the relative charge transfer from Mg to B 38 in B/C ≈ 0.36 very significant. The combined analysis of C-1s and Mg-2p spectra (Fig. 4) clearly indicates that Mg is donating charge to both C and B. As a result of this, the net charge received by C from Mg can decrease (Fig. 4a). Similarly, as B shares its valence electrons with Mg, more likely a lower net charge is received by C from B (Fig. 4a). The presence of both Mg-B and B-C interactions can render B a charge acceptor (from Mg) and a donor (to C), respectively. However, the relative electron richness at octahedral sites (B/C ≈ 0.36, Fig. 3b), which B is likely occupying, suggests that B is negatively charged making it a net charge acceptor. The results indicate that ternary Mg-B-C interactions develop in B/C ≈ 0.36. Similar trends in the interactions are observed in the case of B/C ≈ 0.90 (Fig. 4).
The nanocomposite with B/C ≈ 0.90 exhibits asphericity close to that of B/C = 0 (Fig. 3a). Its ρ Octahedral /ρ max value is negative and similar to that of B/C = 0 (Fig. 3b, Supplementary Fig. S4). Despite these similarities, it shows longer incubation time (Fig. 1) for H-uptake. This suggests that, possibly, further different level of interactions are present in this nanocomposite, as implied by the lowest unit cell volume in this case (Fig. 2). From Fig. 4a, the Mg-C and B-C peaks for B/C of 0 (only Mg-C), ∼ 0.09 and ∼ 0.36 are shifted to lower binding energies compared with those in B/C ≈ 0.90 suggesting stronger interactions. The Mg-B peak is located at ~ 51.3 eV (Fig. 4b).
Interestingly, this is the position where the similar peak in MgB 2 is expected 45 . Moreover, B is expected to be located in the octahedral interstices in Mg unit cell (hcp) in B/C ≈ 0.90 (Fig. 3a), which are the cites where B is located in MgB 2 (also hcp) 46 . However, the XRD pattern ( Supplementary Fig. S1) does not show any MgB 2 peaks [ICSD code: 26675]. Probably, the composition B/C ≈ 0.90 is just about favorable for the formation of MgB 2 . The possible presence of MgB 2 in B/C≈0.90 is also supported by the lowest unit cell volume (Fig. 2), the highest asphericity/covalency (Fig. 3a), the presence of the highest Mg-C and B-C binding energies (Fig. 4a) and a peak corresponding to MgB 2 (51.3 eV) in Mg-2p spectra (Fig. 4b). However, additional characterization is needed to substantiate the presence of MgB 2 in this nanocomposite.
The analysis from the present study is used to propose a mechanism in terms of the structural and local environmental changes in the Mg-B-erGO nanocomposites that help in reducing the incubation time during their H-uptake. This mechanism is schematically shown in Fig. 5. Among the investigated compositions, those at B/C > 0.09 develop ternary Mg-B-C atomic interactions, where C from erGO receives charge from both Mg and B (Mg-C, B-C in Fig. 4a). Interestingly, B also receives charge from Mg developing Mg-B interactions (Fig. 4b). Such a charge reception renders B negative making it a net charge acceptor. This is evidenced by the,  (Fig. 3b, Supplementary Fig. S4). The attraction between the negatively charged B and Mg helps in maintaining the Mg unit cell size in B/C≈0.36 almost the same as that in B/C = 0 (Fig. 2). Despite the charge donation to both C and B, the charge on Mg is < + 2 rendering it difficult to bond with H (dotted line between H and Mg, Fig. 5). This results in the longer incubation times at B/C > 0.09 (Fig. 1). In the case of B/C ≈ 0.90 the longest incubation time can be attributed to the possible presence of MgB 2 . However, this needs to be confirmed with further analysis.
At B/C ≈ 0.09, B acts as a charge donor to C (B-C, Fig. 4a). The possible presence of the positively charged B in the octahedral interstices of Mg unit cell (ρ Octahedral /ρ max , Fig. 3b, Supplementary Fig. S4) repels Mg ('repulsion' , Fig. 5) and causes the lattice expansion (Fig. 2). Here, Mg is more positive relative to that at B/C > 0.09 due to the lower binding energy of Mg-C (Fig. 4a)    Following ball milling the nanocomposites were loaded and sealed in a hydrogenation reactor inside glovebox. Subsequently, the reactor was brought out for H-uptake experiments in a Sievert's type apparatus. Prior to H-uptake, the reactor was purged with Ar. The reactor containing the powder was heated up to ~ 320 °C under vacuum. Eventually, hydrogen gas (99.999%) was permitted in to the reactor at ~ 15 bar and ~ 320 °C. Isothermal H-uptake experiments were conducted on these Mg-B-erGO nanocomposites at these conditions up to saturation. The quantity of H-uptake by the powders was estimated using ideal gas law 20 .
The phase analysis on these nanocomposite powders was performed by X-ray diffraction (XRD) employing PANalytical EMPYREAN goniometer. Cu Kα radiation (wave length: 1.5406 Å) was used for the same. The baseline correction was performed on the obtained data and the phases were indexed using the standard ICSD references. The ICSD database codes used for indexing Mg, MgO and rGO phases are 76748, 104845 and 31170, respectively.
The XRD patterns of the nanocomposites were subjected to Rietveld refinement using FullProf suite (version: 7.20) to estimate phase percentages and to obtain crystallographic data of phases 23 shown in Supplementary Fig. S3 and Tables S2-S5. Baseline for the experimentally obtained data was corrected using winPLOTR program 48 . After baseline correction, the data was refined using Pseudo-Voigt function 49 (Supplementary Eq. S8). The lattice parameters (a, b, c, α, β, γ) of hcp Mg, FWHM (U, V, W, IG), shape (η 0, X) and asymmetry were refined to obtain the best fit with the experimental data for Mg and MgO phases (Supplementary Eqs. S4, S9). For refining the lattice parameters, the initial values of a = 3.2093 Å, b = 3.2093 Å and c = 5.2103 Å were fed as input into EdPCR application of the FullProf suite prior to Rietveld refinement. The electron density maps are used to estimate the local environment within the crystal lattice. Electron density maps for Mg unit cell were developed using GFourier Program (version: 4.06) through Maximum Entropy Method (MEM) 25 . Through MEM the electron density ρ(r) was calculated by Fourier transformation of structure factors obtained after Rietveld refinement 50 (Supplementary Eq. S11).
X-ray photoelectron spectroscopy (XPS) was performed on the nanocomposites employing Axis Supra Photoelectron spectrometer (Kratos Analytical) using Al Kα source. The powders were ultrasonicated in toluene for uniform dispersion following which they were drop casted onto Al foil and exposed to XPS source maintaining 20 eV pass energy. The obtained high resolution XPS spectra for various orbitals were deconvoluted using Gaussian function employing an XPS-specific deconvolution method for estimating the chemical bonds 28 . License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.