Graphene- and Phosphorene-like Boron Layers - eduaossano.com

Graphene- and Phosphorene-like Boron Layers

0 downloads 13 Views 1MB Size Report
molybdenum phosphide (MoP),12 molybdenum carbide. (MoC)22 and molybdenum boride (MoB)16,17 have received a great deal of attention due to their low ...

Communication pubs.acs.org/JACS

Graphene- and Phosphorene-like Boron Layers with Contrasting Activities in Highly Active Mo2B4 for Hydrogen Evolution Hyounmyung Park,†,‡,# Yuemei Zhang,†,§,# Jan P. Scheifers,†,§ Palani R. Jothi,†,§ Andrew Encinas,† and Boniface P. T. Fokwa*,†,‡,§ †

Department of Chemistry, University of California, Riverside, Riverside, California 92521, United States Department of Chemical and Environmental Engineering, University of California, Riverside, Riverside, California 92521, United States § Center for Catalysis, University of California, Riverside, Riverside, California 92521, United States ‡

S Supporting Information *

only a few studies on the HER activity of Mo-based borides.16,17 The main challenge for this class of materials, even at the microscale, is their single-phase synthesis. For example, the commercially available α-MoB is still contaminated by β-MoB.16 The fact that they are already highly active in the bulk accelerates the discovery of HER active borides. Another advantage of metal borides is their stability in acidic and basic solutions. Thus, extensive HER activity investigations for this large group of compounds are particularly promising. Using arc-melting procedures, high purity (>95 wt %, weight percent) of the binary bulk molybdenum borides Mo2B, αMoB, β-MoB and MoB2 could recently be synthesized. The high purity allowed the study of their HER activity, which turned out to increase with increasing boron content in the order Mo2B, α-MoB, β-MoB and MoB2.17 One potentially interesting phase, Mo2B4 (with the same boron content as MoB2), could not be synthesized with sufficient purity by this method. In this work, we have achieved the first single-phase synthesis in the Mo−B system using the tin flux method, and it has enabled the discovery of one of the highly active bulk borides to date, Mo2B4. The chemical composition and structure of Mo2B4 (known since 1947 as “Mo2B5‑x”, see phase diagram in the Supporting Information (SI), Figure S1, top)26 were revised ten years ago using neutron and synchrotron data,27 and later (2013) confirmed by theoretical calculations.28 Its structure contains two types of boron layers; a flat graphene-like layer and a corrugated layer (like the recently discovered phosphorene layer)25 build of cyclohexane-like arm-chair motifs. Furthermore, the B-layers alternate with Mo-layers along the [001] direction (Figure 1, right). This structural arrangement is similar to that of the highly HER active MoB2, the only difference being the absence of the corrugated B-layer in MoB2. Recently, the two modifications α-MoB and β-MoB were found to have similar HER activity in the bulk, which could have been expected because the boron subunits (zigzag chains) are the same, the difference being just their orientations.17 This finding indicates that the structure−activity relationship concept may apply to these borides. Consequently, this would suggest that Mo2B4 may also have a significantly high

ABSTRACT: Two different boron layers, flat (graphenelike) and puckered (phosphorene-like), found in the crystal structure of Mo2B4 show drastically different activities for hydrogen evolution, according to Gibbs free energy calculations of H-adsorption on Mo2B4. The graphene-like B layer is highly active, whereas the phosphorene-like B layer performs very poorly for hydrogen evolution. A new Sn-flux synthesis permits the rapid single-phase synthesis of Mo2B4, and electrochemical analyses show that it is one of the best hydrogen evolution reaction active bulk materials with good long-term cycle stability under acidic conditions. Mo2B4 compensates its smaller density of active sites if compared with highly active bulk MoB2 (which contains only the more active graphene-like boron layers) by a 5-times increase of its surface area.

H

ydrogen has been considered as the most promising, sustainable, clean and renewable energy carrier.1,2 Hydrogen can be produced from the electrochemical water splitting through the hydrogen evolution reaction (HER). This method is highly attractive because it is efficient, clean and sustainable for large scale hydrogen production.3 Among the metals, noble metals such as platinum are the best HER catalysts because they have an overpotential close to zero volts, but low abundance and high cost of these materials limit their large-scale application.4−7 Thus, the development of low-cost, high-performance non-noble-metal catalysts to replace the precious metals is important and necessary to make hydrogen a cost-effective energy carrier. Recently, electrocatalysts based on earth-abundant elements such as sulfides,8,9 nitrides,10,11 phosphides,12,13 carbides,14,15 borides,16−20 were shown to have a high HER catalytic activity. In addition, nonmetallic carbon materials have also shown HER catalytic activity.21 Among these alternatives, molybdenumbased materials such as molybdenum disulfide (MoS2),8 molybdenum phosphide (MoP),12 molybdenum carbide (MoC)22 and molybdenum boride (MoB)16,17 have received a great deal of attention due to their low cost and high activity. To further enhance the HER activity, nanoscale23,24 and porous Mo-based25 materials have been studied. However, there are © 2017 American Chemical Society

Received: July 11, 2017 Published: September 5, 2017 12915

DOI: 10.1021/jacs.7b07247 J. Am. Chem. Soc. 2017, 139, 12915−12918

Communication

Journal of the American Chemical Society

Figure 2. XPS spectra of Mo2B4: Mo 3d (a) and B 1s (b). Experimental data (■), fitting peaks (black line); (a) MoO3 (red), MoO2 (orange), Mo2O3 (green) and Mo peaks (blue) from Mo2B4; (b) B2O3 (red) and B peaks (blue) from Mo2B4.

Figure 1. (left) Refined powder X-ray diffraction data of Mo2B4 showing the measured (red) and calculated (black) intensities, the difference plot (blue) and the reflection positions (green, Mo2B4: top; Sn: bottom). (Right) Crystal structure of Mo2B4 showing two types of boron layers alternating along [001].

activity because of its structural similarity with highly active bulk MoB2. However, the presence of puckered B-layers complicates a reliable activity prediction by the structure− activity relationship (as made in the MoB case). Thus, theoretical calculations will be key to understand the catalytic behavior and contribution of the different boron layers to the HER activity of this new candidate material for HER. The synthesis of bulk Mo2B4 has been a great challenge that was addressed over several decades.29 The most recent procedure by Albert et al.27 illustrates quite well the tenuous process (ca. 17 days) toward high quality samples (see SI for more details). Using a Sn flux, we have now synthesized singlephase samples of Mo2B4 in a single step at 1100 °C. We take advantage of the fact that the molten metal flux (Sn melts at 232 °C) facilitates diffusion of the elements in the molten mixture and that Sn does not form any stable binary boride. In addition, Sn can be easily removed afterward by dissolving it in hydrochloric acid. This method has been used very effectively to prepare several classes of materials including borides, carbides and intermetallics, just to name a few.30,31 The other advantage of tin flux synthesis is that smaller particles are obtained without grinding compared to samples obtained by arc melting. More details on this synthesis procedure are given in the SI. Figure 1 shows the Rietveld refinement plot of the X-ray powder diffraction data, which indicates that Mo2B4 is the only boride phase present (less than 1 wt % of undissolved tin was detected, see SI, Table S1). The refined lattice parameters of Mo2B4, a = 3.012 (3) and c = 20.932 (3) Å, match excellently the reported values (a = 3.010 Å, c = 20.926 Å).27 The morphology of Mo2B4, analyzed by SEM, indicates irregular particle shapes and a broad size distribution ranging from 1 to 10 μm (see SI, Figure S2 left). EDS semiquantitative analysis (see SI, Figure S2 right) confirmed the presence of Mo and B as well as the undissolved Sn impurity (0.8%). The measured Brunauer−Emmett−Teller (BET) surface area (see SI, Figure S3) of 20.45 m2/g is about 5 times larger than for bulk MoB2 obtained by arc melting (4.23 m2/g),17 probably due to the smaller average particle size of the Mo2B4 sample. Figure 2 displays the Mo 3d and B 1s core level XPS spectra of the Mo2B4 surface indicating four different oxidation states of molybdenum (Mo0, Mo3+, Mo4+ and Mo6+), and two for boron (B0 and B3+). The fitted peak positions (Table S2) and a full survey XPS spectrum (Figure S4) are given in the SI. The oxidized Mo- and B-species belong to minor oxides impurities

Figure 3. (a) Polarization curves for amorphous B, Mo, Mo2B4 and Pt/C in 0.5 M H2SO4. (b) Corresponding Tafel plot of Mo2B4 and Pt/ C. (c) Stability measurements of Mo2B4 after 2000 cycles in 0.5 M H2SO4. Scan rates: 1 mV/s for (a) and 100 mV/s for (c).

(MoO3, MoO2, Mo2O3 and B2O3) usually found on the surface of materials when exposed to air. In previous XPS studies on binary molybdenum borides, it was shown that the freshly prepared materials are less contaminated by oxides, indicating that oxidation occurs as the samples are exposed to air.17 Similarly, in our freshly prepared Mo2B4 sample, the oxide peaks are smaller than those found in commercial α-MoB.16 The presence of an oxide layer has been found on the surfaces of many molybdenum-based materials, but they do not affect their HER activity, because those oxides are dissolved during the initial operation cycles in acidic media.16 The electrochemical activities of Mo2B4, Pt on carbon (Pt/ C), carbon sheet, molybdenum and amorphous boron were measured using a three-electrode system in a 0.5 M H2SO4 solution at a scan rate of 1 mV/s under IR drop compensation (Figure 3a). The polarization curves indicate that the carbon sheet exhibits no activity, whereas Pt/C shows the highest activity, as expected, with an overpotential value similar to that reported in the literature, whereas molybdenum is more active than amorphous boron, but both have poor HER activities. The overpotential (270 mV at 3.5 mA/cm2 current density) of Mo2B4 is comparable to that of the best bulk molybdenum boride found so far at similar loading, MoB2 (260 mV at 3.5 12916

DOI: 10.1021/jacs.7b07247 J. Am. Chem. Soc. 2017, 139, 12915−12918

Communication

Journal of the American Chemical Society mA/cm2 current density).17 At first sight, this result seems logical, because the two phases have the same composition, but the 5-times larger surface area of Mo2B4 must be considered in the final evaluation of its catalytic activity, as found recently for the two modifications of MoB.17 However, the crystal structures of α-MoB and β-MoB are very similar as they contain the same type of boron and molybdenum substructures (just different orientations), thus they are expected to have similar active sites. On the contrary, the crystal structures of MoB2 and Mo2B4 have one key difference: Though MoB2 contains only graphene-like boron layers, Mo2B4 contains an additional puckered (phosphorene-like) boron layer. Therefore, there is a possibility to have different active sites for these two structures. The electrochemically active surface area (ECSA) was estimated by the double layer capacitance measurement through cyclic voltammetry (CV) at various scan rates (between 0.1 and 0.2 V vs RHE, see SI, Figure S6).32,33 The obtained double layer capacitance of bulk Mo2B4 (156 μF/cm2) is significantly larger than that of bulk MoB2 (101 μF/cm2). However, as shown above, the overpotential obtained for Mo2B4 is comparable to that of MoB2. The fact that Mo2B4 has a 5-times larger surface area than MoB2 shows that it has less active sites per unit surface area than MoB2, which indicates that the different boron layers found in the Mo2B4 structure may have different activities. Indeed, our density functional theory (DFT) calculations suggest that the graphene-like boron layer is far more active than the phosphorene-like boron layer (see below). The HER mechanism of Mo2B4 was probed by the Tafel analysis (Figure 3b). In acidic solution, three reactions are considered as the rate-determining step (RDS); the Volmer reaction (Tafel slope of ∼120 mV/dec), the Heyrovsky reaction (Tafel slope of ∼40 mV/dec) and the Tafel reaction (Tafel slope of ∼30 mV/dec).34−37 As Figure 3b shows, the obtained Tafel slope for Pt/C is ∼31 mV/dec, indicating that the RDS for Pt/C is the Tafel reaction, which is consistent with previous studies.12,34 The Tafel slope of Mo2B4 (80 mV/dec) does not match exactly any of the ideal Tafel slopes, but it is close to that found for MoB2 (75 mV/dec). Like all other molybdenum borides studied until now, it is difficult to determine the RDS using the Tafel analysis in the current system. This finding suggests that the HER mechanism may be more complex for this class of active bulk materials than for nanoscale materials and Pt/C.38,39 Stability is also a key factor for a good HER catalyst. To evaluate stability, we have performed CV experiments in acid solution (Figure 3c). After 2000 cycles, the activity of Mo2B4 is nearly as good as initially, indicating that the synthesized Mo2B4 sample has good stability in acidic solution. To investigate the above-suggested different activities of the two boron layers present in the structure of Mo2B4, we have calculated the Gibbs free energy (ΔGH) for H adsorption on different surfaces (including molybdenum surfaces) by DFT calculations. The calculated Mo2B4 surfaces correspond to the layers perpendicular to the [001] direction (see Figure 1, right). Because we have a polycrystalline sample, we expect most facets found in XRD (see SI, Figure S1, bottom) to be present, including the B- and Mo-layers ({00n} crystallographic plane sets with n = 3, 6, ...). ΔGH is commonly used and widely accepted as a descriptor of the HER activity.12,40 An optimal HER activity can be achieved at a ΔGH value close to zero, and under this condition the overall reaction of both H adsorption

and H2 desorption has the maximum rate.40 The bonding energy of H on the surface (ΔEH) and ΔGH were calculated for four different surfaces (Mo1-, Mo2-, flat B- and puckered Bterminated surfaces) at 25% H coverage to determine the preferred adsorption sites on each surface (see section III and Table S4 in the SI for more details). Three types of adsorption sites were considered: on top (T) of a Mo or B atom, on a Mo−Mo or B−B bridge site (Bg) and on a hollow (Ho) site. There are two top sites (T1: on top of the outer boron; T2: on top of the inner boron) for the puckered phosphorene-like B surface (see SI, Figure S8). As shown in Table S4 for 25% H coverage, the Ho site of the Mo1 surface, the Bg site of the Mo2 and flat B surfaces, and one of the top (T1) sites of the puckered B surface are the most preferred adsorption sites, because of their high surface binding energy. H binds much stronger on the puckered B layer (−0.81 and −1.14 eV on the Bg and T1 sites, respectively) than on the flat B layer (−0.53 and −0.42 eV on the Bg and T site, respectively). However, this strong binding also makes the puckered B layer less active for the HER than the flat B layer. In fact, the ΔEH value obtained for the flat B layer is close to that calculated for Pt (111), hinting at the possible strong activity of this layer. The two Mo layers also bind with H, but not as strongly as the puckered B layer, an indication of moderate catalytic activity. In addition, ΔGH was examined at different H coverages for the preferred adsorption sites of all Mo- and B-terminated surfaces. The ΔGH values obtained for different H coverages for the two Mo and two B layers are listed in the SI (Table S5) and plotted in Figure 4a. ΔGH increases as H coverage increases on both the Bg and T sites of the flat B layer, and reaches zero at around 50% H coverage for both sites. Therefore, both the Bg and T sites of the flat B layer should be very active for HER. Beside the fact that the puckered B layer binds strongly with H at 25% H coverage, the variation of ΔGH with increasing H coverage up to 100% never leads to a ΔGH value of 0, indicating that the

Figure 4. (a) Gibbs free energy (ΔGH) for H adsorption on different surfaces plotted as a function of hydrogen coverage. (b) Calculated free-energy diagram for HER over different surfaces at 50% H coverage. 12917

DOI: 10.1021/jacs.7b07247 J. Am. Chem. Soc. 2017, 139, 12915−12918

Communication

Journal of the American Chemical Society puckered B layer performs very poorly for the HER. ΔGH for the two Mo surfaces (Figure 4a) behaves similarly to that of the T1 site of the puckered B layer, indicating similar activity. The ΔGH diagram plotted at 50% coverage (Figure 4b) indicates that the graphene-like B layers are the most active, followed by the molybdenum layers and last the phosphorene-like B layers. This theoretical result confirms and even extends the experimental finding that Mo2B4 has less active sites per surface area than MoB2, because 50% of the highly active flat graphene-like B layers have been traded for the less active puckered phosphorene-like layers, thus reducing its HER activity. Nevertheless, Mo2B4 has compensated its lack of active sites by a 5-times increase of its surface area achieved through the Sn-flux synthetic method. In conclusion, single-phase bulk Mo2B4 has been successfully synthesized by the tin flux method for the first time. Electrochemical characterizations show that this phase is among the best bulk HER catalysts in acidic conditions and it also shows good stability. As found by DFT free energy calculations, the active sites in the graphene-like B layer contribute to the high HER activity, in contrast to the phosphorene-like B layer which is found to be far less active.



(8) Hinnemann, B.; Moses, P. G.; Bonde, J.; Jørgensen, K. P.; Nielsen, J. H.; Horch, S.; Chorkendorff, I.; Nørskov, J. K. J. Am. Chem. Soc. 2005, 127, 5308. (9) Jaramillo, T. F.; Jørgensen, K. P.; Bonde, J.; Nielsen, J. H.; Horch, S.; Chorkendorff, I. Science 2007, 317, 100. (10) Chen, W.-F.; Muckerman, J. T.; Fujita, E. Chem. Commun. 2013, 49, 8896. (11) Chen, W. F.; Sasaki, K.; Ma, C.; Frenkel, A. I.; Marinkovic, N.; Muckerman, J. T.; Zhu, Y.; Adzic, R. R. Angew. Chem., Int. Ed. 2012, 51, 6131. (12) Xiao, P.; Sk, M. A.; Thia, L.; Ge, X.; Lim, R. J.; Wang, J.-Y.; Lim, K. H.; Wang, X. Energy Environ. Sci. 2014, 7, 2624. (13) Huang, Z.; Chen, Z.; Chen, Z.; Lv, C.; Humphrey, M. G.; Zhang, C. Nano Energy 2014, 9, 373. (14) Michalsky, R.; Zhang, Y.-J.; Peterson, A. A. ACS Catal. 2014, 4, 1274. (15) Liao, L.; Wang, S.; Xiao, J.; Bian, X.; Zhang, Y.; Scanlon, M. D.; Hu, X.; Tang, Y.; Liu, B.; Girault, H. H. Energy Environ. Sci. 2014, 7, 387. (16) Vrubel, H.; Hu, X. Angew. Chem., Int. Ed. 2012, 51, 12703. (17) Park, H.; Encinas, A.; Scheifers, J. P.; Zhang, Y.; Fokwa, B. Angew. Chem., Int. Ed. 2017, 56, 5575. (18) Gupta, S.; Patel, N.; Miotello, A.; Kothari, D. J. Power Sources 2015, 279, 620. (19) Xu, N.; Cao, G.; Chen, Z.; Kang, Q.; Dai, H.; Wang, P. J. Mater. Chem. A 2017, 5, 12379. (20) Sitler, S. J.; Raja, K. S.; Charit, I. J. Electrochem. Soc. 2016, 163, H1069. (21) Das, R. K.; Wang, Y.; Vasilyeva, S. V.; Donoghue, E.; Pucher, I.; Kamenov, G.; Cheng, H.-P.; Rinzler, A. G. ACS Nano 2014, 8, 8447. (22) Wan, C.; Regmi, Y. N.; Leonard, B. M. Angew. Chem., Int. Ed. 2014, 53, 6407. (23) Tang, C.; Sun, A.; Xu, Y.; Wu, Z.; Wang, D. J. Power Sources 2015, 296, 18. (24) Chen, Z.; Cummins, D.; Reinecke, B. N.; Clark, E.; Sunkara, M. K.; Jaramillo, T. F. Nano Lett. 2011, 11, 4168. (25) Wu, H. B.; Xia, B. Y.; Yu, L.; Yu, X.-Y.; Lou, X. W. D. Nat. Commun. 2015, 6, 6512. (26) Kiessling, R.; Wetterholm, A.; Sillén, L. G.; Linnasalmi, A.; Laukkanen, P. Acta Chem. Scand. 1947, 1, 893. (27) Frotscher, M.; Klein, W.; Bauer, J.; Fang, C. M.; Halet, J. F.; Senyshyn, A.; Baehtz, C.; Albert, B. Z. Anorg. Allg. Chem. 2007, 633, 2626. (28) Hayami, W.; Momozawa, A.; Otani, S. Inorg. Chem. 2013, 52, 7573. (29) Klesnar, H.; Aselage, T.; Morosin, B.; Kwei, G. J. Alloys Compd. 1996, 241, 180. (30) Ade, M.; Hillebrecht, H. Inorg. Chem. 2015, 54, 6122. (31) Kanatzidis, M. G.; Pöttgen, R.; Jeitschko, W. Angew. Chem., Int. Ed. 2005, 44, 6996. (32) Xie, J.; Zhang, J.; Li, S.; Grote, F.; Zhang, X.; Zhang, H.; Wang, R.; Lei, Y.; Pan, B.; Xie, Y. J. Am. Chem. Soc. 2013, 135, 17881. (33) Lukowski, M. A.; Daniel, A. S.; Meng, F.; Forticaux, A.; Li, L.; Jin, S. J. Am. Chem. Soc. 2013, 135, 10274. (34) Li, Y.; Wang, H.; Xie, L.; Liang, Y.; Hong, G.; Dai, H. J. Am. Chem. Soc. 2011, 133, 7296. (35) Ji, S.; Yang, Z.; Zhang, C.; Liu, Z.; Tjiu, W. W.; Phang, I. Y.; Zhang, Z.; Pan, J.; Liu, T. Electrochim. Acta 2013, 109, 269. (36) Tang, Y. J.; Gao, M. R.; Liu, C. H.; Li, S. L.; Jiang, H. L.; Lan, Y. Q.; Han, M.; Yu, S. H. Angew. Chem., Int. Ed. 2015, 54, 12928. (37) Chung, D. Y.; Park, S.-K.; Chung, Y.-H.; Yu, S.-H.; Lim, D.-H.; Jung, N.; Ham, H. C.; Park, H.-Y.; Piao, Y.; Yoo, S. J.; Sung, Y.-E. Nanoscale 2014, 6, 2131. (38) Kibsgaard, J.; Chen, Z.; Reinecke, B. N.; Jaramillo, T. F. Nat. Mater. 2012, 11, 963. (39) Huang, Y.; Gong, Q.; Song, X.; Feng, K.; Nie, K.; Zhao, F.; Wang, Y.; Zeng, M.; Zhong, J.; Li, Y. ACS Nano 2016, 10, 11337. (40) Zheng, Y.; Jiao, Y.; Jaroniec, M.; Qiao, S. Z. Angew. Chem., Int. Ed. 2015, 54, 52.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b07247. Experimental details (PDF)



AUTHOR INFORMATION

Corresponding Author

*[email protected] ORCID

Boniface P. T. Fokwa: 0000-0001-9802-7815 Author Contributions #

H.P. and Y.Z. contributed equally.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank UC Riverside (startup fund to BPTF) for financial support. We acknowledge the San Diego Supercomputer Center (SDSC) and the High-Performance Computing Center (HPCC) at UC Riverside for providing computing resources. The XPS data were collected with an instrument acquired through the NSF MRI program (DMR-0958796). We are grateful to the Conley group (UC Riverside) for the BET measurements.



REFERENCES

(1) Turner, J. A. Science 2004, 305, 972. (2) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 15729. (3) Wang, J.; Cui, W.; Liu, Q.; Xing, Z.; Asiri, A. M.; Sun, X. Adv. Mater. 2016, 28, 215. (4) Morales-Guio, C. G.; Stern, L.-A.; Hu, X. Chem. Soc. Rev. 2014, 43, 6555. (5) Kibler, L. A. ChemPhysChem 2006, 7, 985. (6) Subbaraman, R.; Tripkovic, D.; Strmcnik, D.; Chang, K.-C.; Uchimura, M.; Paulikas, A. P.; Stamenkovic, V.; Markovic, N. M. Science 2011, 334, 1256. (7) Vesborg, P. C.; Jaramillo, T. F. RSC Adv. 2012, 2, 7933. 12918

DOI: 10.1021/jacs.7b07247 J. Am. Chem. Soc. 2017, 139, 12915−12918

Suggest Documents