Design Principles for Dual-Element-Doped Carbon Nanomaterials as

Feb 3, 2016 - In particular, bifunctional catalysts are desirable for catalyzing ORR/OER in rechargeable metal–air batteries and regenerative fuel c...
0 downloads 0 Views 4MB Size
Letter pubs.acs.org/acscatalysis

Design Principles for Dual-Element-Doped Carbon Nanomaterials as Efficient Bifunctional Catalysts for Oxygen Reduction and Evolution Reactions Zhenghang Zhao and Zhenhai Xia* Department of Materials Science and Engineering, Department of Chemistry, University of North Texas, Denton, Texas 76203, United States S Supporting Information *

ABSTRACT: Dual-element-doped carbon nanomaterials are demonstrated to be more efficient bifunctional catalysts than noble metals to catalyze two key chemical reactions: oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) in regenerative fuel cells and metal−air batteries. Moreover, to accelerate the search for the best metal-free catalysts, an activity descriptor is identified for the codoped carbon nanomaterials, which correlates doping structures to their catalytic activities. These predictions are supported by experimental data. Our work also predicts that the synergistic effect of codoping occurs within a certain distance between the codopants. The descriptor enables rational design of new bifunctional catalysts. KEYWORDS: catalysis, oxygen evolution reaction, oxygen reduction reaction, first-principle modeling, doping, graphene

1. INTRODUCTION Clean energy technologies, such as fuel cells, water splitting, and metal−air batteries, are promising alternative energy sources to power automobiles and portable devices. In particular, Li−air batteries, with the ability to power a car for a 500-mile range per discharge, have the potential to rival traditional gasoline powered engines.1 The photoelectrochemical water splitting system combining with fuel cells could continuously provide electricity day and night with zero greenhouse gas emission and pollution.2 However, these energy technologies require noble metal (oxide) (e.g., Pt, Au, RuO2) catalysts to promote key chemical reactions: oxygen reduction reaction (ORR) and oxygen evolution reaction (OER), which generate and store electricity. The limited resources and high cost of the platinum catalysts have hindered the clean energy technologies for commercial applications. In particular, bif unctional catalysts are desirable for catalyzing ORR/OER in rechargeable metal−air batteries and regenerative fuel cells. Recently N and P codoped graphene and carbon nanotubes (CNTs) have been shown to have high activities in catalyzing both ORR and OER as bifunctional catalysts in Zn-air batteries.3,4 In addition to their abundant sources, N and P codoped graphene is more efficient, more stable, and more tolerant to crossover/CO-poisoning effects than Pt and RuO2, the most active catalysts so far for ORR for OER, respectively. Although the superior bifunctional catalytic capabilities of the codoped carbon nanomaterials have been clearly demonstrated, there is a lack of design principles for searching for the best metal-free bifunctional catalysts from numerous possible © XXXX American Chemical Society

combinations of codoping elements in the periodic table. To rationally design a catalyst, it is critical to understand which intrinsic material characteristics, or descriptors, control the catalytic activity of carbon-based catalysts. Several first-principles studies suggest that the ORR activity on metal surfaces5,6 and N-doped graphene is correlated with ORR elementary steps: adsorption of O2, formation of OOH* and O*, and removal of OH*, where the star refers to adsorption. It is suggested that the ORR activity on different transition metals and N-doped carbons can be described as a function of the adsorption energy of OH*, which yields a volcano curve.7 The best catalyst is therefore found near the summit of the volcano for the ORR. Recently, we proposed a descriptor for describing ORR/OER activities of singleelement-doped carbon-based catalysts.8 In this study, we have, for the first time, identified a new intrinsic materials property that serves as the activity descriptor for accurately predicating bifunctional ORR/OER activities of dual elementdoped carbon-based catalysts. Such descriptor has predictive power to design new metal-free catalysts with enhanced ORR/ OER activities, even better than those reported for platinumbased metal catalysts. As supported by a number of reports for ORR activity of p-block element codoped carbon nanomaterials,7−37 this descriptor can also be used as a powerful guidance Received: December 1, 2015 Revised: January 12, 2016

1553

DOI: 10.1021/acscatal.5b02731 ACS Catal. 2016, 6, 1553−1558

Letter

ACS Catalysis

Figure 1. Free energy and overpotentials of various doped graphene nanoribbons. (A) Schematic of the N-X codoped graphene nanoribbons (X = B, P, S, and Cl), showing the possible positions of dopants. Free-energy diagrams of N−X codoped graphene with the best performance (U0 = 0) for (B) OER and (C) ORR in alkaline medium. (D) Reaction energies of the third electron transfer, ΔG3 (eq S18) vs the second electron transfer, ΔG2, (eq S17) on different sites of N−X codoped armchair and zigzag graphene nanoribbons for OER in alkaline medium. (E) The lower limit of OER/ ORR overpotentials for N-, and X-doped, and N-X codoped graphene structures vs descriptor Φ.

Free-energy diagrams (Figure 1B,C) indicate that the third step of OER (eq S14) and the forth step of ORR (eq S19) are the rate-limiting steps of OER and ORR on the doped structures, respectively. Figure 1D shows the free energy of OER in the third electron transfer: O* + OH− → OOH* + e− (eq S18) versus the second one: OH* + OH− → H2O + e− (eq S17) for N-X codoped graphene structures in alkaline media. The free energies of the third reaction of OER is linearly related to that of the second one by ΔG3 = −ΔG2 + CX, (Figure 1D) where the constant CX depends on dopant type X but is independent of the binding strength to the surface. Combing eq S3 with S4 or S13 with S14, ΔG2 and ΔG3 can be transferred into the binding energies of OH* (ΔGOH*) and OOH* (ΔGOOH*), which are widely used for a scaling relationship plot.6,38−40 Since ΔG2 and ΔG3 are related to overpotential calculation more directly (eqs S23−S26), a lower limit of OER overpotentials (Ulimit) can be determined from the relationship, and the results are listed in Supporting Information Table S1.

to develop various new earth-abundant, efficient catalyst materials.

2. RESULTS Overpotential U can be considered as an electrocatalytic parameter that represents catalytic activities of a catalyst.6 From a thermodynamic point of view, a catalyst with lower U would have better performance. To link an intrinsic material’s properties to catalytic activities of the carbon nanomaterials, we calculated the free energy and overpotential for elementary reactions of ORR/OER (Supporting Information eqs S1−S26) for the possible active sites on graphene structures codoped with p-block elements, N and X (X = N, B, P, S, Cl, etc.), and determined the rate-limiting step (more detailed calculations of free energy and overpotential are shown in Supporting Information). The doping positions in each structure were changed with respect to the graphene edge to reveal the effect of doping sites (Figure 1A, Supporting Information Figure S1). 1554

DOI: 10.1021/acscatal.5b02731 ACS Catal. 2016, 6, 1553−1558

Letter

ACS Catalysis

Figure 2. (A) Predicted average ORR/OER overpotentials of the active sites on N−X codoped graphene (X = P, B, S, and Cl), normalized by the overpotential on the same positions in N−N doped graphene nanoribbons, as a function of the descriptor Φ. (B) Average relative onset potential (onset potential of codoped graphene minus onset potential of Pt/C electrode in the same experiment). (C) Average measured limiting current density from the linear scan voltammogram (LSV) curves, normalized by N-doped carbon electrode current density under the same conditions, as a function of descriptor Φ for N−B,11,13,34 N−P,4,26 N−F,10 N−S,36 and N−Cl33 codoped graphene, and for N−B,13 N−P,19 N−Si,28 and N−S41 codoped CNTs.

Having theoretically identified the lower limits of ORR and OER potentials for carbon-based catalysts under ideal conditions, we turned our attention to “real” codoping structures and the descriptor that determines the “realistic” performance of the catalysts. Using the similar methods, we selected several typical codoping structures and calculated the overpotentials over all the possible sites of N−X codoped graphene structures (Supporting Information Figure S1) and identified the active sites with overpotential below ∼1.0 V (Table S2). For comparison, single element doping (e.g., N−N, P−P, or B−B) with two of the same dopants in the same positions as “co-doping” was also studied, and the overpotentials are also listed in Table S2. Overall, compared with single element doping (e.g., N−N), the overpotentials for the N−X codoping reduce by 10%−40%, depending on the type of codopants X. Interestingly, when normalized by the overpotentials for N−N doping, the overpotentials for N−X codoping versus the descriptor Φ yield an inverted volcano relationship (Figure 2A). In general, the larger the difference in Φ between nitrogen and codopant X is, the lower the overpotential becomes. Therefore, the descriptor Φ can also be used to quantitatively describe the catalytic performance, with the summit of the inverted volcano having the lowest catalytic activity. To verify the theoretical predictions, we have summarized the experimental results on ORR using codoped graphene and CNTs as catalysts, reported in the literatures. Since the onset potential and current density were measured under different conditions in these experiments, it could lead to misleading results when absolute data from different sources are used for

Note that the calculations were made under alkaline condition. If the electric field is neglected, there is no difference in the free energy of the ORR/OER intermediates calculated in acidic and alkaline environment at fixed potential on the RHE scale.40 In our previous work, we have identified an effective ORR/ OER activity descriptor for the single p-block element-doped graphene structures. The descriptor is defined as the product of electronegativity (EX) and electron affinity (AX): Φ = (EX/ EC)*(AX/AC), a dimensionless factor, to represent the effect of these characteristics, where the subscript represents element.8 For codoped graphene, there are active sites related to single N, single X, and N−X clusters. Thus, we calculated all the doping structures related to the single and codoping clusters and chose the lowest value among single N, single X, and N−X doping as the lower limit of the codoping structures. As expected, the lower limit of the overpotentials Ulimit versus descriptor give a volcano shape for single doping (Figure 1E). By contrast, the codoping leads to a very shallow inverted volcano and much lower overpotentials for both ORR and OER compared with single element doping. For example, the lower bound of the overpotential for N−P codoping is 0.355 V, lower than that for single N- or single P-doping. This suggests that there is a synergistic effect in codoping, which significantly enhances the catalytic activities of both ORR and OER. Note that from Figure 1E and Table S1, Ulimit predicted for those dopants are much lower than that of noble metal catalysts (∼0.45 V for ORR on Pt6 and ∼0.42 V for OER on RuO238), indicating the potential of p-block element-doped carbon nanomaterials as efficient bifunctional catalysts for fuel cells and metal-air batteries. 1555

DOI: 10.1021/acscatal.5b02731 ACS Catal. 2016, 6, 1553−1558

Letter

ACS Catalysis

Figure 3. Differential charge density (between codoped and undoped graphene) and overpotential distribution of (A) N−P codoped, (B) P−N codoped, (C) N−N codoped, (D), P−P codoped, (E) single P-doped, (F) single N-doped graphene nanoribbons with zigzag edge. Brown and white balls refer to C and H atoms, respectively. Yellow and blue colors indicate the negative and positive values of electron quantities. The isosurface value is set to 0.0015. The ORR (in red) and OER (in black) overpotentials of the active sites (U < 1.5 V) are also listed on them. Compared with singleelement doping, the dual-element doping leads to more active sites for ORR and OER.

close to each other in graphitic structures (Figure 3A−D, Figure S2), p-electron clouds overlap and interact with each other, generating more active sites on neighboring carbon atoms compared with single-element-doped graphene (Figure 3E,F). In general, when the electronegativity of dopants (e.g., B, P, and S) is smaller than that of N, these dopants carry positive charges, and vice versa (Figure S3 and S4). In any case, these codoping clusters generate positive charges on neighboring carbon atoms that are potentially active sites for ORR and OER. In addition to the types of dopants, the distance between dopants also strongly affects the catalytic activity of the doped carbon. We have calculated the ORR/OER overpotentials of the most active centers as a function of the distance between N and X dopants (Figure 4). The possible dopant locations are schematically shown in Figure S5. Generally speaking, the ORR/OER overpotentials largely vary within a critical distance of dc ∼4 Å, but they gradually converge to a stable level beyond the critical distance. The large variations in ORR/OER within 4 Å can be attributed to the edge effects.7 With increasing the

comparison. To minimize the possible errors from different sources, we use a relative onset potential that is defined as the onset potentials for codoped carbon subtracted by the benchmarked Pt/C electrode onset potential measured under the same condition in the same experiment. Similarly, a normalized current density is defined as the measured current density divided by the N-doped electrode current density under the same condition in the same experiment. Figure 2B,C shows the measured relative onset potential and normalized current density of N−X codoped graphene4,10,11,13,26,33,34,36 and CNTs13,19,28,33 (X = B, P, Si, S, Cl, and F), respectively. Both the relative onset potential and the normalized current show inverted volcanoes similar to the theoretical predictions. Thus, our predictions are consistent with the experimental data. The same descriptor Φ for single doping can be used for describe the catalytic behavior of the codoped carbon-based bifunctional catalysts, but it yields an inverted volcano relationship. The improved ORR/OER activity of these codoped carbonbased catalysts can be attributed to the interactions of dual dopants. When two heteroatoms (e.g., N and P) are doped 1556

DOI: 10.1021/acscatal.5b02731 ACS Catal. 2016, 6, 1553−1558

Letter

ACS Catalysis

Figure 4. (A) ORR and (B) OER minimum overpotentials of N−X codoped graphene (X = N, B, P, S, and Cl), normalized by the corresponding ORR/OER overpotentials for single X-doped graphene at the same active sites, versus the distance between dopants N and X.

distance between the dopants, ORR/OER overpotentials for (N−N) doping, keep a small variation, but those for (N−X) doping changes significantly. While the overpotentials for N−P and N−Cl doping decreases, those for N−S and N−B doping increase with increasing the distance. These results indicate that to obtain larger synergistic effect, the distance between N and Cl (ΦCl > ΦN) should be kept large, while the N and B (ΦB < ΦN) should be located more closely. The synergistics effects between the dopants identified above seem to relate well to the adsorption energies of intermediates O*, OH*, and OOH* on the active sites (Figure S6, S7). For example, adsorption energies of these intermediates for N−Cl doping decrease, but the energies increase for N−B doping with increasing the distance, which are consistent with the trends of ORR/OER overpotentials. Since the valence orbital of each active center participates in the bond formation with an oxygen-containing intermediate on the graphene surface (e.g., OH*), the valence orbital level and their interactions should greatly influence its adsorption energy ΔGOH*,38 and thus catalytic activities. Overall, the synergistic interactions become stronger when the difference in descriptor (Φ) between two dopants (N and X) is larger. From the above analysis on the codoped carbon nanomaterials, three design principles for enhancing the catalytic activities can be drawn: (i) Choose codopants with relatively different chemical properties (e.g., electronegativity); (ii) Keep the dopants close to each other if the descriptor number (Φ) of one dopant is much smaller than N, and vice versa; and (iii) Make use of edge effect by doping the element near edges of the graphene. In conclusion, we have identified an intrinsic descriptor that can accurately describe the ORR/OER activities of dualelement-codoped carbon nanomaterials. The ORR and OER activities are directly related to the descriptor in an inverted volcano shape, which is verified by the experimental data. These results can be explained by synergistic interactions of pelectrons between codopants, which reduces the overpotentials and stabilizes the adsorbates, and hence the fast ORR/OER kinetics. This work shows that codoping is a promising strategy in developing highly active metal-free carbon-based bifunctional

catalysts for ORR and OER in fuel cells, metal−air batteries, and water-splitting systems.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.5b02731. Materials and methods, electron affinity and electronegativity data for elements, overpotentials, and other supplement data (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work is supported financially by Air Forces MURI program for the support of this research under the contract no. FA9550-12-1-0037 and the National Science Foundation (NSF-AIR-IIP-1343270, NSF-CMMI-1363123). Computational resources were provided by UNT high performance computing initiative, a project of academic computing and user services within the UNT computing and information technology center.



REFERENCES

(1) Graham-Rowe, D. New Sci. 2012, 213 (2846), 18. (2) Gabbasa, M.; Sopian, K.; Fudholi, A.; Asim, N. Int. J. Hydrogen Energy 2014, 39, 17765−17778. (3) Zhang, J.; Zhao, Z.; Xia, Z.; Dai, L. Nat. Nanotechnol. 2015, 10, 444−452. (4) Li, R.; Wei, Z.; Gou, X. ACS Catal. 2015, 5, 4133. (5) Calle-Vallejo, F.; Koper, M. T. M. Electrochim. Acta 2012, 84, 3− 11. (6) Nørskov, J. K.; Rossmeisl, J.; Logadottir, a.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jónsson, H. J. Phys. Chem. B 2004, 108, 17886− 17892.

1557

DOI: 10.1021/acscatal.5b02731 ACS Catal. 2016, 6, 1553−1558

Letter

ACS Catalysis (7) Li, M.; Zhang, L.; Xu, Q.; Niu, J.; Xia, Z. J. Catal. 2014, 314, 66− 72. (8) Zhao, Z.; Li, M.; Zhang, L.; Dai, L.; Xia, Z. Adv. Mater. 2015, 27, 6834−6840. (9) Sun, X.; Song, P.; Chen, T.; Liu, J.; Xu, W. Chem. Commun. (Cambridge, U. K.) 2013, 49, 10296−10298. (10) Sun, X.; Song, P.; Zhang, Y.; Liu, C.; Xu, W.; Xing, W. Sci. Rep. 2013, 3, 2505. (11) Wang, L.; Yu, P.; Zhao, L.; Tian, C.; Zhao, D.; Zhou, W.; Yin, J.; Wang, R.; Fu, H. Sci. Rep. 2014, 4, 5184. (12) Wang, S.; Zhang, L.; Xia, Z.; Roy, A.; Chang, D. W.; Baek, J.-B.; Dai, L. Angew. Chem., Int. Ed. 2012, 51, 4209−4212. (13) Xue, Y.; Yu, D.; Dai, L.; Wang, R.; Li, D.; Roy, A.; Lu, F.; Chen, H.; Liu, Y.; Qu, J. Phys. Chem. Chem. Phys. 2013, 15, 12220. (14) Yang, D.-S.; Bhattacharjya, D.; Inamdar, S.; Park, J.; Yu, J.-S. J. Am. Chem. Soc. 2012, 134, 16127−16130. (15) Yang, L.; Jiang, S.; Zhao, Y.; Zhu, L.; Chen, S.; Wang, X.; Wu, Q.; Ma, J.; Ma, Y.; Hu, Z. Angew. Chem. 2011, 123, 7270−7273. (16) Yang, S.; Zhi, L.; Tang, K.; Feng, X.; Maier, J.; Müllen, K. Adv. Funct. Mater. 2012, 22, 3634−3640. (17) Yang, Z.; Yao, Z.; Li, G.; Fang, G.; Nie, H.; Liu, Z.; Zhou, X.; Chen, X.; Huang, S. ACS Nano 2012, 6, 205−211. (18) Yao, Z.; Nie, H.; Yang, Z.; Zhou, X.; Liu, Z.; Huang, S. Chem. Commun. (Cambridge, U. K.) 2012, 48, 1027−1029. (19) Yu, D.; Xue, Y.; Dai, L. J. Phys. Chem. Lett. 2012, 3, 2863−2870. (20) Zhang, C.; Mahmood, N.; Yin, H.; Liu, F.; Hou, Y. Adv. Mater. 2013, 25, 4932−4937. (21) Zhang, L.; Xia, Z. J. Phys. Chem. C 2011, 115, 11170−11176. (22) Zhang, Y.; Chu, M.; Yang, L.; Deng, W.; Tan, Y.; Ma, M.; Xie, Q. Chem. Commun. (Cambridge, U. K.) 2014, 50, 6382−6385. (23) Chen, Z.; Higgins, D.; Chen, Z. Carbon 2010, 48, 3057−3065. (24) Geng, D.; Chen, Y.; Chen, Y.; Li, Y.; Li, R.; Sun, X.; Ye, S.; Knights, S. Energy Environ. Sci. 2011, 4, 760. (25) Jeon, I.-Y.; Choi, H.-J.; Choi, M.; Seo, J.-M.; Jung, S.-M.; Kim, M.-J.; Zhang, S.; Zhang, L.; Xia, Z.; Dai, L.; Park, N.; Baek, J.-B. Sci. Rep. 2013, 3, 1810. (26) Jiang, H.; Zhu, Y.; Feng, Q.; Su, Y.; Yang, X.; Li, C. Chem. - Eur. J. 2014, 20, 3106−3112. (27) Jin, Z.; Nie, H.; Yang, Z.; Zhang, J.; Liu, Z.; Xu, X.; Huang, S. Nanoscale 2012, 4, 6455−6460. (28) Liu, Z.; Fu, X.; Li, M.; Wang, F.; Wang, Q.; Kang, G.; Peng, F. J. Mater. Chem. A 2015, 3, 3289−3293. (29) Liu, Z.-W.; Peng, F.; Wang, H.-J.; Yu, H.; Zheng, W.-X.; Yang, J. Angew. Chem. 2011, 123, 3315−3319. (30) Park, M.; Lee, T.; Kim, B.-S. Nanoscale 2013, 5, 12255−12260. (31) Qu, L.; Liu, Y.; Baek, J.-B.; Dai, L. ACS Nano 2010, 4, 1321− 1326. (32) Sheng, Z.-H.; Gao, H.-L.; Bao, W.-J.; Wang, F.-B.; Xia, X.-H. J. Mater. Chem. 2012, 22, 390−395. (33) Shi, J.; Fan, M.; Qiao, J.; Liu, Y. Chem. Lett. 2014, 43, 1484− 1486. (34) Zheng, Y.; Jiao, Y.; Ge, L.; Jaroniec, M.; Qiao, S. Z. Angew. Chem., Int. Ed. 2013, 52, 3110−3116. (35) Chen, S.; Duan, J.; Jaroniec, M.; Qiao, S.-Z. Adv. Mater. 2014, 26, 2925−2930. (36) Liang, J.; Jiao, Y.; Jaroniec, M.; Qiao, S. Z. Angew. Chem., Int. Ed. 2012, 51, 11496−11500. (37) Qu, K.; Zheng, Y.; Dai, S.; Qiao, S. Z. Nano Energy 2016, 19, 373−381. (38) Man, I. C.; Su, H. Y.; Calle-Vallejo, F.; Hansen, H. a.; Martínez, J. I.; Inoglu, N. G.; Kitchin, J.; Jaramillo, T. F.; Nørskov, J. K.; Rossmeisl, J. ChemCatChem 2011, 3, 1159−1165. (39) Rossmeisl, J.; Logadottir, A.; Nørskov, J. K. Chem. Phys. 2005, 319, 178−184. (40) Rossmeisl, J.; Qu, Z.-W.; Zhu, H.; Kroes, G.-J.; Nørskov, J. K. J. Electroanal. Chem. 2007, 607, 83−89. (41) Shi, Q.; Peng, F.; Liao, S.; Wang, H.; Yu, H.; Liu, Z.; Zhang, B.; Su, D. J. Mater. Chem. A 2013, 1, 14853.

1558

DOI: 10.1021/acscatal.5b02731 ACS Catal. 2016, 6, 1553−1558