Theoretical insights on the reaction pathways for oxygen reduction reaction on phosphorus doped graphene

Theoretical insights on the reaction pathways for oxygen reduction reaction on phosphorus doped graphene

Accepted Manuscript Theoretical insights on the reaction pathways for oxygen reduction reaction on phosphorus doped graphene Xiaowan Bai, Erjun Zhao, ...

2MB Sizes 0 Downloads 38 Views

Accepted Manuscript Theoretical insights on the reaction pathways for oxygen reduction reaction on phosphorus doped graphene Xiaowan Bai, Erjun Zhao, Kai Li, Ying Wang, Menggai Jiao, Feng He, Xiaoxu Sun, He Sun, Zhijian Wu PII:

S0008-6223(16)30294-9

DOI:

10.1016/j.carbon.2016.04.033

Reference:

CARBON 10914

To appear in:

Carbon

Received Date: 29 January 2016 Revised Date:

6 April 2016

Accepted Date: 14 April 2016

Please cite this article as: X. Bai, E. Zhao, K. Li, Y. Wang, M. Jiao, F. He, X. Sun, H. Sun, Z. Wu, Theoretical insights on the reaction pathways for oxygen reduction reaction on phosphorus doped graphene, Carbon (2016), doi: 10.1016/j.carbon.2016.04.033. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1 2

Theoretical insights on the reaction pathways for oxygen reduction reaction on

3

phosphorus doped graphene

4

Xiaowan Bai,a,b Erjun Zhao,a,* Kai Li,b Ying Wang,b Menggai Jiao,b Feng He,b Xiaoxu

6

Sun,b He Sun,c Zhijian Wub,*

RI PT

5

7 8

a

9

China

SC

College of Science, Inner Mongolia University of Technology, Hohhot 010051, P. R.

10

b

11

Applied Chemistry, Chinese Academy of Sciences, Changchun 130022, P. R. China

12

c

M AN U

State Key Laboratory of Rare Earth Resource Utilization, Changchun Institute of

The Computing Center of Jilin Province, Changchun 130012, P. R. China

13

Abstract

15

The reaction mechanisms for oxygen reduction reaction (ORR) on phosphorus doped

16

divacancy graphene (P-GDV) are investigated by using the density functional theory

17

method. Our results showed that all of the possible ORR elementary reactions could

18

take place within a small region around the P atom and its adjacent four carbon atoms.

19

The hydrogenation of O2 molecule which forms OOH and hydrogenation of OOH

20

which forms H2O+O have negligible energy barrier. This reaction pathway is also the

21

kinetically most favorable. The rate-determining step is the final step in the pathway,

22

i.e., the hydrogenation of OH into H2O with an energy barrier of 0.85 eV. Therefore,

23

ORR mechanism on P-GDV would be a four electron process. The free energy diagram

24

of the ORR predicted that for the most favorable pathway, the working potential is

25

0.27 V. Consequently, our theoretical study suggests that P doped graphene with

26

intrinsic carbon defects could possess good catalytic activity for ORR.

AC C

EP

TE D

14

27 28 29

* Corresponding authors. E-mail: [email protected] (EJZ); [email protected] (ZJW).

30 1

ACCEPTED MANUSCRIPT 1

1. Introduction Fuel cells (FCs), as highly efficient and environmentally friendly energy

3

converting devices, have been studied extensively in the past decades. For FCs, the

4

electrons from the cathode to anode which makes oxygen reduction reaction (ORR)

5

play a crucial role in energy conversion efficiency [1,2]. Since ORR is a kinetically

6

sluggish process on the cathode [3,4], the cathode catalysts are important for the

7

energy conversion efficiency, operating cost, durability of the FCs, etc.. Up to now,

8

the best performing cathode catalysts for ORR are Pt or Pt-based alloy electrocatalysts

9

[5-8]. However, the high cost, poor durability and stability of Pt hinder its widespread

10

commercial applications. Therefore, it is indispensable to search for low cost and high

11

efficiency electrocatalysts, aiming to substitute Pt or Pt-based catalysts. Recently,

12

heteroatom-doped graphene catalysts, such as non-noble metal [9-12] and metal-free

13

[13-18] doped graphene, attract a great deal of attention due to their low cost and

14

good performance for ORR.

SC

M AN U

15

RI PT

2

Among the heteroatom-doped graphene, N-doping is the most popular choice as the catalytic substrate [19-21] or direct catalyst [22-24]. For instance, Xiong et al. [20]

17

found that N-doped graphene as substrate could improve the durability of Pt catalyst

18

and exhibit higher catalytic activity for CO oxidation. As a direct catalyst, Chai et al.

19

[22] considered the various structures in nitrogen doped carbon catalysts (CACs). The

20

results indicated that N-doped Stone-Wales (SW) defect graphene structure provides

21

good active sites for ORR. Recently, most of the theoretical [10-12] and experimental

22

[25-27] investigations have shown that nitrogen-coordinated transition metals (e.g.,

23

Mn, Fe, and Co) in graphene exhibit also good ORR catalytic performance. These

24

results demonstrated that N-doped graphene can be acted as the alternatives for Pt in

25

ORR. Besides N-doped graphene, researches on graphene doped by other heteroatoms

26

are also available, such as boron (B) [28], sulfur (S) [29,30], selenium (Se) [31] and

27

their mixtures [32,33], etc.. In particular, due to the similarity of nitrogen and

28

phosphorus in structure and chemical properties, phosphorus (P) doped graphene

29

received a great deal of research interests. Both the experimental [34-36] and

30

theoretical studies [37] confirmed that P-doped graphene could be used as a high

AC C

EP

TE D

16

2

ACCEPTED MANUSCRIPT efficient metal-free ORR electrocatalyst. On experimental aspect, Liu et al. [34]

2

synthesized P-doped graphite layers through the thermolysis method and pointed out

3

that P-doped graphite could exhibit high electrocatalytic activity, long-term stability

4

and strong durability in ORR. Li et al. [35] found that doping by P atoms could create

5

new active sites on graphene and the activity is comparable to the commercial

6

benchmark Pt/C catalyst for ORR. On the theoretical aspect, Kaukonen et al. [37] has

7

studied the formation energies for P atoms embedded in graphene with monovacancy

8

and divacancy and adsorption energies of O2 and H2O on the P-doped graphene. The

9

results showed that single P doped graphene with divacancy could be a good

10

candidate for ORR and the weak O2 binding is a main criteria for its good

11

performance [37]. For ORR mechanism on P-doped graphene with monovacancy,

12

Zhang et al. [38] demonstrated that P dopants are active sites and OOH formation and

13

dissociation is the most favorable pathway with the rate-limiting step of OH

14

hydrogenation to H2O. Recently, experimental researches suggested that significant

15

contribution of intrinsic carbon defects could break the symmetry of the carbon

16

framework and promote ORR activity [18,39,40].

TE D

M AN U

SC

RI PT

1

In this work, we have studied the graphene doped by single P atom. By replacing

18

one C atom with P, we obtained three configurations, i.e., P-doped monovacancy

19

graphene (P-GMV) , P-doped divacancy graphene (P-GDV) and P-doped graphene with

20

Stone-Wales defect (P-GSW) (Fig. S1, Supporting Information). The calculated

21

formation energy indicated that P-GDV is the most stable with the formation energy

22

-4.58 eV. While for P-GMV and P-GSW, the formation energy is positive, indicating

23

that they are unstable. Therefore, P-GDV is selected for further study in this work.

24

Besides ORR mechanism, we have also investigated that the effects of the electrode

25

potentials on the activity in acid environment. The calculated results revealed that the

26

phosphorus coordinated intrinsic defect could improve the catalytic activity of

27

graphene for ORR.

AC C

EP

17

28 29 30

2. Computational details All the geometrical and energetic calculations are performed using Vienna ab 3

ACCEPTED MANUSCRIPT initio simulation package (VASP) [41-44]. The Perdew-Burke-Ernzerhof (PBE)

2

exchange-correlation functional within the generalized gradient approximation (GGA)

3

is employed [45], while the Blöchl’s all-electron-like projector augmented wave

4

(PAW) method [46,47] is adopted to describe the interactions between ion cores and

5

valence electrons. The wave functions at each k-point are expanded with a plane wave

6

basis set and the kinetic cutoff energy is set to be 400 eV. The electron occupancies

7

are determined according to the Fermi scheme with an energy smearing of 0.1 eV. The

8

Brillouin zone integration is approximated by a sum over special selected k-points

9

generated from 5×5×1 Monkhorst–Pack method [48]. The convergence tolerance of

10

energy of 1.0×10-5 eV/atom and maximum force of 0.05 eV/Å are employed in all the

11

geometry optimizations. The reaction pathways and the transition state (TS) structures

12

are located using the climbing image nudged elastic band (CI-NEB) method [49]. The

13

minimum energy path is optimized using a force-based conjugate-gradient method [43]

14

until the maximum force is less than 0.05 eV/Å. In order to describe the

15

van-der-Waals (vdW) interactions between the reactants and the substrate, a

16

semiempirical dispersion-corrected density functional theory (DFT-D2) force-field

17

approach [50,51] is used in our calculations.

TE D

M AN U

SC

RI PT

1

A 4×4 graphene hexagonal supercell (containing 32 atoms) with lattice parameters

19

a= b = 9.84 Å are chosen to model the P-doped graphene. A vacuum layer of 12 Å is

20

chosen along the C axis direction normal to the sheet to avoid periodic interactions.

21

For P-GDV, P atom prefers to be anchored at the center of divacancy with the four

22

nearest carbon atoms having the same P-C bond distance (1.85 Å). This is in

23

agreement with the previous study [37].

25

AC C

24

EP

18

The formation energy (∆Ef) is calculated as:

(

∆E f = EP−GDV + µC − EGDV + µP

)

(1)

26

where E P −GDV and EGDV are the total energies for optimized with P-doped and

27

without P-doped divacancy graphene structures, respectively. µC is the chemical

28

potential of carbon in graphene [52,53]. µP is the chemical potential of the 4

ACCEPTED MANUSCRIPT 1

phosphorus in the bulk phase of black phosphorus. The negative formation energy

2

means that the P-doped graphene is stable.

3

The adsorption energy (∆Eads) is calculated as:

(

)

4

∆Eads = Eadsorbate/ P−GDV − Eadsorbate + EP −GDV

5

where Eadsorbate/ P −GDV , Eadsorbate and EP −GDV are the total energies of the P-GDV

6

structure with adsorbate, free adsorbate and P-GDV structure, respectively. All

7

energies are calculated by using the same periodic box and parameter setting.

RI PT

(2)

Since the cathode electrocatalysts work under a positive electricpotential in reality,

9

the free energy changes of the ORR are studied under different electrode potentials.

11

The change in free energy (∆G) of per reaction step is calculated as [54]:

∆G = ∆E + ∆ZPE − T ∆S + ∆GU + ∆G pH + ∆G field

M AN U

10

SC

8

(3)

where ∆E is the change of the total reaction energy obtained from DFT calculations,

13

∆ZPE is the change of the zero-point energy, T is the temperature (298.15 K), and

14

∆S is the change of the entropy. ∆GU = -eU, here, U is the potential at the

15

electrode and e is the transferred charge. ∆G pH = kBT×ln10×pH where kB is the

16

Boltzmann constant and T=300 K. We define pH = 0 for acidic medium [55,56].

17

∆G field = 0 where ∆G field is the free energy correction due to the electrochemical

18

double layer and is neglected as in previous study [54]. The free energy of O2 is

19

obtained from the reaction O2 + 2H2→2H2O, which is 4.92 eV at temperature of

20

298.15 K and pressure of 0.035 bar. The entropies and vibrational frequencies of

21

molecules (including O2, H2, H2O, etc.) in the gas phase are taken from the NIST

22

database. The zero-point energy and the entropies of the possible adsorbed species are

23

calculated from the vibrational frequencies.

AC C

EP

TE D

12

5

ACCEPTED MANUSCRIPT

2

RI PT

1

Fig. 1 Geometrical structure of P-GDV and possible adsorption sites.

4

3. Results and discussion

5

3.1 The adsorption of reaction species

SC

3

The possible adsorption configurations of the various reaction species on P-GDV

7

involved in ORR mechanism, including O2, O+O, O, H, OH, OOH, O+OH, OH+OH,

8

and H2O have been studied first. The surface structure with possible adsorption sites

9

on P-GDV is shown in Fig. 1. After optimization of the reaction species on these

10

possible sites, the obtained stable structures are shown in Fig. S2, Supporting

11

Information. The most stable configurations and the calculated adsorption energies

12

(∆Eads) for each species are listed in Fig. 2. Generally, adsorption of O2 molecule has

13

two configurations, i.e., end-on and side-on. In our work, the end-on configuration

14

(Fig. 2a) is more stable than side-on configuration (with positive ∆Eads = 0.34 eV, Fig.

15

S2). Thus, the end-on O2 is selected as the initial state in ORR in the following

16

calculations. This is different from the case in P-GMV, in which side-on configuration

17

is found to be the most stable [38]. This indicates that the graphene defect type has

18

great influence on the adsorption behavior of O2 molecule. For the two separated O

19

atoms, the most stable configuration is that one O is on the top of P atom, while the

20

other one sits on the bridge site of P-C bond, similar to the case in P-GMV [38]. For

21

the single O atom, the bridge site of P-C is the most stable site, while the site on the

22

top of P atom is slightly unstable (by about 0.33 eV, Fig. S2, Supporting Information).

23

For H, OH and OOH, the most stable site is on the top of P atom (Fig. 2d-f). For H,

24

the second most stable site is on the C atoms adjacent to P atom, which is slightly

AC C

EP

TE D

M AN U

6

6

EP

1

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 2 Atomic structures of the most stable geometries for various ORR chemical

3

species adsorbed. ∆Eads is the adsorption energy. (a) end-on O2, (b) two O atoms, (c)

4

atomic O, (d) atomic H, (e) OH, (f) OOH, (g) atomic O and OH co-adsorption, (h)

5

two OH co-adsorption, and (i) H2O. In the figure, the gray, pink, red, and white balls

6

represent C, P, O, and H atoms, respectively.

AC C

2

7 8

lower by 0.08 eV (Fig. S2, Supporting Information). Therefore, the C atoms adjacent

9

to P atom are the important active site for the hydrogenation in study of ORR

10

mechanism. In the searching of HOOH, our calculations indicate that the

11

hydrogenation of OOH always gives H2O+O. The product of both OH+OH and H2O2 7

ACCEPTED MANUSCRIPT 1

can not be found. Thus, the ORR mechanism on P-GDV could be mainly a four

2

electron process, in agreement with the experimental observations [35,57]. From the calculated adsorption energy, it is seen that O2 molecule only adsorbs at

4

P top site with a relatively small adsorption energy (∆Eads = -0.19 eV). This result is

5

similar to the pervious theoretical result for P-GDV (< 0.1 eV) [37]. The smaller

6

adsorption energy for O2 molecule is also found on Pt/Cu(111) surface (-0.14 eV, Ref.

7

8). According to the Sabatier principle, the adsorption energy of O2 on an ideal

8

catalyst for ORR should be as small as possible to avoid strong adsorption of O2 [58].

9

Therefore, we expect that P-GDV will be a suitable candidate as a good catalyst for

SC

ORR due to its weak adsorption of O2 [37,58].

AC C

EP

TE D

M AN U

10

RI PT

3

11 12

Fig. 3 Possible reaction pathways for ORR. The numbers in parentheses are the 8

ACCEPTED MANUSCRIPT 1

energy barrier and reaction heat in units of eV. For the details of labels from a1 to f1,

2

see Fig. 4; a1' to c1' see Fig. 5; a1" to b1" see Fig. 6. * denotes that the ORR species is

3

adsorbed on the catalyst surface.

4

3.2 The ORR mechanism In ORR mechanism, the chemisorption of O2 on P-GDV is the first necessary step

6

to initialize the reaction. Following the adsorption of O2 molecule on P-GDV, there are

7

two possible reaction pathways, i.e., the dissociation of O2 into two separated O atoms

8

and hydrogenation of O2 into OOH species. During our research, we have found

9

several possible pathways for ORR as shown in Fig. 3. In the following, we shall

SC

10

RI PT

5

describe these reaction pathways in detail.

O2 dissociation. In the first step, the O2 molecule adsorbs on the surface with

12

end-on configuration (Fig. 4). The O-O bond distance is 1.31 Å, slightly larger than

13

its bond distance 1.22 Å in free gas O2 molecule. In transition state, the O-O distance

14

enlarges to 1.86 Å. At the final state, one O atom is adsorbed on top site of the P atom

15

and the other O atom sits on bridge site of P-C bond with O-O bond distance 2.44 Å.

16

The O2→O+O reaction is an exothermic process with an energy of -2.49 eV and an

17

energy barrier of 0.81 eV. This energy barrier is larger than 0.38 eV for O2

18

dissociation on P-GMV [38]. Following the O2 dissociation, the O atom at P top site

19

will be hydrogenated to form O+OH. The energy barrier is 0.64 eV with an

20

exothermic energy of -1.37 eV. The O-H bond distance is 1.46 Å in transition state.

21

After the formation of O+OH, its hydrogenation will give either H2O+O or OH+OH.

TE D

EP

The formation of H2O+O requires an energy barrier of 0.86 eV and the reaction is

AC C

22

M AN U

11

23

exothermic by -0.71 eV (Fig. 4). Once the weak adsorbed H2O molecule is released

24

from P top site, the remaining O at P-C bridge site could diffuse to P top site with a

25

diffusion barrier of 0.77 eV. In transition state, the O-P bond distance is 1.61 Å.

26

Subsequently, it will be hydrogenated to give OH, then the further hydrogenation of

27

OH forms H2O. After the release of the H2O molecule, the catalyst will be refreshed

28

and another cycle of reaction would begin. The reaction of O+H→OH is exothermic

29

by -0.77 eV and the energy barrier is 0.54 eV. The hydrogenation of OH gives H2O

30

with a slightly higher barrier of 0.85 eV and less exothermic energy of -0.29 eV. In 9

ACCEPTED MANUSCRIPT 1 2

Although the hydrogenation of OOH does not produce OH+OH as mentioned earlier, OH+OH species can be generated from the hydrogenation of O+OH. To form

SC

RI PT

3

transition state, the H-O bond distance is 1.24 Å.

6

AC C

EP

5

TE D

M AN U

4

7 10

RI PT

ACCEPTED MANUSCRIPT

M AN U

SC

1

2

Fig. 4 Atomic structures of the initial state (left panel), transition state (middle panel),

4

and final state (right panel) for O2 dissociation (a1), atomic O hydrogenation (b1), first

5

H2O formation (c1), the diffusion of O atom from P bridge site to top site (d1), second

6

OH formation (e1), and second H2O formation (f1).

TE D

3

7

OH+OH, the very large reaction barrier is required (1.95 eV) with less exothermic by

9

-0.10 eV (Fig. 5), indicating that this reaction pathway is difficult. In OH+OH, the

10

most stable configuration is that one OH is on the top of P atom, while the other OH

11

on the top of C atom (adjacent to P atom) (Fig. 2h). After OH+OH is generated, there

12

are two reaction pathways. The first one is the hydrogenation of OH+OH gives

13

OH+H2O with a relatively lower energy barrier of 0.58 eV. This process has large

14

exothermic energy of -1.95 eV. After the release of H2O, the hydrogenation of OH, i.e.,

15

H+OH, will produce another H2O molecule. The reaction has a barrier of 0.85 eV as

16

mentioned above (Fig. 4). For the second reaction OH+OH→H2O+O, the energy

17

barrier is 0.37 and the reaction is exothermic by -0.27 eV (Fig. 5). This means that

18

this reaction is relatively easy compared with OH+OH→OH+H2O. After the

19

desorption of H2O, the two sequential hydrogenations of O give water as discussed in

20

Fig. 4.

AC C

EP

8

11

1

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 5 Atomic structures of the initial state (left panel), transition state (middle panel),

3

and final state (right panel) for two OH formation (a1'), OH hydrogenation to H2O (b1')

4

and two OH disproportionation (c1').

6

AC C

5

EP

2

O2 hydrogenation. In this reaction pathway, the hydrogenation of O2 is very easy

7

due to the negligible reaction barrier (Fig. 6). Similar behavior is also observed for

8

FeN4 doped graphene [10]. This is in agreement with the experimental observation

9

that peroxide has been formed [34,59]. The conclusion is also different from the result

10

on P-GMV, where the hydrogenation of O2 molecule needs a barrier of 0.45 eV [38].

11

After the formation of OOH, the reaction will proceed in two pathways (Fig. 3). The

12

first reaction is the dissociation of OOH. After OOH dissociation, OH is located on

13

the top of P atom, while O is on the bridge site of P-C bond (Fig. 6). The reaction is 12

ACCEPTED MANUSCRIPT exothermic by -2.58 eV with an energy barrier of 0.83 eV. This barrier is close to the

2

one for O2 dissociation (0.81 eV) (Fig. 4). The following reactions from O+OH is the

3

same as mentioned in O2 dissociation (Figs. 3, 4). The second reaction is the

4

hydrogenation of OOH. The product is H2O+O, in which O atom is on the bridge site

5

of P-C bond (Fig. 6). The calculated energy barrier is 0.04 eV. This means that the

6

hydrogenation of OOH is a spontaneous process. After the release of H2O, the

7

hydrogenation of the remaining O produces the second water, as mentioned in Fig. 4.

TE D

M AN U

SC

RI PT

1

EP

8

Fig. 6 Atomic structures of the initial state (left panel), transition state (middle panel),

10

and final state (right panel) for OOH dissociation (a1") and OOH hydrogenation to

11

H2O (b1").

12 13

AC C

9

In a word, it is seen from the reaction pathways shown in Fig. 3 that for O2

14

dissociation, the favorable pathway is the process a1→b1→c1→d1→e1→f1 with the

15

rate-determining step of H2O+O formation (energy barrier 0.86 eV). Nonetheless, we

16

also noticed that O2 dissociation and OH hydrogenation to form H2O have also similar

17

energy barriers (i.e., 0.81 and 0.85 eV, respectively). On the other hand, the formation

18

of OH+OH is not preferred due to the very large energy barrier (1.95 eV). For the O2 13

ACCEPTED MANUSCRIPT 1

hydrogenation,

2

a1"→c1→d1→e1→f1,

3

a1"→c1→d1→e1→f1, the dissociation of OOH has a barrier of 0.83 eV. The

4

rate-determining step is the same as a1→b1→c1→d1→e1→f1 for H2O+O formation

5

with an energy barrier 0.86 eV. For the pathway b1"→d1→e1→f1, however, the

6

formation of OOH and H2O+O is a spontaneous process for ORR with negligible

7

barrier, suggesting that OOH hydrogenation is more easily than OOH dissociation. In

8

this pathway, the rate-determining step is the hydrogenation of OH with a barrier of

9

0.85 eV. The similar energy barriers for a1"→c1→d1→e1→f1 (0.86 eV) and

are

favorable

the

second

pathways. one

The

is

first

one

b1"→d1→e1→f1.

is In

RI PT

while

two

SC

10

there

b1"→d1→e1→f1 (0.85 eV) indicated that they are competitive pathways. The above results indicated that the O2 hydrogenation is much easier than O2

12

dissociation. On the whole, the most favorable reaction pathway would be

13

b1"→d1→e1→f1. The energy barrier of 0.85 eV is similar to that of 0.88 eV on P-GMV

14

[38] and 0.86 eV on Pt (111) surface [8]. Atomic O will be the main species on the

15

catalyst surface. Therefore, our study revealed that the doped P atom could improve

16

the catalytic activity of graphene for the ORR. Recalling that for P-GMV, the most

17

favorable pathway is the dissociation of OOH [38], different from our conclusion.

18

This demonstrated that the defect type in graphene has great influence on reaction

19

pathways due to the different geometries and the active sites.

20

3.3 Effect of electrode potentials on ORR

EP

TE D

M AN U

11

Since the above results are from zero electrode potential, while in reality, the

22

electrochemical systems are operated under positive electrode potentials. Therefore,

23

we have studied the reaction pathways under different electrode potentials. We only

24

give

25

b1"→d1→e1→f1 in Fig. 7, the remaining pathways are shown in Fig. S3, Supporting

26

Information. From Fig. 7, it is seen that at zero potential, all the reactions are

27

exothermic since all the elementary reaction steps are downhill. With the increase of

28

the electrode potentials, some intermediate reactions become less exothermic and

29

there exists the highest electrode potential to keep all the elementary reactions to be

30

exothermic, which is defined as the working potential of the electrocatalyst. For

AC C

21

the

pathways

a1→b1→c1→d1→e1→f1,

14

a1"→c1→d1→e1→f1

and

ACCEPTED MANUSCRIPT instance, in the pathway a1→b1→c1→d1→e1→f1, the working potential is 0.10 V. For

2

potentials higher than 0.10 V, endothermic reaction is found, e.g., the first H2O

3

formation is endothermic at V=0.20 V, while at V=1.23 V, all the reactions are

4

endothermic after the formation of O+OH. Similarly for the pathways

5

a1"→c1→d1→e1→f1 and b1"→d1→e1→f1, the working potentials are 0.10 V and

6

0.27 V, respectively. For the remaining pathways (Fig. S3), the reaction step is uphill

7

for the formation of OH+OH, consistent with its large energy barrier (Fig. 3).

AC C

EP

TE D

M AN U

SC

RI PT

1

8 15

ACCEPTED MANUSCRIPT 1

Fig. 7 The free energy diagrams for the reduction of O2 to H2O at different electrode

2

potential

3

a1"→c1→d1→e1→f1 and b1"→d1→e1→f1; see also Fig. 3 for the detailed reaction

4

pathways. * denotes that the ORR species is adsorbed on the catalyst surface.

5

4. Conclusions

O2

dissociation,

a1→b1→c1→d1→e1→f1;

O2

hydrogenation,

RI PT

U.

DFT computations show that the formation of P-doped graphene with divacancy

7

is energetically favorable compared with monovacancy and Stone Wales defect. By

8

using P-doped graphene with divacancy as catalyst, we performed systematic

9

investigations on possible reaction mechanisms for ORR. We first calculated the

10

adsorption energy of the ORR species and found the most stable configuration for

11

each species. Then, the energy barriers and free energies of all the possible ORR

12

elementary steps are calculated. We have indentified three competitive reaction

13

pathways, in which the energy barrier is around 0.85~0.86 eV for the rate-determining

14

step. Among them, the kinetically most favorable one would be the hydrogenation of

15

O2 molecule to form OOH, then the hydrogenation of OOH gives H2O+O. The

16

rate-determining step for this pathway is the final step, i.e., the hydrogenation of OH

17

to produce H2O with an energy barrier of 0.85 eV. Since the hydrogenation of O2

18

molecule to form OOH and the hydrogenation of OOH to form H2O+O can be

19

happened spontaneously, this implies that atomic O would be the main species on the

20

catalyst surface. The free energy diagram of the ORR predicts that the working

21

potential for P-GDV catalyst is 0.27 V for the most favorable pathway. Therefore, we

22

expect that P cooperating with intrinsic carbon defect could lead to a quite promising

23

alternative non-Pt ORR catalyst and is helpful for designing novel high efficiency

24

catalysts for fuel cells.

25

Acknowledgements

26

This work is supported by the National Natural Science Foundation of China

27

(21261013, 21503210, 21521092), the program for Young Talents of Science and

28

Technology in Universities of Inner Mongolia Autonomous Region (NJYT-15-B16),

29

the Natural Science Foundation of Inner Mongolia Autonomous Region (Grant Nos.

AC C

EP

TE D

M AN U

SC

6

16

ACCEPTED MANUSCRIPT 2015MS0120 and 2011BS0104), the Key Science Research Project of Inner Mongolia

2

University of Technology (Grant No. ZD201517 and ZD201117), Jilin Province

3

Youth Fund (20130522141JH) and Jilin Province Natural Science Foundation

4

(20150101012JC). The computing time is supported by the Performance Computing

5

Center of Jilin University and Special Program for Applied Research on Super

6

Computation of the NSFC-Guangdong Joint Fund (the second phase).

RI PT

1

7 8

Supporting Information

SC

9

Fig. S1 is the geometrical structures and formation energy of the three P doped

11

graphenes. Fig. S2 is possible configurations for each adsorbed species (*O2, *2O,

12

*OOH, *O+*OH, *2OH, *O, *OH and H2O) involved in the ORR. Fig. S3 shows the

13

free energy diagrams for the reduction of O2 to H2O at different electrode potentials.

M AN U

10

14

References

16

[1] Debe MK. Electrocatalyst approaches and challenges for automotive fuel cells.

17

Nature 2012;486(7401):43-51.

18

[2] Dai L, Xue Y, Qu L, Choi HJ, Baek JB. Metal-free catalysts for oxygen reduction

19

reaction. Chem Rev 2015;115(11):4823-92.

20

[3] Steele BCH, Heinzel A. Materials for fuel-cell technologies. Nature

21

2001;414(6861):345-52.

22

[4] Snyder J, Fujita T, Chen MW, Erlebacher J. Oxygen reduction in nanoporous

23

metal-ionic liquid composite electrocatalysts. Nat Mater 2010;9(11):904-07.

24

[5] Genies L, Faure R, Durand R. Electrochemical reduction of oxygen on platinum

25

nanoparticles in alkaline media. Electrochim Acta 1998;44(8):1317-27.

26

[6] Zhang J, Sasaki K, Sutter E, Adzic RR. Stabilization of platinum

27

oxygen-reduction

28

2007;315(5809):220-22.

29

[7] Lim B, Jiang M, Camargo PHC, Cho EC, Tao J, Lu X, et al. Pd-Pt Bimetallic

30

Nanodendrites

AC C

EP

TE D

15

electrocatalysts

with

High

using

Activity 17

for

gold

Oxygen

clusters.

Reduction.

Science

Science

ACCEPTED MANUSCRIPT 2009;324(5932):1302-05.

2

[8] Li K, Li Y, Wang Y, He F, Jiao M, Tang H, Wu Z. The oxygen reduction reaction

3

on Pt(111) and Pt(100) surfaces substituted by subsurface Cu: a theoretical

4

perspective. J Mater Chem A 2015;3(21):11444-52.

5

[9] Kattel S, Atanassov P, Kiefer B. Catalytic activity of Co-Nx/C electrocatalysts for

6

oxygen reduction reaction: a density functional theory study. Phys Chem Chem Phys

7

2013;15(1):148-53.

8

[10] Kattel S, Wang G. Reaction Pathway for Oxygen Reduction on FeN4 Embedded

9

Graphene. J Phys Chem Lett 2014;5(3):452-56.

SC

RI PT

1

[11] Lu Z, Xu G, He C, Wang T, Yang L, Yang Z, Ma D. Novel catalytic activity for

11

oxygen reduction reaction on MnN4 embedded graphene: A dispersion-corrected

12

density functional theory study. Carbon 2015;84:500-08.

13

[12] Li F, Shu H, Hu C, Shi Z, Liu X, Liang P, Chen X. Atomic Mechanism of

14

Electrocatalytically Active Co-N Complexes in Graphene Basal Plane for Oxygen

15

Reduction Reaction. ACS Appl Mater Interfaces 2015;7(49):27405-13.

16

[13] Gong K, Du F, Xia Z, Durstock M, Dai L. Nitrogen-Doped Carbon Nanotube

17

Arrays with High Electrocatalytic Activity for Oxygen Reduction. Science

18

2009;323(5915):760-64.

19

[14] Tang Y, Allen BL, Kauffman DR, Star A. Electrocatalytic Activity of

20

Nitrogen-Doped Carbon Nanotube Cups. J Am Chem Soc 2009;131(37):13200-01.

21

[15] Wang S, Yu D, Dai L. Polyelectrolyte Functionalized Carbon Nanotubes as

22

Efficient Metal-free Electrocatalysts for Oxygen Reduction. J Am Chem Soc

23

2011;133(14):5182-85.

24

[16] Wu J, Yang Z, Li X, Sun Q, Jin C, Strasser P, Yang R. Phosphorus-doped porous

25

carbons as efficient electrocatalysts for oxygen reduction. J Mater Chem A

26

2013;1(34):9889-96.

27

[17] Liao Y, Gao Y, Zhu S, Zheng J, Chen Z, Yin C, et al. Facile Fabrication of

28

N-Doped Graphene as Efficient Electrocatalyst for Oxygen Reduction Reaction. ACS

29

Appl Mater Interfaces 2015;7(35):19619-25.

30

[18] Jiang Y, Yang L, Sun T, Zhao J, Lyu Z, Zhuo O, et al. Significant Contribution of

AC C

EP

TE D

M AN U

10

18

ACCEPTED MANUSCRIPT Intrinsic

Defects

to

Oxygen

Reduction

Activity.

ACS

Catal

2

2015;5(11):6707-12.

3

[19] Groves MN, Malardier-Jugroot C, Jugroot M. Improving Platinum Catalyst

4

Durability with a Doped Graphene Support. J Phys Chem C 2012;116(19):10548-56.

5

[20] Xiong B, Zhou Y, Zhao Y, Wang J, Chen X, O’Hayre R, Shao Z. The use of

6

nitrogen-doped graphene supporting Pt nanoparticles as a catalyst for methanol

7

electrocatalytic oxidation. Carbon 2013;52:181-92.

8

[21] Zhang X, Lu Z, Xu G, Wang T, Ma D, Yang Z, Yang L. Single Pt atom stabilized

9

on nitrogen doped graphene: CO oxidation readily occurs via the tri-molecular

SC

Carbon

RI PT

1

Eley-Rideal mechanism. Phys Chem Chem Phys 2015;17(30):20006-13.

11

[22] Chai G, Hou Z, Shu D, Ikeda T, Terakura K. Active sites and mechanisms for

12

oxygen reduction reaction on nitrogen-doped carbon alloy catalysts: Stone-Wales

13

defect and curvature effect. J Am Chem Soc 2014;136(39):13629-40.

14

[23] Zhang L, Xia Z. Mechanisms of Oxygen Reduction Reaction on Nitrogen-Doped

15

Graphene for Fuel Cells. J Phys Chem C 2011;115(22):11170-76.

16

[24] Lin Z, Waller GH, Liu Y, Liu M, Wong C. Simple preparation of nanoporous

17

few-layer nitrogen-doped graphene for use as an efficient electrocatalyst for oxygen

18

reduction and oxygen evolution reactions. Carbon 2013;53:130-36.

19

[25] Lin L, Zhu Q, Xu A. Noble-Metal-Free Fe-N/C Catalyst for Highly Efficient

20

Oxygen Reduction Reaction under Both Alkaline and Acidic Conditions. J Am Chem

21

Soc 2014;136(31):11027-33.

22

[26] Zhang L, Wang A, Wang W, Huang Y, Liu X, Miao S, et al. Co-N-C Catalyst for

23

C-C Coupling Reactions: On the Catalytic Performance and Active Sites. ACS Catal

24

2015;5(11):6563-72.

25

[27] Zhang R, Zhang J, Ma F, Wang W, Li R. Preparation of Mn-N-C catalyst and its

26

electrocatalytic activity for the oxygen reduction reaction in alkaline medium. J Fuel

27

Chem Tech 2014;42(4):467-75.

28

[28] Zhou Y, Yen CH, Fu S, Yang G, Zhu C, Du D, et al. One-pot synthesis of B-doped

29

three-dimensional reduced graphene oxide via supercritical fluid for oxygen reduction

30

reaction. Green Chem 2015;17(6):3552-60.

AC C

EP

TE D

M AN U

10

19

ACCEPTED MANUSCRIPT 1

[29] Yang Z, Yao Z, Li G, Fang G, Nie H, Liu Z, et al. Sulfur-doped graphene as an

2

efficient

3

2012;6(1):205-11.

4

[30] Zhang L, Niu J, Li M, Xia Z. Catalytic Mechanisms of Sulfur-Doped Graphene as

5

Efficient Oxygen Reduction Reaction Catalysts for Fuel Cells. J Phys Chem C

6

2014;118(7):3545-53.

7

[31] Jin Z, Nie H, Yang Z, Zhang J, Liu Z, Xu X, Huang S. Metal-free selenium doped

8

carbon nanotube/graphene networks as a synergistically improved cathode catalyst for

9

oxygen reduction reaction. Nanoscale 2012;4(20):6455-60.

catalyst

for

oxygen

reduction.

ACS

Nano

RI PT

cathode

SC

metal-free

[32] Choi CH, Chung MW, Kwon HC, Park SH, Woo SI. B, N- and P, N-doped

11

graphene as highly active catalysts for oxygen reduction reactions in acidic media. J

12

Mater Chem A 2013;1(11):3694-99.

13

[33] Zheng Y, Jiao Y, Ge L, Jaroniec M, Qiao SZ. Two-Step Boron and Nitrogen

14

Doping in Graphene for Enhanced Synergistic Catalysis. Angew Chem Int Ed

15

2013;52(11):3110-16.

16

[34] Liu Z, Peng F, Wang H, Yu H, Zheng W, Yang J. Phosphorus-doped graphite

17

layers with high electrocatalytic activity for the O2 reduction in an alkaline medium.

18

Angew Chem Int Ed 2011;50(14):3257-61.

19

[35] Li R, Wei Z, Gou X, Xu W. Phosphorus-doped graphene nanosheets as efficient

20

metal-free oxygen reduction electrocatalysts. RSC Adv 2013;3(25):9978-84.

21

[36] Zhang C, Mahmood N, Yin H, Liu F, Hou Y. Synthesis of phosphorus-doped

22

graphene and its multifunctional applications for oxygen reduction reaction and

23

lithium ion batteries. Adv Mater 2013;25(35):4932-37.

24

[37] Kaukonen M, Krasheninnikov AV, Kauppinen E, Nieminen RM. Doped Graphene

25

as a Material for Oxygen Reduction Reaction in Hydrogen Fuel Cells: A

26

Computational Study. ACS Catal 2013;3(2):159-65.

27

[38] Zhang X, Lu Z, Fu Z, Tang Y, Ma D, Yang Z. The mechanisms of oxygen

28

reduction reaction on phosphorus doped graphene: A first-principles study. J Power

29

Sources 2015;276:222-29.

30

[39] Shen A, Zou Y, Wang Q, Dryfe RAW, Huang X, Dou S, et al. Oxygen reduction

AC C

EP

TE D

M AN U

10

20

ACCEPTED MANUSCRIPT reaction in a droplet on graphite: direct evidence that the edge is more active than the

2

basal plane. Angew Chem Int Ed 2014;53(40):10804-08.

3

[40] Jin H, Huang H, He Y, Feng X, Wang S, Dai L, Wang J. Graphene Quantum Dots

4

Supported by Graphene Nanoribbons with Ultrahigh Electrocatalytic Performance for

5

Oxygen Reduction. J Am Chem Soc 2015;137(24):7588-91.

6

[41] Kresse G, Hafner J. Ab initio molecular-dynamics for liquid-metals. Phys Rev B

7

1993;47(1):558-61.

8

[42] Kresse G, Hafner J. Ab-initio molecular-dynamics simulation of the liquid-metal

9

amorphous-semiconductor

in

germanium.

Phys

Rev

B

SC

transition

RI PT

1

1994;49(20):14251-69.

11

[43] Kresse G, Furthmuller J. Efficient iterative schemes for ab initio total-energy

12

calculations using a plane-wave basis set. Phys Rev B 1996;54(16):11169-86.

13

[44] Kresse G, Furthmuller J. Efficiency of ab-initio total energy calculations for

14

metals and semiconductors using a plane-wave basis set. Phys Rev B

15

1996;6(1):15-50.

16

[45] Perdew JP, Burke K, Ernzerhof M. Generalized gradient approximation made

17

simple. Phys Rev Lett 1996;77(18):3865-68.

18

[46] Blochl

19

1994;50(24):17953-79.

20

[47] Kresse G, Joubert D. From ultrasoft pseudopotentials to the projector

21

augmented-wave method. Phys Rev B 1999;59(3):1758-75.

22

[48] Monkhorst HJ, Pack JD. Special points for brillouin-zone integrations. Phys Rev

23

B 1976;13(12):5188-92.

24

[49] Henkelman G, Uberuaga BP, Jonsson H. A climbing image nudged elastic band

25

method for finding saddle points and minimum energy paths. J Chem Phys

26

2000;113(22):9901-04.

27

[50] Grimme S. Semiempirical GGA-type density functional constructed with a

28

long-range dispersion correction. J Comput Chem 2006;27(15):1787-99.

29

[51] Grimme S, Antony J, Ehrlich S, Krieg H. A consistent and accurate ab initio

30

parametrization of density functional dispersion correction (DFT-D) for the 94

TE D

Projector

augmented-wave

method.

Phys

Rev

B

AC C

EP

PE.

M AN U

10

21

ACCEPTED MANUSCRIPT elements H-Pu. J Chem Phys 2010;132(15):154104.

2

[52] Garcia A, Baltazar SE, Romero Castro AH, Perez Robles JF, Rubio A. Influence

3

of S and P Doping in a Graphene Sheet. J Comput Theor Nanosci

4

2008;5(11):2221-29.

5

[53] Faccio R, Fernandez-Werner L, Pardo H, Goyenola C, Ventura ON, Mombru AW.

6

Electronic and Structural Distortions in Graphene Induced by Carbon Vacancies and

7

Boron Doping. J Phys Chem C 2010;114(44):18961-71.

8

[54] Norskov JK, Rossmeisl J, Logadottir A, Lindqvist L, Kitchin JR, Bligaard T,

9

Jonsson H. Origin of the overpotential for oxygen reduction at a fuel-cell cathode. J

SC

RI PT

1

Phys Chem B 2004;108(46):17886-92.

11

[55] Yu L, Pan X, Cao X, Hu P, Bao X. Oxygen reduction reaction mechanism on

12

nitrogen-doped

13

2011;282(1):183-90.

14

[56] Kattel S, Atanassov P, Kiefer B. Density Functional Theory Study of Ni-Nx/C

15

Electrocatalyst for Oxygen Reduction in Alkaline and Acidic Media. J Phys Chem C

16

2012;116(33):17378-83.

17

[57] Wu J, Yang Z, Sun Q, Li X, Strasser P, Yang R. Synthesis and electrocatalytic

18

activity of phosphorus-doped carbon xerogel for oxygen reduction. Electrochim Acta

19

2014;127:53-60.

20

[58]Chorkendorff I, Niemantsverdriet JW. Concepts of Modern Catalysis and Kinetics.

21

John Wiley & Sons: New York 2003.

22

[59] Liu Z, Peng F, Wang H, Yu H, Zheng W, Wei X. Preparation of phosphorus-doped

23

carbon nanospheres and their electrocatalytic performance for O2 reduction. J Nat Gas

24

Chem 2012;21(3):257-64.

density

functional

theory

study.

J

Catal

EP

TE D

A

AC C

25

graphene:

M AN U

10

26

22