Supplementary Information

A Room-Temperature Sodium Metal Anode Enabled by a Sodiophilic Layer

Shuai Tang, Zhi Qiu, Xue-Yin Wang, Yu Gu, Xia-Guang Zhang, Wei-Wei Wang, Jia-Wei Yan, Ming-Sen Zheng*, Quan-Feng Dong, Bing-Wei Mao*

State Key Laboratory of Physical Chemistry of Solid Surfaces and Department of Chemistry, College of Chemistry and Chemical Engineering, iChEM, Xiamen University, Xiamen

361005 (China).

E-mails: bwmao@xmu.edu.cn, mszheng@xmu.edu.cn

Computational method

All density functional theory calculations were performed using Vienna ab initio simulation

package (VASP) with Pwedew-Burke-Ernzerhof (PBE) functional of the generalized-gradient

approximation (GGA) to the exchange-correlation potential. Projector augmented-wave

method (PAW) was used, and the plane wave basis set was cut off at the energy of 450 eV.

The Methfessel and Paxton method with a broadening factor of 0.1 eV was adopted in

geometry optimization and the convergence criterion was set for the force less than 0.02 eV/Å

in all calculations.

Γ-centered k-point sampling grid of 12×12×12 for primitive cell of Cu and Au lattice was

applied, and calculated lattice constants were 3.635 Å and 4.159 Å, respectively. The five

layers of (2×2) supercell of Cu (111) and Au (111) slab with the bottom two layers fixed was

used to simulate the Cu and Au surface performance. The k-point sampling grid was 6×6×1 Γ-

centered k-points for geometric optimization. The periodic repeated slab and its image were

separated by vacuum layers of 15 Å. For Au2Na alloy, the lattice constants were set

a=b=c=5.607 Å, and α=β=γ=60º, and for AuNa2 alloy, the lattice constants were set

a=b=c=5.915 Å, and α=β=103.5º, γ=122.3º. The bottom two of five layers of (2×2) supercell

of Au2Na (111) and AuNa2 (111) slab were fixed and vacuum layer of 15 Å was set along the z

axis in the supercell. The k-point sampling grid was 8×8×1 Γ-centered k-points for geometric

optimization.

The adsorption energies, Eads, were calculated by

Eads=ENa−slab−E slab−ENa

where ENa-slab is the energy of Na adsorbed on the slab, i.e. Cu, Au or Au2Na slab; Eslab is the

energy of Cu, Au or Au2Na slab and ENa is the energy of Na atom in a sodium crystal.

Figure S1. Cyclic voltammograms of (a), (c), Cu|Na and (b), (d), Cu@Au|Na half cells with

different cathode limit potentials. The cathodic potential limit is 0 V for (a) and (b) and -0.05

V for (c) and (d).

10 20 30 40 50 60 70 80

Stripped to 1.5 V

Stripped to 0.5 V

◇

AuNa2

◇ ◇◇◇

◇◇ ◇

2 theta (degree)

◇

Au2NaCu

Stripped to 0.1 V

Figure S2. The XRD patterns of the Cu@Au substrates after 1st plating and stripping cycle.

The stripping cut-off potentials were set to 0.1, 0.5, 1.5 V, respectively. The standard PDF

card numbers of Au2Na and AuNa2 are 03-0927 and 65-2707. The two peaks at 33.6 and 35.1

degree may correspond to AuNax (x > 2)[1]. AuNa2 and Au2Na are formed for the samples

with stripping potential of 0.1 and 0.5 V, respectively. This suggests that the CV peak at 0.2 V

in Figure S1d corresponds to the transformation from AuNa2 to Au2Na.

When the cathode potential is limited to 0 V (Figure S1a, b), only reduction peaks at 0.8 V

and 0.2 V are observed from both the CV curves recorded for Cu|Na and Cu@Au|Na,

corresponding to the SEI formation process. When the cathode potential limit is set to -0.05 V

(Figure S1c, d), a pair of reduction and oxidation peaks at close to 0 V are observed for both

the Cu|Na and Cu@Au|Na, responding to Na plating and stripping processes. In the case of

Cu@Au|Na, the plating process involves the formation of AuNa2 alloy which is transformed

to Au2Na upon anodic stripping, as indicated by an extra oxidation peak appears at ca. 0.2 V

on the CV. The formation of AuNa2 and Au2Na on Cu@Au substrate is confirmed by XRD

results shown in Figure S2. This leads to a conclusion that when Na is plated on Cu@Au

during the first cycle, it forms alloy of Au2Na and then AuNa2. During the subsequent

stripping process, Na can be stripped from the AuNa2 layer to Au2Na alloy at 0.2 V, but cannot

be stripped from the Au2Na up to 1.5 V.

10 20 30 40 50 60 70 80

Cu

(200

)

(220

)(111

)(2

22)

(311

)

Inte

nsity

(A.U

.)

2 theta (degree)

After 100 cyclesAu2Na

(111

)

Figure S3. The XRD pattern of the Cu@Au2Na after stripping of the 100th cycle. Crystal

phases of Au2Na and Cu were marked in red and green colors respectively. After 100 cycles,

Au2Na remained on the surface of Cu. The standard PDF card number of Au2Na was 03-0927.

Figure S4. XPS spectra (a), (b) and cross section SEM images (c), (d) of 1st Na plating and stripping on Cu@Au. The cross section SEM image (c), (d) showed the Na (the working electrode) at discharge (stripped) and charge (plated) state, respectively.

Figure S5. Binding energies between Na and the substrates. Binding energies of Na with (a)

Cu, (b) Au, (c) Au2Na, (d) AuNa2.

Figure S6. Photographs of molten Na. Photographs (a), (b) of molten Na on bare Cu after

cooling; Photographs (c), (d) of molten Na on Cu@Au after cooling. It showed that molten

Na cannot spread out on bare Cu while covers almost the whole area of Cu@Au foil. It

clearly confirm that Au was more sodiophilic than Cu. Videos recording the whole process of

the experiment were supplied separately.

Figure S7. SEM images of (a), Cu and Au rectangle array, (b), Na electroplated on the Cu and

Au rectangle array. The current density for Na deposition is 1.0 mAcm-2 and areal capacity is

0.1 mAhcm-2. Owing to smaller nucleation barrier on Au than on Cu, Na is selectively

electrodeposited on sodiophilic Au but Cu.

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.5

1.0

1.5

2.0

0.0 0.2 0.4 0.6 0.8 1.0

-0.02

0.00

0.02

Cu Cu@Au

Pot

entia

l/ V

vs.

Na

+/N

a

Areal capacity/ mAhcm-2

Pot

entia

l/ V

vs.

Na+ /N

a

Areal capacity/ mAhcm-2

Cu Cu@Au

Figure S8. The voltage profiles of the 1st Na plating and stripping. Na plating on Cu red line

and Na plating on Cu@Au was shown in blue color. The current density is 1.0 mAcm -2. There

exists a spike for Na plating on Cu responding to nucleation over potential while the over

potential for Na on Cu@Au is not obvious. The corresponding 1st cycle CE for Cu and

Cu@Au is 97.47% and 99.00%, respectively.

Figure S9. SEM images of Na plated on Cu or Cu@Au in large scale. Na of 1.0 mAhcm-2 was

plated on the substrates.

Figure S10. The optical photos (a), (c) and SEM images (b), (d) of Na plated on Cu and

Cu@Au after 250 cycles at 2.0 mAcm-2 with the Na areal capacity of 1.0 mAhcm-2.

0 20 40 60 80 100

60

80

100

0 100 200 300 400 500

92

94

96

98

100(b)

500 nm Au layer 200 nm Au layer 50 nm Au layer 10 nm Au layer

Cou

lom

bic

effic

ienc

y/ %

Cycle

(a)

Cou

lom

bic

effic

ienc

y of

1st

cyc

le/ %

Thickness of Au layer/ nm

Figure S11. The coulombic efficiencies of half cells with different thickness of Au layer on

Cu. a, the coulombic efficiencies of Na plating on Cu substrate with Au layers of different

thicknesses. b, the coulombic efficiencies of the 1st cycle of the half cells in (a). The

coulombic efficiencies are all very high up to 100 cycles for Au layer thickness ranging from

10 to 500 nm. The difference is that the coulombic efficiency of the 1 st cycle decreases as the

thickness of Au increases. The coulombic efficiencies of 1st cycle for the cell with 10, 50, 200,

500 nm Au layer are 99.34%, 99.00%, 97.01%, 92.00%. This is because Au needs more Na to

form alloy phases, from which Na cannot be stripped.

Table S1. The calculation of the effect of Au thickness on CE of 1st cycle

Thickness of Au coated on Cu 10 nm 50 nm 200 nm 500 nm

Capacity loss caused by forming Au2Na/

mAhcm-21.31x10-3 6.56x10-3 2.62x10-2 6.56x10-2

Theoretical maximum CE considering

the irreversible Au2Na formation at 1.0

mAhcm-2

99.87% 99.34% 97.38% 93.44%

Capacity loss caused by forming AuNa2/

mAhcm-25.24x10-3 2.62x10-2 1.05x10-1 2.62x10-1

Theoretical maximum CE considering

the irreversible AuNa2 formation at 1.0

mAhcm-2

99.47% 97.38% 89.05% 73.80%

Real CE(1.0 mAcm-2, 1.0 mAhcm-2) 99.34% 99.00% 97.01% 92.00%

The real 1st CEs are all slightly lower than the calculated maximum CEs based on the Au2Na

alloy with 10, 50, 200 and 500 nm of Au layer, but much higher than the ones based on the

AuNa2 layer with the Au thickness of 50, 200 and 500 nm. This further confirms the

transformation process between AuNa2 and Au2Na alloys.

The details of the calculation (taking 50 nm Au and forming Au2Na as an example) are shown

as below:

Electrode surface area: S= πr2=π x1.32=1.3 cm2

The density of Au: 19.3 gcm-3

The mass of 50 nm Au: 50x10-7 cm x 1.3cm2 x 19.3 gcm-3=1.25 x10-4 g

The relative atomic mass of Au: 196.96 gmol-1

The amount of 50 nm Au: 1.25x10-4 g/196.966 gmol-1= 6.36 x10-7 mol

The amount of Na needing to form Au2Na: 6.36x10-7/2= 3.18 x10-7 mol

The areal capacity loss of Na forming Au2Na:

3.18 x10-7 mol x 96485 Cmol-1/1.3 cm2/3.6 CmAh-1=6.56 x10-3 mAhcm-2

Theoretical maximum CE considering the irreversible Au2Na formation at 1.0 mAhcm-2:

100%-6.56 x10-3/1.0=99.34%

0 100 200 3000

20

40

60

80

100

0 20 40 60 800

20

40

60

80

100

Cou

lom

bic

effic

ienc

y/ %

Cycle

3.0 mAcm-2

1.0 mAhcm-2

(b)

Cou

lom

bic

effic

ienc

y/ %

Cycle

1.0 mAcm-2

5.0 mAhcm-2

(a)

Figure S12. The coulombic efficiencies of Na cycling at different conditions. (a). Na cycling

on Cu@Au at 3.0 mAcm-2 with 1.0 mAhcm-2; (b). Na cycling on Cu@Au at 1.0 mAcm-2 with

5.0 mAhcm-2. The average efficiencies of 99.77% and 99.92% can be obtained, respectively.

0 100 200 3000.5

1.0

1.5

2.0

2.5

3.0

E/ V

vs.

Na+ /N

a

Specific capacity/ mAhg-1

Initial discharge

1st cycle

2nd

cycle

10th cycle

Figure S13. Potential curves of FeS2 cathode presodiation. The current was 0.1 Ag-1 and the

potential range was from 0.8 to 3.0 V. After 10 cycles, the FeS2 cathode maintained capacity

of ca. 300 mAhg-1.

[1] R. Sarmiento-Pérez, T.F.T. Cerqueira, I. Valencia-Jaime, M. Amsler, S. Goedecker, S. Botti, M.A.L. Marques, A.H. Romero, New Journal of Physics 15 (2013) 115007.