advances.sciencemag.org/cgi/content/full/3/12/eaao7233/DC1

Supplementary Materials for

Ultrafast all-climate aluminum-graphene battery with quarter-million

cycle life

Hao Chen, Hanyan Xu, Siyao Wang, Tieqi Huang, Jiabin Xi, Shengying Cai, Fan Guo, Zhen Xu,

Weiwei Gao, Chao Gao

Published 15 December 2017, Sci. Adv. 3, eaao7233 (2017)

DOI: 10.1126/sciadv.aao7233

This PDF file includes:

fig. S1. Additional information for cathode material design, preparation, and

characterizations.

fig. S2. Raman spectra, XRD patterns, and XPS spectra of GF-HC, GF-p, GF-Hp,

and expanded graphite.

fig. S3. HRTEM images of expanded graphite and GF-HC.

fig. S4. SEM images of GF-HC, GF-p, and GF-Hp.

fig. S5. Orientation demonstration of GF-HC and graphene foam with statistic of

cracks in different GF cathodes.

fig. S6. Porosity characteristics of GFs.

fig. S7. Permeability test of ionic liquid electrolyte on GF-HC and GF-Hp in a

glove box.

fig. S8. Dynamic contact angle test of DMF droplet on different GF cathodes.

fig. S9. Mechanical properties of GF.

fig. S10. CV and cutoff voltage optimization of the GF-HC cathode.

fig. S11. Cycling performances of GF-HC and GF-p.

fig. S12. EIS and CV spectra of GF-HC, GF-p, GF-Hp, and graphite.

fig. S13. Element mapping of the charged GF-HC cathode.

fig. S14. Electrochemical performance of the GF-HC cathode at low rates.

fig. S15. Charge/discharge curves of the GF-HC cathode at different

temperatures.

fig. S16. Charge/discharge curves of the GF-HC cathode and ionic conductivity of

[EMIm]AlxCly ionic liquid electrolyte at low temperature.

fig. S17. Comparison of electrochemical performances of GF-HC and GF-p

cathodes at low temperature.

fig. S18. Photograph of flexible Al-GB.

fig. S19. EIS spectra of flexible Al-GB soft pack cell after different bending

cycles.

fig. S20. Additional information on coin cell fabrication, demonstration for the

absence of side reaction, and the electrochemical performance based on mass

loading.

fig. S21. Galvanostatic cycling of the GF-HC cathode with [Et3NH]AlxCly

electrolyte.

table S1. Electrochemical properties of electrode materials from various reports.

Calculations of AlCl4− diffusivity based on CV plot

Calculations of AlCl4− diffusivity based on EIS data

References (50, 51)

f

fig. S1. Additional information for cathode material design, preparation, and

characterizations. (A) The fabrication mechanism of GF-HC, GF-p and GF-Hp. The gas

pressure caused by deoxygenating reaction during annealing created interconnected

channels in GF-HC (under relax state while annealing). Meanwhile the GF-p and GF-Hp

(under mechanical pressure while annealing) afforded less fissures, channels or even

plain surface due to guided gas releasing direction. (14) (B) Photograph of a roll of GF-

HC demonstrating its good flexibility and processibility. (C) XRD patterns of GF-HC.

fig. S2. Raman spectra, XRD patterns, and XPS spectra of GF-HC, GF-p, GF-Hp,

and expanded graphite. (A-C) Raman spectra (A), XRD patterns (B) and XPS spectra

(C) of GF-p and GF-Hp, revealing similiar chemical structure to GF-HC. (D-F) C1s peak

of XPS spectra of GF-HC (D), GF-p (E) and GF-Hp (F). (G) Raman spectra of expanded

graphite showing highly stacked graphite-like 2D peaks with low 2D1 component. (H)

Raman spectra of GF-HC showing few-stacked graphene-like 2D peaks with high 2D1

component (50).

fig. S3. HRTEM images of expanded graphite and GF-HC. (A-C) HRTEM images of

expanded graphite showing highly stacked structure (>>20 layers stacking). (D-L)

HRTEM images of GF-HC showing few-stack graphene structure (2-9 layers stacking).

fig. S4. SEM images of GF-HC, GF-p, and GF-Hp. (A-C) SEM images of GF-HC. (A)

Magnified SEM image of GF-HC corresponding to Fig. 2F, revealing highly aligned

graphene sheets. (B) Sloping cross-section SEM image of GF-HC showing the fissures in

the surface of GF-HC, which were boundaries of graphene sheets. (C) Cross-section SEM

image of GF-HC. (D-F) SEM images of GF-p, revealing fewer channels and fissures than

GF-HC. (G-I) SEM images of GF-Hp, revealing even fewer channels and fissures than

GF-p.

fig. S5. Orientation demonstration of GF-HC and graphene foam with statistic of

cracks in different GF cathodes. (A) SEM image of graphene foam, revealing these

graphene sheet components are non-oriented. (B) Small-angle X-ray scattering (SAXS)

analysis of GF-HC revealing existance of orientation in the graphene sheets of GF-HC.

(C) SAXS analysis of graphene foam revealing absence of orientation. (D) Distribution

of the cracks number in each SEM image of graphene film cathode surface, summarized

from 100 pieces of SEM images (400 μm*270 μm).

fig. S6. Porosity characteristics of GFs. (A) Nitrogen adsorption-desorption isotherms

of GF-HC, GF-p and GF-Hp. The specific surface areas were decreased from 3.1 m2 g-1

for GF-HC, 2.42 m2 g-1 for GF-p to 2 m2 g-1 for GF-Hp. (B-D) Mercury intrusion

porosimetry spectra for graphene films cathode in the macropore range. (B) Plot of

cumulative intrusion volume versus pore diameter. (C) Pore size distribution and (D) Plot

of incremental intrusion volume versus pore diameter. These plots show that GF-HC

possess more large pores (diameter larger than 4 μm), so more mercury intrusion at larger

pores (lower pressure) is achieved. The macroporosities of GF-HC, GF-p and GF-Hp are

67.8%, 60% and 54% respectively.

fig. S7. Permeability test of ionic liquid electrolyte on GF-HC and GF-Hp in a glove

box. (A) Photograph of [EMIm]AlxCly ionic liquid electrolyte droplet on GF-Hp in glove

box, after 18 hours there was no sign of wetting or permeation. (B) The reverse site of

GF-Hp after 18 hours demonstrating no sign of permeation at all, which greatly differs

with Fig. 2J. (C) The cross section of GF-HC with [EMIm]AlxCly ionic liquid droplet,

exhibiting wet permeation area and smaller contact angle. (D-L) The cross section of GF-

Hp with [EMIm]AlxCly ionic liquid droplet, exhibiting neglectable change in contact

angle and permeation.

fig. S8. Dynamic contact angle test of DMF droplet on different GF cathodes. (A-E)

for GF-HC. (F) Time-dependent variation of contact angle of DMF droplet on GF-HC.

(G-I) for GF-p. (J-L) for GF-Hp.

fig. S9. Mechanical properties of GF. (A) Typical tensile stress curves of GF-HC. (B)

Relative electric conductivities of GF-HC with different bending states. (C) Relative

electric conductivities of GF-HC after repeated 180o-bending for 600 times. R0 means

initial resistance.

fig. S10. CV and cutoff voltage optimization of the GF-HC cathode. (A) CV of GF-

HC cathode, revealing an apparent diffusion coefficient of 2.025*10-14 cm2 s-1 calculated

by the cathodic peak at 2.2 V according to Randles-Sevcik equation. The clear redox

peaks demonstrate the cathodic electrochemical redox reaction instead of electrical

double-layer capacitive behaviour. Based on the relationship between peak current and

scan rate in sweep voltammetry, the cathodic peak at 2.25 V and anodic peak at 2.37V

were mostly ascribed to the diffusion-controlled intercalation-based electrochemical

redox reaction, while other peaks at lower potential were ascribed to pseudocapacitive

behavior (25). (B) Charge/discharge curves of GF-HC cathode at different cut-off

voltage. (C) Specific capacities and Coulombic efficiency of GF-HC cathode at different

cut-off voltage, revealing 2.5 V is the optimal choice for cut-off voltage.

Calculations of AlCl4− diffusivity based on CV plot

To evaluate the electrode kinetics of the graphene film cathodes, the AlCl4- diffusivity

(diffusion coefficient) was determined from the results of CV at various scan rates over

the potential range of 0.7-2.5 V at room temperature by the Randles-Sevick equation. In

the typical CVs of the GF-HC cathodes at scan rate of 0.5-10 mVs-1 (fig. S10A). The

peak current (ip) in cathode samples (cathodic peak at 2.2 V in this study) was highly

related to the square root of the scan rate (v) during the de-intercalation and intercalation

of AlCl4-, thus the electrochemical reaction rate would be controlled by a semi-infinite

diffusion process. The diffusivity of the AlCl4- at room temperature (25°C) can be

calculated from Equation (1) as follows

ip = 2.69×105×n1.5×A×DAlCl4

0.5 ×v0.5×C0 (1)

where ip is the peak current (A), n is the number of electrons per reaction species, A is the

apparent area of the electrode (cm2), DAlCl4 is the diffusivity of AlCl4- (cm2 s-1), C0 is the

bulk concentration of AlCl4- in C18AlCl4 (0.006692 mol cm-3 derived from its theoretical

density of 2.63 g cm-3, based on the enhanced weight and expanded layer ratio after

stage-3 intercalation of AlCl4- ion into graphene), and v is the sweep rate (V s-1) (51).

fig. S11. Cycling performances of GF-HC and GF-p. (A) Rate capabilities of GF-HC

and GF-p cathodes at different low rates. The GF-HC cathode maintained remarkable

specific capacity at relatively lower current densities: 115±3 mAh g-1 at 0.1~2 A g-1 and

120 mAh g-1 at 5 A g-1, which also overwhelmed those of GF-p: 93±2 mAh g-1 at 0.1~2 A

g-1 and 70 mAh g-1 at 5 A g-1. (B) Corresponding charge/discharge curves of GF-HC

cathode. (C) Corresponding charge/discharge curves of GF-p cathode. (D)

Charge/discharge curves of GF-HC cathode corresponding to Fig. 3F at different cycles.

(E) Specific capacity and Coulombic efficiency of defective GF-2500 cathode (annealled

at 2500oC, ID/IG=0.05) and GF-1300 cathode (annealed at 1300oC, ID/IG=1.2), delivering

a lower capacity than defect-free GF-HC.

fig. S12. EIS and CV spectra of GF-HC, GF-p, GF-Hp, and graphite. (A) EIS spectra

of GF-HC and GF-p. Calculated by the EIS spectra, the effective diffusion coefficient of

GF-HC and GF-p cathode are 3 *10-14 cm2 s-1 and 9.7*10-15 cm2 s-1 respectively. (B) EIS

spectra of GF-Hp. (C) CV spectra of GF-p at different scan rate, revealing an apparent

diffusion coefficient of 8.1*10-15 cm2 s-1. (D) CV spectra of graphite cathode at different

scan rate, revealing an apparent diffusion coefficient of 4.71*10-15 cm2 s-1. The diffusion

coefficient of GF-HC is higher than highly stacked graphite and less channelled GF-p (8),

supporting the "continuous active material" feature of GF-HC. (E) Relevant equivalent

circuit model for EIS data.

Calculations of AlCl4− diffusivity based on EIS data

EIS results are fitted using an equivalent circuit. In the equivalent circuit, Rs indicates the

ohmic resistance; Rct is attributed to the charge-transfer resistance; CPE represents the

double-layer capacitance and passivation film capacitance. W is the Warburg impedance

caused by a semi-infinite diffusion of AlCl4- ion in the electrode. Zre from EIS is highly

related to the root square of the lower angular frequencies. Through linear fitting, the

Warburg impedance coefficient (σw) can be obtained from the straight lines. The relation

is governed by Eq. (2)

Zre = Rs + Rct +σw * ω-1/2 (2)

After obtaining σw, the diffusivity values of the lithium ions diffusing into the

electrode materials can be further calculated using Eq. (3)

DAlCl4=0.5R

2T2

(AF2Cσw)2 (3)

DAlCl4: AlCl4- diffusivity, R: the gas constant, T: the absolute temperature, F: Faraday's

constant, A: the contact area between active materials and electrolyte, and C: molar

concentration of AlCl4- ions (51).

fig. S13. Element mapping of the charged GF-HC cathode. (A) SEM image of

charged GF-HC corresponding to those following element mapping images, scale bar:

160 μm. (B-H), corresponding element mapping of (B) aluminum, (C) chlorine, (D)

carbon, (E) oxygen, (F) nitrogen, (G) calcium, and (H) silicon. Calcium and silicon

comes from residue separator. Homogeneous distribution of Al, Cl and N species in GF-

HC demonstrate the complete permeation of electrolyte into GF-HC cathode. (I) The

total element distribution, demonstrating the element component of charged GF-HC.

Absence of Fe, Cr or Ni supports no side reaction caused by stainless stell coin cell shell

or nickel current collector.

fig. S14. Electrochemical performance of the GF-HC cathode at low rates. (A)

Cycling performance of GF-HC cathode at 0.5 A g-1 and 0.2 A g-1. (B) Corresponding

charge/discharge curves. (C) Specific capacity and (D) charge/discharge curves of GF-

HC cathode at current densities of 250-400 A g-1.

fig. S15. Charge/discharge curves of the GF-HC cathode at different temperatures.

(A) Charge/discharge curves of GF-HC cathode at 0oC, 25oC and 60oC. (B)

Charge/discharge curves of GF-HC cathode at 80oC with optimization on cut-off voltage.

(C) Charge/discharge curves of GF-HC cathode at 100oC with optimization on cut-off

voltage. (D) Charge/discharge curves of GF-HC cathode at 120oC with optimization on

cut-off voltage. (E) Stable galvanostatic cycling of GF-HC (10 A g-1) over 45,000 cycles

at 100oC.

fig. S16. Charge/discharge curves of the GF-HC cathode and ionic conductivity of

[EMIm]AlxCly ionic liquid electrolyte at low temperature. (A-D) Charge/discharge

curves of GF-HC cathode at different rates at (A) 0oC, (B) -10oC, (C) -20oC, and (D) -

30oC. (E) Relative ionic conductivity of [EMIm]AlxCly ionic liquid electrolyte at

different low temperature below 0oC, comparing to that at 25oC (15 mS cm-1).

fig. S17. Comparison of electrochemical performances of GF-HC and GF-p cathodes

at low temperature. (A) Capacity retention of GF-p cathode (compared with 95 mAh g-1

at 25oC) at different low temperature and different current densities. (B) Capacity

retention of GF-HC cathode (compared with 120 mAh g-1 at 25oC) at different low

temperatures and different current densities.

fig. S18. Photograph of flexible Al-GB. The flexible Al-GB can power LED light under

(A) 0o, (B) 90o and (C) 180o bending. (D-F) The reverse side view (D) and top (E) side

view of the flexible Al-GB watchband connecting to the LED watch. (F) The watchband

battery can successfully power the LED watch while been wrapped around wrist.

fig. S19. EIS spectra of flexible Al-GB soft pack cell after different bending cycles.

fig. S20. Additional information on coin cell fabrication, demonstration for the

absence of side reaction, and the electrochemical performance based on mass

loading. (A) Model of fabricated coin cell, the electrolyte cannot touch the cathode shell,

so that suspected stainless steel-involved side reaction cannot happen. (B) The CV

spectra of Al-ion coin cell with graphene cathode, nickel foil current collector without

graphene cathode, and tantalum current collector without graphene cathode. Inset shows

the magnified CV spectra of nickel foil current collector without graphene cathode, and

tantalum current collector without graphene cathode. The extremely low cathodic peak

currents of current collectors suggest negligible side reaction within those voltage range,

demonstrating the stability of Ni and Ta current collectors within this voltage range. The

very weak anodic peak is due to electrolyte decomposition. (C) Photograph of coin cell

shell after 200,000 cycles, exhibiting no sign of corruption at all. Together with element

mapping result in fig. S13 that no sign of dissoluted Fe, Cr and Ni specie was detected,

the absence of side reaction caused by stainless steel coin cell shell or nickel current

collector is confirmed. (D) Specific capacity of GF-HC cathode at different active

material areal loading (current density of 1 A g-1). (E) SEM image of fresh nickel foil

current collector before being cycling. (F) SEM image of nickel foil current collector

after 50,000 cycles, exhibiting no difference with nickel foil before being cycling.

fig. S21. Galvanostatic cycling of the GF-HC cathode with [Et3NH]AlxCly electrolyte.

(A) Stable cycling of GF-HC cathode at 5 A g-1 within 11,000 cycles. (B) Corresponding

charge and discharge curves within different cycles. (C) Rate performance of different

cathode (GF-HC, GF-p and graphene foam) with [Et3NH]AlxCly electrolyte,

demonstrating much better performance of GF-HC cathode than GF-p and graphene foam.

(D) EIS spectra of different Al-GB with different cathodes and [Et3NH]AlxCly electrolyte.

These results demonstrate that the advance in electrochemical performaces of GF-HC

mainly owes to the "3H3C" design of cathode material rather than electrolyte. Details on

this new electrolyte will be reported later.

Table S1. Electrochemical properties of electrode materials from various reports.

Electrode Rate capability Cycle life

Ref.

Highest

current

density

(A g-1)

Capacity

retention

at highest

current

Cycle number Capacity

retention

Al-

ion

batt

ery

(A

IB) This work 400 91% 250000 91.7%

Graphene

foam 100 74% 25000 97% (2)

Porous 3D

Foam 8 75% 10000 100% (7)

Graphitic foam 6 83% 7500 100% (1)

Graphite 0.792 41% 600 100% (8)

Sod

ium

-ion

batt

ery

NVP 58.5 34% 20000 54% (33)

NVP 11 70% 2000 96% (15)

Bimuth 2 90% 2000 94.4% (34)

Phosphor 5.2 31% 2000 80% (35)

Carbon 20 31% 10000 65% (36)

Lit

hiu

m-i

on

batt

ery

(L

IB)

TiO2 10 46% 2000 87% (37)

LTO 35 45% 3000 92.5% (38)

FeS 10 59% 100000 30% (39)

Silicon 1000 88% (40)

Silicon 10 40% 1000 75% (41)

Sulfur 4.8 46% 4000 94% (42)

Sulfur 8 40% 2000 64% (43)

Sulfur 100 35% (44)

Sulfur 1500 40% (45)

LFP 81 15% 2000 99% (46)

Su

per

cap

aci

tor

(SC

)

Graphene 2000 56% 20000 91% (16)

Graphene 500 53% 1000000 90% (47)

Graphene 40 70% 50000 92% (4)

LDH 100 60% 20000 87% (48)