Advanced rechargeable aluminium ion battery with a high-quality natural graphite cathode

Recently, interest in aluminium ion batteries with aluminium anodes, graphite cathodes and ionic liquid electrolytes has increased; however, much remains to be done to increase the cathode capacity and to understand details of the anion–graphite intercalation mechanism. Here, an aluminium ion battery cell made using pristine natural graphite flakes achieves a specific capacity of ∼110 mAh g−1 with Coulombic efficiency ∼98%, at a current density of 99 mA g−1 (0.9 C) with clear discharge voltage plateaus (2.25–2.0 V and 1.9–1.5 V). The cell has a capacity of 60 mAh g−1 at 6 C, over 6,000 cycles with Coulombic efficiency ∼ 99%. Raman spectroscopy shows two different intercalation processes involving chloroaluminate anions at the two discharging plateaus, while C–Cl bonding on the surface, or edges of natural graphite, is found using X-ray absorption spectroscopy. Finally, theoretical calculations are employed to investigate the intercalation behaviour of choloraluminate anions in the graphite electrode.

Diffusion can create an impedance, known as the Warburg impedance, which is related to the frequency (ω) of the potential perturbation. Equation 1 is the equation for the "infinite" Warburg impedance. With mobile anions, diffusion flux implies a Warburg-like impedance; this impedance is a ω -1/2 function.
(1) On a Nyquist plot the infinite Warburg impedance appears as a diagonal line with a slope of 0.5.
On a Bode plot, the Warburg impedance exhibits a phase shift of 45°. In the above equation, is the Warburg coefficient defined as: (2) ω is radial frequency, D is diffusion coefficient of anions, A is surface area of the electrode, n is number of electrons transferred and C is bulk concentration of the diffusing species (moles cm -3 ).
In Supplementary Figure 6 (b), we found that the slope of the relationship between Z' and ω -1/2 in the battery with high loading amount is higher than that with low loading amount. According to eq.2, this result indicates that the diffusion coefficient of anions in the battery with high loading is less than that with low loading. The result could be attributed to that the diffusion of chloroaluminate ions to graphite in depth might be reduced when the thickness of film is increased, resulting in reducing the specific capacity of the battery with higher graphite loading. (a) Pouch cells were assembled by using a NG cathode (4 mg cm -2 ) and an Al foil (70 mg) anode, which were separated by one layer of glass fiber filter paper to prevent shorting. Polymer coated Ni bars were used as current collectors connected to NG film and Al foil by a carbon tape for the anode and cathode, separately. (b) The two sides of resulting pouch cell were sealed by a heating sealer for further adding electrolyte easily in the glove box. (c) The resulting pouch cell was delivered into the glove box. Then, the electrolyte (2mL, prepared using AlCl 3 /[EMIm]Cl~1.3 by mole) was injected and the cell was sealed by a heating sealer. (d) Finally, the cell was prepared and removed from the glove box for further performance test.  a. AlCl 4 anions at edge position of the carbon layer. All calculations in this work were performed using the CASTEP program, 1 which employs the plane wave pseudopotential method to calculate the total energy within the framework of the Kohn-Sham DFT. 2 Three flakes of graphene and five AlCl 4 anions were considered in the DFT calculation. The models were fully optimized, allowing the relaxation of all atoms. The system was gradually relaxed to achieve a balanced state with the lowest energy. In the charge/discharge process, AlCl 4 anions were supposed to move toward graphite layers. We used the RPBE exchange-correlation functional. 3 The ion-electron interaction was modeled by the non-local real space, ultrasoft pseudopotential 4 with a cutoff energy of 400 eV.

C-Cl
In the DFT calculations, we employed projector-augmented waves (PAW) 4-6 generalized gradient approximation (GGA) 7 as implemented in the Vienna ab initio simulation package (VASP). 8,9 The clusters 5 AlCl 4 and carbon layers were modeled by a three-layer carbon lattice, with a supercell (dimensions of a = b = 2.5 nm, and c = 3.0 nm). All Al x Cl y -carbon layer clusters models were fully optimized, allowing the relaxation of all atoms. The system gradually become relaxed to achieve a balanced state with lowest energy. Total energy calculations were performed as a cluster calculation (k=0) and convergence was achieved when the residual total energy of 0.01 eV and the maximal force of 0.01 eV Å -1 were reached. The optimized geometries of AlCl 3 and AlCl 4 and carbon layer are shown in Supplementary Table 2.
The bond lengths of Al-Cl are listed in Supplementary Table 2. AlCl 4 clusters with tetrahedral and planar quadrangle geometries were inserted into the relaxed graphite separately, as shown in Figure. 6. In the DFT section, the Al-Cl bond length lengths are listed in Supplementary and bond angles. The distortion directly results from the Van der Waals interactions between the graphene layers, which leads to flattening of the AlCl 4 tetrahedron in response to pressure from the c-axis direction. The interlayer graphene spacing was substantially enlarged (6.23Å) after AlCl 4 intercalation, with the geometric size (4.79 Å) of the intercalated anion.