Specific ion effects at graphitic interfaces

Improved understanding of aqueous solutions at graphitic interfaces is critical for energy storage and water desalination. However, many mechanistic details remain unclear, including how interfacial structure and response are dictated by intrinsic properties of solvated ions under applied voltage. In this work, we combine hybrid first-principles/continuum simulations with electrochemical measurements to investigate adsorption of several alkali-metal cations at the interface with graphene and within graphene slit-pores. We confirm that adsorption energy increases with ionic radius, while being highly dependent on the pore size. In addition, in contrast with conventional electrochemical models, we find that interfacial charge transfer contributes non-negligibly to this interaction and can be further enhanced by confinement. We conclude that the measured interfacial capacitance trends result from a complex interplay between voltage, confinement, and specific ion effects-including ion hydration and charge transfer.


Supplementary
: Capacitance measured for HCAMs electrode in 1 M aqueous solutions at a scan rate of 0.5 mVs −1 . Ionic radius, hydration radius of Li + , Na + , K + and Cs + and their calculated adsorption energy on a graphene basal plane with different charges, including 0e, −0.5e, −1.0e and −1.5e, are also included. For the calculations of cations at the interface with graphene, we model the electrode by a 96-atom orthorhombic cell, as discussed in the main text. For the binding energy, negative values indicate a favorable interaction between the cations and graphene.
Ions Ionic radius 1 Hydrated radius 1 Adsorption energy   Supplementary Fig. 4: Iso-surfaces of oxygen density representing solvation structure of Cs + during intercalation, where z is the distance between ion and the surface hydrogen. In particular, configuration at z = 10Å represents solvated Cs + in bulk electrolyte; configurations at 6.0, 5.0, and 4.0Å from the surface represent partially desolvated Cs + at the interface, whereas those at 3.0Å and 0.0Å correspond to fully desolvated Cs + inside the pore.
Supplementary Fig. 5: Pore size characteristics of the hierarchical carbon aerogel monoliths (HCAMs), including volume derivative with respect to the pore width, and cumulative pore volume. To measure the micro-porosity, we carried out N 2 absorption at 77 K with a Micromeritics ASAP 2020. In addition, in order to adequately resolve the micropore structure, the experiment was conducted over a 100 h time period to allow the N 2 time to diffuse into the narrow porous regions. The final pore size distribution (PSD) was obtained using both the NLDFT Standard Slit model and the 2D-NLDFT Heterogeneous Surface model 3 processed in the SAIEUS software package. 4,5 The resulting PSD plots show substantial similarities between the two models, with the key feature being the majority of pore volume located at ∼5-6Å. Supplementary Fig. 10: A schematic description of our hybrid quantum-continuum approach. 6 The explicit cations and carbon electrodes were described at the DFT level of theory, whereas the electrolyte was represented by an implicit solvent model consisting of water molecules and 1 M of salt. The interaction between the explicit "graphene-cation" system and the implicit solution is described through the Lennard-Jones (LJ) and electrostatic potentials. In particular, the electrostatic potential acting on the implicit solution species is computed directly from first-principles, whereas the LJ parameters of universal force fields were employed for the carbon atoms. Atomic charges and LJ potentials of the classical solvent and ions are described through the OPLS all-atom force fields. 7,8 Our calculations of cations on charged electrodes were carried out by fixing the net charge of the explicit quantum mechanical region, which is balanced by the classical ions in the RISM solvation, and therefore the whole system is neutral.

Supplementary Methods
Due to the high resistant in thick monolithic activated HCAM electrode systems, [9][10][11] measuring cyclic voltammetry at higher scan rates provide less accurate capacitance information.
The CV of 1 M LiCl and CsCl at different scan rates was measured using a three-electrode setup, particularly at scan rates below 10 mVs −1 where more accurate capacitance values can be determined, and we observed similar trend compared to the two-electrodes experiments.
We also note that the calculated capacitance with the three-electrodes is lower than the one obtained with the two-electrode set up, this can be explained by the higher resistance observed in the three-electrode measurements, due to poor contact between our thick monolithic electrode and the current collector, as well as the differences in the electrode geometry of both methods. Most importantly, the calculated capacitance with the two electrode measurements accounts for both the cation and the anion, while the calculated capacitance with the three-electrode measurements accounts for the cation or the anions depending on the potential polarity. The values plotted on Supplementary Fig. 13 is only for the cation ions.
Supplementary Fig. 12: Cyclic voltammogram of three-electrodes set up at 10 and 1 mVs −1 using 1 M LiCl and CsCl electrolytes. Carbon aerogel working electrode, platinum wire counter electrode, calomel reference electrode.
Supplementary Fig. 13: Capacitance dependence on potential sweep rates using threeelectrodes set up and 1 M LiCl and CsCl electrolytes.