The generation of voltages in lithium-ion batteries occurs at electrode–electrolyte interfaces as a result of thermodynamic equilibrium which requires equal energies of the Li ⁺ , electrons and neutral species in the adjacent phases. ^{[ 4 , 5 ]} Different electrical energies of the charged species (Li ⁺ ) have to compensate differences in their chemical potentials. When the electrode material is brought in contact with the electrolyte, diffusion of all mobile species will occur from the phase of higher chemical potential to the one of lower chemical potential. This process continues until the transport of the charged species will build up an effective electrical field from the electrode material (space charge region), across the interface to the electrolyte (double layer), in which further diffusion will become counter-balanced. In this case, the electrochemical potentials of all the species (which are involved in lithiation/ delithiation reactions) in the electrode materials equals that in electrolyte. Inside the electrolyte, the chemical potential of the lithium ions remains constant while the chemical potentials of the neutral mobile species and electrons decrease with a constant gradient from the anode to the cathode. ^{[ 4 , 5 ]} As these species have very low concentration in the electrolyte phase, small concentration changes will cause large difference of the chemical potential. Figure  1 illustrates the generation of the electrostatic potential drop across the respective electrode–electrolyte interface in lithium-ion batteries. Upon reaching equilibrium state, such an electrostatic potential difference at both the cathode and anode generates the voltage or the e . m . f . of battery. Upon operation, the electrochemical potential difference of electrons between the cathodic and anodic materials provides the driving force of “lithiation/delithiation” on the cathode and anode during discharge.

Fig. 1.

	Figure Option View Download New Window
	Fig. 1. Chemical potentials of the the neutral mobile component (M), mobile ions (Li + ), and electrons (e − ) in both the cathode and anode, and electrostatic energy ( φ ) across a lithium-ion battery at open-circuit state in the absence of solid electrolyte interface (SEI). [ 4 , 5 ]

All voltage drops occur across the interfaces between the electrodes and electrolyte because there is practically no electrical field built up in the bulk materials considering the large quantities of ionic and electrical species in the bulk phases of both electrodes and electrolyte. This happens when the electrodes are reasonably good electronic conductors and the electrolytes are reasonably good ionic conductor. ^{[ 5 , 6 ]} The electron energy levels can be determined by photoemission spectroscopy (PES) such as x-ray photoelectron spectroscopy (XPS) and ultra-violet photoelectron spectroscopy (UPS). The ion energy level can be deduced by probing the electric field inside a phase or at the interface. ^{[ 7 ]}

The rate of charge and discharge reactions of lithium-ion batteries are determined by several kinetic processes, which include: ^{[ 8 ]} (i) electron transfer between electrode materials and the current collectors; (ii) transport of electron and lithium ion in the electrode materials; (iii) transport of ions in the non-aqueous electrolyte; (iv) interfacial transfer of electrons between the electrode and electrolyte; (v) phase transition in the electrode materials caused by lithiation or delithiation. Process 1 is usually optimized to be fast for commercial battery. The rate of processes 2 and 3 can be increased by engineering means, which mainly involve reducing the distance of transport length. Process 4 is less well understood. The electrochemical equilibrium of Li ⁺ and the electrical potential gradient at the interface between the electrode materials and electrolyte determine the overall electrode potential. Other ionic species at the interface, however, are far from equilibrium. The lack of equilibrium provides a driving force for dissolution, electrolyte decomposition, and the solid electrolyte interface (SEI) formation. ^{[ 6 ]} Future development of high-voltage cathode materials, alloy and conversion reaction based anode materials, as well as improvement of battery safety and cycle life hinges on a better understanding of interfaces.

In Fig.  2 , the energy separation ( E _g ) of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) defines the electrochemical window of the electrolyte. The electrolyte will be either reduced on an anode with above the LUMO or oxidized on a cathode with below the HOMO. The thermodynamic stability of the electrolyte can be achieved by using electrode materials with electrochemical potentials located within the window. ^{[ 9 ]} Therefore,

Fig. 2.

	Figure Option View Download New Window
	Fig. 2. Open-circuit energy diagram of lithium-ion battery in the presence of SEI. and are the electrochemical potential (Fermi energy of electrons) of the anode and cathode, respectively. E g is the window of the electrolyte for thermodynamic stability. V OC is the open-circuit voltage of the lithium-ion battery.

In an ideal lithium-ion battery, where no side reactions occur, lithium ions can be reversibly shuttled between the anode and cathode with 100% efficiency. In reality, however, electrolyte decomposition and SEI formation consume lithium ions and convert them into “electrochemically inactive” lithium compounds (in rare cases, the SEI may have limited reactivity upon charging or discharging), resulting in permanent capacity loss. ^{[ 1 ]} Even though the SEI can prevent the electrolyte from further decomposition, it also slows down the transfer of Li ⁺ at the interfaces during cycling. ^{[ 11 ]} As a result, the cell will have higher impedance and lower cycling efficiency, especially at reduced temperatures. Moreover, decomposition of electrolyte may be accelerated at high temperature, causing fire or possibly explosion, a severe safety concern of lithium-ion battery. The formation and growth of SEI, thereby the battery performance, can be affected by the composition of electrolyte and the cycling temperature during operation. ^{[ 12 ]} In order to improve the performance of lithium-ion batteries, the surface of the electrode materials and the properties of the electrode/electrolyte interface need to be well tuned.

In commercial electrolytes with organic carbonates as solvents, the SEI is generally formed on graphite anode and lithium transition metal oxides cathode during the initial charging process of a fresh lithium-ion battery. The SEI layer consists of organic and inorganic decomposition products of the electrolyte. A few commonly agreed inorganic decomposition products include LiF, Li ₂ O, and Li ₂ CO ₃ . ^{[ 1 ]} SEI formed on cathode materials is usually thinner initially compared to that on graphite, but it tends to increases at a higher rate upon prolonged cycling or at elevated temperature aging. ^{[ 6 , 10 ]}

Fig. 3.

	Figure Option View Download New Window
	Fig. 3. Schematic of the energetic coordinates for charge transfer at the graphite/electrolyte interface. Li + de-solvation is followed by its subsequent migration through the resistive interface.

The charge transfer process at the electrode–electrolyte interface was generally identified as one of the rate-determining steps in the kinetics of lithium-ion battery, with an activation energy of 50–70 kJ·mol ⁻¹ for the case of graphite electrode. ^{[ 13 ]} Abe and Ogumi proposed that the de-solvation of Li ⁺ before it enters the graphite structure is the most sluggish step. Subsequent studies by these authors ^{[ 8 , 14 , 15 ]} and Xu ^{[ 16 – 18 ]} provided corroborative evidence to this speculation, further delineating the step into two sub-processes, which are the de-solvation of Li ⁺ at the interface and the subsequent diffusion of the naked Li ⁺ through the resistive SEI, ^{[ 18 ]} as shown in Fig.  3 . The Li ⁺ de-solvation step requires large activation energy because the solvent sheath, which stabilizes the Li ⁺ , needs to be stripped off before the naked Li ⁺ can be intercalated into the graphite structure. ^{[ 10 , 19 ]} Salient reviews on SEI formation and charge transfer on the surface of SEI have been written by Aurbach et al. in 2007, ^{[ 20 ]} and Xu et al. in 2014. ^{[ 19 ]}

Studies on charge transfer on the surface of active materials without SEI are relatively few, presumably because the most commercialized materials do not fall into this category. Li ₄ Ti ₅ O ₁₂ and LiFePO ₄ are the two important active materials on which it is generally considered that SEI is absent at the operation conditions of the two materials.

The potential of Li ₄ Ti ₅ O ₁₂ (or LTO) lies at around 1.55 V vs. Li/Li ⁺ , well above the reduction decomposition potential of most organic carbonate solvents, and it is assumed that no SEI is formed on the surface of LTO. Xu et al . ^{[ 18 ]} observed that the ac impedance spectra of LTO contain only one semicircle instead of two as those on graphite electrodes. The missing semicircle at high frequency is due to the absence of the migration process of Li ⁺ in the SEI layer. Therefore, the de-solvation process of Li ⁺ is solely responsible for the activation energy barrier of the charge transfer process on LTO electrodes (∼ 50 kJ·mol ⁻¹ ). This value is 20 kJ·mol ⁻¹ less than that on graphite electrodes and it is consistent with the theory that the de-solvation process of Li ⁺ is the most energy consuming step in the interfacial charge transfer process. A similar study was done on anatase TiO ₂ with activation energy of around 20 kJ·mol ⁻¹ . ^{[ 21 ]}

Solvent also plays an important role in the activation energy of the interfacial charge transfer process. Studies have shown that in the presence of SEI, a higher ratio of ethylene carbonate (EC)/dimethyl carbonate (DMC) (up to a certain point) increases the activation energy of the interfacial charge transfer process. ^{[ 16 , 18 ]} The higher activation energy is attributed to stronger interactions between EC and Li ⁺ and different SEI structures formed. In the absence of SEI on LTO, replacing EC/DMC with 3-methoxypropionitrile (MPN) results in much smaller charge transfer resistance. ^{[ 22 , 23 ]} This is a result of weaker interactions between the MPN and Li ⁺ , as compared to that of EC/DMC.

LiFePO ₄ is another active material that has a stable surface in contact with the electrolyte during cycling. Unlike various lithium transition metal oxides, such as LiMn ₂ O ₄ , LiCoO ₂ , and LiNi _0.8 Co _0.2 O ₂ , no decomposition products are detected on LiFePO ₄  surface. ^{[ 24 , 25 ]} The activation energy of charge transfer on the LiFePO ₄ /electrolyte interface was reported to be ∼ 15 kJ·mol ⁻¹ . ^{[ 26 , 27 ]}

Immobilised redox relay molecules have been attempted to enhance the interfacial charge transfer process of battery materials and eliminate the use of carbon additives. Compared with conductive polymers, the relay molecules have greater flexibility in tuning the redox potential. For instance, judiciously synthesized 4-[bis(4-methoxyphenyl)amino]benzylphosphonic acid (BMABP) molecule has an identical redox potential to that of LiFePO ₄ in an EC/DEC (1:1) electrolyte containing 1 M LiPF ₆ . A monolayer of BMABP could be easily formed on the surface of LiFePO ₄ via the chemical adsorption of phosphonic acid group. Upon operation, the surface grafted BMABP molecules will be oxidized/reduced via cross-surface lateral h ⁺ percolation. The change in the concentrations of the surface immobilised BMABP and BMABP ⁺ shifts the equilibrium redox potential of the molecule and provides the driving force for hole injection into the valance band of LiFePO ₄ during charging process and electron injection into the conduction band of FePO ₄ during discharging process, as illustrated in Fig.  4 . ^{[ 30 ]} Strong electronic coupling and low reorganization energy between adjacent BMABP/BMABP ⁺ molecules ensure fast charge propagation. ^{[ 31 ]} As a result, a monolayer of surface relay molecule could effectively wire the insulating battery materials and make the materials electrically addressable even in the absence of conducting additives.

Fig. 4.

	Figure Option View Download New Window
	Fig. 4. Band diagram and density of electronic states curves of the BMABP grafted LiFePO 4 . (a) Charging of cathode by hole injection from BMABP + into the valance band (VB) of LiFePO 4 . (b) Discharging of cathode by electron injection from BMABP into the conduction band (CB) of FePO 4 . [ 30 ]

Alternative to the molecular wiring approach, redox polymers with tethered redox relay molecules have also been studied to solve the stability problem of the surface-grafted redox molecules, which easily come off upon prolonged operation. For instance, redox polymer consisted of poly(4-vinylpyridine) and phenoxazine moiety tethered with a C ₁₁ alkyl chain (PVP-DD-PXZ) was employed to wire the LiFePO ₄ (see Fig. ). ^{[ 32 ]} As the polymer could form robust electrode sheet with the material, no additional binder is required. The tethered phenoxazine molecules “swing” and transfer charges from the current collector to the material. Besides redox polymers, conjugated conducting polymers have also been applied on cathode materials to replace carbon black and inert polymer binder, meanwhile increase the rate capability. ^{[ 33 , 34 ]} Charges propagate in the conductive polymer phase by thermally activated electron hopping between adjacent sites, and are compensated by the diffusion of counter-ions in the electrolyte. Ideal conductive polymers have to be electrochemically stable in the operation window and with good electronic conductivity. ^{[ 35 ]} Several conductive polymers have been investigated in lithium-ion batteries, including polypyrrole (PPy) ^{[ 33 , 34 , 36 – 38 ]} and poly(3,4-ethylenedioxythiophene) (PEDOT). ^{[ 39 – 42 ]}

Desorption of the surface relay molecules and swelling of the redox polymer from the active materials lead to poor electrical contact and consequently limits the cycling performance of battery. Moreover, the rate capability of the grafted electrode is substantially constrained by the small cross-section for lateral charge percolation and by the small driving force for interfacial charge transfer. To circumvent the above issues, redox targeting of battery materials using mobile redox mediators dissolved in the electrolyte was introduced (see Fig.  6 ), ^{[ 2 ]} aimed to solve the problems of instability and sluggish charge percolation of the surface immobilised relay molecules.

Fig. 5.

	Figure Option View Download New Window
	Fig. 5. Schematic illustrating the polymer wiring of insulating electrode material LiFePO 4 . Redox polymer poly(4-((10-(12-dodecyl phenoxazine)vinylpyridinium)-co-4-vinylpyridine) in a molar ratio of 6:1 ( y / x = 5) is wrapped on the surface of the material. [ 32 ]

Fig. 6.

	Figure Option View Download New Window
	Fig. 6. Schematic showing the working principle of redox targeting of LiFePO 4 by a freely diffusing redox mediator S . [ 2 ]

In order for redox targeting to work, the chosen redox mediators must have standard redox potentials that straddle the Fermi energy of the active materials. For instance, two osmium polypyridyl complexes ( and ) were chosen to pair with LiFePO ₄ as the potential of (3.53 V vs. Li/Li ⁺ ) is slightly higher, while that of (3.32 V vs. Li/Li ⁺ ) is lower than the material. During charging process, the formed on current collector oxidizes LiFePO ₄ , which accompanies delithiation of the material. During discharge, the generated on current collector further reduces FePO ₄ for lithiation. As a result, the active material could be reversibly charged/discharged even it is physically detached from the current collector. Compared with the molecular wiring approach, the use of two different redox mediators with potentials further away from the active material provides a larger driving force for the interfacial charge transfer process. More importantly, as the diffusion rate of the redox mediators in electrolyte is much faster than that of the cross-surface charge percolation in the case of molecular wiring, the redox targeting could potentially achieve considerable rate performance for practical applications, despite that a membrane is required to prevent the crossover of the redox molecules from self-discharging at the counter electrode.

In a follow-up study, the reaction rate between another pair of redox mediators, 1,1’-dibromoferrocene (FcBr ₂ ), ferrocene (Fc), and LiFePO ₄ were determined by applying a one-dimensional diffusion-reaction model. ^{[ 43 ]} The redox couple has a standard redox potential (∼ 3.55 V vs. Li/Li ⁺ ) that is 100 mV higher than the Fermi energy of LiFePO ₄ ; therefore, can oxidize LiFePO ₄ during charging process. On the other hand, the Fc ⁺ /Fc redox couple has a standard redox potential (∼ 3.25 V vs. Li/Li ⁺ ) that is 200 mV lower than that of FePO ₄ ; therefore, Fc can reduce FePO ₄ during discharging process. The reactions between the LiFePO ₄ particles and the redox mediators have pseudo-first-order volumetric rate constants in the range of 1–6 s ⁻¹ , which are faster than other kinetic processes in lithium-ion batteries. ^{[ 43 ]}

Besides LiFePO ₄ , redox targeting of anatase TiO ₂ has also been investigated. ^{[ 44 ]} As the lithiation/delithiation potential of TiO ₂ is ∼ 1.80 V vs. Li/Li ⁺ , a different pair of redox mediators cobaltocene (CoCp ₂ ) and bis(pentamethylcyclopentadienyl)cobalt (CoCp* ₂ ) with potentials straddling that of TiO ₂ were selected as the redox mediators for the chemical lithiation/delithiation of the material. Upon discharging, the reduced CoCp* ₂ generated on current collector further reduces TiO ₂ and Li ⁺ is inserted; upon charging, the oxidized CoCp ₂ further oxidizes the lithiated TiO ₂ and Li ⁺ is extracted. As a result, TiO ₂ could be reversibly lithiated and delithiated via redox targeting reactions with the two suitable redox molecules.