Basic operating principle

Figure 1 shows a schematic of a typical Li-ion battery consisting of two electrodes (cathode and anode), a separator, and a liquid electrolyte that permeates the whole system. The general convention is to refer to the cathode, or positive electrode, as the electrode that is lithiated upon discharge, while the anode, or negative electrode, is lithiated on charging. For example, in Figure 1, the battery is fully discharged and has just begun the charging process. In order to charge, energy must be supplied (e.g., when plugging in a phone to charge). When this is done, Li ions are driven out of the cathode and through the ion-conducting separator and electrolyte where they are incorporated into the anode material. Because each Li ion carries a net positive (+1) charge, a driving force is created for the loss of a net negative (–1) charge, an electron, to maintain charge neutrality in the system. However, because the electrolyte does not conduct electrons, the negative charge must take a different path out of the cathode, through an external circuit, before being incorporated into the anode. Thus, for every Li ion removed, a corresponding electron will flow through the external circuit. Consequently, transition metals (TMs) that typically make up the cathode material (e.g., Mn, Ni, Co) will be oxidized (lose electrons) during charging.

Figure 1.

Schematic of a working Li-ion battery utilizing a LiCoO₂ cathode and a graphite anode with aluminum and copper current collectors, respectively. The electrolyte, which permeates the entire system, and separator allow for the diffusion of positively charged Li⁺ ions but not negatively charged electrons, which must travel through the external circuit, constituting an electric current that powers the attached device (load) on discharge.

Ideally, once all of the lithium is moved from the cathode to the anode, the circuit can be broken, and the battery can be maintained in a fully charged state, ready for later use. Therefore, energy has been taken from some source (e.g., coal, gas, solar, wind) and stored in the battery. This is why Li-ion batteries are often referred to as energy-storage devices. When this stored energy is needed, the circuit is again completed, and the whole process happens in reverse. Specifically, lithium ions move back to the cathode, while the corresponding electrons flow back through the external circuit (the TMs are now reduced—gain electrons) until the process is complete and the battery is “dead,” ready for recharging.

Each of the major components of the battery (cathode, anode, and electrolyte) plays a critical role in the overall performance. Furthermore, batteries are highly correlated systems in that the performance of each component is to some degree dependent on the properties and collective performance of the others. As such, the different components must be chosen carefully to suit particular technologies, while no single combination satisfies the needs of all applications.

Cathodes

A list of current, commercially available cathodes and the relevant battery metrics are shown in Table I^{Reference Thackeray, Wolverton and Isaacs7,Reference Meyers8} for comparison. Layered LiCoO₂ (shown in Figure 1) has been the standard Li-ion chemistry for almost 25 years largely due to its high volumetric energy density. This is especially important since space to house the battery in any application is always limited. However, the cost and safety (thermal instability at high states of charge) of LiCoO₂ are major concerns for use in large-format, automotive batteries, as is the cost of pure, NCA-based electrodes (NCA = nickel, cobalt, aluminum). As a practical reference, Chevrolet’s PHEV Volt, with a ∼40 mile all-electric range, uses a physically blended NMC-based/LiMn₂O₄ cathode material (NMC = nickel, manganese, cobalt). In order to drive cost down, and/or the driving range up, significant advancements in cathode technology are needed beyond NMC. However, energy densities are not the only relevant metric when looking for overall advancements in Li-ion systems.

Table I.

Best-case (low rate), practically achievable, working voltages and capacities versus Li/Li⁺ for various cathode materials.^{Reference Thackeray, Wolverton and Isaacs7,Reference Meyers8}

Crystal densities are theoretical values, while tap densities represent typical, practical values, determined experimentally as the actual weight per unit volume occupied by a given material. Volumetric energy densities are calculated using the crystal densities for comparison as the optimum, final electrode densities will vary among materials. *Not yet commercialized. Capacity and voltage targets for 0.5 Li₂MnO₃•0.5 LiMO₂ are based on the US Department of Energy’s end-of-life goals for composite materials.^{Reference Gallagher13} Crystal density is calculated as an average of Li₂MnO₃ and LiMn_0.5Ni_0.5O₂.

Promising options: Lithium- and manganese-rich NMC

There are five driving factors in the world of batteries: safety, cost, life, energy, and power. These drivers almost always work against each other, and some amount of compromise/optimization must be settled upon for a given application. For transportation, safety and cost are major concerns with little room for compromise. For this reason, manganese-rich cathode materials have attracted much attention. Manganese is plentiful, inexpensive, and safe, especially when compared to Co.

One class of materials that is rich in manganese and offers the possibility of high-energy densities is the lithium-and manganese-rich xLi₂MnO₃•(1–x)LiMO₂ (M = Mn, Ni, Co) composite cathode. As the notation suggests, these materials are, ideally, a combination of Li₂MnO₃ and LiMO₂. However, unlike physically blended materials,^{Reference Chikkannanavar, Bernardi and Liu9} the two components of composite materials are integrated at the nanoscale during synthesis, giving unique electrochemical properties to the final material.^{Reference Thackeray, Kang, Johnson, Vaughey, Benedek and Hackney10} These electrodes can deliver capacities approaching ∼300 mAh/g (see Table I to compare conventional system capacity values).^{Reference Ohzuku, Nagayama, Tsuji and Ariyoshi11} One reason for this stems from the fact that the Li₂MnO₃ structure contains “excess” lithium with respect to the LiMO₂ structure, see Figure 2. The remarkable characteristic of composites is that Li₂MnO₃ by itself does not cycle well, in fact, not at all if not charged above ∼4.5 V, after which, its structure quickly changes while losing capacity with cycling.^{Reference Robertson and Bruce12} When integrated at the nanoscale in composite electrodes, however, a portion of the excess lithium can be accessed and cycled for many cycles.

Figure 2.

(Top) Structural diagrams of the two layered components, Li₂MnO₃ (left) and LiMO₂ (right, M = Mn, Ni, Co), that make up the integrated xLi₂MnO₃•(1–x)LiMO₂ composite cathode materials. As shown, the Li₂MnO₃ contains “excess” Li (purple) in the MO₆ layers with respect to the LiMO₂ structure. (Bottom) Second cycle (post-activation) charge and discharge curves for a series of xLi₂MnO₃•(1–x)LiMn_0.5Ni_0.5O₂ electrodes (versus Li metal) along with calculated, specific energy densities. ΔV represents the hysteresis at 50% state of charge. Curves have been offset by 100 mAh/g for clarity. As shown, considerable gains in energy density over current chemistries (Table I) might be realized through these composite materials.

Some of the NMC-based materials previously mentioned contain small Li₂MnO₃ components as well (e.g., 0.05Li₂MnO₃•0.95LiMO₂), which, when cycled below 4.4 V, do not access the lithium from Li₂MnO₃. In that case, Li₂MnO₃ acts as a stabilizer to the overall structure, as it does not change with charge and discharge.^{Reference Thackeray, Kang, Johnson, Vaughey, Benedek and Hackney10}Figure 2 shows the second cycle charge and discharge curves for a series of xLi₂MnO₃•(1–x)LiMn_0.5Ni_0.5O₂ electrodes with x = 0, 0.1, 0.3, and 0.5 along with calculated specific energy densities.^{Reference Gallagher13} As shown, these electrodes allude to the possibility of substantial energy densities compared to current Li-ion chemistries.

There are, however, several issues with composite electrodes that hinder their implementation in automotive applications. As discussed, in order to access the lithium associated with the Li₂MnO₃ component, the material must be charged above ∼4.5 V, a process termed “activation.” During this activation, the material undergoes irreversible structural changes, observed in the first-cycle charge as a plateau at ∼4.5 V proportional in length to the Li₂MnO₃ content, x.^{Reference Croy, Gallagher, Balasubramanian, Long and Thackeray14} Among these changes are believed to be oxygen loss from the lattice and a rearrangement of the TMs.^{Reference Armstrong, Holzapfel, Novak, Johnson, Kang, Thackeray and Bruce15} Activation of the material, followed by high-voltage cycling (needed to access the high capacity), sets in motion continuous structural changes in the bulk of the material with the onset of a significant hysteresis (e.g., a gap in voltage between charge and discharge, Figure 2), again proportional to Li₂MnO₃ content, x.^{Reference Croy, Gallagher, Balasubramanian, Long and Thackeray14,Reference Croy, Gallagher, Balasubramanian, Chen, Ren, Kim, Kang, Dees and Thackeray16} Because the local crystallographic environment of lithium dictates the working voltage of the battery, these structural changes cause a continuous decrease in the average voltage with cycling, despite maintaining high capacities. This drop in voltage, known as voltage fade, decreases the energy output of cells from one cycle to the next.^{Reference Bettge, Li, Gallagher, Zhu, Wu, Lu, Bloom and Abraham17} Therefore, even though the second-cycle (post-activation) energy densities in Figure 2 illustrate the promise of composite electrodes, they are, unfortunately, short-lived, quickly degrading to levels on par, or worse, than those of current chemistries, see Table I. Considerable effort is now being invested on the problem of voltage fade in these electrodes.

The complexity of composite materials, especially under working conditions, however, makes the task of understanding voltage fade non-trivial. Recently, progress along these lines has been made. A study of hysteresis in composite electrodes was linked to studies on voltage fade, and a single model was proposed that might provide insights into both mechanisms.^{Reference Croy, Gallagher, Balasubramanian, Long and Thackeray14,Reference Gallagher, Croy, Balasubramanian, Bettge, Abraham, Burrell and Thackeray18} In addition, Sathiya et al.^{Reference Sathiya, Rousse, Ramesha, Laisa, Vezin, Sougrati, Doublet, Foix, Gonbeau, Walker, Prakash, Ben Hassine, Dupont and Tarascon19} have shown minimal voltage fade for Li₂RuO₃-based electrodes that may lead to clues regarding the degradation mechanisms of Mn-based composites.

Alternatively, synthetic routes are also being pursued in order to create more voltage fade-resilient structures. For example, long-term structural changes are thought to result in the transformation of the layered composite materials into more spinel-like structures. As a countermeasure, a spinel component can be integrated into the composite during synthesis, giving complex materials of the form y[xLi₂MnO₃•(1-x)LiMO₂]•(1–y)LiM₂O₄. Thackeray et al. recently reported some promising results on these electrodes.^{Reference Donghan, Sandi, Croy, Gallagher, Kang, Lee, Slater, Johnson and Thackeray20,Reference Long, Croy and Thackeray21} Although the spinel component typically lowers the average voltage and capacity, improved stability and rate capability could prove to be an advantage.

Finally, an important concern with lithium- and manganese-rich materials is the practical material densities (tap density) that can be achieved, which dictate the volumetric energy densities. Crystal densities of lithium- and manganese-rich materials are inherently lower than those of cobalt-based materials. Furthermore, the addition of manganese and its involvement in redox processes lowers the working voltage of layered cathode materials. Therefore, even with such high practical capacities (∼250 mAh/g), the slightly lower working voltages (<3.7 V) and densities will bring down volumetric energy densities. Nevertheless, it is the final, overall electrode density that dictates volumetric energy, and if the challenges of voltage fade can be successfully met, the promise of high energy, inexpensive, and safe battery materials might be realized.

Anodes

Criteria for anodes in Li-ion batteries include low voltage, high capacity (gravimetric and volumetric), cycleability, rate capability, and safety. The anode material most commonly used is graphite.^{Reference Yazami and Touzain25} Although graphite is a successful anode material, alternatives are being developed to increase capacity and improve overall electrochemical properties.^{Reference Wen and Huggins26–Reference Poizot, Laruelle, Grugeon, Dupont and Tarascon28} In addition, graphite is known to show capacity degradation, especially when the battery is cycled at high temperatures.^{Reference Andresson and Edstrom29}

In order to increase the anode capacity, new materials based on mixtures of alloys and oxides have been investigated. For example, silicon monoxide (SiO) shows better cycle performance than pure micron-size silicon.^{Reference Yang, Takeda, Imanishi, Capiglia, Xie and Yamamoto30} While SiO can yield high capacity (though it suffers from poor cycle life), Sn-Co-C has good cycleability and high tap density, although its reversible capacity (∼400 mAh/g) is not significantly better than that of commercial carbon anodes.^{Reference Todd, Mar and Dahn31,Reference Zhang and Whittingham32} SiO-SnCoC composite materials can be prepared inexpensively via ball milling, which is safer and less expensive than typical sputtering or gas phase separation of silicon-based materials. These materials show a volumetric density of about 1.5 g/cm^–3, which is higher than that of graphite, silicon, or silicon carbon mixtures.

Other composite materials, based on the M_aO_b (M = Si, Ge) oxides and the Sn_xCo(Fe)_yC_z alloys prepared by high-energy ball milling, have also been electrochemically investigated.^{Reference Liu, Abouimrane, Ren, Balasubramanian, Wang, Fang and Amine33–Reference Liu, Abouimrane, Ren, Neuefeind, Fang and Amine35} The studies in References 33–35 are particularly interesting because the raw materials and processing for SiO-SnCoC and SiO-SnFeC composites are economically attractive. Figure 3a–b illustrates x-ray diffraction (XRD) patterns of SiO-SnCoC prepared by high-energy ball milling (UHEM, ultrahigh energy ball milling). The main phases present (e.g., CoSn, CoSn₂, and/or Co) and, therefore, the electrochemistry can be controlled by the type of milling process (low energy ball milling (SPEX) versus UHEM, see References 33–35) used.

Figure 3.

X-ray diffraction patterns of Co-based anodes prepared by (a) high-energy ball milling (UHEM) and (b) low-energy ball milling (SPEX);^{Reference Liu, Abouimrane, Ren, Balasubramanian, Wang, Fang and Amine33} cycling performance of (c) UHEM anode half-cell cycled at rates of 300 mAh/g (∼C/3:three-hour discharge)^{Reference Liu, Abouimrane, Ren, Balasubramanian, Wang, Fang and Amine33} and (d) full cell based on Li_1.2Ni_0.15Co_0.10Mn_0.55O₂ versus 50 wt% SiO-50 wt% Sn₃₀Co₃₀C₄₀ with a stabilized lithium metallic powder electrode coating.^{Reference Liu, Abouimrane, Ren, Neuefeind, Fang and Amine35} (Inset graph shows the voltage profiles of the full cell for the 2nd, 100th, and 200th cycles.) Adapted with permission from Reference 33, © 2012 American Chemical Society; and Reference 35, © 2013 The Electrochemical Society.

In the case of SiO-SnFeC materials, Fe₃Si is only present in materials with a high iron content. However, FeSn₂ is present in all samples, and Fe₃Sn is present when Fe₃Si is absent. The electrochemical performance of SiO-SnFeC materials is shown in Figure 4, where the obtained discharge capacities varied between 400 and 800 mAh/g. The 50 wt% SiO-50 wt% Sn₃₀Co₃₀C₄₀ composite material exhibits a low decay in capacity with cycling (Figure 3c–d), while the electrochemical performance of the iron-based material is less desirable (Figure 4a–b). Both materials, SiO- Sn₃₀Co₃₀C₄₀ and SiO-Sn₃₀Fe₃₀C₄₀, contain a corresponding MSn₂ phase (CoSn₂ and FeSn₂, respectively) and crystallize in the same tetragonal structure within the I4/mcm space group.^{Reference Havinga, Damsma and Hokkeling36} The presence of FeSn₂, however, seems to result in accelerated performance degradation.

Figure 4.

Electrochemical properties of low energy ball milled (SPEX) 50 wt% SiO-50 wt% Sn₃₀Fe₃₀C₄₀ composite: (a) voltage profile of the 1st, 2nd, and 50th cycles showing a first cycle capacity irreversibility of about 33%, and (b) capacities versus cycle performance under C/2 (two hours discharge) and C/6 (six hours discharge). Reprinted with permission from Reference 34. © 2013 Royal Society of Chemistry.

The SiO-Sn₃₀Fe₃₀C₄₀ material delivered more than 950 mAh/g upon discharge, with a first-cycle efficiency of ∼67%. This capacity was sustained for ∼40 cycles, where 900 and 700 mAh/g were obtained at C/6 (e.g., a six-hour discharge) and C/2 (e.g., a two-hour discharge) current rates, respectively. Between the 40th and 50th cycles, the capacities dropped to 800 and 550 mAh/g under C/6 and C/2 current rates, respectively. The electrochemical behavior of the cobalt-based material showed better cycleability. In half-cell tests (Li metal counter electrode), the SiO-Sn₃₀Co₃₀C₄₀ composite electrode delivered a high discharge capacity of 900 mAh/g at a current of C/3 for over 100 cycles (Figure 3c). In the first cycle, the material delivered 1430 mAh/g discharge capacity and 960 mAh/g charge capacity with a 67% Coulombic efficiency. During the discharge, a capacity of 90 mAh/g was delivered before reaching a voltage of 0.75 V. This capacity can be attributed to the formation of a solid electrolyte interphase (SEI) layer, which consists of ethyleneoxide-based oligomers, LiF, Li₂CO₃, and lithium alkyl carbonate (ROCO₂Li, where R is an organic group).

After the first cycle, the curves of subsequent cycles were similar in shape, indicating that the electrode reactions become more reversible. Electrochemical tests (not shown) revealed that an ultrahigh-energy ball milled sample yielded a specific capacity of 930 mAh/g at a current rate of 300 mA/g (∼C/3), much higher than the specific capacity obtained for a similar material prepared using the SPEX milling method (only 650 mAh/g at 300mA/g). This difference was attributed to the presence of small particle sizes in the case of the HEM preparation, which facilitated lithium insertion and enhanced rate capability.

The lithiation of the SiO/Sn₃₀Co₃₀C₄₀ composite was followed by XRD and Pair distribution function analysis (PDF), which can provide nearest-neighbor atoms information.^{Reference Liu, Abouimrane, Ren, Neuefeind, Fang and Amine35} During the discharge, CoSn₂ did not completely react with lithium, and cobalt was electrochemically inactive. In addition, the presence of a broad peak at low scattering angle 2θ suggested the formation of amorphous Li_xSi phases. The main phase observed was a Li₁₃Sn₅ alloy. During the discharge, until 0.15V, the obtained high capacity could not be attributed to SiO, SEI, or CoSn₂, which is still present at that voltage. It was proposed that lithium might be partially inserted into CoSn₂ by the following intercalation/conversion reaction: CoSn₂ + (x + xy)Li ↔ (1+y)Li_xSn + CoSn_1–y. PDF analysis of discharged samples revealed the presence of Li₁₃Sn₅ and an increase in metallic cobalt in the composition of the material. The formation of metallic cobalt can be represented by the following conversion reaction: CoSn₂ + 2xLi ↔ 2Li_xSn + Co.

For practical applications, the anode system was evaluated in a full-cell (e.g., versus a practical counter electrode). Due to the high first-cycle irreversibility of 33%, stabilized lithium metallic powder was coated on an already prepared anode laminate before cycling. The full-cell was assembled using the high-energy composite cathode material HE5050 (Li_1.2Ni_0.15Co_0.10Mn_0.55O₂) (Figure 3d) as the counter electrode and cycled between 2 and 4.5 V. An average voltage of 3.5 V was obtained with ∼234 mAh/g based on the cathode weight. Promising cycleability with fair capacity retention was observed for more than 200 cycles. A discharge capacity of 167 mAh/g was delivered after 200 cycles under a C/12 current rate. This result demonstrates that the anode system can be assembled with high-capacity composite cathode materials without the loss of significant voltage. The high capacity of this anode material, combined with its high tap density (∼1.9 g/cm^–3), will improve the final, full-cell energy density.

Electrolytes

LiPF₆ salt dissolved in organic carbonates, in particular, mixtures of ethylene carbonate (EC) with dimethyl carbonate (DMC), diethyl carbonate (DEC), and/or ethyl methyl carbonate (EMC), are the default Li-ion electrolytes for materials research and cell fabrication due to their superior properties.^{Reference Xu37} However, these carbonate solvents have low voltage stability. By using x-ray photoelectron spectroscopy and Fourier transform infrared (FTIR) spectroscopy, Lucht and co-workers^{Reference Yang, Ravdel and Lucht38} found that EC/DMC/DEC-based electrolytes decompose at 4.5 V versus Li/Li⁺.

Recently, the electrochemical performance of carbonate-based electrolytes in high-voltage graphite/LiNi_0.5Mn_1.5O₄ (LNMO) cells under various charging conditions was investigated.^{Reference Hu, Zhang and Amine39}Figure 5a shows the cycling performance of a 1.2 M LiPF₆ EC/EMC (3/7 in weight, Gen 2, second generation of the carbonated based electrolyte) electrolyte at 55°C. A constant voltage charging step at 4.9 V led to deterioration in performance due to the continuous decomposition of the electrolyte, especially at elevated temperatures. The capacity retention of the Gen 2 cell was 75% at room temperature (RT), whereas the capacity retention at 55°C was only 25%. The Coulombic efficiencies reflect the same trend as illustrated in Figure 5b, suggesting that severe electrochemical decomposition occurs in the cell at such high voltages and elevated temperatures. The voltage instability of conventional carbonate electrolytes limits their application for use with high-energy cathode chemistries that deliver capacities at high charging voltages, such as Ni/Mn spinel LNMO.^{Reference Okada, Ueno, Uebou and Yamaki40} Therefore, the search for a high-voltage electrolyte has become a priority for the development of Li-ion batteries with high-energy densities.

Figure 5.

(a) Cycling profile and (b) Coulombic efficiency of Gen 2 (second-generation electrolyte) LiNi_0.5Mn_1.5O₄ (LNMO)/graphite cells at room temperature (RT) and 55°C. Conditions: constant current between 3.5 and 4.9 V, C/3 = 0.67 mA, then constant voltage at 4.9 V for 5 min or until currentI< 0.067 mA, C/30. Reprinted with permission from Reference 39. © 2013 Elsevier. Note: C/3, three-hour discharge; C/30, charge process will end when the current reaches C/30.

Although an additive tris(hexafluoro-iso-propyl)phosphate for cathode SEI formation was reported to suppress the reactivity between the charged electrode and electrolyte,^{Reference Cresce and Xu41} a superior electrolyte system comprising solvents with intrinsic anodic stability is the ultimate solution for high-voltage electrolytes. Among all the potential, high-oxidation compounds, organic fluoro-compounds are the most promising electrolyte solvent candidates for use in high-voltage Li-ion batteries. This is because fluorinated molecules generally have higher oxidation potentials than their non-fluorinated counterparts due to the strong electron-withdrawing effect of the fluorine atom. In fact, fluorinated, cyclic carbonates have been reported as a co-solvent by McMillan^{Reference McMillan, Slegr, Shu and Wang42} and Nanbu^{Reference Nanbu, Takimoto, Takehara, Ue and Sasaki43} and as a solid electrolyte interface-forming additive for graphite^{Reference Wang, Xu, Yao, Yao and Yang44} and silicon^{Reference Choi, Yew, Lee, Sung, Kim and Kim45} anodes.

Smart et al.^{Reference Smart, Ratnakumar, Ryan-Mowrey, Surampudi, Prakash, Hu and Cheung46,Reference Smart, Ratnakumar, Prakash and Krause47} designed and synthesized a series of fluorinated linear carbonates at NASA’s Jet Propulsion Laboratory as new electrolyte components to improve the low-temperature performance of Li-ion batteries for deep space applications. However, there have been few reports on the high-voltage application of these fluorinated carbonates. Recent research on the high-voltage application of fluorinated carbonates and fluorinated ethers in high-voltage LNMO cells is highlighted in a review article.^{Reference Zhang, Hu, Wu, Weng, Koh, Redfern, Curtiss and Amine48} The chemical structures of studied fluorinated carbonates and ethers are listed in Figure 6.

Figure 6.

Chemical structure of the baseline carbonate (EC and EMC), ethyl propyl ether (EPE), fluorinated cyclic carbonate (F-AEC), fluorinated linear carbonate (F-EMC), and fluorinated ether (F-EPE). Electrochemical stability of Gen 2 and fluorinated electrolytes E1 to E6 at (a) 5.3 V using a 3-electrode electrochemical cell, and (b) capacity retention of LNMO/LTO cells with baseline electrolyte Gen 2 and fluorinated electrolyte E3, E5, and E6 at 55°C. Gen 2: second generation of the carbonated-based electrolyte; E1:EC/EMC/F-EPE (fluorinated ethyl propyl ether) (2/6/2) 1.2 M LiPF₆; E2:EC/EMC/F-EPE (2/5/3) 1.2 M LiPF₆; E4:F-AEC/EC/EMC/F-EPE (1/1/6/2) 1.2 M LiPF₆. Reprinted with permission from Reference 48. © 2013 Royal Society of Chemistry. Note: EC, ethylene carbonate; EMC, ethyl methyl carbonate.

The initial prediction of the oxidation stability of these fluorinated solvents was carried out by density functional theory (DFT) calculations.^{Reference Frisch, Trucks, Schlegel, Scuseria and Robb49} Fluorine substitution in organic carbonates and ethers results in higher oxidation stability and higher reduction potentials. The calculated oxidation potentials of the fluorinated carbonates and ethers are much higher compared with other high-voltage electrolytes. For instance, the theoretical oxidation potential of EMC is greatly increased, from 6.63 V to 7.01 V, through fluorine substitution. For comparison, the oxidation potential of ethyl methyl sulfone, a representative of high-voltage sulfone solvents, was calculated by Shao and co-workers to be around 6.0 V using the same DFT method.^{Reference Shao, Sun, Dai and Jiang50}

The stability of various fluorinated electrolytes against oxidation under high voltage conditions was investigated.^{Reference Zhang, Hu, Wu, Weng, Koh, Redfern, Curtiss and Amine48} Six fluorinated electrolytes were formulated as: E1–1.2 M LiPF₆ in EC/EMC/F-EPE (fluorinated ethyl propyl ether) (2/6/2; ratio by volume); E2–1.2 M LiPF₆ in EC/EMC/F-EPE (2/5/3); E3–1.2 M LiPF₆ in F-AEC (fluorinated cyclic carbonate)/EMC/F-EPE (2/6/2); E4–1.2 M LiPF₆ in F-AEC/EC/EMC/F-EPE (1/1/6/2); E5–1.2 M LiPF₆ in F-AEC/F-EMC (fluorinated linear carbonate)/F-EPE (2/6/2); and E6–1.2 M LiPF₆ in EC/F-EMC/F-EPE (2/6/2). Based on the DFT calculations, E5 was the most stable electrolyte since its components F-AEC, F-EMC, and F-EPE have the highest oxidation potentials. For electrolyte E6, the EC component forms an open cation radical at 6.83 V, which is lower than F-AEC (6.98 V) in E5, suggesting oxidative decomposition in E6 at a lower potential than in E5.

The electrochemical floating test was performed with the fluorinated electrolytes and compared with a Gen 2 electrolyte from 5.3 V to 6.1 V versus Li⁺/Li.^{Reference Zhang, Hu, Wu, Weng, Koh, Redfern, Curtiss and Amine48} As shown in Figure 6a–b, the Gen 2 electrolyte showed much larger leakage current and maintained this current when the potential was raised to 5.7 V. In contrast, the fluorinated electrolytes showed minimal increase in leakage current at this potential. At this high potential, the fluorinated ether F-EPE enhances the electrochemical stability, evidenced by the fact that the major difference between E1 and Gen 2 is the addition of F-EPE in the former. The EMC-based electrolyte compositions E1 to E4 showed a similar increase in leakage current, while the F-EMC-based E5 and E6 retained a minimal leakage current. This indicates that the predominant potential-limiting factor arises from the linear carbonate rather than the cyclic carbonate.

To validate the DFT calculations and electrochemical floating test results, additional electrolyte tests using the LNMO/Li₄Ti₅O₁₂ (LTO) couple, cycled galvanostatically between 2.0 and 3.45 V at a C/2 rate at 55°C, were conducted. In terms of capacity retention, the E3, E5, and E6 electrolyte cells performed much better than Gen 2 cells, with negligible capacity loss after 80 cycles (Figure 6b). This experimental result agrees with the calculated DFT predictions. By comparison, the LNMO/LTO cell with Gen 2 electrolyte showed more than 70% capacity loss in the same cycling span.

Because fluorinated electrolytes have higher reduction potentials, this resulted in instability on the anode side of the high-voltage graphite/LNMO cells.^{Reference Achiha, Nakajima, Ohzawa, Koh, Yamauchi, Kagawa and Aoyama51} LNMO/graphite cell data indicated that electrolytes E5 and E6 were not compatible with the graphite anode. To realize the actual 5 V energy output of LNMO/graphite systems, further modifications are needed. A new fluoroethylene carbonate-based, fluorinated, high-voltage electrolyte (HVE), reported by Hu et al.,^{Reference Hu, Zhang and Amine52} showed much improved cell stability and performance due to its capability of SEI formation on graphitic anodes^{Reference McMillan, Slegr, Shu and Wang42} by reductive decomposition at a potential higher than that of EC.

The formulation of the HVE consisted of 1.0 M LiPF₆ in fluoroethylene carbonate (FEC)/F-EMC/F-EPE with a 3/5/2 ratio by volume. The cycling performance of the HVE cell compared to the Gen 2 cell at RT and 55°C is shown in Figure 7a–b. At 55°C, the HVE cell exhibited considerable improvement in capacity retention over the Gen 2 cell, indicating that electrolyte decomposition was mitigated with the HVE. For the HVE cell, at both RT and 55°C, the Coulombic efficiency remained above 99.5% for the majority of cycles, signaling much less decomposition even at 55°C. Long-term cycling data showed that the HVE cell was able to retain 50% of its initial capacity over 600 cycles at RT and 250 cycles at 55°C at a rate of C/3. These results indicate that although other cell components^{Reference Chen, Xu, Xiao, Engelhard, Ding, Mei, Hu, Zhang and Zhang53} affect the high-voltage cell performance to a certain degree, the electrolyte plays the most significant role.

Figure 7.

Cycling performance of LNMO (LiNi_0.5Mn_1.5O₄)/graphite cells at (a) room temperature (RT) and (b) 55°C with the high-voltage electrolyte (HVE) and Gen 2 electrolytes. (Cutoff voltages of 3.5–4.9 V, rate of C/3, where C/3 is a three-hour discharge.) Fourier transform infrared spectra of pristine and harvested cathodes from cycled LNMO/graphite cells with (c) Gen 2 and (d) HVE electrolytes and harvested anodes from cycled LNMO/graphite cells with (e) Gen 2 and (f) HVE electrolytes. All cells cycled at RT and 55°C. Reprinted with permission from Reference 52. © 2013 Elsevier.

To understand the mechanism behind this improvement, electrochemical diagnosis and characterization of the harvested electrodes and electrolytes are undertaken.^{Reference Hu, Zhang and Amine52} Differential capacity analysis (dQ/dV, where Q, cell capacity; V, cell voltage) of graphite/Li cells with the HVE and Gen 2 electrolytes indicated that formation of a SEI layer occurred at a voltage of about 0.6 V versus Li⁺/Li with the Gen 2 cells and at the much higher voltage of 1.5 V versus Li⁺/Li with the HVE cells, due to the high reduction potential of the FEC in the electrolyte. The higher reduction potential of FEC may contribute to a more stable SEI on the graphite anode at elevated temperatures and high charging voltages.^{Reference Peled, Golodnitsky, Menachem and Bar-Tow54}

Figure 7c–f shows FTIR spectra of the harvested LNMO cathodes and graphite anodes. The FTIR characterization was also in agreement with electron microscopy (SEM) characterization (not shown).^{Reference Hu, Zhang and Amine52} On the harvested LNMO cathodes from Gen 2 cells, while no obvious organic buildup was identified at RT, at 55°C, a prominent peak appeared in the ∼1720 cm^–1 region, corresponding to the carbonyl functional group. This peak signifies considerably more decomposition compared with room-temperature cycling where the decomposition products may be carbonates or carboxylate derivatives. In the case of the HVE, although the trend is similar to that of the Gen 2 electrolyte, the change is much less dramatic, and the carbonyl peak is trivial.

The more interesting case actually lies in the anode. From RT to 55°C, the graphite anode with Gen 2 electrolyte underwent a significant increase in decomposition.^{Reference Hu, Zhang and Amine52} This indicates a significant morphology change of the SEI layer under high-temperature operation, again consistent with SEM data. In comparison, the HVE anode showed less organic residue in the FTIR spectrum (Figure 7f), indicating the presence of a SEI composed of more inorganic species on the graphite surface as observed in the scanning electron microscope. No visual difference was observed between the RT and 55°C data. These results demonstrate the superior high-temperature stability of the SEI layer formed by the HVE on the graphite anode during high-voltage charging and discharging.