ANALYSIS OF INTERPHASE FORMED ON THE ELECTRODES OF LITHIUM ION BATTERIES

Lithium-ion batteries (LIBs) are one of the most widely used energy sources, especially for portable electronics. However, the development of lithium ion batteries for Electric Vehicle (EV) and Plug-in Hybrid Electric Vehicle (PHEV) requires the research and development of improved electrolytes. The development of electrolytes which allow LIBs to perform over a wide temperature range and operating potential are of significant current interest. The interphase formed on the surface of the electrodes generally governs kinetics of charging and discharging and is an important factor in life of LIBs. Favorable electrode interphases can be generated by altering the composition of the electrolyte. Commercially available LIBs have a normal discharge voltage around 3.7V where the electrolyte oxidation on the surface of cathodes is not a significant problem. Recent research in high voltage cathode materials (>4.5 V vs Li/Li) to increase the power and energy density of LIBs for EV applications has raised concerns about the stability of LiPF6/carbonate based electrolyte to oxidation. Furthermore, the flammability of organic electrolyte hinders LIBs’ application in the EV market. Detailed investigations of improved electrolyte systems which can address the above issues will be presented. Components of the interphase are detected using various surface analysis techniques such as XPS, FTIR and SEM.


Typical lithium ion battery
Typical LIB ( Fig.1) consists of anode and cathode separated by a thin separator made up of polyethylene or polypropylene which allows ionic conductance through electrolyte but prevents electrons to pass through. So the separator prevents direct electrical contact between anode and cathode. Anode is a thin graphite sheet deposited on a copper current collector which allows reversible Li + ion intercalation at very low potentials vs.
Li/Li + . Fully lithiated carbon can provide theoretical specific capacity of 372 mAh/g.
Quest for high capacity materials has led to silicon and tin-based anodes which can form alloys with lithium and provide capacity as much as 4200 and 990 mAh/g. 4

Working mechanism of LIBs
Lithium ion batteries are assembled in the discharged state with all the lithium in cathode host. Fig. 1 shows charging (left) and discharging (right) process of typical LIB. (1) It should dissolve salt to sufficient concentration (high dielectric constant, ε).
(2) It should have low viscosity, η and high conductivity.

Introduction
While commercial lithium-ion batteries (LIBs) perform well for most home electronic applications, currently available LIB technology does not satisfy some of the performance goals for aerospace applications or plug-in hybrid electric vehicles because of potential safety concerns. The electrolytes used in commercial LIBs are composed of LiPF 6 dissolved in organic carbonates or esters. 1,2 The organic solvents present in lithium-ion battery electrolytes are volatile and flammable. Battery abuse can initiate numerous highly exothermic reactions of the electrolyte with the electrode materials potentially resulting in fire or explosion. 3 These reactions create significant safety concerns for lithium ion batteries especially for large-scale battery applications. One area of interest is the replacement of liquid electrolytes with solid electrolytes. Many polymer and polymer-gel electrolytes have been investigated and have improved flame resistance. 4 However, the improvements were accompanied with problems including significantly lower conductivity, especially at low temperature. Due to the inherent low conductivity of solid electrolytes they are not practical for many applications where high power and low temperature performance are required.
Recent efforts to address this problem have focused on developing flame retardants (FR) as electrolyte additives or co-solvents. Initial investigations into FR additives for lithium-ion batteries have focused on organophosphorus compounds, which suppress flame propagation and thermal runaway. So far, investigators have focused on trialkylphosphates such as trimethyl phosphate, 5 triethyl phosphate, 6 triphenyl phosphate, 7 tris (2,2,2-trifluoroethyl) phosphate, 8 and bis (2,2,2-trifluoroethyl) methylphosphate, 9 and phosphoramides including hexamethylphosphoramide, 10  were utilized to maintain a constant concentration of EC compared to the baseline.
The SET of the electrolytes was measured as described by Xu and co-workers. 6 cycles at C/10, 4 th , 5 th and 6 th cycles at C/5, 7 th and 8 th cycles at C/10 and remaining cycles at C/5. All cell types were prepared in duplicate or triplicate to confirm reproducibility of cycling behavior. Representative data sets are provided. After 30 cycles the full cells were disassembled in the glove box. The electrodes were rinsed with DMC to remove residual EC and LiPF 6 followed by overnight evacuation.
The XPS spectra were conducted with a PHI 5500 system using Al Kα radiation (hυ=1486.6 eV) under ultra high vacuum. The graphite peak at 284.3 eV was used as a reference for the final adjustment of the energy scale in the spectra. The spectra obtained were analyzed by Multipak 6.1A software and fitted using XPS peak software (version 4.1). A mixture of Lorentzian and Gaussian functions was used for the leastsquares curves fitting procedure.
SEM was conducted on a JEOL-5900 SEM. FTIR spectra were acquired with a Thermo Nicolet IR iS10 spectrometer. Argon was purged during the acquisition process to avoid air exposure. The spectra were acquired in multiple locations on the electrodes in the ATR mode with 4 cm -1 resolution with 128 total scans.

Results and Discussion
Conductivity measurement Figure 1 shows the conductivity of the electrolyte solutions. There is no difference in the conductivity of the solutions containing 15 or 20% of DMMP. The conductivity of the baseline solution is slightly lower than the DMMP containing electrolytes at temperatures above ~25°C and the freezing point of both DMMP containing electrolytes is slightly higher than the baseline electrolyte, but otherwise the conductivities of the different electrolytes are very similar. The specific conductivity (σ) over the temperature range of -40 to +80°C cannot be described by the activation mechanism mathematically expressed as the Arrhenius law for any of the solutions investigated: Where σ 0 is a constant, Ea is the activation energy for the ion's motion in solution, and When ion transport is facilitated by the mobility of the solvent molecule, the conductivity may obey the Vogel-Tammann-Fulcher (VTF) relation. 19,20,21 Where A, B, and T 0 are constants. Although the exponential term looks similar to that in the Arrhenius equation (1), the B term of (2) does not have the same meaning of activation energy. B is related to the activation energy for the creation of critical free volume for ion transport, B = E a , VTF /R. T 0 is the temperature at which the transport function ceases to exist or the solvent structural relaxation becomes zero. T 0 may be considered as the glass transition temperature or a temperature near it. To determine the parameters of the VTF equation a non-linear least squares analysis method was utilized. 22 Figure 1 confirms that the agreement between experimental data (circles) and the VTF approximation (lines) is very good (the relative residual standard deviation was typically less then 1.5%). The VTF parameters are collected in Table 2.

Self Extinguishing Time (SET)s
The flame-retarding effect of DMMP additive is summarized in  Figure 3 presents a cyclic voltommagram of Electrolyte 1 (Table 1) on a glass carbon electrode. During the first cathodic potential sweep, a reduction peak is observed at 1.7 V, which is attributed to the reduction of oxalate-moieties present in the LiBOB salt. 27 On the second sweep, the reduction peak located at 1.7 V is not observed consistent with the formation of a stable oxalate-derived anode SEI during the first cycle. This suggests that the presence of DMMP does not interfere with SEI formation in electrolytes containing LiBOB.

Electrochemical performances of cells with DMMP
The electrochemical stability of DMMP on positive and negative electrodes of LIB was evaluated in lithium metal-MCMB (anode half-cell) and lithium metal-LNCO (cathode half-cell) during galvanostatic cycling. Figure 4 shows the cycling performance of anode half-cells with baseline electrolyte and electrolytes 2 and 3 (Table 1) (Table 1) is provided in Figure 5. The cycling efficiencies of the initial cycles were calculated to be 95% and 97％ for the baseline and 20 % DMMP electrolyte respectively. It is believed that that the irreversible capacity loss during the first cycle is attributed to the irreversible structural change of LiNi 0.8 Co 0.2 O 2 crystalline lattice and the formation of surface layer due to organic solvent oxidation. 28 Thus, it is reasonable to suppose that the electrochemical oxidation of DMMP is responsible for the reversibility decrease (97 vs. 95%). The capacity and coulombic efficiency on the sequential cycles were identical for the cells with both solvent's compositions ( Figure   5). The trends in the discharge capacities of the full cells containing Electrolyte 1, Electrolyte 4 (without LiBOB), and Electrolyte 5 (with LiBOB) ( Table 1)

FTIR
The surfaces of the electrodes were also analyzed by FTIR-ATR spectroscopy. The IR spectra of the anodes and cathodes extracted from cycled full cells are provided in The P2p spectrum contains a peak characteristic of LixPOyFz (134 eV) for all samples.

Conclusions
Non-Flammable electrolytes for lithium ion batteries can be prepared with        20 Electrochemical impedance spectroscopy and ex-situ surface analysis of cycled cells suggest that the addition of LiBOB inhibits the growth of a cathode surface film and improved cell performance. Reflectance (FTIR-ATR) analysis of the cathodes was carried out with Thermo Nicolet iS10 IR spectrometer which has a smart performer accessory with a Germanium crystal.

MANUSCRIPT -II
For each sample, 128 scans were collected at two/three different locations, constantly purged with high purity Argon. SEM analyses were conducted on a JEOL-5900 environmental SEM.

Cycling performance
The discharge capacity of Li/LiNi 0.5 Mn 1.5 O 4 cells containing LiPF 6 /carbonate electrolyte with and without added LiBOB over the first fifteen cycles are depicted in Figure 1. Cells containing baseline electrolyte showed higher discharge capacity after the first cycle but gradual capacity fade is observed upon subsequent cycling consistent with previous reports. 8,9 Cells containing LiBOB showed lower discharge capacity after first cycle but capacity retention was much improved in following cycles. Cells containing 0.25% LiBOB retained almost 93% of the initial discharge capacity as compared to 71% capacity retention for cells containing standard electrolyte. The cell efficiency was also improved with added LiBOB. Cells containing the standard electrolyte have relative low efficiency (~94%). Incorporation of LiBOB significantly improves cell efficiency with efficiencies exceeding 99% for cell with 0.25% LiBOB ( Figure 1).

Electrochemical impedance spectroscopy (EIS)
The peak of Li x PO y F z has a similar bonding energy to PVDF and is not observed due to peak overlap and low intensity.

Li-ion cells containing methyl butyrate-based wide operating temperature range electrolytes Introduction
There is a continued desire to develop improved lithium-ion batteries that can operate efficiently over a wide temperature range, while still providing long life characteristics.
The narrow operating range of the state of art systems has been identified by the

Formation Characteristics
As illustrated in Table 1 good cycle life characteristics can be obtained. It should be noted that the observed trends cannot be entirely attributed to electrolyte effects, and that some small differences in reversible capacity can be partly attributed to cell to cell variability (i.e., <5% variation in electrode weights).

Discharge Characteristics.
After performing the formation cycling, the cells were subjected to systematic discharge rate characterization testing over a wide temperature range. These tests consisted of charging the cells at ambient temperature and then discharging at the desired temperatures after soaking the cells for at least four hours prior to discharge and the results are summarized in Table 2.
Given the rationale of adding the electrolyte additives was to improve the high temperature resilience, it is significant that the discharge rate capability is somewhat comparable for all of the formulations investigated. This suggests that the electrolyte additives are not having a negative impact upon the cell impedance of the cell, which would limit the low temperature performance. For example, as displayed in Fig. 1, when the cells were discharged at ~ C/16 discharge rate at -40 o C, very comparable performance was obtained in all cases with approximately 75% of the room temperature capacity being delivered.
In contrast, in many cases, the cells containing the electrolyte additives actually deliver better performance than the baseline formulation under many conditions. As shown in Fig. 2, more differentiation of the cells can be observed under high rate conditions at low temperatures (i.e., using a C/4 discharge rate at -40 o C). As illustrated, the cells containing the LiBOB, VC, FEC, and lithium oxalate all outperformed the baseline solution without any additive. These results suggest that the lithium intercalation/de-intercalation kinetics are more favorable with the cells containing the electrolyte additives, presumably due to preferable SEI characteristics. In order to decipher the influence of the electrolyte additives upon the kinetics of the respective electrodes, detailed electrochemical characterization of the cells was performed, as described in the section below.

High Temperature Cycling
After performing the rate characterization testing, cycling tests at high temperatures were performed on the cells to determine their high temperature resilience. This consisted of performing 20 cycles at 60 o C, followed by electrochemical characterization, which was in turn followed by performing an additional 20 cycles at 80 o C. As illustrated in Fig. 3, the cells containing the VC and lithium oxalate additives displayed the best capacity retention after being subjected to cycling at 60 o C. This trend followed for the most part even after cycling at 80 o C, with the cell containing the VC additive displaying the best performance. It is, however, likely that the underlying capacity fade mechanisms may be different at these two temperatures, and for the various electrolyte formulations (i.e., the performance characteristics of the anode and cathode may degrade at different rates depending upon the electrolyte type) to account for the observed trends in the high temperature resilience.

Electrochemical characteristics
Based on previous work, it is believed that the improved rate capability at low temperatures of Li-ion cells that utilized these ester-based solutions is primarily attributed to a combination of improved mass transfer characteristics in the electrolyte (higher ionic conductivity) and facile kinetics of lithium intercalation/de-intercalation at the interface due to favorable film formation behavior at the electrode surfaces. 18,20 To enhance this understanding on the impact that the electrolyte additives have upon these factors as well as any benefit they may have to increase the high temperature resilience, we have assessed the electrochemical characteristics of the systems using a number of techniques, including Tafel polarization measurements, Electrochemical Impedance Spectroscopy (EIS), and linear micro-polarization measurements.

Electrochemical impedance spectroscopy (EIS)
To interpret the impedance patterns from Li-ion cell electrodes an equivalent circuit is typically used which consists of two relaxation loops and is comprised of a series resistance, R s, which represents an algebraic sum of the electronic resistance from both the electrodes, leads and the electrolyte ionic resistance, a parallel resistor-capacitor network (C f and R f ) in series for the high frequency relaxation loop associated with surface film between the electrolyte and the electrodes, another resistor-capacitor parallel network in series for the low frequency relaxation loop (C dl and R ct ) and a Warburg impedance (w), to represent a slow solid state diffusion of lithium inside the cathode or anode 25-29 A schematic of the equivalent circuit used in the analysis in the present study is provided in our previous publications. 30 In the event that more than two relaxation loops are observed, the addition of resistor-capacitor networks in series were added to represent the multi-layer lamina formed parallel to the surface. 31 The capacitors are replaced with constant-phase elements (CPE) to represent their non-ideal characteristics, whenever needed. For cathode measurements, an inductor was appropriately added to account for the magnetic properties attributable to the aluminum substrate. The impedance data were analyzed, using the above equivalent circuit and ZSimWin CNLS modeling software from Princeton Applied Research.

Tafel polarization measurements
The lithiation/de-lithiation kinetics were determined for the anodes and the cathodes by conducting Tafel polarization measurements of the MCMB-Li x Ni y Co 1-y O 2 cells in contact with the various electrolytes. The measurements were conducted on fully charged cells (i.e., cells were charged at room temperature and the open circuit potential was above 4.08V) and allowed to equilibrate at the desired temperatures for at least two hour prior to performing the tests. These measurements were performed under potentiodynamic conditions, using slow scan rates (0.2mV/sec), approximating steadystate conditions. In all of these Tafel plots, there are distinct charge-transfer controlled regimes, where the overpotential increases linearly with log (I). The effect of mass transfer seems to be relatively insignificant, such that kinetic parameters, i.e., exchange current and transfer coefficients can be obtained. In cases, where there is mass transfer interference, proper correction was applied in the data analysis. These rate parameters, i.e., the exchange currents and transfer coefficients, for the intercalation/de-intercalation of lithium were determined from the intercept and the slope of the mass-transfer corrected plots (i.e., generated by plotting the logarithm of I/(1-I/I lim ) against the electrode potential).
As illustrated in Fig. 8, with the exception of the cell containing the LiBOB, improved lithium de-intercalation kinetics (i.e. higher limiting currents) were observed for the anodes in contact with the electrolytes possessing the electrolytes additives compared to the baseline solutions, suggesting that desirable surface films have formed in these cases.
As illustrated in Fig. 9, in which the Tafel polarization measurements have been performed on the LiNiCoO 2 cathodes at room temperature, the cell containing the VC and lithium oxalate electrolyte additives displayed enhanced lithium kinetics (i.e., higher limiting current densities) compared to the baseline system that does not contain the additive. This observation, namely that VC has a beneficial effect upon the nature of the SEI layer on the cathode and the corresponding lithium kinetics, has been previously found in our previous studies involving different solvent mixtures.
Generally, the trends with regard to the observed lithium kinetics and how they depend upon electrolyte type tend to track well with temperature. For example, when the Tafel measurements were performed on the MCMB anodes at low temperature, as displayed in Fig. 10, the cells containing the FEC and lithium oxalate additives delivered somewhat improved performance over the baseline solution. However, when the the LiNiCoO 2 cathodes were measured at -30 o C, as shown in Fig. 11, a different trend was observed compared to that displayed at 20 o C, with the cell containing the LiBOB delivering significantly better performance.

Elemental concentration on the MCMB anode surface
The fresh MCMB composite electrode has high concentrations of C and F due to the presence of PVDF binder. There is also a low concentration of O due to the presence of However, an additional decrease in the intensity of peaks characteristic of C=O bonds (~289 eV) is observed. The XPS element spectra for anodes cycled with LiBOB containing electrolyte are very similar to the element spectra for anodes cycled with Li 2 C 2 O 4 , except they contain an additional B1s peak at 192 eV for borates generated from the reduction of LiBOB on anode surface. 33 The anodes extracted from the cells containing VC have similar element spectra, except that the intensity of the peak characteristic of LiF is much greater than for any of the other samples.

FTIR and SEM of cycled MCMB electrodes
The IR spectra of all of the anodes extracted from cells after cycling are dominated by the peaks associated with the PVDF binder at 1400, 1271, 1170, 1070, 877, and 840 cm −1 . All the anodes contain a broad peak at ~ 1600cm -1 corresponding to C=O stretch from oxalate containing species. 34 The anodes cycled with the baseline electrolyte and the baseline with FEC have a strong absorption at 1730 cm -1 characteristic of poly(ethylene carbonate). 35 The anodes cycled with Li 2 C 2 O 4 and LiBOB also contain the peak at 1730 cm -1 although with weaker intensity. The anode from the VC containing electrolyte shows additional small peak at 1778 cm -1 which is due to the presence of poly(VC).
SEM images of cycled anodes indicate that the bulk MCMB particles are covered with a thin film of electrolyte decomposition products characteristic of the SEI. However, the bulk spherical MCMB particles are intact and the surface coverage is similar for anodes cycled with all electrolytes.                 (4212 mAh/g) 14 compared to graphite (372 mAh/g). 4 One of the primary obstacles for the use of silicon as an anode material is the large volume expansion (200-300%) during lithiation. 5 This results in large irreversible capacity during first cycle and poor cycle life. 6 To overcome this problem researchers have employed nano-structured silicon particles, thin film silicon and amorphous silicon anodes. 7 SEM analyses were conducted on a JEOL-5900 environmental SEM.

Electrochemical performance of Si/Li cells
All of the cells show a first cycledischarge capacity >3200 mAh/g. A high dicharge capacity (>2800 mAh/g) is retained for all cells during the first ten cycles. Cells without film forming additives have rapid capacity fade after the first ten cycles. The cell containing PC retains only 21% of initial discharge capacity while the baseline electrolyte, containg EC, retains only 59% of intial discharge capacity after the first 55 cycles. The incorporation of additives improve capacity retention of Si anode cell by about 12 -24% over the baseline electrolyte. The initial discharge capacity is best with the cells containing VC and LiFOB. While the capacity retention is best with cells conatining VC and FEC.
Cycling efficiency of Si/Li half cells is displayed in fig.2 Figure 1.

Surface analysis of Si electrodes
Elemental concentration on electrode surface XPS spectra of cycled Si electrodes extracted from cells after 55 cycles supports changes in the composition of the anode SEI due to changes in electrolyte composition ( Figure 3, Table 2). The elemental concentration of the fresh electrode contains 39 % oxygen and 37 % silicon, a carbon peak is observed associated with universal carbon contamination. After cycling with all EC containing electrolytes, Si is no longer observable on the surface of the electrodes consistent with the generation of a thick SEI.
A low concentration of Si is observed on the surface of the cell cycled with PC based electrolyte consistent with a thin SEI.

XPS of Si electrodes
The fresh electrodes contain a Si peaks at 99.6 and 102.3eV consistent with pure silicon and surface silicon oxide respectively (not shown in Figure 3) species is present at ~532 eV. The F1s spectrum contains a peak for LiF. There are no peaks in the C1s, O1s, or F1s spectra consistent with the presence of C-F bond or poly(vinylene carbonate) (poly(VC)). 15 The surface of the Si electrode cycled with electrolyte containing FEC is covered with a passivation layer composed of lithium alkyl carbonates, polycarbonates, and oxalates. The C1s of the electrode cycled with VC containing electrolyte contains a peak at 291 eV which is due to poly(VC). 15 The corresponding poly(VC) peak is observed in the O1s spectrum at 534.3 eV. The other C1s and O1s peaks observed support the presence of C-O and C=O containing species.
The element spectra of the anodes extracted from LiBOB and LiFOB are similar. The C1s spectrum contains a large peak at ~290 eV characteristic of oxalates. The O1s peak at 531-532 eV is also consistent with the presence of oxalates. Interestingly, the F1s spectra of the anodes extracted from cells containing LiBOB and LiFOB containin significant concentrations of Li x PF y O z . The element spectra for the anode cycled with PC electrolyte is similar to the anode cycled with the baseline electrolyte except the peaks characteristic of C-O and C=O species have greater intensity.

FTIR analysis
The IR spectrum of the anode extracted from the cell containing the baseline electrolyte contains peaks at 1640, 1420, 1330, 1070 and 840 cm -1 (Figure 4). The peaks at 1600-1700 cm -1 and 1300-1350 cm -1 are consistent with the presence of lithium alkyl carbonates and lithium oxalate, the peak centered at ~1420 cm -1 is characteristic of Li 2 CO 3 and the broad peak centered around 1000 cm -1 is characteristic of C-O peaks in ethers and Li x PF y O z . 16 The IR spectra of the anodes extracted from cells cycled with additives are similar to the spectra of the baseline samples with a few important differences. The sample with VC contains a weak peak at ~1800cm -1 due to poly(VC).