Fundamentals of rechargeable lithium ion and beyond lithium ion batteries

As shown in figure 1.1, a conventional intercalation battery setup includes three major components: cathode, anode (a graphite anode, or Gr, is used as an example), and electrolyte. An example of an intercalation cathode material is lithium cobalt oxide (LiCoO₂), or LCO, which is widely adopted for portable electronics. Because cobalt resources are limited and geopolitical tensions arise from cobalt mining, current research is devoted to development of transition metal oxides (TMOs) that are low in cobalt but high in nickel and manganese. However, because of the high volumetric energy density of LCO, it is still favored over other TMOs for portable electronics. EVs, on the other hand, are more forgiving with respect to volumetric energy density, and the emphasis there is on high gravimetric energy density and cost effectiveness.

The anode for lithium ion batteries is usually graphite due to its low potential, safety, and high specific capacity. More specific details are given in section 1.2.

An easy way to characterize a nonaqueous electrolyte is through its electrochemical window, a practical value that can be associated with the theoretical quantity E_g. The traditional view is that the Lowest Unoccupied Molecular Orbital (LUMO) is associated with cathodic stability, the Highest Occupied Molecular Orbital (HOMO) is associated with anodic stability, and the difference between LUMO and HOMO characterizes the electrolyte's electrochemical window. However, it is better to explain the electrochemical window in parallel with the transition-metal energy band, as many of the physical and chemical phenomena in batteries are related. The potentials of the cathode (μ_c) and anode (μ_a) are determined by their Fermi energies (E_F) in the itinerant-electron band, and the μ_c and μ_a consequently determine the cell's open-circuit voltages V_oc = (μ_a−μ_c)/e. As shown in figure 1.2, the μ_c and μ_a should ideally be within the bracket of LUMO and HOMO. However, in most LIBs, because of the low reduction potential of the Gr, 0.1 V versus Li/Li⁺ (all voltages subsequently cited in this paper will use Li/Li⁺ as the reference), almost no electrolytes will have a higher LUMO than lithiated graphite.

Zoom In Zoom Out Reset image size

The practical way to expand HOMO and LUMO is to have a solid–electrolyte interface (SEI) that successfully blocks the reduction of electrolyte in contact with the anode, and yet effectively transports the Li⁺ ions.

In a typical LIB with a TMO cathode and a Gr, different electrochemical reactions take place at the cathode and the anode simultaneously. The general Faraday's Law can be used to calculate the theoretical capacity of electrode materials; it depicts the relationship between the current that runs through the external circuit and the amount of the corresponding active materials that are consumed/produced in the electrochemical reaction. For example, in a typical graphite reduction,

$$\begin{eqnarray}&&{{\rm{C}}}_{6}+{{\rm{Li}}}^{+}+{e}^{-}\to {{\rm{LiC}}}_{6}\end{eqnarray} \tag{ 1.2 }$$

Using Faraday's Law, F = qN_AV: where q is the fundamental unit of charge, and N_AVO is Avogadro's number, Faraday's constant F has a value of 96 485 C mol⁻¹. The molecular weight of C₆ (graphite) is 72 g mol⁻¹, and the theoretical capacity of graphite can be calculated as

$$\begin{eqnarray}&&1\,{{\rm{m}}{\rm{A}}{\rm{h}}{\rm{g}}}^{-1}={10}^{-3}\cdot 1\,{\rm{C}}\cdot {{\rm{s}}}^{-1}\cdot 1\,h\,\cdot 3600\cdot \displaystyle \frac{s}{h}\cdot {{\rm{g}}}^{-1}\end{eqnarray} \tag{ 1.3 }$$

$$\begin{eqnarray}&&1\,{\rm{C}}\cdot {{\rm{g}}}^{-1}=\displaystyle \frac{1}{3.6}\,{{\rm{m}}{\rm{A}}{\rm{h}}{\rm{g}}}^{-1}\end{eqnarray} \tag{ 1.4 }$$

$$\begin{eqnarray}&&{\rm{Capacity}}=\displaystyle \frac{96485\,{\rm{C}}/{\rm{mole}}}{72{\rm{g}}/{\rm{mole}}}=984\,{\rm{C}}\cdot {{\rm{g}}}^{-1}=\displaystyle \frac{984}{3.6}\,{\rm{mAh}}{{\rm{g}}}^{-1}=273\,{\rm{mAh}}{{\rm{g}}}^{-1}\end{eqnarray} \tag{ 1.5 }$$

From this example, a general formula can be deduced for the calculation of a theoretical capacity:

$$\begin{eqnarray}&&{\rm{Capacity}}=\displaystyle \frac{\displaystyle \frac{96485}{3.6}\,\cdot z}{{\rm{Mw}}}=\displaystyle \frac{26800\cdot z}{{\rm{Mw}}}{\rm{mAh}}{{\rm{g}}}^{-1},\end{eqnarray} \tag{ 1.6 }$$

where z is the number of electrons generated per molecule.

The fundamentals of electrochemistry are covered well in the classical electrochemistry books. A recent book by Fuller and Harb also offers both the basics and in-depth study materials for topics such as thermodynamics, electrochemical kinetics, and transport, with examples of how each concept is used [3]. We encourage readers to refer to those books for an understanding of the basics. Three fundamental equations, the Debye–Hückel equation (activity coefficient), Butler–Volmer equation (electrochemical kinetics), and Nernst–Plant equation (transport properties), are given here for the readers' reference:

$$\begin{eqnarray}&&\text{Debye-H\unicode{x000FC}ckel equation}\,{\rm{ln}}\,{\gamma }_{\pm }={z}_{+}{z}_{-}\alpha \sqrt{I}\end{eqnarray} \tag{ 1.7 }$$

$$\begin{eqnarray}&&\text{Butler-Volmer equation}\,i={i}_{a}+{i}_{c}={i}_{0}\left[{\rm{\exp }}\left(\displaystyle \frac{(1-\beta )F{\eta }_{s}}{{RT}}\right)-{\rm{\exp }}\left(-\displaystyle \frac{\beta F{\eta }_{s}}{{RT}}\right)\right]\end{eqnarray} \tag{ 1.8 }$$

$$\begin{eqnarray}&&\text{Nernst-Plant equation}\,{N}_{j}=\mathop{-{z}_{i}{u}_{i}F{c}_{i}\nabla \phi }\limits_{\text{migration}}\mathop{-{D}_{i}\nabla {c}_{i}}\limits_{\,\text{diffusion}}\mathop{+\,{c}_{i}v}\limits_{\,\text{convection}}\end{eqnarray} \tag{ 1.9 }$$

Three parameters are identified as the important criteria for measuring the performance of a battery: coulombic efficiency, specific capacity, and capacity retention. Recently, more detailed studies have been implemented to get a deeper understanding of the operation of a battery, mainly addressing: (1) the individual voltages of cathodes and anodes using a reference cell; (2) detection of lithium slippage; (3) application of dV/dQ and dQ/dV; and (4) impedance studies on both the anodes and cathodes.

Figure 1.3 illustrates the overall voltages of the batteries, as well as the individual voltages experienced on the cathode and anode. The method commonly used to measure the individual voltages is to use a third Li metal reference electrode that can either be made in-house or purchased. The same concept can be adapted for measurement of individual impedance on the cathode and anode. Interestingly, in an NMC//Gr cell, the initial cell voltage after assembly is ~0 V; however, through the initial charge pulse (for example, a current of 1.5 C rate), the cathode stays at a relatively stable voltage of 3.7 V [4].

Zoom In Zoom Out Reset image size

A typical coin-cell anatomy is shown in figure 1.3(A). A more advanced version, a so-called reference cell, is shown in figure 1.3(B). This type of cell consists of two 20.3-cm² electrode disks separated by two layers of a 25-μm-thick microporous separator (Celgard 2325). The unique feature of this reference cell is the Cu–Li alloy wire used as a reference. Lithium metal was plated in situ onto the tip of this thin (25-μm-dia.) Cu wire to form a microprobe reference electrode (RE). Only a 2- to 3-mm portion of this wire was exposed to the electrolyte and coated with Li; the remainder was insulated by a polyurethane coating. The Cu wire was lithiated at 5 μA for 6 h, with the cathode as the source of lithium. The advantage of such a Li–Cu alloy wire is that it provides an accurate reference for both the potential and the impedance. A typical compiled voltage profile consisting of the anode potential, the cathode potential, and the full cell potential is shown in figure 1.3(C).

Cycling protocols (determining which electrochemical condition the cells will experience) are also crucial for battery performance evaluation and mechanism diagnosis. For example, a comprehensive cycling protocol was implemented at Argonne National Laboratory to impose harsher conditions than normal cycling, and triplicate cells were used to provide for repeatable examination of the cells [5]. In general, the cycling procedure can be divided into four electrochemical steps: formation, preparation, HPPC (Hybrid Pulse Power Characterization), and aging. The formation step consists of five slow C/20 cycles to ensure the formation of a robust SEI layer. Preparation consists of a slow C/10 cycle. HPPC cycles are specifically designed to mimic the EV operating conditions occurring at the 'accelerating' or 'braking' stages, with large pulses going through the cell and the impedance recorded at different states of charge (SOCs). Over the course of 119 cycles, periods of slow C/20 cycles were inserted before each of the 20 fast C/3 cycles (classified as the 'aging' step) so that capacity could be indicated with less influence from the impedance. The HPPC segments were purposely reinstituted before each of the 20 cycles of the C/3 aging. A rather aggressive cycling protocol was used: the galvanostatic charge (constant current) was followed by a constant voltage hold of 3 h. Through a standard cycle package of 119 cycles, information is obtained with respect to (1) the Area Specific Impedance (ASI) change during the cycling [6]; (2) aging, reflecting the capacity retention (through both the major C/3 cycles and periodic C/20 cycles); and (3) dQ/dV, from the C/20 cycles [5]. ASI can be measured through the HPPC cycles using the following equation, with the parameters V_t , V₀, t₀, and t₁ shown in figure 1.4(b) [6]:

$$\begin{eqnarray}&&\text{ASI}\,=\,\displaystyle \frac{{V}_{{t}_{0}}-{V}_{{t}_{1}}}{{I}_{{t}_{1}}-{I}_{{t}_{0}}}\end{eqnarray} \tag{ 1.10 }$$

Zoom In Zoom Out Reset image size

Another key concept in battery research is the slippage of voltage. The original voltage profiles for a representative NMR//Gr cell is shown in figure 1.5(a), the V_c0 at 4.22 V and V_a0 at 0.12 V, once the cycled electrodes reach the charge voltage and in the absence of the reference cell, the voltage reflected through the cell potential is the difference between the positive and negative electrode (V_c = V_c−V_a). Using an NMC//Gr-Si cell as an example, the slippage of capacity results in a large increase in the electrode potential of the anode. In one scenario where no Li⁺ inventory loss and no cathode degradation occurs, the endpoint voltages and discharge/charge profiles of the cathode and the anode will remain the same. However, using the NMC cathode as an example, when a charge cutoff of 4.1 V is maintained, as the cathode capacity decreases (anode capacity slips to the right), the anode voltage will slightly increase (V_a1 = 0.15 V). In the next cycle, the cathode needs to reach a higher individual voltage to maintain the V_c1−V_a1, resulting in higher polarization of the cathode (V_c1 > V_c0). The voltage profile of graphite is featured with plateaus at ~ 120 mV and 90 mV, and loss of the capacity when cells are cycled at the same upper cutoff voltages will push the voltage in the anode to the right with respect to the cathode. For example, in figure 1.5(b), as the cell ages, the capacity decreases, as denoted by the gray area of Li⁺ inventory loss, resulting in the increase of anode voltage at 0.15 V. As shown in figure 1.5(c), when a Li⁺ gain occurs in the anode side, slippage towards the right of the anode is observed (V_a2 = 0.09 V).

Zoom In Zoom Out Reset image size

Nowadays, LCO is still the cathode of choice for portable electronics because of its high volumetric energy density. However, the dominant usage of LCO raises concerns over the geopolitical and environmental aspects of cobalt mining. The desire for a reduction in the cobalt component, and for other characteristics including higher voltage, high rate capability, long cycle life, and high energy density, has motivated research on cathodes for LIBs. Concerted efforts to improve the performance of LCO have included structural modifications to improve both its stability and resistivity, and the modification of the electrolyte–electrode interface to reduce cathode impedance rise [15]. In contrast to the high compact density of LCO (4.05–4.15 g cm³) [16], NMC materials only have a compact density in the range of 3.1–3.6 g cm³. Despite the theoretical capacity of 273 mAhg⁻¹, only half of the Li⁺ inventory in LCO is practically usable with a cutoff voltage of less than 4.2 V versus Li/Li⁺, which is necessary for maintaining structural stability over prolonged cycling [15, 17–19]. Recognized as one of the first layered materials through the work of Goodenough [20], LCO has a layered α-NaFeO₂ structure, featuring a rhombohedral structure with an R3m space group, with the oxygens arranged in a cubic close-packed network and lattice constants of a = 2.816 Å and c =14.052. Li and Co³⁺ alternate in the (111) planes to form CoO₆ and LiO₆ octahedra. The CoO₆ octahedra share edges and form CoO₂ sheets, separated by sheets of octahedral Li [21]. During delithiation, the structure first undergoes expansion of the c axis due to the loss of Li and weak Van der Waals expulsion between the CoO₂ sheets; however, further delithiation (to Li_x CoO₂ with x < 0.5) would be detrimental, resulting in structure collapse.

Other layered TMOs with a lower cobalt content include NMC materials with a general formula of LiNi_x Mn_y Co_z O₂ (where x + y + z = 1), including the high-nickel cathodes that are under industrial consideration for the realization of higher-energy-density batteries of 800 Wh kg⁻¹ at the cell level.

Three other severe problems have been noted with cycling LCO at high voltages [22]: (1) the interface between the delithiated LCO and the electrolyte is unstable. As shown in figure 1.2, the thermodynamic stability window of the electrolyte should surpass the electrode electrochemical potential of both the cathode and the anode. As LCO is cycled to a higher voltage (> 4.45 V), the deterioration of electrolyte is exacerbated, with the formation of a resistive passivation layer and a large impedance rise. Coating or doping with oxides such as Al₂O₃ [23] or ZrO₂ [24] reduces the contact area between LCO and electrolyte and improves the performance under high voltages. (2) The Co dissolution increases with a higher cutoff voltage (> 4.2 V), together with the loss of O (peroxo species of O₂, O₂ ⁻). The delithiation process lowers the Fermi level of Co³⁺ until the Co 3d orbital overlaps with the O 2p band (figure 1.4) and facilitates peroxo evolution [25, 26]. Evolution of these peroxo species causes irreversible surface decomposition and phase transformation of the cathodes, which eventually leads to capacity fade. (3) Degradation of other components such as the binders and the separators occurs, and the consequent detachment of cathode particles can lead to capacity loss.

In the context of battery materials, graphite refers to 'graphitic carbon,' with a structure consisting of a hexagonal network of carbon atoms lacking crystalline ordering in the c direction. A typical crystalline structure of graphite is illustrated in figure 1.7, with alternating A and B layers consisting of sp² hexagonal array of carbon atoms. Upon lithium intercalation, each C₆ can accommodate one Li⁺ atom.

Zoom In Zoom Out Reset image size

The basal planes, referring to the hexagonal C₆ planes, have a spacing (d₀₀₂) of 3.35 Å, about twice that of the Van der Waals distance for carbon atoms. Graphite displays a non-perfectness, and the basal planes demonstrate less reactivity than the edge planes. Upon lithiation, the Li⁺ intercalates through the edge plane and imperfect basal plane, with three distinctive plateaus at 208, 116 and 83 mV (figure 1.8(c)). To circumvent the reactivity of the edge plane, SEI formation enables further intercalation at the expense of consumption of Li⁺ inventory.

Zoom In Zoom Out Reset image size

With the development of Li–S batteries and ASSBs, lithium metal anode research has demonstrated the potential for reaching the high theoretical gravimetric capacity of 3860 mAh g⁻¹. However, the promise of Li metal anodes is accompanied by serious issues, such as the extreme electropositivity of Li metal and its resulting reactivity with almost all organic components, as well as the large interfacial resistance of SEIs consisting of LiF, Li₂CO₃, ROLi, and R'COOLi [1].

The prevalent challenges associated with the application of metal anodes include particle pulverization, unstable SEI, and impedance rise from electrolyte decomposition on the interfaces. The large ionic radii of the relevant metal ions (Na⁺, 1.02 Å; K⁺, 1.38 Å; Mg²⁺, 0.72 Å; Ca²⁺, 1.00 Å [32]) and the charge density can also cause difficulty with intercalation into the cathode lattices.

Among the potential metal anodes for batteries, silicon has been researched extensively in both academia and industry. Co-utilization of graphite and silicon is a common practice to alleviate the issues associated with using Si particles alone, and can increase the specific capacity [27].

The concerns about cost and long-term sustainability of Li resources also demand development of alternative low-cost and sustainable energy storage systems. Novel research, targeting energy based on low-cost and naturally abundant elements, is focusing on using Na and K. The abundance and low price of these elements enable the possibility of sustainably produced, low-cost batteries based on Na⁺ and K⁺ ions [8, 33].

Cathode materials utilizing Na⁺ and K⁺ intercalation that have been explored include layered TMOs, polyanionic compounds [33], and Prussian-blue analogs (A_x M_A[M_B(CN)₆]_y , where A = a combination of Li and Na, and M is a transition metal: Fe, Mn, Co, Ni, Cu, etc). Compared with their Li analog, the large size of Na⁺ and K⁺ can be of benefit for slowing the mobility and disordering of the transition metal. Cathodes with the layered structure of O₃−NaNi_0.5Mn_0.5O₂ can show good cycling stability, with the major issues arising mostly from the delithiated state with low Na/K content [8].

Without Yoshino Akira's introduction of the combination of Gr and carbonate (a unique compound with an −O−C(=O)−O moiety that is prone to react with underpotential Li to generate a SEI and impart extra thermodynamic stability), it would be hard to imagine developing the compositions of today's electrolytes. In particular, ethylene carbonate is responsible for formation of a stable SEI on the surface of a Gr. The success of the Gr anode in LIBs has also revealed another truth: the electrolyte has unique capabilities to modify the interlayer surface.

Despite the enormous progress researchers have made regarding cathodes and anodes, the fraction of LIB research devoted to electrolytes seems relatively small. The discussion of electrolytes below will center on the development of (1) lithium salts; (2) additives; (3) solvents; and (4) special applications involving extreme temperature, fast-charging conditions, and non-flammability.

1.2.3.1.1. Lithium salts

Lithium salts are essentially used for the purpose of transporting Li⁺ under chemically and electrochemically active conditions. The electrolyte environment needs to be compatible with current collectors, cathodes—especially when in a delithiated state, with many energetic active sites to cause decomposition—and anodes in a lithiated state with a potential close to the Li redox potential (~ 100 mV).

The most common lithium salt used in commercial LIBs is LiPF₆, which seems to provide the best combination of ionic conductivity, electrochemical stability, stability with respect to the current collector (a safety criterion), and overall LIB performance. The basic properties of LiPF₆ and several other common lithium salts, such as LiTFSI (Li(CF₃SO₂)₂N), LiAsF₆, LiClO₄, LiBF₄, and LiCF₃SO₃, can be found in the literature [34]; these include such properties as ionic conductivity, ionic radii, temperature, and stability towards both Cu (the anode current collector) and Al (the cathode current collector). Relatively new salts—LiN(SO₂C₂F₅)₂ (LiBETI) [35], lithium bis (oxalate) borate (LiBOB) [36], LiPF₃(C₂F₅)₃ (LiFAP) [37], lithium 4,5-dicyano-2-(trifluoromethyl) imidazolide (LiTDI) [38], lithium 4,5-dicyano-2-(pentafluoroethyl) imidazolide (LiPDI), and lithium bis(fluorosulfonyl)imide (LiFSI)—are also available through exploratory design and syntheses, although their compatibility with coin cells and their electrochemical stability are still a matter of debate.

Below, the chemistry of LiPF₆ will be discussed in detail with respect to its reactivity towards H₂O [39] (an unavoidable impurity), solvent [40], and additives such as tris(trimethylsilyl) phosphite [41].

The cascading hydrolysis of LiPF₆ is illustrated in scheme 1.1. Overall, the reactivity of LiPF₆ with water is detrimental, leading to formation of both HF and oxyfluorophosphoric acid, which corrodes the cathodes and causes crossover of transition metal species (Mn²⁺); these species cause the poisoning of the anode, leading to an electrocatalytic reduction of Li⁺ inventory and capacity loss. Recent studies of hydrolysis of LiPF₆ under high voltage suggest that the detrimental catalytic effects of high voltage are much worse, with the majority of the hydrolysis products coming from the reaction between PF₆ ⁻ anion and H₂O under voltages above 4.5 V.

Zoom In Zoom Out Reset image size

1.2.3.1.2. Additives

Additives are designed for task-specific applications such as anode SEI formation, cathode CEI formation, flame retardation, and other unique applications including fast-charge and extreme-temperature applications. The Dahn group, as well as Argonne National Laboratory researchers, have long been developing both singular and combinatorial additives [43, 44].

The reactivity of LiPF₆ with tris(trimethylsilyl) phosphite (TMSPi) was discovered by the Winter group [41]. As shown in scheme 1.2, the reaction involves an astounding F–O exchange between the P−F bond and the P−O−TMS bond, resulting in formation of three compounds: P(OTMS)₂F, P(OTMS)F₂ and PF₃. The intermediate compound O=P(−F)₂−O⁻ can also react with P(OTMS)₃ to generate P(OTMS)₂F and O=P(−F)₂−OTMS; note the presence of an extra O=P in the latter compound. Overall, a spontaneous cascading reaction is observed with aging, with little control of either the speciation or concentration (figure 1.9). In general, it is predicted that P(OTMS)₃ will show a consistent decrease with time, while both the P(OTMS)₂F and P(OTMS)F₂ intermediates will show an initial increase before reaching a maximum concentration, followed by a decrease. The final product PF₃ appears last in the reaction and is expected to show a steady increase until the original P(OTMS)₃ is totally consumed.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The control of the spontaneity of the LiPF₆ reactivity, however, has been showcased in our group recently by a unique in situ method for formation of cathode-protective additives with structures identified as lithium oxyfluorophosphates (2a and 2b), as illustrated in scheme 1.3. These in situ reactions take advantage of the reactivity of LiPF₆ with the bis(trimethylsilyl) esters 1a and 1b, and a series of extremely moisture-sensitive tetrafluorophosphates can be easily obtained at a nearly 100% yield in the solution (scheme 1.4) [45]. In contrast to the reaction in scheme 1.2, there is only one designated product from the reaction between LiPF₆ and bis(trimethylsilyl)ester. As discussed by Yang et al, the strain of the carboxylate linker in the tetrafluorophosphate is important: strong preference goes to the 5-membered and 6-membered rings because of the lesser ring strain and torsional strain. A similar protective effect on the cathode surface was also discovered: a 6-membered ring was found to have a superior effect on the reduction of impedance rise in an NMC//Gr full cell [45].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

LIBs for industrial applications require a combination of excellent thermal stability, wetting capability, and SEI formation capability, resulting in the presence of many additives and co-solvents of propionate compounds [46]. The additives in commercial LIBs are targeted for different applications and can be loosely classified as anode additives, cathode (interfacial) additives, overcharge protection, and alloy additives [43, 47]. Some common additives are shown in figure 1.10.

Zoom In Zoom Out Reset image size

Despite the large number of additives that are already available, new additives and uses for those additives are continuing to be invented [48].

1.2.3.1.3. Solvents

1.2.3.1.4. Special applications: extreme-temperature, fast-charging, and flame-retardant electrolytes

1.2.3.2.1. Lithium metal batteries

1.2.3.2.2. Multivalent ion batteries