Tailoring Lithium Polysulfide Coordination and Clustering Behavior through Cationic Electrostatic Competition

Abstract

The materials chemistry underlying lithium-sulfur (Li-S) batteries is uniquely dependent on the behavior of soluble lithium polysulfide intermediates, which form during operation and mediate the charge-transfer process in solution. The manner by which lithium polysulfides are solvated by surrounding solvent and salt compounds is a critical factor with regards to electrochemical utilization and reversibility of the sulfur active material. Particularly at low-temperature and lean electrolyte conditions, lithium polysulfides tend to coordinate with other polysulfide units in solution, forming large, aggregated clusters that stymie the electrochemical conversion process. However, the tendency to cluster is known to be influenced by the presence of strongly binding anionic species in solution, which present electrostatic competing interactions with Li⁺. The heightened electrostatic competition in turn can dissuade the formation of clustered Li⁺−S_x²⁻ bond networks. Here, we extend that understanding to the influence of distinct cationic species in solution, which can present analogous competing interactions with S_x²⁻ dianions to stymie polysulfide cluster formation. We find that introducing NH₄⁺ cations into solution through an ammonium trifluoroacetate additive positively tailors the polysulfide coordination shell. This improves the electrochemical conversion kinetics at challenging lean electrolyte and subzero low-temperature conditions, and provides a more holistic understanding of polysulfide coordination behavior.

Graphical Abstract

Introduction

The widespread adoption of next-generation energy-dense batteries is poised to enable tremendous advancements in electrified mobility and transportation, particularly for weight-dependent applications within the aviation and space sectors.^1–3 This is especially true for lithium-sulfur (Li-S) batteries, which possess an order of magnitude greater theoretical gravimetric capacity (1,672 mA h g⁻¹) and specific energy (~ 2,600 W h kg⁻¹) than the incumbent lithium-ion batteries.⁴ While this theoretical capability is highly promising, in practice it is quite difficult to achieve even 10% of this theoretical specific energy in state-of-the-art Li-S cells.⁵ The single largest factor leading to this shortcoming ties to the inability of the redox couple to deliver high active material utilization under lean electrolyte conditions.⁶ This property is a manifestation of the inherent solution-mediated nature intrinsic to the lithiation of sulfur active material.

Both elemental sulfur and the Li₂S discharge product are severely ionically and electronically insulating, presenting unfavorable barriers to the surface-based charge-transfer pathways traditionally required for reduction and oxidation of battery materials. The Li-S chemistry overcomes this through the generation of soluble lithium polysulfide (Li₂S_x, 2 < x ≤ 8) intermediates, which can enable facile charge-transfer in solution via solution-mediated reduction pathways.^7–9 Thus, there is a minimum degree of active material dissolution necessitated by the Li-S chemistry for the cell to function and make use of solution-mediated reaction avenues. For this reason, there is often a divergence in the behavior of Li-S cells as a function of the volume of electrolyte contained within the cell. Most notably, high capacities become increasingly difficult to achieve in cells constructed under lean electrolyte conditions. This is quantified by the electrolyte-to-sulfur (E/S) ratio (μL_electrolyte mg⁻¹_sulfur), a commonly reported metric accompanying electrochemical data within the Li-S literature.⁶

The solution-mediated paradigm intrinsic to the Li-S chemistry imparts a variety of unique and chemically complex phenomena during discharge and charge. This is readily demonstrated by the recent surge in research attention directed at the nature of solvation and coordination of individual lithium polysulfide units.^9–12 Under kinetically constrained conditions, such as lean electrolyte volumes or even subzero low-temperature conditions, lithium polysulfides tend to cluster and coordinate with each other disproportionately more than with the surrounding solvent molecules, forming clustered polysulfide aggregates.^10,13 This behavior can proceed to inhibit the favorable solution-mediated kinetics underlying the Li-S chemistry, curtailing lithiation and diffusion of lithium polysulfides. In fact, this kinetic constraint is found to be the dominant limiting mechanism at low temperatures rather than just physical changes to the electrolyte’s ionic conductivity or viscosity like that seen in low-temperature lithium-ion batteries.¹³ Moreover, this behavior disproportionately constrains the electrochemical utilization of sulfur under conditions most relevant to the challenging uses-cases Li-S batteries are best suited for.¹⁴

However, just as the Li-S system possesses a large degree of intrinsic chemical intricacy, it also presents tremendous opportunities for targeted improvements through relatively simple additions to the liquid electrolyte. For example, in our past work we revealed that detrimental clustering behavior can be altered and influenced through the introduction of competing electrostatic interactions in solution.¹³ It was shown that the introduction of lithium salts with highly binding anionic groups can disrupt polysulfide cluster formation. The electrostatic attraction between the highly negative anion and the polysulfide-adjoined lithium cation dissuades the coordination of adjacent polysulfides units, leading to much higher discharge capacity and more favorable kinetic behavior at low-temperature conditions. This result is quite consistent with other reports exemplifying the advantages of strongly binding salt anions for improvements to room-temperature Li-S electrochemical behavior.^15–18

Within Li-S batteries, the underlying materials chemistry is intimately tied to the manner by which distinct ions interact with active material in solution. This considerably differs from traditionally employed systems with insertion-ion electrodes, as the active material in the Li-S battery is in many respects inextricably linked to the electrolyte solvent and salt. In this work, we seek to advance the understanding of how the coordination of dissolved polysulfide active material can be further influenced, beyond just the implementation of highly binding anionic groups. We specifically look at the addition of a strongly binding cationic group, NH₄⁺, into the Li-S electrolyte to further influence and alter the coordination behavior of lithium polysulfides.

Experimental Methods

Polysulfide Solution Preparation:

Inside an argon-filled glovebox, lithium polysulfide samples were prepared in the solvent of interest by mechanically stirring Li₂S and elemental sulfur powders in an amount stoichiometrically equivalent to the concentration of Li₂S₆ desired. These samples were generally stirred over the course of 12 – 24 h at 50 °C. For samples also containing NH₄TFA additive, an amount of NH₄TFA equivalent to the concentration desired was added and stirred (for ~ 30 min) after the lithium polysulfide solution was fully prepared.

Fourier Transform Infrared (FTIR) Spectroscopic Studies:

FTIR spectroscopy studies were conducted in the attenuated totaled reflectance (ATR) configuration with a Thermo Scientific-Nicolet iS5 FTIR spectrometer. Liquid samples were prepared in an argon-filled glovebox. Fresh background scans were performed before running each sample. For each sample, 50 μL was drawn and placed onto the ATR crystal, after which scans were immediately performed. 32 scans were recorded in the 400 to 4000 cm⁻¹ range with a scan resolution of 0.482 cm^-1.

Electrolyte preparation:

Electrolytes were prepared inside an argon-filled glovebox. For electrochemical studies at room-temperature, two different electrolytes were prepared: 1.0 M LiTFSI and 0.2 M LiNO₃ in DOL/DME (1:1, by vol.); and 0.8 M LiTFSI, 0.2 M LiNO₃, and 0.2 M NH₄TFA in DOL/DME. For variable temperature electrochemical studies, three different electrolytes were prepared: 0.8M LiTFSI and 0.2 M LiNO₃ in DOL/DME (85:15, by vol.); 0.5 M LiTFSI, 0.3 M LiTFA, and 0.2 M LiNO₃, in DOL/DME (85:15, by vol.); and 0.5 M LiTFSI, 0.3 M NH₄TFA, and 0.2 M LiNO₃, in DOL/DME (85:15, by vol.).

Li-S Coin Cell Fabrication:

Li-S coin cells assembled in CR-2032 type coin cells consisted of a sulfur cathode, a lithium-metal foil anode, electrolyte, and a Celgard 2500 separator. For room-temperature cycling, the sulfur cathodes possessed an active material loading of 3.4 mg cm⁻², while the cells contained an E/S = 12 μL mg⁻¹. For variable temperature cycling, the sulfur cathodes contained an active material loading of 3.1 mg cm⁻², while the cells had an E/S = 13 μL mg⁻¹. The slurry-cast sulfur cathode consisted of 63% sulfur, 28% carbon, and 9% PEO-PVP binder (4:1, by mass, dissolved in H₂O), cast onto a carbon-coated aluminum foil.

Li-S Pouch Cell Fabrication:

Soft-packaging Li-S pouch cells were assembled inside an argon-filled glovebox. Slurry-cast sulfur cathodes (63% sulfur, 28% carbon, and 9% PEO-PVP binder) with an area of ~ 24 cm² were used. The cell with the NH₄TFA-containing electrolyte had an areal sulfur loading of 4.8 mg cm⁻², while the cell with control electrolyte contained a sulfur loading of 3.5 mg cm^-2. Slight discrepancies in the loading for each cell stemmed from each large-format cathode requiring a separate casting step (as opposed to smaller coin-cell cathodes). Nonetheless, differences in the performance are attributed primarily to electrolyte properties rather than the slight differences in loading given the sizeable variations in electrochemical performance. Li-foil was attached to a Ni-foam current collector to form the anode. Each tab for the cathode and anode was, respectively, welded on to the Al and Ni-foam backings. The pouch cells were injected with electrolyte (E/S = 4.5 μL mg⁻¹) before being sealed.

Electrochemical Cell Testing:

All electrochemical studies were conducted with an Arbin battery cycler. For the galvanostatic coin-cell cycling at room-temperature, a rate of C/10 was employed after 2 formation cycles at C/20 rate, with 1C = 1,672 mA g⁻¹ of sulfur in the cathode. A voltage window between 1.7 and 2.7 V was used for cycling coin cells. Pouch cells were assessed over a single discharge and charge at a C/20 rate, using a voltage window of 1.65 − 2.45 V. Galvanostatic cycling at variable temperature conditions was also employed to assess the variations in the discharge behavior at low-temperature conditions. Cells were stored at each temperature of interest for at least 3 h, after which they were cycled at a C/20 rate. Voltage ranges of 1.7 – 2.7 V, 1.65 – 2.75 V, and 1.5 – 2.8 V were used, respectively, for cycling at temperatures of 25 °C, 0 °C, and −20 °C. Temperature was modulated with an Espec SH-241 bench-top environmental chamber.

⁷Li-Nuclear Magnetic Resonance (NMR) Spectroscopic Studies:

NMR samples consisted of 0.2 M nominal Li₂S₄ with and without 0.2 M NH₄TFA in DME/d-THF (2:3, by vol.) solvent. In an argon-filled glovebox, 690 μL of each solution was deposited in an NMR tube, along with a reference coaxial insert tube. This insert contained 0.2 M LiBr in DME/d-THF (7:3, by vol.), serving as an external reference without affecting or influencing the coordination environment of the sample. The sample was sealed inside the glovebox before being brought out for analysis. Each ⁷Li-NMR scan was performed with a Varian VNMRS 600 MHz spectrometer at both 25 and −50° C. Temperature was modulated to −50° C with a flow line containing liquid nitrogen. 16 scans were run at each temperature, with a relaxation delay of 60 s, a pulse width of 14.5 s, and an acquisition time of 2 s. In the case of overlapping peaks, spectra were deconvoluted to process the raw data and determine peak locations.

Computational Methods:

First-principles calculations were performed with Gaussian 16 Rev. A.03.¹⁹ The hybrid DFT method B3LYP as well as the 6–311++G(d,p) basis set were used for optimizing structures and performing vibrational frequency analyses. Molecular structure files used for inputs in calculations were prepared with Avogadro Version 1.2.0.²⁰ A polarizable continuum model (PCM) was used to simulate the influence of surrounding solvent; THF was used as the chosen solvent given that its properties are consistent with the glyme-based solvents generally used in Li-S batteries. Energy changes and ionic association energies were calculated using the sum of electronic and zero-point energies (outputs from a vibrational frequency analysis) of each optimized structure. Vibrational frequency analysis served as a validity-check to ensure each final output structure exhibited no negative frequencies.

Results and Discussion

Probing Coordination through Cationic Modification:

During the discharge and charge phases of a Li-S cell, the lithium polysulfide coordination environment within the liquid electrolyte plays a critical role in determining both the physical and electrochemical behavior of the active material, including transport and diffusion properties,^10,21 speciation,^7,22,23 and nucleation and growth kinetics.^11,15,24 Within the Li-S electrolyte, there are a wide array of chemical compounds and ions present in solution. This includes the dioxolane (DOL) and dimethoxyethane (DME) solvents, the Li⁺ cation and polysulfide dianions making up lithium polysulfides, as well as the Li⁺, TFSI⁻, and NO₃⁻ ions stemming from the lithium bistriflimide (LiTFSI) and LiNO₃ salts typically in solution. This in turn imparts sizeable electrostatic competition and interactions between the various species in solution. Lithium polysulfide species in particular undergo significant short-range electrostatic coordination not just with other ions, but also with neighboring polysulfide units, as shown through the illustration in Figure 1a. This clustering behavior is found to be a relatively lower energy state, and as discussed, can occur disproportionately more frequently under lean electrolyte or low-temperature conditions.^12–14

Figure 1.

(a) Lithium polysulfides can coordinate and aggregate together in solution, forming a relatively lower energy state Li⁺−S_x²⁻ strongly bound network. (b) The clustered state can be disrupted by other anions present in solution, like trifluoroacetate (TFA⁻), which present competing electrostatic interactions to form Li⁺−TFA⁻ bonds in favor of Li⁺−S_x²⁻ bonds. (c) The electrostatic coordination environment can potentially be further augmented by simultaneously introducing competing cationic interactions, disrupting the Li⁺−S_x²⁻ network to a much greater extent.

The selective coordination of lithium polysulfides to adjacent polysulfides in solution has been shown to be influenced and altered through the addition of strongly binding (high donor number) salt compounds like lithium trifluoroacetate (LiCO₂CF₃, abbreviated hereafter as LiTFA) to the electrolyte.¹³ The highly Lewis basic trifluoroacetate (TFA⁻) anion presents an effective electrostatic competing force to disrupt the strongly bound Li⁺−S_x²⁻ networks in favor of Li⁺−TFA⁻ interactions, as illustrated in Figure 1b. The mitigation of polysulfide clustering in turn has been shown to benefit kinetic behavior and capacity attainment under challenging lean electrolyte or low-temperature conditions.^13,14

The anionic modification to the lithium polysulfide coordination environment presents an excellent platform to develop upon for further scientific insight and engineering improvement. However, rather than utilizing strongly binding anions like TFA⁻ to help coordinate with polysulfide-adjoined Li⁺ cations, it is intriguing to consider whether the opposite approach of utilizing a strongly binding cation could potentially demonstrate similar behavior, as shown in Figure 1c. In this manner, rather than solely utilizing compounds possessing a high electron donor capability,^25,26 this approach assesses the efficacy of electron acceptors towards improving polysulfide coordination behavior. If true, then employing a strongly binding cation and anion simultaneously in solution could feasibly curtail the formation of clustered polysulfide aggregates to a much greater extent than shown before. This would serve to boost solution-mediated kinetics in Li-S batteries to new heights under the challenging conditions most applicable for practical use of Li-S batteries.¹⁴

The ammonium cation, NH₄⁺, was selected as a starting point to assess this strategy based off its favorable use for lean-electrolyte Li-S cells in the literature.²⁷ Building off an in-situ derivatizing additive we introduced in a recent work,²⁸ ammonium trifluoroacetate (NH₄TFA) was identified as an effective electrolyte additive for evaluating the joint effects of NH₄⁺ and TFA⁻ on the solution-coordination behavior of lithium polysulfides. This compound is expected to react in-situ with the lithium polysulfides that inevitably leech into solution during the course of discharge, resulting in the formation of diammonium polysulfides and LiTFA salt according to Figure 2a. Computational chemistry calculations were conducted to thermodynamically assess the feasibility of the forward reaction taking place. The molecular structure of each reactant and product in Figure 2a was optimized using gas-phase hybrid density functional theory calculations, after which the total change in free energy for the reaction was found according to Equation 1. The broad range of lithium polysulfides that appear in solution were nominally represented by Li₂S₆.

Figure 2.

(a) Reaction pathway between NH₄TFA and lithium polysulfides theorized to take place in-situ, with calculated overall change in formation energy. (b) Vials containing 0.1 M Li₂S₆ in DOL/DME solvents, with and without the addition of 0.2 M NH₃TFA. (c) FTIR spectra of a sample containing both Li₂S₆ and NH₄TFA dissolved in THF solvent, as well as samples containing NH₄TFA, Li₂S₆, and LiTFA alone in solution.

$$\text{Δ}G=\left[\left(2G_{LiTFA}+G_{{\left(N{H}_4\right)}_2{S}_6}\right)-\left(2G_{N{H}_{4}TFA}+G_{L{i}_2{S}_6}\right)\right]$$ (1)

The forward reaction is found to take place with a minimization of energy on the order of 19 kJ mol⁻¹, suggesting this additive would readily react with lithium polysulfides in solution. This reaction is also experimentally confirmed to take place. As shown in Figure 2b, lithium polysulfides are visually confirmed to react spontaneously with NH₄TFA added into solution. The black and opaque hue of lithium polysulfides readily transforms into a deep red and translucent color upon contact with NH₄TFA. This is further confirmed through Fourier-transform infrared (FTIR) spectroscopy on liquid samples containing both Li₂S₆ and NH₄TFA dissolved in tetrahydrofuran (THF) solvent, as well as samples containing only NH₄TFA, Li₂S₆, or LiTFA alone in solution. As seen in Figure 2c, the spectra from the combined sample closely matches what is seen for just LiTFA salt dissolved in solution, rather than either the NH₄TFA sample or Li₂S₆ sample. The formation of LiTFA in this sample strongly confirms that NH₄TFA reacts with dissolved lithium polysulfides in a manner consistent with the reaction pathway outlined in Figure 2a. This is further expanded on in Table S1 and the accompanying discussion in the Supporting Information. Thus, NH₄TFA is situated well as a tool for modulating and improving the polysulfide coordination environment.

Electrochemical Variance with Electrolyte Amount:

Equipped with a firm understanding of the role NH₄TFA chemically plays in solution, the resultant polysulfide coordination environment was assessed through electrochemical analysis. Li-S coin and pouch cells were used as platforms to galvanostatically assess the impacts of the additive on electrochemical behavior as a function of electrolyte amount. Specifically, cells were constructed with an electrolyte consisting of 0.2 M NH4TFA, 0.2 M LiNO3, 0.8 M LiTFSI, with the intent of keeping the total solute concentration equal to a control electrolyte consisting of 0.2 M LiNO₃ and 1 M LiTFSI in DOL/DME (1:1, by vol.). Additionally, the electrolyte volumes utilized in coin cells and pouch cells were deliberately varied to understand the behavior of the additive as a function of E/S ratio. Coin cells were constructed with high E/S ratios of 12 μL mg⁻¹ while pouch cells were constructed with low E/S ratios of 4.5 μL mg⁻¹, a lean amount where standard Li-S cells can tend to perform poorly.^6,29 As discussed earlier, the E/S ratio can to a first order approximate the expected degree of active material solvation and utilization. Thus, through these cycling experiments, the specific contributions NH₄TFA yields can be holistically understood as a function of the polysulfide concentration and coordination environment within the electrolyte.

As seen in Figure 3a, the galvanostatic cycling behavior of coin cells (E/S = 12 μL mg⁻¹) with and without NH₄TFA is quite similar, generally maintaining between 800 and 900 mA h g⁻¹ over the course of 100 cycles. Despite the similar capacity retentions, it is seen in Figure 3b that the voltage profiles of each cell exhibit slight differences. Most notably, the cell containing NH₄TFA does exhibit slightly superior kinetic behavior, with a lower overpotential on the order of 40 mV seen during discharge. This is particularly noticeable in the lower voltage plateau conversion from Li₂S₄ to Li₂S. However, a much more substantial difference is seen with both electrolytes when tested under lean electrolyte conditions in large-format pouch cells (E/S = 4.5 μL mg⁻¹), as shown in Figure 3c. Under such conditions, the cell containing NH₄TFA demonstrates substantial improvements over the cell with control electrolyte, which can be seen over the course of a single cycle in Figure 3d. From a capacity standpoint, the NH₄TFA additive enables a capacity of 930 mA h g⁻¹ compared to 250 mA h g⁻¹ for the control cell, a sizeable 370% improvement. This is also clearly reflected in the drastic improvements in the upper and lower voltage plateaus to both capacity and polarization, signifying the more complete utilization of the sulfur active material in the presence of NH₄TFA. While this reflects the hypothesized boosts to solution-mediated kinetics envisioned from the selection of the additive, it is especially compelling that the major improvements seen are primarily under lean electrolyte conditions.

Figure 3.

(a) Cycling performance of coin cells containing 0.8 M LiTFSI, 0.2 M NH₄TFA, and 0.2 M LiNO₃ compared to a control cell containing 1 M LiTFSI and 0.2 M LiNO₃ in DOL/DME. Aberrations in capacity starting at cycle 60 are due to unanticipated changes in the ambient temperature of the room where cycling was performed. (b) Voltage profiles of the NH₄TFA-containing cell and the control coin cell on cycle 50. (c) Image of the pouch cell format that was used to assess cells under lean-electrolyte conditions. (d) Voltage profiles of the NH₄TFA-containing cell and the control cell over a single cycle.

The divergence in performance at low E/S ratios in cells containing NH₄TFA points to the effect the additive may be having on enhancing the polysulfide coordination environment. When presented in a non-stoichiometric manner, the reaction shown in Figure 2a would only be expected to proceed until one of the reactants is consumed. This is demonstrated in Figure S1, where a lithium polysulfide reactant is fully consumed in contact with excess NH₄TFA. Given the high polysulfide concentrations expected to be seen under lean E/S conditions,⁷ the relatively small amount of 0.2 M NH₄TFA would almost certainly serve as the limiting reactant in the lean electrolyte pouch cells from Figure 3c and d. As the reaction would not proceed to completion, the makeup of polysulfides in solution would likely represent a combination of lithium, diammonium, and mixed lithium-ammonium polysulfides, as visualized in Figure 4a. The net result of the competitive ion interactions between S_x²⁻ and either Li⁺ or NH₄⁺ plays an instrumental role in determining the subsequent coordination behavior of the polysulfide molecule. For any given S_x²⁻ dianion, the resultant minimization of energy achieved from ionically associating with a Li⁺ and NH₄⁺ cation can be calculated according to Equation 2.

Figure 4.

(a) The molecular geometries of lithium, diammonium, and mixed lithium-ammonium six-member polysulfides, optimized with computational chemistry. (b) The ionic association energies of lithium diammonium, and lithium-ammonium polysulfides with sulfur chain length equal to 4, 6, and 8. Energies are normalized to a kJ mol⁻¹ S_x²⁻ basis.

$$\text{Δ}{G}_{ionicassociation}=G_{L{i}_n{\left(N{H}_4\right)}_y{S}_x}-\left(n{G}_{L{i}^{+}}+y{G}_{N{H}_4}++G_{S_x{}^{2-}}\right)$$ (2)

Utilizing computational chemistry, the ionic association energy was calculated for lithium, diammonium, and mixed lithium-ammonium polysulfides of length x = 4, 6, and 8 sulfur atoms. For each geometry, the value was normalized to a kJ mol⁻¹ S_x²⁻ basis, where x is the polysulfide chain length of the molecule. As shown in Figure 4b, a far greater driving energy is found for polysulfide dianions to ionically associate with NH₄⁺ than Li⁺, with both diammonium and mixed lithium-ammonium polysulfides demonstrating minimizations of energy ranging from 10 to 40 kJ mol⁻¹ S_x²⁻ greater than pure lithium polysulfides. This is in many respects consistent with the minimization of energy shown through the forward reaction in Figure 2a. When considering the wide-ranging variety of ions and species in solution in the Li-S electrolyte, there is a strong theoretical basis for NH₄⁺ to provide a stronger electrostatic competing force for polysulfides than adjoining Li⁺ cations.

Clustering Tendency with Heightened Electrostatic Competition

The divergence in ionic association energies between NH₄⁺ and Li⁺ directly extends to the clustering and aggregation of lithium polysulfides. The basis for clustering behavior stems from a minimization of energy via adjacent polysulfide units coordinating through short range electrostatic interactions.^30,31 This tends to occur most frequently in short-chain Li₂S₄, which is the dominant species present in solution at the onset of the second plateau conversion to Li₂S.^12,13 This is modelled quantitatively through Equation 3.

$$\text{Δ}G=\frac{1}{n}\times \left[G_{{\left(L{i}_2{S}_4\right)}_n}-n\left(2G_{L{i}^{+}}+G_{S_4{}^{2^{-}}}\right)\right]$$ (3)

As shown from first-principles calculations in Figure 5a, free energy is minimized to a greater extent with increasing sized clusters of Li₂S₄ units coordinating to one another. For instance, a 4-member Li₂S₄ cluster exhibits a free energy 37 kJ mol⁻¹ S₄^2- less than a single Li₂S₄ unit. This significant decrease in energy shifts equilibrium to highly clustered states in the kinetically limiting environments that, for example, low-temperature conditions present.¹⁴ However, as shown through Figure 5b, the greater propensity for diammonium polysulfides to exhibit greater ionic association ability also extends to their ability to cluster. Varying sized clusters of ((NH₄)₂S₄)_n of size n = 1, 2, and 4 display the same trends seen in Li₂S₄, with energy being minimized to the greatest extent in large 4-member clusters. Notably, the energy minimization calculated far exceeds that seen in lithium polysulfide clusters, with ((NH₄)₂S₄)₄ exhibiting a free energy 51 kJ mol⁻¹ S₄^2- less than a single (NH₄)₂S₄ unit. Thus, a liquid solution containing solely diammonium polysulfide species would be expected to exhibit clustering behavior at an even greater frequency than that with lithium polysulfides.

Figure 5.

(a) The variation in free energy normalized to a kJ mol⁻¹ S₄²⁻ basis with increasingly sized molecular clusters of Li₂S₄. (b) The variation in free energy with increasingly sized molecular clusters of (NH₄)₂S₄.

However, as noted under lean electrolyte conditions, the liquid electrolyte environment with NH₄TFA additive likely contains an amalgam of lithium, diammonium, and mixed lithium-ammonium polysulfides. Thus, the drive to minimize energy in the form of a lithium polysulfide cluster may run counter to the competing drive to minimize energy in the form of a diammonium polysulfide cluster. As a result, the competitive and opposed drives to ionically associate may incur a stalemate-like situation, altogether inhibiting the formation of clustered aggregates and preserving a favorable coordination environment for dissolved active material. Another possibility is that the formation of ammonium polysulfides clusters proceeds in the absence of Li⁺ charge carriers, minimizing any detrimental impacts on the vital charge-transfer kinetics that Li-S discharge depends upon. Indeed, the electrochemical behavior seen under the lean-electrolyte conditions in Figure 3d is highly consistent with this proposed mechanism.

In order to supplement the theoretical basis for this hypothesis, ⁷Li-Nuclear Magnetic Resonance (NMR) spectroscopy was employed to further probe the mechanism. NMR can provide detailed information regarding the bond environment surrounding ⁷Li nuclei within each sample and is an excellent tool to survey the dynamic coordination environment within Li-S electrolytes. Two samples were prepared containing 0.2 M Li₂S₄ with and without 0.2 M NH₄TFA dissolved in deuterated tetrahydrofuran (d-THF) and DME solvent. Spectra were obtained at both 25 °C and −50 °C to observe the change in bond environment occurring with the induced formation of polysulfide clusters. As seen in Figure 6a, the peak affiliated with Li₂S₄ at 25 °C noticeably shifts upfield from 0.06 ppm to −0.33 ppm at −50 °C, highly consistent with the increased diamagnetic shielding expected from the heighted surrounding electron density present in large clustered states.¹³ Additionally, this peak drastically reduces in intensity, consistent with the precipitate-like solvate formation of Li₂S₄ clusters. However, when lithium polysulfides are simultaneously accompanied by NH₄TFA in solution, there are a variety of key differences in the ⁷Li-NMR response. The presence of NH₄TFA induces the formation of diammonium polysulfides and LiTFA salt, which serves to shift the room temperature peak in Figure 6b downfield to 0.53 ppm. Even more so, there is distinct change in ⁷Li-NMR response as temperature is lowered to −50 °C compared to what is seen with just Li₂S₄. The peak still shifts upfield, but interestingly, displays a multimodal range of peaks at 0.07, 0.03, and 0.02 ppm overlapping with the LiBr reference.

Figure 6.

(a) The ⁷Li-NMR spectra of 0.2 M Li₂S₄ in d-THF/DME at 25°C and −50° C, with 0.2 M LiBr in d-THF/DME as an external reference. (b) The ⁷Li-NMR spectra of 0.2 M Li₂S₄ and 0.2 M NH₄TFA in d-THF/DME.

The sizeable changes in coordination environments when NH₄TFA is present suggest that the nature of coordination is fundamentally altered. At 25 °C, for example, the large shift downfield implies that the average Li⁺ nuclei in solution is significantly diamagnetically de-shielded; there is a “withdrawing” of the electron density surrounding Li⁺ cations, increasing the effective field experienced by the average ⁷Li nucleus. This is consistent with the heightened electrostatic competition postulated through the addition of NH₄TFA, where Li⁺ becomes less coordinated with surrounding S_x²⁻ ligands. However at −50 °C, the sizeable shift downfield suggests that Li⁺ is still experiencing a strong shift in coordination environment with temperature, and given that this shift is not likely to stem from a change in coordination with the TFA− anion or surrounding solvent,¹³ the only option is that Li⁺ is still partaking in the formation of aggregated polysulfide clusters. The distinct peaks, however, suggest that rather than pure lithium polysulfide aggregates, there are an array of distinct and compositionally variant mixed lithium-ammonium polysulfide aggregates that form at low-temperatures.

This can be better understood using the lens of Pearson’s hard and soft acids and bases (HSAB) theory.^9,26,32 In an ideal solvated lithium polysulfide unit, the hard Li⁺ Lewis acid prefers to coordinate with the relatively hard S_x²⁻ dianion more strongly than with the softer DOL/DME solvent, which accounts for the widespread polysulfide clustering that occurs at low temperatures or lean electrolyte conditions. With the increased coordination to a greater number of S_x²⁻ ligands, each Li⁺ cation is shielded in a softer solvated complex by the increased surrounding electron density, leading to the diamagnetic shielding effect. The substitution of Li⁺ with a slightly softer NH₄⁺ Lewis acid would promote more effective dipole-cation coordination with the relatively soft DOL/DME solvent compared to the harder S_x²⁻ dianion. This would serve to counteract the detrimental Li⁺−S_x²⁻ coordination interactions making up the backbone of large polysulfide clusters, as reflected by the downfield translation in the 25 °C ⁷Li-NMR chemical shift. Even at −50 °C, the clustered aggregates that do form in the presence of NH₄TFA still exhibit chemical shifts downfield of that seen in pure lithium polysulfide clusters, suggesting that the average Li⁺ coordination state is still more favorable. This validates the tendency of NH₄⁺ and TFA⁻ to preferentially coordinate with, respectively, polysulfide dianions and Li⁺ in solution. This heightened electrostatic competition preserves a more favorable and dissociated coordination state of Li⁺, providing a basis by which to boost solution-mediated electrochemical kinetics.

Electrochemical Behavior at Low Temperatures

The strong Li⁺-S_x²⁻ bond network that accompanies the onset of clustered polysulfides forms the backbone for kinetically impeding discharge and charge kinetics at low temperatures or low E/S conditions. Thus, the joint effects of NH₄⁺ and TFA⁻ on the coordination environment of lithium polysulfides can be most accurately ascertained through the electrochemical assessment of Li-S cells at low-temperature conditions. In this manner, the application of a low-temperature environment is used a toggle to intensify and alter the polysulfide coordination shell into a known unfavorable state. This time, the optimized low-temperature electrolyte contained 0.5 M LiTFSI, 0.3 M NH₄TFA, and 0.2 M LiNO₃ in DOL:DME (85:15, by vol.). This was compared to a control electrolyte containing 0.8 M LiTFSI and 0.2 M LiNO₃ in DOL:DME (85:15, by vol.). In addition, a halfway optimized electrolyte containing 0.5 M LiTFSI, 0.3 M LiTFA, and 0.2 M LiNO₃ in DOL:DME (85:15, by vol.) was also tested to deconvolute and separately understand the positive contributions stemming from the high donor LiTFA salt and NH₄⁺-modified polysulfides. The total solute concentrations and solvent ratios in all formulations were chosen to match empirically optimized low-temperature Li-S electrolytes from the literature.^13,33

As seen in the voltage profile in Figure 7a, the NH₄TFA-optimized electrolyte formulation slightly outperforms the other formulations at 25 °C, achieving a capacity that is 13% higher and 9% higher than, respectively, the control and LiTFA-containing electrolytes. This trend becomes slightly stronger at 0 °C (shown in Figure 7b), with the NH₄TFA-optimized electrolyte achieving a capacity that is 26% higher and 15% higher electrolyte than, respectively, the control and LiTFA-containing electrolytes. This data support the hypothesis that NH₄⁺ tends to associate more favorably with solvated polysulfide units than contiguous Li⁺ cations, disrupting the strong Li⁺−S_x²⁻ clustered frameworks that impede discharge behavior at kinetically limiting conditions. At −20 °C, however, the electrochemical behavior enabled by each electrolyte formulation exhibits highly stark differences, beyond just the continued trends in capacity attainment. As shown in Figure 7c, the control electrolyte fails to exhibit a lower voltage plateau at all, indicating highly curtailed conversion to Li₂S as expected. The LiTFA-containing electrolyte, meanwhile, displays a sizeable lower voltage plateau, albeit taking place at a lower-than-expected potential of 1.7 V vs Li/Li⁺. However, while these voltage profiles still somewhat resemble the ideal Li-S voltage profiles seen at 25 °C, the same cannot be said for the NH₄TFA-containing electrolyte at −20 °C.

Figure 7.

The variations in discharge and charge behaviors in cells containing 0.8 M LiTFSI and 0.2 M LiNO₃; 0.5 M LiTFSI, 0.3 M LiTFA, and 0.2 M LiNO₃; or 0.5 M LiTFSI, 0.3 M NH₄TFA, and 0.2 M LiNO₃, shown at (a) 25 °C, (b) 0 °C, and (c) −20 °C.

Rather than the typical two-plateau behavior predominantly seen in Li-S electrochemistry, the discharge taking place in the NH₄TFA-containing electrolyte exhibits an unexpected and fascinating 4-plateau voltage profile at −20 °C. Moreover, the first two plateaus and last two plateaus exhibit a combined capacity attainment on the same order as, respectively, that seen by the first and second voltage plateaus in LiTFA-containing electrolyte. While the first plateau at ~ 2.37 V matches that seen in the control and LiTFA-containing electrolytes, this potential quickly drops to ~ 2.16 V. Meanwhile, a third plateau occurs at a relatively high potential of 1.9 V, before dropping to a sloping 4^th plateau that polarizes to the discharge cutoff. While this additive may be promising as a solution to ameliorate poor low-temperature performance, its unprecedented electrochemical behavior requires further study to understand the coordination and speciation of polysulfides during each plateau. For instance, this behavior may reflect the staged discharge of distinct mixed ammonium-lithium polysulfide species like what is seen in the NMR spectra of Figure 6b. Another possibility ties to how the active material complex is solvated and shielded during the discharge process.

The quasi-equilibrium potential at which conversion takes place is inversely tied to how effectively solvated the active material is.^9,26 This can be seen in the case of high donor number solvents, which more effectively bind and shield Li⁺. This shielding effect increases the relative softness of the cationic complex, which then proceeds to bind to the hard S^2- discharge product less effectively.^9,26 This in turn lowers the discharge potential of the second plateau, which can be interpreted as a change in the chemical activity of the constituent species from the Nernst equation. Similarly, clustering of polysulfides increases the shielding of Li⁺, lowering the quasi-equilibrium potential of reduction. The substitution of NH₄⁺ serves to deter cluster formation, boosting the Li₂S conversion potential as seen in the third plateau at 1.9 V. However, NH₄⁺ is a softer Lewis acid than Li⁺ and is more effectively solvated by surrounding DOL/DME molecules, which may drive the losses in potential seen in the second and fourth reduction plateaus.

This highly atypical and nonintuitive electrochemical behavior corroborates that the presence of NH₄⁺ indeed transforms the coordination environment of dissolved polysulfide species away from the lithium polysulfide clustered aggregates typically seen at low-temperature conditions. The divergence in the impact of this additive under the application of low temperatures and lean electrolyte amounts points to the driving influence of the polysulfide coordination environment on electrochemical performance. Additionally, this speaks to the practical capability of additives like NH₄TFA, which deliberately take advantage of the intrinsic polysulfide crossover to enable an in-situ molecular engineering.²⁸ The additive shown here was selected to assess the joint effects of cationic and anionic modifications to the polysulfide coordination sphere, and sets the stage for future efforts into understanding, modifying, and exploiting the solvation and coordination of active material.

Conclusion

The materials chemistry underlying the Li-S battery system presents a rich array of complexity due to the innate solution-mediated pathways underscoring the electrochemical behavior. Engineering the system for practical applications requires understanding and improving the discharge behavior at challenging conditions like low E/S ratios and low temperatures. Under such conditions, understanding the coordination environment surrounding the dissolved active material is key. This is due to the interplay between polysulfides and surrounding species influencing almost every factor involved in discharge, including materials transport properties, chemical activities of dissolved species, and Li₂S electrodeposition dynamics.

In this work, NH₄TFA was utilized in the electrolyte additive to understand the joint impacts of strongly binding TFA⁻ and NH₄⁺ ions on the coordination sphere of lithium polysulfides. These species are used to assess and modulate the fundamental polysulfide clustering dynamics that underlie behavior at lean electrolyte and low-temperature conditions. Through a variety of electrochemical, spectroscopic, and computational studies, the in-situ conversion of NH₄TFA and lithium polysulfides to LiTFA and diammonium polysulfides is shown to amplify solution-mediated kinetics by deterring the formation of unfavorable polysulfide clustered aggregates. The Li⁺-S_x²⁻ framework making up the core of polysulfide clustered networks is disrupted in favor of NH₄⁺-S_x²⁻ and Li⁺-TFA⁻ ionic associations, amplifying the solution-mediated kinetics that the Li-S battery chemistry relies upon.

Beyond presenting a novel platform for amplifying molecular-scale electrostatic competition, this knowledge strengthens our understanding of the mechanisms underlying the dynamic Li-S electrochemistry. Building off this work, there is a rich, underexplored space for chemical investigation and engineering development within the Li-S electrolyte environment. With continued insight and mechanistically guided augmentation, the promise of Li-S batteries arrives ever closer to reality.