Understanding the Impact of Precipitation Kinetics on the Electrochemical Performance of Lithium − Sulfur Batteries by Operando X ‑ ray Di ﬀ raction

: The complex reaction mechanism of the lithium − sulfur battery system consists of repetitive dissolution and precipitation of the sulfur-containing species in the positive electrode. In particular, the precipitation of lithium sul ﬁ de (Li 2 S) during discharge has been considered a crucial factor for obtaining a high degree of active material utilization. Here, the in ﬂ uence of electrolyte amount, electrode thickness, applied current, and electrolyte salt on the formation of Li 2 S is systematically investigated in a series of operando X-ray di ﬀ raction experiments. Through a combination of simultaneous di ﬀ raction and resistance measurements, the evolution of Li 2 S is directly correlated to the variation in internal resistance and transport properties inside the positive electrode. The correlation indicates that at di ﬀ erent stages the Li 2 S precipitation both facilitates and impedes the discharge process. This information about the kinetics of Li 2 S formation o ﬀ ers mechanistic explanations for the strong impact of di ﬀ erent electrochemical cell parameters on the cell performance and, thus, directions for holistic optimizations to achieve high sulfur utilization.


INTRODUCTION
Lithium−sulfur (Li−S) batteries have attracted extensive research interest in the recent years mainly due to their high theoretical specific energy of 2552 Wh kg −1 and the abundance of sulfur as a byproduct of the petroleum industry. 1,2 However, only about 400 Wh kg −1 of the specific energy at cell level has been achieved in practice, at least for commercial cell formats and during longer cycling. 3−5 Apart from the mass of inactive materials, this discrepancy can be attributed to the incomplete sulfur utilization and the excess of lithium needed due to its low reversibility.
As the positive electrode of a Li−S cell is discharged, elemental sulfur is reduced to lithium polysulfides (Li 2 S x , x = 2−8), which are soluble in the commonly used ether-based electrolyte to different degrees. 1 Subsequently, the insoluble reaction product lithium sulfide (Li 2 S) is formed in the last stages of discharge. Because both elemental sulfur and Li 2 S are insulating, a conductive matrix, usually made from carbonaceous materials, is necessary to facilitate the electrochemical reactions. 6 Nonetheless, the complexity of the multiphase reactions with multiple possible routes is still considered to be the main challenge for elevating the utilization of the active materials in the positive electrode. On the other side of the cell, the metallic lithium electrode suffers from nonuniform stripping and plating, which results in poor reversibility. 7 Besides these issues reducing specific energy, the spontaneous reaction between lithium and dissolved polysulfides also compromises the Coulombic efficiency (CE), 8 which is another major challenge facing this system.
To enhance sulfur utilization, various approaches have been proposed, including encapsulating sulfur in the carbon matrix, 9,10 by using electrolyte with low polysulfide solubility, 11−13 incorporating polysulfide-interacting polar species into the carbon matrix by doping carbon 14 or by using specific types of polymer binders, 15,16 doping carbon with transition metals, 17 and increasing the donor number of the electrolyte solvent 18,19 and/or the anion of the salt. 20,21 All of these methods aim to retain the sulfur species inside the carbon matrix and promote their reduction, which eventually leads to the precipitation of Li 2 S. Microscopic studies of the formation of Li 2 S with scanning electron microscopy (SEM) have revealed that the precipitation follows a nucleation-and-growth behavior, whose kinetic parameters and thereby morphology are dependent on the current and electrolyte solvent. 22, 23 For example, the growth geometry was observed by atomic force microscopy (AFM) 24 and SEM 20,25 to vary with the electrolyte salt. At the electrode level, the formation rate and crystallite/ cluster size were found to change with the states of discharge (SoD), temperature, and current by operando X-ray diffraction (XRD) 26 and X-ray microscopy (XRM). 27 The formation of the Li 2 S was also characterized simultaneously with the cell resistance, which demonstrated the correlation between the precipitation and transport properties inside the positive electrode. 28 Recently, the investigation of Li 2 S formation was extended to promoting three-dimensional growth of Li 2 S by increasing the donicity of the electrolyte. Explanations for this approach are based on Pearson's hard/soft acid base (HSAB) theory, 13,18,20,24,29,30 which states that hard Lewis acids, e.g., Li + , have higher affinity toward hard Lewis bases, e.g., S 2− and S 4 2− , than to the "softer" bases, e.g., S 3 •− , S 6 2− , and S 8 2− . In an electrolyte with high donicity, Li + is more strongly complexed by the solvent or the anion of the salt and thus becomes a softer acid as a cluster. Although several exact mechanisms are proposed, there is a consensus that the softer Li + clusters stabilize different species of polysulfides and thus affects the precipitation of Li 2 S. It has been demonstrated with open/ nonencapsulating carbon matrices that Li 2 S particles are larger, and higher sulfur utilization is attained in electrolytes with solvents 18,19 or anions of the salt 20,21 that have high donor numbers.
Despite this abundance of research on the precipitation of Li 2 S, the diverse range of parameters when constructing and testing Li−S cells makes it difficult to compare results across these different studies to draw systematic conclusions. It is well-recognized that the electrochemical properties of the Li−S system, due to its catholyte nature, are sensitive to several parameters of the cell construction, such as the composition and structure of the positive electrode 16,31,32 and the amount of electrolyte used. 31−33 Actually, the fundamental reason for this sensitivity can be reasonably assumed to be the formation of Li 2 S itself because it involves the polysulfides in the electrolyte 23 and can also alter the transport properties inside the carbon matrix. 28 Therefore, it is crucial to keep other parameters constant when probing the effect of one specific factor on the precipitation of Li 2 S. Moreover, such investigations should be conducted with practically relevant cell configurations to ensure that the results are transferable.
In this work, operando XRD experiments are conducted to investigate the Li 2 S precipitation under various conditions. These cells are assembled with the same metallic lithium electrode and separator and employ an optimized coating recipe with the same sulfur-to-carbon ratio in the positive electrode. 16 The modified coin cells with X-ray transparent windows utilize a previously investigated design, 28 which ensures uniform stack pressure and preserves the electrolyte volume. The real-time resistance measurements by the intermittent current interruption (ICI) method 28,33,34 demonstrate the equivalence of the electrochemical properties of the operando cells and provide instantaneous insights into the effect of the Li 2 S precipitation, which in turn is captured by XRD. The influence of the electrolyte-to-sulfur (E/S) ratio and sulfur loading will be examined first, followed by the impact of the C-rate. Finally, a substantially higher precipitation rate is found by incorporating 50% lithium bromide (LiBr) as a cosalt. Interestingly, the high precipitation rate does not translate into high sulfur utilization in an encapsulating carbon matrix. This study thereby aims to construct the missing link between the performance of the S/C composite electrode and the factors impacting it through analyzing how these factors affect the precipitation of Li 2 S and thus the electrochemical properties of an operating Li−S cell. Lithium metal foil (Li, 125 μm thick, Cyprus Foote Mineral) was also used as received but stored under an Ar atmosphere. Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI, BASF), lithium bromide (LiBr, Sigma-Aldrich), and lithium nitrate (LiNO 3 , Sigma-Aldrich) were dried at 120°C under vacuum for 12 h. 1,2-Dimethoxyethane (DME, BASF) and 1,3dioxolane (DOL, Sigma-Aldrich) were dried with 3 Å molecular sieves for at least 12 h. Celgard 2400 separators were dried under vacuum at 80°C for 8 h.
2.2. Electrode Fabrication and Cell Assembly. The S/ C composite electrodes were fabricated through slurry coating by a doctor blade. Based on a previously optimized recipe, 16 the slurry consists of 65% S, 21% KB, 3.5% C65, 3.5% CNF, 5.6% PEO, and 1.6% PVP by mass. S and KB were first mixed in a mortar and then heated to 155°C for 20 min. After cooling to room temperature, the mixture was blended with the rest of the components and put into a ball mill jar with 20 vol % isopropanol solution in deionized water. After 2 h in a planetary ball mill, the mixture turned into a homogeneous slurry and was coated on the C-coated Al foil with two gap settings on the doctor blade corresponding to two targeted sulfur loadings. After initial drying under atmospheric conditions, the coatings were cut into ⌀ 13 mm discs and further dried under vacuum at 55°C for 12 h.
The design of the modified coin cell is illustrated in Figure 1. The coin cell cases with drilled holes (⌀ 7 mm) were sealed with Al-coated polyimide by the hot melting tape. The rest of the assembly process was the same for both operando XRD cells and standard coin cells for benchmarking, except for the different spacers. The cell stack was composed of ⌀ 13 mm S/ C electrode, ⌀ 17 mm Celgard 2400 separator, and ⌀ 15 mm Li. The electrolyte was then added to the cell according to the The Journal of Physical Chemistry C pubs.acs.org/JPCC Article assigned E/S ratio by an automatic micropipet right before the cell was closed by a crimper. Five different types of cells were constructed by systematically varying the sulfur loading of the electrode (2.5 and 3.3/ 3.4 mg S cm −2 ; called "S-L") and the electrolyte-to-sulfur ratio (6 and 10 μL mg S −1 ; called "E/S"). The results from the operando XRD cells are included in the main text while the cycling data of their conventional/unmodified counterparts are in the Supporting Information. The cells are hereafter termed "S-L = X E/S = Y", where X and Y denote the electrolyte content and sulfur loading, respectively. The "standard" electrolyte formulation was 1 m LiTFSI, 0.25 m LiNO 3 in DME/DOL (1:1, v:v), while the cells marked with "TFSI/Br" used a formulation of 0.5 m LiTFSI, 0.5 m LiBr, 0.25 m LiNO 3 in DME/DOL (1:1, v:v).
2.3. Electrochemical Cycling, Operando XRD, and the ICI Method. XRD measurements were conducted on a STOE STADI P diffractometer in transmission mode with monochromatized Cu Kα 1 radiation (45 kV, 40 mA). The detector system composed of three stationary Dectris Mythen 1K strip detectors (∼18°apart) covering an angular 2θ range from 0°t o 54.8°with an angular resolution of 0.015°(2θ). A photograph of the setup can be found in Figure S1. All the patterns presented were collected for 903 s, except for the first 385 patterns (approximately the first 4.5 cycles) of the cell "S-L = 3.3 E/S = 6", which were collected for 906 s due to a manual error.
All cells started with the same electrochemical testing program. The cells were rested for 6 h after assembly. As the XRD measurements started, the cells were discharged at C/50 (C = 1672 mAh g S −1 ) to 1.9 V and subsequently charged at C/ 25 to 2.6 V in the first cycle. Afterward, the cells were galvanostatically cycled at C/10 between 1.8 and 2.6 V. The constant current was paused for 1 s every 5 min. The cell voltage was recorded every 0.1 s during the current interruption and analyzed by the previously reported ICI method, which renders an internal resistance (R) and a coefficient of diffusion resistance (k). 28,33 The former is a sum of electronic, ionic and charge-transfer resistances while the latter is proportional to the coefficient of the Warburg element 28 which is commonly used in the equivalent circuit models for the analysis of electrochemical impedance spectroscopy data. The cell "S-L = 2.5 E/S = 6" started with the same test protocol but switched to C/5, C/2, C/3, and C/20 for three cycles each after the eighth cycle. The cell "S-L = 3.3 E/S = 6 TFSI/Br" also started with the same test protocol but switched to C/20, C/13.3, and C/20 for three cycles each after the ninth cycle. The operando XRD cells were cycled with either an SP-150 or an SP-240 (Bio-Logic) portable potentiostat while the standard coin cells were cycled with an Arbin BT-2043 potentiostat.
The evolution of the Li 2 S species was investigated by modeling the Li 2 S 111 reflection (Fm3m) in the operando XRD data by using Topas Academic (V6) software. The reflection was modeled by using a single Gaussian peak whose intensity and width were refined with its position constrained to a 24°−29°(2θ) angular range. A fourth degree Chebychev polynomial was used to model the background. To facilitate a comparison between the different cells, the intensity of the Li 2 S 111 reflection was normalized to the 002 reflection of the Kapton tape (polyimide) used as windows for the operando cell, thereby serving as an internal standard.
The raw data from the potentiostats and the diffractometer can be accessed with the fitting results from Topas and the script written in the R-programming language via the data deposit Zenodo. 35

RESULTS AND DISCUSSION
3.1. Effect of Electrolyte Amount (E/S Ratio). The results from the fifth cycle of the operando XRD measurements are summarized for four cells with two sulfur-loadings and two E/S ratios in Figure 2. These cells contain the same standard electrolyte (i.e., without LiBr) and were cycled at C/ 10 in this cycle. The analysis is chosen to start at the fifth cycle because it has been observed in previous work that the properties of these Li−S cells stabilize after the first three cycles. 28 Figure 2 shows the development of the cell voltage (E), internal resistance (R), and diffusion resistance coefficient (k) as well as the normalized integrated intensity of the 111 reflection of Li 2 S (I), displayed as a function of the specific capacity (Q), where Q = 0 corresponds to the 0% state of discharge (SoD). The bottom panel displays the ratio between the change in the normalized intensity and the change in the total charge (dI/dq), which is analogous to the mass per (mole of ) electron (mpe) value often used in kinetic analyses. Note , and its rate of changing with respect to the total charge passing through the cell (dI/dq) for cycle 5 of the cells using the standard electrolyte and plotted against the state of discharge/charge in terms of specific capacity (Q). Cell names are labeled on top of the cell voltage, where S-L and E/S denote sulfur loading and electrolyteto-sulfur ratio, respectively. Data acquired during discharge and charge are plotted in blue and red, respectively. The dI/dq values are calculated based on the average of three consecutive I values to reduce the noise level. The raw data can be found in Figure S5. The R plots are cut off at 50 Ω cm 2 for clarity. Note that this excludes the last data point of the cell "S-L = 3.3 E/S = 6" during charging, which is 99.6 Ω cm 2 .
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article that a comparison between the modified coin cells for operando XRD and conventional coin cells with similar electrochemical cell parameters can be found in Figures S2− S4. It can be concluded that the cell voltage, resistances, discharge capacity, and the Coulombic efficiency of the modified coin cells are comparable to those of their conventional counterparts. The only major difference is in the specific capacity of the cell "S-L = 2.5 E/S = 10", but the cell voltage and resistance profiles are still comparable to those of the commercial coin cell counterpart at corresponding state of charge (SoC). From the two panels on the left in Figure 2, the effect of the E/S ratio can be observed in cells with a lower sulfur-loading of 2.5 mg S cm −2 . A pronounced increase in R can be identified during discharge and at the end of charging in the cell with less electrolyte. The rise in R between the two discharge plateaus has been attributed to the increase in solution resistance caused by the higher viscosity brought about by the higher concentration of dissolved polysulfides. 36−38 As the intensity of the Li 2 S peak starts to increase, R drops as a result of the decreasing concentration of polysulfides in the electrolyte. While R returns to the level before the rise in the cell with E/S = 10, it stays at intermediate values during the lower discharge plateau in the cell with E/S = 6. This discrepancy may be a consequence of the different precipitation rates of Li 2 S, as indicated by dI/dq. A faster Li 2 S formation can lead to a more dense passivation layer on the conductive carbon matrix, which has been observed in cells with higher current densities. 23 Thus, the charge-transfer resistance is increased, which contributes to the higher R on the lower discharge plateau in the cell with E/S = 6. The higher rate of precipitation can also be responsible for the early deterioration in the transport properties, which is seen in the values of k. The formation of insulating Li 2 S can worsen the ionic transport inside the porous carbon matrix by covering the inner surface and occupying the pore space. The former decreases the doublelayer capacitance of the carbon surface (C) while the latter increases the ionic resistance inside the pores (R′, not to be confused with R), which both increase k (or, equivalently, the Warburg coefficient), which can be expressed as the following according to the porous electrode model: 28 Previous analysis proposed that the formation of Li 2 S causes the initial slow growth of k, which is observed here in the "S-L = 2.5 E/S = 6" cell around 450 mAh g S −1 , by decreasing the capacitance of the carbon matrix. The subsequent sharp increase in k can be attributed to the replacement of the electrolyte volume by Li 2 S precipitates, which increases the ionic resistance inside the pores. If the ionic conductance (1/ R′) is proportional to the volume fraction of the electrolyte in the pores, a parabolic increase in R′ is expected from a linear increase in the volume of Li 2 S, which is observed here by XRD.
In the cell with E/S = 10, the slower precipitation leads to lower R and k values, except for the rapid increase of k at the end of discharge. The larger crystallite size seen from the smaller full width at half-maximum (FWHM) in Table 1 suggests more Li 2 S grows in the direction perpendicular to the carbon matrix rather than covering its surface, which renders a less passivating morphology according to previous SEM studies. 23 Therefore, the sharp increase in k at the end of discharge is likely not entirely caused by the deterioration of the transport properties inside the porous electrode. It has been demonstrated that k also characterizes diffusion processes coupled with electrochemical reactions, 39 which is not surprising since k is proportional to the Warburg coefficient. 28 Thus, the high k values observed at the end of discharge may partially stem from the slow mass transport of polysulfides from the catholyte in the separator, as proposed previously. 40 This is more probable to be limiting in a cell with low polysulfide concentration in the electrolyte, which also explains the lower specific capacity and the lower solution resistance seen for the cell with E/S = 10. However, given the considerable Li 2 S intensity at the end of discharge, the effect of Li 2 S occupying space inside the pores, which also increases k as discussed above, cannot be ruled out in the cell "S-L = 2.5 E/S = 10". It is plausible that both factors contribute to the rapid increase in k at the end of discharge.
In summary, the higher E/S ratio reduces the polysulfide concentration and the formation rate of Li 2 S, resulting in both lower resistances and lower specific capacity, as summarized in the scheme in Figure 3. Such effects of the E/S ratio are more Standard deviations in parentheses share the same digit of the last digit of the mean value; e.g., 11.5(3) means the standard deviation is 0.3. The labels of the cells are the same as those used in Figures 2, 4, and 6, where S-L and E/S denote sulfur loading and electrolyte-to-sulfur ratio, respectively. pronounced for S/C electrodes with the higher sulfur loading, as displayed when comparing the two panels on the right in Figure 2. The differences in precipitation amount and rate are more distinct than in the cells with the lower sulfur loading, as also demonstrated in Table 1. The stronger impact of the E/S ratio at the higher sulfur loading may be a consequence of the fixed loss of electrolyte outside the positive electrode, possibly due to reactions with metallic Li, absorption in the separator, and possible diffusion through the X-ray transparent window of the modified coin cell for operando XRD in this case. Because the E/S ratio is scaled by the amount of sulfur, the same increment in the E/S ratio results in more additional electrolyte available for the S/C electrode when the sulfur loading is higher.
On charging, both R and k decrease rapidly in all four cells in Figure 2, but a delay in the drop can be observed in the cells with the lower E/S ratio. This can be expected from the higher degree of passivation by Li 2 S during discharge at lower E/S ratio, as elaborated above. Subsequently, R and k remain low and do not seem to be correlated to either the amount or dissolution rate of Li 2 S. A slight increase in R can be identified between the lower and upper charging plateaus, but the degree of increase is substantially smaller compared to that during discharge, which is presumably due to the different reaction routes. 41,42 At the end of charge, an increase in R can be observed in all cells but is more pronounced in the cells with lower E/S ratio. A recent report on three-electrode resistance measurements shows that this rise in R is contributed by the positive electrode. 43 Therefore, it can be correlated to the formation of elemental sulfur, which can be expected to be more substantial in the cells with lower E/S ratio for the same reason more Li 2 S precipitates. However, with a one-dimensional strip detector as used in this study, it is not feasible to directly quantify elemental sulfur due to its varying preferred orientation.
3.2. Effect of Sulfur Loading. The effect of an increased sulfur-loading on the cells with the same E/S ratio of 6 μL mg S −1 is depicted in the first and third columns of Figure 2. The rise in R between the upper and lower discharge plateaus is lower in the cell with the high sulfur loading. Considering that the rise is caused by the increase in polysulfide concentration, there are two plausible reasons. First, at the same E/S ratio, the cell with higher sulfur loading has effectively more electrolyte, as discussed in the previous section. Second, the upper discharge plateau is shorter for the cell with the higher loading, indicating that a lower degree of sulfur utilization on the upper discharge plateau. This may be caused by the nonuniform reaction progress along the axial direction of the electrode, resulting from the concentration gradient of Li + formed in the same direction.
This heterogeneity brought by the higher sulfur loading may also be responsible for the subsequent increase in R during the rest of discharge, which instead decreases in other cells. Together with the earlier increase in k on the lower discharge plateau, it can be deduced that the transport properties of the porous carbon matrix deteriorate faster despite the similar dI/ dq values of 5 mAh −1 in both cells. This faster deterioration at the same normalized precipitation rate is most likely also a consequence of the increased concentration gradient of Li + in the thicker electrode. The above-discussed effects of the sulfur loading are summarized in Figure 3.
Compared to E/S = 6, at the higher electrolyte volume, E/S = 10, the higher sulfur loading results in a decrease in the normalized rate of precipitation (dI/dq) and more loss in specific capacity, as displayed in the second and fourth columns of Figure 2 and Table 1. On the lower discharge plateau, the increase in sulfur loading results in slightly lower R and k, which means that the nonuniform reactions brought by the higher loading at E/S = 6, are mitigated by the lower polysulfide concentration at E/S = 10. Moreover, the slight decrease in R and k is consistent with the reasoning for the more accessible electrolyte at higher sulfur loading at a given E/S ratio, which is corroborated by the lower precipitation rate. The explanation for the lower specific capacity due to mass transport limitations may also apply here.
3.3. Effect of Specific Current (C-Rate). The operando XRD results of the cell with a sulfur loading of 2.5 mg S cm −2 and E/S ratio of 6 μL mg S −1 are exhibited in Figure 4 for various C-rates. The first column is the same as the first column in Figure 2. The data from cycles at C/3 are omitted here due to the similarities with the ones at C/2. In the voltage profiles, the impact of increasing specific current is obvious above C/10, which results in the shrinkage of the lower discharge plateau at C/5 and its absence at C/2. This premature termination of discharge can be attributed to the higher solution resistance at higher current. The nonuniform current distribution along the axis of the electrode may force The Journal of Physical Chemistry C pubs.acs.org/JPCC Article the reduction reactions to proceed primarily on the separator side, while sulfur species with higher oxidation states are still present on the current collector side. Therefore, the higher concentration of the polysulfides, which make the electrolyte more viscous, 36 can be expected to on the separator side and thus increase the solution resistance. At C/5, a simultaneous rise in k and R between the upper and lower plateaus indicates a mass transport limitation due to the concentration gradients of active materials, while, at C/10 and C/20, the value of k remains low at the same SoC. The increasing rate of precipitation (dI/dq) on the lower discharge plateau at C/20 may be explained by the more homogeneous distribution of polysulfide species across the positive electrode. The lower current allows the interpolysulfide reactions, including redox and disproportionation reactions, to homogenize the polysulfide species in addition to the physical mass transport, which is classified as the "solutionmediated pathway" in the literature. 23 In contrast, at higher currents, the inter-polysulfide reactions cannot keep pace with the electrochemical reduction, so Li 2 S is precipitated near the carbon surface, where the reduced species are formed and concentrated, 23 while higher-order polysulfides are still present. Therefore, a more narrow distribution of the oxidation states of the polysulfides means that Li 2 S precipitates in a more narrow range of SoC, which is consistent with the lower and higher dI/dq values in the first and second halves of the lower discharge plateau, compared to the constant dI/dq at C/10. The lower R at C/20 in the first half of the lower discharge plateau, 300−600 mAh g S −1 , can be attributed to less passivation of the carbon matrix. The higher k during the second half of the lower discharge plateau is expected from the higher ionic resistance inside the pores (R′ in eq 1) due to more short-chained polysulfides and/or the higher precipitation rate. The higher R accompanying the rise in k suggests that the charge-transfer is impeded by either electrolyte viscosity or passivation of the conductive surface.
At C/5, the precipitation of Li 2 S is initiated in time to overcome the large peak in R between two discharge plateaus, as opposed to the case at C/2. On the lower discharge plateau, R is still higher in comparison to the case at C/10. Because the similar dI/dq means that the intensity of Li 2 S increases at a double rate with respect to time, the higher R values may indicate that fast formation of Li 2 S causes an increased passivation of the carbon matrix. The earlier and more rapid growth of k agrees with the higher degree of passivation. The discharge process stops at a lower k in comparison to the cycles at lower C-rates due to the higher overpotential contribution from R (iR drop). Also, these phenomena are illustrated in Figure 5. The decrease in I for the last XRD pattern before charging is due to the switch from discharging to charging during the 15 min data collection time.
3.4. Effect of Salt Anions with High Donor Number. The substitution of 50% of LiTFSI with LiBr in the electrolyte makes a substantial difference in the formation of Li 2 S and thereby the resistance profiles. An outward bulge of the Alcoated polyimide films indicated that more gas was generated in the cell when employing the LiBr salt, presumably on the interface between the electrolyte and metallic Li. This extra air pressure in fact burst the cell in our first attempt. The sealing of the second cell was therefore reinforced by applying an extra layer of hot-melting polymer by a glue gun after cell assembly. However, a higher loss of electrolyte can still be expected through the polyimide films due to the higher pressure. In addition, gas evolution may also worsen the contact resistance over cycling. Both factors may have affected the rate capability of the cell after the third cycle. Therefore, the data of both cells cycling at C/10 in Figure 6 are extracted from the second cycle.
In Figure 6, the operando XRD results from the cell, where 50% of LiTFSI is replaced by LiBr, are compared with those from the cell with only LiTFSI. An evidently higher rate of precipitation during the second half of the lower discharge plateau can be observed in the cell with LiBr. Moreover, the continuously increasing dI/dq is similar to the case at C/20 in the previous section, which suggests a more homogeneous distribution of the oxidation states of the polysulfides. This can be a consequence of the different dominating polysulfide species. 18,30 However, this alteration also increases the ionic resistance both outside and inside the carbon matrix, as shown by the increase in R and k at the beginning of the lower discharge plateau. Compared to the cell with pure LiTFSI, the crystallite size is smaller in the cell with 50% LiBr, as indicated by the larger FWHM in Table 1. This trend contradicts a previous report with a nonencapsulating carbon matrix, 20,21 which has much lower porosity and surface area than the encapsulating carbon matrix, KB, applied here. This demonstrates how the pore structure of the positive electrode can affect the precipitation process.
At the end of charge, the rise in R and k, which are observed in cells with pure LiTFSI, are absent in the cell with 50% LiBr. , and its rate of changing with respect to the total charge passing through the cell (dI/dq) of the cell "S-L = 2.5 E/ S = 6" cycled at C/10, C/5, C/2, and C/20 and plotted against the state of discharge/charge in terms of specific capacity (Q). S-L and E/ S denote sulfur loading and electrolyte-to-sulfur ratio, respectively. Data acquired during discharge and charge are plotted in blue and red, respectively. The dI/dq values are calculated based on the average of three consecutive I values to reduce the noise level. The raw data can be found in Figure S6.
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article This observation, together with the longer upper discharge plateau, agrees with the stabilization of S 6 2− and S 8 2− in the electrolyte with higher donicity in the previous reports. 30 In summary, the effects of electrolyte with LiBr are consistent with how the polysulfide species varies with the donor number of the electrolyte. However, the advantage of promoting the solution-mediated pathway of Li 2 S formation, as demonstrated with nonencapsulating cathodes, 20,21 may be compromised by the worse ionic resistance and thus transport properties inside an encapsulating S/C electrode, which stem from the high concentration of short-chain polysulfides and high precipitation rate of Li 2 S.

CONCLUSIONS
In this work, a combination of operando XRD and electrochemical analysis is employed to study how the precipitation of Li 2 S is affected by the Li−S cell parameters and operating conditions and its effect, in turn, on the electrochemical properties of the cell. It can be concluded that the formation of Li 2 S has two major roles in the discharge process. First, it regulates the concentration of polysulfides at the start of the lower discharge plateau, which can be critical in circumstances where the internal resistance surges between the two discharge plateaus, e.g., at high C-rates. Second, as the Li 2 S precipitates accumulate, they can passivate the conductive surface of the carbon matrix and reduce the volume fraction of electrolyte inside the porous matrix, both of which lead to the deterioration of the transport properties inside the positive electrode. The morphology of the precipitates, which are affected by the cell components and current density, adds further complexity to the effect of Li 2 S formation on the transport properties in the porous electrode. It was also demonstrated that decreasing the precipitation rate by increasing the E/S ratio also decreases the specific capacity, while increasing the precipitation rate by adding LiBr causes a surge in the overpotential. Therefore, a delicate balance has to be struck to achieve the optimal performance of a Li−S cell with the desired properties, e.g., a certain areal capacity or power capability.

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcc.1c10197. Figure 5. Effects of specific current (C-rate) on the dissolution and precipitation processes and their impacts on the internal resistance (R) and diffusion resistance coefficient (k), summarized from the experimental results in Figure 4. Figure 6. Cell voltage (E), internal resistance (R), diffusion resistance coefficient (k), normalized integrated intensity under the 111 reflection of Li 2 S (I), and its rate of changing with respect to the total charge passing through the cell (dI/dq) of the cell with standard electrolyte "S-L = 3.3 E/S = 6" cycled at C/10 and the cell with 50% LiTFSI 50% LiBr as the electrolyte salt "S-L = 3.3 E/S = 6 TFSI/Br" cycled at C/10, C/20, and C/13, plotted against the state of discharge/charge in terms of specific capacity (Q). S-L and E/S denote sulfur loading and electrolyte-to-sulfur ratio, respectively. Data acquired during discharge and charge are plotted in blue and red, respectively. The dI/dq values are calculated based on the average of three consecutive I values to reduce the noise level. The raw data can be found in Figure S7.