Differences in Electrochemistry between Fibrous SPAN and Fibrous S/C Cathodes Relevant to Cycle Stability and Capacity

Two different Li/S cathodes are compared in terms of capacity (mA . h . g sulfur − 1 ) and intermediates during discharge and charge. One cathode material is based on ﬁbrous SPAN, a sulfur-containing material obtained via the thermal conversion of poly(acrylonitrile), PAN, in the presence of sulfur. In this material, sulfur is covalently bound to the polymeric backbone. The second cathode material is based on porous activated carbon ﬁbers (ACFs) with elemental sulfur embedded inside the ACFs’ micropores. Cyclic voltammetry clearly indicates different discharge and charge chemistry of the two materials. While S-containing ACFs show the expected redox- chemistry of sulfur, SPAN does not form long-chain polysulﬁdes during discharge; instead, sulﬁde is chopped off the polymer-bound sulfur chains to directly form Li 2 S. The high reversibility of this process accounts for both the high cycle stability and capacity of SPAN-based cathode materials.

Lithium-sulfur (Li/S) batteries offer great potential for efficient, safe and economic cell systems. 1 Generally, the potential for commercialization of a cell system depends on the price, the safety of the device, energy density, longevity, cycle stability and rate capability. The reversible reaction in Li/S batteries is S 8 + 16Li 8Li 2 S and gives a high theoretical specific energy of approx. 2600 W·h·kg -1 , which is higher than the specific energies of Li-ion-batteries. 1 This high specific energy density of Li/S batteries results from the high specific capacities of the elemental sulfur cathode with 1672 mA·h·g -1 for the reaction S + 2e -→ S 2and the lithium metal anode with 3860 mA·h·g -1 for 2Li → 2Li + + 2e -. Nonetheless, Li/S cells also face several challenges such as a low sulfur utilization and low rate capability, which are caused by the insulating character of cyclo-S 8 and its slow redox-kinetics. 2 Another issue is the polysulfide shuttle. Polysulfides (S x 2-, x = 3-8) are formed as intermediates during charging and discharging. They dissolve in ether-and glyme-based electrolytes, migrate to the anode and form a passivating layer on the Li-surface. This results in a loss in capacity and decreases the cycle stability of the battery. 1,3 In view of these issues, it is not surprising that commercialization of Li/S cells is still limited by poor rate capability and comparable fast degradation. In an effort to overcome these problems, several successful approaches have been presented [4][5][6] and entail the incorporation of the insulating sulfur into a conducting framework as well as retention of sulfur in constrained geometries, porous organic frameworks, 7 silicon-carbons 8-10 and nitrogen-doped carbonbased composite materials. 11,12 Also microporous, 13 mesoporous 14 or porous (hollow) carbonaceous materials [15][16][17] or hierarchically nanostructured carbonaceous materials [18][19][20] have to be mentioned. Unfortunately, most, if not all of them, are too sophisticated to be of any real commercial usefulness. Based on the system originally proposed by Wang et al. 21,22 we recently presented sulfurized poly(methyl methacrylate)/ poly(acrylonitrile) (PMMA/PAN) compound fibers, referred to as SPAN, which contain up to 46 wt.-% sulfur and their high potential as Li/S cathode material. 6 In combination with 3M lithium bis(trifluorosulfonyl)imide (LiTFSI) in fluoroethylene carbonate (FEC): 1,3-dioxolane (DOL, 1:1) as electrolyte, it allows for the highly reversible charging/discharging up to 8 C over >1200 cycles. This z E-mail: michael.buchmeiser@ipoc.uni-stuttgart.de high cycle stability is related to the very unique structure of SPAN, which has been elucidated earlier by our group (Figure 1). 6,[23][24][25] One of the most important structural features of SPAN is the presence of covalently bound sulfur in form of (vinylogous) thioamides, which exist in their enolate form, thereby allowing for the formation of intra-and intermolecular oligosulfide chains. In turn, during charging, the enolic thioamides in their anionic form have been proposed to serve as "docking sites" for sulfur during recharge, which allow for the reversible reformation of the above-outlined oligosulfides. One drawback, however, is the comparable low sulfur loading of SPAN, which is limited to ca. 46 wt.-%.
To shed more light on the very particular discharging and charging chemistry of SPAN, we carried out a comprehensive comparative study with SPAN and activated carbon fiber (ACF)-based sulfur cathode material containing up to 56 wt.-% of elemental sulfur. Notably, this sulfur loading is significantly higher than the one of SPAN (up to approximately 46 wt.-%). Investigations addressed differences in morphology, differences in cyclovoltammetry (CV) during charging and discharging as well as in discharge curves. A mechanism for the discharge/charge chemistry of SPAN is presented.

Experimental
Synthesis of active materials.-Fibrous SPAN was prepared from PMMA/PAN-based nonwovens (1:3) in the presence of excess S 8 at 550 • C for 3 h in a nitrogen atmosphere. 4 Excess sulfur was removed via overnight Soxhlet extraction with toluene. S/ACF-based cathode material was synthesized by using Kynol ACF-1603-20 activated carbon fibers (2000 m 2 ·g −1 ; pores <20 Å). Those were degassed at 100 • C in vacuum for 1 hour. S 8 was added to the carbon fibers in different ratios (Table S2). The sulfur was encapsulated into the carbon fibers under static vacuum (p = 0.01 bar) for 2 hours at 160 • C. 26,27 Residual sulfur was evaporated (p = 0.01 bar) at 250 • C under inert gas.
Electrochemical characterization.-Electrodes were prepared by coating a 70:15:15 wt-% mixture of fibrous SPAN or S/ACF: Super C65:PVDF in NMP on Al-foil (200 μm wet ). After drying of the electrode sheet at 60 • C, electrode coins 12 mm in diameter were punched out and transferred to Swagelok T-type cells using a Freudenberg FS 2190 membrane and/or a Cellgrad separator 2325 and freshly prepared electrolyte. Electrochemical testing was performed on a BasyTec XCTS-LAB. Current density and specific were calculated based on the mass of sulfur in the cathode (1C = 1672 mA·h·g -1 = 1 mA·h·cm -2 ).

Results and Discussion
Synthesis of cathode materials.-Fibrous SPAN was synthesized as reported. 6,23-25 Elemental analysis (Table S1) revealed S-contents between 38 and 44 wt.-%. S/ACF-based cathode material was synthesized using Kynol ACF-1603-20 (2000 m 2 g −1 ; pores < 20 Å). ACFs were degassed at 100 • C in vacuo (p = 0.01 bar) for 1 hour. Next, various amounts of S 8 were added (Table S2). The sulfur was encapsulated into ACFs under static vacuum (p = 0.01 bar) for 2 hours at 160 • C. 26,27 Residual sulfur was evaporated at 250 • C under nitrogen. The maximum amount of sulfur (56 wt.-% S) was reached with an S:ACF-ratio of 1:1. SEM images in combination with EDX (Figures S1 and S2, S.I.) revealed the absence of any sulfur clusters on the surface of the fibers. Notably, as revealed by XRD ( Figure S3, S.I.), the small pore diameter of the ACFs effectively suppressed the formation of crystalline sulfur modifications (α-/β-S 8 ) inside the fiber.
Structural comparison.-Despite the fibrous structure of both cathode materials, these significantly differed from each other. Fibrous SPAN 6 had a diameter around 15 μm compared to ACFs having a di-ameter of 9 μm. Also, SPAN had a specific surface area of 22 m 2 ·g −1 compared to ACFs, which had a specific surface area of 2000 m 2 ·g −1 resulting from a mostly microporous structure with pore diameters <2 nm (Table S1, S.I.). These small pores were used for infiltration with elemental sulfur, with the goal that the small pores would successfully impede the formation of large, crystalline sulfur entities, which in turn should improve redox kinetics of these cathodes. By contrast, as outlined above, SPAN-fibers consisted of a polymeric, aromatic structure with covalently bound sulfur. Generally, short-chain poly(sulfide)s (Li 2 S x ; x ≤ 3) that form upon discharge have a lower solubility in carbonates than in ethers. 23,[27][28][29][30] Vice versa, long-chain poly(sulfide)s are soluble in both carbonates and ethers. However, carbonates can react with the dissolved poly(sulfide)s via nucleophilic addition. 28 Consequently, carbonates are usually considered incompatible with classical Li/S batteries based on S 8 . As outlined earlier, the small pores in ACF fibers allow only for the presence of small, amorphous sulfur fragments inside the pores. These small amorphous sulfur fragments in ACFs but also the  covalently bound oligomeric sulfur units in SPAN should form shortchain but no long chain polysulfides. 29,31 It therefore appeared beneficial to use FEC for both systems, not only to suppress polysulfide shuttle but also to form a stable solid electrolyte interfaces (SEI) on the anode and the cathode. 7,23,28,[31][32][33][34] As can be seen in Figure 2, SPAN gives a high initial capacity of approximately 1370 mA·h·g -1 while S/ACF delivers ca. 1400 mA·h·g -1 . Both cells show high cycle stability for at least 50 cycles with Coulomb efficiencies > 99% both for the SPAN-and ACFbased electrodes using an FEC-based electrolyte. Nonetheless, the SPAN cathode clearly outperforms the ACF-based one with the chosen electrolyte giving a reversible discharge capacity around 1200 mAh·g -1 compared to S/ACF with 260 mA·h·g -1 . With S/ACF, the substantial loss in discharge capacity between the initial and the second cycle can be attributed to the disability of ACFs to reintegrate all the Li 2 S formed into to fiber during charging. By contrast, the formed Li 2 S in the SPAN cathodes stays inside the porous fibrous network and can be reintegrated more easily and reversible. To understand the electrochemical differences between the two different cathode active materials, cyclic voltammetry (CV) was conducted, both with an FECand a 1,2-dimethoxyethane-(DME) containing electrolyte. Because CV curves did not change after the second cycle, only cycles 1 and 2 are shown (Figure 3). Table S3 summarizes the different poly(sulfide)s formed during discharge together with the corresponding voltage at which they are formed. [35][36][37][38] Upon discharge, S/ACF in 1M LiTFSI in DME/DOL as electrolyte shows a maximum in the range of 2.3-2.4 V, which correspond to long-chain poly(sulfide)s, i.e. Li 2 S 8 and Li 2 S 4 , respectively. Another maximum at 2.05 V corresponds to the formation of Li 2 S 2 (Figure 3a). During oxidation, a maximum at 2.4 V occurs. All maxima in the cyclovoltamogram correspond to the observed plateaus in the discharge curves with the DME/DOL electrolyte in Figure S4. The intensity of the maximum for the long-chain poly(sulfide)s at 2.3 V is relatively low compared to the next maximum at 2.05 V. While the XRD pattern clearly reveals the absence of any crystalline form of S 8 , CV data strongly suggest the presence of elemental sulfur since the signal for long-chained poly(sulfide)s is present in the CV of the ACF-based cathode with the DME/DOL electrolyte and the Li 2 S 8 signal at 2.3 V must be a reaction product of elemental sulfur. Notably, the reduction to Li 2 S, which usually occurs at ca. 1.6 V, could not be detected. In fact, reduction apparently stops at Li 2 S 2 , which serves as good explanation for the lower discharge capacities.
By contrast, the CV of SPAN (Figure 3b) shows only one maximum in the first discharge (U = 1.35 V), which is substantially less pronounced in the second discharge. There, weak signals in the range of ca. 2-1.6 V occur, which cannot be unambiguously attributed to a distinct sulfide species. Upon charging, a broad signal starting 2.1 V with one maximum at 2.2 V and a second signal at 2.45 V is observed. This lack of any pronounced signals at U > 2 V, which can in contrast to S/ACF not be assigned to individual sulfide species indicates the absence of any long-chain poly(sulfide)s such as Li 2 S 8 , Li 2 S 6 and Li 2 S 4 during discharge. The signal starting at 1.7 V is attributed to the formation of Li 2 S. As outlined in Figure 1, the sulfur chains in SPAN are covalently bound to the polymeric backbone. The absence of the signals for long-chain poly(sulfide)s indicates a reduction mechanism, which neither starts from S 8 nor via the formation of S x 2− with x ≥ 4. Instead, data strongly suggest that the polymer-bound sulfur chains in SPAN, whether intra-or intermolecular, are first reductively broken at some point to form terminal SPAN−S x -S − moieties. These are then degraded stepwise from the chain end with concomitant formation of Li 2 S until the sulfur chain is completely reduced according to SPAN−S x − + 2Li + + 2e − → SPAN−S x-1 − + Li 2 S (starting from 2 ≤ x ≤ 7 and ending at x = 2). This is supported by the continuous discharge curve shown in Figure S6. The diffuse maximum during charging shows the reversible incorporation of sulfur into the polymer backbone. By contrast, S/ACF with the FEC-based electrolyte shows a broad maximum at 1.6 V during the discharge cycle (Figure 3c) indicating the formation of Li 2 S. A small maximum at 2.35 V in the first cycle indicates the formation of long-chained poly(sulfides). These must react with the carbonate-based electrolyte since they cannot be detected in the second cycle. 39 This proposed loss of sulfur is in accordance with the observed drop in capacity ( Figure 2). The absence of long-chain poly(sulfides) leads to a short plateau in the discharge curves ( Figure S5) leading to comparably low capacities (260 mA·h·g -1 ).
Results for the SPAN cathodes in combination with FEC-based electrolyte are shown in Figure 3d. Only one signal can be detected at 1.35 V in the first cycle. In all further cycles, this maximum is shifted to 1.7 V. This indicates the improvement of the reversibility of the cell after the first cycle. The signal for the oxidation at 2.3 V corresponds to the signal for the formation of sulfur chains in SPAN according to SPAN−S − Li + + x Li 2 S → SPAN−S x+1 − Li + + 2x Li (1 ≤ x ≤ 7). The broad signal clearly indicates the different lengths of the sulfur chains in SPAN (4 ≤ x ≤ 8).
In summary, we have presented two fiber-based cathode materials for Li/S cells. The fibers strongly differ in the way the sulfur is integrated. CV clearly revealed differences in the discharge/charge behavior. Despite the good cycle stability of the S/ACF (50 cycles) with different electrolytes, cathodes based on this material show a substantially lower capacity than SPAN-based fibers, which gave better cycle stability and a higher rate capability with higher discharge capacities. CV clearly revealed the different discharge/charge pathways in these two cathode materials.