- Email: [email protected]
Li cells
Accepted Manuscript Title: Influence of catholyte composition on the performances of VACNT/Polysulfides/Li cells Author: S´ebastien Liatard Kamal Benhamouda Adeline Fournier Jean Dijon C´eline Barchasz PII: DOI: Reference:
S0013-4686(15)30885-9 http://dx.doi.org/doi:10.1016/j.electacta.2015.11.113 EA 26113
To appear in:
Electrochimica Acta
Received date: Revised date: Accepted date:
7-7-2015 13-10-2015 21-11-2015
Please cite this article as: S´ebastien Liatard, Kamal Benhamouda, Adeline Fournier, Jean Dijon, C´eline Barchasz, Influence of catholyte composition on the performances of VACNT/Polysulfides/Li cells, Electrochimica Acta http://dx.doi.org/10.1016/j.electacta.2015.11.113 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Influence of catholyte composition on the performances of VACNT / Polysulfides / Li cells Sébastien Liatarda, Kamal Benhamoudaa, Adeline Fournierb, Jean Dijonb, Céline Barchasza,* a
CEA, LITEN, Department of Electricity and Hydrogen for Transportation, 17 rue des Martyrs, F-38054
Grenoble, France. b
CEA, LITEN, Department of Technologies for NanoMaterials, 17 rue des Martyrs, F-38054 Grenoble, France.
*
Corresponding author
Mail: [email protected] Phone: 0033 438 78 90 36 Fax: 0033 438 78 51 98
Keywords Lithium/sulfur rechargeable batteries Catholyte composition Electrolyte additive Lithium polysulfide concentration Vertically-aligned carbon nanotubes Abstract Elemental sulfur has become a particularly attractive material for lithium metal batteries owing to its low cost, availability, non-toxicity and high specific capacity of sulfur active material. Problematics of the Li/S cell arise from its particular mechanism, based on dissolution/precipitation phenomena of active materials along cycling. The phase transitions endured by active material are responsible for dramatic changes in the morphology of the
positive electrode upon cycling, and results in severe capacity fading. In this work, verticallyaligned carbon nanotubes (VACNT) were investigated as a positive electrode 3D current collector for Li/S cells, combined with a semi-liquid VACNT / Polysulfides (catholyte) / Li battery configuration. The influence of the catholyte composition and the sulfur concentration, and the impact of electrolyte solvents and additives in the catholyte were investigated, yielding significant improvements in terms of practical surface capacity of the cells.
Introduction Lithium-ion batteries have been under intense studies for the past decades, due to the growing need of power for portable electronics, electric cars, or grid energy storage.1,2 For over two decades, Li-ion technologies have revolutionized the portable electronics market due to their unequalled volumetric energy density (650 Wh L-1).3 However, the development of Li-ionpowered electric vehicles has been hindered by the high costs and the low capacity of positive electrode materials (250 mAh g-1), resulting in a limited practical gravimetric energy (maximum of 250 Wh kg-1).4 To go beyond Li-ion batteries performances, today’s research is more and more focused on post-Li-ion systems, such as lithium/sulfur (Li/S) or lithium/oxygen (Li/O2) technologies.1,5 In particular, elemental sulfur has become a particularly attractive material for lithium metal batteries owing to its low cost, availability, non-toxicity and high specific capacity of sulfur active material.
The reaction between elemental sulfur (S8) and metallic lithium involves the exchange of 16 electrons per mole of S8, which allows for high theoretical specific discharge capacity (1675 mAh gsulfur-1) and high theoretical gravimetric energy density (2600 Wh kgsulfur-1). In practical, gravimetric energy densities of 300-600 Wh kgcell-1 are expected, thus making the Li/S cell
fully competitive with Li-ion counterparts.6 The Li/S rechargeable cell has been studied since the 70’s,7 but Li/S technology is still not commercialized. In fact, it disappointingly suffers rapid capacity fading, low practical discharge capacity and low coulombic efficiency.8 Problematics arise from the particular mechanism of the Li/S cell during charge/discharge processes, which are based on dissolution/precipitation phenomena of active materials along cycling, while Li-ion systems rather rely on intercalation/deintercalation reactions. The phase transitions endured by the active material in the battery during charge and discharge are mainly responsible for dramatic changes in the morphology of the positive electrode upon cycling, and results in severe capacity fading during the first cycles.8 Limitations in the Li/S performances are also linked to the strong insulating properties of solid sulfur compounds (i.e. S8 and final discharge products Li2S2/Li2S), which rapidly passivate the positive electrode when formed and lead to incomplete active material utilization. Last but not least, some of the problematics also originate in the use of organic liquid electrolyte and lithium metal negative electrode, which results in electrolyte consumption upon cycling, low coulombic efficiency, sudden cell fading upon prolonged cycling and possible safety issue. As a consequence, more and more attention is paid today to the Li metal electrode, its reliability, its potential replacements or its compatibility with liquid electrolyte.9,10
To address these issues, different approaches are considered in today’s research. A popular strategy involves developing sophisticated cathode architectures, through the preparation of sulfur/carbon composites, sulfur/polymer materials or a mixture of these three, and impressive results have been obtained this way in terms of cycle life, sulfur utilization and rate capability.11,12,13,14 In particular, the confinement approach is widely studied,15,16,17 and intends to prevent sulfur and polysulfides dissolution into the electrolyte upon cycling. However, capacity fading is still frequently observed, since polysulfides are likely to diffuse into the
electrolyte in the case of prolonged cycling, thus indicating that the confinement of sulfur may never be complete.18,19,20
Among alternative solutions for the positive electrode side, the interest of 3D current collectors has already been demonstrated as an efficient concept to replace classical 2D aluminum foil.21,22,23 In particular, we already demonstrated the beneficial effect of using vertically-aligned carbon nanotubes (VACNT) structure as a positive electrode current collector for Li/S cells, both in terms of practical discharge capacity and retention, and when employed in combination with a catholyte approach.22,24 First proposed in the early days of Li/S developments, the concept of catholyte recently gained increasing attention.7,25 In this approach, the electrolyte contains the active material in the form of dissolved polysulfides, and can be associated with a 3D current collector as a positive electrode. For this purpose, metal foams or carbon papers are promising substrates for enhancing both discharge capacity and capacity retention of Li/S cells,22,26,27 through the stabilization of positive electrode morphology during cycling. In previous work, we described a light-weight, high surface current collector made of VACNT grown on an aluminum foil, which was successfully used in a semi-liquid VACNT / Polysulfides / Li battery configuration (Figure 1). In this study, we investigate the influence of the catholyte composition on the performances of these semiliquid cells. The impact of electrolyte solvents on the discharge capacity was clearly evidenced, while capacity retention was found to be strongly correlated with the electrolyte additives. Finally, sulfur concentration in the catholyte was increased, yielding significant improvements in terms of practical surface capacity of the cells. Beneficial effects of VACNT and optimization of the catholyte were demonstrated, while the limitations coming from other cell components were also evidenced.
Experimental section Fabrication of VACNTs on aluminum: Vertically-aligned carbon nanotubes were synthesized by chemical vapor deposition technique on a 20 µm-thick aluminum substrate (Sumikei Al, Japan).24 Al foil was cleaned thoroughly by sonication in acetone and isopropanol baths, and by a plasma treatment (O2 0.1 mbar, 100 W, 2 min). A nano-layer of iron catalyst was deposited on Al using an electron beam physical vapor deposition (EBPVD) technique. Growth of VACNT was performed in a low pressure (~ 1 mbar) CVD chamber equipped with an array of 10 conductive filaments connected to a power source. Acetylene (C2H2, 20 sccm) was introduced in the reactor along with hydrogen (H2, 50 sccm) and helium (110 sccm). The chamber was heated up to 450 °C, and the temperature was sustained for 30 min. The length of VACNT could be tuned from few microns up to 150 µm, by modifying the residence time of samples into the CVD chamber. Images of the electrodes structure were realized within a Zeiss LEO 1530 scanning electron microscope. Catholyte preparation: Sulfur was introduced into the batteries in the form of polysulfides (average formula Li2S6) dissolved in the electrolyte, this system being called ‘catholyte’. In an argon-filled glove box, metallic lithium and sulfur powder (99.5%, 325 mesh, Alfa Aesar) were precisely weighted and introduced in the electrolyte solvent or mixture of solvents. Stoichiometric amounts of lithium and sulfur powders were weighted in order to obtain an average composition of Li2S6, and the dispersions were let reacting at 50°C over few days to ensure complete reaction between powders. Different catholyte compositions were prepared, based on the use of 1,3-dioxolane (Diox, anhydrous, 99,8%, Aldrich), 1,2-dimethoxyethane (DME, anhydrous, 99.5%, Aldrich) or tetraethylene glycol dimethyl ether (TEGDME, 99%, Adrich). TEGDME/DIOX mixture was used with a volume ratio of 1/1. DME was employed as a pure solvent while also in a mixture with DIOX (9/1 vol. ratio of DME/DIOX). The concentration of Li2S6 was varied between 0.25 M to 0.6 M in order to adjust the sulfur
loading, by dissolving the proper amounts of sulfur and metallic lithium materials. Lithium bis(trifluoromethane sulfone)imide (LiTFSI, 99.95%, Aldrich) and lithium nitrate (LiNO3, 99.99%, Aldrich) were dissolved in the catholyte at 1 M and 0.1 M respectively. For comparison purpose, the concentration of LiNO3 was also increased to 0.2 M, in order to investigate its effect on cell behavior.28 At least 48h of stirring at 50°C were necessary to dissolve all solid compounds and to form the catholytes. Electrochemical characterization: VACNT structures on Al were punched to 14 mm diameter electrodes and then used directly as positive electrodes in CR2032 coin cells. No sulfur deposition was performed on this current collector prior to cell assembly, since the positive active material was introduced within the electrolyte. The VACNT / Polysulfides / Li coin cells were assembled in an argon-filled glove box and a schematic representation of the cell design is presented in Figure 1. Coin cells were galvanostatically cycled between 3 V vs. Li+/Li and 1.8 V (or 1.5 V depending on the applied C-rate), either at constant current, or at a varying current density for rate capability tests (from C/50 up to C/2). A disk of lithium metal (16 mm) was used as a negative electrode. A Celgard 2400® served as the separator, assisted by a non-woven Viledon® layer to provide an electrolyte reservoir. 75 µL of catholyte was introduced in the coin cells, thus leading to theoretically loadings from 3.9 to 9.4 mAh cm-2 (from 2.3 to 5.6 mgsulfur cm-2).
Results and Discussion Figure 1 shows a schematic representation of the cell design that was considered in this study, as well as a SEM picture of a VACNT structure before cycling. In this configuration, VACNT serves both as a current collector and a positive electrode, while active material is initially dissolved in the electrolyte. Electrodeposition of sulfur (or Li2S) material is performed in situ in the battery by initial charging (or discharging) of the Li/S cell. Lithium metal was used as a
negative electrode in this study, but VACNT can also be combined with other negative electrode materials such as silicon or hard carbon,29 as the catholyte also contains a source of lithium ions to first charge the battery. VACNT structures produced by CVD on Al are composed of multi-walled (mainly double-walled) carbon nanotubes, with a mean diameter of 3.3 nm, that are vertically-aligned on the Al foil and organized into bundles of about 15-35 nm.24 VACNT could be produced with tube length ranging from few microns up to 150 µm. The performances of these structures were evaluated as a function of VACNT height for two different catholyte compositions: 1) Li2S6 0.25 M + LiTFSI 1 M + LiNO3 0.1 M + TEGDME/DIOX 1/1 and 2) Li2S6 0.25 M + LiTFSI 1 M + LiNO3 0.1 M + DME/DIOX 9/1. The two compositions were selected owing to their different solvation properties, their different viscosity, the first one being used as a reference in our studies (mainly with sulfur composite electrodes),8 while the second one being usually used as a reference in literature.9,10 Indeed, TEGDME has been reported for its good affinity with lithium polysulfides,30,31 but suffers a high viscosity (~4 cP). On the contrary, DME has a lower affinity with polysulfides,30 but also a much lower viscosity (0.455 cP), thus being the most common solvent in Li/S batteries. The electrochemical performances of the two catholyte compositions were compared in terms of sulfur utilization, rate capability and cyclability, with varying VACNT lengths, and the results are presented in Figure 2. Firstly, practical discharge capacity values at low C-rate (C/50, 0.12 mA cm-2) were always higher for DME-based cells whatever the VACNT length (Figure 2a), ranging from ~400 mAh gsulfur-1 for short VACNT up to about 1100 mAh gsulfur-1 for longer ones. This indicates that sulfur utilization in the DME-based catholyte is higher than in the TEGDME-based one. Maximum discharge capacity with increasing VACNT length was found to stabilize at 1100 mAh gsulfur-1 for DME-based cells, while stabilizing at 800 mAh gsulfur-1 in TEGDME only. The differences in sulfur utilization could be explained by the different electrolyte viscosities: DME solvent may allow better
electrolyte penetration into the VACNT structure thanks to its low viscosity, thus leading to better wettability of the positive electrode, and larger available conductive surface for higher sulfur utilization. In turn, the maximum capacity value, obtained when increasing VACNT height, is higher in case of DME, pointing out that this ‘saturated’ capacity strongly depends on the catholyte composition. On the contrary, it is really likely that using TEGDME, the accessibility of VACNT is much lower, and we can propose that discharge products mainly precipitate on the positive electrode surface than in the VACNT porosity, which strongly impacts the discharge capacity. This hypothesis is currently under investigation. In any case, it is worth noticing that both positive electrode and electrolyte compositions are strongly interconnected to each other, and both components make an inherent pair of materials that need to be optimized jointly. Thanks to the low weight of VACNT structures as compare to sulfur weight, the discharge capacity related to the masses of both S8 and VACNT was relatively high, of ~ 900 mAh gsulfur+CNT-1 even with VACNT as high as 90 µm. Then, due to the saturation of discharge capacity at about 1100 mAh gsulfur-1, the value referring to both S8 and VACNT masses is then progressively impacted by further increase of the VACNT height, pointing out the need of optimizing sulfur-to-carbon ratio. Thus, DME-based catholyte was found to provide high practical capacity at low C-rate in combination with VACNT structures, and the optimum VACNT height was found to be ~90 µm, resulting in a maximum practical capacity in terms of mAh gsulfur-1 and mAh gsulfur+CNT-1. This optimized length was taken into account for subsequent investigations of catholytes. One should note that a precise VACNT height was relatively difficult to produce over a large number of electrodes using this CVD process. As a matter of fact, and due to the small dispersion observed in Figure 2 for such range, it was assumed that considering a range of VACNT height between 80 and 100 µm would be good
enough to ensure best performances (i.e. without any large dispersion of the practical discharge capacity in mAh gsulfur-1), while affording a slight error bar in the VACNT height.
Different viscosities between catholytes also influenced rate capability tests, as displayed in Figure 2b. As previously discussed, TEGDME-based electrolyte gave lower discharge capacity values at low C-rate, as compared with DME. When increasing the C-rate from C/50 up to C/2 (3 mA cm-2), using 90 µm high VACNT, the cell polarization increases for both electrolytes, which directly reflects the relatively poor performances of the Li/S system at high current densities. However, the increase is much more dramatic in case of TEGDMEbased electrolyte, with a polarization of ~ 0.3 V and a capacity of 400 mAh gsulfur-1 at C/2 only, while relatively good performances can be maintained on the contrary for DME-based cell, with a polarization of less than 0.2 V and a discharge capacity of more than 700 mAh gsulfur-1.
Capacity fading of VACNT / Polysulfides / Li cells was also evaluated during prolonged cycling at C/10 (Figure 2c), and the use of DME, in combination of DIOX, was found to be as good as with TEGDME solvent, with a capacity fading being lower than 0.03% per cycle. One should note that coulombic efficiency was below 100% for both electrolytes. Despite the use of LiNO3 additive in the electrolyte, the presence of dissolved polysulfides may induce some shuttle mechanism right from the beginning of cycling, but this problematic is discussed below. Finally, owing to its low viscosity, DME was found to be a solvent of choice to be combined with VACNT structures, providing high practical capacities, even at C/2 (3 mA cm2
) rate, and excellent capacity retention. It is worth noticing that these results are in relative
disagreement with our previous work, which was claiming improved electrochemical performances in TEGDME (or even PEGDME)-based electrolytes.30 In fact, as the
charge/discharge mechanisms strongly depend on the electrolyte composition and on species dissolution phenomena, the Li/S system is quite complex and optimizations should never be done on only one component. On the contrary, all the cell components are strongly interconnected, and research should always aim at finding the best electrolyte/electrodes couples. Through this study, the good compatibility of DME and VACNT is explained by the beneficial effect of low viscosity DME regarding the accessibility of the VACNT structure and its high surface area. While VACNT require low viscosity solvents in order to achieve high active material utilization, non-optimized sulfur composite electrodes require on the contrary the use of such highly solvating solvents (e.g. TEGDME). These results are in agreement with recent publications, pointing out the strong links existing between positive electrode composition/structure and the electrolyte nature/properties.32
To deeper understand the impact of catholyte on the cell performances, especially on capacity retention, the electrolyte composition was modified, either by removal of DIOX co-solvent or thought the increase in LiNO3 additive concentration (Figure 3). DME having already a low viscosity, DIOX co-solvent was removed from the electrolyte composition, and electrochemical performances at C/50 were evaluated with 85 µm high VACNT (Figure 3a). One can see that the charge profile rapidly became noisy, phenomenon which was already attributed to dendrites formation.33 This test allows to confirm the key properties of DIOX as a co-solvent for Li/S batteries, and especially for lithium metal negative electrode. As already proposed by Aurbach et al.,34,35 DIOX is indeed an effective solvent for lithium metal passivation, and its polymerization on the negative electrode may allow to prevent rapid dendrites formation in Li/S system. As a matter of fact, a small portion of DIOX was always used in our electrolyte formulations afterwards, especially for catholytes and even for low viscosity compositions, and DME/DIOX 9/1 (volume ratio) mixture was selected as an
optimized catholyte composition for VACNT structures. The high content of DME in this catholyte composition allowed higher concentrations of polysulfides than the classical 1/1 volume ratio,9,10 since DIOX is not a good solvent for polysulfides.
One key additive to Li/S technology development is LiNO3.34 Effect of LiNO3 was investigated on the present system, by varying its concentration in the catholyte. Figure 3b clearly proves that the increase in additive concentration was beneficial for capacity retention and coulombic efficiency. In fact, after 200 cycles, capacity was maintained above 500 mAh gsulfur-1 with 0.2 M LiNO3, and efficiency at that stage was still above 90%. On the contrary, the discharge capacity and coulombic efficiency fell down to 200 mAh gsulfur-1 and 80% respectively with 0.1 M LiNO3. The deterioration of electrochemical performances may be linked to the gradual consumption of the additive upon cycling. Since fresh lithium is deposited on the surface of the anode at each cycle, LiNO3 reacts with the newly deposited lithium surface to form a SEI. These results indicate the strong impact of lithium metal passivation when aiming at long cycle life Li/S cells. And with the catholyte design, the protection of lithium electrode is even more crucial, as dictating both the cyclability and the performances (capacity and self-discharge) of the Li/S cells. The increase in LiNO3 content in the electrolyte composition seems to be an effective solution to improve the cell behavior within a limited number of cycles.36 It may be interesting to investigate other LiNO3 concentrations, to see if it would be possible to further enhance the electrochemical performances by using higher concentrations for example. However, this electrolyte additive may have a limited effect upon prolonged cycling, as it will eventually get fully consumed at some moment. As a matter of fact, while most of the research currently focuses on the positive electrode side, efforts should be made for the development of efficient electrolyte and lithium metal protection solutions, in order to extend Li/S cell cycle life noticeably.
One of the greater advantages of Li/S batteries is the large gravimetric energy density. However, in order to take advantage of the outstanding properties of sulfur, one needs to use large quantities of this active material to provide high gravimetric energy density of an entire cell. For that purpose, in the case of the present VACNT / catholyte / Li cells, the sulfur concentration in the DME/DIOX-based catholyte was increased from 0.25 M of Li2S6 up to 0.6 M (Figure 4), which corresponds to sulfur loadings between 2.3 and 5.6 mgsulfur cm-2. The other electrolyte components were kept identical, while long VACNT structures were chosen (from 100 µm for 0.25 M Li2S6 up to 150 µm for 0.6 M Li2S6), aiming at providing the system with a large surface area. Indeed, in this last study, the aim was to select VACNT height that would not limit the practical discharge capacity in term of sulfur utilization. As a matter of fact, VACNT were selected with a growing height when increasing the sulfur concentration in the catholyte, not aiming at optimizing the S/CNT ratio in this case, but rather ensuring high enough VACNT surface to be able to ensure the best practical performances of each catholyte (i.e. saturated capacity of Figure 2 independent from VACNT height). The cells were first cycled at low C-rate (C/50), and the current density was calculated based on the lowest sulfur loading (i.e. 0.12 mA cm-2) and applied for all cells. Then the cells were also launched for prolonged cycling at higher regime, 0.6 mA cm-2 (corresponding to C/10 rate for the lowest sulfur loading). Finally, rate capability tests were also performed, from 0.12 up to 3 mA cm-2. The voltage profiles of first complete cycles are shown on Figure 4a. One can clearly see that the initial discharge capacity is dependent on sulfur loading in the cell, and that sulfur utilization is maximal for the lower concentration (1100 mAh gsulfur-1 for 0.25 M concentration). On the contrary, when progressively increasing Li2S6 concentration, sulfur utilization is slightly decreasing (from ~1000 mAh gsulfur-1 down to 900 mAh gsulfur-1 for 0.3 M and 0.5 M Li2S6 respectively). In any cases, practical discharge
capacity remains above 800 mAh gsulfur-1, even with 0.6 M Li2S6 concentration, indicating good performances of VACNT with polysulfide-concentrated electrolytes. Best practical surface capacities were even obtained with both 0.5 M and 0.6 M Li2S6 concentrations, and more than 4.75 mAh cm-2 could be delivered at low current density. Cell polarization was also slightly modified by the increase in Li2S6 concentration, for both charge and discharge steps, especially when reaching 0.6 M concentration. Indeed, polysulfide species are responsible for the increase in electrolyte viscosity, which is then responsible for higher catholyte resistivity. However, these results are promising since they confirm the potential interest of VACNT / catholyte / Li cells towards high energy Li/S cells.
Cells were then launched for prolonged cycling at high current density (0.6 mA cm-2), and results are presented in Figure 4b, referring to discharge capacity in mAh cm-2. Indeed, values in mAh cm-2 is interesting as it allows comparison of the catholytes regarding practical application of Li/S cells. In fact, surface capacity values directly reflect both sulfur utilization and practical loading in the cells. First 50 cycles show remarkable capacity retention, with less than 0.1% loss per cycle. The surface capacity was found to be dependent on the Li2S6 concentration. Firstly, the capacity increases with the sulfur content, ranging from ~1.8 mAh cm-2 for Li2S6 0.25 M, up to 4 mAh cm-2 when reaching Li2S6 0.5 M. Secondly, the 0.5 M polysulfide concentration is an optimum, since the practical capacity then remains constant at ~4 mAh cm-2 for the 0.6 M catholyte too. This can either be explained by the high viscosity of more concentrated electrolyte, which then does not fully access to VACNT structure, and does not offer satisfying sulfur utilization at 0.6 mA cm-2. Another reason can also relate to the non-adequate VACNT height, which may not be high enough for such a content of sulfur. One can think about the possibility to increase the VACNT height even further, to investigate with highly concentrated catholytes. Nonetheless, in this configuration, an optimum was
found between sulfur utilization and loading, which was composed of Li2S6 0.5 M and VACNT of ~150 µm high, leading to a practical surface capacity of about 4 mAh cm-2. With such capacity retention, these results are highly promising for long cycle life and high energy Li/S cells.
Catholytes with varying Li2S6 concentrations were then further compared at different current densities (Figure 4c). Current was progressively increased each 10 cycles from 0.12 up to 3 mA cm-2, then the current was decreased back to 0.12 mA cm-2 for the subsequent 50 cycles. As already mentioned regarding voltage profile, the discharge capacity in mAh gsulfur-1, reflecting sulfur utilization, was found to be dependent on the sulfur concentration in the electrolyte, and higher capacity values were obtained with lower Li2S6 concentrations. In addition, rate capability is also impacted by the increase in polysulfides content, mainly due to different electrolyte viscosities. In fact, at 3 mA cm-2, discharge capacities corresponding to 0.5 and 0.6 M concentrations dramatically decrease as compared to low current densities (≤ 400 mAh gsulfur-1), while less concentrated electrolytes still allow to provide more than 500 mAh gsulfur-1 (even more than 700 mAh gsulfur-1 for both 0.3 and 0.4 M). As a matter of fact, while the increase in polysulfides concentration within electrolyte is a key parameter to allow for high surface capacity values, one should also keep in mind that the cell polarization and rate capability performances will also be impacted, and that an optimum of concentration may exist and need to be found for optimizing the VACNT / catholyte / Li design.
In the best case, 4 mAh cm-2 was maintained for 50 cycles, with less than 0.1% capacity loss per cycle. However, the cells launched at 0.6 mA cm-2 for prolonged cycling showed evidence of the system limitations, as shown in Figure 5a. In fact, after a critical number of cycles, this number depending on the sulfur loading and practical capacity, the performances of the cells
started to degrade and a large capacity fading was observed. First, after about 60 cycles, high capacity cells corresponding to 0.6 M and 0.5 M concentrations revealed a rapid capacity loss, with almost no more capacity after 120 and 200 cycles respectively. Lower polysulfide concentration also showed a capacity fade, but later on upon cycling and in a less pronounced way. This phenomenon was attributed to lithium failure, which was already evidenced to be responsible for cell sudden death.33 In fact, when going towards high surface capacity values, e.g. when increasing polysulfides content in the electrolyte, the lithium stripping/plating process induces more stress on the lithium electrode, and its initial morphology may rapidly turn into a foamy or dendritic formation, as we already evidenced and discussed previously in this study. Thus, an important conclusion of this work is related to the fact that the better the initial electrode performances are, the sooner the cell death will happen with the present cell configuration, which emphasizes the need to work and improve the behavior of highly loaded sulfur electrodes and to tackle high areal capacity cells. In addition, it is also really likely that, using such a catholyte configuration, a slight part of sulfur active material is progressively consumed on the lithium electrode in contact with polysulfides. This progressive loss of sulfur in the SEI of lithium may allow to explain the slight capacity fading which is observed for some cells, especially the one showing high initial sulfur utilization. As a matter of fact, a strategy of lithium protection would make sense when using such a catholyte configuration.
In order to confirm lithium failure in our VACNT / catholyte / Li system (Figure 5b), a cell suffering from long overcharge attributed to lithium dendrites was opened. The lithium electrode was replaced by a new one, and the cell was launched for cycling again. The charge/discharge profiles went back to normal, and no further abnormal charge events were observed. This experiment confirmed that lithium failure was the origin of the cell rapid degradation after a critical number of cycles. This phenomenon was found to be particularly
pronounced in the case of high capacity VACNT / catholyte / Li design. We believe that more research efforts should focus on the issues linked to lithium cyclability, through the development of efficient electrolyte compositions and/or lithium electrode protections. Controlling the degradation of the lithium electrode will be key for next generation of high energy Li/S cells.
Conclusion To conclude, VACNT / catholyte / Li cells were successfully cycled with different catholyte compositions, aiming at improving discharge capacity (sulfur utilization), practical surface capacity (energy stored within the cell), rate capability and cycle life. It was found that DME solvent was a good choice since it allowed for high sulfur utilization, thanks to its low viscosity adapted to the fine porosity of VACNT. DIOX co-solvent and LiNO3 additive were found to be key components for cyclability and cell efficiency, as they provided improved lithium electrode passivation and capacity retention. The increase in Li2S6 concentration in the catholyte composition allowed an efficient increase in practical surface capacity, with more than 4 mAh cm-2 obtained for 50 cycles, and with less than 0.1% capacity loss over that period. However, cell death was observed after a certain number of cycles, influenced by the sulfur loading. In turn, an optimum of the system can be drawn from these results: while high VACNT structures favor high sulfur utilization, the height of VACNT should always be optimized in order to get the best carbon/sulfur ratio. In addition, increasing the sulfur content in the electrolyte allows for high surface capacity values to be obtained, but it may also be detrimental for capacity retention and rate capability of the cell. Thus, intermediate concentration, such as 0.4 M Li2S6, is the best compromise presently. Finally, the issue of lithium metal should not be minimized, and future research should focus on this cell component, its protection or its potential replacement.
Acknowledgements The authors would like to thank “Energies du futur” Carnot Institute for funding. Figure captions
Figure 1. Schematic representation of the Li/S cell design that was considered in this study, composed of VACNT, a catholyte containing Li2S6 active material and Li metal negative electrode, so called VACNT / catholyte / Li cell.24 On right side, an example of VACNT SEM picture before cycling.
Figure 2. a) Discharge capacity as a function of VACNT height in two different catholyte compositions: in TEGDME/DIOX 1 /1 and in DME/DIOX 9/1. Discharge capacity values were measured at RT and C/50 (0.12 mA cm-2), and are given in mAh gsulfur-1 and mAh gVACNT+sulfur-1. Electrochemical performances at RT of VACNT (~ 90 µm) / catholyte / Li cells in these two different catholyte compositions: b) voltage profiles at both C/50 and C/2 (0.12 mA cm-2 and 3 mA cm-2), and c) capacity retention and coulombic efficiency at C/10 (0.6 mA cm-2).
Figure 3. a) Voltage profile of a VACNT / catholyte / Li coin cell at RT and C/50 (0.12 mA cm-2) when containing DME only as a catholyte solvent and with ~ 85 µm high VACNT. b) Capacity retention and coulombic efficiency during cycling at RT and C/5 (1.2 mA cm-2) of VACNT / catholyte / Li cells containing different catholyte compositions: using respectively 0.1 and 0.2 M LiNO3 concentrations with ~ 65 µm high VACNT.
Figure 4. Electrochemical performances at RT of VACNT / catholyte / Li cells containing different Li2S6 concentrations in a DME/DIOX 9/1 catholyte: a) voltage profiles at both 0.12 mA cm-2, b) capacity retention at 0.6 mA cm-2 and c) rate capability tests from C/50 to C/2 rates (0.12 up to 3 mA cm-2).
Figure 5. a) Capacity retention at 0.6 mA cm-2 and RT of VACNT / catholyte / Li cells containing different Li2S6 concentrations. b) Voltage profile of a VACNT / catholyte / Li coin cell at RT and C/100 (0.012 mA cm-2), containing a DME/DIOX based catholyte: due to suspicion of dendrites formation, Li metal electrode was replaced during cycling and the corresponding Li/S cell was launched again for cycling.
References
1. P. G. Bruce, S. A. Freunberger, L. J. Hardwick, J.-M. Tarascon, Nat. Mat., 11, 2012, 19-29 2. B. Dunn, H. Kamath, J.-M. Tarascon, Science, 334, 2011, 928-935 3. M. S. Whittingham, Chem. Rev. 2004, 104, 4271 4. B. Scrosati, J. Hassoun, Y.-K. Sun, Energy Environ. Sci., 4, 2011, 3287 5. K. Amine , R. Kanno , Y. Tzeng , MRS Bull., 39, 2014, 395-401 6. R. Van Noorden, Nature, 507, 2014, 26-28 7. R. D. Rauh, K. M. Abraham, G. F. Pearson, J. K. Surprenant, S. B. Brummer, J. Electrochem. Soc., 126, 1979, 523-527
8. C. Barchasz, J.-C. Leprêtre, F. Alloin, S. Patoux, J. Power sources, 199, 2012, 322 9. J. Brückner, S. Thieme, H. T. Grossmann, S. Dörfler, H. Althues, S. Kaskel, J. Power Sources, 268, 2014, 8287
10. M. Hagen, P. Fanz, J. Tübke, J. Power Sources, 264, 2014, 30-34 11. J. Wang, Y.-S. He, J. Yang, Adv. Mater., 27, 2015, 569-575 12. A. Manthiram, Y. Fu, S.-H. Chung, C. Zu, Y.-S. Su, Chem. Rev., 114, 2014, 11751-11787 13. N. Jayaprakash, J. Shen, S. S. Moganty, A. Corona, L. A. Archer, Angew. Chem. Int. Ed., 50, 2011, 5904 14. W. Li, Q. Zhang, G. Zheng, Z. W. Seh, H. Yao, Y. Cui, Nano Lett., 13, 2013, 5534-5540 15. G. Zheng, Q. Zhang, J. J. Cha, Y. Yang, W. Li, Z. W. Seh, Y. Cui, Nano Lett., 13, 2013, 1265 16. J. Chen, D. Wu, E. Walter, M. Engelhard, P. Bhattacharya, H. Pan, Y. Shao, F. Gao, J. Xiao, J. Liu, Nano Energy, 13, 2015, 267–274 17. G. Zhou, Y. Zhao, A. Manthiram, Adv. Energy Mater., 2015, 1402263 18. C. Liang, N. J. Dudney, J. Y. Howe, Chem. Mater., 21, 2009, 4724 19. R. Demir-Cakan, M. Morcrette, Gangulibabu, A. Guéguen, R. Dedryvère, J.-M. Tarascon, Energy Environ. Sci., 6, 2013, 176 20. Z. Li, L. Yuan, Z. Yi, Y. Sun, Y. Liu, Y. Jiang, Y. Shen, Y. Xin, Z. Zhang, Y. Huang, Adv. Energy Mater., 4, 2014, 1301473 21. R. Elazari, G. Salitra, A. Garsuch, A. Panchenko, D. Aurbach, Adv Mater., 23, 2011, 5641-5644
22. C. Barchasz, F. Mesguich, J. Dijon, J.-C. Leprêtre, S. Patoux, F. Alloin, J. Power Sources, 211, 2012, 19-26 23. S.-H. Chung, A. Manthiram, J. Mater. Chem. A, 1, 2013, 959 24. S. Liatard, K. Benhamouda, A. Fournier, R. Ramos, C. Barchasz, J. Dijon, Chem Comm., 51, 2015, 7749 25. Y. Fu, Y.-S. Su, A. Manthiram, Angew. Chem. Int. Ed., 52, 2013, 6930. 26. M. Hagen, S. Dörfler, P. Fanz, T. Berger, R. Speck, J. Tübke, H. Althues, M.J. Hoffmann, C. Scherr, S. Kaskel, J. Power Sources, 224, 2013, 260-268 27. S. S. Zhang, D. T. Tran, J. Power Sources, 211, 2012, 169-172 28. S. S. Zhang, Electrochim. Acta, 70, 2012, 344 29. J. Brückner, S. Thieme, F. Böttger-Hiller, I. Bauer, H. T. Grossmann, P. Strubel, H. Althues, S. Spange, S. Kaskel, Adv. Funct. Mater., 24, 2014, 1284-1289 30. C. Barchasz, J.-C. Leprêtre, S. Patoux, F. Alloin, Electrochim. Acta, 89, 2013, 737-743 31. M.J. Lacey, F. Jeschull, K. Edström, D. Brandell, Chem. Commun., 49, 2013, 8531 32. M. Barghamadi, A.S. Best, A.I. Bhatt, A.F. Hollenkamp, M. Musameh, R.J. Reesc, T. Rüther, Energy Environ. Sci., 7, 2014, 3902 33. M. Hagen, E. Quiroga-González, S. Dörfler, G. Fahrer, J. Tübke, M.J. Hoffmann, H. Althues, R. Speck, M. Krampfert, S. Kaskel, H. Föll, J. Power Sources, 248, 2014, 1058-1066 34. D. Aurbach, E. Pollak, R. Elazari, G. Salitra, C. S. Kelley, J. Affinito, J. Electrochem. Soc., 156, 2009, A694-A702 35. D. Aurbach, O. Youngman, Y. Gofer, A. Meitav, Electrochim. Acta, 35, 1990, 625-638 36. S.S. Zhang, Electrochim. Acta, 70, 2012, 344-348
S0013-4686(15)30885-9 http://dx.doi.org/doi:10.1016/j.electacta.2015.11.113 EA 26113
To appear in:
Electrochimica Acta
Received date: Revised date: Accepted date:
7-7-2015 13-10-2015 21-11-2015
Please cite this article as: S´ebastien Liatard, Kamal Benhamouda, Adeline Fournier, Jean Dijon, C´eline Barchasz, Influence of catholyte composition on the performances of VACNT/Polysulfides/Li cells, Electrochimica Acta http://dx.doi.org/10.1016/j.electacta.2015.11.113 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Influence of catholyte composition on the performances of VACNT / Polysulfides / Li cells Sébastien Liatarda, Kamal Benhamoudaa, Adeline Fournierb, Jean Dijonb, Céline Barchasza,* a
CEA, LITEN, Department of Electricity and Hydrogen for Transportation, 17 rue des Martyrs, F-38054
Grenoble, France. b
CEA, LITEN, Department of Technologies for NanoMaterials, 17 rue des Martyrs, F-38054 Grenoble, France.
*
Corresponding author
Mail: [email protected] Phone: 0033 438 78 90 36 Fax: 0033 438 78 51 98
Keywords Lithium/sulfur rechargeable batteries Catholyte composition Electrolyte additive Lithium polysulfide concentration Vertically-aligned carbon nanotubes Abstract Elemental sulfur has become a particularly attractive material for lithium metal batteries owing to its low cost, availability, non-toxicity and high specific capacity of sulfur active material. Problematics of the Li/S cell arise from its particular mechanism, based on dissolution/precipitation phenomena of active materials along cycling. The phase transitions endured by active material are responsible for dramatic changes in the morphology of the
positive electrode upon cycling, and results in severe capacity fading. In this work, verticallyaligned carbon nanotubes (VACNT) were investigated as a positive electrode 3D current collector for Li/S cells, combined with a semi-liquid VACNT / Polysulfides (catholyte) / Li battery configuration. The influence of the catholyte composition and the sulfur concentration, and the impact of electrolyte solvents and additives in the catholyte were investigated, yielding significant improvements in terms of practical surface capacity of the cells.
Introduction Lithium-ion batteries have been under intense studies for the past decades, due to the growing need of power for portable electronics, electric cars, or grid energy storage.1,2 For over two decades, Li-ion technologies have revolutionized the portable electronics market due to their unequalled volumetric energy density (650 Wh L-1).3 However, the development of Li-ionpowered electric vehicles has been hindered by the high costs and the low capacity of positive electrode materials (250 mAh g-1), resulting in a limited practical gravimetric energy (maximum of 250 Wh kg-1).4 To go beyond Li-ion batteries performances, today’s research is more and more focused on post-Li-ion systems, such as lithium/sulfur (Li/S) or lithium/oxygen (Li/O2) technologies.1,5 In particular, elemental sulfur has become a particularly attractive material for lithium metal batteries owing to its low cost, availability, non-toxicity and high specific capacity of sulfur active material.
The reaction between elemental sulfur (S8) and metallic lithium involves the exchange of 16 electrons per mole of S8, which allows for high theoretical specific discharge capacity (1675 mAh gsulfur-1) and high theoretical gravimetric energy density (2600 Wh kgsulfur-1). In practical, gravimetric energy densities of 300-600 Wh kgcell-1 are expected, thus making the Li/S cell
fully competitive with Li-ion counterparts.6 The Li/S rechargeable cell has been studied since the 70’s,7 but Li/S technology is still not commercialized. In fact, it disappointingly suffers rapid capacity fading, low practical discharge capacity and low coulombic efficiency.8 Problematics arise from the particular mechanism of the Li/S cell during charge/discharge processes, which are based on dissolution/precipitation phenomena of active materials along cycling, while Li-ion systems rather rely on intercalation/deintercalation reactions. The phase transitions endured by the active material in the battery during charge and discharge are mainly responsible for dramatic changes in the morphology of the positive electrode upon cycling, and results in severe capacity fading during the first cycles.8 Limitations in the Li/S performances are also linked to the strong insulating properties of solid sulfur compounds (i.e. S8 and final discharge products Li2S2/Li2S), which rapidly passivate the positive electrode when formed and lead to incomplete active material utilization. Last but not least, some of the problematics also originate in the use of organic liquid electrolyte and lithium metal negative electrode, which results in electrolyte consumption upon cycling, low coulombic efficiency, sudden cell fading upon prolonged cycling and possible safety issue. As a consequence, more and more attention is paid today to the Li metal electrode, its reliability, its potential replacements or its compatibility with liquid electrolyte.9,10
To address these issues, different approaches are considered in today’s research. A popular strategy involves developing sophisticated cathode architectures, through the preparation of sulfur/carbon composites, sulfur/polymer materials or a mixture of these three, and impressive results have been obtained this way in terms of cycle life, sulfur utilization and rate capability.11,12,13,14 In particular, the confinement approach is widely studied,15,16,17 and intends to prevent sulfur and polysulfides dissolution into the electrolyte upon cycling. However, capacity fading is still frequently observed, since polysulfides are likely to diffuse into the
electrolyte in the case of prolonged cycling, thus indicating that the confinement of sulfur may never be complete.18,19,20
Among alternative solutions for the positive electrode side, the interest of 3D current collectors has already been demonstrated as an efficient concept to replace classical 2D aluminum foil.21,22,23 In particular, we already demonstrated the beneficial effect of using vertically-aligned carbon nanotubes (VACNT) structure as a positive electrode current collector for Li/S cells, both in terms of practical discharge capacity and retention, and when employed in combination with a catholyte approach.22,24 First proposed in the early days of Li/S developments, the concept of catholyte recently gained increasing attention.7,25 In this approach, the electrolyte contains the active material in the form of dissolved polysulfides, and can be associated with a 3D current collector as a positive electrode. For this purpose, metal foams or carbon papers are promising substrates for enhancing both discharge capacity and capacity retention of Li/S cells,22,26,27 through the stabilization of positive electrode morphology during cycling. In previous work, we described a light-weight, high surface current collector made of VACNT grown on an aluminum foil, which was successfully used in a semi-liquid VACNT / Polysulfides / Li battery configuration (Figure 1). In this study, we investigate the influence of the catholyte composition on the performances of these semiliquid cells. The impact of electrolyte solvents on the discharge capacity was clearly evidenced, while capacity retention was found to be strongly correlated with the electrolyte additives. Finally, sulfur concentration in the catholyte was increased, yielding significant improvements in terms of practical surface capacity of the cells. Beneficial effects of VACNT and optimization of the catholyte were demonstrated, while the limitations coming from other cell components were also evidenced.
Experimental section Fabrication of VACNTs on aluminum: Vertically-aligned carbon nanotubes were synthesized by chemical vapor deposition technique on a 20 µm-thick aluminum substrate (Sumikei Al, Japan).24 Al foil was cleaned thoroughly by sonication in acetone and isopropanol baths, and by a plasma treatment (O2 0.1 mbar, 100 W, 2 min). A nano-layer of iron catalyst was deposited on Al using an electron beam physical vapor deposition (EBPVD) technique. Growth of VACNT was performed in a low pressure (~ 1 mbar) CVD chamber equipped with an array of 10 conductive filaments connected to a power source. Acetylene (C2H2, 20 sccm) was introduced in the reactor along with hydrogen (H2, 50 sccm) and helium (110 sccm). The chamber was heated up to 450 °C, and the temperature was sustained for 30 min. The length of VACNT could be tuned from few microns up to 150 µm, by modifying the residence time of samples into the CVD chamber. Images of the electrodes structure were realized within a Zeiss LEO 1530 scanning electron microscope. Catholyte preparation: Sulfur was introduced into the batteries in the form of polysulfides (average formula Li2S6) dissolved in the electrolyte, this system being called ‘catholyte’. In an argon-filled glove box, metallic lithium and sulfur powder (99.5%, 325 mesh, Alfa Aesar) were precisely weighted and introduced in the electrolyte solvent or mixture of solvents. Stoichiometric amounts of lithium and sulfur powders were weighted in order to obtain an average composition of Li2S6, and the dispersions were let reacting at 50°C over few days to ensure complete reaction between powders. Different catholyte compositions were prepared, based on the use of 1,3-dioxolane (Diox, anhydrous, 99,8%, Aldrich), 1,2-dimethoxyethane (DME, anhydrous, 99.5%, Aldrich) or tetraethylene glycol dimethyl ether (TEGDME, 99%, Adrich). TEGDME/DIOX mixture was used with a volume ratio of 1/1. DME was employed as a pure solvent while also in a mixture with DIOX (9/1 vol. ratio of DME/DIOX). The concentration of Li2S6 was varied between 0.25 M to 0.6 M in order to adjust the sulfur
loading, by dissolving the proper amounts of sulfur and metallic lithium materials. Lithium bis(trifluoromethane sulfone)imide (LiTFSI, 99.95%, Aldrich) and lithium nitrate (LiNO3, 99.99%, Aldrich) were dissolved in the catholyte at 1 M and 0.1 M respectively. For comparison purpose, the concentration of LiNO3 was also increased to 0.2 M, in order to investigate its effect on cell behavior.28 At least 48h of stirring at 50°C were necessary to dissolve all solid compounds and to form the catholytes. Electrochemical characterization: VACNT structures on Al were punched to 14 mm diameter electrodes and then used directly as positive electrodes in CR2032 coin cells. No sulfur deposition was performed on this current collector prior to cell assembly, since the positive active material was introduced within the electrolyte. The VACNT / Polysulfides / Li coin cells were assembled in an argon-filled glove box and a schematic representation of the cell design is presented in Figure 1. Coin cells were galvanostatically cycled between 3 V vs. Li+/Li and 1.8 V (or 1.5 V depending on the applied C-rate), either at constant current, or at a varying current density for rate capability tests (from C/50 up to C/2). A disk of lithium metal (16 mm) was used as a negative electrode. A Celgard 2400® served as the separator, assisted by a non-woven Viledon® layer to provide an electrolyte reservoir. 75 µL of catholyte was introduced in the coin cells, thus leading to theoretically loadings from 3.9 to 9.4 mAh cm-2 (from 2.3 to 5.6 mgsulfur cm-2).
Results and Discussion Figure 1 shows a schematic representation of the cell design that was considered in this study, as well as a SEM picture of a VACNT structure before cycling. In this configuration, VACNT serves both as a current collector and a positive electrode, while active material is initially dissolved in the electrolyte. Electrodeposition of sulfur (or Li2S) material is performed in situ in the battery by initial charging (or discharging) of the Li/S cell. Lithium metal was used as a
negative electrode in this study, but VACNT can also be combined with other negative electrode materials such as silicon or hard carbon,29 as the catholyte also contains a source of lithium ions to first charge the battery. VACNT structures produced by CVD on Al are composed of multi-walled (mainly double-walled) carbon nanotubes, with a mean diameter of 3.3 nm, that are vertically-aligned on the Al foil and organized into bundles of about 15-35 nm.24 VACNT could be produced with tube length ranging from few microns up to 150 µm. The performances of these structures were evaluated as a function of VACNT height for two different catholyte compositions: 1) Li2S6 0.25 M + LiTFSI 1 M + LiNO3 0.1 M + TEGDME/DIOX 1/1 and 2) Li2S6 0.25 M + LiTFSI 1 M + LiNO3 0.1 M + DME/DIOX 9/1. The two compositions were selected owing to their different solvation properties, their different viscosity, the first one being used as a reference in our studies (mainly with sulfur composite electrodes),8 while the second one being usually used as a reference in literature.9,10 Indeed, TEGDME has been reported for its good affinity with lithium polysulfides,30,31 but suffers a high viscosity (~4 cP). On the contrary, DME has a lower affinity with polysulfides,30 but also a much lower viscosity (0.455 cP), thus being the most common solvent in Li/S batteries. The electrochemical performances of the two catholyte compositions were compared in terms of sulfur utilization, rate capability and cyclability, with varying VACNT lengths, and the results are presented in Figure 2. Firstly, practical discharge capacity values at low C-rate (C/50, 0.12 mA cm-2) were always higher for DME-based cells whatever the VACNT length (Figure 2a), ranging from ~400 mAh gsulfur-1 for short VACNT up to about 1100 mAh gsulfur-1 for longer ones. This indicates that sulfur utilization in the DME-based catholyte is higher than in the TEGDME-based one. Maximum discharge capacity with increasing VACNT length was found to stabilize at 1100 mAh gsulfur-1 for DME-based cells, while stabilizing at 800 mAh gsulfur-1 in TEGDME only. The differences in sulfur utilization could be explained by the different electrolyte viscosities: DME solvent may allow better
electrolyte penetration into the VACNT structure thanks to its low viscosity, thus leading to better wettability of the positive electrode, and larger available conductive surface for higher sulfur utilization. In turn, the maximum capacity value, obtained when increasing VACNT height, is higher in case of DME, pointing out that this ‘saturated’ capacity strongly depends on the catholyte composition. On the contrary, it is really likely that using TEGDME, the accessibility of VACNT is much lower, and we can propose that discharge products mainly precipitate on the positive electrode surface than in the VACNT porosity, which strongly impacts the discharge capacity. This hypothesis is currently under investigation. In any case, it is worth noticing that both positive electrode and electrolyte compositions are strongly interconnected to each other, and both components make an inherent pair of materials that need to be optimized jointly. Thanks to the low weight of VACNT structures as compare to sulfur weight, the discharge capacity related to the masses of both S8 and VACNT was relatively high, of ~ 900 mAh gsulfur+CNT-1 even with VACNT as high as 90 µm. Then, due to the saturation of discharge capacity at about 1100 mAh gsulfur-1, the value referring to both S8 and VACNT masses is then progressively impacted by further increase of the VACNT height, pointing out the need of optimizing sulfur-to-carbon ratio. Thus, DME-based catholyte was found to provide high practical capacity at low C-rate in combination with VACNT structures, and the optimum VACNT height was found to be ~90 µm, resulting in a maximum practical capacity in terms of mAh gsulfur-1 and mAh gsulfur+CNT-1. This optimized length was taken into account for subsequent investigations of catholytes. One should note that a precise VACNT height was relatively difficult to produce over a large number of electrodes using this CVD process. As a matter of fact, and due to the small dispersion observed in Figure 2 for such range, it was assumed that considering a range of VACNT height between 80 and 100 µm would be good
enough to ensure best performances (i.e. without any large dispersion of the practical discharge capacity in mAh gsulfur-1), while affording a slight error bar in the VACNT height.
Different viscosities between catholytes also influenced rate capability tests, as displayed in Figure 2b. As previously discussed, TEGDME-based electrolyte gave lower discharge capacity values at low C-rate, as compared with DME. When increasing the C-rate from C/50 up to C/2 (3 mA cm-2), using 90 µm high VACNT, the cell polarization increases for both electrolytes, which directly reflects the relatively poor performances of the Li/S system at high current densities. However, the increase is much more dramatic in case of TEGDMEbased electrolyte, with a polarization of ~ 0.3 V and a capacity of 400 mAh gsulfur-1 at C/2 only, while relatively good performances can be maintained on the contrary for DME-based cell, with a polarization of less than 0.2 V and a discharge capacity of more than 700 mAh gsulfur-1.
Capacity fading of VACNT / Polysulfides / Li cells was also evaluated during prolonged cycling at C/10 (Figure 2c), and the use of DME, in combination of DIOX, was found to be as good as with TEGDME solvent, with a capacity fading being lower than 0.03% per cycle. One should note that coulombic efficiency was below 100% for both electrolytes. Despite the use of LiNO3 additive in the electrolyte, the presence of dissolved polysulfides may induce some shuttle mechanism right from the beginning of cycling, but this problematic is discussed below. Finally, owing to its low viscosity, DME was found to be a solvent of choice to be combined with VACNT structures, providing high practical capacities, even at C/2 (3 mA cm2
) rate, and excellent capacity retention. It is worth noticing that these results are in relative
disagreement with our previous work, which was claiming improved electrochemical performances in TEGDME (or even PEGDME)-based electrolytes.30 In fact, as the
charge/discharge mechanisms strongly depend on the electrolyte composition and on species dissolution phenomena, the Li/S system is quite complex and optimizations should never be done on only one component. On the contrary, all the cell components are strongly interconnected, and research should always aim at finding the best electrolyte/electrodes couples. Through this study, the good compatibility of DME and VACNT is explained by the beneficial effect of low viscosity DME regarding the accessibility of the VACNT structure and its high surface area. While VACNT require low viscosity solvents in order to achieve high active material utilization, non-optimized sulfur composite electrodes require on the contrary the use of such highly solvating solvents (e.g. TEGDME). These results are in agreement with recent publications, pointing out the strong links existing between positive electrode composition/structure and the electrolyte nature/properties.32
To deeper understand the impact of catholyte on the cell performances, especially on capacity retention, the electrolyte composition was modified, either by removal of DIOX co-solvent or thought the increase in LiNO3 additive concentration (Figure 3). DME having already a low viscosity, DIOX co-solvent was removed from the electrolyte composition, and electrochemical performances at C/50 were evaluated with 85 µm high VACNT (Figure 3a). One can see that the charge profile rapidly became noisy, phenomenon which was already attributed to dendrites formation.33 This test allows to confirm the key properties of DIOX as a co-solvent for Li/S batteries, and especially for lithium metal negative electrode. As already proposed by Aurbach et al.,34,35 DIOX is indeed an effective solvent for lithium metal passivation, and its polymerization on the negative electrode may allow to prevent rapid dendrites formation in Li/S system. As a matter of fact, a small portion of DIOX was always used in our electrolyte formulations afterwards, especially for catholytes and even for low viscosity compositions, and DME/DIOX 9/1 (volume ratio) mixture was selected as an
optimized catholyte composition for VACNT structures. The high content of DME in this catholyte composition allowed higher concentrations of polysulfides than the classical 1/1 volume ratio,9,10 since DIOX is not a good solvent for polysulfides.
One key additive to Li/S technology development is LiNO3.34 Effect of LiNO3 was investigated on the present system, by varying its concentration in the catholyte. Figure 3b clearly proves that the increase in additive concentration was beneficial for capacity retention and coulombic efficiency. In fact, after 200 cycles, capacity was maintained above 500 mAh gsulfur-1 with 0.2 M LiNO3, and efficiency at that stage was still above 90%. On the contrary, the discharge capacity and coulombic efficiency fell down to 200 mAh gsulfur-1 and 80% respectively with 0.1 M LiNO3. The deterioration of electrochemical performances may be linked to the gradual consumption of the additive upon cycling. Since fresh lithium is deposited on the surface of the anode at each cycle, LiNO3 reacts with the newly deposited lithium surface to form a SEI. These results indicate the strong impact of lithium metal passivation when aiming at long cycle life Li/S cells. And with the catholyte design, the protection of lithium electrode is even more crucial, as dictating both the cyclability and the performances (capacity and self-discharge) of the Li/S cells. The increase in LiNO3 content in the electrolyte composition seems to be an effective solution to improve the cell behavior within a limited number of cycles.36 It may be interesting to investigate other LiNO3 concentrations, to see if it would be possible to further enhance the electrochemical performances by using higher concentrations for example. However, this electrolyte additive may have a limited effect upon prolonged cycling, as it will eventually get fully consumed at some moment. As a matter of fact, while most of the research currently focuses on the positive electrode side, efforts should be made for the development of efficient electrolyte and lithium metal protection solutions, in order to extend Li/S cell cycle life noticeably.
One of the greater advantages of Li/S batteries is the large gravimetric energy density. However, in order to take advantage of the outstanding properties of sulfur, one needs to use large quantities of this active material to provide high gravimetric energy density of an entire cell. For that purpose, in the case of the present VACNT / catholyte / Li cells, the sulfur concentration in the DME/DIOX-based catholyte was increased from 0.25 M of Li2S6 up to 0.6 M (Figure 4), which corresponds to sulfur loadings between 2.3 and 5.6 mgsulfur cm-2. The other electrolyte components were kept identical, while long VACNT structures were chosen (from 100 µm for 0.25 M Li2S6 up to 150 µm for 0.6 M Li2S6), aiming at providing the system with a large surface area. Indeed, in this last study, the aim was to select VACNT height that would not limit the practical discharge capacity in term of sulfur utilization. As a matter of fact, VACNT were selected with a growing height when increasing the sulfur concentration in the catholyte, not aiming at optimizing the S/CNT ratio in this case, but rather ensuring high enough VACNT surface to be able to ensure the best practical performances of each catholyte (i.e. saturated capacity of Figure 2 independent from VACNT height). The cells were first cycled at low C-rate (C/50), and the current density was calculated based on the lowest sulfur loading (i.e. 0.12 mA cm-2) and applied for all cells. Then the cells were also launched for prolonged cycling at higher regime, 0.6 mA cm-2 (corresponding to C/10 rate for the lowest sulfur loading). Finally, rate capability tests were also performed, from 0.12 up to 3 mA cm-2. The voltage profiles of first complete cycles are shown on Figure 4a. One can clearly see that the initial discharge capacity is dependent on sulfur loading in the cell, and that sulfur utilization is maximal for the lower concentration (1100 mAh gsulfur-1 for 0.25 M concentration). On the contrary, when progressively increasing Li2S6 concentration, sulfur utilization is slightly decreasing (from ~1000 mAh gsulfur-1 down to 900 mAh gsulfur-1 for 0.3 M and 0.5 M Li2S6 respectively). In any cases, practical discharge
capacity remains above 800 mAh gsulfur-1, even with 0.6 M Li2S6 concentration, indicating good performances of VACNT with polysulfide-concentrated electrolytes. Best practical surface capacities were even obtained with both 0.5 M and 0.6 M Li2S6 concentrations, and more than 4.75 mAh cm-2 could be delivered at low current density. Cell polarization was also slightly modified by the increase in Li2S6 concentration, for both charge and discharge steps, especially when reaching 0.6 M concentration. Indeed, polysulfide species are responsible for the increase in electrolyte viscosity, which is then responsible for higher catholyte resistivity. However, these results are promising since they confirm the potential interest of VACNT / catholyte / Li cells towards high energy Li/S cells.
Cells were then launched for prolonged cycling at high current density (0.6 mA cm-2), and results are presented in Figure 4b, referring to discharge capacity in mAh cm-2. Indeed, values in mAh cm-2 is interesting as it allows comparison of the catholytes regarding practical application of Li/S cells. In fact, surface capacity values directly reflect both sulfur utilization and practical loading in the cells. First 50 cycles show remarkable capacity retention, with less than 0.1% loss per cycle. The surface capacity was found to be dependent on the Li2S6 concentration. Firstly, the capacity increases with the sulfur content, ranging from ~1.8 mAh cm-2 for Li2S6 0.25 M, up to 4 mAh cm-2 when reaching Li2S6 0.5 M. Secondly, the 0.5 M polysulfide concentration is an optimum, since the practical capacity then remains constant at ~4 mAh cm-2 for the 0.6 M catholyte too. This can either be explained by the high viscosity of more concentrated electrolyte, which then does not fully access to VACNT structure, and does not offer satisfying sulfur utilization at 0.6 mA cm-2. Another reason can also relate to the non-adequate VACNT height, which may not be high enough for such a content of sulfur. One can think about the possibility to increase the VACNT height even further, to investigate with highly concentrated catholytes. Nonetheless, in this configuration, an optimum was
found between sulfur utilization and loading, which was composed of Li2S6 0.5 M and VACNT of ~150 µm high, leading to a practical surface capacity of about 4 mAh cm-2. With such capacity retention, these results are highly promising for long cycle life and high energy Li/S cells.
Catholytes with varying Li2S6 concentrations were then further compared at different current densities (Figure 4c). Current was progressively increased each 10 cycles from 0.12 up to 3 mA cm-2, then the current was decreased back to 0.12 mA cm-2 for the subsequent 50 cycles. As already mentioned regarding voltage profile, the discharge capacity in mAh gsulfur-1, reflecting sulfur utilization, was found to be dependent on the sulfur concentration in the electrolyte, and higher capacity values were obtained with lower Li2S6 concentrations. In addition, rate capability is also impacted by the increase in polysulfides content, mainly due to different electrolyte viscosities. In fact, at 3 mA cm-2, discharge capacities corresponding to 0.5 and 0.6 M concentrations dramatically decrease as compared to low current densities (≤ 400 mAh gsulfur-1), while less concentrated electrolytes still allow to provide more than 500 mAh gsulfur-1 (even more than 700 mAh gsulfur-1 for both 0.3 and 0.4 M). As a matter of fact, while the increase in polysulfides concentration within electrolyte is a key parameter to allow for high surface capacity values, one should also keep in mind that the cell polarization and rate capability performances will also be impacted, and that an optimum of concentration may exist and need to be found for optimizing the VACNT / catholyte / Li design.
In the best case, 4 mAh cm-2 was maintained for 50 cycles, with less than 0.1% capacity loss per cycle. However, the cells launched at 0.6 mA cm-2 for prolonged cycling showed evidence of the system limitations, as shown in Figure 5a. In fact, after a critical number of cycles, this number depending on the sulfur loading and practical capacity, the performances of the cells
started to degrade and a large capacity fading was observed. First, after about 60 cycles, high capacity cells corresponding to 0.6 M and 0.5 M concentrations revealed a rapid capacity loss, with almost no more capacity after 120 and 200 cycles respectively. Lower polysulfide concentration also showed a capacity fade, but later on upon cycling and in a less pronounced way. This phenomenon was attributed to lithium failure, which was already evidenced to be responsible for cell sudden death.33 In fact, when going towards high surface capacity values, e.g. when increasing polysulfides content in the electrolyte, the lithium stripping/plating process induces more stress on the lithium electrode, and its initial morphology may rapidly turn into a foamy or dendritic formation, as we already evidenced and discussed previously in this study. Thus, an important conclusion of this work is related to the fact that the better the initial electrode performances are, the sooner the cell death will happen with the present cell configuration, which emphasizes the need to work and improve the behavior of highly loaded sulfur electrodes and to tackle high areal capacity cells. In addition, it is also really likely that, using such a catholyte configuration, a slight part of sulfur active material is progressively consumed on the lithium electrode in contact with polysulfides. This progressive loss of sulfur in the SEI of lithium may allow to explain the slight capacity fading which is observed for some cells, especially the one showing high initial sulfur utilization. As a matter of fact, a strategy of lithium protection would make sense when using such a catholyte configuration.
In order to confirm lithium failure in our VACNT / catholyte / Li system (Figure 5b), a cell suffering from long overcharge attributed to lithium dendrites was opened. The lithium electrode was replaced by a new one, and the cell was launched for cycling again. The charge/discharge profiles went back to normal, and no further abnormal charge events were observed. This experiment confirmed that lithium failure was the origin of the cell rapid degradation after a critical number of cycles. This phenomenon was found to be particularly
pronounced in the case of high capacity VACNT / catholyte / Li design. We believe that more research efforts should focus on the issues linked to lithium cyclability, through the development of efficient electrolyte compositions and/or lithium electrode protections. Controlling the degradation of the lithium electrode will be key for next generation of high energy Li/S cells.
Conclusion To conclude, VACNT / catholyte / Li cells were successfully cycled with different catholyte compositions, aiming at improving discharge capacity (sulfur utilization), practical surface capacity (energy stored within the cell), rate capability and cycle life. It was found that DME solvent was a good choice since it allowed for high sulfur utilization, thanks to its low viscosity adapted to the fine porosity of VACNT. DIOX co-solvent and LiNO3 additive were found to be key components for cyclability and cell efficiency, as they provided improved lithium electrode passivation and capacity retention. The increase in Li2S6 concentration in the catholyte composition allowed an efficient increase in practical surface capacity, with more than 4 mAh cm-2 obtained for 50 cycles, and with less than 0.1% capacity loss over that period. However, cell death was observed after a certain number of cycles, influenced by the sulfur loading. In turn, an optimum of the system can be drawn from these results: while high VACNT structures favor high sulfur utilization, the height of VACNT should always be optimized in order to get the best carbon/sulfur ratio. In addition, increasing the sulfur content in the electrolyte allows for high surface capacity values to be obtained, but it may also be detrimental for capacity retention and rate capability of the cell. Thus, intermediate concentration, such as 0.4 M Li2S6, is the best compromise presently. Finally, the issue of lithium metal should not be minimized, and future research should focus on this cell component, its protection or its potential replacement.
Acknowledgements The authors would like to thank “Energies du futur” Carnot Institute for funding. Figure captions
Figure 1. Schematic representation of the Li/S cell design that was considered in this study, composed of VACNT, a catholyte containing Li2S6 active material and Li metal negative electrode, so called VACNT / catholyte / Li cell.24 On right side, an example of VACNT SEM picture before cycling.
Figure 2. a) Discharge capacity as a function of VACNT height in two different catholyte compositions: in TEGDME/DIOX 1 /1 and in DME/DIOX 9/1. Discharge capacity values were measured at RT and C/50 (0.12 mA cm-2), and are given in mAh gsulfur-1 and mAh gVACNT+sulfur-1. Electrochemical performances at RT of VACNT (~ 90 µm) / catholyte / Li cells in these two different catholyte compositions: b) voltage profiles at both C/50 and C/2 (0.12 mA cm-2 and 3 mA cm-2), and c) capacity retention and coulombic efficiency at C/10 (0.6 mA cm-2).
Figure 3. a) Voltage profile of a VACNT / catholyte / Li coin cell at RT and C/50 (0.12 mA cm-2) when containing DME only as a catholyte solvent and with ~ 85 µm high VACNT. b) Capacity retention and coulombic efficiency during cycling at RT and C/5 (1.2 mA cm-2) of VACNT / catholyte / Li cells containing different catholyte compositions: using respectively 0.1 and 0.2 M LiNO3 concentrations with ~ 65 µm high VACNT.
Figure 4. Electrochemical performances at RT of VACNT / catholyte / Li cells containing different Li2S6 concentrations in a DME/DIOX 9/1 catholyte: a) voltage profiles at both 0.12 mA cm-2, b) capacity retention at 0.6 mA cm-2 and c) rate capability tests from C/50 to C/2 rates (0.12 up to 3 mA cm-2).
Figure 5. a) Capacity retention at 0.6 mA cm-2 and RT of VACNT / catholyte / Li cells containing different Li2S6 concentrations. b) Voltage profile of a VACNT / catholyte / Li coin cell at RT and C/100 (0.012 mA cm-2), containing a DME/DIOX based catholyte: due to suspicion of dendrites formation, Li metal electrode was replaced during cycling and the corresponding Li/S cell was launched again for cycling.
References
1. P. G. Bruce, S. A. Freunberger, L. J. Hardwick, J.-M. Tarascon, Nat. Mat., 11, 2012, 19-29 2. B. Dunn, H. Kamath, J.-M. Tarascon, Science, 334, 2011, 928-935 3. M. S. Whittingham, Chem. Rev. 2004, 104, 4271 4. B. Scrosati, J. Hassoun, Y.-K. Sun, Energy Environ. Sci., 4, 2011, 3287 5. K. Amine , R. Kanno , Y. Tzeng , MRS Bull., 39, 2014, 395-401 6. R. Van Noorden, Nature, 507, 2014, 26-28 7. R. D. Rauh, K. M. Abraham, G. F. Pearson, J. K. Surprenant, S. B. Brummer, J. Electrochem. Soc., 126, 1979, 523-527
8. C. Barchasz, J.-C. Leprêtre, F. Alloin, S. Patoux, J. Power sources, 199, 2012, 322 9. J. Brückner, S. Thieme, H. T. Grossmann, S. Dörfler, H. Althues, S. Kaskel, J. Power Sources, 268, 2014, 8287
10. M. Hagen, P. Fanz, J. Tübke, J. Power Sources, 264, 2014, 30-34 11. J. Wang, Y.-S. He, J. Yang, Adv. Mater., 27, 2015, 569-575 12. A. Manthiram, Y. Fu, S.-H. Chung, C. Zu, Y.-S. Su, Chem. Rev., 114, 2014, 11751-11787 13. N. Jayaprakash, J. Shen, S. S. Moganty, A. Corona, L. A. Archer, Angew. Chem. Int. Ed., 50, 2011, 5904 14. W. Li, Q. Zhang, G. Zheng, Z. W. Seh, H. Yao, Y. Cui, Nano Lett., 13, 2013, 5534-5540 15. G. Zheng, Q. Zhang, J. J. Cha, Y. Yang, W. Li, Z. W. Seh, Y. Cui, Nano Lett., 13, 2013, 1265 16. J. Chen, D. Wu, E. Walter, M. Engelhard, P. Bhattacharya, H. Pan, Y. Shao, F. Gao, J. Xiao, J. Liu, Nano Energy, 13, 2015, 267–274 17. G. Zhou, Y. Zhao, A. Manthiram, Adv. Energy Mater., 2015, 1402263 18. C. Liang, N. J. Dudney, J. Y. Howe, Chem. Mater., 21, 2009, 4724 19. R. Demir-Cakan, M. Morcrette, Gangulibabu, A. Guéguen, R. Dedryvère, J.-M. Tarascon, Energy Environ. Sci., 6, 2013, 176 20. Z. Li, L. Yuan, Z. Yi, Y. Sun, Y. Liu, Y. Jiang, Y. Shen, Y. Xin, Z. Zhang, Y. Huang, Adv. Energy Mater., 4, 2014, 1301473 21. R. Elazari, G. Salitra, A. Garsuch, A. Panchenko, D. Aurbach, Adv Mater., 23, 2011, 5641-5644
22. C. Barchasz, F. Mesguich, J. Dijon, J.-C. Leprêtre, S. Patoux, F. Alloin, J. Power Sources, 211, 2012, 19-26 23. S.-H. Chung, A. Manthiram, J. Mater. Chem. A, 1, 2013, 959 24. S. Liatard, K. Benhamouda, A. Fournier, R. Ramos, C. Barchasz, J. Dijon, Chem Comm., 51, 2015, 7749 25. Y. Fu, Y.-S. Su, A. Manthiram, Angew. Chem. Int. Ed., 52, 2013, 6930. 26. M. Hagen, S. Dörfler, P. Fanz, T. Berger, R. Speck, J. Tübke, H. Althues, M.J. Hoffmann, C. Scherr, S. Kaskel, J. Power Sources, 224, 2013, 260-268 27. S. S. Zhang, D. T. Tran, J. Power Sources, 211, 2012, 169-172 28. S. S. Zhang, Electrochim. Acta, 70, 2012, 344 29. J. Brückner, S. Thieme, F. Böttger-Hiller, I. Bauer, H. T. Grossmann, P. Strubel, H. Althues, S. Spange, S. Kaskel, Adv. Funct. Mater., 24, 2014, 1284-1289 30. C. Barchasz, J.-C. Leprêtre, S. Patoux, F. Alloin, Electrochim. Acta, 89, 2013, 737-743 31. M.J. Lacey, F. Jeschull, K. Edström, D. Brandell, Chem. Commun., 49, 2013, 8531 32. M. Barghamadi, A.S. Best, A.I. Bhatt, A.F. Hollenkamp, M. Musameh, R.J. Reesc, T. Rüther, Energy Environ. Sci., 7, 2014, 3902 33. M. Hagen, E. Quiroga-González, S. Dörfler, G. Fahrer, J. Tübke, M.J. Hoffmann, H. Althues, R. Speck, M. Krampfert, S. Kaskel, H. Föll, J. Power Sources, 248, 2014, 1058-1066 34. D. Aurbach, E. Pollak, R. Elazari, G. Salitra, C. S. Kelley, J. Affinito, J. Electrochem. Soc., 156, 2009, A694-A702 35. D. Aurbach, O. Youngman, Y. Gofer, A. Meitav, Electrochim. Acta, 35, 1990, 625-638 36. S.S. Zhang, Electrochim. Acta, 70, 2012, 344-348