Sulfur Cell Discharge Mechanism: An Original Approach for

Article pubs.acs.org/ac

Lithium/Sulfur Cell Discharge Mechanism: An Original Approach for Intermediate Species Identification Céline Barchasz,*,†,‡ Florian Molton,§ Carole Duboc,§ Jean-Claude Leprêtre,‡ Sébastien Patoux,† and Fannie Alloin‡ †

French Atomic Energy and Alternative Energy Agency (CEA), Laboratory of Innovation for New Energy Technologies and Nanomaterials (LITEN), 17 Rue des Martyrs, 38054 Grenoble Cedex 9, France ‡ Laboratoire d’Electrochimie et Physicochimie des Matériaux et Interfaces (LEPMI), UMR 5279, CNRS, Grenoble INP, Université de Savoie, Université Joseph Fourier, 1130 Rue de la Piscine, BP75, 38402 Saint Martin d’Hères, France § Département de Chimie Moléculaire (DCM) UMR 5250, CNRS, Université Joseph Fourier, 301, Rue de la Chimie, BP53, 38041 Grenoble Cedex 9, France S Supporting Information *

ABSTRACT: The lithium/sulfur battery is a promising electrochemical system that has a high theoretical capacity of 1675 mAh g−1, but its discharge mechanism is well-known to be a complex multistep process. As the active material dissolves during cycling, this discharge mechanism was investigated through the electrolyte characterization. Using high-performance liquid chromatography, UV−visible absorption, and electron spin resonance spectroscopies, we investigated the electrolyte composition at different discharge potentials in a TEGDME-based electrolyte. In this study, we propose a possible mechanism for sulfur reduction consisting of three steps. Long polysulfide chains are produced during the first reduction step (2.4−2.2 V vs Li+/Li), such as S82− and S62−, as evidenced by UV and HPLC data. The S3•− radical can also be found in solution because of a disproportionation reaction. S42− is produced during the second reduction step (2.15−2.1 V vs Li+/Li), thus pointing out the gradual decrease of the polysulfide chain lengths. Finally, short polysulfide species, such as S32−, S22−, and S2−, are produced at the end of the reduction process, i.e., between 2.1 and 1.9 V vs Li+/Li. The precipitation of the poorly soluble and insulating short polysulfide compounds was evidenced, thus leading to the positive electrode passivation and explaining the early end of discharge.

R

echargeable lithium-ion batteries have been under intense research over the past 20 years because of the increasing energy consumption of portable devices. Lithium transition metal oxides, especially LiCoO2 and its counterparts, are currently dominating the commercial lithium-ion battery market, due to their advantages such as high energy density, high operating voltage, and low self-discharge rate.1 However, the gravimetric energy density of lithium-ion batteries is known to be limited to 200−250 Wh kg−1, which is not sufficient to meet the electric vehicles battery requirements for extended ranges. Moreover, cobalt is toxic and expensive, and layered oxides can have safety issues.2 The lithium/sulfur (Li/S) battery is a promising electrochemical system that has a high theoretical capacity of 1675 mAh g−1, thus attracting the attention of electrochemistry community for many years and much greater than the 100−250 mAh g−1 capacity attainable with the conventional lithium-ion positive electrode materials.3 The discharge potential is around 2.1 V, and the complete Li/S system should allow a gravimetric energy density of 500 Wh kg−1 to be reached. In addition, elemental sulfur is readily available and nontoxic, advantages that should allow cheap and safe high-energy batteries to be produced.4 © 2012 American Chemical Society

Sulfur reduction is a multistep electrochemical process that can involve different intermediate species.5,6 Lithium metal reacts with elemental sulfur (S 8 ) to produce lithium polysulfides with a general formula Li2Sn. It is well-known that long polysulfides chain lengths are first produced, such as Li2S8 and Li2S6. During discharge, the polysulfide chain length is shortened as the sulfur is being further reduced. At the end of the discharge, the final product is lithium sulfide (Li2S), and the overall reaction is7

16Li + S8 → 8Li 2S

(1) 4,8,9

Despite 3 decades of research, the discharge mechanism of the Li/S rechargeable cell is still controversial. As many polysulfides anions or radicals are known to exist in solution, different sulfur reduction mechanisms could be found in the literature,7,10−15 and the authors do not agree about the intermediate species that might be formed during the electrochemical process. Some publications discuss quite simple Received: December 7, 2011 Accepted: April 8, 2012 Published: April 9, 2012 3973

dx.doi.org/10.1021/ac2032244 | Anal. Chem. 2012, 84, 3973−3980

Analytical Chemistry

Article

Figure 1. Galvanostatic potential profile ((a) S/C/binder at 45/45/10 wt %−C/10 rate) and cyclic voltammograms ((b) S/C/binder at 30/60/10 wt %−20 μV s−1) obtained in a liquid electrolyte (LiTFSI 1 mol L−1 + TEGDME/DIOX 50/50).

technique was coupled with UV−visible absorption spectroscopy (UV), high-performance liquid chromatography (HPLC), and electron spin resonance spectroscopy (ESR), in order to investigate the electrolyte composition at different discharge potentials. As conventional lithium-ion carbonate-based electrolytes turn out to be unusable in Li/S cells,31 the study was performed in an ether-based electrolyte. Tetraethylene glycol dimethylether (TEGDME) was used as a solvent, this “glyme” family being commonly used in Li/S batteries and enabling the active material to be well-dissolved.

processes, as Mikhaylik et al. or Shim et al., involving only some of the existing polysulfides, such as S42− or S22− as intermediate species.10,12,15 On the other hand, other authors describe a more complex reduction mechanism, involving many polysulfides intermediate species, such as S82−, S62−, S42−, S32−, S3•−, S22−.7,13,14 In addition, the literature often reports a two plateau reduction process.12,16−18 However, this is not in agreement with the multistep electrochemical process reported for elemental sulfur reduction. Indeed, if involving many equilibrium reactions (chemical and electrochemical reactions), the sulfur reduction would theoretically induce many discharge plateaus. As proof, Mikhaylik et al. show a multiplateau discharge profile when reporting the low-temperature performances of their Li/S batteries.19 As a conclusion, there might be some needs to better understand the Li/S cell discharge process, in regard to the number of steps involved in sulfur reduction as well as the composition of intermediate species. This study aimed at better understanding the sulfur electrochemical reduction occurring in the Li/S rechargeable cells. Because of active material dissolution, the investigation of the discharge mechanism consists of the characterization of the electrolyte composition. To this purpose, different analysis techniques were found in the literature, enabling one to characterize the dissolved lithium polysulfides produced in the electrolyte during cycling. UV−visible absorption spectroscopy was chosen, as the lithium polysulfide compounds are wellknown to strongly absorb the UV−visible radiation.13,14,20,21 Indeed, some publications already report the use of UV−visible absorption spectroscopy for investigation of the sulfur and lithium polysulfide reduction.13,14,20,22−24 Chromatography techniques, such as high-performance liquid chromatography, was employed, as enabling the lithium polysulfides to be separated depending on their chain lengths.25−27 So far, this analysis technique had not been reported in a view to investigate the Li/S reduction. Finally, lithium polysulfide solutions are known to contain some sulfide radicals. Thus, electron spin resonance spectroscopy finally helped to fully characterize the electrolyte composition, as already reported in the literature.20,28−30 First, a preliminary study enabled the UV bands and HPLC peak attributions to be performed. Then, chronoamperometry

■

RESULTS AND DISCUSSION Two-Electrode Cell Measurements. As previously discussed, elemental sulfur reduction mechanism has already been extensively investigated,7,10,12−15,22 but several studies are still in contradiction. In addition, publications often report a two plateau discharge profile for the Li/S rechargeable cell.12,16−18 However, the galvanostatic potential profiles and cyclic voltammograms, as presented in Figure 1, point out three electrochemical processes, i.e., three discharge plateaus and three cathodic peaks (at ∼2.4, ∼2.1 and ∼2 V vs Li+/Li) during the first cycle. Modification of voltammograms is observed during successive cycles for cathodic and anodic processes, indicating complex mechanisms such as chemical and electrochemical coupled processes. The signal at 2.1 V vs Li+/Li is easily detected for sulfur electrodes having a high carbon content. Decreasing the additive content, the signal becomes less detectable, even almost disappearing due to other electrochemical processes. This behavior was also observed for other electrolytes such as PEGDME + LiTFSI 1 mol L−1, thus being associated with electrochemical processes of sulfur species. Some cyclic voltammograms were recorded using three different scan rates, as presented in Figure S-1 in the Supporting Information. The high voltage system (cathodic peak at ∼2.4 V vs Li+/Li) is linked to a reversible redox process, as illustrated by the relative low ΔE value (100 mV, efficiency of about 98%). On the other hand, the second and third redox processes are found to be quite different. The second redox process (2.1 V) becomes almost nondetectable at low scan rates, while relatively intense at higher scan rates. This 3974



Article

particular behavior unambiguously shows that this signal is related to poorly stable intermediate species produced during the first redox process. Concerning the third system, it exhibits high ΔE value (i.e., 300 mV), indicating that chemical reactions are presumably coupled to the electron(s) transfer at this potential (e.g., disproportionation reactions). These results are quite consistent with the fact that the Li/S cell discharge mechanism is complex and involves many intermediates compounds. While being further reduced, the polysulfide chains are gradually shortened, meaning that the second and third reduction steps are deeply dependent on the first one. If using a slower scan rate, the intermediate species, such as radicals (e.g., S3•−) or poorly stable polysulfides, have enough time to evolve before being involved in the second and third reduction steps. As a result, the low voltage reduction processes are modified as compared to faster scan rates and these two other systems are more complex to fully characterize. Preliminary Study. Ex situ analysis techniques, such as UV−visible absorption spectroscopy and HPLC, were used to characterize the active species produced in the electrolyte during cycling. Thus, this preliminary study aimed at characterizing the starting materials UV and HPLC responses, i.e., elemental sulfur, lithium sulfide, solvents, and lithium polysulfides. To this purpose, elemental sulfur and Li2S were dispersed in TEGDME, and the UV−visible absorption spectra of dispersions were recorded. As noticeable in Figure S-2 in the Supporting Information, elemental sulfur indeed dissolves in TEGDME and presents two absorption bands around 270 and 280 nm, in agreement with publications.30,32 On the other hand, as lithium sulfide is insoluble in ether solvents, no UV response could be obtained even after filtration (1 μm, Whatman). A 10−2 mol L−1 concentration of equivalent lithium octasulfide (Li2S8) solution was prepared in TEGDME in order to achieve the polysulfide UV band and HPLC peak attribution. As presented in Figure S-3 in the Supporting Information, different UV bands could be detected for this equivalent Li2S8 solution. As previously mentioned, the polysulfide compounds are well-known to easily disproportionate in solution, leading to a distribution of different species chain lengths.14,20,21,30 In particular, some bands are decreasing over time (617, 470, 450, 280 nm...) while others are increasing (340, 265 nm), and these UV measurements enabled one to point out the changes in lithium polysulfide composition versus time. As a result, some isosbestic points can be noticed on the spectra, which also tend to change over time. For shorter time (some hours), a first chemical process is displayed by isosbestic points at λ = 267, 323, and 415 nm, whereas two others are detected at λ = 279 and 394 nm for longer time (after 1 day evolution). This indicates that successive equilibrium reactions occur in solution. As a consequence, a single polysulfide chain length cannot be obtained pure, as the species easily disproportionate in solution, giving many polysulfide compounds. Disproportionation reactions 4−6 can be summarized as follows n−m S8 8

(4)

2Sn 2 − → Sn + m 2 − + Sn − m 2 −

(5)

Sn 2 − ↔ 2Sn /2•−

(6)

2Sn 2 − ↔ Sm 2 − +

It was found that the 415 nm absorbance is almost invariant over time, thus independent of the disproportionation processes and enabling some quantitative measurements to be carried out. Looking at the 415 nm absorbance versus sulfur concentration, it is possible to obtain a linear regression following Beer−Lambert’s law, indicating the proportional relation of absorbance with the solution concentration at this specific wavelength. Looking at the shape of spectra, absorption bands prove to be quite broad and close to each other. As a result, it was difficult to identify the exact number of bands and consequently the species present. Besides, as it is not possible to isolate a population of single polysulfide chain lengths, it becomes tricky to completely characterize the polysulfide system using only UV−visible absorption spectroscopy. Anyway, a list of the approximate band wavelengths, and their possible attribution, is also given in Figure S-3 in the Supporting Information. This hypothetical assignment was based on literature data20,22,32 but also on the assumption that the polysulfide absorption energy may decrease with an increasing sulfur chain length, leading to shorter wavelengths for shorter polysulfides. As well described in the literature,14,20 different UV bands are attributed to the S62− species, i.e., the 470, 350, 300, and 260 nm bands. As regards to the other polysulfide chain lengths, one single band is assigned to theses compounds. However, this does not mean that they do not present many absorption bands. This single band attribution is rather due to a lack of knowledge and supporting literature. The 617 nm band is attributed to the S3•− radical, as commonly reported to exist in polysulfide solutions, giving them a blue color.14,20,33 In order to fully characterize the lithium polysulfide solutions, another analysis technique was implemented, i.e., HPLC, as reported to be a successful tool to separate the different lithium polysulfide chain lengths.25−27 Using a C18 reverse phase HPLC column, polysulfide compounds (methylated beforehand) could be eluted according to their chain length, with a retention time increasing with the number of sulfur atoms in the chain. As for UV, the raw materials were characterized first, in order to achieve the peak attribution. TEGDME and MeTf were directly analyzed as-received, while solid Li2S and elemental sulfur materials were first dispersed in TEGDME. The methylation agent and solvents are quickly eluted with retention times close to 3 min. The insoluble Li2S material does not give any HPLC peak, while elemental sulfur can be detected at around 80 min. The polysulfide-containing solution was also investigated by HPLC, and the resulting chromatogram is presented in Figure S-4 in the Supporting Information. As the polysulfide compounds are eluted according to the number of sulfur atoms in the chain, the peak attribution is quite easy to perform. To check the peak attribution, publications report that the logarithm of retention time should be proportional to the number of sulfur atoms, that is to say to the n of Li2Sn.25−27 As presented in Figure S-5 in the Supporting Information, a linear regression can be obtained with a satisfying coefficient of determination and the peak attribution could be confirmed. Given their low and similar polarity, methylated lithium octasulfide and elemental sulfur are closely eluted, leading to the superimposition of their broad signals (at 72 and 76 min, respectively). As for spectroscopic measurements, different concentrations were prepared for Li2S8-containing solutions and characterized by HPLC. However, because of disproportionation reactions, 3975



Article

Figure 2. Chromatograms of catholyte samples obtained after the cell polarization at different potentials.

production of short polysulfides and elemental sulfur. Also, as some polysulfides are concurrently produced (from S32− to S62−), this indicates that the redox potentials of these kinds of product may be close to each other and may prevent the chemical processes to be isolated and the different discharge plateaus to be detectable. Chronoamperometry Measurements. Thanks to the preliminary study, the ex situ analysis techniques could be coupled in order to investigate the sulfur reduction process and to characterize the electrolyte composition during cycling. In order to recreate this electrolyte composition, a glass cell was used and filled with a polysulfide-containing electrolyte. Specific potential values were applied to this catholyte (i.e., 3, 2.4, 2.3, 2.1, 1.95, 1.5 V vs Li+/Li), and the resulting current was measured as a function of time, waiting for a quasi equilibrium state to be reached (i < 10 μA). HPLC, UV, and ESR data are presented in Figures 2, 3, and 4, respectively. Looking at the 3 V measurements, HPLC chromatogram indicates that only S8, S82−, and S42− species exist in solution. UV data are in agreement with these results, as the strong absorption of elemental sulfur is detected around 270−280 nm, while weaker bands are noticed around 420 nm (S42−) and 560 nm (S82−). The presence of S82− indicates that the sulfur species might not be fully converted into elemental sulfur despite the 100 h of cell polarization at 3 V vs Li+/Li. Indeed, 3 V vs Li+/Li

no proportional relation can be found between one single peak area and the sulfur concentration. On the other hand, when looking at the whole integration area (calculated from integration of all peaks), it is possible to find a proportional relation with the overall sulfur concentration (evidenced in Figure S-5 in the Supporting Information). The chain length distribution as a function of the overall sulfur concentration is given in Figure S-6 in the Supporting Information. Indeed, the polysulfide distribution (chain lengths) is changing with this sulfur concentration. Looking at low sulfur content (i.e., 4.10−3 and 6.10−3 mol L−1 solutions), Li2S2 and S8 are found to be the most predominant compounds. On the contrary, looking at the most concentrated solution (i.e., 10−1 mol L−1), the intense peaks are rather linked to Li2S3 and Li2S4 compounds, while the S8 peak is dramatically decreased. The disproportionation reactions conveniently explain the decrease in elemental sulfur content and the change in polysulfides distribution when the overall sulfur concentration is increased. Obviously, the disproportionation equilibriums deeply depend on the initial overall sulfur concentration, and shorter polysulfide compounds are found to be more favorable in dilute solutions. Anyway, no polysulfide solutions seem to contain any long polysulfide compounds, such as Li2S7 and Li2S8, for example. Indeed, these long polysulfide species disproportionate in solution, leading to the 3976



Article

Figure 3. UV−visible absorption spectra of catholyte samples obtained after the cell polarization at different potentials. Different sizes were employed for the quartz cells, as reported on insets of the graphs, in order to achieve a reasonable UV absorption on the whole 200−800 nm range.

potential is notably higher than the oxidation potential of S82−, i.e., close to 2.4 V vs Li+/Li. However, even after catholyte polarization at 3 V vs Li+/Li for several days, the current did not fall down to zero. This may be associated with the wellknown shuttle mechanism, which induces the presence of residual S82− species in the electrolyte. Also, lithium tetrasulfide was not expected at this potential, as sulfur should exist in its highest oxidized states. However, as evidenced by UV measurements, the S42− band is not really intense, and the tetrasulfide content might be low. Thus, this compound can arise from the S82− disproportionation reaction, as described below

S82 − ↔ S4 2 − +

1 S8 2

(7)

Looking at the 2.4 V measurements, the UV bands associated with elemental sulfur (270−280 nm) tend to decrease, while the 470, 450, 350, and 260 nm bands are now distinctively visible on the spectra. As a matter of fact, the UV data point out the consumption of sulfur and the formation of long polysulfide chain lengths, such as S52− and S62− compounds. These data may be in disagreement with publications, as often reported the S82− formation during the first reduction step. In fact, it is assumed that lithium octasulfide is indeed produced during the first step of sulfur electrochemical reduction, as follows S8 + 2e− → S82 − 3977

(8)



Article

band of the S3•− radical, this latter is surprisingly increasing, while the one of S62− is rather decreasing. As S62− may be consumed during the second reduction step, it can be assumed that this dianion is involved in two equilibriums: one fast electrochemical reaction leading to the formation of S42− compound (eq 13-fast), while also being involved in the slower S3•− association (eq 12-slow), as follow. 2S3•− → S6 2 −

(12-slow)

2S6 2 − + 2e− → 3S4 2 −

(13-fast)

If the first reaction 12-slow is slower than the second one (eq 13-fast), this conveniently could explain why S62− dianion could not be found in solution during this second reduction step anymore. The S32− compound can also be detected in solution. As a matter of fact, these above-mentioned reactions can be completed by another disproportionation reaction, such as

2S4 2 − → S52 − + S32 −

Also as lithium trisulfide may not be soluble enough in the electrolyte,34 the dianion may partially precipitate, thus explaining why the recording of UV spectra had to be performed before the end of polarization. Anyway, the production of poorly stable species, such as radicals, during the second and third reduction steps, is consistent with the cyclic voltammogram data, as pointed out that theses processes may be related to some intermediate species formation. Looking at the 1.95 V measurements, HPLC analysis was performed on two samples: one electrolyte sample was taken and analyzed during chronoamperometry measurements (i.e., i ≈ −80 μA), while a second one was sampled at the end of the experiment (i.e., |i| < 10 μA). This additional measurement aimed at better understanding what may occur during this third electrochemical step. The resulting chromatograms are presented in Figure 2 and were indeed quite different from each other. HPLC peaks could be detected for the first electrolyte sample, and the S 2 2− , S 3 2− , S 4 2− , and S 5 2− compounds could be identified. On the other hand, the second sample did not show any peaks. Likewise, the chromatogram corresponding to the 1.5 V measurements does not present any peaks too. Thus, these HPLC data nicely point out (1) the gradual decrease of the polysulfide chain when further reducing elemental sulfur. Thus, the electrochemical reactions, which may occur during the third reduction step, can be described as follows

Figure 4. X-band ESR spectra (vertical shifted) obtained at 100 K for catholyte samples after the cell polarization at different potentials.

However, the compound may not be stable enough in solution and may disproportionate as follows 1 S82 − ↔ S6 2 − + S8 (9) 4 3 S8 (10) 8 Also if the electrochemical reaction is slower than the disproportionation one, this could explain why the S82− intermediate specie cannot be found in solution, whereas S62− and S52− do. Looking at the 2.3 V measurements, the optical response is still related to the growth of the 470, 450, 350, 300, and 260 nm bands, thus indicating the increase of the S52− and S62− concentrations. The electrochemical and disproportionation reactions remain unchanged, consisting of S82− formation, which quickly disproportionates in solution producing S62− and S52− intermediate species. Another band starts to emerge, i.e., the 617 nm band, which was previously supposed to be related to the S3•− radical. Indeed, this species is well-known to be in equilibrium with the S62− compound, as produced by the lithium hexasulfide disproportionation (called dissociation in the literature): S82 − ↔ S52 − +

S6 2 − → 2S3•−

(14)

(11)

With respect to the 2.1 V measurements, the UV spectra were not recorded at the end of the chronoamperometry measurements, as the polysulfide compounds were found to precipitate at the end of polarization. The catholyte was rather sampled at i ≈ −80 μA, i.e., before the polysulfide precipitation and the corresponding optical response is now changing. The 260, 300, 350, and 470 nm bands linked to S62− are decreasing, while two new absorption bands, attributed to S32− and S42− species, are emerging at 340 and 420 nm. As a result, the UV spectra point out the consumption of S82−, S62−, and S52− compounds and the production of S42− and S32− dianions. Thus, the second step is linked to the shortening of the lithium polysulfide chain when further going into sulfur reduction. Regarding the absorption

3S4 2 − + 2e− → 4S32 −

(15)

2S32 − + 2e− → 3S2 2 −

(16)

These reactions can occur concurrently as well as successively. Another mechanism can be given as follow for this third reduction step, involving both electrochemical (eq 17) and disproportionation (eq 18) reactions: S4 2 − + 2e− → 2S2 2 −

(17)

S2 2 − + S4 2 − ↔ 2S32 −

(18)

(2) The precipitation of lithium polysulfides at the end of discharge, as no product could be detected when sulfur was fully reduced and conveniently agrees with other publications.10,16,35,36 3978



Article

composition, as it dictates the species equilibriums. Thus, if changing the solvent, for example, electrochemical and disproportionation reactions may be modified, resulting in a different discharge mechanism. A schematic representation of the reduction process is given in Figure 5, where the proposed

Indeed, the literature data often reports the precipitation of short polysulfides at the end of discharge, which may be one of the major capacity limitations for the Li/S cells.35−37 Indeed, sulfide compounds are highly insulating and may quickly passivate the positive electrode if precipitating. Discussions on Li/S cell capacity limitations are not the aim of this paper, as it has already been discussed in previous ones.31,37 However, these data nicely pointed out the short polysulfide precipitation, thus corroborating the hypothesis of the end of discharge because of the positive electrode passivation. As all polysulfide products were found to be insoluble at the end of discharge, lithium sulfide compound must be produced during this third reduction step, even if it was not detected by HPLC and other analysis techniques. Therefore, another electrochemical reaction can be given for this third step, as follows

S2 2 − + 2e− → 2S2 −

(19)

Spectroscopic data of the first 1.95 V sample (i ≈ −80 μA) are presented in Figure 3. Optical responses of the 1.95 V (|i| < 10 μA)) and 1.5 V samples could not be recorded, as the catholyte samples were mainly composed of solid products. The two spectra (0.1 and 1 mm large cells) clearly points out the decrease of the S3•−, S62−, S52−, and S42− bands. On the other hand, the S22− and S32− bands are significantly increased, which are visible at 265 and 340 nm, respectively. The UV data indicate once again the gradual shortening of the polysulfide chain length along sulfur reduction and confirm the HPLC data interpretation. As previously mentioned, ESR spectroscopy was used to identify the possible formation of radical species during the sulfur reduction process. A very intense ESR signal is observed for the 2.3 and 2.1 V measurements, which corresponds to a single population of radical species, with a signature characteristic of a rhombic S = 1/2 system (gx = 2.0011, gy = 2.0329, gx = 2.0529 leading to gav = 2.029). On the basis of literature data,28−30,38 only the opened form of S3•− could exist in the investigated solution. Indeed, the recorded ESR spectrum cannot be related either to the closed form of S3•− (isotropic or quasi-axial ESR signal) or to S2•− (gy ≈ 2.8). On the other hand, g-factors are found to be consistent with all previous experimental measurements performed on ultramarine typepigments or DFT calculations that were made on the opened form of S3•−. In addition, this assignment well agrees with the appearance of the 617 nm band on UV spectra, thus confirming the ESR signal attribution. As this signal significantly decreases during the 2.1 V measurements, these ESR data point out that the radical is consumed during the second reduction step, thus confirming the mechanism that was mentioned before (eqs 12slow and 13-fast). The S3•− consumption was indeed confirmed by the 1.95 V sample measurements, as no signal could be detected anymore. As a conclusion, a possible mechanism for sulfur reduction in TEGDME/LiTFSI-based electrolyte could be proposed, coupling many electrochemical and disproportionation reaction equations as presented in the equation in the Supporting Information. Looking at this proposed reduction process, it is worth noticing that the mechanism involves complex and successive equilibriums, which may depend on the first reduction step, on the intermediate species’ stability, and on the scan rate (or galvanostatic rate), thus explaining the cyclic voltammogram evolution. It is also worth noticing that this discharge mechanism must depend on the electrolyte

Figure 5. Proposed sulfur reduction mechanism, involving disproportionation and electrochemical reactions. Major lithium polysulfide compounds are listed on the figure, as well as the specific capacities corresponding to each step.

disproportionation and electrochemical reactions are summed up. The main products are also given on this figure as well as the specific capacities calculated for each step. One can note that the experimental results seem to be in accordance with the calculated capacities, thus confirming the proposed mechanism and the UV−visible band attribution. Only the third step does not fit with the calculated capacity, as predicted about 1256 mAh g−1, while obtaining only 700 mAh g−1 experimentally. These observations are in agreement with the fact that the discharge capacity may be limited by the precipitation process and the positive electrode passivation occurring at the end of the discharge. More details can be found in previous papers about this limitation, which explains the incomplete sulfur active material utilization and capacity.37

■

CONCLUSIONS This communication aimed at providing a better understanding of the sulfur reduction process, also pointing out the great interest of UV, HPLC, and ESR analysis techniques for the investigation of the electrolyte composition. Using the 3979



Article

(8) Aurbach, D.; Pollak, E.; Elazari, R.; Salitra, G.; Kelley, C. S.; Affinito, J. J. Electrochem. Soc. 2009, 156, A694−A702. (9) Choi, Y. J.; Chung, Y. D.; Baek, C. Y.; Kim, K. W.; Ahn, H. J.; Ahn, J. H. J. Power Sources 2008, 184, 548−552. (10) Barchasz, C.; Leprêtre, J. C.; Alloin, F.; Patoux, S.; International Battery Association Meeting, Waikoloa, Hawaii, 2010 (11) Shim, J.; Striebel, K. A.; Cairns, E. J. Electrochem. Soc. 2002, 149, A1321−A1325. (12) Mikhaylik, Y. V.; Akridge, J. R. J. Electrochem. Soc. 2004, 151, A1969−A1976. (13) Kim, B. S.; Park, S. M. J. Electrochem. Soc. 1993, 140, 115−122. (14) Han, D. H.; Kim, B. S.; Choi, S. J.; Jung, Y.; Kwak, J.; Park, S. M. J. Electrochem. Soc. 2004, 151, E283−E290. (15) Wang, C.; Chen, J.; Shi, Y.; Zheng, M.; Dong, Q. Electrochim. Acta 2010, 55, 7010−7015. (16) Kolosnitsyn, V. S.; Karaseva, E. V. Russ. J. Electrochem. 2008, 44, 506−509. (17) Kumaresan, K.; Mikhaylik, Y.; White, R. E. J. Electrochem. Soc. 2008, 155, A576−A582. (18) Ryu, H. S.; Guo, Z.; Ahn, H. J.; Cho, G. B.; Liu, H. J. Power Sources 2009, 189, 1179−1183. (19) Mikhaylik, Y. V.; Akridge, J. R. J. Electrochem. Soc. 2003, 150, A306−A311. (20) Leghié, P.; Levillain, E.; Lelieur, J. P.; Lorriaux, A. New J. Chem. 1996, 20, 1121−1130. (21) Rauh, R. D.; Shuker, F. S.; Marstin, J. M.; Brummer, S. B. J. Inorg. Nucl. Chem. 1977, 39, 1761−1766. (22) Li, Y.; Zhan, H.; Liu, S.; Huang, K.; Zhou, Y. J. Power Sources 2010, 195, 2945−2949. (23) Dominko, R.; Demir-Cakan, R.; Morcrette, M.; Tarascon, J. M. Electrochem. Commun. 2011, 13, 117−120. (24) Gaillard, F.; Levillain, E.; Dhamelincourt, M. C.; Dhamelincourt, P.; Lelieur, J. P. J. Raman Spectrosc. 1997, 28, 511−517. (25) Kamyshny, A.; Ekeltchik, I.; Gun, J.; Lev, O. Anal. Chem. 2006, 78, 2631−2639. (26) Rizkov, D.; Lev, O.; Gun, J.; Anisimov, B.; Kuselman, I. Accred. Qual. Assur. 2004, 9, 399−403. (27) Kamyshny, A.; Goifman, A.; Gun, J.; Rizkov, D.; Lev, O. Environ. Sci. Technol. 2004, 38, 6633−6644. (28) Goslar, J.; Lijewski, S.; Hoffmann, S. K.; Jankowska, A.; Kowalak, S. J. Chem. Phys. 2009, 130, 204504-1−204504-11. (29) Arieli, D.; Vaughan, D. E. W.; Goldfarb, D. J. Am. Chem. Soc. 2004, 126, 5776−5788. (30) Tobishima, S. I.; Yamamoto, H.; Matsuda, M. Electrochim. Acta 1997, 42, 1019−1029. (31) Barchasz, C.; Leprêtre, J.-C.; Alloin, F.; Patoux, S. 218th Electrochemical Society Meeting, Las Vegas, NV, 2010 (32) Heatley, N. G.; Page, E. J. Anal. Chem. 1952, 24, 1854. (33) Boros, E.; Earle, M. J.; Gilea, M. A.; Metlen, A.; Mudring, A. V.; Rieger, F.; Robertson, A. J.; Seddon, K. R.; Tomaszowska, A. A.; Trusov, L.; Vyle, J. S. Chem. Commun. 2010, 46, 716−718. (34) Kolosnitsyn, V. S.; Karaseva, E. V.; Seung, D. Y.; Cho, M. D. Russ. J. Electrochem. 2002, 38, 1314−1318. (35) Cheon, S.; Ko, K.; Cho, J.; Kim, S.; Chin, E.; Kim, H. J. Electrochem. Soc. 2003, 150, A796−A799. (36) Yuan, L.; Qiu, X.; Chen, L.; Zhu, W. J. Power Sources 2009, 189, 127−132. (37) Barchasz, C.; Leprêtre, J.-C.; Patoux, S.; Alloin, F. J. Power Sources 2011, 199, 322−330. (38) Fehrmann, R.; Von Winbush, S.; Papatheodorou, G. N.; Berg, R. W.; Bjerrum, N. J. Inor. Chem. 1982, 21, 3396−3400.

chronoamperometry technique, the electrolyte composition was characterized at different discharge potentials in a TEGDME-based electrolyte. Many polysulfides species were found to exist in solution due to disproportionation reactions. As demonstrated by UV measurements, S82− and S62− dianions are produced during the first reduction step, in agreement with the 280 mAh g−1 corresponding capacity, while the S52− and S3•− presence arises from disproportionation reactions. Through the reduction of S62− and S3•− species, S42− and S32− are produced during the second step, while the amount of longer polysulfide species decreases and in agreement with the 140 mAh g−1 capacity. Finally, shorter polysulfide compounds, such as S32−, S22−, and S2−, can be found at the end of the reduction process, which tend to precipitate in the electrolyte. Indeed, the precipitation of the poorly soluble and insulating short polysulfide compounds was evidenced by HPLC measurements, enabling one to explain the incomplete sulfur active material utilization (third step giving only 700 mAh g−1 experimentally, instead of 1256 mAh g−1 theoretical). Furthermore, ESR data enabled one to point out that one single sulfur radical, i.e., S3•−, is involved in the sulfur reduction process, being produced during the first step and consumed during the second one. Thanks to these coupled measurements, a discharge mechanism could be proposed, which turns to be specific to the investigated system, i.e., to the TEGDME/ LiTFSI-based electrolyte.

■

ASSOCIATED CONTENT

S Supporting Information *

Experimental procedures such as coin cell preparation, chronoamperometry setup and characterization technique procedures, details of data analysis: two-electrodes cell measurements, preliminary study data, equations for the proposed sulfur reduction mechanism, and Figures S-1−S-6. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to acknowledge the CEA-INSTN for supporting part of this study (Ph.D. funding awarded to Céline Barchasz). The authors would also like to thank the “TGE Réseau National de RPE Interdisciplinaire”, Grant FR-CNRS 3443.

■

REFERENCES

(1) Tarascon, J. M.; Armand, M. Nature 2001, 414, 359−367. (2) Armand, M.; Tarascon, J. M. Nature 2008, 451, 652−657. (3) Ellis, B. L.; Lee, K. T.; Nazar, L. F. Chem. Mater. 2010, 22, 691− 714. (4) Rauh, R. D.; Abraham, K. M.; Pearson, G. F.; Surprenant, J. K.; Brummer, S. B. J. Electrochem. Soc. 1979, 126, 523. (5) Yamin, H.; Penciner, J.; Gorenshtein, A.; Elam, M.; Peled, E. J. Power Sources 1985, 14, 129−134. (6) Yamin, H.; Gorenshtein, A.; Penciner, J.; Sternberg, Y.; Peled, E. J. Electrochem. Soc. 1988, 135, 1045−1048. (7) Jung, Y.; Kim, S. Electrochem. Commun. 2007, 9, 249−254. 3980


Sulfur Cell Discharge Mechanism: An Original Approach for

Recommend Documents