Healing High-Loading Sulfur Electrodes with ... - ACS Publications

Mar 16, 2017 - The inert SMiP is activated by the solubilization effect of polysulfides whereas the unstable phase transfer is mediated by mitigated s...
0 downloads 4 Views 6MB Size
Article pubs.acs.org/JACS

Healing High-Loading Sulfur Electrodes with Unprecedented Long Cycling Life: Spatial Heterogeneity Control Hong-Jie Peng, Jia-Qi Huang, Xin-Yan Liu, Xin-Bing Cheng, Wen-Tao Xu, Chen-Zi Zhao, Fei Wei, and Qiang Zhang* Beijing Key Laboratory of Green Chemical Reaction Engineering and Technology, Department of Chemical Engineering, Tsinghua University, Beijing 100084, China S Supporting Information *

ABSTRACT: Self-healing capability helps biological systems to maintain their survivability and extend their lifespan. Similarly, self-healing is also beneficial to nextgeneration secondary batteries because high-capacity electrode materials, especially the cathodes such as oxygen or sulfur, suffer from shortened cycle lives resulting from irreversible and unstable phase transfer. Herein, by mimicking a biological selfhealing process, f ibrinolysis, we introduced an extrinsic healing agent, polysulfide, to enable the stable operation of sulfur microparticle (SMiP) cathodes. An optimized capacity (∼3.7 mAh cm−2) with almost no decay after 2000 cycles at a high sulfur loading of 5.6 mg(S) cm−2 was attained. The inert SMiP is activated by the solubilization effect of polysulfides whereas the unstable phase transfer is mediated by mitigated spatial heterogeneity of polysulfides, which induces uniform nucleation and growth of solid compounds. The comprehensive understanding of the healing process, as well as of the spatial heterogeneity, could further guide the design of novel healing agents (e.g., lithium iodine) toward high-performance rechargeable batteries.

1. INTRODUCTION Self-healing, the ability to repair damage spontaneously, is a ubiquitous feature in nature and plays a vital role in increasing the life expectancy of organisms. This inspires many promising applications toward smart electronics, artificial skin, and energy storage.1−3 For example, the self-healing characteristic is extremely desirable for long-life energy storage devices. On one hand, there are growing interests in adopting new battery chemistry to replace conventional lithium-ion batteries that are almost approaching the theoretical limit.4,5 On the other hand, the high-capacity electrode materials such as silicon and sulfur always suffer from rapid capacity fading and short life. This dilemma is primarily ascribed to the mechanical fracture and structural collapse.6 If damage of these electrode materials can be spontaneously healed, their huge potential will be well translated into reality. Recently, Wang et al. pioneered the work where self-healing chemistry was adopted to design a skin-inspired polymer binder for silicon microparticle anodes. This binder spontaneously repaired the mechanical fracture produced during silicon expansion/contraction and thus enabled stable operation.2 Their strategy offers an intriguing way to enable negative electrode materials tortured by large volume expansion. However, most high-capacity positive electrode materials (e.g., sulfur, oxygen, carbon dioxide) exert conversion chemistry, and thus they often possess an inherently “hostless” feature and virtually “infinite” structural change upon electrochemical reactions. Neither has such skin-inspired self-healing © 2017 American Chemical Society

strategy been demonstrated effective nor are there any other clear self-healing concepts proposed from fundamentals to expand battery life. In principle, conversion electrochemistry involves multielectron transfer, which offers not only innately high energy but unfortunately also drastic phase transfer (e.g., solid sulfur to soluble polysulfides or gaseous oxygen to solid lithium peroxides in aprotic electrolytes).5 The thermodynamic and kinetic barriers of phase transfer render the battery unstable and irreversible. More specifically, the accumulation of solid compounds exaggerates this issue. For instance, sulfur, an alternative positive electrode material (theoretical capacity of 1672 mAh g−1) to lithium-ion battery cathodes (capacities less than 250 mAh g−1), regrettably suffers from severe issues including (1) extremely low electron/ion conductivity of solid compounds, (2) dissolution, diffusion, and parasitic reactions of intermediate polysulfides, as well as induced “shuttle effect”, and (3) dramatic structural change upon dissolution/ redeposition.7−14 Particularly, owing to the rampant diffusion and uncontrolled deposition of intermediate polysulfides,7,8,11,15 the structural-change issue is much more serious than the hypothetical solid−solid lithiation of sulfur to lithium disulfide (Li2S). It is obvious that the intrinsic but uncontrolled phase transfer between solid compounds (sulfur and Li2S) and liquid polysulfides accounts for poor cycling stability and Received: December 3, 2016 Published: March 16, 2017 8458

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Article

Journal of the American Chemical Society

Figure 1. Schematic of healing mechanism for SMiPs. (A) top, simplified schematic of coagulation cascade toward formation of thrombus (blood clot) and f ibrinolysis of thrombus into soluble f ibrin fragments, which are modulated by two trypsins: thrombin and plasmin; bottom, schematic of deposition/dissolution of micrometer-sized sulfur/Li2S particles with limited electrical contact, hindering full electrochemical utilization. To enable the stable and reversible utilization, a complementary route is introduced, through chemical reactions between an extrinsic healing agent in solution and insoluble sulfur compounds. (B) Schematic of how spatial heterogeneity impacts the solution−solid phase transfer. Gibbs−Thomson theory is employed as simple implementation to describe to solution−solid process of Li2S nucleation and growth under galvanostatic condition.

effect of concentration and amount of polysulfide additives.34 The polysulfide electrolyte was further coupled with various nonmetallic anode materials such as Si/SiOx and carbon to build high-energy and highly efficient Li−S batteries.35,36 The synergistic effect on stabilizing solid electrolyte interphase, provided by polysulfide and LiNO3, on lithium anode was also suggested.37−39 Nevertheless, the underlying mechanism of how polysulfides “heal” a conventional sulfur cathode still remained concealed due to either limited cycling number or insufficient sulfur loading presented in the previous works.33,34,40 Typically, the buffering effect is widely accepted as a straightforward explanation while its persistency remains doubtful, since the concentration gradient changes constantly and drastically at high sulfur loading (Figure S1). If a mechanistic understanding can be acquired on self-healing Li−S batteries with sufficient level of cycling numbers and

electrode reversibility. Therefore, rationally understanding and manipulating the phase transfer process in lithium−sulfur (Li− S) batteries is a kernel in improving battery performance. Besides the strategies either to physically confine polysulfides into a porous framework16−26 or to block the diffusion of polysulfides by an interlayer or functional separators,27−32 healing sulfur positive electrodes is another intriguing approach since self-healing chemistry has achieved great success in negative electrodes.2 However, the skin-inspired polymer binder, in principle, is insufficient to repair the structural change induced by phase transfer. Therefore, an alternative healing strategy is under urgent demand to control phase transfer between sulfur/Li2S and polysulfides. Toward selfhealable Li−S batteries, Xu et al. first proposed a novel etherbased electrolyte employing only polysulfides and LiNO3 as the cosalts to operate a simple sulfur cathode with stable cycling over 50 cycles.33 Chen et al. also systematically investigated the 8459

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Article

Journal of the American Chemical Society

Figure 2. Diffusion, distribution, and deposition characteristics of SHE. (A) In-situ monitoring of polysulfide diffusion in routine electrolyte (RE) and SHE. U-shape quartz cells were set up for visiblization demonstration, where SMiPs loaded on nickel foam and a lithium foil served as cathode and anode, respectively. (B) Corresponding gray level of randomly selected regions (indicated in Figure S7) as a function of time duration, illustrating the temporal and spatial variations of polysulfide concentration in RE and SHE. Scanning electron microscopy (SEM) images and corresponding energy dispersive spectroscopic (EDS) mapping of elemental nickel and sulfur of (C) a pristine nickel foam/SMiP cathode and discharged cathodes in (D) RE and (E) SHE. Scale bars, 100 μm.

composed of sulfur nanoparticles (SNPs), attains an optimized capacity (∼3.7 mAh cm−2) with almost no decay after 2000 cycles at a high sulfur loading of 5.7 mg(S) cm−2.

sulfur loading, a series of novel healing agents will thereafter be designed under rational guidance. In this contribution, we propose a general scheme to explain the mechanism of self-healing electrolytes (SHEs) preloaded with polysulfides and to engineer extrinsic healing agents toward self-healable Li−S batteries. The healing mechanism mimics biological processes, fibrinolysis, within blood vessels. The uncontrolled deposition and accumulation of deactivated solid products are analogous to coagulation of thrombus that undesirably obstructs flow of blood to healthy vessels (Figure 1A).41 Polysulfides, which mimic the basic function of an enzyme called plasmin to solubilize the thrombus, are capable of transferring these solid compounds into solution phase and enabling their subsequent reparticipation in electrochemical cycles. More importantly, similar to plasmin, the ability of polysulfides to self-regenerate guarantees their persistency to work in Li−S batteries. Additionally, a classical crystal growth model is first introduced to explain how polysulfides mediate the phase transfer and heal the sulfur cathode. Spatial heterogeneity plays a vital role in the conversion-type electrochemical reactions, agreeing with the insertion-type ones coincidently.58 The sulfur microparticle (SMiP) positive electrode, which is simple and cost-effective compared to that

2. RESULTS AND DISCUSSION There are two categories of self-healing mechanisms: intrinsic self-healing and extrinsic self-healing.3 Our strategy herein is assigned to the second one because it is dependent on external healing chemicals. In principle, the long-term function of extrinsic self-healing is realized through providing healing agents sustainably. Therefore, in our case, the healing agent must possess the following attributes: (1) the possibility of being regenerated, (2) the capability to mediate solution-tosolid phase transfer controllably, and (3) the ability to restore deactivated solid compounds. The first aspect is unambiguously guaranteed by polysulfides whereas the other two require further demonstration. For the second trait, the mechanism of nucleation−growth from a solution phase is the kernel and therefore needs to be understood first.42 Here we provide a simplified mechanism according to a classical supersaturation theory (Gibbs− Thomson effect).42 The theoretical analysis implies that polysulfides with higher concentration precipitate at lower 8460

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Article

Journal of the American Chemical Society

Figure 3. Self-healable SMiP cathodes based on SHE. (A) CV profiles of SMiPs with RE and SHE-1 M. (B) Galvanosatic discharge−charge curves of SMiPs with RE, SHE-0.33 M, and SHE-1 M, for the initial discharge and the first full cycle at a current density of 0.7 mA cm−2 (0.17 A g(Scathode)−1). (C) Ultralong stability evaluation of RE/SMiP, RE/SNP, and SHE−0.33 M/SMiP at a current density of 1.2 mA cm−2. (D) Self-healable capacities of SHE-1 M/SMiP upon long-term cycling at different current densities. The solid sulfur loading (Scathode) is indicated in each figure; while the sulfur loading in SHE is 0.64 and 1.92 mg(SSHE) cm−2 for SHE-0.33 and 1 M/SMiP (4.2 mg(Scathode) cm−2), respectively. For electrodes with 1.4 mg(Scathode) cm−2, the sulfur loading in SHE-0.33 M is 0.16 mg(SSHE) cm−2.

transfer in a Li−S cell, whereas the simple buffering effect is not persistent nor impactful. The capability of polysulfides to decrease the concentration variation within the cathode stems from its ability to solubilize solid sulfur species with poor dispersion, even when those particles are large and irregular. Although it has been widely known that stoichiometric Li2Sx solution could be prepared from the spontaneous reactions between sulfur and Li2S in organic solvents through heating and stirring, these reactions have not been directly proven under ambient and static conditions yet. Model experiments were therefore designed to indicate the facile reactions under ambient and static condition (Figure S3−S5). SHEs with polysulfide concentration of 0.33, 1.0, and 2.0 mol L−1 were prepared and denoted as SHE-0.33, -1, and -2 M, respectively. The reaction duration was within the range of electrochemical measurements, allowing their implementation in a working cell. As indicated by ex situ ultraviolet−visible light adsorption spectroscopy, these reactions mainly involve the transformation between high-order and low-order polysulfides. Higher concentration of preloaded polysulfides renders smaller variation in the spectra and thus solution composition (Figure S6). This consequently implies that a higher phase-transfer durability could be attained using high-concentration SHEs under a static condition and fixed time scale (Supplementary Discussion II).

overpotential and with a higher rate (Figure S2 and Supplementary Discussion I). If the distribution of polysulfides is inhomogeneous, their deposition will also be uneven. The nonuniform electrodeposition, on one hand, is unable to fully utilize the exposed conductive surface and, on the other hand, renders large aggregates that easily lose electrical contact (Figure 1B). Hence the spatial heterogeneity of precursors is of great significance for controlled nucleation and growth of Li2S. Such a theoretical analysis is in good accordance with the overpotential-correlated Li2S nucleation/growth model proposed by Chiang and co-workers.43 Previous papers employing polysulfide electrolytes commonly held the view that concentration gradient between cathode and preexisting polysulfides buffer the “shuttle” effect.33,40 However, it has been rarely demonstrated that the “shuttle” effect virtually occurs not only between the opposite electrodes but also within the positive electrode region, leading to spatial heterogeneity of polysulfide distribution (Figure 1B). According to the aforementioned theoretical model (Supplementary Discussion I), this heterogeneity undoubtedly results in unsynchronized nucleation and growth rates of Li2S in different regions. The resultant inefficient utilization of conductive scaffolds and irreversible electrical disconnection of solid products consequently lower the overall capacity and cyclability. Homogenizing the concentration within the cathode region is thus crucial for controlling solution−solid phase 8461

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Article

Journal of the American Chemical Society Although many in situ diffraction or spectroscopic techniques have been developed to characterize the time-dependent evolution of sulfur species in a working cell,44−47 it is still very challenging to monitor the spatial distribution of polysulfides and corresponding solid deposition at microscale. Therefore, we designed a visiblization experiment to investigate the phase transfer behaviors of SMiPs mediated by polysulfides (Figure 2A). It can be observed that the polysulfide concentration, produced electrochemically from sulfur cathode, exhibits inhomogeneous distribution both temporally and spatially. In contrast, the SHE-based cell barely displays any visible changes in color. Statistical analysis and recorded movie of the entire process further validate not only temporal but also spatial heterogeneities of polysulfides in RE (Figure 2B, S7, and Supplementary Movie). These heterogeneities, however, can be well addressed by SHE. The impact of as-validated heterogeneity on phase transfer in a working Li−S cell was examined by postmodern SEM combined with EDS mapping. The same electrodes as in the visiblization set-ups were discharged galvanostatically at a current density of 0.1 mA cm−2. In the pristine nickel foam/ SMiP positive electrode, SMiPs were almost separated from the porous scaffolds and exhibited lateral size larger than 100 μm (Figure 2C). After discharging in RE, the sulfur particle still remained sizable but consisted of many primary particles with an average size of ca. 20 μm. This observation is attributed to the recrystallization (Figure 2D). Nevertheless, a large portion of conductive surface was still unoccupied by sulfur species. After being discharged in SHE, however, the nickel foam backbone was uniformly coated with a copious film of sulfur species, indicating much more uniform phase transfer in SHE than in RE (Figure 2E). Apparently, as we discussed above, such controlled deposition was directed by the uniform spatial distribution of polysulfides when employing SHE. Such a deduction was further evidenced by the analogous deposition behavior of pure SHE with the absence of SMiPs (Figure S8). Nickel foam scaffold is favorable for SEM characterization. But the low surface area discourages its application in a practical Li−S cell. Herein, carbon nanotubes (CNTs), an already commercially available material with a cost lower than $100 kg−1, were employed to build flexible, mechanically robust, highly conductive, and hierarchical porous current collectors through a solution-processed method (see Supplemental Experimental Section). The crystalline structure, surface area, porosity, and mechanical flexibility were well examined (Figure S9−S11). SMiPs were cast onto the CNT current collector with an areal loading of 1.4 or 4.2 mg(Scathode) cm−2. To prevent the initial corrosion of lithium anodes by dissolved polysulfides and short circuits of assembled Li−S cells (Figure S12), RE was first injected in the anode chamber, and the fresh lithium anode could be passivated by the LiNO3. Cyclic voltammetry (CV) was performed to compare the electrochemical kinetics of SMiP cathodes in RE and SHE-1 M (Figure 3A). It is well understood that the pristine SMiP in RE manifests inferior kinetics as a result of poor dispersion and electrical contact, as indicated by the broad redox peaks and the low discharge voltage of the initial cycle. Such a sluggish kinetics is surprisingly propelled by SHE. The discharge potential shifts to a more positive position and the redox peak assigned to conversion of polysulfides to Li2S is more intensive. Interestingly, a third cathodic peak emerges below 1.9 V, which we attribute to the kinetically limited lithiation of bulk solid phase due to the huge size of SMiPs. In the second cycle,

the redox peak of sulfur-to-polysulfide transition (2.35 V) rises, implying the increase of electrochemically available sulfur phase. Such an increase can only result from the activation of bulk SMiPs in the first cycle. LiNO3 oxidation is excluded to account for the third anodic peak (Figure S13).48 The activation of bulk SMiPs is attributed to their partial solubilization in SHE since they could now access the bare conductive surface after entering solution phase. This hypothesis will be further validated in the next part. Galvanostatic charge−discharge curves illustrate this unique behavior as well (Figure 3B). The above emerging third redox peak is in good correspondence with the discharging slope below 1.9 V herein. The gradual downhill feature indicates the kinetic limitation that requires additional overpotential to drive the sluggish conversion to insulating bulk Li2S. The length of this slope decreases in the following cycles while the first plateau corresponding to the reduction of sulfur broadens. SMiP positive electrodes in both SHE-0.33 M and SHE-1 M possess much higher discharge capacities. And such an increase in capacity is mostly obtained from the second plateau, namely, the polysulfide−Li2S phase transfer. This improvement, again, accentuates the significance of preloaded polysulfides in mediating phase transfer and facilitating utilization of SMiPs. The biggest advantage afforded by SHE is the unprecedented shelf life of the SMiP cathode. As shown in Figure 3C, with little addition of 0.33 M [S] of Li2S5 in SHE, the cycling life of SMiP cathode was dramatically extended to 7500 cycles at a current density of 1.2 mA cm−2 while the initial capacity was 889 mAh g(Scell)−1. Besides the unparalleled cycling duration, the average Coulombic efficiency is above 99%, and the capacity decay rate is only 0.010% per cycle. Most significantly, the cycling life of SHE-0.33 M/SMiP was almost 2 orders of magnitude as that of RE/SMiP; while compared to RE/SNP, there was still at least 1 order of magnitude enhancement. The benchmark of battery cycling performance is 80% capacity retention per 1000 cycles.49 To reach this target, the concentration of polysulfides in SHE has to increase because in aforementioned model experiments, higher concentration has been proven more effective to improve phase transfer. Furthermore, the sulfur loading of SHE 0.33M/SMiP cathodes for 7500-cycle operation was set to 1.4 mg(Scathode) cm−2 only for fair comparison with RE/SMiP, which was incapable of being cycled with high sulfur loading. Although this loading value is comparable to many previously reported sulfur cathodes with 1000-cycle life or more, much higher sulfur loading above 4 mg cm−2 must be launched to enhance the energy density of working cells as discussed below.50 Based on these considerations, we built more advanced Li−S batteries with a “self-healing” feature virtually, no matter upon long-term cycling or under static storage. SHE-1 M (with a sulfur amount in solution phase equivalent to 1.4 mg(Selectrolyte) cm−2) and SMiP positive electrode with a sulfur loading of 4.2 mg(Scathode) cm−2 were herein employed (see Supplemental Experimental Section for details). Operated under various current densities, cells based on SHE-1 M/SMiP all exhibited similar “self-healing” cycling behaviors as depicted below. The initial capacity decay was followed by the restoration to initial value or even higher (Figure 3D). Interestingly, this self-healing behavior can be correlated to cycling duration rather than cycle numbers, as revealed by nearly monotonously increased cycle numbers needed for capacity recovery as a function of current density (Figure S14). At a relatively high current density of 1.5 A g(Scell)−1, a high capacity of ∼3.7 mAh cm−2 can be retained 8462

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Article

Journal of the American Chemical Society

Figure 4. Structural evolution of SMiPs with SHE. SEM images of (A) pristine SMiPs/CNT current collector, (B) cycled RE/SMiPs for 120 cycles at a current density of 0.7 mA cm−2, (C) initial SHE-1 M/SMiPs before cycling, cycled SHE-1 M/SMiPs for (D) 20 and (E) 120 cycles at a current density of 0.7 mA cm−2, and (F) for 2000 cycles at a current density of 8.5 mA cm−2. The scale bars in panels A−C, D, E, and F are 1 μm, 300 nm, 10 μm, and 1 μm, respectively. (G) XRD patterns of sulfur, CNTs, SMiPs/CNTs, RE/SMiPs, and SHE-1 M/SMiPs. (H) Electrochemical impedance spectra (EIS) of initial RE/SMiPs, initial SHE-1 M/SMiPs, cycled RE/SMiPs for 120 cycles, and cycled SHE-1 M/SMiPs for 20/120 cycles. The applied current density is 0.7 mA cm−2.

additional discharge slope below 1.9 V was observed. At a higher current density (1.5 A g(Scell)−1), however, the inner phase hardly exhibited any activity, and consequently the initial capacity was much lower. Upon cycling, the inert bulk phase became electrochemically available since it was gradually exposed to the SHE and solubilized into polysulfides. During subsequent cycling, the spatial heterogeneity was also mitigated by SHE because of its durability to concentration changes induced by sulfur/Li2S solubilization. Simultaneously, the overpotential concomitantly decreased, evidencing the mitigated heterogeneity (Figure S2). Additional analysis of the capacity contributions of different phase-transfer processes provided extra evidence. The ratio of capacities contributed by polysulfide−Li2S conversion (QII) and sulfur−polysulfide conversion (QI) decreased at the early cycling stage but increased significantly upon further cycling in SHE, indicating the enhancement of polysulfide−Li2S phasetransfer (Figure S19). On the contrary, RE/SMiPs underwent continuous decay of both QI and QII (Figure S20). Before the spatial heterogeneity of electrochemically generated polysulfides was eliminated, substantial active materials had lost the electrical contact. Therefore, unlike initially preexisting polysulfides in SHE, the amount of newly produced polysulfides through electrochemical reactions was too low to

after 2000 cycles, which exhibited no decay compared to its initial value and was fairly comparable to the benchmarks of batteries (Figure S15A). Even after a static operation for 72 h, the capacities manifested no decay but slight increase at both low and high current densities, further indicating the unique self-healing feature (Figure S15). More interestingly, a cycled Li−S cell with SMiP/RE exhibited restored capacity and stable cycling after being injected with the SHE, suggesting an unambiguous healing capability of the SHE (Figure S16). Upon replacement of CNTs with commercial carbon black networks, the stable cycling behavior was still attained, but the capacity was much lower (Figure S17). If a highly efficient “sulfiphilic” host was employed to support SMiP/SHE, further enhancement in sulfur utilization can be expected.51−54 We attribute the self-healing merit to the polysulfidemediated phase transfer since such a phenomenon is plausibly electrochemically independent, as indicated by the nearly linear correlation between self-healing duration and current density (Figure S14). The evolution of galvanostatic discharge−charge curves under both low and high current densities further validated the mediated phase-transfer process by SHEs (Figure S18). As a result of the large size of pristine SMiPs, the inner phase of bulk SMiPs was only electrochemically active under extremely low current density (0.13 A g(Scell)−1), at which an 8463

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Article

Journal of the American Chemical Society

driven buffering effect has intrinsic weakness owing to periodic changes in polysulfide concentration. In this content, however, we highlight the role of spatial distribution of intermediates (e.g., polysulfides) in modulating the phase transfer in conversion-type electrochemical reactions, which coincidently agrees with the effect in insertion-type electrochemical reactions. Very recently, Chueh et al. explored how spatial heterogeneities in reaction rates accounted for the decayed performance of a LiFePO4 cathode through operando X-ray techniques.58 Nevertheless, the analogous effect of spatial heterogeneity was rarely investigated in Li−S batteries or other conversion reaction-based systems. Increasing the concentration of preloaded polysulfides in SHEs, which provides more reduced spatial variation, significantly promotes the mediation of solution−solid phase transfer and thus stabilizes the positive electrode upon cycling whether based on SMiPs or catholyte (Figure S25A,B). Generally, the SMiP/SHE exhibited higher specific capacity than its catholyte counterpart with the same concentration of polysulfides. This excludes the possibility that the capacity is simply restricted by the limited conductive surface or porosity.59,60 Moreover, there are 22.5−56.6% synergistic increases in overall capacities of SMiPs if the contribution from bare SHEs is subtracted (Figure S26). Actually the net contribution of SMiP to SMiP/SHE still exhibited much improved capacity and stability than RE/SMiP (Figure S25C). Therefore, adding polysulfides is neither simply a compensation for the capacity loss nor interference to the overall electrochemical performance. In sum, our strategy sheds new light on the fundamentals of the crucial phase transfer process and provides an unambiguous method to control it rationally, that is, through homogenizing the spatial distribution of precursor monomers. The main difference between preloaded and electrochemically generated polysulfides lies in how they are initially distributed. Again, the initial spatial homogeneity matters as it dominates the electrodeposition behaviors in subsequent cycles. Besides, initially uniform deposition also guarantees that the amount of electrochemically generated polysulfides during subsequent cycling is sufficiently high to solubilize deactivated solid phases. Further modification on other components in Gibbs−Thomson theory (e.g., surface tensile γ in eqs 1−3 in the Supporting Information) will be the next step. Cui et al. have already illustrated such an opportunity for spatially controlling the deposition of discharge products through the employment of both a hybrid polar scaffold and polysulfide catholytes.52 To fully demonstrate the concept of engineering an extrinsic healing agent to mediate the phase transfer and to differ this work from a bunch of previous ones where only polysulfides were employed, we also explored other healing agents for highly stable Li−S batteries. Here the redox couple of LiI/I2 was exerted. The model experiments indicated that I2 was able to rapidly oxidize the bulk Li2S to highly colloidal sulfur particles (I2 + Li2S = 2LiI + S), further promoting the subsequent solubilization of Li2S into polysulfides (Li2S + (x − 1)S = Li2Sx) (Figure 5A). The I2 was generated by charging the cell to 3.2 V, where part of the electrochemical oxidation (charge capacity) was possibly offset by the chemical reduction of I2 by isolated Li2S (Figure 5B). Consequently, analogous self-healing ability was attained (Figure 5C). Although some previous reports have grafted the LiI/I2 redox couple to Li−S systems,61,62 none indicated such a desirable self-healing effect. Unlike previous reports that only focused on polysulfide

restore the irreversible sediments. Previous reports showed that constraining either discharge or charge capacity to a moderate but constant value (500−1000 mAh g−1) can also lead to enhancement in cycling performance.55,56 Constant-capacity cycling performance at 500 mAh g(Scell)−1, which was lower than that obtained on SMiP/SHE-1 M, was also evaluated using RE (Figure S21). Nevertheless, the shelf life is much shorter than that of SMiP/SHE, suggesting that the limited capacity is not the reason accounting for the exceptional self-healable cycling of SMiP/SHE-1 M. Postmodern analysis offers further insights into structural evolution of SHE/SMiPs upon cycling. The pristine SMiP positive electrode clearly possesses an unfavorable distribution of sulfur with large size ranging from 1 to 5 μm on the porous CNT current collector, with a large quantity of CNTs left uncoated (Figure 4A). X-ray diffraction (XRD) pattern also confirms the existence of large crystalline domain of SMiPs (Figure 4G). After cycling in RE for 120 cycles at a current density of 0.7 mA cm−2, SMiP cathodes were covered by a dense and impermeable layer with no exposure of underneath CNT current collector. This therefore suggests the irreversible deposition and aggregation of solid sulfur compounds (Figure 4B). In contrast, the SMiP exhibited favorable decrease in sulfur particle size and the CNT network was somewhat coated by solid deposits after being immersed in the SHE for only 24 h, which was ascribed to the partial solubilization and redistribution of bulk SMiPs (Figure 4C). The dissolution of bulk phase can be verified by the absence of X-ray reflection of sulfur in SHE-1/SMiP (Figure 4G). After 20 cycles, it was found that LixS NPs were uniformly deposited and tightly attached to the outer surface of CNTs, which was in sharp contrast to its pristine microsized morphology (Figure 4D). Unlike RE/SMiPs, the SHE/SMiP during subsequent cycling did not display severe aggregation of solid phases and blockage of electron/ion pathways (Figure 4E). Even extremely longterm cycling of over 2000 cycles with a larger current density only gives rise to the formation of copious solid sulfur species whereas the porous CNT architecture was preserved and still observable (Figure 4F). Compared to REs, SHEs also left a positive impact on the morphology of lithium anodes and rendered a LiF-rich solid electrolyte interphase, which further strengthened the battery performance (Figure S22, S23).57 Nevertheless, to differentiate the effects of polysulfides on the cathode and anode, the Li−S cell with RE and a SHE-treated lithium anode was evaluated. Stabilized cycling after initial capacity decay but no capacity healing was obtained (Figure S24). EIS was employed to monitor the changes in cell inner resistance related to the structural and electrochemical evolution of SMiPs upon cycling (Figure 4H). It was obvious that with SHE, SMiPs underwent a dramatic decrease in charge transfer resistance, which was attributed to the gradual dissolution of electrochemically inert bulk phase of SMiPs and the enhanced electrical contact consequently. After 120 cycles, the newly emerging semicircle in the Nyquist plot of RE/SMiPs indicated the formation of another resistive phase, which was in good accordance with that suggested by SEM image (Figure 4B). Since polysulfides are naturally produced in Li−S batteries, it is very interesting to probe the difference between preloaded and electrochemically generated polysulfides. Though the selfhealing role of polysulfides has been proposed previously, the so far commonly accepted explanation based on equilibrium8464

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Journal of the American Chemical Society



Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.6b12358. Complete experimental details, additional characterizations, and discussions of implementation of Gibbs− Thomson effect and ex situ UV−vis analysis (PDF) Video demonstrating spatial heterogeneities (AVI)



AUTHOR INFORMATION

Corresponding Author

*[email protected] ORCID

Qiang Zhang: 0000-0002-3929-1541 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the National Key Research and Development Program (No. 2016YFA0202500) and Natural Scientific Foundation of China (Nos. 21422604, 21676160, and 21561130151). The authors appreciate helpful discussion from Ze-Wen Zhang and Ge Zhang.



REFERENCES

(1) Tee, B. C. K.; Wang, C.; Allen, R.; Bao, Z. Nat. Nanotechnol. 2012, 7, 825−832. (2) Wang, C.; Wu, H.; Chen, Z.; McDowell, M. T.; Cui, Y.; Bao, Z. Nat. Chem. 2013, 5, 1042−1048. (3) Diesendruck, C. E.; Sottos, N. R.; Moore, J. S.; White, S. R. Angew. Chem., Int. Ed. 2015, 54, 10428−10447. (4) Armand, M.; Tarascon, J. M. Nature 2008, 451, 652−657. (5) Bruce, P. G.; Freunberger, S. A.; Hardwick, L. J.; Tarascon, J.-M. Nat. Mater. 2012, 11, 19−29. (6) Chan, C. K.; Peng, H.; Liu, G.; McIlwrath, K.; Zhang, X. F.; Huggins, R. A.; Cui, Y. Nat. Nanotechnol. 2008, 3, 31−35. (7) Evers, S.; Nazar, L. F. Acc. Chem. Res. 2013, 46, 1135−1143. (8) Yin, Y.-X.; Xin, S.; Guo, Y.-G.; Wan, L.-J. Angew. Chem., Int. Ed. 2013, 52, 13186−13200. (9) Zhang, S. S. J. Power Sources 2013, 231, 153−162. (10) Zhou, W.; Yu, Y.; Chen, H.; DiSalvo, F. J.; Abruna, H. D. J. Am. Chem. Soc. 2013, 135, 16736−16743. (11) Manthiram, A.; Fu, Y.; Chung, S.-H.; Zu, C.; Su, Y.-S. Chem. Rev. 2014, 114, 11751−11787. (12) Pang, Q.; Liang, X.; Kwok, C. Y.; Nazar, L. F. Nat. Energy 2016, 1, 16132. (13) Borchardt, L.; Oschatz, M.; Kaskel, S. Chem. - Eur. J. 2016, 22, 7324−7351. (14) Seh, Z. W.; Sun, Y.; Zhang, Q.; Cui, Y. Chem. Soc. Rev. 2016, 45, 5605−5634. (15) Peng, H.-J.; Zhang, Z.-W.; Huang, J.-Q.; Zhang, G.; Xie, J.; Xu, W.-T.; Shi, J.-L.; Chen, X.; Cheng, X.-B.; Zhang, Q. Adv. Mater. 2016, 28, 9551−9558. (16) Ji, X.; Lee, K. T.; Nazar, L. F. Nat. Mater. 2009, 8, 500−506. (17) Ji, L.; Rao, M.; Zheng, H.; Zhang, L.; Li, Y.; Duan, W.; Guo, J.; Cairns, E. J.; Zhang, Y. J. Am. Chem. Soc. 2011, 133, 18522−18525. (18) Xin, S.; Gu, L.; Zhao, N.-H.; Yin, Y.-X.; Zhou, L.-J.; Guo, Y.-G.; Wan, L.-J. J. Am. Chem. Soc. 2012, 134, 18510−18513. (19) Guo, J.; Yang, Z.; Yu, Y.; Abruna, H. D.; Archer, L. A. J. Am. Chem. Soc. 2013, 135, 763−767. (20) Wei Seh, Z. W.; Li, W.; Cha, J. J.; Zheng, G.; Yang, Y.; McDowell, M. T.; Hsu, P.-C.; Cui, Y. Nat. Commun. 2013, 4, 1331. (21) Zhao, M.-Q.; Zhang, Q.; Huang, J.-Q.; Tian, G.-L.; Nie, J.-Q.; Peng, H.-J.; Wei, F. Nat. Commun. 2014, 5, 3410.

Figure 5. LiI/I2 redox couple as extrinsic healing agents for Li−S batteries. (A) Model experiment showing the reaction between I2 and Li2S. The Li2S was rapidly oxidized by I2 (reaction i), while the excess Li2S completely consumed the I2 and some Li2S residue remained, which was then dissolved with solution turning light yellow simultaneously (reaction ii in which polysulfides were generated). The model experiments illustrated the fast response of I2-mediated solubilization of Li2S. (B) Galvanostatic discharge−charge profile of SMiPs with 0.05 M LiI added in RE. (C) Cycling performance of SMiP/RE with or without 0.05 M LiI addition at a current density of 0.7 mA cm−2.

additives, the general understanding gained from crystal growth theory is inspiring and extensible for the design of various extrinsic healing agents that meet the desirable attributes such as self-generativity and phase-transfer tunability.

3. CONCLUSIONS We proposed a healing approach inspired by blood vessels to shed light on electrochemical energy storage systems that involve complex phase transfer. Taking micrometer-sized sulfur-based positive electrodes as an example, we then well demonstrated its fundamentals and advantages. The kernel lies in the mediation of phase transfer by the extrinsic healing agent. This strategy also guides the design and application of other healing agents for Li−S batteries. Although it is extremely challenging to mimic the high efficiency, specificity, or precise controllability of biological processes, novel healing agents that are smart, sustainable, and rapidly responsive hold future promises. This innovative approach can also be feasibly and facilely implemented in other high-energy electrochemical storage/conversion systems such as metal−O2 batteries and fuel cells. 8465

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466

Article

Journal of the American Chemical Society (22) Li, Z.; Jiang, Y.; Yuan, L.; Yi, Z.; Wu, C.; Liu, Y.; Strasser, P.; Huang, Y. ACS Nano 2014, 8, 9295−9303. (23) Wang, Z.; Dong, Y.; Li, H.; Zhao, Z.; Wu, H. B.; Hao, C.; Liu, S.; Qiu, J.; Lou, X. W. Nat. Commun. 2014, 5, 5002. (24) Wu, F. X.; Lee, J. T.; Fan, F. F.; Nitta, N.; Kim, H.; Zhu, T.; Yushin, G. Adv. Mater. 2015, 27, 5579−5586. (25) Song, J.; Gordin, M. L.; Xu, T.; Chen, S.; Yu, Z.; Sohn, H.; Lu, J.; Ren, Y.; Duan, Y.; Wang, D. Angew. Chem., Int. Ed. 2015, 54, 4325− 4329. (26) Liang, J.; Sun, Z.-H.; Li, F.; Cheng, H.-M. Energy Storage Mater. 2016, 2, 76−106. (27) Huang, J.-Q.; Zhang, Q.; Wei, F. Energy Storage Mater. 2015, 1, 127−145. (28) Manthiram, A.; Chung, S.-H.; Zu, C. Adv. Mater. 2015, 27, 1980−2006. (29) Zhou, G.; Pei, S.; Li, L.; Wang, D.-W.; Wang, S.; Huang, K.; Yin, L.-C.; Li, F.; Cheng, H.-M. Adv. Mater. 2014, 26, 625−631. (30) Chung, S.-H.; Han, P.; Singhal, R.; Kalra, V.; Manthiram, A. Adv. Energy Mater. 2015, 5, 1500738. (31) Balach, J.; Jaumann, T.; Klose, M.; Oswald, S.; Eckert, J.; Giebeler, L. Adv. Funct. Mater. 2015, 25, 5285−5291. (32) Bai, S.; Liu, X.; Zhu, K.; Wu, S.; Zhou, H. Nat. Energy 2016, 1, 16094. (33) Xu, R.; Belharouak, I.; Li, J. C. M.; Zhang, X.; Bloom, I.; Bareno, J. Adv. Energy Mater.o 2013, 3, 833−838. (34) Chen, S.; Dai, F.; Gordin, M. L.; Wang, D. RSC Adv. 2013, 3, 3540−3543. (35) Agostini, M.; Scrosati, B.; Hassoun, J. Adv. Energy Mater. 2015, 5, 1500481. (36) Lee, S. K.; Oh, S. M.; Park, E.; Scrosati, B.; Hassoun, J.; Park, M. S.; Kim, Y. J.; Kim, H.; Belharouak, I.; Sun, Y. K. Nano Lett. 2015, 15, 2863−2868. (37) Li, W.; Yao, H.; Yan, K.; Zheng, G.; Liang, Z.; Chiang, Y.-M.; Cui, Y. Nat. Commun. 2015, 6, 7436. (38) Zhao, C.-Z.; Cheng, X.-B.; Zhang, R.; Peng, H.-J.; Huang, J.-Q.; Ran, R.; Huang, Z.-H.; Wei, F.; Zhang, Q. Energy Storage Mater. 2016, 3, 77−84. (39) Yan, C.; Cheng, X.-B.; Zhao, C.-Z.; Huang, J.-Q.; Yang, S.-T.; Zhang, Q. J. Power Sources 2016, 327, 212−220. (40) Lee, D.-J.; Agostini, M.; Park, J.-W.; Sun, Y.-K.; Hassoun, J.; Scrosati, B. ChemSusChem 2013, 6, 2245−2248. (41) Van Gaal, L. F.; Mertens, I. L.; De Block, C. E. Nature 2006, 444, 875−880. (42) Xia, Y.; Xiong, Y.; Lim, B.; Skrabalak, S. E. Angew. Chem., Int. Ed. 2009, 48, 60−103. (43) Fan, F. Y.; Carter, W. C.; Chiang, Y.-M. Adv. Mater. 2015, 27, 5203−5209. (44) Xu, R.; Lu, J.; Amine, K. Adv. Energy Mater. 2015, 5, 1500408. (45) Nelson, J.; Misra, S.; Yang, Y.; Jackson, A.; Liu, Y.; Wang, H.; Dai, H.; Andrews, J. C.; Cui, Y.; Toney, M. F. J. Am. Chem. Soc. 2012, 134, 6337−6343. (46) Cuisinier, M.; Cabelguen, P.-E.; Evers, S.; He, G.; Kolbeck, M.; Garsuch, A.; Bolin, T.; Balasubramanian, M.; Nazar, L. F. J. Phys. Chem. Lett. 2013, 4, 3227−3232. (47) Xiao, J.; Hu, J. Z.; Chen, H.; Vijayakumar, M.; Zheng, J.; Pan, H.; Walter, E. D.; Hu, M.; Deng, X.; Feng, J.; Liaw, B. Y.; Gu, M.; Deng, Z. D.; Lu, D.; Xu, S.; Wang, C.; Liu, J. Nano Lett. 2015, 15, 3309−3316. (48) Aurbach, D.; Pollak, E.; Elazari, R.; Salitra, G.; Kelley, C. S.; Affinito, J. J. Electrochem. Soc. 2009, 156, A694−A702. (49) Lu, L.; Han, X.; Li, J.; Hua, J.; Ouyang, M. J. Power Sources 2013, 226, 272−288. (50) Peng, H.-J.; Huang, J.-Q.; Cheng, X.-B.; Zhang, Q. Adv. Energy Mater. 2017, 7, 1700260. (51) Pang, Q.; Kundu, D.; Cuisinier, M.; Nazar, L. F. Nat. Commun. 2014, 5, 4759. (52) Yao, H.; Zheng, G.; Hsu, P.-C.; Kong, D.; Cha, J. J.; Li, W.; Seh, Z. W.; McDowell, M. T.; Yan, K.; Liang, Z.; Narasimhan, V. K.; Cui, Y. Nat. Commun. 2014, 5, 3943.

(53) Yuan, Z.; Peng, H.-J.; Hou, T.-Z.; Huang, J.-Q.; Chen, C.-M.; Wang, D.-W.; Cheng, X.-B.; Wei, F.; Zhang, Q. Nano Lett. 2016, 16, 519−527. (54) Peng, H.-J.; Zhang, G.; Chen, X.; Zhang, Z.-W.; Xu, W.-T.; Huang, J.-Q.; Zhang, Q. Angew. Chem., Int. Ed. 2016, 55, 12990− 12995. (55) Zheng, J.; Gu, M.; Wang, C.; Zuo, P.; Koech, P. K.; Zhang, J.-G.; Liu, J.; Xiao, J. J. Electrochem. Soc. 2013, 160, A1992−A1996. (56) Su, Y.-S.; Fu, Y.; Cochell, T.; Manthiram, A. Nat. Commun. 2013, 4, 2985. (57) Lu, Y.; Tu, Z.; Archer, L. A. Nat. Mater. 2014, 13, 961−969. (58) Lim, J.; Li, Y.; Alsem, D. H.; So, H.; Lee, S. C.; Bai, P.; Cogswell, D. A.; Liu, X.; Jin, N.; Yu, Y.-s.; Salmon, N. J.; Shapiro, D. A.; Bazant, M. Z.; Tyliszczak, M. Z.; Chueh, W. C. Science 2016, 353, 566−571. (59) Fu, Y.; Su, Y.-S.; Manthiram, A. Angew. Chem., Int. Ed. 2013, 52, 6930−6935. (60) Zhou, G.; Paek, E.; Hwang, G. S.; Manthiram, A. Nat. Commun. 2015, 6, 7760. (61) Wu, F. X.; Lee, J. T.; Nitta, N.; Kim, H.; Borodin, O.; Yushin, G. Adv. Mater. 2015, 27, 101−108. (62) Chen, H. N.; Lu, Y. C. Adv. Energy Mater. 2016, 6, 1502183.

8466

DOI: 10.1021/jacs.6b12358 J. Am. Chem. Soc. 2017, 139, 8458−8466