Engineering Solid Electrolyte Interphase for

0 downloads 0 Views 4MB Size Report
Apr 17, 2018 - between surface properties and electrochemical performance are still not ..... bonate (DMC) electrolyte with 1 wt% fluoroethylene carbonate.
FULL PAPER Sodium Ion Batteries

www.afm-journal.de

Engineering Solid Electrolyte Interphase for Pseudocapacitive Anatase TiO2 Anodes in Sodium-Ion Batteries Zheng-Long Xu, Kyungmi Lim, Kyu-Young Park, Gabin Yoon, Won Mo Seong, and Kisuk Kang* and conversion reactions are capable of delivering high specific capacities, the huge volume expansion induced by Na ion insertion causes structural degradation and poor cyclic stability. The carbonaceous materials, especially hard carbon, always show very low Na storage potential, inducing the challenging issue of Na dendrite formation during cycles. TiO2 materials have also been proposed as notable anodes for SIBs, due to their intriguing properties of the large sodium storage capacities, the cost-effectiveness, nontoxicity, and natural abundancy of Ti-based oxides. More importantly, TiO2 anodes are expected to present excellent cyclic stability based on intercalation reactions. While various polymorphic TiO2 phases have been explored such as anatase, rutile, brookite, and hollandite,[12,13] the anatase TiO2 is particularly appealing due to the suitable sized pathways (3.725 Å × 3.785 Å) for sodium ion diffusion, a moderate working potential of about 0.8 V versus Na+/Na and a high theoretical sodium storage capacity.[14–16] During the past decade, extensive efforts have been devoted to boosting the electrochemical performance of anatase TiO2 anodes in SIBs, through designing and preparing various TiO2 nanostructures, such as nanorods,[17,18] nanotubes,[19] nanospheres,[20,21] nanofibers,[22] as well as TiO2/carbon nanocomposites including N-doped yolk-shell carbon/TiO2[23] and carbon coated TiO2 nanostructures.[24–26] While the electrochemical performance could be enhanced from these approaches, the fundamental understanding on the electrochemical behaviors of the anatase TiO2 with sodium still remain elusive with controversial statements in literature, which puts a vital obstacle on the way toward developing high performance TiO2 anodes. Up to now, several sodium storage mechanisms have been proposed such as i) reversible insertion/extraction of Na ions from anatase TiO2 with a minimal lattice change,[17] ii) irreversible amorphization and the subsequent conversion reaction of TiO2 to metallic Ti and NaO2[16] or Na2CO3,[27] and iii) formation of disordered layered-like NaxTiO2 phase after the first sodiation and reversible phase transition between amorphous TiO2 and NaxTiO2 in following cycles.[28] Although these claims were supported by valid evidences from respectively different

Anatase TiO2 is considered as one of the promising anodes for sodium-ion batteries because of its large sodium storage capacities with potentially low cost. However, the precise reaction mechanisms and the interplay between surface properties and electrochemical performance are still not elucidated. Using multimethod analyses, it is herein demonstrated that the TiO2 electrode undergoes amorphization during the first sodiation and the amorphous phase exhibits pseudocapacitive sodium storage behaviors in subsequent cycles. It is also shown that the pseudocapacitive sodium storage performance is sensitive to the nature of solid electrolyte interphase (SEI) layers. For the first time, it is found that ether-based electrolytes enable the formation of thin (≈2.5 nm) and robust SEI layers, in contrast to the thick (≈10 nm) and growing SEI from conventional carbonate-based electrolytes. First principle calculations suggest that the higher lowest unoccupied molecular orbital energies of ether solvents/ion complexes are responsible for the difference. TiO2 electrodes in ether-based electrolyte present an impressive capacity of 192 mAh g−1 at 0.1 A g−1 after 500 cycles, much higher than that in carbonate-based electrolyte. This work offers the clarified picture of electrochemical sodiation mechanisms of anatase TiO2 and guides on strategies about interfacial control for high performance anodes.

1. Introduction Driven by the concerns of limited lithium resources, sodiumion battery (SIB) technologies have taken the privilege over lithium-ion batteries (LIBs) on meeting the demanding applications of large-scale energy storage systems, due to the inexhaustible sodium resources on earth crust and in sea water.[1–4] With respect to the anode materials for SIBs, a number of potential materials have been explored, including metal (oxides/chalcogenides),[5] phosphorus,[6] and carbonaceous materials.[7–11] Although the anode materials undergoing alloying/dealloying Dr. Z.-L. Xu, K. Lim, Dr. K.-Y. Park, G. Yoon, W. M. Seong, Prof. K. Kang Department of Materials Science and Engineering Research Institute of Advanced Materials (RIAM) Seoul National University 1 Gwanak-ro, Gwanak-gu, Seoul 151-742, Republic of Korea E-mail: [email protected] The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adfm.201802099.

DOI: 10.1002/adfm.201802099

Adv. Funct. Mater. 2018, 1802099

1802099  (1 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

characterization tools, a consensus on the reaction mechanisms has not been obtained. Therefore, a further study on the sodiation/desodiation behaviors of anatase TiO2 with comprehensive characterizations is necessary, which is a prerequisite for optimizing the electrochemical performance of TiO2 anodes. In dealing with the nanosize electrode materials, a critical factor to consider in achieving the high performance is the surface property and its compatibility with the electrolyte including a stable solid electrolyte interphase (SEI) formation. In LIBs, TiO2 has been considered as an SEI-free anode because the operating potential of TiO2 is ≈1.7 V (vs Li+/Li), which is far higher than the reduction potential of most electrolyte systems.[29] Thus, there has been no strong motivation to study the SEI for TiO2 anodes in LIBs. However, in SIBs, TiO2 electrodes present an initial discharge plateau at 0.2 V (vs Na+/Na) and an average working voltage of 0.8 V in the following cycles.[16,17,27,28] Moreover, to fully utilize the sodium storage capacity of TiO2 anodes, the cutoff voltages are usually set at 0.01 V (vs Na+/Na).[19–24] These low potentials are ready to induce the decomposition of electrolytes and trigger the formation of SEI layers on TiO2 surface.[21,30,31] It should be noted that the undesirable accumulation of SEI would cause a high polarization, low Coulombic efficiencies, and a rapid capacity degradation.[15,21] In this respect, it would be critically important to design reliable SEI layers for TiO2 anodes in SIBs, unlike the case of TiO2 electrodes in LIBs. Nevertheless, the understanding on the relationship between the SEI property and the electrochemical performance of TiO2 anodes in SIBs is still elusive and has not been systematically studied so far. To address the above challenging issues, this work is dedicated to examining the sodiation/desodiation behaviors of anatase TiO2 using a suite of characterizations such as synchrotron X-ray diffraction (XRD), high resolution transmission electron microscopy (HRTEM), ex situ electrochemical impedance spectroscopy (EIS), and X-ray absorption structure (XAS) techniques. We find that the anatase TiO2 is amorphized during the first sodiation and maintains the amorphous nature during following cycles. The amorphous phases are revealed to display evident pseudocapacitive sodium storage behaviors. It is further demonstrated that the pseudocapacitive reaction can be facilitated by engineering the nature of SEI layer; ether-based electrolytes enable the formation of thinner and more stable SEI layers than conventional carbonate-based electrolytes, whose origin is elucidated through the first principles calculations. We show that the TiO2 electrodes with engineered SEI present an exceptional cyclic stability and high rate capabilities. This new understanding on the reaction mechanisms and the interfacial chemistry is expected to expedite the development of the high performance TiO2 anodes for SIBs.

2. Results and Discussion 2.1. Sodiation/Desodiation Mechanism To investigate the phase evolution of TiO2 anodes during sodiation/desodiation processes, in situ synchrotron XRD experiment was conducted using a pin–hole half-cell containing TiO2 electrode, sodium foil counter electrode, and 100 µL 1 m NaPF6 Adv. Funct. Mater. 2018, 1802099

diethylene glycol dimethyl ether (DEGDME) electrolyte.[32] XRD patterns were periodically collected in every ≈3 min when the electrochemical cell was discharged/charged at a constant current density of 40 mA g−1. During discharging from the open circuit voltage to 0.5 V, there was almost no change of the major peak for the TiO2 (101) with respect to the intensity and shape (Figure 1a), indicating the retention of the pristine tetragonal anatase phase. The absence of the bulk structural evolution suggests that the capacity in this region can be attributed to the formation of SEI layers and/or the absorption of sodium ions on the surface of TiO2 nanoparticles.[15,16,33] During discharging from 0.25 to 0.01 V, it was evidently observed that the (101) peak gradually vanished, implying the loss of the long-range ordering in the anatase phase with the sodium insertion.[16] The marginal blueshift of the (101) peak was observed even though it was not nonambiguous due to the substantial broadening of the XRD peak.[15,16] When charging back to 3.0 V, the tetragonal anatase phase was not recovered, illustrating an irreversible amorphization of TiO2 crystals during the first sodiation. To complement the bulk structural information provided by XRD patterns, HRTEM images of individual TiO2 nanoparticles were examined at different reaction stages (Figure 1b). The ex situ HRTEM images for samples during discharging from 0.25 to 0.01 V show that the crystalline domain of TiO2 gradually shrinks from the shell to the core of the particle, and the crystalline domain size gradually decreases from ≈20 nm to negligible size, consistent with the attenuation of (101) peak in XRD patterns and the amorphization of TiO2 crystals during the first sodiation. Some residual ultrafine TiO2 crystal (≈1.6 nm) in the core of the particle at the end of the discharge is ascribed to the incomplete sodiation at relatively high current rates. It is noted that further studies, such as in situ transmission electron microscopy (TEM) examination,[34] are still needed to clarify the detailed sodiation/desodiation processes of an individual TiO2 particle. We could not observe any noticeable lattice expansion of the (101) plane in HRTEM images during the sodium ion insertion before the amorphization in consistent with the XRD results. The charging of the TiO2 electrode to 3.0 V did not recover the crystalline anatase TiO2 phase according to the HRTEM image, in agreement with the XRD result of the irreversible amorphization. In order to confirm the observed phase evolution and exclude the influence of electrolyte types, we repeated the same experiments in other electrolytes, i.e., 1 m NaPF6 dissolved in dimethyl ether (DME), ethylene carbonate: diethyl carbonate (EC/DEC) and ethylene carbonate: propylene carbonate (EC/PC) solvents. Morphological and structural characterizations reveal that anatase TiO2 electrodes also present irreversible amorphization during the initial sodiation/desodiation processes in EC/PC electrolyte (Figure S1, Supporting Information), in full agreement with the previous report.[16] The additional two electrodes exhibited similar trend of forming the amorphous phases after cycling (Figure S1, Supporting Information), verifying the amorphization of TiO2 crystals during sodiation/desodiation processes regardless of the electrolyte types. To further investigate the local bonding nature of the amorphized electrode material during discharge/charge, we carried out Ti K-edge X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) experiments. In the pre-edge region (4968–4977 eV) of XANES spectra, the

1802099  (2 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

Figure 1. Sodium storage mechanisms of TiO2 electrodes: a) in situ synchrotron XRD analysis of the phase evolution of TiO2 electrodes, b) ex situ HRTEM images of individual TiO2 nanoparticles at different discharge/charge voltages, c) Ti-K edge XANES spectra of TiO2 during the first two cycles, d) FT magnitude of k3-weighted EXAFS spectra for TiO2 electrodes, e) log-scale plots of peak current versus scan rate to determine b values, f) the capacitive and diffusion controlled contributions to Na storage in TiO2 at a scan rate of 2 mV s−1, and g) contribution ratios of the capacitive controlled charge at different scan rates.

pristine TiO2 electrode presented three dominant peaks (inset Figure 1c), namely, P1 at 4968.6 eV, P3 at 4971.7 eV, and P4 at 4974.1 eV, which refer to the quadrupole transition from Ti1s to t2g states, the dipole transitions from Ti1s to p states accompanied with the t2g states of Ti neighbor and eg states of Ti absorber, and the diploe transition from Ti1s to p states with the eg states of Ti neighbor, respectively.[27,35–37] After the first sodiation, however, a new peak P2 at 4970.7 eV was observed, Adv. Funct. Mater. 2018, 1802099

which was identified as a typical feature of amorphous sodium titanate.[36] Even after the desodiation, the electrode presented a broad peak near to P2, similar to the feature of amorphous TiO2 in the literature but has not been unambiguously interpreted.[27] The absence of the fingerprint of the anatase TiO2 after the first charge process supports the statement of the irreversible amorphization of the TiO2 electrode with the sodiation. During the subsequent discharge/charge cycle, only red/blue shifts

1802099  (3 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

of the broad peaks near P2 were observable, referring to Ti4+/ Ti3+ redox reactions of the amorphous phases, which could be also supported by X-ray photoelectron spectroscopy (XPS) results (Figure S2, Supporting Information). The Fourier-transformed (FT) magnitude derived from EXAFS could present a more detailed information on the short-range orderings in TiO2 during charge and discharge as shown in Figure 1d.[38,39] After the first discharge, the peak at 1.50 Å corresponding to the first neighbor TiO bonds in the pristine TiO2 splits into double peaks at 1.35 and 1.83 Å, implying the coexistence of two different TiO bonding natures in the first shell of Ti ions. The splitting of the first TiO peak was also previously observed in lithiated TiO2, which was explained as the distortion of local structure of Ti sites and the mixed states of Ti4+-O and Ti3+-O.[35,37] After the first charge, the first TiO and TiTi peak positions recovered to the same positions as those for pristine TiO2, suggesting the similar local structure of the charged product as the anatase TiO2. However, significantly weaker signals of the first neighbor peaks and the loss of long-range order in the region from 3 to 6 Å demonstrate that the charged material is highly amorphous,[28] consistent with the XRD and HRTEM analyses. The electrochemical response of the cycled TiO2 electrode was examined to further understand its structure after the amorphization by the cyclic voltammetry (CV) scans from 0.1 to 5 mV s−1 (Figure S3, Supporting Information). The redox peak currents (i) in the CV curves obey a power-law relationship with the scan rate (v), which shows an almost linear proportionality of i  = avb with b-values of about 0.93 at anodic and cathodic peaks (Figure 1e), indicating that the sodium storage mechanisms rely primarily on capacitive (b = 1.0) reaction other than diffusion-controlled (b = 0.5) reaction. The ratio between these two contributions at a certain scan rate, i.e., 2 mV s−1, can be quantitatively determined by comparing the area of a capacitive region (shade part) with the overall capacity (Figure 1f) (see detailed methods in Figure S3, Supporting Information).[40,41] It was found that 77% of the observed capacity was attributed to the capacitive reaction, presenting the pseudocapacitive behavior. Moreover, the relative ratios between capacitive and diffusion-controlled reactions could also be quantified at other scan rates, as shown in Figure 1g, which displayed gradually increasing capacitive capacity with higher scan rates, reaching 86% capacitive reaction at 5 mV s−1. We found that TiO2 electrodes also exhibit pseudocapacitive responses in other electrolytes such as DME and EC/DEC, indicating the similar electrochemical behavior of TiO2 electrode regardless of the electrolyte types (Figure S4, Supporting Information). The contribution of pseudocapacitance to Na storage in TiO2 anodes before cycling was also measured (Figure S5, Supporting Information). The b values of 0.82 and 0.81 are quantified from the cathodic and anodic peaks, respectively, indicating pseudocapacitive behaviors. The b values and the corresponding capacitive contribution ratios are smaller than those for cycled TiO2 (Figure 1e,g). It suggests that the cycled TiO2 with amorphous structure present enhanced capacitive behavior, which is attributed to the high surface potential of disordered TiO2 nanoparticles for sodium ion absorption, as evidenced by a study about the correlation between the atomic orderliness of TiO2 and the electrochemical performance.[21] Therefore, The Adv. Funct. Mater. 2018, 1802099

pseudocapacitive behavior of TiO2 anode can be concluded to be an intrinsic property in SIBs. In addition, the pseudocapacitive performance has been reported to be significantly regulated by designing smart nanostructures or/and compositions, such as sulfur-doped TiO2 nanorods,[19] TiO2/carbon nanotube (CNT)/C nanorods,[25] and graphene-coupled TiO2 particles.[33] 2.2. Interfacial Chemistry of TiO2 and Tailoring the Solid Electrolyte Interphase Pseudocapacitive reaction is known to be sensitively affected by the interfacial chemistry of electrode materials and its compatibility with the electrolyte.[18,19,21,33] Previous studies also showed that when disordered TiO2 nanosphere electrodes are discharged to 0.01 V in SIBs,[19,21] the electrode presents a rapid capacity degradation due to the unstable and thickening SEI layers arising from excessive side reactions between electrolyte and TiO2 surface. In SIBs, carbonate-based electrolytes have been developed as the most commonly used electrolyte due to their suitable voltage windows, stable physicochemical properties, and its widespread usage in LIBs.[31,42,43] However, the stability of SEI layers formed from carbonates is often unreliable in SIBs and is dependent on the electrode material.[42] Thus, it requires the electrolyte additives to modify the SEI formation for the long term cycle stability.[42] On the other hand, etherbased electrolytes, which have presented attractive features in forming stable SEI layers in potassium-ion batteries (KIBs),[44] have rarely been visited for SIBs. Moreover, our recent findings on the graphite anodes in SIBs indicated that the SEI layer is notably stable in nature, allowing fast ionic transport in etherbased electrolytes,[45] which implies the potential promise of the ether-based electrolytes in SIBs. In our comparative study in this regard, ether- (i.e., DEGDME, DME) and carbonate-based (i.e., EC/DEC, EC/PC) electrolytes were prepared by dissolving 1 m NaPF6 in respective solvents, in which TiO2 electrodes were cycled at 0.1 A g−1 for 100 cycles. The morphological evolution of SEI layers on TiO2 surface during cycles were studied by HRTEM and scanning electron microscopy (SEM) characterizations. HRTEM images show that the thickness of the SEI layer on TiO2 particles was ≈2.5 nm after the second cycle (Figure 2a) and remained less than 4.0 nm after 100 cycles in DEGDME electrolyte (Figure 2b). The overall porous morphology of the pristine electrode maintained intact after cycling in DEGDME electrolyte for 100 cycles as shown in SEM images Figure 2c,d. TiO2 electrodes cycled in DME electrolyte also presented a thin SEI layer of ≈3 nm in thickness with the porous electrode morphology retained after 100 cycles (Figure S6, Supporting Information). However, a substantially thicker SEI layer of ≈10 nm was observed on the surface of TiO2 particles after the second cycle in EC/DEC electrolyte (Figure 2e), which further increased to ≈17 nm after 100 cycles (Figure 2f). The probe of the electrode morphology in Figure 2,h shows that, as a result of the growing SEI layers on the TiO2 particles, the electrode was fully covered by thick SEI layers after 100 cycles (Figure 2h), losing the pristine porous structure of the electrodes (Figure S6, Supporting Information). Similarly, thick and growing SEI layers were also observed on TiO2 anodes cycled in EC/PC electrolyte, indicating

1802099  (4 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

Figure 2.  Morphological characterizations of SEI layers: HRTEM and SEM images of TiO2 after a,c) 2 and b,d) 100 cycles in DEGDME electrolyte, HRTEM and SEM images of TiO2 after e,g) 2 and f,h) 100 cycles in EC/DEC electrolyte.

the general instability of the SEI in carbonate-based electrolytes (Figure S6, Supporting Information). The chemical compositions of each SEI layer were explored by XPS assisted with Ar ion sputtering. Figure 3 and Figure S7 in the Supporting Information illustrate the deconvoluted O1s, C1s, and F1s spectra for pristine and cycled TiO2 electrodes and their corresponding depth profiles, respectively. In the O1s spectra of the pristine electrode (bottom in Figure 3a), a peak at 529.8 eV was observed, referring to TiO bonds from TiO2.[18,46] After cycling in DEGDME electrolyte, an additional peak at 531.5 eV corresponding to CO and HO

bonds increased prominently, an indication of the formation of sodium alkoxides (RCH2ONa), Na2CO3, and NaOH according to the previous XPS studies.[47–50] This assignment was further confirmed by the evolution of the peak at 289.1 eV in the C1s spectra corresponding to RCH2ONa (Figure 3b), along with a new peak at 290.1 eV that can be assigned to Na2CO3.[48,49] Two evident peaks at 684.5 and 687.9 eV in the F1s spectra of cycled electrodes identified NaF and PF, CF bonds (Figure 3c), respectively, corresponding to NaF, NaPF6 species, and polyvinylidene fluoride (PVDF) binder.[44,47] The series of chemical information suggests that the SEI film on TiO2 electrode cycled

Figure 3.  Chemical characterization of SEI layers by XPS: a) O 1s, b) C 1s, and c) F 1s spectra for pristine (bottom) and cycled TiO2 anodes in DEGDME (middle) and EC/DEC (up) electrolytes.

Adv. Funct. Mater. 2018, 1802099

1802099  (5 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

in DEGDME electrolyte contains both organic (RCH2ONa) and inorganic (NaOH, NaF, NaPF6, Na2CO3, and phosphates) components. The deconvoluted XPS spectra of TiO2 electrodes cycled in EC/DEC electrolyte revealed that the SEI layer consisted of similar organic (RCH2OCO2Na) and inorganic (Na2CO3, NaF, NaOH, Na2O, and NaPF6) species as those formed in DEGDME electrolyte. However, the main difference of the SEI films lies in their relative proportions and the overall quantity in carbon-containing byproducts, which indicates a substantially large portion of Na2CO3 in the case of the carbonate-based electrolytes, while negligible difference was found for the fluorine-containing byproducts. The larger quantity of the carbon-containing byproducts in SEI layer is also confirmed by the etching of the SEI film by Ar ion sputtering, which is consistent with the HRTEM observations (Figure S6, Supporting Information) and suggests the critical solvent dependency in the thickness of SEI layer in SIBs.[50] These results were further supported by the energy dispersive X-ray spectroscopy mapping analyses, which presented that the SEI layer formed in EC/DEC electrolyte contains more organic/inorganic species than that in DEGDME electrolyte (Figure S8, Supporting Information). The formation and growth of SEI films could be further probed by in situ and ex situ EIS experiments in the comparative inspection of the interfacial property of the TiO2 electrodes in ether and carbonate electrolytes. Figure 4a shows

the in situ EIS curves recorded when a cell containing a TiO2 electrode was discharged in DEGDME electrolyte at a current density of 40 mA g−1. It indicates that no noticeable increase in the impedance is observed during the discharge in spite of the expected SEI layer formation. The quantitative analysis of Nyquist plots using an equivalent circuit in Figure 4b also supports the stability of the electrochemical impedance of the cell. R0, Rct1, and Rct2, which correspond to the system resistance, the interfacial resistance between electrolyte and electrode, and the charge transfer resistance, respectively, remain almost constant.[51–53] Slight reduction of Rct1 is due to the ameliorating wettability between electrolyte and electrolyte. Upon the subsequent cycles, the cell presented marginally increased resistances, Rct1/Rct2 = 9.5/1.5, 16.5/2.1, and 20.8/2.5 Ω for 2, 50, and 100 cycles, respectively (Figure 4c and Figure S9, Supporting Information), an implication of the growth and stabilization of an SEI film on the electrode surface. In contrast, the electrochemical cells cycled in EC/DEC electrolyte displayed a dramatical increase in resistances, Rct1/Rct2  = 382.3/15.2, 3057/518, and 5165/1036 Ω, after 2, 50, and 100 cycles (Figure 4d), which are hundreds of times higher than those for cells cycled in DEGDME electrolyte. This apparently different behavior suggests that the charge-transfer kinetics of TiO2 electrode remains much faster in ether-based electrolyte than in carbonate-based electrolyte. It is attributed to a stable SEI layer observed on the surface of TiO2 in ether-based electrolyte, suppressing the

Figure 4.  EIS study of the formation and growth of SEI layers: a) in situ EIS of TiO2 electrodes during the 1st sodiation process in DEGDME electrolyte, b) equivalent circuit and derived resistances at different depth of discharge (DOD) from the Nyquist plots in (a), Nyquist plots after the 2nd, 50th, and 100th cycles of TiO2 electrodes in c) DEGDME and d) EC/DEC electrolytes.

Adv. Funct. Mater. 2018, 1802099

1802099  (6 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

increase in the electrode/electrolyte interfacial resistance during cycles. Additionally, it is expected to have been benefitted from better wettability between ether electrolyte and TiO2 electrodes (the wettability comparison is provided in Figure S10, Supporting Information).[32] Similar trend was also consistently found in the comparative study of DME electrolyte and EC/ PC electrolyte during cycles, where cells with DME electrolyte presented much lower EIS impedance (Figure S11, Supporting Information). The aforementioned results suggest that the SEI layer is formed in ether-based electrolytes with more robust characteristics and is capable of providing a desirable electrolyte/electrode interface for pseudocapacitive TiO2 anodes. The formation of SEI on anodes is known to be closely related to the stability of electrolyte components such as solvent, salts, and the solvent/ ion complex.[29,54,55] In a typical electrolyte system such as the case in this study, SEI layer is mainly derived from the decomposition of solvents and solvent/ion complexes because the inner Helmholtz layer is dominated by these components in diluted electrolytes. Thus, the electrochemical reduction of an electrolyte is generally linked to the relative electronic states of solvent and solvent–ion complex against the applied potential, i.e., their relative LUMO (lowest unoccupied molecule orbital) energies. In our first principles study of each electrolyte system, the LUMO and HOMO (highest occupied molecule orbital) energy values were comparatively calculated with this respect for EC, DEC, PC, DME, and DEGDME along with their corresponding solvent–ion complexes as shown in Figure S12 in the Supporting Information. Ether-based solvents and complexes such as DEGDME and DEGDME-Na+ presented much higher LUMO levels than those for DEC, EC, PC solvents, and their solvent–ion complexes, indicating that carbonate solvents and solvent–ion complexes are more readily reduced to form SEI films during discharge. The relative instability of the carbonates compared with the ethers during the reductive process implies that more byproducts can be formed with the given applied potential, leading to a thicker SEI film. Moreover, it is also noteworthy that components of SEI are known to present higher solubility in carbonates than in ethers due to the Lewis acid character of sodium.[42,43,48] It potentially leads to the frequent formation of local pit in the SEI films, which also results in exposing the bare electrode. In this circumstance, less stable carbonate solvents and/or solvent–ion complexes would be further reduced on the surface depositing extra byproducts, which would consequently trigger the continuous growth of the film over time.

2.3. Electrochemical Properties The electrochemical performances of TiO2 anodes in ether and carbonate electrolytes are comparatively shown in Figure 5. TiO2 electrodes in DEGDME electrolyte presented an initial Coulombic efficiency (ICE) of 56% (Figure S13, Supporting Information), which is higher than 40% for electrodes in EC/DEC electrolytes as well as the reported values in literature: ≈40% for porous TiO2 nanofibers in 1 m NaClO4 EC/dimethyl carbonate (DMC) electrolyte with 1 wt% fluoroethylene carbonate additive[22] and 31.4% for TiO2/graphene composite in 1 m NaClO4 EC/PC electrolyte.[36] The higher ICE is consistent Adv. Funct. Mater. 2018, 1802099

with the thinner SEI layer in the DEGDME electrolyte, a testament to the HRTEM and XPS results. Figure 5a displays the electrochemical profiles for the subsequent ten cycles of TiO2 electrodes, which shows that 216 and 180 mAh g−1 of capacities could be delivered in ether and carbonate electrolytes, respectively. The extended cycle test of the TiO2 anodes in DEGDME electrolyte (Figure 5b) demonstrates that a reversible capacity of 192 mAh g−1 was obtained at 0.1 A g−1 after 500 cycles with an exceptionally high capacity retention of 96% versus the highest reversible capacity during cycles. On the contrary, a rapid capacity degradation was observed for TiO2 electrodes in EC/DEC electrolyte, giving rise to a low capacity retention of 43 mAh g−1 after 500 cycles. The fast capacity degradation can be attributed to the growing SEI layer with the continuous consumption of the electrolyte and the consequent increase in the cell impedance during cycles. The rate performance in Figure 5c illustrates that TiO2 electrodes in DEGDME electrolyte present a reversible capacity of 217 mAh g−1 at 0.05 A g−1 and maintain 93.1%, 87.1%, 80.6%, 72.4%, 66.3%, and 59.4% of the capacity at 0.1, 0.2, 0.5, 1, 2, and 4 A g−1, respectively. On the other hand, under the same measurement conditions, the capacity retention was only 17.7% for the TiO2 electrodes in EC/DEC electrolyte, i.e., 181 mAh g−1 at 0.05 A g−1 and 32 mAh g−1 at 4 A g−1 (Figure 5d), consistent with previous reports.[56,57] To further demonstrate the high-rate capability of TiO2 electrodes in ether electrolytes, the cells were cycled at a current density of 2 A g−1 for 600 cycles as shown in Figure 5e. The reversible capacity was 136 mAh g−1 after 600 cycles with a remarkable capacity retention of 89% and high Coulombic efficiencies of 99.98%. It is worth noting that after 600 cycles, there are fluctuations of the discharge/charge capacities, possibly due to the accumulated side reactions between electrolyte and the Na metal electrode. The stability of cyclic capacity can be recovered by refreshing the coin cells with new electrolyte and Na metal (Figure S14, Supporting Information). TiO2 anodes also present excellent electrochemical performance in the DME electrolyte with an impressive capacity retention of 92% after 500 cycles at 0.1 A g−1 and remarkable high rate capabilities (Figure S15, Supporting Information). To estimate the standing of the current TiO2 electrodes cycled in ether electrolyte, we summarized the electrochemical performance of nanostructural TiO2 electrodes in different electrolyte systems in Table S1 in the Supporting Information. It shows that the electrochemical performance of present TiO2 electrode in ether electrolyte is among the best in terms of cyclic capacities and power capability. The superior battery performance of TiO2 electrodes in ether electrolyte can be attributed to the thin and robust SEI layer formed from ether electrolyte, enabling stable and rapid sodium storage reactions with fast charge-transfer kinetics at the interface and low electrode/electrolyte interfacial impedances. To make this work more comprehensive, some aspects warrant further discussion. The ICE of TiO2 electrodes in DEGDME electrolyte is 56% in this work. Even though it is higher than those in literature,[15–17,20–22,36] it is still insufficient to meet the demand of above 90% for commercial batteries. Strategies such as presodiation, surface chemical pretreatment,[58] and/or optimizing the binder[36] should be further carried out to increasing the ICEs. The boiling points (Tb) of DME and DEGDME are

1802099  (7 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

Figure 5.  Electrochemical performance of TiO2 anodes in SIBs: a) first 10 cycles’ discharge/charge profiles of TiO2 in DEGDME and EC/DEC electrolytes at 0.05 A g−1, b) cyclic capacities and Coulombic efficiencies of TiO2 cycled at 0.1 A g−1, voltage profiles of TiO2 electrodes cycled at 0.05, 0.1, 0.2, 0.5, 1, 2, and 4 A g−1 in c) DEGDME and d) EC/DEC electrolytes, and e) high rate capability of TiO2 electrodes cycled at 2 A g−1 in DEGDME electrolyte.

reported to be 84 and 162 °C, respectively, which are lower than the 245 and 250 °C for EC/DEC and EC/PC solvents.[59,60] It is expected that the low Tb of ether solvents may lead to poor thermal stability of an electrolyte and the SIBs, causing potential safety issues. To solve this challenge, two directions of research are suggested in the future study, namely, i) to comprehensively understand the thermal stability of ether electrolytes with different salts and solvents as well as the thermal stability of batteries when the electrolytes and electrode interact with each other[31] and ii) to explore functional additives to improve the thermal stability, similar to the (ethoxy)pentafluorocyclotriphosphazene polymer developed for nonflammable electrolyte in LIBs.[61] In addition, the ether-based electrolyte presented considerable electrochemical stability up to 4.5 V

Adv. Funct. Mater. 2018, 1802099

(Figure S16, Supporting Information), which is competitive with carbonated-based electrolyte, suggesting their promising application in full cells. Compare with the widely used carbonate electrolyte in SIBs, the study of ether electrolyte is still at its infant stage. We expect that the promising results of the reliable SEI formed on TiO2 anodes from ether electrolyte in this work would stimulate further studies on ether electrolytes in SIBs. While the main effort in this work was to address the challenges of TiO2 electrodes involving both the sodiation/ desodiation mechanisms and the interfacial chemistry (as schematically shown in Figure S17, Supporting Information), it is believed that further improvement can be achieved by optimizations in other components of SIBs, incorporating our findings on the tailoring the SEI layers.

1802099  (8 of 10)

© 2018 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com

www.afm-journal.de

3. Conclusion The sodium storage mechanism in nano-TiO2 electrode was clearly unveiled through a suite of characterizations in this work. It was found that the anatase TiO2 crystals were amorphized during the first sodiation and maintained the disordered structure in the following cycles. The sodium storage was dominantly based on the pseudocapacitive reaction in the amorphous phase. Motivated by the findings on the pseudocapacitive sodium storage, we attempted to engineer the interface between the electrolyte and the electrode to maximize the performance of the electrode, and, by exploiting the ether-based electrolyte which has rarely been visited in SIBs, it was found that robust SEI layers could be formed on the TiO2 electrode. The probe of the morphological, chemically compositional, and electrochemical properties of the SEI layers suggested that ether electrolytes were more favorable in forming thin and stable SEI layers during cycles than conventional carbonatebased electrolyte. Consequently, TiO2 electrodes presented high Coulombic efficiencies, cyclic stability, and remarkable high rate capabilities, which excels the previously reported performances for TiO2 electrodes. This work demonstrates the importance of the interfacial engineering to induce proper SEI layers in the development of the high-performance and low-voltage anodes for SIBs.

4. Experimental Section Preparation of Electrolytes: Electrolytes were prepared at low moisture content (H2O < 20 ppm). Sodium salts (NaPF6) and molecular sieves were dried in vacuum oven for 8 h at 180 °C. Then, the dried salts were dissolved in DME, DEGDME, EC: DEC (1/1 v/v), and EC: PC (1/1 v/v) solvents at 1 m, respectively. The solutions were stirred at 50 °C for 8 h in glove box before adding molecular sieves to remove the residual H2O in electrolytes. Electrodes Preparation and Electrochemical Characterization: The TiO2 anodes were prepared by mixing the commercial anatase TiO2 nanoparticles (Sigma-Aldrich,