Vertically oriented TiS2−x nanobelt arrays as binder- and carbon-free intercalation electrodes for Li- and Na-based energy storage devices

Casey G. Hawkins and Luisa Whittaker-Brooks *
Department of Chemistry, University of Utah, 315 South 1400 East, Rm 2020, Salt Lake City, Utah 84112, USA. E-mail: luisa.whittaker@utah.edu

Received 13th June 2018 , Accepted 26th July 2018

First published on 26th July 2018


Titanium(IV) sulfide (TiS2) is a promising 2D layered material for energy storage applications. However, its electrochemical performance has been hindered by its low ion diffusion coefficient and inconsistencies in determining its electrical properties. To overcome these challenges, bulk TiS2 cathodes are normally mixed with conductive additives (typically carbon) and polymer binders (typically polyvinylidene fluoride – PVDF) to yield a paste that is subsequently cast onto a current collector. The electrochemical performance of the electrode is lowered due to the extra weight of all the inactive components (i.e., additives, polymer binder, and metal substrates) introduced during the fabrication process. An alternative to the use of pasted electrodes is the direct growth of well-defined nanostructures on a conducting substrate. In this study, we report the synthesis, characterization, and electrochemical performance of carbon- and binder-free cathodes comprising highly conducting TiS2 nanobelts. The fabricated TiS2 nanobelts are highly anisotropic and are vertically grown directly on the current collector to yield a spatially controlled array. The short ion diffusion paths, high electrical conductivity and absence of additives that hinder ion migration lead to Li- and Na-based TiS2 electrochemical devices exhibiting high specific capacity, less capacity fade, and resilience under higher cycling rates. We also present the effects of sulfur vacancies on the electrochemical performance of both vertically oriented Li- and Na-based TiS2−x nanobelt array cathode insertion hosts. It is also worth mentioning that Na-ion half-cells comprising our vertically oriented TiS2 nanobelt arrays exhibit a discharge capacity of 217 mA h g−1 (theoretical specific capacity of 239 mA h g−1). This discharge capacity equates to 0.91 Na per TiS2 unit – exceeding the maximum loading ever obtained of 0.8 Na per TiS2 under practical operation. We thus observe that scaling to finite sizes as well as fabricating highly anisotropic nanobelt arrays are found to profoundly alter both the kinetics of Li- and Na-ion insertion and the electrochemical performance of TiS2.


Introduction

With the dramatic increase of portable electronics, the demand for pure or hybrid electric vehicles, and the need for grid-scale energy storage for renewable energy technologies, there is an urgent necessity to develop rechargeable energy storage technologies with high energy density, high power density, and long-life cycles. Rechargeable energy storage technologies such as lithium (Li)-ion batteries have become the most widely implemented platform since commercialized in the early 1990s.1,2 While Li-ion batteries have dominated the rechargeable energy storage market, there are several hurdles currently hindering their widespread use in electric vehicles and in grid-scale electrical storage devices.3 One primary hurdle pertains to safety issues presented by using Li and flammable organic electrolytes. Even if these safety issues are resolved, due to the limited number of Li deposits globally, there are concerns that Li is not a sustainable resource.3,4 Forecasting the energy storage landscape for Li-ion technologies, their high demand, and the limited supply of Li sources may increase their cost and jeopardize their further development.

Several ions such as monovalent sodium (Na+) and potassium (K+),5–10 divalent magnesium (Mg2+) and calcium (Ca2+),11–14 and trivalent aluminum (Al3+),15–18 have been investigated as possible alternatives to Li+. Na+ is an attractive candidate due to its relative earth abundance as well as the widespread availability of Na sources around the globe. The stark similarity between the Na+ and Li+ ions in terms of their physical and chemical properties allows for a “lithium-like approach” where most of the research advancements obtained with Li-ion technologies may be fully adaptable to Na-ion technologies. Shifting to Na-ion energy storage technologies may still require the development of efficient cathode and anode materials that can overcome the sluggish kinetics of the Na+ intercalation/de-intercalation reaction caused by the relatively larger ionic radius of Na+ (102 pm) versus Li+ (76 pm).19–21 Various layered metal oxides,22–26 phosphates/fluorophosphates,27–29 metal fluorides,30–32 and metal chalcogenides33–37 have been developed as reversible intercalation/de-intercalation cathodes for Na-ion energy storage devices. However, most of these potential electrode materials suffer from limited specific capacities and rate capabilities. Among these materials, metal chalcogenides have drawn significant interests because of their promise of overcoming these limitations due to their high electrical conductivity, large interlamellar spacing, and short ion/charge diffusion paths. As an archetypal layered metal chalcogenide, titanium(IV) sulfide (TiS2) possesses several properties that are advantageous for energy storage applications such as earth abundance, high electrical conductivity (104 S m−1), and higher theoretical specific capacity (239 mA h g−1) than that of the structurally limited specific capacity for commercially ubiquitous LiCoO2 (140 mA h g−1, when cycled between 3.0–4.5 V).38,39

TiS2 crystallizes in a simple hexagonal structure comprising layers of [TiS6] octahedra sharing edges and corners. The layers are weakly bound by van der Waals' electrostatic forces along the c-axis and the spacing between the layers (5.69 Å) provides abundant sites for the intercalation/de-intercalation of organic and inorganic guest species.40–45 In the mid-1970s, Whittingham first demonstrated the possibility of using metallic lithium, coupled with a TiS2 single-crystal as a viable secondary energy storage material.46 While using a single-crystal provides key fundamental insight into the properties of TiS2, this is not a practical synthesis that would allow commercialization on any appreciable scale. Correspondingly, in the late 1970s and early 1980s, Whittingham, Newman, and Klemann also investigated the electrochemical properties of TiS2 as a possible intercalation host for Na-ions.46,47 Both reports indicate that TiS2 would not be suitable as a cathode due the loading of Na+ to be limited to x = 0.8 for NaxTiS2. Additionally, a potential phase change in the TiS2 crystal structure was observed upon insertion and de-insertion of Na+ ions that results in unrecoverable capacity loss.47 Nonetheless, a recent study has shown that the size of the TiS2 crystallites has an impact on the stability and Na-ion loading. Through a colloid synthesis route, Liu et al. fabricated TiS2 nanoplatelets with diameters of 100–200 nm and thicknesses around 5 nm. Batteries employing these nanoplatelets exhibit high Na-ion loading (186 mA h g−1) and good cycling stability (76% capacity retention after 300 cycles at a 2C rate).48

The solid-state synthesis of TiS2 from elemental Ti and S typically yields micron-sized hexagonal platelets. As these platelets are free-flowing, a polymeric binder, typically polyvinylidene fluoride (PVDF), is used to create a slurry that is tape-cast onto a current collector. Since PVDF is an insulator, conductive carbon is usually added to the slurry to overcome its low electrical conductivity. While the addition of conductivity enhancers (additives) is the current procedure used to overcome the low conductivity of oxide-based cathode materials such as LiCoO2, LixMn2O4 and LiFePO4, there are some inherent issues with this approach.39,49,50 First, the addition of an electrochemically inert binder and a conductivity enhancer lowers the overall gravimetric capacity of the cathode. As some formulations call for as high as 10% by/wt loading of the binder and carbon, a significant reduction in gravimetric capacity occurs. In addition, the loss in capacity makes it difficult to achieve high volumetric energy solutions. Finally, studies have shown that the carbon surrounding the active material can slow down the diffusion of the ions in and out of the active material.51 In order to increase the stability and maintain the highest specific and gravimetric capacities, a new synthesis method for the fabrication of TiS2 cathode insertion hosts that does not require the use of any binder or conductive additives is needed. Furthermore, scaling TiS2 to nanoscale dimensions would offer the potential for increased power and energy densities due to an improved interfacial contact between TiS2 and the electrolyte and shorter solid-state diffusion paths. Moreover, the fabrication of well-defined one-dimensional nanostructured TiS2 arrays allows for better control of volume changes upon intercalation/de-intercalation without fracturing, a phenomenon that is often observed in bulk or micron-sized materials.

In a previous publication, we reported the synthesis, electrical characterization, electrochemical properties, and the control of sulfur vacancies of TiS2 cathode insertion hosts. As demonstrated in our earlier work, the systematic solid-state transformation of TiS3 to TiS2 nanobelts allows for the control of the nonstoichiometry in TiS2−x nanostructures.52 Such control demonstrates that cathodes based on sulfur-deficient TiS2−x nanobelts deliver efficient Li+ intercalation/de-intercalation activity, excellent cycling life, enhanced specific capacity, and excellent rate capability pointing to the importance of carefully controlling defects and stoichiometries in materials as a way to favorably tune their electronic properties. In this work, TiS3 nanobelts are grown in a vertically oriented array directly on a metallic current collector substrate. Via a pyrolysis fabrication technique, the TiS3 nanobelt array is further converted into a TiS2−x nanobelt array that can be used as a cathode without the addition of binders or carbon additives. The vertically oriented nanobelt array geometry allows for each TiS2−x nanobelt that is electrically connected to the metallic current collector substrate to contribute to the capacity. Moreover, each TiS2−x nanobelt has the ability to access various anisotropic 1D electronic pathways allowing for efficient charge transport. We also report the effects that varied pyrolysis times have on the crystalline structure, induced S vacancies, and specific capacity of the vertically oriented TiS2−x cathode insertion hosts. Similarly, we report the electrochemical properties of vertically oriented TiS2−x nanobelt arrays, the effect of different charge and discharge rates on their capacity, and their long-term stability when incorporated into both Li-ion and Na-ion battery systems. Finally, we discuss the changes in diffusion rates for Li- and Na-ions upon their insertion into vertically oriented TiS2−x nanobelt arrays via studies involving galvanostatic intermittent titration techniques (GITT).

Experimental methods

Synthesis

Titanium (Ti) sputtering target (Kurt J. Lesker 99%, 1.0′′ Dia. × 0.125′′ thick) and sulfur (S) powder (Alfa Aesar 99.5%, ∼100 mesh) were used as received without any purification or further treatment. All work was performed using Schlenk-line techniques or performed in an argon-filled glovebox to limit oxygen and water exposure. To synthesize TiS3 nanobelt arrays, a 1.6 μm-thick Ti film was first sputtered onto a carbon-coated Al (C–Al) foil (MTI) using a Denton Discover 18 sputtering system. This process yielded a 5/8′′ x 3′′ piece of Ti-coated C–Al foil that was subsequently placed into a borosilicate ampule with ≈85 mg of S powder. The ampule was evacuated to ≈10−3 torr using a vacuum line prior to flame-sealing. The sealed ampule was heated in a furnace to 450 °C at 10 °C min−1 and then isothermed at 450 °C for 20 h. The ampule was then removed from the oven and placed on an Al block to induce rapid quenching of the reaction. For the conversion of TiS3 nanobelts to TiS2−x nanobelts, the as-prepared TiS3-coated substrate was transferred to a new borosilicate ampule. A glass wool plug was added into the ampule as a constriction to prevent the precursors from coming into physical contact during handling. Subsequently, ≈150 mg of Ti powder (Alfa Aesar 99%, ∼325 mesh) was added while being kept separated from the TiS3 precursor. This was performed to allow Ti to react with volatilized sulfur that has been pyrolyzed from the TiS3 sacrificial template. The ampule containing the TiS3 sacrificial template along with the Ti powder precursor was evacuated and flame-sealed under vacuum. The reaction ampule was placed in a furnace and heated to 450 °C at 10 °C min−1 and then isothermed at 450 °C for 48–144 h. The reaction ampule containing the TiS2−x nanobelts was allowed to cool to ambient temperature inside of the furnace.

Characterization

The morphologies of the resulting products were examined using a scanning electron microscope (SEM) equipped with a field-emission gun (FEI Nova Nano, FE-SEM 630) operated at an accelerating voltage of 10 keV. Phase identification and purity of the as-synthesized TiS3 and TiS2−x nanobelts were obtained by X-ray diffraction (XRD) using a Bruker D8 Advance diffractometer, at a scanning rate of 1.2° min−1 in a 2θ range between 5° and 70° using monochromated Cu Kα radiation (λ = 1.5406 Å). The operating voltage and current were kept at 40 kV and 40 mA, respectively. The dimensions and crystallinity of the as-synthesized nanobelts were examined using a JEOL JEM 2800 field-emission gun transmission electron microscope (FE-TEM) operated at 200 keV. High-resolution TEM (HRTEM) images and selected area electron diffraction (SAED) patterns were independently acquired and provided insights into the crystal growth habits and crystallinity of the TiS3 and TiS2−x nanobelts, respectively. To prepare the samples for TEM/SAED analysis, the nanobelts were dispersed in 2-propanol via sonication and then deposited onto a 400-mesh holey carbon-coated copper grid. For spectroscopy measurements, all films fabricated were handled in an inert Ar atmosphere and never exposed to oxygen and water levels above 1 ppm before introduction into the ultra-high vacuum (UHV) system for X-ray photoelectron (XPS) and ultraviolet photoelectron (UPS) spectroscopy measurements. XPS and UPS spectra were acquired in a custom-built UHV chamber equipped with a cylindrical mirror electron analyzer and operated at a base pressure of 10−10 torr. UPS was performed by using a He I (21.22 eV) excitation line of a He plasma in a discharge lamp. Spectra were taken at a pass energy of 5 eV for a nominal experimental resolution smaller than 150 meV. XPS was performed on an additional sample series, using the Al Kα line (1487.6 eV) with a resolution of 0.8 eV. A pass energy of 100 eV was used for survey scans, while a 40 eV pass energy was used for detailed scans. All measurements were carried out at normal take-off angles. The acquired spectra were calibrated against an adventitious carbon peak at 284.6 eV. Curve fitting was carried out using CasaXPS software with a Gaussian–Lorentzian product function and a non-linear Shirley background.

Electrochemical studies

Cathodes were fabricated by punching 1/2′′ diameter discs from the TiS2−x-coated C–Al foil substrate. Each disc had approximately 1.5–1.7 mg of TiS2−x nanobelts. CR2032 coin cells were assembled in an Ar-filled glovebox using the cathode prepared above and Li(Na) foil as the anode. Celgard (2400) was used as the separator. A 1.0 M LiPF6 (for Li-ion battery) and 1.0 M NaPF6 (for Na-ion battery) in ethylene carbonate/diethyl carbonate (50/50 v/v) solutions were used as the electrolytes. The batteries were constructed in the following manner: bottom cup, cathode, electrolyte (25 μl for Li-ion and 30 μl for Na-ion), separator, electrolyte (25 μl for Li-ion and 30 μl for Na-ion), anode, Ni foam, and the top cap.

Galvanostatic measurements were conducted at room temperature using a Neware BT-4008 battery testing system. Li-ion cells were cycled between 1.0 V and 3.0 V versus Li+/Li. Na-ion cells were cycled between 0.5 V and 2.5 V versus Na+/Na. The capacity retention study was cycled at different rates ranging from 0.2–5C. The long-term capacity fade study was cycled at a theoretical rate of 0.1C. The specific capacity and coulombic efficiency calculations are determined within a ± 1% uncertainty error. The GITT study was discharged at a theoretical rate of 0.2C. All batteries tested for the GITT study were discharged for 1 minute and then allowed to rest for 2 hours before the next discharge.

Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) studies were performed on a CH Instrument (CHI660E) electrochemical workstation. Each test was performed using a CR2032 coin cell. Frequency ranges from 100 MHz to 0.1 Hz and an amplitude of 10 mV were used. EIS spectra were fit to the equivalent electrical circuit presented in Fig. 7 using a CHI Version 14.09 software.

Results and discussion

Both TiS3 and TiS2 exhibit a lamellar structure that makes them ideal intercalation host materials. TiS3 has a monoclinic crystal structure that consists of infinite 2D double layers of TiS8 polyhedra. Each polyhedron is comprised of two S22− ions in a rectangular face and four noncoplanar S2− ions. The double layers are formed by sharing the S2−–S2− edges between the 1D single layers. Conversely, the space between the (S22−)2 ion rectangular faces is empty, thus providing the necessary room for the intercalation and deintercalation of Li+ and Na+ ions.53 TiS3 has a higher theoretical specific capacity (556 mA h g−1) than that of TiS2 (239 mA h g−1). However, when TiS3 is incorporated into liquid-type Li-ion cells, its experimental specific capacity quickly fades from ≈350 mA h g−1 to ≈75 mA h g−1 after the 5th cycle.54 This abrupt capacity fading is caused by the structural degradation of TiS3 upon the irreversible reduction of S22− to S2− in the presence of organic liquid electrolytes.54 TiS2 is comprised only of S2− ions in its crystal structure and does not suffer from any structural degradation upon cycling. For these particular reasons, TiS2 has been widely studied over TiS3 as a cathode insertion host despite having a lower theoretical specific capacity. Furthermore, given the remarkable enhancements of charge capacities and life cycles observed for anodes and cathodes comprising materials that are susceptible to finite size effects, it is now widely accepted that scaling materials – particularly 2D metal chalcogenides – to nanoscale dimensions may provide an incredible bounty of advantages such as improved accommodation of strain defects upon electrochemical lithiation/sodiation, increased electrode/electrolyte interaction, and shorter diffusion path lengths.55,56 TiS3 tends to crystallize as anisotropic structures due to its crystal structure having two easy cleavage planes both of which ease the reduction of bulk TiS3 into belt-like nanostructures. TiS2, in contrast, tend to crystallize as isotropic structures that often yield very irregular and undefined plate-like structures. Since these plate-like structures tend to be very inhomogeneous and poorly size controllable, it is extremely challenging to accurately determine how quantum confinement, morphology, and orientation affect the electrochemical properties of TiS2. Here, we developed a fabrication method for the synthesis of well-defined and vertically oriented TiS3 nanobelt arrays. These TiS3 nanobelt arrays form the sacrificial scaffolding that is then pyrolyzed to yield TiS2−x while preserving the nanobelt morphology and vertically aligned array orientation. The pyrolysis conversion (desulfurization) of TiS3 nanobelts into TiS2−x nanobelts proceeds according to the following chemical reaction:
 
image file: c8ta05645e-t1.tif(1)
 
image file: c8ta05645e-t2.tif(2)

During the desulfurization of TiS3, nucleation seeds comprising TiS2 facets are formed on the surface of the TiS3 nanobelts and are propagated inward at a rate that is strongly dependent on the reaction time and temperature. As depicted in Fig. 1, TiS3 is initially formed by the interaction of sulfur (vapor-phase) with elemental titanium (solid-phase) atop a C–Al foil substrate. This interaction yields a vertically oriented TiS3 nanobelt array. Crucial to the fabrication of vertically oriented TiS2−x nanobelt arrays is the control of two competing reaction dynamics, i.e., the high-temperature removal of S from TiS3 and the possible disruption of the nanostructure morphologies due to these high temperatures. When heated, sulfur leaches out from the TiS3 sacrificial template. Here, a titanium source is introduced to avoid the re-absorption of sulfur by the TiS3 sacrificial template. A reaction temperature of 450 °C enables the conversion of TiS3 to TiS2−x while preserving both the 1D morphology and vertically aligned orientation.52


image file: c8ta05645e-f1.tif
Fig. 1 Schematic depiction for the synthesis of vertically oriented TiS3 nanobelt arrays and their subsequent conversion into vertically oriented TiS2−x nanobelt arrays.

The SEM micrographs presented in Fig. 2 illustrate the morphological features of the bare substrates and those of the vertically aligned nanostructures. Fig. 2A shows the SEM image for a bare C–Al foil substrate. This C–Al foil substrate is highly porous with multiple voids being interconnected by a C–C network. Previous reports have demonstrated that cathodes made using C–Al foils as substrates exhibit higher initial capacity as well as less degradation over the cycling lifetime. This performance boost is attributed to greater adhesion strength of the current collector foils when carbon is present as compared to pristine current collector foils.57,58 As shown in Fig. 2B, upon sputtering Ti onto the C–Al foil, the voids are filled in with a thin layer of the metal, resulting in a pinhole free and non-porous surface. Fig. 2C depicts the SEM image for an as-synthesized TiS3 vertically oriented nanobelt array. These nanobelt arrays are ≈75 ± 10 μm thick (based on statistical length distributions obtained from cross-sectional SEM images, Fig. S1). The growth of the nanobelts is thought to proceed via a vapor-solid process involving the reaction between sulfur (in the vapor phase) and titanium (in the solid phase). Individual nanobelts obtained by the vapor transport reaction are ≈65 ± 10 nm in width (based on statistical distributions obtained for 500 nanobelts) and ≈85 ± 3 nm thick (from tapping mode atomic force microscopy measurements). Both dimensions are smaller than the average length of the nanobelt (≈75 ± 10 μm). Fig. 2D shows the SEM image for a dense vertically aligned TiS2−x nanobelt arrays showing the smooth walls and uniform morphology and sizes along the length of the nanobelts. The resulting nanobelts obtained from the desulfurization of the TiS3 sacrificial template are ≈60 ± 10 nm in width and ≈90 ± 5 nm thick suggesting that the pyrolysis process has little to no impact on the morphology and sizes of the final product, i.e., the TiS2−x nanobelt arrays.


image file: c8ta05645e-f2.tif
Fig. 2 SEM images for a pristine C–Al foil substrate (A), Ti-coated C–Al foil substrate (B), as-synthesized TiS3 (C) and TiS2−x vertically oriented nanobelt arrays (D).

Definitive phase identification of the structures is obtained from XRD measurements. Fig. 3 depicts the XRD patterns of the vertically oriented TiS3 nanobelt array substrate upon annealing at 450 °C and being exposed to different desulfurization reaction times (48–144 h). As a control, the XRD spectrum for a bare C–Al foil is also presented. Besides from the peaks ascribed to the C–Al substrate (denoted with a “C”), all experimental reflection peaks can be assigned either to TiS3 (Joint Committee of Powder Diffraction Standards (JCPDS) no. 15-0783) or TiS2 (JCPDS no. 15-0853). As illustrated in Fig. 3, the initial contact between sulfur and titanium only yields TiS3 as the main product. When pyrolyzed for 48 h, TiS3 is fully converted to a phase that is closer to that of stoichiometric TiS2. As previously determined, fabricating stoichiometric TiS2 may be challenging due to its tendency to develop sulfur vacancies and/or titanium adatoms.52 Since it is challenging to control the stoichiometry of TiS2, it is also troublesome to accurately determine its electrical and physicochemical properties as these are mostly interrelated to the amount of vacancies and defects introduced during the fabrication process. For this purpose, we will refer to the as-pyrolyzed products as TiS2−x, where the prolonged desulfurization exposure produces a Ti/S stoichiometry ratio that is farther away from the nominal TiS2 stoichiometry.


image file: c8ta05645e-f3.tif
Fig. 3 XRD spectra for a C–Al substrate, TiS3, and TiS2−x as a function of pyrolysis (desulfurization) reaction time. Peaks from the C–Al foil substrate are denoted with the letter “C”. The JCPDS for TiS3 and TiS2 crystal structures are presented as solid vertical lines at the bottom of the XRD patterns.

Further structural and morphological studies on individual TiS3 and TiS2−x nanobelts have been undertaken using lattice-resolved HRTEM and SAED (Fig. 4). The patterns obtained from both HRTEM and SAED allow for the elucidation of the single-crystalline nature of the TiS3 sacrificial template and its pyrolysis to yield TiS2−x nanobelt arrays. The lattice-resolved HRTEM image shown in Fig. 4A shows three distinct lattice fringes with interplanar spacings of 2.01, 2.28, and 4.93 Å. The observed spacing between the lattice fringes can be assigned to the (210), (112) and (100) crystallographic planes of the TiS3 crystal structure, respectively. Fig. 4B shows the SAED pattern for an individual TiS3 nanobelt. This pattern is consistent along the length of the nanobelt corroborating its single-crystalline nature. The diffraction spots observed in the SAED pattern presented in Fig. 4B are associated with the crystallographic reflection planes from the TiS3 monoclinic crystal structure. The full conversion of TiS3 to TiS2−x nanobelt arrays also results in the formation of a single-crystalline product as corroborated in the lattice-resolved HRTEM and SAED patterns presented in Fig. 4C and D. The lattice-fringes observed in the HRTEM micrograph presented in Fig. 4C are spaced at 1.70 and 2.95 Å, which corresponds to the lattice spacing between the (110) and (100) planes of the TiS2 crystal structure, respectively. The SAED pattern presented in Fig. 4D exhibits a six-fold symmetry (∼60° oriented with respect to the (010) and (100) planes) that is characteristic of TiS2 having a hexagonal crystal structure. Based on both HRTEM and SAED analyses, TiS3 and TiS2−x nanobelts have a preferred growth direction along their c-axis. This preferred growth direction can readily facilitate the cleavage of TiS3 and TiS2−x layered structures into few atoms-thick layers thus potentially unraveling unconventional properties due to quantum confinement effects.


image file: c8ta05645e-f4.tif
Fig. 4 HRTEM images for TiS3 and TiS2−x single nanobelts. (A) Lattice-resolved HRTEM image for an individual TiS3 nanobelt. (B) Indexed SAED pattern for TiS3. (C) Lattice-resolved HRTEM image for an individual TiS2−x nanobelt. (D) Indexed SAED pattern for TiS2−x. The sample used for acquiring the HRTEM data in (C and D) was pyrolyzed at 450 °C for 48 h.

To further confirm the composition, purity, and coordination environment of the samples fully converted from TiS3 to the TiS2−x nanobelt arrays, high-resolution XPS analyses for sulfur and titanium binding energy regions were performed. Fig. 5A exhibits the XPS spectrum acquired for the Ti 2p region of the TiS3 sacrificial template. Here, the XPS spectrum features two distinct peaks centered at ≈456.3 and 462.4 eV. These two peaks are ascribed to Ti 2p3/2 and Ti 2p1/2, respectively, with a spin–orbit doublet splitting of 6.1 eV. Such spin–orbit doublet splitting is characteristic of an oxidation state of 4+ for the Ti atoms present in TiS3.54,59Fig. 5B depicts a slightly more convoluted XPS spectrum of the S 2p region for the as-prepared TiS3 nanobelts. The S 2p XPS spectrum is composed of three peaks at ≈161.2, 162.4 and 163.6 eV. TiS3 is notorious for having two different sulfur species, i.e., S2− and S22− which can be clearly identified by XPS. By analyzing the XPS spectrum presented in Fig. 5B, we observe that the peak centered at ≈161.2 eV can be assigned to the S2− 2p3/2 binding mode. The peak centered at ≈162.4 eV is a combination of the S2− 2p1/2 and the S22− 2p3/2 binding modes. Finally, the peak centered at ≈163.6 eV can be assigned to the S22− 2p1/2 binding mode. The fit ratio of the S22−[thin space (1/6-em)]:[thin space (1/6-em)]S2− peak areas is 2[thin space (1/6-em)]:[thin space (1/6-em)]1, indicating that TiS3 is comprised of one S22− and one S2− units.54,59Fig. 5C shows the Ti 2p XPS spectrum obtained for a pyrolyzed TiS3 substrate upon full conversion to TiS2−x (reaction temperature: 450 °C; pyrolysis time exposure: 48 h). Here, we observe the evolution of the Ti 2p3/2 and Ti 2p1/2 doublet peaks centered at ≈456.3 and 462.4 eV, respectively. These peaks, are again, characteristics of Ti4+. Fig. 5D shows the XPS spectrum of the S 2p region for the fully converted TiS2−x sample. The S 2p peaks are a slightly merged doublet that is fit to S 2p3/2 and S 2p1/2 centered at 161.2 and 162.4, respectively, with a spin–orbit doublet splitting of 1.2 eV. As depicted in Fig. 5D, the lack of any peaks associated with S22− further corroborates the full conversion of TiS3 to TiS2−x. It is worth mentioning, that the XPS spectra presented in Fig. 5C and D are representative of all samples obtained by the pyrolysis of TiS3 for 48–144 h. We do not observe the evolution of any peaks that could potentially be associated with the decomposition of TiS2−x into sub-stoichiometric Ti/S structures upon prolonged desulfurization.


image file: c8ta05645e-f5.tif
Fig. 5 XPS spectra for vertically oriented TiS3 and TiS2−x nanobelt arrays. (A and B) Ti 2p and S 2p regions, respectively, for TiS3. (C and D) Ti 2p and S 2p regions, respectively, for TiS2−x. The sample used for acquiring the XPS data in C and D was pyrolyzed at 450 °C for 48 h.

As discussed above, a problem that has deterred TiS2 from being considered a viable material for energy storage – despite its superb electrochemical properties – is the inability to synthesize well-controlled stoichiometric TiS2 structures. In the early 1960s, work by Benard and Jeannin argued that Ti atoms would occupy interstitial sites and therefore the TiS2 structure would be metal-rich.60 They claimed that the metallic nature of TiS2 arises from degenerate states created by excess Ti atoms in the crystal structure. Further investigations done by Thompson et al., proved that stoichiometric TiS2 does exist and that the metallic behavior must arise from some other source.61 Another plausible explanation is that instead of being metal rich due to Ti atoms occupying interstitial sites, it may be that TiS2 is sulfur deficient thus, creating S vacancies. If a sulfur vacancy occurs, the electron participating in the bonding is transferred to the Ti4+ atom thus reducing it to Ti3+. One way to probe the electronic and defect states of TiS2−x is via ultraviolet photoelectron spectroscopy (UPS) studies.

Using ultraviolet radiation, the occupied stated at the top of the valence band and below the Fermi level are probed. Fig. 6 shows the direct photoemission spectra – around the valence band onset – of TiS2−x with different sulfur vacancy concentrations. The Fermi level is referenced to a binding energy of 0 eV. Besides the normal secondary electron energy cutoff observed at a binding energy of ≈1 eV below the Fermi level, there is the evolution of a peak at a binding energy of ≈0.3 eV around the Fermi edge. This feature is attributed to Ti3+ 3d band defect states that arise when S vacancies are present. To determine how the Ti3+ 3d band defect states changes during the desulfurization process, the signals were normalized using the secondary energy cutoff as the base. The inset provided in Fig. 6 shows the growth of the Ti3+ 3d band defect state feature as the pyrolysis time is increased. This suggest that upon prolonged desulfurization time exposure of TiS3 in order to convert it to TiS2−x, the likelihood for the formation of sulfur-deficient nonstoichiometric TiS2 structures is increased. This tendency of developing anion vacancies in titanium compounds has been previously demonstrated for oxygen-deficient TiOx structures via similar UPS studies.62,63


image file: c8ta05645e-f6.tif
Fig. 6 UPS spectra near the Fermi level for vertically oriented TiS3 and TiS2−x nanobelt arrays. Shoulder features near the Fermi level indicate the presence of Ti3+ band defect states. Inset shows background subtracted peaks illustrating the ingrowth of the Ti3+ 3d band defect peak upon prolonged desulfurization time exposure.

While the electrochemical performance of an energy storage device may be improved through the incorporation of novel materials with unconventional properties, designing conformal energy storage devices comprising 1D nanoarrays in a 3D architecture may fundamentally provide superior advantages (vide infra). As a model architecture in high power Li(Na)-ion batteries, self-supported 1D nanoarrays that can serve as vertically oriented electrodes enable high power densities as a result of their increased surface area. Theoretically, Li+(Na+) diffusion time is directly proportional to the square of the diffusion length. Thus, reducing the Li+(Na+) diffusion pathway decreases the diffusion time scales. Specifically, fabricating an electrode in a vertically oriented architecture where the ion diffusion path is perpendicular to the current collector holds tantalizing possibilities for enhanced ion delivery while maintaining fast electron transport. Here, we fabricated binder- and carbon-free vertically oriented TiS2−x nanoarray electrodes directly grown on the current collector. Fig. 7 shows the electrochemical performance for our vertically oriented TiS2−x nanobelt arrays as cathode insertion hosts in a half-cell configuration (Li/Na–TiS2−x). Upon cycling both Li- and Na-based TiS2−x half-cells, no significant changes in the cyclic voltammetry (CV) characteristics are observed (Fig. S2). This suggests that TiS2−x nanobelt arrays do not undergo any phase transformations upon cycling or during the formation of the SEI layer. This result is in accordance with a previous report demonstrating the effective intercalation of Li+ ions within the layers of TiS2 nanobelts without any significant structural disruption.64


image file: c8ta05645e-f7.tif
Fig. 7 Cyclic voltammetry characteristics of the second and third cycles for a vertically oriented TiS2−x nanobelt array cathode in an assembled CR3032 cell with Li (A) and Na (B) as the anodes. Electrochemical impedance spectra for the Li- and Na–TiS2 half cells shown in (C) and (D), respectively. The simulated Randle's equivalent circuit used for fitting the experimental data is also included in the figure. In this circuit, Rs is equivalent series resistance, RCT is the charge-transfer resistance, RSEI is the resistance of the electrode–electrolyte interphase, Zw is the Warburg impedance element, CSEI and CPE are the SEI and double layer capacitances, respectively. Each cathode was pyrolyzed at 450 °C for 48 h.

The CV curves for the Li-based half-cell within the potential region of 1–3 V (vs. Li/Li+) at a scan rate of 5 mV s−1 is shown in Fig. 7A. The discharge profile for the Li-based TiS2−x half-cell exhibits two peaks centered at ≈1.8 and 1.6 V. Within this voltage window, the Ti4+/Ti3+ is the only observable redox couple. The two peaks upon lithium intercalation are attributed to lithium rearrangement within the TiS2−x host matrix as the lithium loading approaches one Li per TiS2 unit. This behavior is often observed in high-voltage oxide-based intercalation cathodes. The de-intercalation of lithium within the TiS2−x host matrix occurs at 2.3 V. Taking into consideration the redox waves for the lithium intercalation and de-intercalation process taking place at 1.7 (median value) and 2.3 V, respectively, a charge–discharge overpotential of ≈0.6 V is observed. Conversely, as shown in Fig. 7B, the CV curves for the Na-based TiS2−x half-cell within the potential region of 0.5–2.5 V (vs. Na/Na+) at a scan rate of 5 mV s−1 display a single peak at 1.1 V for the intercalation of Na into the TiS2−x matrix followed by a de-intercalation peak at 1.9 V. A charge/discharge over potential of ≈0.8 V is observed for the Na-based TiS2−x system. Due to Na having a larger ionic radius than that of Li, Na intercalation/de-intercalation would require a higher driving force to occur. Previous demonstrations using LiNi1/3Co1/3Mn1/3O2 as an active cathode material have shown that an increased overpotential results in a decrease in specific capacity.65 As such, when fabricating batteries and understanding the electrochemical performance of our vertically oriented TiS2 nanobelt arrays as cathode insertion hosts, it would not be surprising that Na-based TiS2 devices will show lower discharge specific capacities than that of Li-based TiS2 devices (vide infra).

Electrochemical impedance spectroscopy (EIS) measurements yield insights into the charge transport, resistances, and interfacial processes in our Li(Na)-based TiS2−x electrochemical systems. Fig. 7C and D show the EIS data in the form of Nyquist plots for Li–TiS2−x and Na–TiS2−x half-cells. The EIS measurements were acquired at the open circuit potential for each cell after the 50th charge–discharge cycle. For both type of cells, we observe the formation of a quasi-semicircle at high frequencies. The low frequency region, however, is comprised of a quasi-linear tail. The front edge of the semi-circle is a function of the resistance and capacitance of the formed solid-electrolyte interface. The later part of the semi-circle is derived from the charge-transfer resistance and it is associated to the double-layer capacitance. The quasi-linear tail observed at low frequencies is a product of the Warburg impedance element associated with ion diffusion. When fit to the equivalent circuit also presented in Fig. 7, the EIS data allow for the determination of the resistances within the Li(Na)-based TiS2−x half-cells. Based on the equivalent circuit, Rs represents the ohmic series resistance of the electrolyte (resistances from the current collectors, separators, and electrodes may play a minor role in the magnitude of the ohmic resistance), RSEI denotes the resistance of the solid-electrolyte interphase (SEI) formed on the electrode surfaces, CSEI is the capacitance of the SEI layer, RCT is the resistance developed upon charge transfer, CCT is the double layer capacitance, and W corresponds to the Warburg impedance due to the diffusion of Li+(Na+) ions into the electrode. By fitting the EIS data, we obtain a low Rs of ≈4.2 Ω, which is indicative of the good electrochemical cycling stability of our vertically oriented TiS2−x nanobelt arrays as cathode insertion hosts for Li-ion batteries. Also, a low RCT of ≈6.9 Ω is obtained which suggests a fast rate of redox reactions at the electrode–electrolyte interface and it is also indicative of fast Li-ion transport within the vertically oriented TiS2−x scaffold. In addition, a low RCT could also be associated with an increased contact area at the electrode–electrolyte interface upon introducing the vertically oriented nanobelt arrays. The RSEI which is the leakage resistance at the SEI layer is usually very high and can be ignored in the circuit. In contrast, as shown in Fig. 7D, a larger semi-circle radius is obtained for the Na-based TiS2−x half cells. The fit data yields a Rs and RCT of ≈5.0 and 151.4 Ω, respectively. The resistance values calculated for Na-based TiS2−x are higher than those obtained for Li-based TiS2−x indicating that there is a larger electron transfer resistance for Na-ion transport between the vertically oriented TiS2−x matrix host and the electrolyte.

Since the fabrication of our vertically oriented TiS2−x arrays provide a well-defined percolating network comprising highly conducting nanobelts, it is now possible to utilize them as carbon- and binder-free cathode insertion hosts for Li- and Na-ion energy storage devices. The electrochemical performance of our vertically oriented TiS2−x nanobelt arrays was evaluated using CR2032 coin cells in a half-cell configuration using either Li or Na as counter/reference electrodes. The galvanostatic charge–discharge performances of our vertically oriented TiS2−x nanobelt arrays, when used as electrodes in Li(Na)-ion batteries are presented in Fig. 8. The cycling stability and rate capabilities at different current densities and as a function of desulfurization time are studied in depth. The vertically oriented TiS2−x nanobelt arrays used as cathodes were pyrolyzed for 48–144 h. Each cell was galvanostatically charged/discharged at a current density between 0.2–5C, for ten cycles at each current density. Fig. 8A shows the specific discharge capacities at various current densities for Li-based TiS2−x half-cells. At a current density of 0.2C, a discharge capacity of ≈216 mA h g−1 is obtained for a cathode that is pyrolyzed for 48 h. At the same current density (0.2C), there is almost a 50% loss in capacity (≈110 mA h g−1) when the TiS2−x cathodes are further pyrolyzed for 144 h. To explain the drop-off in capacity, we will have to refer to the following reaction using the lithiation process as a model:

 
xLi+ + TiS2 + xe → LixTiS2(3)


image file: c8ta05645e-f8.tif
Fig. 8 Specific discharge capacities at various current densities for vertically oriented TiS2−x nanobelt arrays used as cathode insertion hosts in Li (A) and Na (B) half-coin cells. Cathodes were pyrolyzed at 450 °C for several time intervals ranging from 48–144 h.

According to reaction (3), the Ti atom is reduced from a 4+ state to a 3+ state upon Li+ insertion. If there are S vacancies present and the Ti atom already exists in the 3+ oxidation state, that Ti atom can no longer contribute to the electrochemical capacity of the electrode. This would imply that the greater the number of S vacancies, the lower the overall discharge of the TiS2−x cathode. Additionally, as the current density rate is increased from 0.2 to 5C, a typically observed drop-off in specific capacity is also observed (regardless of the desulfurization exposure time). However, when the half-cells are again charged/discharged at a rate of 0.2C, we observe the recovery of the specific capacity closer to that of the theoretical capacity (239 mA h g−1). This capacity recovery suggests that the rapid intercalation/de-intercalation at higher current densities does not irreversibly affect the crystal structure and layered nature of the TiS2−x nanobelts. Furthermore, as previously discussed, the specific capacity for Li-based half-cell comprising our vertically oriented TiS2−x nanobelt arrays pyrolyzed for 48 h is 216 mA h g−1. Using the theoretical specific capacity for TiS2 of 239 mA h g−1, the loading of Li would be Li0.90TiS2−x. Fig. 8B shows a similar trend for the Na-based TiS2−x half-cells. The Na-ion intercalation proceeds similarly to that of what is observed for Li-ion intercalation (reaction (2)). However, due to the increased size of the Na-ion, it is argued that there is a limit on the amount of Na that can be incorporated into TiS2. This limit has been previously determined to be 0.8 Na equivalent atoms per TiS2 formula thus it is expected that the specific discharge capacity for Na-based TiS2 would be lower than that obtained for Li-based TiS2 half-cells.46,47 As depicted in Fig. 8B, at a 0.2C rate, the specific capacity for Na-based TiS2−x pyrolyzed for 48 h is 190 mA h g−1. Using the theoretical specific capacity, the loading of Na would be Na0.80TiS2−x. Ignoring any capacity losses due to sulfur vacancies, which are relatively low for the cathodes pyrolyzed for 48 h, the Na loading is in agreement with previously reported values.46,47 In terms of morphology and possible structural changes that our TiS2−x electrodes may undergo upon cycling, we observe that after closing the coin cell, the very top layer of the nanobelt array is completely pressed against the cap. However, the overall nanobelt morphology and structure are preserved before and after electrochemical cycling (Fig. S3).

The long-term cycling stability of the vertically oriented TiS2 cathode was studied by cycling the Li- and Na-based TiS2−x half-cells at a rate of 0.1C for 100 cycles. Fig. 9 shows the cycling stability for cathodes pyrolyzed at 450 °C for 48 h. The initial specific discharge capacity for a Li-based TiS2−x system is 217 mA h g−1 (Fig. 9A). The second cycle exhibits a discharge capacity of 221 mA h g−1 which is equivalent to an energy density of 433 W h kg−1. An initial lower discharge capacity followed by a stabilization of the capacity at higher values is quite unusual. Many cathode systems exhibit their highest capacity in the first discharge with an immediate drop in discharge in the second cycle before stabilizing.32,33 This drop in capacity between the first two discharge cycles is attributed to the growth of an interfacial layer on the surface of the cathode material. This layer is commonly referred to the solid electrolyte interface or to the more recently proposed solid permeable interface.34 At cycle 10, the discharge capacity of the battery reaches a steady value of 225 mA h g−1. Over the next 90 cycles, the TiS2 cathode shows remarkable robustness with a discharge capacity of 224 mA h g−1 which is translated to Li loading of Li0.94TiS2−x with a capacity retention of 99.5%. This significant discharge capacity retention alludes to the long-term stability of the Li-based TiS2−x system.


image file: c8ta05645e-f9.tif
Fig. 9 Cycling performance and coulombic efficiency for a half-cell comprising vertically oriented TiS2−x cathodes. The half-cells were charged/discharged at a 0.1C rate for 100 cycles. Cathodes were pyrolyzed at 450 °C for 48 h.

When discharged at a 0.1C rate, Na-based TiS2−x half-cells, on the other hand, yield an initial specific discharge capacity of ≈217 mA h g−1 (Fig. 9B). This specific discharge capacity is equivalent to a Na loading of Na0.91TiS2. This loading is higher than that reported as the top limit of only 0.8 Na-ions per TiS2 unit.47,66 Although we do not observe any initial irregularities in the discharge capacity upon cycling the Na-based TiS2 half-cells, their long-term stability is not as robust as that observed for the Li-based TiS2−x system. Over 100 cycles, 89.7% of the initial specific discharge capacity is maintained. Also, upon cycling, the specific discharge capacity does not fall below the 0.8 Na-ions per TiS2 unit value of 191.2 mA h g−1. This specific capacity is equivalent to an energy density of 372 W h kg−1. It is important to note that previous studies on TiS2 have predominantly focused on elucidating the electrochemical properties of bulk platelets with micron-sized dimensions. Diffusion of the Na-ion into the center of these platelets likely becomes energetically unfavorable. However, by reducing the dimensionality of individual TiS2 nanobelts as well as the fabrication of a well-defined hierarchical structure, the diffusion length to the middle of the nanobelt may be significantly reduced. In addition, the nanoscale morphology allows for greater lattice expansion without fracturing the crystal lattice.

To further investigate the role that Li- and Na- ion diffusion has on the electrochemical performance of our vertically oriented TiS2−x nanobelt arrays, GITT studies were performed. As described by Weppner and Huggins, GITT combines both transient and steady-state measurements to study the kinetic properties of an electrode.67 During GITT analysis, the coin cells are subjected to a short pulse of fixed current and the voltage response is measured as a function of time. The coin cells are then allowed to rest for a length of time until the current reaches steady-state. Once steady-state (cutoff limit) is reached, a current pulse is again applied to the coin cells. Fig. 10 shows the GITT discharge curves for both Li- and Na-based TiS2−x systems. Here, a 0.1C current was applied for 60 seconds and then the coin cells were allowed to rest for 2 hours. From the GITT measurements, the diffusion of the Li-ion or Na-ion into the TiS2−x matrix host can be calculated according to the following equation:

 
image file: c8ta05645e-t3.tif(4)
where D is the Li+ or Na+ ion diffusion coefficient, I is the applied current pulse, n is the number of electrons involved during ion diffusion (n = 1 for both Li+ and Na+), F is the Faraday constant, S is the electrode/electrolyte contact area (5.4 m2 g−1 for vertically oriented TiS2 nanobelt arrays, determined from N2 adsorption–desorption experiments), VM is the molar volume of the vertically oriented TiS2 nanobelt arrays, x is the loading of the Li or Na in the cathode material (LixTiS2−x and NaxTiS2−x), and τ is the current pulse time. The value for dE/dx is obtained by measuring the change in the equilibrium electrode potential after each current pulse. The value dE/d√t is determined from the voltage response plotted against the square root of the time during each current pulse. As depicted in Fig. 10A, a GITT capacity of 238 mA h g−1 is obtained in the first charge/discharge cycle for the Li-based TiS2 half-cell. This GITT capacity is slightly higher than the one obtained from the galvanostatic measurements (Fig. 8 and 9). Such discrepancy is due to the equilibration of the half-cell prior to applying a current pulse (the cell equilibration is also observed in the Na-based TiS2−x system, vide infra). The log diffusion coefficients (D) of Li+ into the vertically oriented TiS2−x nanobelt array cathode insertion host, as calculated by employing GITT are also presented in Fig. 10A. Here, the log diffusion coefficients for the Li-ion upon being inserted into the TiS2−x cathode show a rapid decrease at lower specific capacities followed by an increase in diffusion rate between 20–200 mA h g−1.


image file: c8ta05645e-f10.tif
Fig. 10 Galvanostatic intermittent titration technique (GITT) curves for Li (A) and Na (B) based TiS2 half-cells.

Finally, we observe a decrease in the diffusion rate as the cathode became fully intercalated (238 mA h g−1). Using the slope of the transient-voltage GITT curve (dE/dx) and the slope of the equilibrium voltage (dE/d√t), we calculate a mean diffusion rate of 1.4 × 10−11 cm2 s−1 for the Li-ions intercalated into our vertically oriented TiS2−x cathode insertion hosts. The calculated mean diffusion rate is on par with previously reported values for Li+ diffusion in layered TiS2 and cubic Ti2S4 electrodes.68,69 In contrast, the log diffusion coefficient of the Na-ion into the vertically oriented TiS2 cathode shows a steadier trend (Fig. 10B). However, there is greater variance in the diffusion rates. The mean diffusion rate for the intercalation of Na-ions into the TiS2−x cathode is calculated to be 5.5 × 10−11 cm2 s−1, which is comparable to previous reports using TiS2 nanoplates as the electrode.48 Thus by increasing the surface area of the electrode upon reducing its size to the nanoscale regime as well as the fabrication of anisotropic structures strongly favor Li(Na) insertion into the TiS2−x cathode with the advantage of having a highly conducting matrix host that does not require the addition of a binder or a conductive additive.

Conclusions

In summary, we present the synthesis and electrochemical characterization of vertically oriented TiS2−x nanobelt arrays. These highly anisotropic structures serve as binder- and carbon-free cathode insertion hosts for Li- and Na-ion energy storage devices. The control of the spatial geometry as well as the amount of sulfur vacancies present in the TiS2−x nanobelt arrays enable high rate electrochemical performances and high stabilities for both Li+ and Na+ insertion. We believe that the direct growth of vertically oriented TiS2 nanobelt arrays onto the current collector as well as the increased electrode/electrolyte surface contact area suppress the uncontrollable growth of an SEI layer and allow high Li(Na) diffusion coefficients. As a result, our vertically oriented TiS2 nanobelt arrays, when used as cathode insertion hosts deliver high and stable capacities without the need of any additives. Our findings provide fundamental insights for guiding the rational design of other highly anisotropic metal chalcogenide energy storage materials that could serve as binder- and carbon-free cathode (anode) insertion hosts for both monovalent and heterovalent energy storage devices.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the NSF MRSEC program at the Univ. of Utah under Grant No. DMR 1121252. C. G. H. acknowledges funding from the Utah Governor's Office of Energy Development. L. W. B. would also like to acknowledge the financial support from the Marion Milligan Mason Award for Women in the Chemical Sciences administered by the American Association for the Advancement of Science and from the Research Corporation for Science Advancement through a Cottrell Scholar Award. The authors gratefully acknowledge Mr David M. Parker for his help with acquiring the HRTEM data.

References

  1. G. E. Blomgren, J. Electrochem. Soc., 2017, 164, A5019–A5025 CrossRef.
  2. C. Julien, A. Mauger, A. Vijh and K. Zaghib, in Lithium Batteries: Science and Technology, ed. C. Julien, A. Mauger, A. Vijh and K. Zaghib, Springer International Publishing, Cham, 2016, pp. 29–68 Search PubMed.
  3. N. Nitta, F. Wu, J. T. Lee and G. Yushin, Mater. Today, 2015, 18, 252–264 CrossRef.
  4. A. Elia Giuseppe, K. Marquardt, K. Hoeppner, S. Fantini, R. Lin, E. Knipping, W. Peters, J. F. Drillet, S. Passerini and R. Hahn, Adv. Mater., 2016, 28, 7564–7579 CrossRef PubMed.
  5. B. Dunn, H. Kamath and J.-M. Tarascon, Science, 2011, 334, 928 CrossRef PubMed.
  6. N. Yabuuchi, K. Kubota, M. Dahbi and S. Komaba, Chem. Rev., 2014, 114, 11636–11682 CrossRef PubMed.
  7. C. Nithya and S. Gopukumar, Wiley Interdiscip. Rev.: Energy Environ., 2015, 4, 253–278 Search PubMed.
  8. I. Hasa, D. Buchholz, S. Passerini and J. Hassoun, ACS Appl. Mater. Interfaces, 2015, 7, 5206–5212 CrossRef PubMed.
  9. Q. Zhao, Y. Hu, K. Zhang and J. Chen, Inorg. Chem., 2014, 53, 9000–9005 CrossRef PubMed.
  10. X. Ren and Y. Wu, J. Am. Chem. Soc., 2013, 135, 2923–2926 CrossRef PubMed.
  11. D. Aurbach, H. Gizbar, A. Schechter, O. Chusid, H. E. Gottlieb, Y. Gofer and I. Goldberg, J. Electrochem. Soc., 2002, 149, A115–A121 CrossRef.
  12. D. Aurbach, Z. Lu, A. Schechter, Y. Gofer, H. Gizbar, R. Turgeman, Y. Cohen, M. Moshkovich and E. Levi, Nature, 2000, 407, 724 CrossRef PubMed.
  13. A. Ponrouch, C. Frontera, F. Bardé and M. R. Palacín, Nat. Mater., 2015, 15, 169 CrossRef PubMed.
  14. K. See, J. A. Gerbec, Y.-S. Jun, F. Wudl, G. D. Stucky and R. Seshadri, Adv. Energy Mater., 2013, 3, 1056–1061 CrossRef.
  15. M.-C. Lin, M. Gong, B. Lu, Y. Wu, D.-Y. Wang, M. Guan, M. Angell, C. Chen, J. Yang, B.-J. Hwang and H. Dai, Nature, 2015, 520, 324 CrossRef PubMed.
  16. H. Sun, W. Wang, Z. Yu, Y. Yuan, S. Wang and S. Jiao, Chem. Commun., 2015, 51, 11892–11895 RSC.
  17. E. I. Shkolnikov, A. Z. Zhuk and M. S. Vlaskin, Renewable Sustainable Energy Rev., 2011, 15, 4611–4623 CrossRef.
  18. Q. Li and N. J. Bjerrum, J. Power Sources, 2002, 110, 1–10 CrossRef.
  19. D. A. Stevens and J. R. Dahn, J. Electrochem. Soc., 2001, 148, A803–A811 CrossRef.
  20. Y. Wang, R. Xiao, Y.-S. Hu, M. Avdeev and L. Chen, Nat. Commun., 2015, 6, 6954 CrossRef PubMed.
  21. J. Sun, H.-W. Lee, M. Pasta, H. Yuan, G. Zheng, Y. Sun, Y. Li and Y. Cui, Nat. Nanotechnol., 2015, 10, 980 CrossRef PubMed.
  22. X. Xia and J. R. Dahn, Electrochem. Solid-State Lett., 2011, 15, A1–A4 CrossRef.
  23. M. Sathiya, K. Hemalatha, K. Ramesha, J. M. Tarascon and A. S. Prakash, Chem. Mater., 2012, 24, 1846–1853 CrossRef.
  24. N. Yabuuchi, M. Kajiyama, J. Iwatate, H. Nishikawa, S. Hitomi, R. Okuyama, R. Usui, Y. Yamada and S. Komaba, Nat. Mater., 2012, 11, 512 CrossRef PubMed.
  25. D. W. Han, J. H. Ku, R. H. Kim, D. J. Yun, S. S. Lee and S. G. Doo, ChemSusChem, 2014, 7, 1870–1875 CrossRef PubMed.
  26. X. Kong, T. Zhu, F. Cheng, M. Zhu, X. Cao, S. Liang, G. Cao and A. Pan, ACS Appl. Mater. Interfaces, 2018, 10, 8730–8738 CrossRef PubMed.
  27. K. T. Lee, T. N. Ramesh, F. Nan, G. Botton and L. F. Nazar, Chem. Mater., 2011, 23, 3593–3600 CrossRef.
  28. P. Moreau, D. Guyomard, J. Gaubicher and F. Boucher, Chem. Mater., 2010, 22, 4126–4128 CrossRef.
  29. Y. Fang, J. Zhang, L. Xiao, X. Ai, Y. Cao and H. Yang, Adv. Sci., 2017, 4, 1600392 CrossRef PubMed.
  30. F. Wang, S.-W. Kim, D.-H. Seo, K. Kang, L. Wang, D. Su, J. J. Vajo, J. Wang and J. Graetz, Nat. Commun., 2015, 6, 6668 CrossRef PubMed.
  31. D. Deng, ChemNanoMat, 2016, 3, 146–159 CrossRef.
  32. L. P. Wang, L. Yu, X. Wang, M. Srinivasan and Z. J. Xu, J. Mater. Chem. A, 2015, 3, 9353–9378 RSC.
  33. X. Wang, D. Chen, Z. Yang, X. Zhang, C. Wang, J. Chen, X. Zhang and M. Xue, Adv. Mater., 2016, 28, 8645–8650 CrossRef PubMed.
  34. E. Yang, H. Ji and Y. Jung, J. Phys. Chem. C, 2015, 119, 26374–26380 CrossRef.
  35. J. S. Cho, S. Y. Lee and Y. C. Kang, Sci. Rep., 2016, 6, 23338 CrossRef PubMed.
  36. I.-H. Chu, C. S. Kompella, H. Nguyen, Z. Zhu, S. Hy, Z. Deng, Y. S. Meng and S. P. Ong, Sci. Rep., 2016, 6, 33733 CrossRef PubMed.
  37. W. Luo, A. Calas, C. Tang, F. Li, L. Zhou and L. Mai, ACS Appl. Mater. Interfaces, 2016, 8, 35219–35226 CrossRef PubMed.
  38. B. Huang, Y. I. Jang, Y. M. Chiang and D. R. Sadoway, J. Appl. Electrochem., 1998, 28, 1365–1369 CrossRef.
  39. A. Zhou, J. Xu, X. Dai, B. Yang, Y. Lu, L. Wang, C. Fan and J. Li, J. Power Sources, 2016, 322, 10–16 CrossRef.
  40. C. Wan, X. Gu, F. Dang, T. Itoh, Y. Wang, H. Sasaki, M. Kondo, K. Koga, K. Yabuki, G. J. Snyder, R. Yang and K. Koumoto, Nat. Mater., 2015, 14, 622 CrossRef PubMed.
  41. S.-H. Chung, L. Luo and A. Manthiram, ACS Energy Lett., 2018, 3, 568–573 CrossRef.
  42. L. Wang, J. Zou, S. Chen, G. Zhou, J. Bai, P. Gao, Y. Wang, X. Yu, J. Li, Y.-S. Hu and H. Li, Energy Storage Mater., 2018, 12, 216–222 CrossRef.
  43. L. Ma, S. Wei, H. L. Zhuang, K. E. Hendrickson, R. G. Hennig and L. A. Archer, J. Mater. Chem. A, 2015, 3, 19857–19866 RSC.
  44. D. Y. Oh, Y. E. Choi, D. H. Kim, Y.-G. Lee, B.-S. Kim, J. Park, H. Sohn and Y. S. Jung, J. Mater. Chem. A, 2016, 4, 10329–10335 RSC.
  45. K. Sun, Q. Zhang, D. C. Bock, X. Tong, D. Su, A. C. Marschilok, K. J. Takeuchi, E. S. Takeuchi and H. Gan, J. Electrochem. Soc., 2017, 164, A1291–A1297 CrossRef.
  46. M. S. Whittingham, Science, 1976, 192, 1126 CrossRef PubMed.
  47. G. H. Newman and L. P. Klemann, J. Electrochem. Soc., 1980, 127, 2097–2099 CrossRef.
  48. Y. Liu, H. Wang, L. Cheng, N. Han, F. Zhao, P. Li, C. Jin and Y. Li, Nano Energy, 2016, 20, 168–175 CrossRef.
  49. T. Eriksson, A. M. Andersson, C. Gejke, T. Gustafsson and J. O. Thomas, Langmuir, 2002, 18, 3609–3619 CrossRef.
  50. L. Bao, L. Li, G. Xu, J. Wang, R. Zhao, G. Shen, G. Han and S. Zhou, Electrochim. Acta, 2016, 222, 685–692 CrossRef.
  51. N. Böckenfeld and A. Balducci, J. Appl. Electrochem., 2014, 44, 467–474 CrossRef.
  52. C. G. Hawkins and L. Whittaker-Brooks, ACS Appl. Nano Mater., 2018, 1, 851–859 CrossRef.
  53. J. O. Island, R. Biele, M. Barawi, J. M. Clamagirand, J. R. Ares, C. Sánchez, H. S. J. van der Zant, I. J. Ferrer, R. D'Agosta and A. Castellanos-Gomez, Sci. Rep., 2016, 6, 22214 CrossRef PubMed.
  54. M. H. Lindic, H. Martinez, A. Benayad, B. Pecquenard, P. Vinatier, A. Levasseur and D. Gonbeau, Solid State Ionics, 2005, 176, 1529–1537 CrossRef.
  55. M. S. Whittingham, Dalton Trans., 2008, 5424–5431 RSC.
  56. A. S. Aricò, P. Bruce, B. Scrosati, J.-M. Tarascon and W. van Schalkwijk, Nat. Mater., 2005, 4, 366 CrossRef PubMed.
  57. H.-C. Wu, H.-C. Wu, E. Lee and N.-L. Wu, Electrochem. Commun., 2010, 12, 488–491 CrossRef.
  58. I. Doberdò, N. Löffler, N. Laszczynski, D. Cericola, N. Penazzi, S. Bodoardo, G.-T. Kim and S. Passerini, J. Power Sources, 2014, 248, 1000–1006 CrossRef.
  59. M. E. Fleet, S. L. Harmer, X. Liu and H. W. Nesbitt, Surf. Sci., 2005, 584, 133–145 CrossRef.
  60. J. Benard and Y. Jeannin, Adv. Chem. Ser., 1963, 39, 191–203 CrossRef.
  61. A. H. Thompson, F. R. Gamble and C. R. Symon, Mater. Res. Bull., 1975, 10, 915–919 CrossRef.
  62. S. Gutmann, M. A. Wolak, M. Conrad, M. M. Beerbom and R. Schlaf, J. Appl. Phys., 2010, 107, 103705 CrossRef.
  63. S. Krischok, J. Günster, D. W. Goodman, O. Höfft and V. Kempter, Surf. Interface Anal., 2005, 37, 77–82 CrossRef.
  64. C. Jun, T. Zhan-Liang and L. Suo-Long, Angew. Chem., Int. Ed., 2003, 42, 2147–2151 CrossRef PubMed.
  65. J. Kasnatscheew, U. Rodehorst, B. Streipert, S. Wiemers-Meyer, R. Jakelski, R. Wagner, I. C. Laskovic and M. Winter, J. Electrochem. Soc., 2016, 163, A2943–A2950 CrossRef.
  66. B. G. Silbernagel and M. S. Whittingham, Mater. Res. Bull., 1976, 11, 29–36 CrossRef.
  67. W. Weppner and R. A. Huggins, J. Electrochem. Soc., 1978, 124, 1569–1578 CrossRef.
  68. A. J. Vaccaro, T. Palanisamy, R. L. Kerr and J. T. Maloy, Solid State Ionics, 1981, 2, 337–340 CrossRef.
  69. A. C. W. P. James, J. B. Goodenough and N. J. Clayden, J. Solid State Chem., 1988, 77, 356–365 CrossRef.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ta05645e

This journal is © The Royal Society of Chemistry 2018