1 Introduction

The rapid miniaturization of consumer electronics has led to several technological marvels like compact cell phones and virtual reality headsets. The batteries that power these devices need to be compact, lightweight, affordable, and energy dense with a long cycle life. These requirements can be fulfilled by the next generation of lithium-based batteries, i.e., lithium sulfur batteries (LSBs). LSBs afford several benefits like (i) a high specific capacity (theoretical: 1675 mAh/g), (ii) high energy density (theoretical: 2600 Wh/kg), (iii) low cost, due to high earth abundance and environmental friendliness of sulfur, and (iv) better safety [1, 2].

One of the principal hindrances to their widespread adoption is due to the high resistivity of sulfur (~ 10–31 S cm−1) and Li2S (the species formed upon discharging the cell). This issue is generally overcome by utilizing nano-sulfur and incorporating it into some porous conductive matrix. The conducting matrices employed, thus, far include reduced graphene oxide (rGO) and graphene frameworks [3], carbon nanotubes [4], conducting polymers like polyaniline [5] and polypyrrole [6], as well as conducting inorganic compounds like MXenes [7].

The notorious “polysulfide shuttle” [1] in LSBs is addressed by various approaches such as incorporation of polar species in the cathode that can restrain the soluble polysulfides by absorption/adsorption. The chemical species may also be loaded on the separator that can then restrain the polysulfides in the catholyte region, thus, disrupting the shuttle and enhancing electrochemical utilization. The species used for this purpose are widely varying, ranging from carbon based (such as CNT, graphene oxide), inorganic compounds (MoS2 [8], WP [9], TiOx [10]), intermetallic catalysts (CoTe [11]), porous metal–organic frameworks [12, 13], as well as metal clusters on doped carbon hosts [14, 15].

The complete conversion of sulfur to Li2S during discharge is accompanied by a very large (~ 80%) volume expansion which poses a huge roadblock for long-term cycling of the material. There have been several efforts to mitigate capacity fade due to the mechanical degradation and loss of conduction pathways during cycling by means of loading sulfur in porous conducting frameworks [16]. This approach ameliorates capacity fade at the cost of active mass dilution by the encapsulating porous framework.

Moreover, conventional electrode preparation involves making a slurry composed of active material, conductive carbon, and a binder which is then coated onto a metallic current collector. Due to large volume changes during cycling, the coated material peels off from current collector and there is loss of active mass and consequent capacity fade results. Typically, the active material is a composite of sulfur with another material which contributes conductivity and offers some form of physical/chemical hindrance to polysulfide migration. Since sulfur itself is a good binder [17] and there is a conducting component already present in our active composite, we propose to eliminate the metallic current collector, binder as well conducting carbon from an assembled cell. The active material is directly deposited by scalable vacuum filtration method onto one side of glass microfiber separator. This coated separator now takes on the role of (i) a porous framework holding the active material and accommodating volume changes during cycling, (ii) enable higher loading due to increased surface area compared to a metallic current collector, (iii) act as current collector by pressing intimately with stainless steel casing, and (iv) act as the separator itself. This approach leads to significantly higher active material (sulfur) loading on the final coated separator compared to conventional routes.

One of the approaches utilized to eliminate the Al current collector is to fabricate a free-standing cathode. Park et al. [18] have reported such an approach by fusing conductive carbon nanotubes (CNTs) with cellulose to make a free-standing 3D matrix for sulfur loading. These free-standing electrodes are then used alongside conventional glass microfiber separator for application in lithium sulfur batteries. They reported good initial specific capacity of 940 mAh/g at an areal current density of 1.57 mA/cm2. N-doped carbon nano-fiber sheets impregnated with Mn3O4 nanoparticles were used as sulfur host by Chen et al. [19] to fabricate free-standing cathode. The synthesis protocol consisted of multiple steps, including electrospinning of template polymer with oxide precursor, carbonization under inert atmosphere and subsequent sulfur loading. They achieved a reversible areal capacity of over 8 mAh/cm2. Bharti et al. [20] reported carbonized bacterial cellulose as a free-standing sulfur host that when coupled with bacterial cellulose as separator delivers a capacity of 740 mAh/g at 1C.

While this approach eliminates the need for aluminum current collector, the synthesis route for free-standing electrodes is generally cumbersome and a separator as well as binder are still needed. In this work, the performance and cycling characteristics of coin cell utilizing a coated separator with a high loading of S8, owing to the lack of binder, conducting carbon as well as aluminum current collector in the present design, have been assessed.

2 Experimental

2.1 Synthesis

The detailed synthesis of MXene (Ti3C2Tx), graphene oxide (GO), and reduced graphene oxide (rGO), and ternary composite MXene/rGO/S (MGS) are discussed in our previous report [7]. Briefly, the synthesis of MGS involves addition of sodium thiosulfate and oxalic acid to an aqueous dispersion of MXene and rGO (refer Fig. 1, for a graphical schematic). The acid catalyzed disproportionation of thiosulfate leads to formation of nano-sulfur on the surface of rGO and MXene. The formed composite is collected by centrifugation, washed with distilled water, and then re-dispersed in distilled water for further use. Synthesis procedures for binary MXene/S and rGO/S samples may be found in supporting information.

Fig. 1
figure 1

Synthesis schematic for fabrication of MGS loaded on glass microfiber disks for use as electrode cum separator

Full size image

2.2 Electrode preparation

To enable higher mass loading without the use of binder, conductive carbon, or metallic current collector, the aqueous dispersion of composite was filtered onto a glass microfiber separator (Whatman Glass Microfiber GF/D; Product No.-1823-047) assisted by a vacuum pump. The coated separator was then dried in hot air oven at 50 °C for 12 h, and thereafter, 12-mm-coated disks were cut using a circular die punch. A schematic of the procedure for synthesis of MGS composite as well as electrode preparation is depicted in Fig. 1.

2.3 Physical characterization

Powder X-ray diffractograms (PXRD) of all samples were recorded by using Cu Kα radiation (λ = 1.5406 Å) using a scan rate of 2°/min on a PANalytical X’Pert Pro unit with an external silicon standard. The sulfur loading in the composites was estimated by means of thermogravimetry analysis using SETSYS Evolution 1750 (SETARAM, France). The composite samples were heated under argon flow to 230 °C in alumina crucibles at a heating rate of 10 °C /min and held isothermally till a flat baseline was achieved.

The morphology of samples was observed by means of scanning electron microscopy (SEM) images recorded on Carl Zeiss Auriga 4553. Fourier transform infrared (FTIR) spectra of samples were recorded on a Bruker instrument (Model Alpha II) in the range 400–4000 cm−1 in attenuated total reflectance mode.

2.4 Electrochemical evaluation

Coin cells (CR2032 size) were assembled using MGS-loaded microfiber separator disks along with Li metal disks as counter electrode in an Argon-filled glove box (Mbraun). 1 M lithium bis(-trifluoromethanesulphonyl) imide (LiTFSI) in 1:1 v/v DOL/DME containing 1 wt% LiNO3 additive was employed as electrolyte. A 24-h rest was allowed for all assembled cells, and all electrochemical measurements were made at room temperature. The galvanostatic charge–discharge cycles were performed on a multichannel battery tester (Neware Battery Test System, Shenzhen Neware Electronic Co, China), at varying current densities, in the potential range of 1.5 to 3 V.

Electrochemical impedance spectroscopy (EIS) as well as Cyclic Voltammetry on the coin cells were performed using an electrochemical workstation (OrigaFlex; Origalys). The EIS was done in the frequency range of 0.1 Hz to 1 MHz while CV was performed at varying scan rates in the potential window ranging from 1.5 to 3 V.

3 Results and discussions

The powder X-ray diffractograms (PXRD) of rGO, synthesized MXene (Ti3C2Tx), and MXene/rGO/S (MGS) composite, along with PCPDF database entry for sulfur are been presented in Fig. 2a. The peaks in XRD pattern of MGS composite can be indexed to crystalline S8 (refer bottom panel for PCPDF card 08-0247) which is the major component of the composite. The broad peaks at 8.66° and 17.95° in the PXRD trace for MXene can be attributed to (002) and (004) reflections, respectively. Also, notably the major (104) peak of MAX phase at 38.54° is absent indicating complete removal of Al layers of MAX phase to form MXene [7, 21]. The PXRD trace for rGO shows a broad peak at 24° with no peaks around 10° (due to graphene oxide) indicating complete reduction of GO to rGO [22].

Fig. 2
figure 2

a PXRD patterns of MGS, MXene, rGO, and PCPDF entry b TG of MXene/rGO/S (MGS)-loaded glass microfiber separator at 230 °C under Ar flow. c FTIR spectra of Mxene and rGO

Full size image

The TG of MGS-loaded separator under Ar flow at 230 °C (Fig. 2b) shows mass loss occurring in a single step, indicating a sulfur loading of 17 wt%. Thus, the 12 mm disks of MGS had a sulfur loading of 3.63 mg (17 wt% of MGS-loaded glass microfiber disk, refer supporting information for detailed calculation) as determined by thermogravimetry. This leads to an areal loading of 2.52 mg cm−2 (3.63 mg/1.44 cm2). Each assembled coin cell utilizes this 12 mm disk as both electrode and separator, and hence, each cell contains a sulfur loading of 3.63 mg of S8 on the cathode.

The chemical bonds in rGO and MXene were characterized by means of FTIR spectra depicted in Fig. 2c. A broad band around 3400 cm−1 due to absorbed water is observed for both rGO and MXene [23], with MXene also exhibiting the Ti–C–Ti skeletal vibrations in the broad region from 2600 to 3100 cm−1.[23] The FTIR spectra of MXene also exhibit peaks at 670, 1126, and 1551 cm−1 due to Ti–O deformation, C–F stretch, and –C=C– stretch, respectively. The peaks at 1426 and 1120 cm−1 due to terminal C–OH and C–H groups, respectively, are seen for both rGO and MXene. The weak peaks at 1040 and 1710 cm−1 are due to C–O–C and C=O stretching of residual oxide functionalities remaining in rGO after hydrothermal reduction from graphene oxide [24].

The SEM images of rGO, MXene, as well as the sulfur loaded composite MGS are depicted in Fig. 3a–d. The rGO flakes exhibit a crumpled sheet like morphology while the layered structure of MXene can be clearly visualized in the image (annotated with a yellow box). The image (Fig. 3c) of composite MGS (powder) shows bright flakes of MXene covered by rGO sheets. The surface of rGO in MGS appears smoother compared to bare rGO hinting at a surface coating of sulfur. Figure 3d depicts an image of MGS-loaded microfiber separator, where a bright chunk of MGS can be seen embedded in a mesh of glass microfibers. This three-dimensional distribution of MGS chunks on the microfiber separator should assist in better electrolyte wetting and faster lithium diffusion. The detailed physical characterization of the MGS composite, including Raman and XPS spectra, may be found in our earlier report [7].

Fig. 3
figure 3

SEM images of a rGO, b MXene, c MGS, and d MGS on glass microfiber

Full size image

To visualize the distribution of sulfur in each of the sulfur loaded composites, namely MGS, MXene/Sulfur (MS), and rGO/sulfur (GS), the EDAX elemental mapping is depicted in Fig. 4. The elemental maps of carbon, oxygen, sulfur, and titanium are depicted in red, yellow, green, and blue colors, respectively. Sulfur along with oxygen is found to be uniformly distributed in all samples, indicating that a uniform coating on MXene or rGO is achieved in all cases, with the presence of oxygen indicating some surface oxidation. The signal from carbon and titanium (in MGS and MS) samples is comparatively weaker, due to the coating of sulfur onto particles of rGO and MXene. Their distribution, however, allows for an unbroken conductive chain to aid in conductivity.

Fig. 4
figure 4

EDAX elemental mapping of a MXene/rGO/sulfur, b MXene/sulfur, and c rGO/sulfur composite samples

Full size image

3.1 Electrochemical characterization

To assess the reversibility of the MGS-loaded separator electrodes during electrochemical cycling, cyclic voltammetry studies were conducted. The cyclic voltammogram (CV) of a coin cell in the potential window ranging from 1.5 to 3 V at a scan rate of 0.1 mV/s is depicted in Fig. 5. The first cycle (reproduced as inset for better clarity in Fig. 5) exhibits two broad cathodic peaks at 1.99 and 1.85 V, which merge into a single broader peak at 1.9 V from second cycle onwards. Conventional MGS electrodes made by coating a slurry of MGS, conductive carbon (super P) and sodium alginate binder onto aluminum current collectors shows cathodic peaks at 2.22 V and 1.99 V [7]. The peaks in present case are broader and shifted ~ 0.23 V and 0.14 V compared to peaks due to conventional MGS electrodes. The sharpness of CV peaks in case of conventional electrodes may be attributed to the higher electronic conductivity of conventional-coated cathodes, due to addition of conductive carbon as well as lower overall loading of resistive sulfur. The two cathodic peaks (observed in 1st cycle) may be assigned to the reduction of sulfur to higher-order (Li2Sx; 8 ≤ x ≤ 4) soluble lithium polysulfides (LiPS) and their subsequent reduction to insoluble Li2S2 and Li2S, respectively. Ideally, these peaks should be in the ratio 1:3 since the 1st peak (i.e., reduction of S8 to higher-order LiPS) contributes 25% of theoretical capacity while the subsequent peak at lower potential (i.e., due to reduction of higher-order LiPS to Li2S) yields the remaining 75% [25]. In this case, it is observed that the second peak is only slightly larger than the first during the first cycle and are thereafter merged into a single broad peak. A similar broad single cathodic peak for sulfur has been reported previously by Jeddi et al. [26]

Fig. 5
figure 5

The cyclic voltammograms of the MGS cell at 0.1 mV/s scan rate. Inset shows first cycle separately

Full size image

A single anodic peak is observed for all cycles at 2.78 V as compared to 2.49 V for the conventional-coated electrode [7]. The single peak corresponds to the conversion of lithiated Li2S to S8 via the intermediate polysulfide phases. All cycles except the first, overlap each-other without any significant peak shift indicating good stability and reversible lithiation/de-lithiation during cycling.

The galvanostatic charge/discharge performance of coin cells assembled with MGS and GS-loaded separator based electrodes at varying current densities is depicted in Fig. 6a. The initial discharge capacity exhibited by MGS (GS) at 100 mA/g current density is 593 (284) mAh/g which declines during the initial few cycles for MGS before stabilizing at 472 (358) after 10 cycles. Thus, the average specific discharge capacities afforded by MGS (GS) are 472 (358), 423 (264), 389 (41), and 330 (3) mAh/g at a discharge rate of 100, 200, 300, and 500 mA/g, respectively. The introduction of MXene in ternary MGS is clearly more beneficial at higher current densities.

Fig. 6
figure 6

a Galvanostatic cycling of varying current densities. Charge/discharge voltage profiles for b MGS c GS. The legend for b and c is shared

Full size image

It is also pertinent to note that cells assembled with MS failed to cycle even at the lowest current density. MXene by itself (~ 10 wt% in the composite) is probably unable to overcome the high resistivity of S8 and allow for proper electrochemical utilization of sulfur.

The voltage profiles of MGS when cycled at varying current densities are depicted in Fig. 6b. In case of MGS, two plateau regions can be clearly visualized at 2.25 V and 2 V in the discharge curve at 100 mA/g, due to reduction of S8 to soluble higher-order LiPS and their subsequent reduction to insoluble Li2S, respectively. A single plateau at 2.46 V is observed in the charge curve. The discharge (charge) plateaus shift towards lower (higher) voltages with increasing current density, pointing towards polarization in the material and inadequate conductivity. The charge and discharge plateaus in case of conventional-coated electrodes did not show any shift with increasing current density [7]. Moreover, the slope of the line joining the two plateaus was much higher in case of the coated electrodes compared to present separator loaded ones, hinting at sluggish electron transfer due to lower conductivity compared to coated electrodes.

The voltage profiles of GS are depicted in Fig. 6c, which shows a single broad plateau during discharge and charge steps, owing to high sulfur loading and resultant high resistance. The specific capacities observed at higher current densities (300 and 500 mA/g) are so small that no clear features in the voltage profile are discernable.

To probe the stability of the cells under extended cycling, MGS and GS were galvanostatically cycled at 100 mA/g for 200 cycles (Fig. 7a). The discharge capacities for MGS and GS cells cycled at 100 mA/g after 5 cycles were 495 and 295 mAh/g, with 483 and 281 mAh/g remaining after 200 cycles yielding a capacity fade of 0.012 and 0.023% per cycle, respectively.

Fig. 7
figure 7

Galvanostatic cycling performance at a 100 mA/g b 500 mA/g

Full size image

The cycling stability of MGS at higher current density was probed by cycling the cells at 500 mA/g (Fig. 7b). At the higher discharge rate, MGS exhibits an initial specific discharge capacity of 378 mAh/g, with 239 mAh/g retained after 750 cycles. This corresponds to a capacity fade of 0.05% per cycle at higher current density.

A comparison of cycling performance of separator loaded MGS vis-à-vis conventional-coated electrodes is tabulated in Table 1. Despite the higher sulfur loading on electrode in the present work, both the specific discharge capacity and capacity fade of separator loaded MGS is found to be superior to that of conventional-coated cathode at low current density. The 3D mesh-like morphology of glass microfiber separator upon which MGS composite is loaded allows for better electrolyte wetting of active material. The electrochemical utilization of active material is also improved due to physical retention of soluble polysulfides formed during cycling in the cathode/separator manifold.

Table 1 Comparison of galvanostatic cycling performance for conventional-coated cathodes and separator loaded cathode
Full size table

3.2 Post-mortem analysis

The digital photographs of lithium disk anode and anode facing side of MGS-loaded separator from a cycled cell are depicted in Fig. S1. The lack of yellow color on the anode facing side of the separator demonstrates the efficacy of MGS in restraining the soluble LiPS to the catholyte region. The circular black imprint on the Li anode disk corresponding to the shape of MGS-loaded separator is due to continuous Li+ stripping and plating during galvanostatic cycling.

The electrochemical impedance spectra of both fresh and cycled cells of MGS are depicted in Fig. 8a, with both exhibiting semi-circular depressions in the high- and medium-frequency regions. Only the fresh cell shows an oblique line attributed to Warburg contribution in the low-frequency region. The semi-circles are due to resistance contributions from solid electrolyte interphase (RSEI), resistance due to charge transfer process (RCT) as well as the resistance due to all other cell components in the conduction pathway (RS). The sloping Warburg line is due to Li+ diffusion process in the electrode material. The equivalent circuit used for fitting impedance data of fresh cell is depicted in Fig. 8c. A similar circuit without the Warburg element is used for fitting the cycled cell’s Nyquist curve, and the results are tabulated in Table 2.

Fig. 8
figure 8

a Nyquist plots of fresh and cycled MGS cells. b Z’ – ω−1/2 plots for fresh MGS cell in low-frequency region c Equivalent Circuit used for fitting Nyquist plots, fit values in Table 2

Full size image
Table 2 Resistance values and calculated diffusion coefficients obtained from fitting impedance data to equivalent circuit
Full size table

It is observed that the resistance of the cells clearly reduces upon cycling with RSEI component showing the largest decline due to reduction in particle size with cycling. Table 2 also lists the Warburg factor (\(\sigma\)) and lithium diffusion coefficient (\({D}_{Li+}\)) that were estimated from Fig. 8b using the following equations [5]:

$$Z^{\prime} = R_{{\text{S}}} + R_{{{\text{CT}}}} + \sigma \omega^{ - 0.5} ,$$
$$D_{{{\text{Li}} + }} = \frac{{R^{2} T^{2} }}{{2A^{2} n^{4} F^{4} C^{2} \sigma^{2} }},$$

wherein, R, T, A, n, F, C and \(\sigma\) are the universal gas constant, absolute temperature, electrode surface area, number of redox active electrons per molecule, Faraday constant, lithium-ion concentration, and Warburg constant, respectively. \(\sigma\) here is obtained from the slope of Fig. 8b. The lithium diffusion coefficient for a fresh cell was estimated to be 4.34 × 10–7 cm2 s−1 from the low-frequency region of the impedance data.

4 Conclusions

An alternative to the traditional approach of slurry coating of active material on metallic current collector was proposed. Since the composite is directly filtered onto the glass microfiber separator, the role of aluminum current collector, binder, as well as conductive carbon are eliminated. This leads to higher sulfur loading as well as enhanced electrochemical utilization in comparison to conventional-coated electrodes. At a current density of 100 mA/g, after 5 cycles, a specific discharge capacity of 495 mAh/g is obtained, with 97.5% capacity retained after 200 cycles, i.e., a capacity fade of 0.01% per cycle. The specific discharge capacity obtained at a higher discharge rate of 500 mA/g was 378 mAh/g, with 239 mAh/g retained after 750 cycles. The superior performance and capacity retention can be explained by the better electrolyte wetting and electrochemical utilization of active material dispersed on the three-dimensional mesh-like framework of glass microfiber separator. The relatively higher impedance values of the separator loaded electrodes compared to conventional-coated electrodes may be improved upon by addition of other conductive agents to the MGS composite.