1 Introduction

Rechargeable batteries with high energy density, long cycle life, and enhanced safety are urgently required for the utilization in electric vehicles and stationary energy storage systems [1, 2]. Despite having a high theoretical specific capacity and energy density, the development of lithium–sulfur (Li–S) batteries for commercial use is impeded by technical bottlenecks such as low conductivity of elemental sulfur, and lithium polysulfide shuttling between the two electrodes [3,4,5,6]. The introduction of permselective membranes has also widely been used to confine lithium polysulfides [7]. However, the introduction of permselective membranes adversely increases the cell resistance and reduces the Coloumbic efficiency. In order to address this issue, electrolyte additives are commonly used particularly LiNO3 has been commonly explored as a co-salt or electrolyte additive which protects the lithium metal anode by creating a surface protective film [8,9,10]. Although this protective film can efficiently suppress the redox shuttle reactions, the reduced products adversely deposited at the cathode surface and deteriorates its performance below 1.6 V. Being a strong oxidant in the acidic medium, it also increases the risk of safety issues when it is not completely consumed [11].

The formation of a robust SEI layer as a consequence of electrolyte reduction reaction serves a vital role in ensuring the cycle life and Columbic efficiency of lithium batteries. Therefore, achieving SEI with optimum chemical, thermal and mechanical properties in addition to both electronic and ionic conductivities is a paramount issue [12]. In order to address these challenges numerous studies have been conducted on anodes that substitute lithium metal with new metal alloys [10]. Compared to the state-of-the art graphite-based anode materials, silicon-based anodes are superior owing to their higher theoretical capacity (4200 mAh g−1 for Li4.4Si), and lower lithiation voltage platform (0.4–0.6 V vs Li/Li+) relative to other potential anode materials such as Sn, Ge, Sb, Al, etc. [13]. Yet their prospective uses in lithium ion and Li–S batteries are severely hampered by the enormous volume expansion (approximately 300%) upon charge–discharge cycling, low electrical conductivity, and highly unstable SEI layer formation [14].

This volume change leads silicon particles to pulverize, resulting in extra electrolyte consumption for the development of a new SEI layer. This increases cell impedance, lowering lithium battery performance with increased cycling. Different approaches have been adopted to conquer these challenges that include facile development of highly porous electrode materials comprising nano- and micro particles sized particles [15, 16], incorporation of electrolyte additives [17] and novel bi-functional binders [18].

Because of the high specific discharge capacity, lower working potential, low cost, and being environmental benign, silicon monoxide and its derivatives have been widely employed used as anode materials for lithium-ion and Li–S batteries [19]. Lithium oxide (Li2O) and lithium silicates (LiSiO4) are generated irreversibly when lithium is inserted into SiOx. The generation of Li2O and LiSiO4 results in the development of a buffer matrix that allows for volume expansion while also forming stable SEI layer [20]. Even though Li2O and LiSiO4 formation is advantageous in preventing the pulverisation of SiOx complexes, many challenges remain, including poor Coulombic efficiency, inferior rate capability, and exceedingly short cycling life at higher current rates due to intrinsically low electrical conductivity and poor Li-ion diffusion [12]. Specifically, the stability of interface between SiOx and electrolyte when cycling with various electrolyte additives remains a difficult task [21].

Choi et al. examined the role of fluoroethylene carbonate (FEC) as additive in the electrochemical performance of Si thin film/Li half-cells [22]. It was revealed that cycling the battery using FEC resulted in a lesser porous SEI layer covering the thin-film electrode of silicon, which reduced the overall surface area uncovered to the electrolyte for side reactions. While electrodes with an electrolyte free of FEC created unstable surfaces like LiPxFy and derivatives of ethylene carbonate (EC), the presence of FEC made it easier to develop a stable structure of SEI layer composed of Li–F and Si–F. Similar to this, Chen et al. investigated how the inclusion of vinylene carbonate (VC) to an electrolyte affected the cycling performance of a Si film anode by deposition of ion-beam sputtering, and the investigators found that the presence of VC improved cycling performance [23]. Electrodes with VC additives showed increased film resistance for lithium-ion migration, implying the unsuitability of VC for high power applications [24].

In the case of Li–S batteries, electrolyte with different additives is extensively used to create thick layer to shield lithium anode from the deposition of lithium polysulfide. As an example, LiNO3 was recognized as a possible component that can generate protective layer on Li metal anode [25, 26]. Even though recent research indicated that sulfur composite cathodes with lithium metal anodes have substantial energy densities and cycleability, it is critically needed to investigate adopting alloy-type anode materials before they are introduced to the market for rechargeable batteries. Very recently the composite electrodes composed of SiOx–Si–C has been widely explored as anode materials.

In the current investigation, a high-capacity anode made of a SiOx–Si–C composite was chosen. The mechanically stable amorphous phase SiOx (5 micron) serves as a shield to compensate the volume changes of the Si (10 nm) while the added carbon particles improve the electrical contact and thus cycling stability of the anode. Finally, the interfacial and electrochemical characteristics of the composite anode composed of SiOx–Si–C cycled with 1 M LiTFSI in (DOX) and (TEGDME, 1/1, v/v) containing 5 wt% LiFSI and LiBOB, as electrolyte additives, have been studied systematically. LiBOB is not only a potential salt for Li-ion batteries but also thermally more stable and environmentally benign because the hydrolytic decomposition products are non-toxic and corrosion resistant [27]. The SEI formed on graphite demonstrated better electrochemical performance, owing to its denseness and tenacity. On the other hand, LiFSI has been recognized as a potential salt owing to its lower sensitivity to hydrolysis and with enhanced thermal and electrochemical stabilities [28, 29].

2 Experimental

2.1 Materials

Silicon monoxide (SiOx of 5-micron, OTC, Japan), silicon nanopowders (10 nm TCI, Japan), LiTFSI, poly acrylic acid (PAA) (Mw ~ 4,000,000, Sigma Aldrich, USA) were used as such. The non-aqueous electrolyte comprising of 1 M LiTFSI in very high purity solvents 1,3-dioxolane and tetraethylene glycol dimethyl ether (both Spectrochem) in a ratio of (1:1, v/v). As a source for conducting carbon, the Super P was employed. The content of electrolyte additives, lithium bis(fluorosulfonyl)imide (LiFSI), and lithium bis(oxalato)borate (LiBOB) was fixed as 3 wt%.

2.2 Preparation of SiOx electrodes and their characterizations

With 5% of PAA as a binder, an electrode slurry of SiOx, silicon, and Super P in the proportions 70:12.5:12.5 was prepared. The slurry was coated on a copper foil to prepare the anodes, which were then dried for 12 h at 100 °C. With polypropylene (Celgard CG) as separator, the lithium metal disc as the counter electrode, the standard CR2032 coin cells were assembled in an-argon filled glove box for further characterization. The electrochemical workstation VSP3 Biologic (France) was used to conduct the EIS experiments in the frequency range between 1 MHz and10 mHz. By establishing the cut off potentials in the voltage range of 0.005–2.3 V at 0.1 C rate, galvanostatic charge–discharge investigations were carried out using an Arbin Instrument Testing System model BT-2000 (USA).

SEM (TESCAN VEGA3, USA) was used to examine the changes in morphology of the SiOx–Si–C electrode after being subjected to cycling with different electrolyte additives. To examine the surface properties of composite electrode, FT-IR (Bruker, Germany) was employed. The XPS (Thermo Scientific, USA) was done upon cycling and the data were analyzed by CASA software. The solid-state nuclear magnetic resonance (NMR) for 7Li and 19F, 29 Si, 13C and 1H was performed by a Bruker model (AVANCE III HD) at 16,000 rpm equipped with a rotor.

3 Results and discussion

The 1st charge–discharge profiles of Li/SiOx–Si–C lithium half-cell with additive-free, LiFSI and LiBOB are shown in Fig. 1i–iii. The voltage profile corresponds to the initial discharge (lithiation) of SiOx–Si–C with additive-free, LiFSI, and LiBOB-added electrolytes appear similar as shown in the Fig. 1. There is a significant change in the values of discharge capacity which suggests that the additive has a vital role in the formation of a SEI layer at the alloying and de-alloying processes. The discharge capacity vs. cycle number of Li/SiOx–Si–C cells is illustrated in Fig. 2i–iii with different electrolytes. The Li/SiOx–Si–C cell with LiFSI additive electrolyte exhibited better capacity retention. On its first and 50th cycles, the cell achieved a discharge capacity of 1900 and 585 mAh g−1, respectively, with a 95% Coloumbic efficiency. The Li/SiOx–Si–C cell without additive, on the other hand, provided a discharge capacity of 300 mAh g−1 on its 50th cycle while having an initial discharge capacity of 1300 mAh g−1. Almost a similar trend has been observed in the Li/SiOx–Si–C cell with LiBOB-added electrolytes. Generally, SiOx or lithiated silicon electrodes allow for hundreds of cycles. The cycling trials in the present study, however, were stopped after 50 charge–discharge cycles as this number is adequate to study the morphological and surface chemistry of SiOx–Si–C electrode with the nature of electrolyte additives. The observed differences in cell cycling performance indicated differences in the surface property of SEI layers generated, depending on the nature of the electrolytes used. The cycled electrode materials were cautiously deconstructed in a glove box filled with argon and investigated using SEM, FTIR, solid NMR, and XPS analyses in order to obtain more information on the surface properties.

Fig. 1
figure 1

Charge–discharge studies (i) without additive (ii) with LiFSI and (iii) with LiBOB

Full size image
Fig. 2
figure 2

Cycle number versus discharge capacity of Li/SiOx–Si–C half cell (i) without additive (ii) with LiFSI (iii) with LiBOB

Full size image

The cyclic voltammograms of Li/SiOx–Si–C cell with additive-free, LiFSI and LiBOB-added non-aqueous liquid electrolytes are shown respectively in Figure SI 1i–iii. The cyclic voltammograms of Li/SiOx–Si–C cell in three types of electrolytes show complete reversible Li–Si alloying and de-alloying processes upon surface layer generation. The current peaks have been recorded at 0.16 and 0.03 V during the first cathodic sweep, which were attributed to lithium–silicon alloying process [30]. The matching anodic current peaks, which resulted from the Li–Si dealloying, were recorded respectively at 0.46 and 0.30 V. The peak in the second cycle disappeared, proving that the cathodic peaks in the initial cycle were caused by an irreversible reaction that turned elemental Si into lithium silicates (Li2Si2O5, Li2SiO3, and Li4SiO4) [31,32,33]. For the Li/SiOx–Si–C cell with LiFSI-added electrolyte, not notable difference is seen in the CV cures. The magnitude of the oxidation and reduction peaks increased as the cycle progressed on for the LiBOB and additive-free electrolytes (Figure SIi and iii), which is explained by the pulverisation of silicon and amorphization during cycling [34, 35].

3.1 XPS studies

Surface films formed play an important part in defining the electrochemical performance of Li–S cells and the properties of the films are primarily determined by the electrolyte compositions such as the kind of lithium salt and additives. In this study, the cycled electrodes were disassembled in an argon filled glove box and carefully washed twice with TEGDME and then vacuum dried to eliminate the presence of added lithium salt (LiTFSI), if any. The post-cycling high-resolution XPS spectra recorded from SiOx–Si–C-composite electrodes with additive-free, LiFSI and LiBOB-added non-aqueous liquid electrolytes are shown respectively in Figs. 3, 4 and 5.

Fig. 3
figure 3

XPS of C 1S of SiOx–Si–C electrodes before and after cycling. Row A (i) bare electrode, (ii) without additive after (ii) 1st (iii) 5th (iv) 20th cycles. Row B (ii) after (ii) 1st (iii) 5th (iv) 20th cycles with LiFSI additive. Row C (ii) after (ii) 1st (iii) 5th (iv) 20th cycles with LiBOB additive

Full size image
Fig. 4
figure 4

XPS of O 1S of SiOx–Si–C electrodes before and after cycling. Row A (i) bare electrode, (ii) without additive (ii) after 1st (iii) 5th (iv) 20th cycles. Row B (ii) after 1st (iii) 5th (iv) 20th cycles with LiFSI additive. Row C (ii) after 1st (iii) 5th (iv) 20th cycles with LiBOB additive

Full size image
Fig. 5
figure 5

XPS of F 1S of SiOx–Si–C electrodes before and after cycling. Row A (i) bare electrode, (ii) without additive after (ii) 1st (iii) 5th (iv) 20th cycles. Row B (ii) after (ii) 1st (iii) 5th (iv) 20th cycles with LiFSI additive. Row C (ii) after (ii) 1st (iii) 5th (iv) 20th cycles with LiBOB additive

Full size image

It is clear from the Fig. 3ai, that the spectra appeared for the SiOx–Si–C composite electrode at 284.44, 285.07 and 288.15 eV for C 1s can be ascribed to C–C, C–O and C=O correspondingly. Upon 1st cycle the observed new peak at 293.19 eV can be ascribed to the development of CF3 group. Furthermore, no discernible changes could be observed in the XPS peaks after 5th and 20th cycle (Fig. 3 Row A–C).0 The formation of a similar compound is identified for the LiFSI and LiBOB-added electrolytes. Additionally, a new peak at 289.88 to 289.99 has been observed after 1st, 5th and 20th cycle for both LiFSI and LiBOB-added electrolytes which is attributed to the Li2CO3 (Fig. 3 Row A–C).

Similarly, the surface films of O1s XPS spectra formed on the SiOx–Si–C composite electrodes upon cycling are shown in Fig. 4 for the bare, upon 1st, 5th and 20th cycles respectively. The O1s spectra of the SiOx–Si–C-composite electrode cycled with electrolyte having no additive shows a deconvoluted peak at 531.22 eV, credited with the development of Li2CO3 and ROCO2Li species (Fig. 4 Row A(i)). The nature of the additives affects the weight of the deconvoluted peaks. A similar appearance of peaks is seen for all the systems examined (Fig. 4 Row Bii–iv). It is worth noting an addition peak is observed for the LiFSI and LiBOB-added electrolytes at 530.0 eV which is assigned to the formation metal oxides. However, it disappears in the subsequent cycles for the LiFSI-added electrolytes and remain unchanged for the LiBOB-added electrolytes.

In a similar way, regardless of the additive utilized, the development of F 1s peak at 688.0 eV shows the generation of organic fluorides on the electrode surface (Fig. 5 Row). This has been further complimented by 19F NMR results (Fig. 6) which show the presence of substituted alkyl flurides with corresponding peaks at − 71, − 72 and − 79 ppm. The peak at − 79 ppm correspond to LiTFSI, whereas the peaks at − 71 and − 72 ppm correspond to substituted alkyl fluorides, a decomposed product (solvent decomposition and fluorination). 7Li NMR shows (Figure SI 2) a sharp singlet at δ 0.17 ppm in all the cases indicates the formation of Li2CO3 resultant of the decomposition of electrolytes and absence of LixSi alloy which is further confirmed by XPS data (Fig. 4a–c). The Li/SiOx–Si–C lithium half-cell with LiFSI as additive showed an additional peak at − 123 corresponding to LiF [36, 37].

Fig. 6:
figure 6

19F NMR spectroscopy of Li/SiOx–Si–C half cell (i) without additive (ii) with LiFSI (iii) with LiBOB

Full size image

Further, an additional broad shoulder in the range of δ-70–100 indicating the presence of SiO3C (− 67 ppm) and SiSi4 (− 85 ppm) species in addition to SiO4. (XPS Li2SiO3—530.3 Li4SiO4—531.8) (Figure SI 3) [38, 39]. 29Si NMR also shows that the electrode materials cycled without additive and LiBOB showed a broad peak around δ -109 indicating the presence of SiO4 species. (XPS Li4SiO4-531.8) It is also worth mentioning that the 13C characteristic peaks (Figure SI 4) observed at δ -94,71,69,64 and 58 corresponding to 1,3 dioxalane and TEGDME indicate further no decomposition of the electrolyte solvents even after cycling.

3.2 FTIR

Cycled SiOx–Si–C electrodes were further characterized by FTIR spectroscopy, given in Figure SI 5 (i-iii). The FTIR spectrum of bare SiOx–Si–C electrode before cycling showed characteristic bands of Si–C/Si–O, Si–C, and Si–O–Si stretching at 1056, 775, 447 cm−1 respectively. Upon cycling, the electrodes without additive (containing 1,3-dioxolane + TEGDME + LiTFSI) as well as the electrodes with LiFSI and LiBOB as additives showed similar profile indicating that the decomposed products formed were similar., i.e., The silicon electrode cycled showed representative peaks of lithium alkyl carbonates, ROCO2Li at ~ 1660, ~ 1330 and ~ 850 cm−1. The peak at ~ 1090 cm−1 is attributed to the C–O–C asymmetric stretching. Li2CO3 peaks were noticed at ~ 1450 and ~ 850 cm−1. Characteristic peak of Si–O was observed at ~ 470 cm−1. Regardless of the electrolyte additives, the FTIR patterns of the electrodes with and without additives (LiFSI) after cycling were identical, demonstrating that electrolyte decomposition plays a substantial role. The LiBOB-added SiOx electrode showed slightly varied characteristics. The LiBOB added electrode displayed peaks at 1668 and 1357 cm−1 ascribed to Li oxalate species. Notably, LiBOB-added electrode showed characteristic peaks of Li2CO3 and Si–O at 1668 and 1357 cm−1 respectively which are further confirmed by NMR data (Figure SI 2). Earlier studies have shown that through the charge–discharge of LIBs, alkyl carbonate electrolytes experience a reduction process of organic and inorganic Li salts, resulting in the formation of both solid and gaseous products such as ROCO2Li, ROLi, ethylene, and others [40]. Here, the solvent molecules such as 1,3-dioxolane and TEGDME go through reduction process to produce HCOOLi, ethane and RO(CH2CH2OCH2O)Li. The lithium salts such as lithium sulphite, lithium nitride, lithium oxide, lithium sulphide, and other salts are expected to develop when electrolyte decomposition takes place.

3.3 Scanning electron microscope

Scanning electron microscopy was used to examine surface layers that were created on SiOx–Si–C electrodes of different electrolyte additives and are depicted in Fig. 7a–c. The roughness of the copper current collector is the cause of the rough surface morphology of the electrodes [41]. It is obvious from Fig. 7a, in the as-prepared electrode, the active materials Si, SiOx, and carbon particles are homogeneously spread with adequate porosity that provides better wetting property to the electrode for Li-ion transport. Additionally, the surface films created after cycling with additive-free solutions are highly thick and non-homogeneous (Fig. 7b–d) after the first, fifth, and twentieth cycles. As a result of the active components being separated from the conducting agent and binder during cycling, the morphology of the surface is substantially altered, and further agglomeration negatively impacts the conductivity of the anode [42]. On the other hand, the surface coating generated on SiOx–Si–C electrode with LiFSI additive comprising electrolyte solution (Fig. 7b) shows an unlike morphology in which the active materials are loosely packed even after 20 cycles (Fig. 7b). This is the significant influence of LiFSI on the SiOx–Si–C electrode's surface chemistry. The impact of LiFSI is projected to minimize the overall charge transfer resistance of the SiOx–Si–C electrodes, which might make it easier to generate a thin surface film with an exceptional surface chemistry for Li-ion movement by the surface films generated on active mass. LiFSI in the electrolyte not only causes a thinner surface coating to develop, but it also changes the surface chemistry of the anode. The structural deterioration of the generated SEI layer with increasing thickness, which gives higher impedance upon cycling, is expected to be a major cause of capacity fading in Li/SiOx–Si–C cells with LiBOB and electrolytes without additives.

Fig. 7
figure 7

SEM images of SiOx–Si–C electrodes before and after cycling. Row A (i) bare electrode, (ii) without additive after (ii) 1st (iii) 5th (iv) 20th cycles. Row B (ii) after (ii) 1st (iii) 5th (iv) 20th cycles with LiFSI additive. Row C (ii) after (ii) 1st (iii) 5th (iv) 20th cycles with LiBOB additive

Full size image

The additive-free electrolyte obviously exhibits some surface fractures and deterioration as illustrated in Fig. 7b, c. The shrinking of Si and SiOx particles during Li-ion extraction may be responsible for the development of fractures. The active Si and SiOx phases are isolated by the creation of a non-uniform SEI layer within the cracked surface, which may eventually result in deprived contact between the active materials and the capacity decay [23].

When PAA is employed as a binder, every Si-based anode is susceptible to cracking after 10 cycles, according to Komaba et al. [43], however, the silicon and carbon particles are retained together on the surface of Cu-current collector and hence not influencing the cycling properties. According to Chen et al. the appearance of cracks can be normally in the silicon-based electrolytes in the valleys between the hills which is due to the shrinkage during de-lithiation [23]. The majority of the contraction occurs when lithium is removed from the electrode and is perpendicular to the surface of the current collector. For the additive-free electrolyte, the cracks arise in the electrode because of the shrinkage of the two sides of the valley walls, which is in agreement with the cycling performance.

3.4 Electrochemical impedance spectroscopy studies

Electrochemical impedance spectroscopy (EIS) has always been utilized as a unique tool to investigate the different complex mechanisms of lithium-ion and lithium–sulfur batteries while electrochemical reactions. It is commonly used to describe Li–S batteries as a whole [44], the compositions of the electrolyte [45] and their impact on their performance [46]. Figure 8a–d accordingly illustrates the Nyquist spectra of Li/SiOx–Si–C half-cell with additive-free, LiFSI, LiBOB-incorporated non-aqueous electrolytes. The semicircles are clearly formed in the spectra at higher frequencies. The high- to medium-frequency semicircles and low-frequency lines, which are inclined as the frequency approaches very low values, the characteristic spectral features. Aurbach’s group demonstrated that the surface morphology of silicon-based anode materials is quite complex and that it is quite challenging to assign the semi-circles precisely [26, 41]. The surface chemistry of silicon-based is often represented by high frequency, whereas the low frequency shed light on the diffusion processes such as the diffusion of lithium ions into silicon. Because of the potential dependence of the lithium concentration in the Li–S alloys, the impedance of the electrodes may exhibit a capacitive characteristic at extremely low frequencies. The values of solution resistance, Rs, has been measured as 5.80, 18.08 and 6.77 ohms respectively for the Li/SiOx–Si–C half-cell with additive-free, LiFSI, LiBOB-added non-aqueous liquid electrolytes before cycling. Upon 1st cycle the values of Rs has been increased to 15 ohms for the additive free electrolyte and reduced to 7.24 and 3.86 ohms respectively for the LiFSI and LiBOB-added electrolytes. Apparently, from Fig. 8d that the Li/SiOx–Si–C cell with LiFSI as additive offered lowest Rs (11 ohms) and Rct (13 ohms) which is diminished when compared additive-free and LiBOB-added electrolytes. Cycling causes a noticeable gradient drop, which points to lithium ion diffusion inside the active material [42]. The homogenous lithiation of the SiOx–Si–C electrode in the LiFSI-added electrolytes is responsible for the reduction in the impedance, and these findings are consistent with cycling measurements where improved charge–discharge performance was attained. The production of more resistant surface films, which slows the Li–Si alloying kinetics and consequently resulted in incomplete lithiation, is what origins the obtained poor cycling performance of Li/SiOx–Si–C cells with additive-free electrolyte.

Fig. 8
figure 8

EIS of Li/SiOx–Si–C half cell a before cycling b after 1st cycle (c) after 5th cycle (d) after 20th cycle

Full size image

4 Conclusions

In summary, the electrochemical and the surface characteristics of SiOx–Si–C composite electrodes cycled with LiFSI and LiBOB as electrolyte additives were explored. In general, hundreds of cycles are feasible with lithiated silicon or SiOx electrodes. Among the two additives examined, LiFSI-added electrolytes exhibited better charge–discharge properties. Even after 50 cycles, the SiOx–Si–C electrode cycled with LiFSI-added electrolyte had a homogeneous and crack-free surface morphology. The establishment of a stable SEI by LiFSI-added electrolytes was related to the improved charge–discharge studies of the Li/SiOx–Si–C cell, which was further supported by electrochemical impedance spectroscopy studies. The development of reductive products such as Li2CO3, ROCO2Li have was confirmed by XPS and FT-IR analyses and was further substantiated by solid NMR studies.