Bi₂Se₃ Nanostructured Thin Films as Perspective Anodes for Aqueous Rechargeable Lithium-Ion Batteries

Abstract

In recent years, aqueous rechargeable lithium-ion batteries (ARLIBs) have attracted attention as an alternative technology for electrical storage. One of the perspective battery anode materials for application in ARLIBs is Bi₂Se₃, which has already shown good perspectives in the application of conventional lithium-ion batteries (LIBs) that use organic electrolytes. In this study, the electrochemical properties of Bi₂Se₃ thin films with two different layers on the electrode surface—the solid electrolyte interphase (SEI) and the Bi₂O₃ layer—were investigated. The results of this work show that the formation of the SEI layer on the surface of Bi₂Se₃ thin films ensures high diffusivity of Li⁺, high electrochemical stability, and high capacity up to 100 cycles, demonstrating the perspectives of Bi₂Se₃ as anode material for ARLIBs.

1. Introduction

LIBs have dominated the field of battery technology in the last few decades due to their remarkable advantages such as high energy density (>200 Wh kg⁻¹), high Coulombic efficiency (>95%), and long cycle life (3000 cycles at the deep discharge of 20%) [1,2,3]. The disadvantages of lithium-ion batteries include the flammability of non-aqueous solvents (e.g., ethylene carbonate, dimethyl carbonate, etc.), which may lead to accidents, and the high costs of fabrication (the average LIB cell cost in 2021 was EUR 104/kWh) [4]. Alternatively, in 1994, Dahn’s group proposed aqueous rechargeable lithium-ion batteries (ARLIBs) consisting of LiMn₂O₄/VO₂ in the 5 M LiNO₃ electrolyte [5]. This battery demonstrated an average voltage of 1.5 V with an energy density of 75 Wh kg⁻¹, which was higher than the energy density of Pb-acid batteries (30 Wh kg⁻¹) [5].

For a long time, investigations on the development of aqua-based batteries were suspended due to several serious limitations—poor cycling performance, low voltage potential window (1.23 V), and decomposition of the water, which made these batteries commercially unviable [6]. In the last 5 years, the requirement for cheap, safe, and environmentally friendly energy storage devices has generated a great deal of interest. ARLIBs have been shown to be one of the potential candidates for the most suitable battery systems to meet these requirements. Thus, the interest in research into this battery technology has only increased [7]. Such a rapid development might allow for the production of ARLIBs with properties that are competitive in the worldwide market [8].

While significant progress has been achieved in the development of cathode materials (e.g., enhancement of mechanical stability), the research on anode materials has been limited [9]. Among a variety of material candidates for the ARLIBs anodes (e.g., Li_1.2V₃O₈, VO₂, TiP₂O₇) [10,11,12], the TiP₂O₇ is considered to be the most promising [10,13]; however, it suffers significant capacity fading during charge/discharge processes, which might be related to the deterioration of the anode material [11].

Another possible perspective anode material for ARLIBs comes from bismuth chalcogenides Bi₂X₃ (X = S, Te, Se), which are unique lamellar materials with high ionic conductivity that have shown great performance in organic electrolyte LIBs [12]. Among the chalcogenides, Bi₂Se₃ has the highest density and theoretical volumetric capacity (3667 mAh cm⁻³) [14], which, being coupled with its high electrical conductivity (10.6 S cm⁻¹ at 300 K) [15], allows for the possibility of fabricating compact electrical devices [14,16].

The crystal lattice of Bi₂Se₃ consists of a sequential layer arrangement that is perpendicular to the trigonal C-axis [17]. Each molecule consists of five atomic planes arranged in the following pattern: Se¹-Bi-Se²-Bi-Se¹, where van der Waals gaps are located between Se¹ atoms [18]. The thickness of such a quintuple layer is about 1 nm. The arrangement of quintuple layers provides enough space for Li⁺ during intercalation and deintercalation processes, which makes them suitable anode materials for LIBs [19].

While both Bi₂Se₃ and TiP₂O₇ have higher potential for Li⁺ intercalation than for H₂ evolution, which allows for avoiding unwanted side reactions [20], Bi₂Se₃ is still considered a more unique material due to its quintuple layer structure. Bi₂Se₃ has a ~4 times higher theoretical gravimetric capacity than TiP₂O₇ (491 mAh g⁻¹ [12] vs. 121 mAh g⁻¹ [21]).

Additionally, in anode material research, special attention should be paid to the formation of the solid-electrolyte interphase (SEI) layer, which impacts the safety and stability of the anode electrode [22]. In aqueous electrolytes, the SEI layer consists of two layers: Li₂O (inner layer) and Li₂CO₃ (outer layer), which are formed by Li⁺ reactions with dissolved O₂ and CO₂ [23,24,25,26,27]. For the ARLIBs, it is necessary to use concentrated aqueous electrolytes, which will ensure the formation of stable SEI layers as, in the diluted electrolytes, the SEI layer quickly undergoes hydrolysis and dissolves [25,27].

In this work, the electrochemical properties of Bi₂Se₃ thin films are investigated for the first time to evaluate their possible application as anode electrodes in ARLIBs. To examine the formation of the SEI layer and its effect on the performance of the Bi₂Se₃ anode, measurements were carried out in two different potential ranges. One of the potential ranges ensured the formation of the SEI layer on the surface of Bi₂Se₃, and the other did not. 5 M LiNO₃ solution, which is electrochemically more stable than LiCl and Li₂SO₄ [28] and can be prepared in high concentrations (up to 9 M) [6], was used as an electrolyte.

2. Material and methods

2.1. Synthesis and Characterization of Bi₂Se₃

Bi₂Se₃ thin films were synthesized on microscope glass slides (25 × 75 mm) using the physical vapor deposition method [29,30,31]. The synthesis procedure was performed in a programmable quartz tube (length = 60 cm, diameter = 4.5 cm) furnace (GSL-1100x-S, MTI Corp.) with adjustable valves on both sides.

In total, 15 mg of raw Bi₂Se₃ (Sigma-Aldrich, Burlington, MA, USA, 99.999%) was weighted in ceramic boats using analytical scales (KERN “ABP 200-5DM”, ±0.01 mg, Balingen, Germany). Inside the quartz tube at certain temperature regions, microscope glass slides (330–380 °C) and raw Bi₂Se₃ (585 °C) were inserted. The quartz tube was connected to the N₂ gas cylinder (Linde, Berlin, Germany, 99.999%) and a vacuum pump. Before the synthesis, the quartz tube was purged with N₂ and vacuumed. The synthesis of Bi₂Se₃ thin films was performed according to the following protocol: first, linear heating of the quartz tube for 45 min up to a temperature of 585 °C; second, maintaining a constant temperature for 15 min (pressure 2–3 Torr); and third, gradual cooling down to 470 °C, reinjecting N₂ gas into the tube up to atmospheric pressure, and cooling down to room temperature.

The morphology and stoichiometry of the samples were investigated with a scanning electron microscope (SEM) (Hitachi FE-SEM S-4800, Marunouchi, Tokyo, Japan) equipped with an energy-dispersive X-ray (EDX) (Bruker XFLASH 5010, Billerica, MA, USA) detector, X-ray diffraction analysis (XRD) (Bruker D8 Discover, Billerica, MA, USA), and X-ray photoelectron spectroscopy (XPS) (ThermoFisher Escalab 250Xi⁺, Walthman, MA, USA) with monochromatic Al-K_α X-ray source).

2.2. Electrochemical Measurements

The electrochemical measurements were performed at room temperature using a 3-electrode cell with 3 M Ag/AgCl as a reference electrode (RE), Pt wire as a counter electrode (CE)—(ItalSens, PalmSens, Houten, The Netherlands), and Bi₂Se₃ thin film on the glass substrate as a working electrode (WE)—(Supplementary Figure S1). 5 M LiNO₃ (ThermoFisher, Walthman, MA, USA, analytical grade 99%) was used as an electrolyte. To ensure appropriate electrical contact, copper wire was fixed to the surface of the working electrode with silver conductive paint (Electrolube, Ashby-de-la-Zouch, UK). Cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), and galvanostatic charge/discharge measurements were performed by potentiostat PalmSens4. CV curves were measured at different scan rates (0.1–1.0 mV s⁻¹). The galvanostatic charge/discharge measurements were carried out at room temperature with a C-rate of 1 C. EIS spectra were measured in the frequency range of 0.01–10000 Hz at open circuit potential before and between the galvanostatic charge/discharge cycles (after the 1st, 2nd, 5th, 50th, and 100th cycle). After CV measurements, the Bi₂Se₃ samples were washed ~4–5 times with deionized water, dried, and investigated using the SEM, EDX, and XPS analysis methods.

3. Results and Discussion

3.1. Characterization of Synthesized Bi₂Se₃ Thin Film

To increase the effective interaction surface between the electrode and the electrolyte, the synthesis method for Bi₂Se₃ thin films with nanoplates that grow at different angles (partly disordered) to the substrate surface was selected [32,33] (Figure 1a). This type of Bi₂Se₃ morphology has no dead volume or gap spacing shrinkage [34] which could lead to the decrease of the anode active sites, thus making it a promising candidate as anode material for ARLIBs. The size of the nanoplates varied between 1.0 and 8.0 μm. The thickness of the synthesized Bi₂Se₃ thin films with partly disordered nanoplates was between 350 and 500 nm. The EDX spectra (Supplementary Figure S2) showed intensive peaks of C, O, Se, Si, and Bi. Si, O, and C peaks were the background signals from the microscope glass slides on which Bi₂Se₃ was synthesized. The Se/Bi atomic ratio determined from the EDX spectra was 1.44 ± 0.03, which is close to the theoretical stoichiometric ratio of Bi₂Se₃ (1.50) and indicates a rather uniform chemical composition of the films. The XRD pattern (Figure 1b) of Bi₂Se₃ powder obtained by scratching the synthesized thin film off the substrate showed Bi₂Se₃ diffraction peaks, which refer to the rhombohedral (R-3 m) crystal system of Bi₂Se₃ (Ref. card No. PDF 01-085-9274): a = b = 4.13850 Å; c = 28.62400 Å, as well as cubic γ-Bi₂O₃ (I23) (Ref. card. No. PDF 01-074-1375) with crystal system a = b = c = 10.08000 Å. It should be noted that the difference in the peak intensity ratios between the reference cards and the actual XRD patterns obtained for the powdered Bi₂Se₃ thin films most likely are related to the impact of the substrate on the growth of Bi₂Se₃ thin films, resulting in the domination of certain crystallographic growth directions [32,33]. The formation of Bi₂O₃ could be explained by the oxidation of Bi₂Se₃ powder, which, due to its increased surface area after scratching, is more prone to oxidation in an ambient environment.

3.2. Electrochemical processes and mechanisms of Bi₂Se₃ thin film

Cyclic voltammetry of Bi₂Se₃ thin films in 5 M LiNO₃ at the scan rate of 0.25 mV s⁻¹ in the potential range (−1.0 ÷ 1.3 V vs. Ag/AgCl) showed two cathodic peaks (I, II) and four anodic peaks (III, IV, V, VI) (Figure 2a). The peaks I (−0.65 V vs. Ag/AgCl) and III (0.02 V vs. Ag/AgCl) represent the intercalation/deintercalation processes of Li⁺ according to the reaction in Equation (1) [19]. The peaks II (−0.71 V vs. Ag/AgCl) and IV (−0.65 V vs. Ag/AgCl) are related to the substitution reaction of Bi₂Se₃ and Li₂Se (Equation (2)) [18,35]. Anodic peak V (0.61 V vs. Ag/AgCl) might describe a formation of NO₂⁻ from NO₃⁻ (Equation (3)) [36]. NO₂⁻ can be rapidly oxidized back to NO₃⁻ due to the presence of dissolved O₂ in the electrolyte [37]. The peak VI (0.84 V vs. Ag/AgCl) appeared only in the first cycle and was attributed to the formation of the SEI layer, which, according to the literature on the other aqueous Li⁺ systems (Mo₆S₈, carbon-coated TiO₂, and carbon electrodes), is most probably composed of Li₂O and Li₂CO₃ (Equations (4) and (5)) [24,25,26,38,39].

To avoid the formation of NO₂⁻ in the solution (at 0.61 V vs. Ag/AgCl) and to investigate the impact of the SEI layer on the stability and operation of the Bi₂Se₃ electrode in the aqueous electrolyte, parallel experiments were performed in the narrower potential range (−1.0 ÷ 0.5 V vs. Ag/AgCl) (Figure 2b). In this case, a new low-intensity anodic peak VII (−0.05 V vs. Ag/AgCl) appeared starting from the second cycle. This peak might be related to the formation of the Bi₂O₃ layer on the surface of Bi₂Se₃ by the reaction with dissolved oxygen in the electrolyte (Equation (6)) [40,41]. For the wide potential range (−1.0 ÷ 1.3 V vs. Ag/AgCl), this peak was not observed, which indicates that, after the formation of the SEI layer on the surface of Bi₂Se₃, further formation of Bi₂O₃ was inhibited. Further in the text, both potential ranges are referred to as “samples with the SEI layer” (−1.0 ÷ 1.3 V vs. Ag/AgCl) and “samples with the Bi₂O₃ layer” (−1.0 ÷ 0.5 V vs. Ag/AgCl).

It is important to note that the CV peak potential values obtained in this work may differ from the values reported in the literature by 0.6–1.0 V in terms of the standard hydrogen potential scale. This is due to the fact that, in this work, the 3-electrode cell configuration was used, which allowed us to measure the properties of the anode material solely, while in the literature, the 2-electrode cell configuration was applied, thus measuring the properties of the whole cell.

Bi₂Se₃ + xLi⁺ + xe⁻ ↔ Li⁺_x[Bi₂Se₃]^x−,

(1)

Bi₂Se₃ + 6Li⁺ + 6e⁻ ↔ 3Li₂Se + 2Bi,

(2)

NO₃⁻ + 2H⁺ +2e⁻ → NO₂⁻ + H₂O,

(3)

O₂ + 4Li⁺ + 4e⁻ → 2Li₂O,

(4)

2Li⁺ + ½O₂ + CO₂ + 2e⁻ → Li₂CO₃,

(5)

4Bi³⁺ + 3O₂ + 6e⁻ → 2Bi₂O₃.

(6)

For both samples with the formed SEI layer and with the Bi₂O₃ layer, the intensities of peaks corresponding to Li⁺ intercalation/deintercalation processes (I, III) in subsequent cycles increased (Figure 2a,b). The increase in the intensity of Li⁺ intercalation/deintercalation processes may be related to the pre-treatment of Bi₂Se₃. The pre-treatment involves the stabilization of the SEI (−1.0 ÷ 1.3 V vs. Ag/AgCl) and the Bi₂O₃ (−1.0 ÷ 0.5 V vs. Ag/AgCl) layers until their continuous coverage of the entire electrode’s surface.

Charge/discharge curves obtained for the Bi₂Se₃ thin films in the potential range (−1.0 ÷ 1.3 V vs. Ag/AgCl) at the C-rate 1C (Figure 3a) exhibited, in the first cycle, a plateau near 0.6 V. This confirmed the formation of the SEI layer. In the next cycles, this plateau was not observed, which confirms that the layer formation process was finished. These findings are in line with the data reported by the other groups, pointing out that the fully developed SEI layer, which is stable and ensures long-term capacity retention and high Coulombic efficiency, forms in the first cycle due to an irreversible electrochemical reaction [42]. In the subsequent cycles, charge (−0.55 V) and discharge (−0.02 V) plateaus, corresponding to Li⁺ intercalation and deintercalation, respectively, were observed (Figure 3a). Although the SEI layer was formed in the first cycle, the charge/discharge profiles were changing until the third cycle (samples with the SEI layer), indicating that, during the first three cycles, the surface pre-treatment took place. After the pre-treatment of the Bi₂Se₃, the charge/discharge profiles did not change significantly (Figure 3a).

Additionally, the voltammograms (Figure 2a) showed that after the formation of the SEI layer in the first cycle, the peaks corresponding to the processes of Li⁺ intercalation/deintercalation slightly shifted towards the negative potential, which might be related to the structural and textural modification of Bi₂Se₃ [43]. A similar tendency was also observed for the anodic peak (V) associated with NO₂⁻ formation, where, after the first cycle, the peak had shifted from 0.67 V to 0.61 V vs. Ag/AgCl (Figure 2a).

For the samples with the Bi₂O₃ layer, the plateau corresponding to the intercalation process was observed already in the first cycle, while the plateau corresponding to the deintercalation process was observed starting with the second cycle (Figure 3b), as there were no Li⁺ ions intercalated into Bi₂Se₃ at the beginning of the first anodic (discharge) scan from −1 V to 6 + 0.5 V vs. Ag/AgCl. The change in the charge/discharge profiles was observed up to the ninth cycle (Figure 3b), indicating slower pre-treatment of the Bi₂Se₃ surface in comparison to the samples with the SEI layer, as well as the formation of the Bi₂O₃ layer.

After pre-treatment, the formed SEI or Bi₂O₃ layers on the surfaces of the Bi₂Se₃ electrodes inhibited further substitution reaction (Equation (2)), which is characterized by decreased peak intensities (II, IV).

The comparison of the heights of the peaks corresponding to Li⁺ intercalation/deintercalation (I, III) showed that they were up to 10 times higher for the samples with formed SEI layers than for the samples with formed Bi₂O₃ layers, which could have been related to the higher transportation rate of Li⁺ in the samples with the SEI layer. This may mean that, while Bi₂O₃ has been shown as a suitable additive in Bi₂Se₃ water-based battery systems [44], its performance as an “SEI” layer is not optimal as its high density inhibits ion transport.

SEM images and XRD patterns acquired from the samples’ surface XPS spectra demonstrated the structural differences of the Bi₂Se₃ thin film surface before and after 5 cycles and 10 cycles for the samples covered with continuous SEI and Bi₂O₃ layers, respectively (Figure 4 and Figure 5).

According to the observation by SEM, the surface of the sample with the SEI layer after five cycles was coated with amorphous material (Figure 4a). The XPS data (Figure 5a) showed a noticeable Li 1s peak, which may indicate the presence of Li₂O and Li₂CO₃. The water-soluble Li₂O could have been preserved if the Li₂CO₃ had grown on top of it and had shielded it from the electrolyte (Equations (4) and (5)). While the XRD pattern of the sample with the SEI layer demonstrated pure rhombohedral Bi₂Se₃ (R-3 m) without the initial cubic y-Bi₂O₃ signal (Figure 4c), the XPS data still showed a Bi₂O₃ signal (Figure 5b). This indicates that the initial Bi₂O₃ has remained, but further growth of the Bi₂O₃ was inhibited by the formation of the SEI layer.

In turn, the sample with the Bi₂O₃ layer after 10 cycles had a smooth surface (Figure 4b), which could mean that the Bi₂Se₃ thin film was covered with the thin and dense Bi₂O₃ layer. This assumption is supported by both XRD and XPS data where a small cubic γ-Bi₂O₃ (I23) peak (Figure 4c) and very pronounced peaks related to the oxidized Bi (Bi 4f, Figure 5d) were detected. It should be noted that only a very small peak of Bi₂Se₃ is observed in the surface layer of the sample (Figure 5c), which confirms that it is mainly composed of Bi₂O₃ and proves the formation of continuous dense layers of Bi₂O₃. In addition, the representing SEI layer Li 1s peak is negligible for the samples coated with the Bi₂O₃ layer (Figure 5c), which proves the absence of SEI layers composed of Li₂O and Li₂CO₃ on the surface of these samples.

These data support the claim that, during cycling in the potential window (−1.0 ÷ 0.5 V), the dense Bi₂O₃ layer is formed, and this layer may inhibit Li⁺ intercalation/deintercalation processes, which results in the decrease of the heights of the CV peaks for the samples coated with the Bi₂O₃ layer (Figure 2).

The stability of Bi₂Se₃ thin films in terms of the chemical composition of Bi and Se was checked with EDX before cycling and after the fifth (sample with the SEI layer) and ninth (sample with the Bi₂O₃ layer) cycles. While it is well known that Bi₂Se₃ tends to lose selenium during the cycling [15], the results of this study demonstrate that the dissolution of Se for both samples covered with SEI and Bi₂O₃ layers was negligible (

Table 1. The atomic composition of Bi₂Se₃ thin films in 5 M LiNO₃ before and after CV measurements.

Element	Before	After 5 Cycles (SEI Layer)	After 10 Cycles (Bi2O3 Layer)
Bi	41 ± 1	46 ± 3	41 ± 1
Se	59 ± 1	54 ± 3	59 ± 1

), which indicates the good stability of the Bi₂Se₃ electrode in the LiNO₃ for the first 5 (SEI layer) and 10 (Bi₂O₃ layer) cycles.

Charge/discharge profiles up to 100 cycles for the samples with the SEI and Bi₂O₃ layers demonstrated Li⁺ intercalation/deintercalation plateaus from the 5th to 25th cycle (Figure 3c,d). After the 25th cycle, these plateaus were not observable anymore, which might be due to the fast process of Li⁺ intercalation/deintercalation [45].

To investigate the mechanisms of Li⁺ intercalation/deintercalation processes in the Bi₂Se₃ thin film, the CV curves were obtained at different scan rates (0.1–1.0 mV s⁻¹) after the surface pre-treatment was finished (the third cycle for the samples with SEI layer (Figure 6a) and ninth cycle for the samples with Bi₂O₃ (Figure 6b)).

The dominant stage of Li⁺ intercalation/deintercalation processes (diffusion-controlled or capacitive process) was determined by Equation (7) [46]:

$$I=a{v}^b,$$

(7)

where I is the current of the Li⁺ intercalation/deintercalation process (mA), v is the scan rate (mV s⁻¹), and a, b are the adjustable fitting values.

Calculated from the linear regression slopes (Figure 7a,b), the b-values for both samples with SEI and Bi₂O₃ layers varied from 1.05 to 1.28, indicating the dominant contribution of the capacitive process (pseudo-capacitance and electrical double-layer capacitance) to the Li⁺ intercalation/deintercalation in both potential ranges. The quantitative contribution of the capacitive and diffusion-based processes in Li⁺ intercalation/deintercalation at different scan rates (Figure 7c,d) was determined by Equation (8):

$$i\left(V\right)=k_{1}v+k_2{v}^{\frac{1}{2}},$$

(8)

where i(V) is the current of the Li⁺ intercalation/deintercalation process, k₁v is the contribution of the capacitive process, and k₂v^1/2 is the contribution of the diffusion-controlled process [47,48].

At the scan rate of 0.1 mV s⁻¹, the contribution of the capacitive process to Li⁺ intercalation/deintercalation was similar (51.4% and 53.9%) for the samples with SEI and Bi₂O₃ layers, respectively (Figure 7c,d). With the increasing scan rate, the contribution of the capacitive process increased, and at the scan rate of 1.0 mV s⁻¹, it reached 78.3% for the samples with the SEI layer and 75.4% for the samples with the Bi₂O₃ layer. The higher contribution of the capacitive processes to the Li⁺ intercalation/deintercalation process in the scan rate range from 0.25 to 1.00 mV s⁻¹ indicates the removal of limitations on charge transfer processes for both SEI and Bi₂O₃ layers.

As the capacitive processes of Li⁺ intercalation/deintercalation were similar for the samples with the SEI and Bi₂O₃ layers, and thus could not explain the differences between them, the diffusion coefficients for Li⁺ intercalation/deintercalation were determined. The diffusion coefficients of Li⁺ intercalation/deintercalation were calculated from the slope fitted values (Figure 8) using Equation (9) [49]:

$$D={\left(\frac{slope}{0.446nFAC}\right)}^2\left(\frac{RT}{nF}\right),$$

(9)

where D is the diffusion coefficient (cm² s⁻¹), slope is the slope value, n is the number of electrons, F is the Faraday constant (C mol⁻¹), A is the surface area (cm²), C is the electrolyte concentration (mol cm⁻³), R is the universal gas constant (J K⁻¹ mol⁻¹), and T is the temperature (K).

Calculated diffusion coefficient values (

Table 2. Comparison of Li⁺ intercalation and deintercalation diffusion coefficients (cm² s⁻¹) for different anodes and electrolytes.

Electrode	Electrolyte	Intercalation	Deintercalation	Reference
Bi2Se3 thin film (SEI layer)	5 M LiNO3	3.3 × 10−12	2.2 × 10−12	This study
Bi2Se3 thin film (Bi2O3 layer)	5 M LiNO3	4.3 × 10−13	4.5 × 10−13	This study
Li0.3V2O5	5 M LiNO3	10−11–10−12		[50]
TiP2O7	1 M Li2SO4	3.80 × 10−15	1.77 × 10−15	[51]
TiNb6O17	1 M LiPF6 EC/DMC b (1:2)	3.43 × 10−13	3.72 × 10−13	[52]
TiNb2O7	1 M LiPF6 EC/DMC b (1:2)	1.08 × 10−14	3.01 × 10−14

^a ethylene carbonate/diethyl carbonate. ^b ethylene carbonate/dimethyl carbonate.

) show that, in the first approximation, Li⁺ diffusion for samples with the SEI layer was higher than for the samples with the Bi₂O₃ layer by an order of magnitude.

Thus, it can be assumed that the formation of less dense Li₂O and Li₂CO₃ SEI layers ensures the faster transportation of Li⁺ in comparison to the Bi₂O₃ layer and can be the contributing factor to the difference in the peak intensities in CV curves for both samples (Figure 2).

Estimations for the Bi₂Se₃ thin films with SEI layer Li⁺ diffusion coefficients were comparable with the values reported for the Li_0.3V₂O₅ in Li⁺ aqueous electrolyte while exceeding the diffusion coefficients of TiP₂O₇ (1 M Li₂SO₄), as well as the diffusion coefficients reported for other anode materials—TiNb₆O₁₇ and TiNb₂O₇ tested in lithium non-aqueous electrolytes (Table 2). These results demonstrate that Li⁺ diffusion coefficients for Bi₂Se₃ anodes in the 5 M LiNO₃ electrolyte are sufficient for their perspective application in ARLIBs systems.

To investigate the changes in the electrochemical properties of the Bi₂Se₃ thin film electrode over 100 cycles for the samples with formed SEI (Figure 9a) and Bi₂O₃ (Figure 9b) layers, EIS measurements were carried out. The obtained impedance hodographs were described using the standard equivalent circuit schemes, which represent the electrochemical properties at the electrolyte/electrode interface. The schemes were complicated by the presence of a constant phase element (CPE₁), which, in addition to the capacity of the double layer, characterized the inhomogeneity of the surface of the working electrode. The Warburg element indicated the diffusion of the reacting ions and/or molecules to the electrode surface. R_el values demonstrated not only the resistance of the electrolyte but also the penetrability of the electrolyte within the electrode [53] (Figure 9c). During the cycling, the SEI layer for the first sample and the Bi₂O₃ layer for the second sample were modeled as a capacitor formed by a CPE (characterizing the heterogeneity and the double-layer capacitance of the layer), with a parallel resistance for the charge transfer reaction (Figure 9d).

The resistance values of the electrochemical processes (

Table 3. Cycle-dependent fitted resistance values of Bi₂Se₃ thin films with SEI and with the Bi₂O₃ layer.

Layer	Resistance	Before Cycling	1st	5th	10th	25th	50th	100th
SEI	Rel, kΩ cm2	0.5	0.9	2.3	2.8	2.9	3.0	3.1
	Rlayer, kΩ cm2	-	2162	1515	772	252	208	225
	Rct, kΩ cm2	584	154	379	275	1246	1965	490
Bi2O3	Rel, kΩ cm2	0.1	0.4	0.8	1.3	1.4	1.7	2.0
	Rlayer, kΩ cm2	-	112	130	170	196	39	26
	Rct, kΩ cm2	111	13	10	17	25	17	20

) were calculated using the Levenberg–Marquardt algorithm [54].

The values of R_el remained almost unchanged during the cycling, indicating good stability of the electrolyte over the whole 100 cycles for both samples. The R_el values were 2.8 ± 0.3 kΩ cm² and 1.6 ± 0.3 kΩ cm² for the samples with the SEI layer and the Bi₂O₃ layer, respectively (Table 3). However, in the first five cycles, the R_el values for both cases were significantly lower, which might be related to the pre-treatment processes of the Bi₂Se₃ thin films, including the stabilization of the surface layer which was also confirmed by the CV curves (Figure 2). For the sample with the SEI layer, R_el was ~2 times higher than for the sample with the Bi₂O₃ layer, which might indicate that the processes occurring with the participation of the electrolyte are more complicated in the first case.

The semicircle at the medium-frequency range (Figure 9a,b) represents the charge-transfer resistance R_ct of Bi₂Se₃ thin films. Starting from the first cycle, R_ct values decreased ~4 times and ~9 times for the samples with SEI and Bi₂O₃ layers, respectively, which may be related to the beginning of textural and structural changes of the Bi₂Se₃ surface. After the 2nd cycle, the R_ct for the sample with the Bi₂O₃ layer was almost constant (17 ± 5 kΩ cm²) until the 100th cycle, while, for the sample with the SEI layer, these values increased (from 154 kΩ cm² to 1965 kΩ cm²). This might indicate the continuation of the textural and structural changes of the Bi₂Se₃ thin films with the formed SEI layer. The ~4 times decrease of R_ct after the 100th cycle most likely corresponds to the electrochemical and mechanical degradation of the electrode.

For the samples with the SEI layer, the R_ct was ~10–50 times higher than for the samples with the Bi₂O₃ layer, which might be related to the different structures of the SEI and Bi₂O₃ layers.

From the 1st to the 100th cycle, a small semicircle corresponding to R_layer was observed in the low-frequency range (Figure 9a,b). For the sample with the formed SEI layer, R_layer gradually decreased ~10 times from the 1st to the 100th cycle, probably indicating the mechanical and electrochemical degradation of the SEI layer by becoming more porous with cracks and cavities on the electrode surface [55,56].

For the samples with the Bi₂O₃ layer, the R_layer gradually increased from the 1st to the 25th cycle. This represents the gradual formation of the Bi₂O₃ layer, including the pre-treatment processes (stabilization of the Bi₂O₃ layer) till the 9th cycle (Figure 2) and possible increase of the thickness of the Bi₂O₃ layer between the 10th and 25th cycles. After the 25th cycle, the R_layer of the sample with the Bi₂O₃ layer started to gradually decrease from 196 kΩ cm² (25th cycle) to 39 kΩ cm² (50th cycle), which might be related to the degradation (formation of mechanical cracks and cavities) of the Bi₂O₃ layer. The comparison between the SEI and the Bi₂O₃ layers showed that R_layer was ~5–20 times higher for the sample with the SEI layer. This probably indicates a higher affinity of energy for an electron of the Bi₂O₃ layer in comparison with the SEI layer.

3.3. Electrochemical Performance of Bi₂Se₃ Thin Films

To investigate the possibility of using Bi₂Se₃ thin films as anodes in the ARLIBs systems, the charge/discharge capacity changes during the cycling were evaluated. After the first cycle, the initial charge and discharge capacities of the Bi₂Se₃ thin film in the samples with the SEI layer were 985 mAh g⁻¹ and 404 mAh g⁻¹ (Figure 10a). For the samples with the Bi₂O₃ layer, these capacities were lower at 250 mAh g⁻¹ (charge) and 115 mAh g⁻¹ (discharge), confirming that the formation of Bi₂O₃ makes the electrode less efficient in comparison with the samples covered by the SEI layer consisting of Li₂O and Li₂CO₃ (Figure 9c). During the first 30 cycles, the charge/discharge capacities of the samples with the SEI layer decreased by ~3 times, which might be associated with the gradual degradation of the SEI layer, as Li₂O and Li₂CO₃ are soluble in aqueous electrolyte compounds. Re-formation of the SEI layer might become more difficult due to the significant consumption of the dissolved O₂ and CO₂ during the first cycle of the SEI layer formation. For the sample with a Bi₂O₃ layer, the capacity increase (till the 10th cycle) and decrease (till the 30th cycle) were observed (Figure 10c), possibly indicating the gradual formation and degradation (the formation of mechanical cracks and cavities) of the Bi₂O₃ layer. Such tendencies of the re-formation of the SEI and Bi₂O₃ layers were also confirmed by the increment and decrement of the R_layer values (Table 3). With the further cycling, the discharge capacities of the samples with the SEI layer also continued to decline, but not as rapidly (~2 times) as in the previous 30 cycles, from 151 mAh g⁻¹ (30th cycle) to 79 mAh g⁻¹ (100th cycle). For the samples with the Bi₂O₃ layer, the values of both charge and discharge capacities remained constant after the 30th cycle and up to the 100th cycle. The initial Coulombic efficiency for the samples with the SEI layer was 41%. However, after cycling up to the 30th cycle, it increased up to 67% and remained almost constant up to the 100th cycle (73%) (Figure 10b). For the samples with the Bi₂O₃ layer, the Coulombic efficiency was almost constant until the 97th cycle (~40%) and decreased down to 33% during the next three cycles (Figure 10d).

The comparison of the discharge capacities of the Bi₂Se₃ thin films with different previously reported anodes (TiO₂, TiP₂O₇, L_0.3V₂O₅, L_1.2V₃O₈) in the lithium aqueous electrolytes studied in this work (

Table 4. Comparison of galvanostatic discharge cycling performances of different anodes in lithium aqueous electrolyte.

Electrode	Electrolyte	C-rate	Initial Specific Capacity	Cycles	Specific Capacity	Capacity Retention	Coulombic Efficiency	Reference
Bi2Se3 thin film (SEI layer)	5 M LiNO3	1 C	404 mAh g−1	30	151 mAh g−1	40%	70%	This study
Bi2Se3 thin film (Bi2O3 layer)	5 M LiNO3	1 C	115 mAh g−1	30	96 mAh g−1	83%	43%	This study
TiO2	21 M LiTFSI	0.2 C	115 mAh g−1	40	8 mAh g−1	7%	70%	[24]
TiP2O7	5 M LiNO3	0.1 C	42 mAh g−1	25	15 mAh g−1	35%	-	[11]
Li0.3V2O5	5 M LiNO3	1 C	75 mAh g−1	50	38 mAh g−1	51%	99%	[50]
Li1.2V3O8	1 M Li2SO4	1 C	101 mAh g−1	30	70 mAh g−1	70%	-	[9]

) show that the Bi₂Se₃ thin films with SEI and Bi₂O₃ layers demonstrate the highest specific capacities at the fastest charge/discharge rate (1 C), indicating their perspective application as anode electrodes in ARLIBs.

4. Conclusions

Bi₂Se₃ thin films were synthesized and investigated for their application in ARLIBs in aqueous the 5 M LiNO₃ electrolyte in two potential ranges which enabled two different interface layers, (−1.0 V ÷ 1.3 V—for Li₂O and Li₂CO₃ SEI) and (−1.0 V ÷ 0.5 V—for Bi₂O₃). With the formed SEI layer, Bi₂Se₃ thin films demonstrated rapid pre-treatment (formation of stable SEI layer), high reversibility, and high transportation of Li⁺ during the intercalation/deintercalation processes. For both SEI and Bi₂O₃ layers, the capacitive process dominated (>50%) Li⁺ intercalation/deintercalation, which indicates the removal of limitations on charge transfer processes. In addition, the SEI layer enabled an Li⁺ diffusion coefficient an order of magnitude higher than the Bi₂O₃ layer, which was in line or higher than that reported in the literature. The EIS investigation showed relatively stable and similar electrochemical performances for both the SEI and Bi₂O₃ layers up to the 100th cycle for electrolyte (R_el), while charge transfer (R_ct) and layer (R_layer) resistances changed during the cycling and were significantly lower for the Bi₂O₃. This may be attributed to the different structures and electron affinities of the SEI and the Bi₂O₃ layers.

The galvanostatic charge/discharge curves obtained at an intensive rate of 1 C for 100 cycles demonstrated that the formation of the SEI layer in the first cycle plays a crucial role in the electrode stabilization, which in turn, ensures high Coulombic efficiency (73%), high charge/discharge capacity, and protection from the mechanical and electrochemical degradation up to 100 cycles. Overall, the Bi₂Se₃ thin films with a formed SEI layer demonstrated better electrochemical performance in comparison to the ones with a formed Bi₂O₃ layer. These results demonstrate that the Bi₂Se₃ thin films can be promising candidates for use as anode materials for ARLIBs; however, further research is needed to optimize the electrochemical properties (e.g., capacity fading) of the system.