1. Introduction
The improvement of small, low-power devices (biosensors [
1], smart watches, radio-frequency identification RFID tags, Internet of Things, etc., with power requirements below 10 mW [
2]) can be achieved by the development of power sources to provide autonomous operation. Lithium-ion batteries (LIBs), owing to their high energy density, cycle-life, and operational temperature range, are widely applied to power portable electronics. Considering these advantages, LIBs can be regarded as perspective power sources for the abovementioned small-sized devices. The power supply requirements are determined by device construction, functions, and operating conditions. Compact LIBs can be fabricated using traditional electrode manufacturing technology, such as a conventional casting approach [
3]. For instance, Wyon produces lithium-ion cells of 6.3 mm
3 with 160 μAh capacity and 94 Wh/L [
4] energy density. Smaller LIBs can be manufactured using semiconductor technology.
LIBs with thin-film solid-state construction (TFSSLIBs) have been under development for many years [
5]. Some prototypes and products are commercially available but have not yet appeared on the mass market. The capacity of TFSSLIBs varies from 50 to 10,000 μAh, and the energy density lies in the range of 2–28 Wh/L. A description of highly cited products and prototypes can be found elsewhere [
6].
TFSSLIB components (positive electrodes, negative electrodes, protection layers, solid electrolytes/separators) can be fabricated using various methods of producing thin films: atomic layer deposition (ALD), chemical vapor deposition (CVD), magnetron sputtering, physical vapor deposition, pulsed-laser deposition, and others. A review of recent achievements in the research and development of TFSSLIBs can be found elsewhere [
5,
7,
8].
Among the above methods, ALD is one of the most promising. ALD is based on self-terminating chemical reactions between surface species of a solid substrate and volatile precursor [
9,
10,
11]. Owing to its self-limiting nature, ALD allows growing of conformal and uniform coatings on the surface of not only planar but high-aspect and porous structures [
12,
13]. Additionally, ALD allows controlling the thickness and composition of films with high precision and can be used for manufacturing not only separate battery components (anodes, cathodes, or solid-state electrolyte), but also entire TFSSLIBs [
8].
ALD has been successfully used for preparing active anode materials such as metal oxides TiO
2 [
14,
15], SnO
2 [
16,
17], combinations of metal oxides (SnO
2/ZnO
2 [
18], Fe
2O
3–SnO
2 [
19]), and lithium titanate [
20,
21,
22]. Thin films of solid electrolytes, such as Li
7La
3Zr
2O
12 [
23], Li
7La
2.75Ca
0.25Zr
1.75Nb
0.25O
12 [
24], LiPON [
25,
26,
27,
28], LiNbO
3 [
29,
30,
31], lithium phosphate (LPO) [
15], LiTaO
3 [
31,
32,
33,
34], LiAlO
x [
35,
36], lanthanum titanate, lithium lanthanum titanate (LLT) [
37], lithium silicates [
38], Li
2O-Al
2O
3 [
39], and Li
3BO
3-Li
2CO
3 [
40] were also effectuated by ALD. The possibility of obtaining operable cathode materials composed of LiFePO
4 [
41,
42,
43], LiCoO
2 [
44,
45], Li
xMn
2O
4 [
46], β-MnO
2 [
47], MnO/LiMn
2O
4 [
48], and V
2O
5, Li
0.2V
2O
5 [
49,
50,
51,
52] was also demonstrated.
However, non-lithiated cathodes do not contain lithium ions, and therefore cannot be used in full cells when combined with an anode (such as metal oxides prepared by ALD) [
53]. Therefore, only lithium-containing ALD-grown cathode materials have realistic prospects for use. Nevertheless, there are several difficulties in obtaining lithium-containing materials by the common ALD processes. First, stable lithium-containing precursors that have high vapor pressure are not currently known. The most frequently used precursors, such as Li(thd), Li(O
tBu), and LIHMDS, require significant heating to achieve the required vapor pressure [
54]. Second, in the synthesis of lithium-containing materials, the use of water as a counter-reagent has some limitations. The point is that lithium not only easily hydrolyzes and forms hydroxide, but also serves as a reservoir for water, which takes part in the proceeding reactions and hampers the surface-limited mechanism of deposition [
54]. To solve this problem, it is necessary to use ozone or oxygen plasma as a counter-reagent. Third, lithium is highly mobile in the temperature range (200–300 °C) usually used for ALD, which may lead to not only ALD surface-limited reactions but also to bulk-controlled growth.
Moreover, the cathode materials are ternary (LiCoO
2, LiMn
2O
4, Li
0.2V
2O
5) or quaternary (LiFePO
4). Ideally, the composition and growth rate of ternary/quaternary compounds should be a linear combination of the growth rates and compositions of binary compounds obtained by ALD processes [
55]. However, owing to a variety of nonidealities, including nucleation effects and precursor ligand interactions, the experimental characteristics of the materials differ significantly from the expected properties.
For the ALD of lithium-containing cathodes, several approaches have been commonly used:
- (1)
Supercycle approach. An ALD supercycle is defined as the minimum sequence of individual binary cycles that are repeated over the course of the ALD process [
55]. For example, one ALD supercycle for the deposition of A
xB
yO
z is composed of a linear combination ALD subcycles for binary compound deposition, i.e., (n × A
xO + k × B
yO), where n and k are the numbers of ALD deposition cycles of the binary compounds A
xO and B
yO, respectively. The ratio and the sequence of subcycles can be chosen, considering the growth rates of the binary compounds [
33] and the appearance of layers in the crystal structure [
42].
- (2)
ALD process, which uses multiconstituent precursors, i.e., precursors containing two or more elements of the resulting films. This approach was successfully used for ALD of lithium phosphates using lithium tert-butoxide (LiO
tBu) as a lithium source and trimethyl phosphate (TMPO) as phosphate source [
15].
- (3)
ALD of multilayered films of lithium oxides and metal oxides followed by annealing [
45]. In some cases, lithiation can occur without annealing, but rather directly during the ALD of lithium oxide on the surface of already deposited β-MnO
2.
The capacity, potential vs. lithium, density of electrode active layer, and construction affect the energy density of LIBs. One of the major parameters is the capacity of the active electrode layer. As stated above, many cathode materials that are widely used in LIBs (LiCoO
2, 140 mAh/g [
56]; LiMn
2O
4, 120–130 mAh/g [
57]; LiFePO
4, 150 mAh/g [
58]) have been deposited by ALD to study their electrochemical characteristics. However, to date, no study has been published on ALD of lithium nikelates or nickel-rich layered cathode materials; these might exhibit higher capacities (200–215 mAh/g) and possess an average potential near 3.6 V, which could improve the energy density of TFSSLIBs. Taking into account the results of our previous research on ALD of lithium oxide [
59] and nickel oxide [
60], we have tried to obtain lithium/nickel-based active cathode material and study its properties.
2. Materials and Methods
Monocrystalline silicon substrates (surface orientation 100, 40 × 40 mm
2, Telecom-STV Co., Ltd., Zelenograd, Russia) and stainless-steel plates (316SS, 16 mm diameter, Tob New Energy Technology Co., Ltd., Xiamen, China) were used as substrates for ALD. Prior to deposition, the silicon and stainless-steel substrates were cleaned in an ultrasonic bath in acetone and deionized water for 10 and 5 min, respectively. After that, the silicon substrates were immersed for 5 min in 10% HF to remove the native silicone oxide. Then, the substrates were cleaned using piranha solution (H
2SO
4/H
2O
2, volume ratio 7:3) for 20 min to remove organic contaminants and produce a hydroxylic surface. Finally, the silicon substrates were rinsed in double deionized (DI) water and dried under an argon atmosphere [
16].
Lithium–nickel oxide (LNO), nickel oxide (NO), and lithium oxide (LO) were deposited by ALD with a commercial R-150 setup (Picosun Oy, Espoo, Finland) at a temperature 300 °С and a reactor base pressure of 8–12 hPa. Lithium hexamethyldisilazide (LiN[(CH
3)
3Si]
2, LiHMDS; 99%, Sigma-Aldrich, St. Louis, MO, USA) and bis(cyclopentadienyl) nickel(II) (Ni(Cp)
2; 99%, Dalchem, Nizhny Novgorod, Russia) were used as the lithium and nickel-containing precursors, respectively. The LiHMDS and Ni(Cp)
2 were kept in stainless-steel bottles (PicohotTM 200, Picosun Oy) and heated 160 and 110 °С, respectively. The pulse times were 0.1 and 5 s for LiHMDS and Ni(Cp)
2, respectively. Remote oxygen plasma was used as a counter-reagent. The plasma power was 3000 W, with a frequency range of 1.9–3.2 MHz. The total oxygen plasma pulse time was 19.5 s (Ar purge during 0.5 s with flow rate 40 sccm; Ar and O
2 plasma purge during 14 s with flow rate 90 sccm; Ar purge during 5 s with flow rate 40 sccm). These deposition conditions were selected based on our previous studies devoted to obtaining lithium and nickel oxide [
59,
60].
We used two approaches for the deposition of lithium–nickel oxides. The first was the supercycle approach. This approach combines the normal ALD cycles of sequential precursor and co-reactant pulses for each constituent process into a cycle of cycles [
55]. We adopted the following ALD cycle ratios in a supercycle: LiHMDS–O
2 plasma/NiCp
2–O
2 plasma equal to 1/1, 1/2, 1/3, and 1/10 (short names: LNO-1/1, LNO-1/2, LNO-1/3, LNO-1/10). The second approach was ALD of the multilayered sample (LNO-M), which was obtained by the following process: NO (2500 cycles) + LNO-1/3 (100 cycles) + LO (300 cycles) + LiNiO-1/3 (100 cycles) + NO (100 cycles). Then, the as-deposited LNO-M and LNO-1/N (where N = 1, 2, 3, 10) samples were annealed at 400–900 °C for 15 min under an air atmosphere.
The spectroscopic ellipsometry (SE) measurements of amplitude ratio (Ψ) and the phase difference (Δ) for the films deposited on a silicon substrate were carried out with an Ellips-1891 SAG ellipsometer (CNT, Novosibirsk, Russia) in a wavelength range from 370 to 1000 nm and an incidence angle of 70°. The Spectr software package (1.10, CNT, Novosibirsk, Russia) was used to construct and fit a structural–optical model function. After fitting the parameters of the optical model and experimental spectra, the thicknesses of the films were calculated. The errors of the film thickness calculation were no more than 0.3 nm. The gradient of the thickness (GT) was calculated using Equation (1):
where T
max and T
min are the maximum and minimum film thicknesses, respectively [
16].
X-ray reflectometry (XRR) and x-ray diffraction (XRD) studies were performed on a D8 DISCOVER (Cu-Kα) diffractometer (Bruker, Billerica, MA, USA). Surface-sensitive grazing incidence XRD (GIXRD) modes was used for XRD measurements using 2θ range of 15–65° with a step of 0.1°. Exposure time at each step was 1 s. The incidence angle of the primary X-ray beam was 0.7°. XRR measurements were performed in an angles range 0.3–5° (increment 0.01°) using symmetric scattering geometry. The results of XRD measurements were processed by the Rietveld method using the TOPAS software package (5, Bruker, Billerica, MA, USA)and XRR curves were fitted by the simplex method using LEPTOS (ver. 7.7, Bruker, Billerica, MA, USA).
X-ray photoelectron spectra (XPS) were acquired with a Escalab 250Xi spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). For depth profiling studies, the samples were sputtered by Ar+ ions with an energy of 3 keV, for sputtering times ranging from 15 to 1035 s. The samples were excited by Al Kα (1486.7 eV) X-rays in a vacuum no more 7 × 10−8 Pa.
Elemental depth profiling was also carried out with a time-of-flight secondary ion mass spectrometer (TOF SIMS 5 instrument, ION-TOF GmbH, Münster, Germany). Cs (0.5 keV, area 120 × 120 µm2) and O2 (0.5 keV, area 150 × 150 µm2) were used for sputtering. The measurements of depth profiles were performed by dynamic SIMS mode using the primary ion gun (Bi+ at an energy of 30 keV and a probe measured sample current of 3.1 pA, detection area 100 × 100 µm2).
Scanning electron micrographs of planar and cross-sectional views were obtained using a Merlin scanning electron microscope (SEM, Zeiss, Oerzen, Germany) with a Gemini-II column and a field emission cathode, and a Mira3 SEM (Tescan, Brno, Czech Republic). Both SEMs had field emission cathodes. The SEM spatial resolution was approximately 0.8 nm at an accelerating voltage of 15 kV. A total of 3–4 randomly selected positions on the surface of the sample were investigated. Everhart–Thornley and InLens secondary electron detectors were used for SEM studies. Energy-dispersive X-ray (EDX) analysis was performed using an INCA X-act (Oxford Instruments, High Wycombe, UK) installed on the Zeiss Merlin SEM. A Zeiss Auriga focused ion beam scanning electron microscope (FIB-SEM) dual-beam station was used for lamella preparation for transmission electron microscopy (TEM) studies. The atomic structure, EDX, and electron diffraction patterns were investigated on a Zeiss LIBRA 200FE TEM with an accelerating voltage of 200 kV equipped with an Oxford Instruments INCA X-Max system.
The electrochemical activity of stainless-steel substrates with deposited LNO and LNO-M coatings before and after annealing was studied in coin cells (CR2032). Metallic lithium was used as the counter electrode. Porous polyolefin film (2325, Celgard, Charlotte, NC, USA), and TC-E918 (Tinci, Guangzhou, China) were used as the separator, and electrolyte, respectively. The composition of TC-E918 was a 1-M solution of LiPF6 in a mixture of organic carbonates. The coin cells were assembled using an OMNI-LAB argon glove box (VAC, Hawthorne, CA, USA); the H2O content was less than five ppm. Cyclic voltammetry (CV) was studied using a potentiostat PGSTAT302N+ (Autolab, Utrecht, The Netherlands) in a range of 2.5–4.3 V with 0.5 mV/s scan rate. Charge/discharge cycling was performed using calibrated channels of a CT-3008W-5 V 10 mA battery testing system (Neware, Shenzhen, China) at room temperature, in potential and current ranges of 3.0–4.3 V and 20–80 µA, respectively.
4. Conclusions
Atomic layer deposition of lithium–nickel–silicon oxide thin films was performed using LiHMDS and NiCp2 as precursors and remote oxygen plasma as a counter-reagent. XRD, EDX, and XPS data indicated that LNO nanofilms deposited using the supercycle approach either do not contain Ni (LNO-1/1 and LNO-1/2) or only contain Ni in small amounts (LNO-1/3, LNO-1/10). However, these films contain lithium carbonates on the surface and a large amount of silicon in the bulk. The latter forms crystalline Li2SiO3 and Li2Si2O5 upon annealing at 800 °C. Obviously, silicon arises from LiHMDS, a silicon-containing precursor.
A multilayered LNO-M sample was successfully deposited using ALD of NO, LNO, LO, LNO, and NO layers. XRD data showed that the LNO-M sample contains crystalline NiO and annealing at 800 °C leads to the formation of Li2Si2O5 and probably Li5Si22 intermetallide. Local analysis of LNO-M-800 by SAED also showed the presence of LiNiO2. According to XPS and TOF-SIMS depth profiling, the annealing caused interdiffusion of layers, which leads to homogenization of the layer composition. In addition to homogenization, iron, chromium, and nickel diffuse from the stainless-steel substrate into the film. However, STEM analysis showed that the annealed films are not homogeneous at the micro/nanoscale. Submicron-scale crystallites of NiO/LiNiO2 and predominantly amorphous silicon enriched layers were found.
Based on the shapes of the CV curves and the discharge curves, it can be assumed that the discharge capacity of deposited films is due to the intercalation of lithium ions, during which the charge of nickel ions changes. We assume that the LiNiO2 detected by SAED is this electrochemically active phase. The values of the specific capacities for annealed LNO-M samples were in the range of 20–26 μAh·μm−1·cm−2 (at discharge currents of 5–7 C) and they are lower than that calculated for a dense LiNiO2 film (0.5 C, 103 μAh·μAm−1·cm−2), but very close to 27 μAh·μAm−1·cm−2, as obtained for ALD-deposited LiCoO2 thin film.
Thus, using ALD method, the positive electrodes were prepared that can be discharged by relatively high currents and potentially suitable for creating thin-film lithium-ion batteries with increased power density. Considering that the ALD method allows coating on substrates with high aspect ratio trenches (high specific surface area), the specific weight of active material applied per unit of geometric area of the substrate can be augmented. As a result, the working time interval can be extended due to higher energy density.