Effect of RF magnetron sputtering parameters on the optimization of the discharge capacity of ternary lithium oxide thin films

The increasing demand for lithium-ion batteries has stimulated the investigation of new compounds in order to reduce the costs and the toxicity of their cathodes. Materials constituted of ternary lithiated oxide compounds are a successful alternative to cobalt-rich cathodes. The main disadvantage of ternary compound materials (TCM) is that the maximum amount of electrical charge is only achieved at high redox potentials, a limiting factor if we consider the current development in electrolyte technology. In this work, we investigated the influence of sputtering deposition parameters on the charge capacity of TCM thin films, restraining their electrochemical potential to conventional values. To do so, we analyzed the impact that small changes in crystalline and morphological structures have on the charge capacity at low cell potentials. For this, we performed the RF mA gnetron sputtering of TCM thin films and carried out a factorial design of experiments to investigate their electrochemical properties, while limiting the charging potential to 4.20 V vs. Li|Li+. The films were deposited onto a rigid and conductive substrate with different parameters (power and pressure at room temperature). Electrochemical results showed that the discharge capacity is strongly influenced by the deposition parameters, reaching 250 mA h g−1 even at 4.20 V vs. Li. This value is superior to the ones of the conventional cobalt cathode and the bulk ternary electrode. Both deposition parameters exhibited a synergic dependency, which means that they need to be simultaneously varied for a response optimization. The discharge capacity of the analyzed samples was highly affected by the surface morphology of the film and its crystallographic properties, and not by its elemental composition. High discharge capacity was obtained without additional thermal treatments, which favors the manufacture of films over polymeric substrates for future electronic applications.


Introduction
Among all transition metal oxides used as cathode material in lithium-ion batteries, the lithium cobalt oxide (LiCoO 2 , LCO) stands out due to its high discharge capacity (140 mAhg − 1 ), speci c energy (250 Whkg − 1 ), good cycling stability and easy production. This compound was originally synthesized by Mizushima and Goodenough [1] and, since the 1990s, it has been widely employed in the composition of commercial rechargeable lithium-ion batteries (LIBs). However, natural cobalt reserves are scarce and concentrated in a few regions of the planet, which not only makes the technology more expensive, but also unfeasible.

SPUTTERING TARGET
The sputtering target was produced of synthesized NMC333 powder, pressed into a 2-inch-diameter cylindrical stainless steel cup container with a height of 2 mm and a wall thickness of 1 mm. The NMC333 powder was synthesized via sol-gel method [23], with lithium, nickel, manganese and cobalt acetates as precursor materials (Dinamica Chemistry Contemporary Ltd and Labsynth), in stoichiometric quantities. The gel, obtained by drying the metals-containing solution with citric acid, was pre-calcined at 450°C for 6 hours, ground and re-calcined at 700°C for 2 hours. The stainless steel container not only provided physical support to the NMC powder, but also increased the thermal conductivity at the base of the target.
The base pressure for chamber cleaning was 2x10 − 6 mbar. Titanium was used as an anchoring layer between aluminum and glass. The analyzed deposition parameters were power and deposition pressure, scrutinized in two levels, as shown in Table 1. Table 1 Variables and levels of the 2 2 factorial design, for TCM thin lms deposition by RFMS The NMC333 sputtering was carried out with an analytical AR (White Martins) controlled by a owmeter, using 100 W (-) and 200 W (+) as power limits, and 1.02x10 − 2 mbar (-) and 1.35x10 − 2 mbar (+) as pressure limits (Pr) for the deposition. The sputtering plasma was maintained by auto-tuning the impedance of the whole system. The thickness of the lms (100 nm) was monitored in-situ by a quartz crystal microbalance. The mass of the lms was estimated with basis on the density of the bulk NMC333 compound (ρ NMC333 = 3.7 g/cm 3 ).

CRYSTALLOGRAPHY
The X-ray diffraction analyses of NMC333 powder and TCM lms was performed using a Panalytical XPert PRO MPD diffractometer, equipped with a Bragg-Brentano geometry of CuKα radiation source (λ = 1.5419 Å), positioned in an angular step of 0.02° with 2 seconds per step, in a 2θ range of 10 to 90 degrees. The powder diffractogram was performed through the re nement of the crystalline structures via Rietveld´s method, using the Xpert Highscore Plus program. The Scherrer equation was employed to evaluate the crystallite size of the lms, using a Gaussian function for estimating the full width at half maximum (FWHM). Polycrystalline silicon was used as instrumental-standard sample to determine the instrumental broadening effect factor. Errors were estimated based on FWHM deviation.

STOICHIOMETRY
Both the target and the lm stoichiometries were carried out by X-ray uorescence (XRF) in a Shimadzu EDX-720 energy-dispersive spectrometer (Shimadzu Co., Kyoto, Japan). The spectrometer was equipped with an Rh tube and a Si (Li) detector, with an active window of 25 mm 2 and 0.5 mm thick, having a 135 eV resolution at Mn-K line. The measurement time was 500 seconds.

SURFACE MORPHOLOGY
The topography and the surface roughness of the lms were characterized by an atomic force microscope with Nanosurf system ( exAFM), consisting of a modular exploration probe that can be updated to improve the measurement capacity in real-time. The measurements were performed employing contact force mode, through a exible cantilever (TAP1990AL) that provided a sweep of 5 µm 2 , with 512 points per scan line.

FOURIER TRANSFORM INFRA-RED SPECTROSCOPY
Fourier-Transform Infra-Red (FTIR) spectra ranging from 400 to 4000 cm − 1 were obtained using a Perkin-Elmer Frontier spectrometer with Attenuated Total Re ectance (ATR). The Fourier Transform was performed over 16 scans, providing an averaged single spectrum as result.

ELECTROCHEMICAL PERFORMANCE
The electrochemical performance was assessed via chronopotentiometry, using a multi-channel potentiostat of Arbin Instruments (MSTAT 8000/BT 2000). In the analysis, we used a three-electrode cell with a counter electrode (CE), a lithium metal wire as a reference electrode (RE), and a TCM lm as a working electrode (WE). The employed electrolyte was LiClO 4 /EC/DMC (1 mol/L), and all utilized chemical reagents were produced by Sigma Aldrich. The cell was assembled at room temperature, under an inert atmosphere, inside an MBRAUN glove box with water and oxygen concentrations below 5 ppm.
Measurements were performed at a potential range between 2.20 and 4.20 V (vs. Li + |Li), with a current density of 15.0 mA g − 1 , for 9 cycles.
3 Results And Discussions 3.1 CRYSTALLOGRAPHY Figure 2a shows the diffractogram of the NMC333 powder used to produce the target. Narrow and intense peaks indicate a compound with a high order of crystallinity, indexed to a α-NaFeO 2 hexagonal layered structure, with space group R-3m [7,9,26]. Based on the diffractogram and applying the Scherrer equation, while using silicon as an instrumental-standard sample for determining the FWHM value, the crystallite size of the target was estimated in 24.5 nm [17].
The diffraction patterns of as-deposited TCM lms under different conditions are shown in Fig. 2b. All diffractograms exhibited three peaks: the rst, around 35º (2θ), belonging to the Ti phase; the second, at 38.0° (2θ), corresponding to the cubic phase of a conductor aluminum lm with preferred orientation towards the direction of [111]. The highest intensity peak of aluminum can be attributed to the structure factor (F (hkl) ) of an FCC-type lattice, which implies an increase in the intensity of the scattered X-rays. The third, at 44.0° (2θ), is related to the TCM lm whose Miler index is (104). The set of planes appears to be spatially oriented in parallel to the substrate, in the [104] direction, and his preferential orientation may be due to the minimization of the surface energy, since these lms have a lesser thickness [20,27,28].
The crystallite size of the TCM lms are shown in Table 2. Table 3, by its turn, presents the correlations between the deposition parameters (power and pressure). It can be seen that the power has a subtle, but statistically signi cant, effect on the size of the crystallite, while the pressure does not seem to in uence this response. The positive value of the main effect related to power indicates that an increase in this factor causes an increase in the crystallite size. Such occurrence is expected because, with the rise of the sputtering power, the atoms ejected from the target reach the substrate with greater kinetic energy, favoring the crystallization of the lm. The pressure main effect and the incidence of interaction values close to zero indicate that these effects exert no in uence on the crystallite size. The intrinsic crystallinity of the lm obtained via RFMS deposition is an important factor, which demonstrates the feasibility of this process in obtaining crystalline lms. Furthermore, such factor is crucial for eliminating post-thermal treatments, allowing for the manufacture of electrodes especially designed for exible micro-devices, since they do not support high temperatures. Table 4 presents elemental quantitative results obtained from the X-ray uorescence of the ternary lms. Through the analysis of the data, it is possible to notice that the lms presented a stoichiometry different from that of the target. All TCM lms showed a high concentration (in mol %) of manganese (45%), an intermediate concentration of cobalt (30%), and a low concentration of nickel (25%). The discrepancy between stoichiometry values (1/3:1/3:1/3) can be due to the different sputtering yields for each atomic species. According to the factorial planning, and considering that no speci c order was followed during the depositions, XRF results revealed that the deposition parameters (power and pressure) do not have a great impact on the stoichiometry of the lms.  Figure 3 exhibits the atomic force microscopy (AFM) images of the surface of (a) clean glass -g; (b) titanium-coated glass -g/Ti; (c) aluminum-coated glass -g/Al; and (d) glass initially covered with titanium, then by aluminum -g/Ti/Al. Titanium was used as an anchoring layer not only due to its strong bond with the oxygen present in the glass substrate, but also due to its small atomic size, close to those of aluminum and silicon, which strengthen the adherence of the aluminum layer, contributing to lower surface tensions. Roughness values are shown in Table 5. The titanium lm seems to make surfaces smoother, whether it is deposited on top of glass (b) or under aluminum (d). This effect causes a decrease in the number of dispersion centers for conduction electrons, located in grain boundaries, causing a decrease in the resistivity [29].  Fig. 4, in a way to facilitate the analysis of the in uence of deposition parameters on the roughness and on the shape of surface grains. When horizontally comparing the images, we can observe the in uence of the power on the morphology of the lm, while a vertical comparison allows for an analysis of the in uence exerted by the deposition pressure. Roughness values are presented in Table 6 and geometrically interpreted in Fig. 4. Results indicate that the highest roughness value was obtained under (Po-, Pr+) conditions. Additionally, the main effect of power occurs due to the kinetic energy of the ions colliding with the target. An increase in power (Po+) causes the atoms to be ejected from the target with greater kinetic energy, which leads to a faster coalescence and favors the formation of a atter surface. Table 7 shows the analysis concerning the main effect of Po and Pr, along with the interactions between the two parameters. Power and pressure in uenced the roughness response, but the negative value (-7.2) assumed for their interaction not only suggests a synergic effect between them, but also an inversely proportional dependence encompassed by both parameters and the roughness coe cients. Therefore, considering their interdependency, it is not recommended to individually analyze each parameter, i.e., only their main effects. However, we can not neglect the signi cant main effect of power (-15.3), which indicates a major in uence on roughness, but in an inverse mode (should the power increase, the roughness decreases). Po + Pr-6.5 ± 0.9

ATOMIC FORCE MICROSCOPY
Po-Pr+ 29.5 ± 4.7 A scanning electron microscopy (SEM), detailed in Fig. 5, revealed that the surface of the lms was free from deep cracks, which indicates a good adhesion of the Ti/Al lms to the glass substrate. The spherical structures observed in all images differ in diameter, suggesting that pressure was the main acting factor, once Pr + lms presented smaller grains than those of Pr-(vertical analyses). Additionally, if we compare the images concerning the deposition power, it is possible to observe that Po + lms tend to be smoother and more compact, with feasible alterations in their density. However, no appreciable difference was observed between Po-and Po+ (horizontal analyses). After confronting Po-Pr-and Po + Pr-, it was possible to observe the presence of spheres with a larger diameter when compared to those of the samples of the top row. Furthermore, Po + Pr-showed grains linked to each other. Po + Pr+, by its turn, also presented joint grains, if compared to Po-Pr+. Such ndings are in agreement with the roughness obtained via AFM, showing that these samples are smoother than Po-Pr + and Po-Pr-.  showed a signi cant improvement in capacity. In general, the obtained discharge capacities were higher than those of conventional cathode composites, which require the addition of binders and electronic conductors [6,26]. Furthermore, the attained values were close to the theoretical capacity, showing that all lithium ions could be reversibly de-intercalated from the host material structure. The Po-Pr-sample, on the other hand, showed good reversibility up to the ninth cycle and presented a high discharge capacity (250 mAhg − 1 ), even with a cut-off potential of 4.20 V. The increase in the discharge capacity, observed from the rst to the ninth cycle, can be attributed to the lattice accommodation and to the activation of more ionic intercalation sites. The discharges capacities are summarized in Table 8. Po-Pr+ 180 ± 2 Table 9 describes the effects exerted by deposition parameters on the discharge capacity. The high negative value of the power main effect (-167.1) indicates that an increase in this parameter would cause a decrease in the discharge capacity, whereas the opposite is valid for the pressure (since it exhibited a positive value). Due to its high value, the power effect exerted more in uence on the response of the discharge capacity. On the other hand, as far as roughness is concerned, the interaction effect was positive, showing a synergistic interaction between Po and Pr. In this case, though, the positive value demonstrates that both parameters directly in uenced the response.

Conclusions
In this work, we investigated the correlation between the deposition parameters of RF magnetron sputtering and the electrochemical response of thin lms synthetized with lithium ternary transition metal oxides (LiMO 2 , M = Mn, Co, and Ni) at 4.20 V, using a 2 2 factorial design of experiments.
The lms and the sputtering target did not exhibit the same elemental composition. However, with basis on electrochemical results, it became evident that, even with an excessive amount of manganese, the delivered discharge capacity of 250 mAhg − 1 was almost two times higher than that of traditional bulk LiCoO2, even when limiting the cell potential to 4.20 V during the charging process.
High discharge capacity was obtained within the potential window conventionally used in LIB´s, with no need of additional thermal treatments. Such conditions favor the manufacture of RF magnetron sputtered TCM lms on polymeric substrates, typically employed in exible devices.

Declarations Funding
This article was partially supported by the following funding agencies: Conselho Nacional de Pesquisa (CNPq) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), which contributed with the nancial support for this work, and Financiadora de Estudos e Projetos (FINEP), for providing equipment infrastructure.

Con icts of interest/Competing interests
Urbano]. All authors commented on previous versions of the article, and read and approved the nal manuscript. Figure 1 Half electrochemical cell illustration, portraying the monolithic arrangement of Ti, Al and TCM thin lms onto glass substrate Chronopotentiometric charge and discharge cycles of TCM lms, tested at a current density of 15.0 mAg-1.