Plasma processes in the preparation of lithium-ion battery electrodes and separators

In a radio frequency (RF) inductively coupled plasma (ICP) torch, the plasma is generated by coupling the electromagnetic field induced by a coil into the cylindrical discharge cavity. RF-ICP plasmas are characterized by the absence of electrodes and can be operated under inert, oxidizing, reducing or other reactive atmospheres. Soft vacuum conditions usually prevail ($20{-}100$ kPa). Pure Ar or Ar-containing mixtures are generally used as central gases (Ar being easily ionized), while reactive gases, if any, are injected into the sheath stream. The LTE plasma temperature can reach $8000{-}10\,000$ K, while the plasma velocity remains under $100$ m s⁻¹, unless a de Laval nozzle is used (in which case the flow can reach supersonic velocities). One of the largest applications of RF-ICP torches is the synthesis of nanomaterials with high purity (the electrodeless plasma generation reduces the risks of contamination) and with controlled chemistry and morphology [24].

Two sequences of physical and/or chemical processes can occur within the plasma and lead to the formation of nanoparticles: evaporation-condensation and evaporation-reaction-condensation [24]. Feedstocks (powders, suspensions, solutions, catalysts and/or gases) are injected axially into the plasma discharge cavity, heated in flight, melted and vaporized. Depending on the nature of the resulting vapors and injected gases, one or more species can react in-flight within the plasma jet. Then, a rapid quenching of the vapor cloud inside a cylindrical reactor (figure 3) results in condensation and nucleation of nanoparticles, which size, morphology and composition (e.g. nanospheres, nanowires, nanotubes, nanoplatelets, pure metals, oxides, carbides, etc) will depend on the quenching conditions [43]. For example, long isotherms in a reducing atmosphere are suitable for the growth of carbon nanotubes [44], while a rapid quench in an oxidizing atmosphere can lead to small particles having polyhedral shapes [45]. Generally, smaller particles are obtained by limiting their recirculation within the quench zone [43, 46]. The nanoparticles are collected onto the reactor walls or onto a downstream porous filter. Reactive gases can be used to modify or to functionalize the particles in flight, as demonstrated with carbon nanotubes in the presence of NH₃ downstream of the growth zone [47]. This is an interesting feature for LIB materials. ICP torches can also be used to spray coatings under controlled atmospheres (figure 3).

Zoom In Zoom Out Reset image size

Direct current (DC) plasma torches generate plasma by striking an arc between the cathode and the anode nozzle. Electrode erosion is unavoidable, which is a source of contamination limiting the use of DC plasma torches for the synthesis of high purity nanomaterials. The plasma gas mixtures usually contain a primary heavy gas for momentum transfer (Ar or N₂) and a secondary gas for heat transfer (He or H₂). Peak LTE temperatures reach $12\,000-15\,000$ K and velocities range between $500{-}2500$ m s⁻¹. Generally, DC plasma torches operate at atmospheric pressure in air, which makes them suitable to spray thick oxide coatings over large areas (see section 2.2.4).

In its simplest form, a hybrid plasma torch consists of a DC plasma torch mounted onto an ICP torch (figure 4). The DC plasma jet flows axially into the discharge cavity of the ICP torch, where the induction coil provides energy to increase the volume of the discharge. The main advantage of the hybrid plasma torch is that the feedstock (powder) can be injected into a large volume, high-temperature plasma jet without being disturbed by the recirculation eddies [49]. Powders are thus efficiently heated and vaporized, making the hybrid torch a valuable tool for plasma spray physical vapor deposition (PS-PVD) or to synthesize nanoparticles. Adding a secondary precursor to the plasma (for example CH₄ as a source of carbon) enables the formation of a composite powder from the co-condensation of vapors.

Zoom In Zoom Out Reset image size

In conventional plasma spraying, a thermal plasma torch (DC or ICP) provides the energy required to melt the powdered feedstock material (diameters in the 10–100 µm range). Powders are injected radially (DC torches) or axially (ICP torches) into the plasma where particles are heated, melted and accelerated towards a substrate (figure 5). Upon contact with the substrate, partially-melted and melted droplets flatten and solidify, forming splats. Thick coatings (1–500 µm) result from the stochastic buildup of several splat layers, as illustrated in figure 5. Although finely structured or nanostructured coatings could be obtained from reduced-size feedstock powders, conventional carrier gas injection of solid nanoparticles within the plasma jet is inefficient: the flow rates required to impart sufficient momentum to the nanoparticles interfere detrimentally with the plasma [50]. Nanostructures are generally obtained with nanoparticle suspensions or precursors in solution as feedstock; the resulting processes are named SPS and SPPS, respectively [51]. Nanoparticles are formed in-flight from the precursors in SPPS, which reduces the health risks associated with their handling. In both processes, the liquid feedstock (suspension or solution) is injected in the plasma and complex plasma-liquid interactions dictate the coating microstructure in a 3-step process: (i) droplet fragmentation and solvent vaporization, (ii) surface precipitation and pyrolysis (in SPPS) and iii) acceleration and heating of nanoparticles/melted droplets to form a coating onto a substrate [52]. The spraying conditions in SPS and SPPS can be adjusted to form either dense or highly porous, cauliflower-like, coatings.

Zoom In Zoom Out Reset image size

Plasma spray is a promising process to deposit binder-free electrode materials for LIB. In this case, the substrate consists of a flexible metal foil that will act as a current collector for the Li-ion battery. The feedstocks can be powders, suspensions or solution precursors of active materials that will be melted in-flight to form a coating onto the current collector, eliminating the need for a subsequent casting of the active material with a polymeric binder. Examples are provided in section 3.

Plasma spray physical vapor deposition (PS-PVD) is similar to conventional plasma spray, but the chamber pressure is reduced to values down to $0.1{-}1$ kPa. At high plasma torch powers, this ensures that all the feedstock material is vaporized, enabling the deposition of a coating from the vapor phase [53]. In fact, columnar coatings similar to those obtained by electron beam physical vapor deposition (EB-PVD) have been produced by PS-PVD, albeit with larger deposition rates. It is also possible to rapidly quench the vapors to form nanomaterials, similarly to what is achieved with an ICP torch.

A glow discharge is a self-sustained luminous plasma generated between two electrodes to which a DC voltage is applied (figure 6). The electrodes are usually located inside a low pressure chamber (0.1–10 Torr). This simple and inexpensive plasma source produces electrons, excited molecules, atoms and ions that efficiently trigger selective chemical reactions. This is useful for the treatment of surfaces, sputtering and film deposition. The cathode region of the discharge is used for sputtering and deposition, whereas the negative glow will be preferred for chemistry [54]. However, the glow to arc transition limits the power that can be coupled into a glow discharge (if an arc discharge forms, the current must be increased to sustain the power, which in turns lead to the thermalization of the gas as in a DC plasma torch). To avoid the glow to arc transition, one can used either pulsed nanosecond discharges or RF discharges.

Zoom In Zoom Out Reset image size

Glow discharges are also limited to low gas pressures unless they are operated in fast flows, leading to a plasma source commonly referred to as atmospheric pressure glow discharge plasma jet or torch. Note that dielectric barrier discharges (DBD) operated at atmospheric pressure can also be configured in such a way that the discharge looks like a glow (absence of microdischarges). These are often interchangeably called glow discharge plasma jet or DBD plasma jet.

In a dielectric barrier discharge, the electrodes are excited by an AC (0.5–500 kHz) driving voltage. A dielectric covers one or both electrodes, or can be introduced in the low pressure discharge cavity, which prevents the formation of an arc (figure 7); in fact, the discharge is self-extinguished by the charge build-up on the surface of the dielectric before an arc could strike. At atmospheric pressure, DBD discharges are filamentary, each individual filament consisting in a breakdown channel called streamer or microdischarge. At lower pressures and higher frequencies, homogeneous diffuse glow discharges are obtained; they are referred to as RF glow discharges [55]. DBDs are simple to operate at powers greater than DC glow discharges.

Zoom In Zoom Out Reset image size

Non-thermal radio frequency plasmas are usually generated by capacitively coupling the RF electromagnetic field (1–500 MHz, commonly 13.56 MHz since it is a worldwide dedicated industrial, scientific and medical radio band) into the discharge cavity. They are referred to as capacitively coupled plasmas (CCP). In a CCP, a RF generator is connected to the electrodes, which may not be covered by any dielectric, as opposed to a DBD. Electrodes can be inside (figure 8) or outside the discharge cavity. Throughout the RF range, the electrons are able to follow the instantaneous RF field whereas the heavy particles cannot, creating nonequilibrium conditions interesting to stimulate chemical reactions in low-temperature conditions [39]. RF-CCP can also generate plasma jets called RF atmospheric glow discharges or atmospheric-pressure plasma jets (APPJ). CCPs have found several applications in plasma-enhanced chemical vapor deposition (PECVD) of silicon-based materials among others, which makes them interesting for LIB anode materials. When further increasing the frequency in the 0.5–10 GHz range, one reaches the microwave (MW) band and generates microwave plasmas. At the commonly used 2.45 GHz frequency, the corresponding wavelength (12.24 cm) is comparable to most plasma reactor sizes. If the MW frequency equals the electron cyclotron frequency, power absorption is maximized. The absorbed power in a RF or MW discharge also depends on the electron-neutral collision frequency and, consequently, on the pressure. Coupling is more efficient at lower pressures (electromagnetic waves with frequencies below the electron plasma frequency are reflected, which is proportional to the square root of the electronic density), but microwave discharges are also possible at atmospheric pressure. In this case, the energy will be coupled via the skin effect, but with a skin depth much thinner for MW compared to RF [54]. Other distinctions between RF and MW plasmas are reviewed elsewhere [56]. The most common configuration of microwave discharges has a rectangular waveguide crossing the discharge cavity enclosed in a dielectric.

Zoom In Zoom Out Reset image size

Coatings and films can be produced by plasma-enhanced chemical vapor deposition (PECVD). In such processes, the plasma works as a source of specific chemical radicals that are deposited on a substrate. To sustain large deposition rates, relatively high pressures and flow rates are required, which challenges the stability of the plasma discharge. Stable parallel-plate RF-CCP discharges are often used for this task. Amorphous silicon, silicon dioxide and carbon nanotubes have been successfully grown by PECVD, although not specifically targeting LIB applications. When cycling the RF-plasma, chemical reactions can be sufficiently well controlled to deposit thin films with atomic layer precision at low temperatures. The process is called plasma-enhanced atomic layer deposition (PEALD).

Sputtering is a process in which atoms are ejected from a surface owing to energetic ion bombardment. The presence of the magnetic field, perpendicular to the electric field, increases power absorption, a suitable feature to sputter material. This is done by trapping electrons within a magnetic field near the target, increasing the efficiency of the ionization process (electrons travel or spiral over longer distances) and allowing plasmas to be both generated at lower pressures and confined near the target with a higher density of ions. Ejected atoms become the source for a following thin film deposition. Atoms can be directly deposited without further modification (physical sputtering) or can react with other chemically active species while in flight in the plasma (reactive sputtering). Non-thermal plasma sources employed in sputtering processes are generally DC discharges for conductive films (DC planar magnetrons), and RF-CCP discharges or RF-driven planar magnetrons for insulating films. A DC planar magnetron is a glow discharge in which the negative plasma glow is trapped in the magnetic mirror formed by magnetron magnets (figure 9) [42]. Atoms can also be ejected from a target (material to be deposited) by other means, including electron and laser beams. In the latter case, a high-energy pulsed laser beam strikes the target surface, vaporize and ionize the material into the plasma state. The technique is called pulsed-laser deposition (PLD).

Zoom In Zoom Out Reset image size

The versatility of thermal plasma processing for LIB electrode materials was shown in a recent report where the authors synthesized lithium metal oxides (Mn, Co, Cr) in a single step, using an ICP torch with an applied power of 20 kW [61]. Powdered precursors (Li₂CO₃ and metal oxide) were injected into the plasma using Ar as carrier gas and a mixture of Ar and O₂ as the plasma forming gas. In the case of LiCoO₂, the particles obtained where found to have a size distribution between 50–80 nm with polyhedral shapes such as pentagonal, hexagonal and quadrangular, all morphologies that could be interesting for electrochemical testing. In fact, the electrochemical performance of electrodes can be tailored through particle morphologies since the crystal structure of a given compound has an important influence on the ion diffusion path [62–65].

Electrodes are conventionally prepared by casting a slurry containing the active material, carbon black and a polymeric binder and solvent on a current collector, generally Al, Ni or Cu foil and dried at 80–110 °C overnight. In the electrode preparation, the binder has the function to provide mechanical stability to the electrode, but it does not participate to the electrochemical reactions, adding an inert weight to the electrode. In this sense, direct deposition of the active material onto the current collector can simplify the electrode processing since the preparation of thin films is generally achieved in one step. Moreover, eliminating the passive binder has the potential to increase the energy density of LIBs. LiCoO₂ thin film electrodes were deposited by various plasma processes [66–68] and sputtering [69–71]. In this sense, a recent report showed the preparation of nanostructured LiCoO₂ electrode through the plasma spraying of metal Co on a stainless steel substrate. The deposition of 20–25 µm-thick coatings was achieved by 3–4 consecutive passes of the DC plasma torch. In a second step, the as-prepared Co coating was covered with a 1 M LiNO₃ solution and thermally treated at 810 °C under ambient conditions at atmospheric pressure and no additional gas supply [72]. Samples were treated for different time periods, from 10 s to 12 min, and XRD analysis showed that, after 5 min, hexagonal LiCoO₂ with space group R3m was already formed. LiCoO₂ electrodes thermally treated for 5 and 12 min were electrochemically tested, showing similar electrochemical performances, with a discharge capacity $\sim 100$ mAhg⁻¹ (75% of the nominal capacity). This capacity decreased with subsequent charge-discharge cycles, most likely due to the dissolution of Co in the electrolyte caused by the lack of carbon coating that would increase the electronic conductivity of the active material and promote the formation of a passivation layer capable of preventing parasitic reactions with the electrolyte.

Improving the electrode performance by adding homogeneous and thickness-controlled coating on particles is key to the development of more stable electrodes, a feature that can be achieved by plasma processes. Such is the case of the deposition of carbon layers on the surface of LiCoO₂ by PECVD [73]. The two-step process consisted in (i) preparation of a pellet electrode by mixing LiOH, H₂O, and Co(OH)_x precursors that were heated at 900 °C for 12 h in air and later mixed with synthetic graphite and polyvinylidene fluoride (PVDF), and (ii) the deposition of a carbon coating on the pellet electrode by reduction of ethylene using a glow discharge (PECVD process) with a chamber pressure of 0.08 Torr and a voltage of 1.2 kV. SEM and XPS confirmed the homogeneous deposition of a 0.1 µm-thick carbon layer that was characterized as diamond-like carbon using Raman spectroscopy. Additional to carbon coatings, different metal oxides have been deposited onto LiCoO₂ [74–79] to effectively prevent direct contact with the electrolyte solution. The objective is to suppress phase transitions, to improve the structural stability and to decrease the disorder of the cations in the crystal sites. For example, commercial LiCoO₂ was coated with ZnO by PECVD [80]. The room temperature process used diethyl zinc ((C₂H₅)₂Zn) as precursor and a RF plasma power of 200 W. During ZnO deposition, LiCoO₂ particles were mixed using a rake-shaped mixer inside the reaction chamber, which allowed the formation of uniformly distributed ZnO nanospheres on the surface of the particles and resulted in a coating of $\sim 25$ nm. Charge and discharge cycling was performed between 3.0–4.5 V to evaluate the effect of the ZnO coating. It was shown that during the first few cycles, ZnO coating had no effect on the electrochemical performance of the oxide when compared to uncoated sample. However, optimally coated samples showed better capacity retention over prolonged cycling, the ZnO layer being effective to prevent the dissolution of Co in the electrolyte.

Taking advantage of the versatility and fast synthesis of materials by means of plasma technologies, other layered lithium metal oxide compositions suitable for electrochemical storage have been investigated. For example, Li(Ni_1/3Co_1/3Mn_1/3)O₂ was prepared by a plasma-assisted solid-state method [81], in which a RF discharge in oxygen increases the reaction kinetics. The same group considered sulfur doping by mixing a Li(Ni_1/3Co_1/3Mn_1/3)O₂ powder with thiourea and exposing the mixture to a RF discharge in argon [82]. The doped material showed higher initial discharge capacity, better cycleability and an improved performance at high temperature [82]. Carbon coatings were also successfully deposited onto Li(Ni_1/3Co_1/3Mn_1/3)O₂ using microwave PECVD [83].

Spinel LiMn₂O₄ has also been largely investigated and used as positive electrode material for commercial LIB applications. Recent report on the synthesis of the spinel has targeted the control of both the particle size distribution and the morphology to enhance its electrochemical performance, which generally shows a low cycleability due to the dissolution of Mn³⁺ in the electrolyte. Nanosized LiMn₂O₄ with well-defined polyhedral morphology was synthesized using a plasma-enhanced low temperature solid-state strategy (PE-LTSS) (see figure 10) [84]. Starting materials (MnO₂ and LiOH) were initially ball milled and later injected into the PECVD tube furnace that was set at 500 °C with an applied RF power of 200 W. Oxygen was flown into the furnace for 30 min to complete the reaction. The electrochemical performance of LiMn₂O₄ was tested on a coin cell against Li metal and LiPF₆ in ethylene carbonate as electrolyte, and compared to a conventional solid-state-prepared LiMn₂O₄. Sample produced by this plasma-enhanced method yielded an initial discharge capacity of 130 mAhg⁻¹ and retention of the starting capacity of 95.5%, while the one prepared by solid-state reaction only achieved 115 mAhg⁻¹ and a capacity retention of 75.6%. Such improvement in the electrochemical performance was attributed to the finer particle size, narrower particle size distribution and controlled morphology.

Zoom In Zoom Out Reset image size

Also, Li₂MnO₃ nanowires were synthesized by first producing MnO₂ nanowires using the solvo-plasma technique [85], followed by a solid-state reaction. The protocol consisted in the deposition of an aqueous slurry of MnO₂ and KCl on a stainless steel substrate. The water was removed from the substrate at 70 °C and exposed to an atmospheric microwave plasma jet [86] for 5 min. To complete the synthesis, the powder was removed from the substrate and alloyed with LiOH at 480 °C. The electrochemical performance of Li₂MnO₃ nanowires presented a pseudoplateau at 4.1 V, corresponding to the Li extraction from spinel LiMn₂O₄. This evidenced a phase transformation of the electrode material upon cycling, as shown in the charge-discharge profile (figure 11). The material was found to have great capacity retention upon cycling at different C rates going from 1 C to 20 C and capacities ranging from 135 mAhg⁻¹ to 110 mAhg⁻¹. The electrochemical performance of the Li₂MnO₃ nanowires represent an improvement when compared to other morphologies [87–89].

Zoom In Zoom Out Reset image size

As mentioned earlier, carbon has proved to enhance the electronic conductivity of the active materials, which are otherwise highly insulating. DC-pulsed plasma treatment of laminated LiMn₂O₄-carbon black composite electrodes improved the stability of the active material upon cycling [90]. Commercial LiMn₂O₄ was mixed with carbon black and PVDF, later laminated on Al foils and dried at 120 °C for 24 h to obtain electrodes $\sim 20$ µm-thick. After drying the electrode, a 400 V pulsed-DC plasma treatment was carried out for different exposure times under an O₂ atmosphere. XRD patterns showed that, after plasma processing, the LiMn₂O₄ crystal structure is kept. Still, the morphology was affected. Untreated LiMn₂O₄ particles showed well-defined grain separations, while treated particles presented a smooth surface where the boundaries between grains disappeared, forming a coating. Electrochemical tests of the two samples showed that plasma-treated LiMn₂O₄ had a capacity retention of 60% upon 40 cycles, while the untreated LiMn₂O₄ only retained 40% of its initial capacity after 40 cycles. Such improvement in the cycleability of the electrode is attributed to a variation in morphology that led to the formation of a protective layer, analogous to conventional carbon or metal oxide coatings.

In a similar report, Wang et al [91] showed that it is possible to deposit a graphite coating onto commercial LiMn₂O₄ using magnetron sputtering. Typical slurry containing the active material, carbon black and PVDF in n-methyl-2-pyrrolidone (NMP) was casted on an Al foil, followed by a drying step. The graphite coating was deposited on the surface of the as-prepared film by magnetron sputtering (figure 12) with different growth times (from 0 to 60 min), aiming to form a graphite layer on the surface that would prevent Mn from dissolving into the electrolyte upon electrochemical cycling. XRD pattern showed no changes in the corresponding spinel peaks, indicating that the crystal structure was preserved during the plasma treatment. Moreover, TEM images allowed identifying and measuring the graphite coating, which was found to have a thickness around 50 nm. Electrochemical performance of the different films prepared at 0, 10, 30 and 60 min treatment time showed that carbon-coated LiMn₂O₄ for 30 min presented the largest improvement in cell polarization and capacity retention. This is attributed to an optimal carbon layer that allowed good electrode wettability and also protected LiMn₂O₄ from HF attack (HF results from the electrolyte decomposition), avoiding changes in the surface structure of the electrode.

Zoom In Zoom Out Reset image size

Direct deposition of active material on the current collector was also performed to obtain binder-free LiMn₂O₄ thin films in a two-step process: (i) sol-gel synthesis of a LiMn₂O₄ thin film electrode and (ii) pulsed-DC plasma treatment [92] under pure O₂ atmosphere to induce the formation of a dense nanocrystalline surface layer that would increase both the stability and the ionic conductivity of the active material. Plasma treatment was carried out for different time periods (0, 5, 30 min) and the resulting electrodes were electrochemically tested at elevated C-rates that went from C/5 to 45 C. Samples treated for 5 and 30 min showed a 65% capacity retention at 10 C and 32% at 35 C, while non-treated sample showed a capacity retention of only 50% and 12%, respectively.

The synthesis of LiFePO₄, usually carried out via solvothermal or the sol-gel methods, has also been addressed by innovative plasma technologies. For example, electric discharge assisted mechanical milling (EDAMM) was used to produce LiFePO₄ [93]. In this milling process, stoichiometric amounts of mixed precursors (Li₂CO₃, FeC₂O₄·2H₂O and (NH₄)H₂PO₄) were fed into a reaction vessel adapted with two curved stainless steel electrodes (figure 13). The pulverization of the powder was obtained through the vibration of one of the electrodes at 10 Hz for different time periods, while an AC power was supplied to generate $0.1{-}1$ kV, 70 Hz-impulses at 100–300 mA. The discharge, in the form of a glow or a spark, is known to accelerate the sintering process [94]. As such, fast synthesis of the olivine was demonstrated; after only 1 min of reaction, the characteristic peaks of LiFePO₄ were already observed in the XRD pattern, together with unreacted precursors. Pure phase was obtained after a 10 min reaction, as shown in figure 14. Although the particle size distribution of the samples was found to be irregular, changing the vibration time changes the morphology, opening the door for deeper studies on the tuning of particle crystallinity, shape and size.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Our research group recently reported the ICP torch synthesis of LiFePO₄ nanospheres of approximately 50–100 nm in diameter, using FePO₄·2H₂O, LiOH and oxalic acid as precursors (conventional sol-gel precursors) [48]. During the synthesis, the sheath gas consisted in a mixture of H₂ and Ar to provide a reducing atmosphere that would prevent iron from oxidizing. In a single step procedure, carbon-coated LiFePO₄ particles were prepared (figure 15(a)). Oxalic acid in the precursor solution acted as a carbon source that led to the formation of a coating onto the nanoparticles, likely improving the electrical conductivity of the material. The possibility of preparing carbon-coated particles in a single step makes this process attractive for scaling up the production of such active materials. In fact, carbon coatings are usually deposited in a second step by the pyrolysis of organic precursors like sucrose, glucose, oxalic acid, ascorbic acid, etc, or by post-synthetic procedures such as chemical vapor deposition from toluene or physical vapor deposition [95, 96]. In the same report, we have detailed the preparation of plasma-sprayed LiFePO₄ coatings (figure 15(b)) using a plasma torch equipped with a supersonic nozzle. Schematic view of the inductively coupled thermal plasma reactor used for the synthesis and deposition of LiFePO₄ was presented in figure 3. Deposition was performed on a 50 µm-thick Ni foil and using the same solution of precursors as for the powdered LiFePO₄. Conventionally-casted electrodes containing the plasma-prepared LiFePO₄ powder were compared to the binder-free plasma-sprayed electrodes by means of cyclic voltammetry (figures 15(c) and (d)). Characteristic positive anodic peak (charging phase) around 3.5 V and a negative cathodic peak (discharging phase) between 3.2 and 3.4 V were obtained in both cases. The best reversibility was observed for the plasma-obtained LiFePO₄ powder.

Zoom In Zoom Out Reset image size

Another approach for the synthesis of LiFePO₄ has been reported by Dobbelaere et al [97]. Plasma-enhanced atomic layer deposition was performed on different 3D platinum-coated silicon substrates to produce micropillar electrodes. The synthesis was carried out by the sequential exposure to trimethyl phosphate (TMP), O₂ and tert-butylferrocene (TBF) inside a plasma, as shown in figure 16(a). The first two compounds created a phosphoric acid-like precursor species on the surface of the substrate, while the latter completed the formation of the olivine. During the process, the plasma power was set to 200 W and 300 W for TMP and O₂ plasmas, respectively. A schematic representation of the proposed reaction path is shown in figure 16(b). The authors developed a comprehensive study on the reaction path and its dependence on the temperature used; the optimal deposition was obtained at 300 °C. The electrochemical performance of the prepared micropillar was compared to a planar electrode at different C-rates (see figure 16(c)). The capacity of the 3D electrode was found to be 20 times higher than that of a planar electrode, which was attributed to the increase in surface area for the micropillar electrodes and, in both cases, capacity faded almost linearly with the increase in C-rate.

Zoom In Zoom Out Reset image size

Silicon nanoparticles based anodes have been synthesized by plasma processes (PS-PVD, RF-ICP, DC arc plasma jet, PECVD, etc) over the last few years, as reviewed by Doğan and Van De Sanden [21]. Silicon anodes have a theoretical capacity of 4200 mAhg⁻¹, an order of magnitude larger than that of graphite-based anodes. Still, this high capacity cannot be maintained for more than a few tens of charge/discharge cycles, as the lithiation process of silicon comes with a detrimentally large volume expansion (~300%) that results in mechanical damages and capacity losses. This volume expansion can be accommodated by controlling the size and morphology of the nanostructured silicon, which can be achieved by plasma processes.

Promising Si negative electrode material has been recently prepared by a DC discharge between a Si bar and a Pt wire through a solution (electrolyte) [101]. A study of different electrolytes and different voltages was carried and, in most cases, spherical particles were obtained, as shown in figure 17. When the applied voltage was low (160 V for 0.1 M HCl and K₂CO₃), particle sizes smaller than 1 µm were obtained; for voltages above 200 V, the particles produced had an average diameter over 1 µm. In the presence of acidic electrolyte, crystalline SiO₂ was obtained, while amorphous SiO₂ resulted from basic electrolyte solutions. This synthetic method allows customization of particle size for anode applications.

Zoom In Zoom Out Reset image size

In the case of the work reported by Lamontagne et al [102], Si nanowires were synthesized by the carboreduction of silica fume in the presence of Al, Ni, Fe and Y₂O₃ as catalysts, using an ICP thermal plasma reactor similar to that shown in figure 3. The composition of the sheath gas went from He to a mixture of Ar and H₂. In this work, it was shown that samples prepared without catalyst contained less Si and no Si nanowires were formed, while samples prepared using a metal catalyst showed a higher efficiency in the carboreduction of SiO₂ to Si and in the growth of Si nanowires, as shown in figure 18. The hypothesized growth path of Si nanowires thus requires silicon to be solubilized in the catalysts in the high temperature zone and then precipitated as its solubility diminishes in the quenching zone. The nature of the catalyst also modified the growth of the nanowires: Al formed packed Si nanowires, whereas Ni and Fe favored an individual growth (see figure 18).

Zoom In Zoom Out Reset image size

A DC-ICP hybrid torch operating in PS-PVD mode (8 kW DC and 90 kW ICP) was employed by Kambara et al to produce Si, SiC and carbon-coated Si (core–shell) aggregates of 200–300 nm in size made of 20–40 nm particles at a throughput  >350 g h⁻¹ [50]. The feedstock consisted in metallurgical grade silicon powders with a mean diameter of 19 µm and CH₄ gas was co-injected as a carbon precursor. The carbon-to-silicon ratio was found to have an incidence on the capacity, as shown in figure 19. The capacity fading of the electrode was found to increase as the C/Si ratio increased due to the irreversible formation of SiC.

Zoom In Zoom Out Reset image size

Composite metal-carbon materials have also been considered. For example, spherical Si particles can be obtained by RF thermal plasma in one step by injecting Si powder with Ar as carried gas [103]. A representation of the Si synthesis is shown in figure 20, together with a secondary step to disperse the prepared powder in a carbon matrix. Si nanospheres (SiNS) dispersed in carbon by ball milling of a mixture Si/C (4:1) showed an outstanding electrochemical performance, with an initial capacity of 2388 mAhg⁻¹. Moreover, after 100 cycles, the remaining capacity was 778 mAhg⁻¹, two times the theoretical capacity of graphite.

Zoom In Zoom Out Reset image size

Si nanoparticles [104] were prepared in a matrix of carbon nanotubes using an arc plasma. The authors used two powder feeders to separately inject the precursors: (i) Si was injected directly into the plasma torch, where particles were evaporated to form nanoparticles, and (ii) commercial MWCNT were injected in a cooler downstream region to preserve their structural integrity. After treatment, the Si-MWCNT nanocomposite collected in the reactor chamber was found to have Si nanospheres uniformly distributed along the column of MWCNTs, which resulted in a suppression of the detrimental Si volume expansion under electrochemical cycling and in a discharge capacity of 1447 mAhg⁻¹. Composite Si/graphene nanosheets (Si/GNs) with a hierarchical structure, prepared using a plasma-assisted milling process similar to EDAMM, also showed interesting cycling capabilities, maintaining a capacity of 1000 mAhg⁻¹ over 350 cycles (figure 21) [105]. For a capacity of 1000 mAhg⁻¹, the electrode was cycled at different C-rates and showed good capacity retention, mainly attributed to the small particle size and hierarchical morphology accommodating stress/strain during the lithiation/delithiation process and buffering the volume changes upon cycling.

Zoom In Zoom Out Reset image size

Sn/SnO₂/MWCNT nanocomposites have been prepared by means of thermal evaporation of metallic Sn onto MWCNT buckypaper, followed by a subsequent RF plasma oxidation process aiming at reducing the SnO₂ volume changes associated to the lithiation/delithiation process [6]. Plasma oxidation of the Sn films was conducted using a mixture of O₂ and Ar in a 1:1 ratio, a total chamber pressure of 1.6 Pa, a RF power of 80 W and a duration of 45 min. The electrochemical performance of the Sn/SnO₂/MWCNT free-standing electrode was tested during charge-discharge cycles, which showed a maximum discharge capacity or 1544 mAhg⁻¹. Upon 10 cycles, the capacity retention was 51% (788 mAhg⁻¹), while it decreased to 24% (374 mAhg⁻¹) after 100 cycles.

A similar approach was taken by Thomas et al [106]. in the synthesis of SnO₂-graphene nanocomposites. Initially, highly porous graphene nanosheets were synthesized by decomposition of acetylene in a microwave PECVD process at 700 °C for 30 min, followed by the sputtering of a 3 nm-thick layer of gold over the vertically-deposited graphene nanosheets (GNS). Annealing of the gold-coated GNS led to the formation of 5 nm gold nanoparticles that served as catalyst for the formation of SnO₂ nanowires. SnO₂ was deposited using reactive electron beam evaporation of Sn in an oxygen-rich atmosphere onto the substrate, which was held at 620 °C. Figure 22 shows a series of SEM images depicting the different stages of the preparation of the nanocomposite. The electrochemical performance of the as-prepared electrode was first evaluated by cyclic voltammetry between 0.01–2.0 V, where the characteristic graphene peaks were identified during cycling. Such observation indicated that Li⁺ intercalated into the GNS. The first discharge and charge cycles showed capacities of 1335 and 4930 mAhg⁻¹, respectively, exceeding theoretical values. Such high capacities were attributed to the contribution of SnO₂ nanowires and of the porous GNS. However, large capacity fading was observed in the second cycle, most likely due to the irreversible decomposition of SnO₂ and of the electrolyte. Subsequent cycling resulted in roughly 70% of the initial charge-discharge capacities, as can be seen in figure 23.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Another strategy to prevent Sn to suffer a dramatic volume change during its charge-discharge process is to alloy Sn with transition metals. The insertion of an inactive element would act as a support to avoid losing the microstructure. DC arc discharge has been used to synthesize binary alloys Sn–M (M  =  Fe, Al, Ni) using a mixture of powdered Sn and M that was compressed and used as anode [107]. Sn–M was evaporated by the arc discharge and condensed after a temperature drop. The alloys prepared showed a typical spherical morphology with a metal core and an oxide shell, as observed with high resolution TEM. The particle size distribution was in the range of 50–150 nm. Sn–Fe was found to have lower charge and discharge capacities (256 mAhg⁻¹ and 338 mAhg⁻¹, respectively). However, great cycleability was obtained after 20 cycles with a capacity fading of only 11%.

The synthesis of titanates by means of pulsed-laser deposition [108], APPJ [109] and ICP [110, 111] was also considered. Starting with solid Li₂CO₃ and TiO₂ precursors that were axially injected into a 40 kW plasma discharge, Quesnel et al obtained nanoparticles, nanowires and nanoplatelets by changing two main experimental parameters: the Li:Ti ratio and the sheath gas composition (mixture or Ar and H₂) [111]. It was found that the presence of H₂ promoted the formation of nanosized materials and that the production yield could be increased with the addition of seeding material (in this case Li₄Ti₅O₁₂).

Tummala et al built on the binder-free plasma spray process of LiCoO₂ coatings described above and reported the synthesis of a flexible Co₃O₄ electrode [112]. A cobalt acetate tetrahydrate solution was radially injected into the plasma plume and sprayed onto a stainless steel substrate, producing a flexible Co₃O₄ coating, as shown in figure 24. The as-prepared electrode was electrochemically tested in a coin cell versus Li/Li⁺ at C/10, which delivered an initial discharge capacity of $\sim 1000$ mAhg⁻¹. Subsequent charge and discharge cycles resulted in capacities close to the theoretical value (890 mAhg⁻¹). Although a large capacity is obtained, considerable capacity fading is observed when increasing the cycling rate to C/2, 1.25 C and 16.5 C, most likely due to the lack of a carbon additive that would increase the electronic conductivity and alleviate the volumetric expansion undergone with continuous cycling. It is expected that, by adding a carbon source to the precursor mixture or in a second step, the electrochemical performance of Co₃O₄ could be improved.

Zoom In Zoom Out Reset image size