*4.2. Atomic Layer Deposition of Conventional Lithium-Ion Battery Materials*

In this section, a few examples of the more conventional lithium-ion battery materials deposited by ALD are introduced. Examples of binary oxide electrodes (specifically V2O5 and TiO2) and lithium containing materials are presented. Lithium containing materials are quite a new addition to the ALD materials toolbox—the first paper on the subject was published in 2009 [70]. Since then, this area of research has expanded very rapidly. Due to the mobility and reactivity of Li<sup>+</sup> ions, ALD of lithium containing materials has additional process development issues in comparison to most other ALD processes and some of these issues are discussed in the following subsections. In this section emphasis is given to potential solid electrolyte materials, as these are generally considered the most difficult materials to deposit. For a more thorough review of this subject, both review articles and books are available [71–79]. In recent years, ALD has also been studied extensively as a method to modify the interfaces between the electrodes and the electrolyte by forming an artificial SEI-layer [75,80,81]. ALD Al2O3 is generally used for this application and it has been found that a few ALD cycles can improve the cycling capability and capacity retention of the electrodes [80,81]. These results will not be discussed further here but the reader is advised that a lot of literature on this subject is available [80–83]. Some examples of ALD-made metal fluorides as artificial SEI layers will be shortly mentioned in Section 5.

#### 4.2.1. Cathodes

Table 4 includes examples of some of the conventional lithium-ion battery cathode materials deposited by ALD. Both binary and ternary oxides and phosphates have been deposited for conventional batteries. Li2S has been envisioned as a cathode for lithium-sulphur batteries [84].

**Table 4.** Examples of conventional lithium-ion battery cathode materials deposited by ALD. Abbreviations used: O<sup>i</sup> Pr = *iso*-propoxide, O<sup>t</sup> Bu = *tert*-butoxide, Cp = cyclopentadienyl, TMPO = trimethyl phosphate, thd = 2,2,6,6-tetramethyl-3,5-heptanedionato.


Vanadium oxide was one of the earliest materials to be deposited by ALD for lithium-ion batteries [85,86,92]. Vanadyl tris-*iso*-propoxide with either water or ozone as co-reactant have been used as precursors for this material. Using water produces amorphous films, while with ozone crystalline films can be obtained. Crystalline and amorphous films produce different capacities depending on the extent of lithium intercalation, with crystalline films having higher capacities when 1 or 2 Li-ions intercalate per one V2O5 unit [86]. A surprisingly high capacity of 455 mAh/g has been reported for 200 nm of amorphous V2O5 between 1.5 and 4.0 V (Li/Li+) [85]. This high value is related to the large potential range used for cycling, resulting in 3 Li-ions intercalating per one V2O5 [85]. For a thicker amorphous film of 450 nm, a capacity of 275 mAh/g was obtained in the same range. Both the thicker and the thinner films showed reasonable capacity retention after 90 cycles. For crystalline films, capacities of 127–142 mAh/g have been obtained in the potential range 2.6–4.0 (1 Li per V2O5) [86]. Between 1.5 and 4.0 V (3 Li per V2O5), a capacity of 440 mAh/g is obtainable but the capacity degrades to 389 mAh/g already in the second cycle.

As already discussed, lithium cobalt oxide LiCoO2 is currently the most often used cathode material in lithium-ion batteries. Despite this, only two reports from one group on the deposition of LiCoO2 by ALD can be found in the literature [87,88]. It appears that the challenges in cobalt oxide deposition have had an effect on the research of the lithiated material. The reported LiCoO2 process makes use of oxygen plasma combined with CoCp2 (cobaltocene) and LiO<sup>t</sup> Bu (lithium *tert*-butoxide). The deposition supercycle consists of Co3O4 and Li2CO3 subcycles and the effect of different pulsing ratios on film properties was studied. The process showed saturation with both metal precursors with Li:Co = 1:1 pulsing ratio and the film thickness increased fairly linearly with the number of cycles when using a 1:4 pulsing ratio. After annealing the films consisted of the hexagonal phase of LiCoO2 according to both Raman and GIXRD (grazing incidence X-ray diffraction) measurements. Electrochemical characterization revealed that a 12% capacity loss was evident between charge and discharge cycles. For a film deposited with a 1:4 pulsing ratio, the capacity was only about 60% of the theoretical value. For a 1:2 ratio film, the capacity was even lower, which might be explained by the higher impurity contents in this film. With the 1:4 pulsing ratio, the composition of the films was Li1.2CoO3.5, as determined by elastic backscattering [88].

The potential cathode material lithium iron phosphate, LiFePO4, has also been the subject of ALD studies [89,93] The material has been deposited at 300 ◦C on silicon substrates using ferrocene and ozone as precursors for the Fe2O3 subcycle, trimethylphosphate (TMPO) and water for PO*<sup>x</sup>* and lithium *tert*-butoxide and water for the Li2O/LiOH subcycle [89] Iron oxide and the phosphate were pulsed sequentially for five cycles, after which one cycle of Li2O/LiOH was applied. The resulting films were amorphous and showed a linear increase in thickness as a function of deposited supercycles. The material could also be deposited onto carbon nanotubes (CNTs) [87] The CNT-based films were amorphous but crystallization to orthorhombic LiFePO4 was observed after annealing in argon at 700 ◦C for 5 h. The Fe:P ratio in the annealed film was 0.9, as determined by EDX (energy-dispersive X-ray spectroscopy). Unfortunately, no compositional information on Li content was given. The LiFePO4 film deposited onto CNTs showed good electrical performance, with clear redox peaks in a cyclic voltammetry curve at 3.5 V and 3.3 V (vs. Li/Li+) and a discharge capacity of 150 mAh/g at 0.1 C [89]. Encouragingly, the material could maintain a discharge capacity of 120 mAh/g at 1 C even after 2000 cycles.

LiFePO4 has also been deposited using metal-thd complexes [93]. Pulsing Lithd (lithium 2,2,6,6-tetramethyl-3,5-heptanedionate) and ozone between subcycles of Fe(thd)3 + O3 and TMPO + O3 + H2O resulted in stoichiometric LiFePO4 when the fraction of Li2CO3 subcycles was 37.5%. The as-deposited films were amorphous but could be crystallized in 10/90 H2/Ar atmosphere at 500 ◦C. These films were reported to show poor electrical conductivity, as expected with this material [19], however very little additional information was given. It should be noted that the same research group has also published an ALD process for the de-lithiated cathode material FePO4 [94]. They reported an initial electrochemical capacity of 159 mAh/g for the as-deposited, amorphous 46 nm thick FePO4 film. The capacity increased to 175 mAh/g after 230 charge-discharge cycles and after 600 cycles the capacity was still 165 mAh/g.

Lithium manganese spinel Li*x*Mn2O4 is an interesting cathode material for lithium-ion batteries due to its low volume change during (de)lithiation, high voltage and environmentally benign elements. The material has been deposited by ALD by Miikkulainen et al. using various methods [90]. Firstly, Mn(thd)3 and ozone were used as precursors for manganese oxide and this process was combined with the Lithd + O3 process for lithium incorporation. Interestingly, even with exceedingly small numbers of Li2CO3 subcycles, high Li+ incorporation was achieved, with only a 5% Li2CO3 pulsing leading to a Li:Mn ratio of 1:1. This was in fact the maximum content of lithium obtained: using larger numbers of Li2CO3 subcycles led to a decrease in uniformity. To achieve the stoichiometric lithium level for LiMn2O4, 1% of Li2CO3 pulsing was sufficient. All the films showed the crystalline spinel phase as-deposited, with MnO2 impurities present in the films with the lowest lithium concentrations. Crystalline spinel LiMn2O4 was also obtained by using LiO<sup>t</sup> Bu and water as precursors, however little else was reported on this process.

The lithium manganese spinel process is unique in that while the growth rate of the films stays rather constant at below 0.3 Å/cycle, the lithium content increases very rapidly and reaches a high value with very small Li-subcycle numbers [90]. This indicates that the mechanism of this process differs significantly from conventional ternary ALD processes. Another clue about the mechanism was given by ToF-ERDA (time-of-flight elastic recoil detection analysis) elemental depth profiles which showed uniform film composition, albeit with a lithium deficiency on the film surface. To achieve such high lithium concentrations, either more than one monolayer should be deposited in one subcycle, or the growth should include a bulk component in addition to the normal surface reactions. Multilayer growth could be assumed to lead to lithium excess on the film surface, since lithium carbonate was always the last material deposited. Therefore, the bulk must be playing a role in the deposition process. Miikkulainen et al. postulated that the reduction needed for manganese to change from +IV in MnO2 to +III/+IV in LiMn2O4 takes place during the Lithd pulse, which affects not only the surface but also deeper parts of the film [90]. The following ozone pulse removes organic residues from the surface and re-oxidizes the topmost manganese ions on the surface. This reaction then drives lithium ions deeper into the film, resulting in a uniform elemental distribution with a slightly lithium deficient surface.

Miikkulainen et al. continued their studies on LiMn2O4, using both LiO<sup>t</sup> Bu and water and Lithd and ozone exposures on MnO2 at 225 ◦C [90]. Interestingly, 110 nm of manganese oxide could be converted to the spinel phase with only 100 cycles of the lithium carbonate process applied on top of the film. The carbonate was not present in the X-ray diffractogram. Lithiation was achieved to some extent also without ozone pulses. The manganese oxide films lithiated with LiO<sup>t</sup> Bu and water in a similar manner showed the best electrochemical storage properties, with a capacity of 230 mAh/g at 50 μA. The capacity retention was very good up to 550 cycles at 200 μA. Similar to LiMn2O4, vanadium oxide V2O5 could also be lithiated by pulsing either LiOt Bu and water or Lithd and ozone on top of the oxide film [90]. Using the Lithd and ozone precursors, lithium contents as high as 15 at% were obtained with only 100 cycles of the Li2CO3 process applied on the 200 nm oxide film.

Lithium sulphide, Li2S, is an attractive cathode material for high capacity lithium-sulphur batteries. It has recently been deposited by ALD [91] and requires inert atmosphere during sample handling to prevent reactions with ambient air, in a similar way to pure Li2O [95]. Li2S has been deposited using LiO<sup>t</sup> Bu and hydrogen sulphide between 150 and 300 ◦C. Unlike most lithium containing processes, this one produced a constant growth rate over the whole deposition temperature range studied. The refractive index of the films was much lower than the value for bulk crystalline Li2S, indicating a lower density of the films. The films were amorphous and could not be crystallized with annealing in inert atmosphere. Both XRF (X-ray fluorescence) and XPS (X-ray photoelectron spectroscopy) gave a Li:S ratio of 2:1, with no carbon contamination in the Li2S layer. Thus, the reaction between the precursors was very efficient. The Li2S films produced high capacities of 800 mAh/g when deposited onto mesocarbon microbeans and a somewhat lower capacity of 500 mAh/g when deposited directly onto a 2D Cu current collector. In both cases the Coulombic efficiency was ~100%, indicating that the material could indeed be used as a cathode in lithium-sulphur batteries. However, film thickness had a large effect on the capacity, with thicker films producing smaller capacities per gram, as Li2S is rather insulating. In addition, reactions with the copper current collector affected the charge-discharge profiles, indicating the formation of Cu*x*S.

#### 4.2.2. Anodes

Table 5 presents examples of lithium-ion battery anode materials deposited by ALD. The selection of materials is quite a bit more limited than in the cathode case, most likely illustrating the consensus that improvements in battery energy density are more easily obtained with improved cathode materials. ALD-made TiO2 has been studied as an anode material mostly in various 3D-constructions, illustrating the conformal coating ability of ALD [78,96–98]. Using 3D-structures the areal capacity of the titania anode can be greatly improved [96,97] Generally, TiO2 can intercalate 0.5 Li, resulting in a capacity of 170 mAh/g [97] However, using nanomaterials more lithium can be intercalated and capacities of 330 mAh/g have been obtained with anatase nanotubes with a wall thickness of 5 nm [98]. These nanotubes also showed excellent capacity retention.

**Table 5.** Examples of conventional lithium-ion battery anode materials deposited by ALD. Abbreviations used: Oi Pr = *iso*-propoxide, O<sup>t</sup> Bu = *tert*-butoxide, thd = 2,2,6,6-tetramethyl-3,5 heptanedionato, TPA = terephthalic acid, LiTP = lithium terephthalate.


Attempts on the deposition of lithium titanate spinel, Li4Ti5O12, have been made using both titanium tetrachloride TiCl4 [99] and titanium tetra-*iso*-propoxide Ti(O<sup>i</sup> Pr)4 as precursors [99–101]. In both cases, LiOt Bu was used as the lithium source and water was used as the oxygen source. Titanate

films deposited using TiCl4 reacted rapidly in air [99]. The films were amorphous as determined with X-ray diffraction and showed only very small amounts of lithium in ERDA measurements. In contrast, when using Ti(Oi Pr)4 as the titanium source and applying a long pulse time for this precursor, uniform titanate films with higher lithium contents could be deposited [99]. These films also reacted with air, however the reaction was much slower than when using TiCl4 as the titanium precursor. The growth rate of the films did not depend much on the pulsing ratio of the two metal precursors, being approximately 0.7 Å/cycle at 225 ◦C [99]. In another report using the same precursors, the growth rate was said to be slightly different at 0.6 Å/cycle at 250 ◦C [100]. For the process at 225 ◦C, ERDA measurements revealed that the lithium content of the films could be routinely tuned over a wide range by changing the metal pulsing ratio [99]. For example, with 33% lithium cycles the film stoichiometry was Li1.19TiO2.48 and the carbon and hydrogen impurity contents were low. XPS and ERDA revealed that in this material lithium was enriched on the film surface, most likely forming a carbonate layer: carbonate peaks were visible both in FTIR (Fourier transform infrared spectroscopy) and XPS [99,101]. Despite the carbonate formation, the films showed the Li4Ti5O12 spinel phase in XRD measurements also in the as-deposited state. The crystallinity could be improved by annealing in nitrogen at 640–700 ◦C. The annealed titanate films showed electrochemical activity but the capacity remained low at 40 mAh/g [101]. However, this low value might be related to uncertainties in the calculation of film mass. For the film deposited at 250 ◦C, the Li:Ti ratio was reported as 2:1 with 44% lithium pulsing [100], which could indicate Li2TiO3 formation—a well-known impurity phase for Li4Ti5O12 [103,104]. After annealing in argon at 850 and 950 ◦C these films showed XRD peaks belonging to Li4Ti5O12 [100].

In addition to purely inorganic materials, ALD can also be used to deposit hybrid materials using organic molecules as the second precursor [105]. Lithium terephthalate (LiTP) has been deposited using Lithd and terephthalic acid as precursors between 200 and 280 ◦C [102]. This material has been proposed as a possible Li-ion battery anode due to its high theoretical capacity of 300 mAh/g and a low potential of 0.8 V (vs. Li+/Li) [106]. The ALD process for LiTP showed saturation but no ALD window or constant growth rate as a function of the number of cycles [102]. The changing growth rate, accompanied by changes in the film density, could be related to the island growth mechanism of the film. The films were crystalline as deposited, which is unusual for ALD hybrid films. The films were electrochemically active and showed high rate capabilities with good capacity retention. The electrochemical properties of the films could further be enhanced by a protective LiPON layer on top of the electrode. The higher than theoretical capacity of 350 mAh/g is partly explained by difficulties in determining the electrode film thickness in electrochemical analyses. Recently, this anode material was combined with an organic cathode material dilithium-1,4-benzenediolate to produce a functioning atomic layer/molecular layer deposited thin film battery [107].
