Lithium ion batteries are one of the most promising energy storage methods in modern society, and significant research efforts have been focused on developing cathode materials with high electrochemical performance [1][2][3][4]. Li 2 MnO 3 is a promising cathode material for high energy density applications due to its high theoretical capacity [5][6][7]. However, Li 2 MnO 3 has a very low electrical conductivity (~10 À9 S cm À1 ) and it limits the power density of this cathode material. Most of the current attempts to improve the electrochemical performance are not satisfying [6][7][8][9][10][11]. The most commonly applied approach was to use carbon coating. For examples, Xiong et al. applied additional 10 wt% of carbon coating on the Li 2 MnO 3 particles on top of 10 wt% of carbon black but it only delivered 52% capacity retention when the testing rate was increased from 0.09C to 3.40C [12]. Another approach is to integrate layered Li 2 MnO 3 phase with spinel LiMn 2 O 4 phase. He et al. utilized such integration approach, but the material only showed 45% capacity retention from 0.1C to 10C [11]. Another common technique is to reduce particle dimensions. Vendra et al. synthesized Li 2 MnO 3 nanowires with an average diameter of 50 nm and length of 1 μm and demonstrated superior rate performance [9]. Taminato et al. also demonstrated high capacity under high current densities with thin film Li 2 MnO 3 at 12.6 nm [13]. Despite the apparent effectiveness of the approach, the small dimension could limit the energy density of the cathodes.
Nanocomposite materials contain multiphase building blocks with dimensions in nanometer range and exhibit improved physical properties. As a unique type of nanocomposites, vertically aligned nanocomposite (VAN) introduces vertical secondary phase domains in the matrix and presents a wide range of applications in tuning electrical transport properties [14][15][16][17], magnetic properties [15,18], and optical properties [19] due to the unique coexisting in-plane and out-of-plane strain and interface coupling. Furthermore, VAN is known to accommodate strain through vertically coupled domains [20][21][22][23]. Therefore, VAN-based thin film electrodes could present great promise in achieving high conductivity and minimizing film delamination at film-to-substrate interfaces caused by volume change.
In this work, for the first time, a novel design of Au nanopillars incorporated into Li 2 MnO 3 thin films through an Oblique Angle Deposition (OAD) method in pulsed laser deposition (PLD) has been achieved. PLD offers advantage in precise stoichiometry control of film composition which is ideal for complex oxides and high temperature ceramics [24]. The OAD technique is introduced in this work because a conventional co-growth of oxide-metal by PLD generally leads to metal particles in oxide matrix [17,19,24,26] with few cases growing as metal pillar in matrix [17,19,25,26]. OAD, on the other hand, introduces the shadowing effect by creating certain angle between the incoming source flux and the substrate, and the additional driving force can help achieve various film morphologies [27][28][29][30]. The schematic drawing of the OAD design in PLD is shown in Fig. 1b. Different from the previously reported particles-in-matrix design [17] (illustrated in Fig. 1a), the proposed Li 2 MnO 3 -Au pillar configuration can provide more effective, continuous pathways for electrical and ionic transport across the entire cathode, and the more effective oxide-metal interfaces can provide mechanical integrity. Detailed microstructural analysis was conducted and assembled coin cell batteries using the unique Li 2 MnO 3 -Au nanocomposite cathode were tested to explore the effects of the tilted Au nanopillars. Anisotropic optical properties originated from Au nanopillars were confirmed by optical measurement. Along with the nanoindentation experiments for exploring the mechanical integrity of the Li 2 MnO 3 -Au nanocomposite cathodes, the work presents a promising approach for future incorporation of nanocomposite cathode designs for mechanically strong and high performance thin film batteries.
Microstructural characterization of the Li 2 MnO 3 -Au composite thin film was performed by X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and highangle annular dark-field scanning transmission electron microscopy (HAADF-STEM). The successful growth of the Au nanopillars is clearly demonstrated in the TEM/STEM images of Li 2 MnO 3 -Au on the α-Al 2 O 3 sample (Fig. 2). Specifically, Fig. 2a shows the low magnification TEM image of the sample. The Au nanopillars are clearly tilted 19 � away from the out-of-plane direction and they are roughly 50 nm separated from each other with the average diameter of ~6 nm as shown in Fig. 2a anda1. It is interesting to note that there are long nanopillars across the entire film thickness and short ones embedded in the film. The overall aspect ratio varies from 40 to 4, estimated from Fig. 2a. Fig. 2b shows the plan-view image of the sample, which further demonstrates the uniform distribution of Au nanopillars in-plane. Such tilted nanopillars are formed due to the OAD growth illustrated in Fig. 1. Due to the shadowing effects during OAD, the Au nanopillars are tilted 19 � away from the incident angle. High resolution TEM image in Fig. 2a1 is presented to further illustrate the high quality epitaxial growth of such Au nanopillars in Li 2 MnO 3 matrix. In cross-sectional STEM (Fig. 2d), Au-map (2d1) and Mn-map (2d2), there are some short Au nanorods near the film-substrate interface region suggesting the initial nucleation and seeding process of the Au nanopillars. In addition, there are some discontinuity between some of the nanopillars along the pillar growth direction. First, the small Au nanorods near the film-substrate interface are Au nucleates formed at the beginning stage of the growth but terminated to grow later due to shadowing effect [31]. Second, the discontinuity along nanopillar growth direction is due to the adatom diffusion. In general, OAD could introduce a large amount of porosity due to the limited adatom diffusion and shadowing effect, which commonly result in columnar-like morphology [32]. Several common types of morphologies formed with mixed sources are multilayered columns [33], zigzag-shaped multilayer columns [34], top/sides coated columns [35][36][37],tilted columns with porosity [38], and laterally assembled columns [39]. However, in this case, Li 2 MnO 3 grows continuously as the matrix without obvious porosity while the Au grows as embedded tilted pillars. The Au adatoms could act as "adatoms diffusion facilitators" which facilitates the diffusion of Li .762 Å on Au-SS, which is equal to a compressive strain of 0.14% and tensile strain 0.25% in c-axis direction compared to reported bulk value, respectively [40]. The different strain states could be from different substrates. Besides, the Li 2 MnO 3 peaks have overall lower intensity on Au-SS than that on Al 2 O 3 , which also matches our previous results [17]. As the Au existence in Al 2 O 3 substrate was proved through TEM/STEM techniques, SEM images were taken for Li 2 MnO 3- -Au on Au-SS substrate to prove the existence of Au. In Fig. S1b, the plan-view SEM image confirms the typical layered oxide growth morphology of Li 2 MnO 3 with some embedded Au secondary phase, which are Au pillars based on the detailed information from both cross-sectional and plan-view images. We believe that the structural relationship of the sample on Au-SS substrate is very similar to that on Al 2 O 3 substrate which is consistent with previous reports [17,41,42].
The electrochemical performance of the Li 2 MnO 3 -Au on Au-SS substrate has been evaluated in 2032R coin cells. The cycling behavior is better than thin film Li 2 MnO 3 cathodes prepared by PLD and with thickness 6-20 times thicker than the reported work [13,43]. Fig. 3a shows the cycling performance of the thin film battery, and it shows a 35.78 μAh cm À2 μm À1 first cycle discharge capacity from constant current discharging step and total discharge capacity of 41.2 μAh cm À2 μm À1 . It is worth noting that the cell exhibits 62.32 μAh cm À2 μm À1 constant current discharge capacity and 71.64 μAh cm À2 μm À1 total capacity at 100th cycle, which is roughly a 74% increase compared to the first cycle. This phenomenon was reported in literature as gradual phase transformation upon cycling [9]. However, the first cycle shows an unexpected large total charge capacity of 445.27 μAh cm À2 μm À1 , which can be translated to 1196.96 mAh⋅g À1 and it exceeds the theoretical capacity 459 mAh⋅g À1 for Li 2 MnO 3 . Similar phenomenon was observed in pure Li 2 MnO 3 , which is shown in Fig. S2a. However, this phenomenon is much more significant in pure Li 2 MnO 3 and the cell failed at 10th cycle. This can be the proof of the increased film conductivity as the first overlarge charge capacity were also reported in other thin film work and were previously reported to be attributed to electrolyte decomposition [44,45].
Two plateaus can be observed in the 1st cycle charge curve but not in the 100th cycle charge curve. The plateau that appears around 3.75 V might be attributed to the diffusion limit of the lithium ions, representing the step which lithium ions overcome the energy barrier and be extracted from the cathode. It shows the slow activation nature for thin film batteries and only appears in the first charge cycle. Another plateau around 4.2 V can be from the electrolyte decomposition and the SEI formation because it gradually fades upon cycling. Besides, it is observed that the capacity increase starts from the 60th cycle. The cycle was performed after a cyclic voltammetry measurement at 59th cycle with sweep rate of 10 μV/s as shown in Fig. 3b. This might be due to the deep reaction depth from the very slow scan rate, which can be converted to a very slow C-rate of 0.01C compared to the 0.5C of regular cycles. This can explain why the capacity starts fading after the 60th cycle but increased again at 80th cycle and continues to fade until the 100th cycle, as the deep cycle increased the depth of reaction and thus the amount of material that undergoes phase transformation. In addition, the extra capacity can be ascribed from the extended plateau around 3V, which is the lithium intercalation at 16c octahedral sites of the spinel phase, again, proves the abovementioned statement. It was previously reported that dislocation can form during cycling, which causes transport of trapped oxidized O 2À species to the surface and release O 2 [46]. Therefore, the improved cycling stability could be attributed to the Li 2 MnO 3 -Au interfaces as it can act as defect sinks to absorb defects during cycling [47], which was not observed in the case of pure Li 2 MnO 3 .
Cyclic voltammetry (CV) measurement was conducted to study the redox reaction in the nanocomposite thin film electrode. Fig. 3b demonstrates the CV results at the 9th, 59th, and 79th cycle. The CV shows typical redox reactions of spinel type structure, where the redox pair around 3V is the lithium insertion in empty 16c octahedral sites and the peaks around 4V is that in empty 8a tetrahedral sites. Similar to the cycling test, the unexpected large anodic peak and the abnormal cathodic peak near 4.0 V confirms the existence of electrolyte decomposition. Similar behavior has been observed in pure Li 2 MnO 3 , as shown in Fig. S2b, therefore the feature could be from the insulating nature of Li 2 MnO 3 , especially in the dense thin film morphology [48]. However, while the pure Li 2 MnO 3 failed after the 9th cycle, the Li 2 MnO 3 -Au composite can be cycled for at least 100 cycles and the electrochemical performance increases upon cycling. It indicates that the Au pillars increase the conductivity of the overall composite film and the cathode will be gradually activated by cycles. The CV measurement at different sweep rates of 10 μV/s, 20 μV/s, 50 μV/s, 75 μV/s, 100 μV/s, 200 μV/s is shown in Fig. S2c. There are two sets of redox reaction pairs appearing in the curves. The redox couples at 4.04V/3.95V and 3.04V/2.88V are the Li intercalation at 8a tetrahedral and 16c octahedral interstitial sites, respectively, and there are also some shoulder peaks next to the two main redox couples, which is typical in spinel type Mn-based materials [49]. Furthermore, the oxidation peaks at 4.2 V are irrelevant to the sweep rates, which are the sign of electrolyte decomposition due to slow kinetics abovementioned, and the higher peak intensity at slower sweep rate suggests the slow reaction constant. However, the reaction is suppressed compared to pure Li 2 MnO 3 , indicating the kinetics improvement from Au nanopillars. It is expected that such a dense structure could hinder the electrolyte penetration in the cathode and thus introducing porosity in the cathode could further improve the performance.
The rate performance at the current density of 5 μA cm À2 to 80 μA cm À2 is shown in Fig. 3c, and the C-rate values are estimated and labeled at each corresponding current density. The Li 2 MnO 3 -Au possesses excellent rate performance with 61% capacity retention at 80 μA cm À2 (~14.8C) compared to 5 μA cm À2 (~0.57C), which is higher than some of the previous reports [11,12]. The discharge capacity at 0.57C is of 36.45 μAh cm À2 μm À1 and the columbic efficiency is about 89.67%.
Besides, the columbic efficiency doesn't decrease until the current density is at 40 μA cm À2 , where the discharge capacity retains ~73.15% of its value at 0.57C. However, it is surprising that four cycles at the end of stepwise rate performance measurements show unexpectedly large charge capacity, and the charge-discharge curves of these four cycles are presented in Fig. S2d. It can be observed that the voltage first decreases then recovers during the charge process, which is not the diffusion limitation as discussed for the 1st cycle but instead the presence of side reaction [50]. This can be the potential explanation that the electrolyte decomposition peak around 4.2V disappears at the 59th cycle but shows up in the 79th cycle, presented in Fig. 3b, as well as the reason that the 81st cycle experienced much significant increase regarding only charge capacity. Electrochemical Impedance Spectroscopy measurements of the cell at 100th cycle is presented in Fig. 3d. The model used to fit the spectrum contains three R-C circuits that can be possibly ascribed to cathode electrolyte interphase, charge transfer process in the cell, and the inhomogeneous Au/Li 2 MnO 3 interfaces [51,52].
Apart from the traditional electrochemical performance, optical anisotropy due to the tilted plasmonic Au nanopillars is expected and thus evaluated to study the tunable optical permittivity. Dielectric permittivity of the pure Li 2 MnO 3 and Li 2 MnO 3 -Au samples are calculated by fitting the angular dependent ellipsometry data in the wavelength of 500 nm-1500 nm, and the results are presented in Fig. 4a-c.
The ellipsometer parameter ψ and Δ was fitted with the use of general oscillator models to make it Kramers-Kronig consistent (see Methods section). The permittivity of Li 2 MnO 3 (Fig. 4c) shows a normal dispersion curve that is characteristic of dielectric materials. Interestingly, the Li 2 MnO 3 -Au sample shows anisotropic permittivity in the in-plane and out-of-plane directions. The in-plane permittivity (ε' k ) shows a typical dielectric behavior while the out-of-plane permittivity ðε' ? Þ shows the decrease in intensity indicating the existence of Au pillars in the out-ofplane direction. The anisotropic behavior of Li 2 MnO 3 -Au grants the potential of real-time online monitoring of battery charge-discharge behavior. The Au appears not just as pillars in the Li 2 MnO 3 matrix but a certain degree of doping [26] in the Li 2 MnO 3 lattice, as evidenced by the decrease of the bandgap for Li 2 MnO 3 -Au composite from 2.15 eV to 1.95 eV calculated using the Tauc plot in Fig. 4d. The measured transmittance plot and simulated curves by ellipsometry data are shown in Fig. S3, where the shape of the curves matches well.
Mechanical integrity of the cathodes is also critical for the long term performance of the battery [53,54]. The mechanical property of both pure Li 2 MnO 3 and Li 2 MnO 3 -Au pillars was evaluated and compared by nano-indentation and the result is presented in Fig. S4a-b. It is seen that the hardness of Li 2 MnO 3 -Au has increased from averagely 4.7 GPa-10.7 GPa, which is about twice of pure Li 2 MnO 3 . The mean value increment can be attributed to the strengthening of interface between Li 2 MnO 3 matrix and Au pillars. Note that the deviation parameter of Li MnO 3 -Au is higher than that of pure Li 2 MnO 3 , which is presumably caused by structure nonuniformity.
It is interesting to note that the tilting angle can be effectively tuned by the PLD growth parameters as illustrated in Fig. 1. TEM samples of films grown at different tilt angles α were prepared, observed using HAADF technique and the result is shown in Fig. 5. The relationship between angles α and β is not fully understood and depends on both material properties and geometrical effects. The setup shown in Fig. 5a have target-to-substrate holder distance d 0 , actually target-to-substrate distance d', and tilt angle α 1 . The composite film possesses thickness about 240 nm, long Au pillars with aspect ratio larger than 20, and pillar-substrate angle β 1 about 70 � . Besides, it can be measured from the detailed view of individual pillar in Fig. S4e that the width is about 5.5 nm. On the other hand, the film exhibits thinner film (~180 nm) at a smaller tilt angle α 2 . Besides, the pillars are shorter yet wider (~11.1 nm) compared to the configuration in Fig. 5b, and the pillar-substrate angle β 2 is about 80 � . Furthermore, the short pillars also hold different strain state compared to the long pillars. As mentioned, the long pillars experience a 0.14% compressive strain in Li 2 MnO 3 along [001] direction and 2.77% compressive stain in Au along [111] direction, whereas the short pillars have 0.53% tensile strain of Li 2 MnO 3 (001) and 1.66% tensile strain of Au (111). The reason that the smaller tilt angle renders this result can be explained by the growth rate controlled by the tilt angle. Smaller tilt angle indicates slower growth rate, further indicating increased stability of in-plane growth and suppression of out-of-plane growth [55]. This matches with the difference in thickness between the two configurations. It is known that smaller tilt angle α 2 result in less significant shadowing effect (more close to normal growth configuration). Therefore, the Li 2 MnO 3 adatoms have higher chance to disrupt the continuous growth certain Au pillars, forming Au morphology that is more close to particle [27,55]. It is expected that the particle-in-matrix structure is less effective than pillar-in-matrix structure due to the inefficient conduction path and the detailed electrochemical performance study is still ongoing. Furthermore, the Au pillars of these two films demonstrate distinct optical behavior as observed in Fig. 5g andh. The permittivity (ε' k ) along the in-plane direction presents a blue shift which might be due to the increased in-plane dimension of the Au pillars. Additionally, the out-of-plane permittivity ðε' ? Þ of shorter Au pillars decays faster than longer pillars due to its relatively higher volumetric percent.
Epitaxial Li 2 MnO 3 -Au thin film cathodes are obtained using the oblique angle deposition method in pulsed laser deposition. The film morphologies such as film thickness, porosity, Au concentration, pillar tilt angle, pillar interspacing, pillar width and length can be effectively tuned by controlling the tilting angle and growth parameters. Such a unique nanopillars-in-matrix form presents the potential of creating even more complicated and novel structures by selecting the optimal materials combination, where new applications can be explored. The battery performance measured by cyclic voltammetry measurements, Crate measurements, and impedance analysis all suggest the enhanced electrochemical performances due to the Au tilted nanopillars. Besides, the successful demonstration of angular tunability and correlation between thin film structure and optical properties suggests such optoelectrochemical systems can be engineered to achieve real-time battery performance monitoring and evaluation through optical approaches.
Composite film growth. Commercial Li 2 MnO 3 powder from Pfaltz & Bauer was mixed with 15% Li 2 CO 3 (Alfa Aesar) to process the target. The Li 2 MnO 3 target was sintered at 900 � C under oxygen flow for 24 h. The Li 2 MnO 3 -Au composite target was prepared by attaching a piece of thin Au strip on central line of the Li 2 MnO 3 target and the target was concentrically rotating during the deposition. The composite Li 2 MnO 3- -Au film was deposited on both Al 2 O 3 single crystalline substrates for structural analysis and stainless steel substrates buffered with Au (Au-SS) for electrochemical measurements. All depositions were conducted using pulsed laser deposition (λ ¼ 248 nm, KrF source, Lambda Physik Compex Pro 205) at laser frequency of 10 Hz and energy of 2.5 J cm À2 . The target surface to substrate surface distance is kept at 5 cm. The temperature was set at 750 � C and oxygen partial pressure was maintained at 50 mTorr during deposition. Substrates were tilted at certain angle away from the substrate holder to achieve the oblique angle deposition configuration. The deposited film was cooled under 15 Torr O 2 at rate of 10 � C/min.
Structural characterization. X-ray diffraction pattern (XRD) was measured using a Cu Kα1 (λ ¼ 1.5406 Å) source (PANalytical Empyrean Diffractometer) to characterize the crystallinity of the samples. The measurement used 0.05 � step size and 0.5 s step dwelling time with 2 theta angle ranging between 10 � to 90 � . Scanning electron microscopy (SEM, NovaNano SEM) was used to probe the surface morphology of composite film on Au-SS substrates. The acceleration voltage was set at 5 kV and spot size of 2.0 was used to be able to obtain high magnification images. Transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM) (under high angle annular dark field, HAADF), and elemental distribution spectrum (EDS) were taken on sample deposited on Al 2 O 3 using a FEI TALOS 200X system operated at 200 kV.
Coin cell battery assembly and electrochemical characterization. CR2032 coin cell was assembled in a glovebox filled with Ar (MBraun Labmaster, O 2 < 0.1 ppm, H 2 O < 0.1 ppm) using Lithium metal disks (Sigma Aldrich) as anode, Celgard 2400 PP (Celgard) as separator, and 1 M LiPF 6 salt dissolved in 1:1 vol ratio EC:DEC (Sigma Aldrich) as electrolyte. The assembled coin cells were evaluated using a Arbin BTS2000 battery testing system. The cutoff voltage window was set between 2.0 V and 4.8 V as other reported literatures for better comparison [11,12]. Cycling performance was measured using a constant current/constant voltage program, which was cycled firstly under 5 μA/cm 2 and constant voltage until the current decreased below 1 μA.
The Cyclic Voltammetry (CV) at different cycles were measured at 20 μV/s, and CV at different voltage sweep rate ranging from 10 μV/s to 200 μV/s was also measured to evaluate the kinetics property of the battery. The rate performance measurement was conducted at current density respectively of 5 μA/cm 2 , 10 μA/cm 2 , 20 μA/cm 2 , and 80 μA/cm 2 , and the C-rate values of each current density were estimated using the actual discharge time. Electrochemical impedance spectrum was measured using Gamry Series G300 Potentiostat between 500 μHz and 200000 Hz at 4V at 100th cycle. The cell was aged at 4V constant voltage for 2 h to reach stable potential before the measurement started.
Optical characterization. A spectroscopic ellipsometry (JA Woollam RC2) was applied to evaluate the optical dielectric permittivity of all films was evaluated. The ellipsometer parameters ψ and Δ () were fitted using the CompleteEASE software with the mathematical relationship of r p /r s ¼ tan(ψ)e (iΔ) where r s and r p are the reflection coefficient for the spolarization and p-polarization light, respectively The incident angles were 50 � ,60 � , and 70 � . The ψ and Δ are measured at different angles to improve the accuracy of the fitted model. Li 2 MnO 3 film was modeled using a Tauc-Lorentz oscillator and a Lorentz oscillator. Li 2 MnO 3 -Au film was assumed to anisotropic since the Au pillar grows as a tilted pillar. Its in-plane permittivity was modeled using same as Li 2 MnO 3 film while the out-of-plane permittivity was built using a Tauc-Lorentz oscillator and a Drude-Lorentz model.
Mechanical property measurement. Nanoindentation test was performed on a Hysitron TI 950 2000XYp with a Berkovich indenter. A loading-unloading function was applied to obtain force-displacement curves, based on which the film hardness at various depths was measured using instrumented nanoindentation method [56].