Origin of the Memory Effect in Electrochromic Sputtered WO3 Films: Composition, Structure, or Morphology?

The memory effect property, described as a reversible color persistence while the potential is withdrawn, is of particular importance to reaching zero‐energy consumption electrochromic devices. Nevertheless, often observed in organic materials, it is poorly studied in oxides. In this study, electrochromic tungsten oxide thin films are elaborated at different working pressures by radiofrequency magnetron sputtering. The influence of the deposition pressure on the electrochemical properties of WO3 films as well as on their memory effect is investigated. Three kinds of WO3 films can be distinguished: with irreversible blue coloration, with reversible coloration presenting a high memory effect, and with reversible coloration with a low memory effect. The origin of these discrepancies is studied through the composition, the local atomic environment, and the morphology of the WO3 thin films. An increase of transmittance at 550 nm as low as 15.8% in 48 h in air and 5.2% in 24 h in the electrolyte is recorded. This study highlights a better understanding of the memory effect property of electrochromic oxides for low‐consumption energy electrochromic devices, pointing out morphology as a key parameter.


Introduction
The increase in carbon dioxide emission from primary energy sources from fossil fuels (oil, gas, and coal) leads to global warming, atmospheric pollution, and rising sea levels.These environmental challenges urge the search for novel energysaving technologies.Buildings cover 30% to 40% of the world primary energy consumption, most of it is used for temperature control, lighting, and appliances. [1]lazing are poor energy-efficient components in buildings leading to use heaters or air conditioning to regulate indoor comfort.In order to preserve outdoor contact and sunlight while reducing energy consumption, smart windows are a relevant option as they allow to control incoming light radiation. [2,3]Smart windows technology is often based on electrochromic devices.Electrochromism is the ability to tune the optical properties of materials by applying an external voltage. [4]Among electrochromic inorganic materials, tungsten trioxide (WO 3 ) remains the most investigated. [5][8] Its electrochromic properties are strongly dependent on its crystallinity.Amorphous WO 3 is well-known for high color modulation and switching speed due to disordered structure while crystalline WO 3 exhibits higher cycling stability. [9,10]A self-bleaching behavior in its colored blue state leads to an increase of the transmittance in opencircuit conditions in air or in electrolyte. [11,12]To overcome this issue, an increase in the color retention of WO 3 based devices when the applied voltage is withdrawn can be obtained by adding a Ta 2 O 5 layer on top of the electrochromic film. [13,14]This memory effect property is defined as the retention of the colored (or bleached) state for cathodic (anodic) electrochromic materials under open circuit conditions while being reversible. [15]It thus becomes unnecessary to apply a voltage continuously in order to keep the desired colored state thus minimizing the energy required to achieve coloration.If such property is often discussed for organic materials, [16] it remains poorly addressed for inorganic ones and in particular for oxides.By using a hybrid-cationic based aluminum electrolyte, Guo et al. reported a decrease of the self-bleaching of WO 3 films due to the growth of an aluminum hydroxide based interphase at the solid electrolyte interface during cycling. [17]Alternatively, by controlling the degree of crystallinity, WO 3 sputtered films displaying a memory effect in air or in electrolyte were successfully deposited without additional layers. [18]rom earlier work, [19] we investigate the control of the memory effect in radiofrequency RF sputtered WO 3 thin films by tuning mainly three deposition parameters, namely the power, total pressure, and oxygen partial pressure.From a combination of experimental and machine learning approaches, three kinds of WO 3 films were identified based on their color retention and electrochemical reversibility.The three categories were irreversible blue coloration, high color retention with electrochemical reversibility, or low color retention with electrochemical reversibility.However, at this stage, no explanation on the origin of the memory effect in respect of the various deposition conditions were established.To fill this gap, the current study focuses on a combination of advanced characterization techniques, including X-ray absorption spectroscopy (XAS), Glow Discharge Optical Emission Spectroscopy (GD-OES), and Time of Flight Secondary Ion Mass Spectrometry (ToF-SIMS) analysis, aiming at a better evaluation of the relationship between the physico-chemical properties of the sputtered WO 3 films and their electrochromic performance in regards in particular of the memory effect.For this purpose, the memory effect was measured in ambient simulated conditions and in electrolyte.The relationship between the stoichiometry, the structure, the morphology of the WO 3 films and the memory effect is discussed.

Electrochromic Characterization
WO 3 thin films were elaborated by RF-sputtering at various total pressures.The samples are named depending on the total pressure during deposition as 1, 2, and 4 Pa (Table S1, Supporting Information).WO 3 films were electrochemically characterized by cyclic voltammetry (CV) at 20 mV s −1 and chronoamperometry (CA) in a lithium based electrolyte (Figure 1).During the oxidation process (j > 0 mA cm −2 ), the overall CV shape, characteristic of amorphous WO 3 , differs for the three samples.The 1Pa-WO 3 curve becomes flat after a broad and low-intensity oxidation peak.The CV oxidation curve of the 2Pa-WO 3 film never reaches 0 mA cm −2 at +1 V whereas the 4Pa-WO 3 one exhibits a sharper decrease in current until 0.7 V from which the current density reaches ≈0 mA cm −2 .Shift in oxidation peak position toward higher potential for the 2 Pa compared to 4 Pa one is related to a higher resistance. [20,21]he insertion/deinsertion Li equation in WO 3 film can be written as The inserted/desinserted Li amount during reduction/oxidation for blue coloration/bleaching can be calculated by:  where Q is the capacity (C cm −2 ) calculated in reduction/oxidation from ∫ j.dV∕v, j is the current density, v is the scan rate (20 mV s −1 ), M is the molar mass of WO 3 (231.84g mol −1 ), F is the Faraday's constant (96 485 C mol −1 ),  is the density estimated to be 75% of the theoretical WO 3 density (0.75*7.16 g cm −3 ) and t is the film thickness.Capacity in reduction as well as estimated x values and electrochemical reversibility (Q oxidation /Q reduction ) extracted from the first CV cycle are collected in Table 1.For the 1 Pa total pressure, as suggested by the CV shape, a poor reversibility of 23.8% is associated with an irreversible blue coloration.It means that at +1 V, there is no further bleaching of the film to a colorless state after the first lithiation.The x values for 2 and 4 Pa are similar around x ≈0.5.
The CV shapes of the 2 Pa-and 4 Pa-WO 3 films remain quasiidentical after the first cycle indicating a good cycling reversibility and stability.For the 1 Pa-WO 3 , the oxidation peak shifts to lower potential, and the capacity in oxidation increases indicating an easier diffusion for the 5th cycle. [22]Indeed, after five cycles, the electrochemical process becomes quite reversible (Table 1, i.e., Q ox /Q red (%) = 98.0%).The estimated amount of trapped lithium during the 1st cycle corresponds to x ≈0.3 leading to a blue color on oxidation.The reversibility of 98.0% on the 5th cycle for the initially irreversible 1Pa-WO 3 results from reversible sites that can be deinserted while some lithium irreversibly trapped cannot be removed even using a +1 V bias leading to a blue color at +1 V. [23] From chronoamperograms (Figure 1b), the shapes in oxidation and in reduction are different for the three WO 3 films.In oxidation, the decrease of current density toward equilibrium shows that it takes 40 s for 2Pa-WO 3 to reach j ≈0 mA cm −2 (considered as <0.02 mA cm −2 ) and only 10 s for 4Pa-WO 3 .In the case of 1Pa-WO 3 this value is never reached.Higher peak current and faster decline of 4Pa-WO 3 compared to 2Pa-WO 3 indicates a faster electrochemical process. [24,25]CV measurements were performed at various scan rates going from 50 to 5 mV s −1 in order to evaluate the lithium diffusion in WO 3 films (Figure 1c-e).The peaks current density in oxidation increased and they shifted toward higher voltage under scan rate increase due to charge polarization of the working electrode. [26]From Figure 1f, the peak current density shows a linear relationship with the square root of scan rate, which indicates a Li + ion insertion/deinsertion limited by diffusion-controlled process.Hence, according to CV data, the Li + diffusion coefficient can be determined by the Randles-Sevcik equation: [10] where j is the peak current density in oxidation, n is the number of exchanged electrons (considered as one), D is the diffusion coefficient of Li + ions, C is the concentration of Li + ions, and v is the scan rate.Lithium diffusion coefficients (D) range from 1.1 × 10 −10 cm 2 s −1 , 2.8 × 10 −9 cm 2 s −1 to 8.5 × 10 −9 cm 2 s −1 for 1Pa-, 2Pa-, and 4Pa-WO 3 , respectively.The increase of diffusion coefficient with total pressure during WO 3 deposition indicates faster kinetics.This is in agreement with the CV shapes in oxidation for which the current density reached almost 0 mA cm −2 at lower potential in the case of 4Pa-WO 3 as compared to 2Pa-WO 3 .
In situ optical measurements were performed to analyze optical modulation as well as kinetic behavior.Figure S1 (Supporting Information) presents the transmittance evolution at 550 nm for the three films during CA between −1 and +1 V.In the case of 1Pa-WO 3 , the transmittance initially decreases from the as-deposited state to −1 V going from 90.1% to 14.1%.The subsequent +1 V pulses and cycles do not allow to fully re-oxidize the film after the first coloration (T ≈ 25%).Upon cycling, the optical contrast ΔT (ΔT = T bleached − T colored ) increases.As observed with the increase of capacity in oxidation during cycling in Figure 1, further cycles help to eliminate trapped Li + in the film.In the case of 2Pa-and 4Pa-WO 3 , the modulations in transmittance during CA are stable.The transmittance spectra in the visible range in oxidized and reduced states of 2Pa-and 4Pa-WO 3 are shown in Figure 2a.
High contrast ≈77.8% and 77.0% are reached at 550 nm.The coloration and bleaching times were deduced from the time needed to reach 90% of the full contrast at 550 nm (Figure 2b). [27]n increase in total pressure from 2 to 4 Pa leads to a significant decrease of the bleaching times from 12.2 to 5.2 s whereas the coloration times slightly decrease from 16.6 to 15.3 s.As related to its lower Li + diffusion coefficient, the 2Pa-WO 3 film presents longer switching times than the 4Pa-WO 3 one.Except for WO 3 -1 Pa (Figure S2, Supporting Information) bleaching times are faster than coloring times as related to the current density in CA that are higher in bleaching than in coloring.[30] Coloration efficiency (CE) is an essential parameter to describe EC performance, it can be expressed as With log(T b /T c ) corresponds to the Optical Density change (ΔOD) and Q is the electrochemical capacity in C cm −2 .CE is obtained by the slope of ΔOD = f(Q) in the linear region (Figure S3, Supporting Information).A high CE indicates a large contrast is reached with a small amount of inserted charges.Here both CEs are in the same range ≈58 and 53 cm 2 −1 for 2Pa-and 4Pa-WO 3 , respectively, and in accordance with literature on WO 3 thin films grown by sputtering. [18,31,32]The optical memory effect was measured after blue coloration by CA at −1 V.The optical transmittance of blue Li x WO 3 films was recorded before and after 48 h storage in a climatic chamber at 25 °C and 50% relative Humidity (R.H) as shown in Figure 3.
In air controlled environment, the transmittance at 550 nm increases by 15.8% and 61.1% for the 2 and 4 Pa films, respectively.After 48 h in the climatic chamber, the 4Pa-WO 3 film was totally bleached whereas the 2Pa-WO 3 remains blue.In the case of 1Pa-WO 3 the transmittance spectrum after −1 V coloration is rather different with a flat shape after absorption edge.The transmittance increase corresponds to 10.4% that is lower than for the others.This film presents a high color persistence as for the 2Pa-WO 3 .However, the difference lies in the electrochemical reversibility.2Pa-WO 3 film exhibits both color retention and re- versible electrochemical behavior corresponding to the so-called memory effect. [19]1Pa-WO 3 films have an irreversible coloration and 4Pa-WO 3 shows a low memory effect.

Origin of the Memory Effect
Determining the origin of the memory effect requires investigation of the physico-chemical properties of the films.Whatever the total pressure, all as-deposited WO 3 films are amorphous (featureless X-ray diffraction pattern Figure S4, Supporting Information).
X-ray absorption spectroscopy (XAS) is a powerful tool to investigate the atomic and electronic structures that dominate the chemical and physical properties of materials, even those that are amorphous.To elucidate the local atomic and electronic structures of WO 3 electrochromic thin films, W L 3 -edge XAS was conducted, as shown in Figure 4a.
The W L 3 -edge probes the electronic transition from W 2p 3/2 to 5d unoccupied states that are hybridized with O 2p states.All the sputtered films, regardless of their total pressure, had similar absorption energy to that of reference WO 3 and dissimilar to that of W powder, indicating a charge state of ≈+6.This oxidation degree was confirmed at least in surface by XPS (X-ray Photoelectron Spectroscopy) analysis (Figure S5, Supporting Information).Notably, as the total pressure increases, the width of the main peak becomes broader.As displayed in the inset of Figure 4a, the main peak of 4Pa-WO 3 is slightly broader than that of 1Pa-WO 3 .
The XAS profile is highly sensitive to the local atomic structural symmetry, and the theoretical prediction indicates that under the octahedral symmetry, the W 5d states of WO 3 split into 5d-t 2g and 5d-e g states. [33]To unveil changes in the spectra, Figure 4b presents the second derivatives of the W L 3 -edge, which reveal two peaks corresponding to W 5d-t 2g and 5d-e g states, respectively.The crystal field splitting of both WO 3 and 1Pa-WO 3 is close to 4.2 eV, a similar value that has been reported elsewhere. [34]However, the crystal field splitting decreases to 3.4 eV for 4Pa-WO 3 , implying the local structural deformation.Notably, the e g states, which are associated with dz 2 and dx 2 -y 2 orbitals that point directly to the six corners of WO 6 octahedron, can reflect the local atomic structural deformation.As indicated in Figure 4b, the e g states become broader and smear in the main O K-edge probes the electronic transition from O1s to empty states.Due to the hybridization between O 2p and W 5d states the measurement of the O K-edge allows to study the modification of the electronic structure around W. Figure 4c presents the O K-edge, which contains three main features.The feature at ≈530-532 eV is attributed to the hybridized O 2p-W 5d-t 2g states, while the feature appearing at ≈534-539 eV results from the hybridized O 2p-W 5d-e g states.The feature ≈544 eV is ascribed to the O 2p hybridized with rather extended W 6sp states.The spectral profiles of all sputtered films resemble that of h-WO 3 or slightly deviate from stoichiometric WO 3 . [35]The overlapped O K-edge spectra are shown in Figure 4d and the intensity of the main peak decreases as the total pressure is increased.Notably, the more intense peak in O 2p-W 5d states suggested the greater degree of overlap between O 2p and W 5d orbitals, which suggests the higher covalency.Then by going from 1Pa-to 4Pa-WO 3 , the covalency decreases.Additionally, the appearance of a shoulder-like feature at ≈532 eV indicates the anisotropic 5d-t 2g states that result from the non-equivalent site of oxygen in the WO 6 octahedron. [35]hus, further EXAFS of W L 3 -edge was performed to explore the local coordination environment.Figure 5a) presents the EX-AFS oscillations of k-space.The oscillated profiles of sputtered films are consistent with that of reference WO 3 indicating the sputtered films have WO 6 octahedral atomic environments.The Fourier-transformed (FT) k 3 -weighted EXAFS at W L 3 -edge presented in Figure 5b further supports all the sputtered films  exhibit similar coordination environments with reference WO 3 .A predominant peak originates from the W─O bonds, which comprise two types of bonds: four bonds in the basal plane (W-O1) and two bonds (W-O2) in the apical direction.To gain a better understanding of the structural information, the FT-EXAFS curve fitting was conducted, as shown in Figure S6 (Supporting Information), and the resulting quantitative structural parameters are included in Table S2 (Supporting Information).The analytic results demonstrate that the total coordination number of W─O remains almost the same for 1 Pa-, 2Pa-, and 4Pa-WO 3 .Moreover, the W─O bond distance changes more significantly for the bond in the apical direction than in basal plane, leading to structural deformation that results in a smaller crystal field splitting of 5d-t 2g and 5d-e g as the film is sputtered at a higher total pressure.Notably, the defect states can be easily observed within the bandgap of semiconductors or oxide materials.Figure S7 (Supporting Information) shows the X-ray absorption and X-ray emission spectra (absorption-emission), which can be utilized to determine energy gap between conduction and valence bands through the derivative of the O K-edge absorption-emission spectra. [36]The bandgap for all the films is estimated to be ≈3.38 eV, which is in agreement with the value obtained from UV-vis spectrum Figure S8 (Supporting Information).No additional states within bandgap were observed, suggesting that the total pressure does not significantly control the different degree of oxygen vacancies during sputtering.
Since the WO3 films were deposited using different total pressures, one can expect beside a change of stoichiometry an evolution of the morphologies of the films. [37]orphology of the films were examined by SEM and AFM in Figure 6.The cross-sections images present compact structure without voids and confirmed the thickness of 180-200 nm of the films.The Root Mean Square (RMS) roughness of 1Pa-, 2Pa-, and 4Pa-WO 3 films are 1.72, 1.65, and 2.38 nm respectively.At 1 Pa, the surface is very smooth with some dispersed particles.By increasing the pressure to 2 Pa, the surface is still compact but with a higher roughness and discernible particles boundaries.At 4 Pa, larger particles are visible with porosities corresponding to a rougher surface than for the two other films.The cracks observed may facilitate the contact with electrolyte and ion insertion. [10]This morphology evolution by increasing the total pressure is to be correlated to the decrease of sputtered particles kinetic energy when reaching the substrate due to higher number of collisions. [38]n increase of the total pressure is associated with a good reversibility (Figure 1).As referred to the morphology, the smooth and dense surface of the 1 Pa film makes difficult the lithium insertion as illustrated by the lower capacity in reduction and lithium trapping that leads to irreversible coloration. [38]An increase in pressure to 2 Pa leads to reversibility because of easier lithium insertion and extraction but still more difficult than for the 4Pa-WO 3 for which the highest diffusion coefficient was obtained.
In order to have a better understanding of the Lithium insertion/deinsertion self-bleaching mechanism, composition depth profile of 2Pa-WO 3 film in various states was analyzed by combining GD-OES and ToF-SIMS.Figure 7a presents the GD-OES profile of 2Pa-WO 3 in as-deposited state, oxidized state at +1 V, after reduction at −1 V and one week after coloration at −1 V while stored at 25 °C and 50% R.H.This longer time of 7 days allows the 2Pa-WO 3 film to bleach as shown by the increase in transmittance of ≈63% (Figure S9, Supporting Information).The asdeposited film consists of no lithium signal on the whole profile as expected.For the cycled films, the peak intensity at the very beginning of the profile curves comes from electrolyte residual traces present at the surface.The lithium distribution of oxidized WO 3 film (+1 V) exhibits a very low intensity on depth until near the WO 3 /ITO interface where an increase of Li signal appears.This high intensity of lithium near the interface is due to a modification of the sputtering rate when passing from the film layer to the ITO substrate.Nevertheless, it indicates that some lithium is still present in the film.For the reduced film Li x WO 3 at −1 V the overall Li intensity increases as compared to +1 V as there is a higher content of inserted lithium.After self-bleaching of the Li x WO 3 film during 7 days (−1V7d), lithium is still present in significant proportion (related to high intensity) whereas the film came back to a colorless state (self-bleaching).However, the lithium signal is lower than in the freshly reduced −1 V state in the upper part of the film than similar near the ITO substrate.This indicates a self-bleaching of the film without a total lithium loss in simulated ambient atmosphere.
As a comparative analysis, ToF-SIMS was used to investigate elemental depth distribution (Figure 7b).The intensities shown are only qualitative since ToF-SIMS is not able to give quantitative information without the use of standards.Hence the direct comparison of W + and Li + intensities is not possible. [39]Nevertheless, considering the same matrix and charging effects for the various states of the film, the intensities for a given element can be compared. [40]Li + was detected on the whole depth of the oxidized film at +1 V meaning that some lithium ions stay trapped during successive cycling.This is in agreement with the GD-OES profile where the Li intensity is not exactly to zero for this film before the increase at the WO 3 /ITO interface.After reduction at −1 V, the Li + intensity is greater than for the oxidized +1 V as expected with the formation of a Li x WO 3 phase.The self-bleached film (−1V7d) presents similar intensities as the −1 V one except at the beginning of the profile, where less Li signal is measured in accordance to the GD-OES profiling.For better comparison of the lithium depth distribution inside the film, the Li + /W + intensity ratio profile is presented Figure 7c.As expected from the electrochemical oxidation, the Li + /W + intensity is close to zero for the oxidized film.By comparing profiles of the reduced and self-bleached film, it seems that there is less lithium in the upper part of the film after 7 days of self-bleaching whereas the Li + /W + intensity ratio is similar in the deeper part of the profile.Even if the lithium quantification is not possible in our case, both GD-OES and ToF-SIMS show that lithium is present after self-bleaching and its relative quantity compared to the freshly reduced film has decreased.
The self-bleaching in air is first characterized by an increase of the transmittance within time.The 2Pa-WO 3 film with the highest memory effect goes from a blue color to colorless after 7 days.Hence during the self-bleaching in air W 5+ color centers are oxidized to W 6+ and lithium is still present in the selfbleached Li x WO 3 film.In terms of morphology, the more porous surface of 4Pa-WO 3 compared to others can allow easy oxygen diffusion the film during ageing in air that leads to a faster self-bleaching rate.The oxidation of W 5+ color centers into colorless W 6+ is then facilitated by the porous morphology compared to 1Pa-WO 3 presenting a compact surface.

Memory Effect in Electrolyte Environment
Since in air atmosphere, large discrepancies on the color persistence are observed, transmittance evolution in time was also recorded in the electrolyte to prevent air oxidation.tance during self-bleaching in electrolyte depends on the degree of coloration. [17]The less the film is colored, related to less involved electrochemical capacity, the more the self-bleaching will be important.As shown from Table 1, the 1 Pa film presents the smallest capacity due to trapped lithium in first reduction.Hence, this could explain its faster increase of transmittance in electrolyte.At the end of the 24 h measurements, all the films remain blue indicating a different speed of self-bleaching in electrolyte compared to air environment.After the 24 h selfbleaching experiment at open-circuit, Li x WO 3 thin films were subjected to a chronoamperometry at +1 V (Figure S10, Supporting Information).The difference between inserted and extracted amount of charge before and after self-bleaching leads to estimate the amount of charges that has been lost.These differences are 13.0, 7.5, and 7.7 mC cm −2 for 1Pa-, 2Pa-, and 4Pa-WO 3 , respectively, which validates the loss of lithium from films during self-bleaching.After the self-bleaching, CA oxidation leads back to the initial oxidized state (Figure S11, Supporting Information).Even if the open-circuit memory effects are reported for different wavelengths in literature, the current WO 3 films present high memory effect in lithium based electrolyte as to compared to most of others amorphous WO 3 as reported in Table S3 (Supporting Information).

Conclusion
In this study, electrochromic WO 3 thin films were deposited by RF magnetron sputtering at different total pressures.The increase of pressure leads from a dense surface morphology to a porous one.From the electrochemical properties, three different behaviors were observed.At low pressure, WO 3 films present a low electrochemical reversibility with an irreversible blue coloration after the first reduction.By increasing the total pressure films with reversible switching coloration and high coloration efficiency up to 58 cm 2 −1 were obtained.To conclude about the origin of the memory effect in WO 3 film, it seems to be greatly affected by the morphology more than their composition or structure.The memory effect property was shown to be tunable as in air or in electrolyte.Thin film with high memory effect values allows to reduce the energy consumption since the voltage is not applied continuously to maintain desired color state, which is highly desirable for applications.

Experimental Section
Thin Film Deposition: WO 3 films were prepared by Radiofrequency Magnetron Sputtering in a PLASSYS MP700S apparatus using a WO 3 target (99.9%purity purchased from Neyco) of 75 mm of diameter and 3 mm thick.The ITO (In 2 O 3 : Sn, 30 Ω sq −1 ) coated glass substrates purchased from SOLEM were ultrasonically cleaned with ethanol and distilled water before their introduction in the chamber.Prior to the deposition process the base pressure in the chamber was set below 3.10 −5 Pa.The target to substrate distance was fixed at 10 cm and the total O 2 +Ar gas flow at 50 sccm.A plasma stabilization step of 20 min was done with a shutter, then the shutter was open to allow deposition that was carried without intentional heating of the substrate.At the end of the sputtering process, the WO 3 target was re-oxidized at 75 W at 5 Pa and 10% O 2 for 1 h.The sputtering time was adjusted to reach films thickness in the 200 ± 20 nm range as measured with a Dektak profilometer.The samples were named depending on the total pressure during deposition as 1, 2, and 4 Pa.Deposition parameters are summarized in Table S1 (Supporting Information).
Optical transmittance of films was recorded using a Varian Cary 5000 UV-vis-NIR spectrophotometer.Acquisition speed was 120 nm min −1 between 300 and 800 nm.
For ex situ measurements, air was taken as baseline.Memory effect measurements protocol consist in five cyclic voltammetry (CV) in between −1 and +1 V followed by five chronoamperometry (CA) at +1 V and −1 V.After blue coloration at −1 V electrolyte traces were removed with ethanol then ex situ transmittance measurement was performed on lithiated Li x WO 3 films.The films were placed in a climatic test chamber (MKF 56, Binder co.) at 25 °C and 50% relative humidity for 48 h before next optical measurement.Measured humidity and temperature were in the range of ±2.5% and ±1.3 °C, respectively.Ex situ optical measurements on lithiated samples after 48 h was carefully performed on the same films area.
For in situ optical transmission measurements, glass cell and electrolyte without electrodes was taken as baseline.
The surface morphology of the films was examined by scanning electron microscopy (JEOL 6700F) in secondary electron operating at 5 kV and by Atomic Force Microscopy (AFM) using a ezAFM+ instrument in tapping mode.
The X-ray Absorption Spectroscopy (XAS) was recorded at Taiwan Light Source (TLS) and the Taiwan Photon Source (TPS) at the National Synchrotron Radiation Research Center (NSRRC), Taiwan.The W L 3 -edge was collected at TLS-BL17C, and the O K-edge X-ray absorption and X-ray emission (absorption-emission) spectra were made at TPS-45A2.The X-ray emission spectra were measured using a 550 eV excitation energy.The reference WO 3 was provided by Acros Organics/Thermo Scientific Chemicals (Lot number A0417679) as powder form with purity of 99.9%.
Glow Discharge Optical Emission Spectroscopy (GD-OES) was performed using GD-Profiler 2 (HORIBA Jobin Yvon) instrument.The sputtering of the films was done under a radio frequency of 3000 Hz with a duty time of 0.25% at 50 W power under 300 Pa Argon atmosphere.The emission of Li was measured at 670 nm.
Time of Flight Secondary Ion Mass Spectrometry (ToF-SIMS) measurements were carried out using a TOF.SIMS five time-of-flight secondary ion mass spectrometer (IONTOF GmbH), equipped with a dual-source sputter gun and a liquid metal ion gun (LMIG) orientated at 45°to the sample surface.The LMIG provided a 10 micron-diameter spot 30 keV Bi + ion beam, operating at an ion current of 2 pA, raster scanned randomly in the region of interest.Depth profiles were performed in a non-interlaced mode with sputtering cycles using 2 keV O 2 + ions at a current of 600 nA and cycle time of 50 μs, over an area of 300 × 300 μm, interleaved with secondary ion images generated by the LMIG over an area of 100 × 100 μm in the center of the sputtered region.Surface potential was adjusted to optimize the analyzer transmission.
XPS was performed using a K-ALPHA equipment (ThermoFisher Scientific) with a monochromatic Al K X-ray source (1486.6 eV) and a 400 μm spot size.A pressure of 10 −5 Pa was reached in the chamber when transferring the films.High-resolution spectra were quantified and fitted using the AVANTAGE software provided by ThermoFisher Scientific (Scofield sensitivity factors applied).High resolution spectra were fitted using a convolution of Lorentzian and Gaussian functions after Shirley background subtraction.All spectra were shifted versus W4f 7/2 maximum peak at 35.9 eV (i.e., WO 3 bonds).

Figure 7 .
Figure 7. a) Glow Discharge Optical Emission Spectroscopy (GD-OES) of Lithium in 2Pa-WO 3 on as-deposited state, in oxidized state at +1 V after cycling, after lithium insertion at −1 V and 7 days after the lithium insertion at −1 V while stored at 25 °C and 50% R.H. b) Corresponding ToF-SIMS profile of W + and Li + ions.c) Li + /W + ion intensity profile calculated from ToF-SIMS measurements.In order to compare the different states, the sputtering times are normalized to their respective time required to reach the substrate.

Figure 8 Figure 8 .
Figure 8. Evolution of transmittance at 550 nm in open circuit voltage (OCV) after coloration at −1 V.

Table 1 .
Comparison of electrochemical capacity in reduction Q, inserted Li amount x, reversibility, and lithium diffusion coefficient of WO 3 films deposited at 1, 2, and 4 Pa.