Investigation of the Memory Effect in Electrochromic Sputtered WO₃ Thin Films—A Combined Approach of Experiments and Machine Learning

Electrochromic materials and devices, able to modify reversibly their optical properties in response to an applied voltage are recognized as one of the key green technologies for sustainability and energy savings. The state of the art of electrochromic oxide is WO₃, which undergoes coloration upon reduction with insertion of small cations (H⁺, Li⁺ or Na⁺) associated with a visual color change from colorless to dark blue. Aiming at low energy consumption applications, the unusual property of so-called the memory effect, known as the persistence of the colored state for electrochromic materials when the applied voltage is withdrawn, has recently received particular attention. This unusual property for electrochromic metal oxides allows to reduce power consumption of devices by avoiding to apply a permanent voltage in order to keep the desired color or optical state. Herein, amorphous WO₃ thin films were grown by Radio-Frequency Magnetron Sputtering (RFMS) using a WO₃ ceramic target. EC properties were estimated by cycling the films in lithium-based electrolyte while the memory effect was characterized by the evolution of the optical transmittance ΔT at 550 nm in 48h of Li_xWO₃ colored films in air-controlled environment (25°C and 50% R.H). By adjusting sputtering parameters such as power density, oxygen and total pressure films with a low (ΔT_550nm~61%) or high color persistence (ΔT_550nm~16%) were obtained. Nevertheless, it was found that high color persistence WO₃ films could also present a low electrochemical reversibility (~23%) hence limiting practical applications. In order to optimize the compromise between these two properties, machine learning, ML, was used as a tool to design prediction maps depending on the sputtering conditions. By combining color persistence and reversibility in a single map, high performance WO₃films with high color persistence and reversibility were targeted. From this ML mapping, WO₃ films with three different electrochromic properties were found : namely, films showing <i>(i)</i> an irreversible blue coloration and high color persistence, <i>(ii)</i> films with high reversibility but poor color persistence and <i>(iii)</i> films with a memory effect which present both high values [1]. As a result from the ML prediction maps, the total pressure during the deposition plays a key role on the memory effect since the three categories of WO₃ films were obtained by varying this parameter. In the meantime, by increasing the total pressure, films progressively go from a compact morphology to a porous one. As the morphology is more open, faster lithium diffusion and switching times are recorded. In return, the duration of the memory effect is shortened. Hence WO₃ films with a high memory effect with good EC properties requires a careful control of the morphology allowing to avoid irreversible coloration or poor color persistence. The relationship between the deposition conditions, morphology, stoichiometry and the memory effect property in WO₃ films will be discussed.<br/><br/>[1] B. Faceira, L. Teule-Gay, G.-M. Rignanese, A. Rougier, Toward the Prediction of Electrochromic Properties of WO ₃ Films: Combination of Experimental and Machine Learning Approaches, J. Phys. Chem. Lett., (2022) 8111–8115.