Towards Self-Driving Laboratory for Perovskite Ceramics: An Automated Rapid-Sintering and Dielectric-Analysis Platform (ASAP)

The convergence of machine learning (ML) and automated experimentation offers promising avenues for advancements in materials and manufacturing. As a result, an entirely integrated automated process facilitates the creation of the autonomous "self-driving laboratory" (SDL) for material discovery.^[1]Yet, challenges remain in creating comprehensive maps connecting the Process-Structure Properties (PSPs) of materials, largely due to the intricate relationship between a material's microstructure and its manufacturing process, as well as associated costs and energy consumption of trial-and-error experimentation. One exemplar is the Solid-State Reaction (SSR), a standard method for producing 3D bulk ferroelectrics. This method requires significant energy, driving up both costs and environmental impact, particularly with the need for prolonged high-temperature sintering and subsequent reheating for dielectric characterization of ML-predicted compositions.^[2]Addressing these challenges, we introduce the Automated Rapid Sintering and Dielectric Analysis Platform (ASAP), designed to streamline the dielectric characterization process of 3D tunable perovskites and the discovery of new disordered layered materials. This platform adopts a holistic approach to sample production, sidestepping potential bottlenecks, particularly in operations carried out by robotic arms connected to off-the-shelf lab equipment. Validated successfully with previously known samples, ASAP drastically reduces processing times to minutes compared to traditional methods that can span hours or days. Moreover, ASAP's capabilities extend to the efficient validation of innovative perovskite solutions, as demonstrated with samples from the barium family, particularly (Ba_xSr_1-x)CeO₃identified using ML-based combinatorial chemical space screening presented in our previous publication.^[3] We used ASAP to create relation between sintering condition and Phase structure of these particular unknown samples. In addition, the ASAP characterisation tool monitored the dielectric performance of number of ferroelectric (such as: BST and BTS samples) rapidly sintered samples over the frequency range of 0.2 to 3GHz with no sample surface modification required. Using this probe, the robot recorded resonance frequency, permittivity and temperature tunability for each sample. Looking ahead, we anticipate a growing trend towards user-friendly automation, open sourcing, and the use of collaborative robots (cobots) in laboratories. These advancements will undoubtedly accelerate the adoption of automation, enabling more efficient exploration in materials science.<br/><br/><b>References.</b><br/>[1] S. T. S. Bukkapatnam, <i>IISE Trans</i> <b>2023</b>, <i>55</i>, 75.<br/>[2] H. Zhang, H. Giddens, Y. Yue, X. Xu, V. Araullo-Peters, V. Koval, M. Palma, I. Abrahams, H. Yan, Y. Hao, <i>J Eur Ceram Soc</i> <b>2020</b>, <i>40</i>, 3996.<br/>[3] A. Ihalage, Y. Hao, <i>NPJ Comput Mater</i> <b>2021</b>, <i>7</i>, 75.