| Pacific Rim Symposium on Surfaces, Coatings and Interfaces (PacSurf 2018) | |
| Thin Films | Wednesday Sessions |
| Session TF-WeP |
| Session: | Thin Films Poster Session II |
| Presenter: | Takuya Sakurai, Tokyo University of Agriculture & Technology, Japan |
| Authors: | T. Sakurai, Tokyo University of Agriculture & Technology, Japan Y. Hirata, Tokyo University of Agriculture & Technology, Japan K Takebayashi, Tokyo University of Agriculture & Technology, Japan Y. Iwata, Tokyo University of Agriculture & Technology, Japan J. Shirakashi, Tokyo University of Agriculture & Technology, Japan |
| Correspondent: | Click to Email |
Much progress towards artificial intelligence (AI) technique is due to the rapid growth of data size and accessibility in recent years. Thus, AI technique has been applied to many control systems. Meanwhile, feedback-controlled electromigration (FCE) has been employed to create atomic junctions with quantized conductance [1]. Previously, we have proposed ultrafast FCE system using field programmable gate array (FPGA) to adjust quantized conductances of Au atomic junctions [2]. Because of many experimental parameters in FCE procedure, it is difficult to optimize them by rules of thumb in control of quantum states. In this report, we investigated AI-assisted FCE system implemented on FPGA to immediately and precisely fabricate Au atomic junctions.
Au nanowires were fabricated using conventional electron-beam lithography and lift-off process. They were patterned on resist-coated SiO2/Si substrates using electron-beam lithography. Then, electron-beam evaporation of Ti (5 nm) and Au (20 nm) was carried out using a developed resist patterned as template.
AI-assisted FCE system is composed of four engines; learning, evaluation, inference and FCE engines. First, the feature values of conductance quantization obtained from previous experiments were stored in an initial database in learning engine. Then, FCE parameters were evaluated using cost function in evaluation engine. After the evaluation, FCE parameters were optimized by genetic algorithm (GA)-based selection methods in inference engine. Finally, FCE procedure using optimized experimental parameters was applied to Au nanowires in FCE engine. As a result, Au atomic junctions were fabricated by AI-assisted FCE procedure and the conductances of the junctions were successfully quantized at room temperature without catastrophic breaks of the structures. Furthermore, the FCE scheme was performed within an order of millisecond due to the use of FPGA. Therefore, these results imply that Au atomic junctions can be elaborately fabricated with improved controllability of quantized conductance using AI-assisted FCE implemented on FPGA.
References
[1] D. R. Strachan, D. E. Smith, D. E. Johnston, T.-H. Park, M. J. Therien, D. A. Bonnell, and A. T. Johnson, Appl. Phys. Lett., 86, 043109 (2005).
[2] Y. Kanamaru, M. Ando, and J. Shirakashi, J. Vac. Sci. Technol. B, 33, 02B106 (2015).