Presentation Information
[WBP1-15]Optimization of Pulsed Laser Deposition Parameters for HoBCO Thin Films Using Definitive Screening Design
*Takumi Takamura1, Kota Shimodaira1, Iori Nishimura1, Noriyuki Taoka1, Yoshiyuki Seike1, Tatsuo Mori1, Yusuke Ichino1,5, Keiichi Horio2,5, Ataru Ichinose3,5, Tomoya Horide4,5, Kaname Matsumoto4,5, Yutaka Yoshida4,5 (1. Aichi Inst. of Technol. (Japan), 2. Kyushu Inst. of Technol. (Japan), 3. Cent. Res. Inst. of Elect. Power Ind. (Japan), 4. Nagoya Univ. (Japan), 5. JST-CREST (Japan))
Keywords:
HoBCO,definitive screening design,optimization of deposition conditions
A common issue with the pulsed laser deposition (PLD) method for producing REBa2Cu3Oy(REBCO) superconducting wires is that optimal conditions vary for different rare-earth elements. This requires a time-consuming optimization process for each REBCO due to the many deposition parameters involved. To address this, this study used a definitive screening design (DSD) to rapidly identify the most influential factors from a large number of parameters with a limited number of experiments. The research focused on developing a quick optimization method for holmium barium copper oxide (HoBCO).
The experimental setup involved using a HoBCO target to deposit thin films on a strontium titanate (SrTiO3) substrate. Three parameters were varied at three different levels: heater temperature (Th), oxygen pressure (pO2), and deposition time. Instead of the 27 experiments required for a full factorial design (33), the DSD method reduced the number of deposition runs to just 9. The resulting films were evaluated using X-ray diffraction (XRD) and the DC four-probe method. The evaluation data was then used to construct a mathematical model based on multiple regression analysis to identify the most influential parameters. The superconducting transition temperature (Tc) and critical current density (Jc) of the nine samples were evaluated, and the coefficients for the mathematical model were determined for both properties. For the Tc model, the largest coefficients indicated a parabolic trend with respect to oxygen pressure and deposition time. The Jc model showed a negative linear correlation with deposition time, which became steeper as oxygen pressure increased. The usefulness of both models was confirmed by their high determination coefficients (R2), which were 0.9980 for Tc and 0.8726 for Jc.
Contour plots, derived from the mathematical models, visualized the relationship between Tc and Jc and the deposition parameters. The plots identified the optimal conditions for high Tc as a heater temperature of 950℃, oxygen pressure of 40 Pa, and a deposition time of 30 min. The optimal conditions for high Jc were a heater temperature of 910℃, oxygen pressure of 50 Pa, and a deposition time of 20 min. The difference in the plots for Tc and Jc is likely due to the formation of a-axis-oriented grains, which can reduce the superconducting current path and consequently decrease Jc. The observed negative correlation between Jc and deposition time is probably because longer deposition times lead to thicker films with a higher ratio of a-axis-oriented grains. The study concludes by stating that future work will involve using Bayesian optimization to conduct a more detailed search for optimal conditions, building on the screening data from the DSD.
The experimental setup involved using a HoBCO target to deposit thin films on a strontium titanate (SrTiO3) substrate. Three parameters were varied at three different levels: heater temperature (Th), oxygen pressure (pO2), and deposition time. Instead of the 27 experiments required for a full factorial design (33), the DSD method reduced the number of deposition runs to just 9. The resulting films were evaluated using X-ray diffraction (XRD) and the DC four-probe method. The evaluation data was then used to construct a mathematical model based on multiple regression analysis to identify the most influential parameters. The superconducting transition temperature (Tc) and critical current density (Jc) of the nine samples were evaluated, and the coefficients for the mathematical model were determined for both properties. For the Tc model, the largest coefficients indicated a parabolic trend with respect to oxygen pressure and deposition time. The Jc model showed a negative linear correlation with deposition time, which became steeper as oxygen pressure increased. The usefulness of both models was confirmed by their high determination coefficients (R2), which were 0.9980 for Tc and 0.8726 for Jc.
Contour plots, derived from the mathematical models, visualized the relationship between Tc and Jc and the deposition parameters. The plots identified the optimal conditions for high Tc as a heater temperature of 950℃, oxygen pressure of 40 Pa, and a deposition time of 30 min. The optimal conditions for high Jc were a heater temperature of 910℃, oxygen pressure of 50 Pa, and a deposition time of 20 min. The difference in the plots for Tc and Jc is likely due to the formation of a-axis-oriented grains, which can reduce the superconducting current path and consequently decrease Jc. The observed negative correlation between Jc and deposition time is probably because longer deposition times lead to thicker films with a higher ratio of a-axis-oriented grains. The study concludes by stating that future work will involve using Bayesian optimization to conduct a more detailed search for optimal conditions, building on the screening data from the DSD.