Presentation Information
[WBP1-12]Correlation analysis of YBCO film between a-axis oriented grain characteristics based on AFM image analysis and XRD
*Ibuki Kato1, Shin Okumura1, Tomoya Horide1, Yutaka Yoshida1 (1. Nagoya University (Japan))
Keywords:
REBCO,a-axis oriented grain,microstructure,AFM,XRD,image analysis
[Purpose]
XRD is a method for analyzing the entire microstructure, including the amount of each substance contained in the crystal and the direction of orientation from a macroscopic perspective. On the other hand, by observing the crystal surface using atomic force microscopy (AFM), it is possible to analyze information about the crystal growth environment, such as material surface diffusion, nucleation frequency, and supersaturation during film deposition, from a microscopic perspective. Although there should be a correlation between the analysis results of both methods in crystal growth, a comprehensive understanding has not yet been achieved. In this study, we aimed to correlate the features analyzable from crystal surface images with the microstructure of the crystal and investigated the correlation between the information on the thin film surface obtained by AFM and the structural information obtained by XRD.
[Method]
Using pulse lazor deposition, YBa2Cu3Oy (YBCO) thin films were fabricated on SrTiO3 (100) substates. The film deposition temperatures were 820℃, 870℃, and 920℃, and the laser frequencies were 1 Hz, 3 Hz, and 10 Hz. For each of the nine samples, 54 AFM images were taken, and the tone of the AFM images was adjusted to emphasize the a-axis oriented grains. The particle analysis was done for each tone-adjusted image using the image analysis software ImageJ, and particles with an aspect ratio above the threshold were classified as a-axis oriented grains, with their number density and average size calculated. Additionally, XRD measurements were performed on each sample to determine the 2θ/ω values, and the a-axis oriented grain mixing ratio was calculated from the intensity ratio of YBCO (005) to YBCO (200). Linear regression, exponential regression, and random forest (RF) regression were examined, using the film deposit temperature, laser frequency, and the a-axis oriented particle number density and average size analyzed by AFM as input variables, and the a-axis oriented particle mixing ratio (a-axis percentage) obtained from XRD as the target variable.
[Results and Discussion]
The linear regression curve and exponential regression curve including AFM information and film deposition condition information had an R2 value greater than 0.93, which indicates an improvement over the regression curves containing only film deposition condition information. On the other hand, no improvement in the R2 value was observed in the RF regression. When relying solely on deposition conditions for regression, minor errors in the deposition process cannot be adequately reflected. However, by incorporating actual sample information obtained from AFM measurements, the correlation with XRD results can be improved, leading to higher prediction accuracy. The correlation coefficient between the number density of a-axis oriented grains and the a-axis orientation grain mixing ratio was calculated to be 0.90, indicating that the microstructure can be estimated from the AFM analysis results. These results demonstrate the correlation between the film deposition conditions and the a-axis data (number density and average size) analyzed from the AFM images, and the a-axis percentage obtained from XRD.
XRD is a method for analyzing the entire microstructure, including the amount of each substance contained in the crystal and the direction of orientation from a macroscopic perspective. On the other hand, by observing the crystal surface using atomic force microscopy (AFM), it is possible to analyze information about the crystal growth environment, such as material surface diffusion, nucleation frequency, and supersaturation during film deposition, from a microscopic perspective. Although there should be a correlation between the analysis results of both methods in crystal growth, a comprehensive understanding has not yet been achieved. In this study, we aimed to correlate the features analyzable from crystal surface images with the microstructure of the crystal and investigated the correlation between the information on the thin film surface obtained by AFM and the structural information obtained by XRD.
[Method]
Using pulse lazor deposition, YBa2Cu3Oy (YBCO) thin films were fabricated on SrTiO3 (100) substates. The film deposition temperatures were 820℃, 870℃, and 920℃, and the laser frequencies were 1 Hz, 3 Hz, and 10 Hz. For each of the nine samples, 54 AFM images were taken, and the tone of the AFM images was adjusted to emphasize the a-axis oriented grains. The particle analysis was done for each tone-adjusted image using the image analysis software ImageJ, and particles with an aspect ratio above the threshold were classified as a-axis oriented grains, with their number density and average size calculated. Additionally, XRD measurements were performed on each sample to determine the 2θ/ω values, and the a-axis oriented grain mixing ratio was calculated from the intensity ratio of YBCO (005) to YBCO (200). Linear regression, exponential regression, and random forest (RF) regression were examined, using the film deposit temperature, laser frequency, and the a-axis oriented particle number density and average size analyzed by AFM as input variables, and the a-axis oriented particle mixing ratio (a-axis percentage) obtained from XRD as the target variable.
[Results and Discussion]
The linear regression curve and exponential regression curve including AFM information and film deposition condition information had an R2 value greater than 0.93, which indicates an improvement over the regression curves containing only film deposition condition information. On the other hand, no improvement in the R2 value was observed in the RF regression. When relying solely on deposition conditions for regression, minor errors in the deposition process cannot be adequately reflected. However, by incorporating actual sample information obtained from AFM measurements, the correlation with XRD results can be improved, leading to higher prediction accuracy. The correlation coefficient between the number density of a-axis oriented grains and the a-axis orientation grain mixing ratio was calculated to be 0.90, indicating that the microstructure can be estimated from the AFM analysis results. These results demonstrate the correlation between the film deposition conditions and the a-axis data (number density and average size) analyzed from the AFM images, and the a-axis percentage obtained from XRD.