Detecting Yield Excursions with Spatiotemporal Correlation in Planar Lightwave Circuit Fabrication
Open Access DepositedPlanar Lightwave Circuits (PLCs) are passive optical chips built using optical fiber techniques and large-scale integrated circuit (LSI) patterning. These processes enable low-loss transmission and require tolerances on the order of femtoseconds to meet operational specifications. Strict process controls and high-precision measurements are used to control fabrication. Increasing the frequency and coverage of measurements improves defect detection rates
however, such measures are cost-prohibitive for smaller fabrication facilities. This lack of comprehensive measurement coverage across fabrication steps complicates efforts to establish correlations between process parameters and test outcomes. These limitations result in spatially and temporally sparse datasets which impact the ability to correlate yield losses with fabrication excursions. This research proposes exploiting the changes in multivariate relationships to enhance process control and reduce excursion investigation cycle time. Dimensionality reduction was performed using wafer maps, analogous to spatial kernels, to pool median aggregate chip results into wafer regions. Temporal dimensionality reduction was performed by joining each wafer spatial measurement (left, top, center, bottom, right) to all its future region pooled results. Multivariate relationships were distilled with rolling 30-day spatiotemporal correlation, creating a time series of fabrication relationship changes. Inter-series relationships were determined through clustering and excursion event labeling. The result is the framework for extensible system for monitoring and labeling the changes in multivariate relationships with excursions in fabrication. This novel approach to spatiotemporal correlation control (SCC) proposed in this research has the potential to augment statistical process control (SPC) by incorporating multivariate temporal observations that reflect historical and leading indicators of process deviations. These methods are intended to assist small and medium business (SMB) PLC Fabricators in scaling production by reducing cycle time, localizing yield excursions, and enhancing accuracy in monitoring and yield estimation. The proposed solutions achieve these objectives without requiring hardware-level metrology innovations, direct modifications to fabrication processes, or capital-intensive interventions such as expanded physical measurement infrastructure or advanced automation systems.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
Gyarmathy_gwu_0075A_17455.pdf | 2025-12-11 | Open Access |
|