PROJECT TITLE :

A Strategy to Characterize Nanofabrication Processes With Large RPM (Experimental Run, Physics, and Measurement) Uncertainties

ABSTRACT:

The bottom-up fabrication of nanostructures can simultaneously face large uncertainties from experimental runs (R), physical understanding (P), and measurement (M). No systematic strategy has been reported to manage these 3 sorts of uncertainties, abbreviated as RPM, concurrently to attain better understanding of nano fabrication processes. Previously, we developed cross-domain model building and validation (CDMV) approach to handle massive physical and measurement (PM) uncertainties in nano fabrication process modeling. During this paper, we propose to prioritize RPM uncertainties and to include the analysis of run variations into method modeling. Under a Bayesian hierarchical framework, this new strategy can initial handle PM uncertainties at the basic level to spot a model structure using CDMV approach. The rationale is that the uncertainty because of experimental runs ought to not fundamentally amendment the process physics or the model structure, however impacts on the model parameters. At a lower hierarchy, process model parameters varying or invariant to runs are treated as random effects or fastened effects to be identified respectively. Demonstrated during a nanowire growth method example, the new strategy not solely assists to determine an improved method model, however conjointly to uncover the variation sources contributing to giant run variations. The obtained physical insights will guide additional process investigation. Note to practitioners: experimental investigation of nanofabrication processes usually encounters giant uncertainties because of a lack of conclusive understanding of process physics, measurement noise, and variability among experimental runs. Trial-and-error strategy is usually adopted beneath this scenario to explore the method physics with very little steering, ensuing in increased price of experimentation or fabrication. This paper presents another strategy to make more economical use of data to manage massive RPM uncertainties and achieve higher method understa- dings for method improvement.