In-Situ Monitoring for Quality Assurance and Machine Learning in Direct-Write Additive Manufacturing
Abstract
Approved for Public Release: Although a central necessity of additively manufactured (AM) materials is that they must satisfy perfor,mance requirements, the microstructure, properties, and performance of AM materials cannot presently be assured with the same certai,nty as their traditionally manufactured equivalents. Further, as it relates to advanced RF generally, while 5G is currently being de,ployed in the lower 4G adjacent frequency ranges, significant and numerous challenges remain for the higher frequency bands, beginni,ng with materials for the various network components. The principal near-term objective of the PI is to use in-situ monitoring to de,fine the processing rules that determine the microstructure, macrostructure, and high-frequency dielectric response of electronic ce,ramic materials for advanced RF "grown" by laser-assisted direct-write AM. The specific aims of this research are to:1. Optimize imp,lementation of in-situ monitoring by Raman and IR spectroscopies in a customized direct-write AM platform for the measurement of tem,perature, phase, structure, and stress during printing of RF electronic ceramics. These studies will determine how particle size and, distribution, shape, slurry chemical composition, and slurry rheological properties together with the direct-write parameters and h,eat from 1.06 micrometre laser irradiation influence the basic chemical and physical mechanism(s) that govern sintering, densificati,on, crystallization, and domain growth in films of electronic ceramics.2. Establish ex-situ validation using post-manufacture struct,ure and properties characterization to correlate and verify in-situ determinations. Phase precipitation and segregation as well as i,mpurity and defect segregation often occur at boundaries and can have outsized effects on properties. Therefore, an emphasis of this,relation with the aforementioned processing and resultant structure, including defect and interfacial structure, and the fundamental, mechanisms of dipole formation, polarization, and dielectric response in the 5-40 GHz range. Crystallinity, defects, the existence, of permanent dipoles, and the mobility of free carriers all contribute to the dielectric response.4. Adopt processing models based, on multi-physics, three-dimensional, computational thermal modeling to simulate and obtain deeper insight into the thermokinetics (,thermodynamics and kinetics) of laser-matter interactions in the AM process. In-situ measurements will be correlated and validated w,ith ex-situ characterization, and these data will iteratively guide synthesis and processing optimization. In parallel, the experime,ntal measurements will be used as inputs for the phy,-situverification are limited by experimental measurements, whereas the datasets generated by the models will be large and governed, by computing resources. The computations and modeling, therefore, result in the amplification of experimentally validated data. Pre,dictions from the experimentally validated models will in turn: 1) guide synthesis and processing optimization and, 2) train and con,strain ML. Thus, quality assurance is achieved. This science-based toolset can be applied to any laser-assisted AM method that facil,itates the integration of in-situ monitoring and diagnostic probes, and is expected to provide a broad, transformative capability fo,r customizable, agile, affordable, high-throughput additive manufacturing of materials with designed structure, properties, and perf,ormance. In the longer-term, the new knowledge, methodology and data generated will be the foundation for training of ML models.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 08, 2022
- Source ID
- N000142212720
Entities
People
- Nigel Shepherd
Organizations
- Office of Naval Research
- United States Navy
- University of North Texas