03/11/2026
By Danielle Fretwell

The Francis College of Engineering, Department of Mechanical Engineering, invites you to attend a Master's Thesis defense by Jonathan Reardon on: "Formulating, printing, and characterization of dielectrics for high frequency applications."

Candidate Name: Jonathan Reardon
Degree: Master’s
Defense Date: Friday, March 13, 2026
Time: 10 a.m.-noon
Location: Southwick 240

Committee:

  • Advisor: Christopher Hansen, Professor (Chair), Mechanical and Industrial Engineering, UMass Lowell
  • Guy Demartinis, Director/Principal Investigator, Physics, UMass Lowell, Submillimeter Wave Technology Laboratory
  • Scott Stapleton, Professor, Mechanical and Industrial Engineering, UMass Lowell

Abstract:
Digital light processing (DLP) additive manufacturing offers a promising pathway for fabricating dielectric polymer composites with tailored electromagnetic properties. However, printing highly loaded ceramic–resin suspensions remains challenging due to viscosity increases, particle settling, and strong optical attenuation that limits cure depth and dimensional accuracy. This thesis investigates the formulation, DLP printing, and dielectric characterization of ceramic-filled photopolymer suspensions using alumina (Al₂O₃), strontium titanate (SrTiO₃), and barium titanate (BaTiO₃) as functional fillers. Single- and multi-component ceramic loadings were formulated and evaluated to assess how ceramic composition influences suspension handling, print performance, and the resulting dielectric response of printed parts. The main objective of this thesis is observe how multi-component loaded systems can be used to tune dielectric properties and if these properties can be predicted with machine learning.

A printing and process development workflow was established on a Bison 1000 DLP printer, including resin preparation, dispersion strategy, and exposure parameter selection based on cure-depth behavior described by Jacob’s working curve. Printed samples were characterized using broadband transmission measurements, and a numerical fitting approach was implemented to extract the complex dielectric constant and optical constants across the measurement band. Transmission data were fit using a Debye relaxation model coupled with a plane-parallel slab optical model. Independent low-frequency permittivity values obtained from a Keysight measurement suite near 75 GHz were used as reference constraints to stabilize parameter extraction and ensure continuity with the Bruker spectrometer measurement range.

To enable composition-driven prediction of dielectric properties prior to printing, an ensemble physics-informed neural network (PINN) was developed to map ceramic/resin volume fractions to Debye parameters (ϵ_∞, Δϵ, τ). The trained ensemble provides predicted dielectric spectra and uncertainty estimates through run-to-run variance, enabling identification of compositions where additional experimental data would most improve model confidence. Overall, this work demonstrates a combined formulation–printing–characterization pipeline for DLP-printed ceramic–polymer dielectrics and shows how multi-component loading can expand the tunability of dielectric properties while balancing printability and material performance.