Ph.D. Candidate · Georgia Institute of Technology

Minji Shon

I work on reliability- and thermal-aware DTCO/STCO for advanced 3D logic and memory. Currently a Design Quality & Reliability Engineer Intern at Intel, backed by 7 years of Samsung Design-for-Reliability across FinFET and GAA nanosheet technology.

Georgia Tech Intel Samsung Electronics
Portrait of Minji Shon
7+ yrsSamsung & Intel reliability
Advanced logic & memoryFinFET · GAA nanosheet · CFET · ferroelectric FET
8first/co-first author papers including IEDM/IRPS
Dec 2027expected Ph.D. graduation
Open to 2027 roles

Graduating December 2027. I am looking for roles in semiconductor reliability engineering, DTCO/STCO and design technology enablement, or physical design (STA, power analysis).

Work authorization: F-1, CPT/OPT eligible; EB-2 NIW I-140 approved (preparing green card application), so no employer sponsorship is needed for a green card.

01

About

I am a Ph.D. candidate in Electrical and Computer Engineering at the Georgia Institute of Technology, advised by Dr. Shimeng Yu in the Laboratory for Emerging Devices and Circuits, and an SRC Research Scholar (SPARC). My thesis is Next-Generation 3D Scaling: Reliability-Aware Device-to-System Co-Optimization for Advanced CMOS Logic and Memory.

Logic transistor density in transistors per square millimetre against year of mass production from 2015 to 2040, rising through the FinFET era to 3 nm, the nanosheet era to 1 nm with backside power delivery and high-NA EUV, and the nanostack era from 7A to 1A with vertically stacked gate-all-around transistors
Logic transistor density against year of mass production, through the FinFET, nanosheet, and nanostack eras. Pitch scaling carried density as far as the 3 nm node; beyond it, n/p separation and track height cap any further cell-height reduction, so density has to come from stacking devices vertically. Source: 2026 VLSI Short Course (IBM).

Before graduate school I spent seven years at Samsung Electronics as a Technology Quality & Reliability engineer, building the aging-aware design environments, reliability models, and sign-off methodologies behind Exynos SoCs from 14 nm through 4 nm and 3 nm GAA nanosheet technology. That industry background is what shapes my research question: once devices stack and power delivery moves to the backside, heat has a longer path out of the die and reliability margins tighten, so self-heating and aging have to be budgeted up front, the way timing and power already are.

Three linked renders: a package stack with heat sink, thermal interface layers, lid, silicon carrier, SoC, HBM, redistribution layer, interposer and substrate with the heat path marked; a close-up of backside power delivery network package integration showing silicon bonding, BEOL, FEOL and BSPDN layers; and the CFET and multi-tier CFET devices at the bottom of that stack
The same question at three scales. Heat leaves a modern package through a long series path (left); backside power delivery moves the power network underneath the device and changes that path (middle); and the devices at the bottom of it are stacked CFET and multi-tier CFET (right). My work follows the thermal and reliability consequences down this chain.

I am currently a Design Quality & Reliability Engineer Intern at Intel in Hillsboro, OR, developing physics-informed machine learning methods for pre-silicon reliability of analog IP and memory circuits, targeting CPU, server, and ASIC products.

02

Research

MATRIX-PDK project overview: a device, circuit, array, process and architecture wheel around a stacked-die icon, beside the research areas of advanced memory PDK, advanced logic PDK, and technology development and benchmarking, with the goal of covering logic, DRAM and flash

CFET DTCO & Open PDK

TCAD-calibrated device models and multi-tier standard-cell libraries that make academic DTCO research possible — cell- to block-level PPA and routing-congestion benchmarking on commercial Cadence/Synopsys flows, driven through a RISC-V core and a systolic array. [L1, L2, L3]

Two device cross-sections side by side: a conventional stack on a bulk silicon substrate with heat symbols at the channels, and the same stack on a thin silicon substrate with buried power rail, nanoTSV and a backside power delivery network beneath, showing more trapped heat

Electro-Thermal Co-Design

Device-to-system electro-thermal frameworks for multi-tier CFET and GAA nanosheet with backside power delivery — including GPU-accelerated multi-physics simulation and layout-resolved thermal analysis to address thermal constraints in 3D scaling. [R1, R2]

Memory window loss against memory window for four ferroelectric gate stacks, from a 19 nm HZO reference at 2.3 volts and 6.6 percent loss to an 8 nm ferroelectric, 3 nm tunnel dielectric, 8 nm ferroelectric stack reaching 7.5 volts at zero loss, with TEM cross-sections of each stack

Ferroelectric Devices

Time-dependent Sentaurus models of the trap and polarization interplay in FeFET and Fe-NAND stacks, spanning gate-stack and channel-side interlayer engineering, with memory window, retention, and disturb quantified. [F1, F2, F3, F4]

Digital compute-in-memory macro: word-line driver and input switch matrix feeding SRAM sub-arrays with adder trees and a shift-and-add stage, with insets of the bit cell with its NOR output and of a 4-bit to 7-bit adder tree

3D Compute-in-Memory

Digital compute-in-memory benchmarking on A5 CFET technology, extended to look-up-table-based power efficient design, targeting the memory wall across the memory pyramid. [L4]

Open process design kits compared by institution, node, device, integration, standard-cell library, multi-threshold-voltage flavors, temperature corners, RTL-to-GDS flow, backside power delivery and availability.
Capability ASAP7 ASAP5 FreePDK3 GT3 USC 3N-2D MATRIX-A5(this work)
Institution ASU ASU NCSU Georgia Tech USC (SCCAD) Georgia Tech
Node 7 nm 5 nm 3 nm 3 nm 3 nm A5
Device FinFET NWFET GAA NS GAA NS GAA NS CFET
Integration 2D 2D 2D 2D 2D 3D stacked device
Standard-cell library ✓ ✓ ✗ ✓ ✓ ✓
Multi-Vth flavors ✓ ✗ ✗ ✗ ✗ ✓
Temperature corners ✓ ✗ ✗ ✗ ✗ ✓
RTL-to-GDS flow ✓ ✓ ✓ ✓ ✓ ✓
Backside power (BSPDN) ✗ ✗ ✗ ✓ ✓ ✓
Availability open open open open open will be open
supported not supported
Where MATRIX-A5 sits among open research PDKs. The others are 2D integration on FinFET, nanowire or gate-all-around nanosheet devices; MATRIX-A5 is the only CFET-based, 3D stacked-device kit, and the only one that carries multi-Vth flavors, temperature corners, a full RTL-to-GDS flow and backside power delivery together.
03

Experience

Design Quality & Reliability Engineer Intern

Aug 2026 – Jan 2027
Intel · Hillsboro, OR
  • Product Quality Network team: developing physics-informed machine learning methods for pre-silicon reliability of analog IP and memory circuits, targeting CPU, server, and ASIC products.

Graduate Research Assistant

Aug 2023 – Present
Georgia Institute of Technology · Atlanta, GA · SRC Research Scholar (SPARC)
  • Open-source CFET PDK and multi-tier DTCO/STCO — TCAD-calibrated device models and the first open-source sequential-CFET PDK (six 67-cell library variants) taken through a complete RTL-to-GDSII flow; 40% logic and 43% SRAM cell-footprint reduction from 2-tier to 4-tier. [L1, L2, L3]
  • Block-level technology benchmark — a RISC-V core and a 16×16 systolic array through synthesis, place-and-route, and timing and power analysis: CFET cut standard-cell area 27.5% against a GAA-nanosheet baseline for a 14.5% power cost at iso-frequency, and a hybrid-row track-mixing scheme (CFlex) closed a 1.5 ns clock that a dense-only library could not meet. [L1]
  • Reliability-aware co-optimization — device-to-circuit electro-thermal framework (TCAD to self-heating-aware compact model to ring-oscillator analysis) showing that backside power delivery traps heat: over 10% ring-oscillator frequency degradation at 1.3 V, against under 5% with frontside delivery; GPU-accelerated multi-physics (DeepSim) extended to multi-tier CFET for layout-resolved analysis. [R1, R2]
  • Memory devices and in-memory compute — CFET digital compute-in-memory (DCIM) against the memory wall, and ferroelectric gate-stack and channel-side interlayer modeling in FeFET and Fe-NAND, where a gate-injection-layer stack more than doubles the memory window. [L4, F1, F3, F4]
CFET DTCO / STCO TCAD RTL-to-GDSII Physical design STA Electro-thermal

Technology Quality & Reliability Engineer

Aug 2016 – Jul 2023
Samsung Electronics · South Korea
Pre-silicon reliability sign-off & aging-aware design verification
  • Delivered aging- and variability-aware design methodologies for Samsung Exynos SoCs (14 nm – 4 nm) under advanced reliability sign-off.
  • Developed statistical aging simulation modeling and methodology, enabling AEC-Q100 automotive-grade design-for-reliability in 8 nm FinFET products.
  • Partnered with Design Technology Enablement to deliver critical-path reliability timing margin from aging simulation directly into the Exynos SoC implementation flow.
  • Built aging-aware circuit design environments and PDK components for 3 nm GAA nanosheet technology.
Post-silicon correlation & reliability model validation
  • Built and validated reliability models for HCI, BTI, TDDB, and self-heating, improving prediction accuracy through model-to-hardware correlation against ring-oscillator silicon benchmarks with the WLR and PDK teams.
  • Provided DFR for real product operating conditions (multi-step aging, power-down bias instability) rather than idealized stress, closing the gap between qualification and field use.
IP- & memory-level reliability verification
  • Verified overdrive-voltage reliability in non-volatile memory (eFuse, OTP, MRAM) under product stress conditions.
  • Prior internship (Jul – Aug 2016): implemented enhanced Dynamic Device Voltage Check (DDVC); strengthened safe-operating-area (SOA) validation and voltage-stress reliability sign-off across advanced nodes.
HCI / BTI / TDDB Self-heating AEC-Q100 GAA nanosheet Exynos SoC
04

Education

Georgia Institute of Technology

Aug 2023 – Dec 2027 (expected)

Ph.D., Electrical and Computer Engineering · Atlanta, GA

Sogang University

Mar 2012 – Aug 2016

B.S., Electronic Engineering · Seoul, South Korea · summa cum laude

University of Connecticut

Jan – May 2015

International Student Exchange, Electrical and Computer Engineering · Storrs, CT

05

Technical Skills

Focus
Reliability- and thermal-aware DTCO/STCO; multi-tier CFET and 3D integration; physical design and timing closure; ferroelectric device modeling; aging and reliability sign-off.
Reliability physics & sign-off
HCI, BTI, TDDB, self-heating, SOA; statistical aging simulation; model-to-hardware correlation; AEC-Q100 automotive qualification.
TCAD & device
Sentaurus electro-thermal device/process; BSIM-CMG compact modeling.
VLSI design & implementation
Synopsys Design Compiler; Cadence Innovus (place-and-route, floorplan, CTS, congestion and timing closure); static timing analysis (Cadence Tempus); power analysis; parasitic extraction (Quantus, StarRC); custom cell layout (Cadence Virtuoso); full PDK build and verification (LIB/LEF, LPE, DRC/LVS); reliability sign-off and critical-path timing-margin analysis.
Circuit simulation
SPICE (HSPICE, FineSim, Spectre) with aging engines (RelXpert, MOSRA); Dynamic Device Voltage Check (DDVC); multi-step and power-down-bias stress modeling.
Technologies
FinFET, GAA nanosheet, multi-tier CFET, backside power delivery, ferroelectric-based memory.
Programming
Python, TCL, Perl, C, MATLAB, SystemVerilog.
06

Honors & Awards

  • Best Paper Award, Runner-Up — IEEE Device Research Conference (DRC) Jun 2025
  • Silicon Creations ISSCC Travel Grant Feb 2024
  • Conference Travel Grants — APCCAS'24, IRPS'25, DRC'25, IEDM'25 (presentation) 2024 – 2025
  • CSE Department Fellowship (offered) — University of Minnesota–Twin Cities Jan 2023
  • Summa cum laude — Sogang University; Dean's List Aug 2016
  • Full National Scholarship (Merit) — Korea Student Aid Foundation 2014 – 2016
  • Honors Scholarship (Merit) — Sogang University 2012 – 2015
07

Leadership & Service

  • Ministry Team Leader, Grad CRU — Georgia Institute of Technology 2024 – Present
  • Education Mentor / Sub-Course Manager — Samsung Electronics 2019, 2020, 2022 Delivered onboarding for 200+ new hires: "Global Business Manners," "Communication & Creative Thinking."