제 목: Material Science for 3D AI Hardware

■ 연 사: 배상훈 (Washington University, Assistant Professor)

■ 일 시: 2026년 06월 09일(화) 오후 4시             

■ 장 소: W1(응용공학동) 1층 영상강의

■ 호 스 트: 김경민 교수님

■ Abstract : The planar geometry that has defined semiconductor architecture for sixty years is approaching its limits, and with it, the materials science that underpins it. As electronics moves into three dimensions, the challenge is no longer simply arranging familiar materials into new configurations. Vertical 3D integration requires rethinking how crystalline layers and devices are grown, released, and assembled, and invites new questions about the physics that emerges when atomically precise interfaces are stacked and coupled beyond planar constraints. Two material classes are central to this shift: freestanding single-crystalline nanomembranes and 2D materials. 2D materials provide van der Waals interfaces free of dangling bonds, gate-tunable band structures, and strong excitonic coupling, enabling physics inaccessible in bulk systems. Nanomembranes remain single-crystalline yet substrate-independent, preserving bulk properties while enabling strain relaxation and transfer onto diverse platforms. Both exhibit ultralow internal stress and mechanical compliance, enabling deterministic heterogeneous 3D assembly. Together, they allow precise heterostructure engineering and access to rich interface physics. At 2D/3D boundaries, electrostatic, elastic, and quantum interactions can be engineered with high fidelity, while stacked layers support coupled phonon, exciton, and magnon modes yielding emergent functionalities such as nonlinear response and correlated transport. Realizing this potential requires advances in scalable synthesis, atomic-precision layer transfer/assembly, and co-design across electronic, optical, thermal, and mechanical domains, forming the foundation for next-generation 3D integrated AI hardware.

In this talk, I will present our group’s work on the fundamental materials science and interfacial physics underlying 3D integrated systems. I will highlight advances in epitaxial growth, layer transfer, nanomembrane release, and heterostructure assembly as platforms to control interfacial boundary conditions. These approaches enable probing of electrostatic, elastic, and quantum interactions across 2D/3D interfaces, revealing emergent ferroic behavior and non-equilibrium coupling phenomena absent in bulk systems. By linking interface structure, charge dynamics, and collective excitations, this work establishes design principles for coupled material systems. Device-level demonstrations, including in-sensor computing, 3D DRAM, and AI accelerators, validate these mechanisms and position interface and interference engineering as a fundamental paradigm for next-generation 3D materials systems.