BEOL-Compatible 4F² Oscillator Using Vertical InGaAs Biristor for Highly Scalable Monolithic 3D Ising Solver.

Area of Science:

• Materials Science
• Computational Physics
• Electrical Engineering

Background:

• Non-deterministic polynomial-time (NP)-hard combinatorial optimization problems (COPs) require efficient and scalable solvers.
• Current hardware-based Ising solvers face significant scaling limitations.

Purpose of the Study:

• To experimentally demonstrate a scalable oscillator-based Ising solver using 4F² InGaAs bi-stable resistors (biristors).
• To explore the potential of InGaAs biristors for solving complex optimization challenges.

Main Methods:

• Investigated the oscillation behavior of InGaAs biristors.
• Emulated classical Ising spins using sub-harmonic injection locking (SHIL).
• Demonstrated capacitive and resistive coupling between InGaAs biristors.

Main Results:

• Established that InGaAs biristors can emulate Ising spins via SHIL.
• Achieved successful coupling (in-phase and out-of-phase) between biristors.
• Experimentally solved the MaxCUT problem using the biristor-based Ising solver, validated by simulations.

Conclusions:

• The InGaAs biristor-based Ising solver presents a scalable and compact architecture for addressing NP-hard COPs.
• This technology offers a promising pathway to overcome current hardware scaling limitations.
• Represents a significant advancement in developing efficient solutions for complex optimization problems.