Unified computational framework for high-sensitivity MTJ magnetometry: Integrating sub-poissonian shot noise, thermal drift, and array scaling

Magnetic Tunnel Junctions (MTJs) are promising weak magnetic field sensors, yet designing optimal device configurations requires accurate electromagnetic and thermal modeling. This work presents a quasi-static electrical-thermal simulation for MTJ sensors, integrating magnetotransport models, sub-Poissonian shot noise, and Bloch-law TMR degradation. The framework characterizes single devices, 2 × 2 and 4 × 4 parallel arrays, and 4-series configurations. For single devices, the simulation predicts a detectivity floor of 64.1 pT/ Hz . Parallel 2 × 2 and 4 × 4 arrays achieve 32.05 pT/ Hz and 16.02 pT/ Hz respectively, validating theoretical 1 / N uncorrelated noise scaling under ideal conditions. Temperature analysis reveals that while sensitivity degrades according to magnon excitation models, offset drift—driven by magnetic layer asymmetry—constitutes the primary accuracy limitation for DC applications, necessitating active compensation. This framework provides a critical computational baseline for system-level trade-off analysis, bridging the gap between isolated device physics and the design of optimized sensor arrays.