Maximizing Output Power in p–n Junction Betavoltaic Batteries via Monte Carlo and Physics-Based Compact Model Cosimulation

Abstract

Betavoltaic nuclear batteries show promise as compact and enduring power sources for microelectromechanical systems (MEMS). Current theoretical calculations often overlook practical diode characteristics like surface recombination (S), bulk recombination within the space-charge region (R-SCR), series resistance ( $R_{s}$ ), and shunt resistance ( $R_{\text {sh}}$ ), resulting in significant discrepancies between theoretical predictions and experimental outcomes, with differences in $J_{\text {SC}}$ , $V_{\text {OC}}$ , or converter efficiency up to ten-fold. To address this, a practical diode model, integrating these practical characteristics, is developed via Monte Carlo and physics-based compact model cosimulation. We quantitatively analyze the differential impacts and synergistic effects of these practical characteristics on $J_{\text {SC}}$ , $V_{\text {OC}}$ , FF, and $P_{\text {out}}$ , highlighting the detrimental effects of S, R-SCR, and $R_{s}$ , while emphasizing the beneficial role of $R_{\text {sh}}$ . Further analysis of the degree of influence of S, $R_{s}$ , and $R_{\text {sh}}$ on output power reveals a priority ranking order of $R_{s}$ , S, and $R_{\text {sh}}$ for Si-based batteries, and S, $R_{\text {sh}}$ , and $R_{s}$ for SiC-based batteries. This approach effectively bridges the theoretical–experimental gap, evidenced by J–V curves closely matching tested batteries and minute relative errors of −0.8% to 0.6% between $P_{\text {out}}$ values and their tested counterparts, emphasizing its accuracy in predictions. We predict output performance across material qualities, obtaining achievable powers of 16.82 and $73.90~{\mathrm {nW/cm^{2}}}$ for planar Si- and SiC-based batteries, and evaluate the quality levels of current batteries. Furthermore, our model can forecast the performance of 3-D batteries by incorporating an extended electron–hole pair (EHP) generation rate model into 3-D structures, achieving $28~\mu $ W/cm3 for the ${}^{63}$ Ni–Si-based multilayer battery, surpassing planar silicon and suitable for MEMS applications.