Lead-free perovskite/HfSe2 surface plasmon resonance biosensor design via multi-metric optimization, Monte Carlo robustness analysis, and machine learning
Abstract
Replacing toxic lead in perovskite-enhanced surface plasmon resonance (SPR) biosensors is critical for environmentally sustainable sensing, yet the performance of lead-free alternatives in multilayer architectures remains poorly characterized. We numerically investigate a lead-free perovskite/HfSe2 SPR biosensor within a CaF2/Ag/perovskite/HfSe2/sensing-medium Kretschmann stack. Four perovskites are evaluated: MAPBr3 as a Pb-based benchmark and the lead-free candidates Cs2AgBiBr6, Cs2AgBiCl6, and CsSnBr3. The optical response is calculated at 633 nm using a recursive Fresnel transfer-matrix method. Two refractive-index windows are optimized independently: a generic aqueous interval (1.3300–1.3350) and a bio/SARS-like interval (1.3348–1.3398). A multi-metric optimization framework balancing angular sensitivity, figure of merit (FoM), resonance depth, and linewidth is employed to identify balanced sensor designs. For the generic sensing interval, the optimal balanced lead-free design uses Cs2AgBiCl6 with Ag/Cs2AgBiCl6/HfSe2 thicknesses of 47/3.0/1.0 nm, achieving a sensitivity of 397.0 deg RIU−1 and a FoM of 103.66 RIU−1. For the bio/SARS-like interval, the optimal balanced design switches to CsSnBr3 (43/2.5/1.0 nm), providing 447.8 deg RIU−1 sensitivity and a FoM of 90.05 RIU−1, while MAPBr3 remains the highest-sensitivity material overall. Complementary Monte Carlo robustness analysis, local machine-learning surrogates, electric-field visualization, and layer-isolation studies are used to interpret the optimized sensor response. Our results demonstrate that optimal material selection is regime-dependent: Cs2AgBiCl6 performs best for generic aqueous sensing, whereas CsSnBr3 is superior for bio-relevant indices, providing application-specific design guidelines for lead-free SPR biosensors.
Related
- Spectral Optimization of Programmable LED Illumination for Smartphone-Based Colorimetric Biosensor Readout.
- Design and Optimization of AlGaN/GaN Graded Barrier MOS-HEMT Biosensor for Breast Cancer Biomolecule detection
- Quasi-2D Perovskite Photonic Crystals for Single-Mode Lasing via Synergistic Material and Structural Optimization by Nanoimprint Lithography
- Perovskite Nanocrystal-Based Scintillators: Synthesis Routes, Matrix Engineering, and Performance Optimization for Radiation Detection
- Charting the thermodynamic stability of hybrid perovskite alloys with machine learning
BibTeX
@article{kouz2026leadfree,
title = {{Lead-free perovskite/HfSe2 surface plasmon resonance biosensor design via multi-metric optimization, Monte Carlo robustness analysis, and machine learning}},
author = {Sadok Kouz},
journal = {Sensing and Bio-Sensing Research},
year = {2026},
month = aug,
doi = {10.1016/j.sbsr.2026.101077},
url = {https://hal.science/hal-05837688v1},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Lead-free perovskite/HfSe2 surface plasmon resonance biosensor design via multi-metric optimization, Monte Carlo robustness analysis, and machine learning
AU - Sadok Kouz
PY - 2026
DA - 2026/08/19
JO - Sensing and Bio-Sensing Research
DO - 10.1016/j.sbsr.2026.101077
UR - https://hal.science/hal-05837688v1
AB - Replacing toxic lead in perovskite-enhanced surface plasmon resonance (SPR) biosensors is critical for environmentally sustainable sensing, yet the performance of lead-free alternatives in multilayer architectures remains poorly characterized. We numerically investigate a lead-free perovskite/HfSe2 SPR biosensor within a CaF2/Ag/perovskite/HfSe2/sensing-medium Kretschmann stack. Four perovskites are evaluated: MAPBr3 as a Pb-based benchmark and the lead-free candidates Cs2AgBiBr6, Cs2AgBiCl6, and CsSnBr3. The optical response is calculated at 633 nm using a recursive Fresnel transfer-matrix method. Two refractive-index windows are optimized independently: a generic aqueous interval (1.3300–1.3350) and a bio/SARS-like interval (1.3348–1.3398). A multi-metric optimization framework balancing angular sensitivity, figure of merit (FoM), resonance depth, and linewidth is employed to identify balanced sensor designs. For the generic sensing interval, the optimal balanced lead-free design uses Cs2AgBiCl6 with Ag/Cs2AgBiCl6/HfSe2 thicknesses of 47/3.0/1.0 nm, achieving a sensitivity of 397.0 deg RIU−1 and a FoM of 103.66 RIU−1. For the bio/SARS-like interval, the optimal balanced design switches to CsSnBr3 (43/2.5/1.0 nm), providing 447.8 deg RIU−1 sensitivity and a FoM of 90.05 RIU−1, while MAPBr3 remains the highest-sensitivity material overall. Complementary Monte Carlo robustness analysis, local machine-learning surrogates, electric-field visualization, and layer-isolation studies are used to interpret the optimized sensor response. Our results demonstrate that optimal material selection is regime-dependent: Cs2AgBiCl6 performs best for generic aqueous sensing, whereas CsSnBr3 is superior for bio-relevant indices, providing application-specific design guidelines for lead-free SPR biosensors.
SN - 2214-1804
ER -