Ultra-high-performance monitoring of deuterium in drinking water via 1D ternary photonic crystal: A study on roughness and fabrication tolerance.

This work reported the development of an ultra-high-performance optical detector based on a one-dimensional ternary photonic crystal (1D-TPC) for the precision monitoring of deuterium weight percentages in drinking water. Meanwhile, the proposed architecture utilizes asymmetric periodic multilayered structure of N (Si/SiO2/ZnSe) unit cells with a central porous silicon (PSi) defect layer, optimized via the Transfer Matrix Method (TMM). The numerical results investigate the emergence of a resonant mode at 575.56 nm at normal incidence, deuterium concentration of 0 wt%, and defect layer thickness of 1.8 μm. A distinct shift in the spectral position of the defect mode is expected, driven by changes in the effective refractive index of the porous silicon (PSi) layer when infiltrated with the microfluidic analyte (drinking water). Notably, variations in the deuterium concentration of the drinking water alter the refractive indices of both the water and the PSi layer. Furthermore, an extensive optimization procedure was carried out to maximize the performance of the proposed sensor. By increasing the refractive index grading of the ternary layers, the optimized sensor achieves high-performance metrics, including a sensitivity of 310.52 nm/RIU, an ultra-high-quality factor (Q) of 1.62 × 107, and an exceptional figure of merit (FoM) of 6.14 × 106 RIU-1. The detection limit is pushed down to 1.125 × 10-8 RIU (≈2.46 ppm), enabling the identification of minute isotopic variations across a full deuterated water concentration range of 0 -100 wt% with a dynamic range of 78262. In order to ensure fabrication viability, a rigorous Monte Carlo analysis with 1000 iterations has utilized integrated with Debye-Waller surface roughness modeling, revealing a remarkable self-compensating effect with a mean resonance shift of only -0.80 nm under significant manufacturing fluctuations. Furthermore, a robust, and highly accurate multi-variable empirical formula with a remarkably high coefficient of determination R2 = 0.99898 and an exceptionally low root-mean-square error RMSE = 1.25, facilitating a direct and efficient readout for real-time monitoring. These findings underscore the sensor's potential for transformative applications in environmental hydrology, metabolic medicine, and nuclear isotope monitoring, offering a cost-effective and stochastically stable platform compatible with standard CMOS nanofabrication.