Loss-Engineered Exceptional-Point Physics and Radiation-Force Particle Transport in a Sub-Wavelength Acoustic Cavity Chain

Fine particulate matter (PM2.5) is a critical air pollutant due to its ability to reach the deeper parts of the respiratory tract because of its small aerodynamic diameter (≤ 2.5 µm). Acoustic particle manipulation is a possible path to controlling airborne particles, rather than using only conventional filtration. The connection between non-Hermitian acoustic field engineering and particle-scale transport, however, is not well developed. This work aims to explore the potential of changing the acoustic spectrum, redistributing the acoustic energy, and affecting the motion of the particles in a finite chain of coupled sub-wavelength resonators with spatially graded acoustic losses.

A reduced-order framework is based on a non-Hermitian coupled-mode model, a one-dimensional interpolated acoustic pressure field, and a classical Gor'kov radiation-force potential combined with overdamped particle tracking. The two-cavity system is found to reproduce an analytical EP at r = 1, while the longer chains are found to show numerically identified pairwise EP-like candidates at r = 1. The first candidate is found close to r = 0.3254 for a ten-cavity chain, and the eigenvector overlap is greater than 0.9999. The calculated field-energy asymmetry with the increase of the loss gradient is 67.3% at r = 1.9, while the reference-particle fraction in the low-loss half only changes from 50.1% to 47.1%. The force landscape has several local reversals, which restrict the net directional redistribution. In addition, the eigenvalue gap between the first EP-like candidate grows rapidly when the frequency disorder occurs.
The results prove that the acoustic spectrum and the acoustic field distribution can be significantly changed by the loss engineering, while providing efficient particle transport is not a general rule for strong field asymmetry. Before separating the performance of PM2.5 can be realized in practice, geometry-resolved modelling and experimental validation will be needed.