Virtual impactors are widely used for particulate matter (PM) classification due to their advantages of small cut-off particle size, simple structural design, ease of operation, and high particle handling capability, enabling subsequent analysis based on the desired aerodynamic diameter. Existing studies have mainly focused on the effects of particle size and structural parameters on classification performance, whereas systematic investigations into the regulatory mechanisms of fluid medium properties and ambient temperature variations on cut-off particle size remain relatively limited. Particularly under low-temperature gas conditions, variations in gas dynamic viscosity may significantly influence the dynamics of inertial particle separation, thereby altering the classification performance of virtual impactors. In this study, a low-temperature carbon dioxide-driven virtual impactor is proposed. By regulating the physicochemical properties of low-temperature gas, effective control over the particle inertial separation process is achieved, thereby expanding the tunable range of classification performance in virtual impactors. Numerical simulation results indicate that under low-temperature CO2 conditions, the virtual impactor can achieve a cut-off particle size classification capability of approximately 1.8 μm for fine particles. Under identical channel dimensions, a comparative analysis between conventional rectangular main channels and trapezoidal main channels was conducted, quantitatively showing that wall loss decreased from 44% to 24%. Based on the trapezoidal main channel configuration, further parametric studies on the horizontal inlet geometric dimensions were performed, revealing their influence on separation efficiency and wall loss. To validate the reliability of the numerical simulation results, particle separation experiments were conducted using polystyrene microspheres with particle sizes of 2 μm and 5 μm. Experimental results demonstrate that the virtual impactor can achieve stable particle separation and confirm the reliability of simulation-predicted particle classification trends. The results further show that, when driven by low-temperature CO2 combined with trapezoidal main channel structural optimization, the cut-off particle size of the virtual impactor decreases by approximately 26%, from 2.5 μm to about 1.8 μm. The trapezoidal channel structure significantly reduces particle wall loss under specific cut-off particle size conditions, while the low dynamic viscosity characteristic of low-temperature CO2 lowers the internal gas temperature environment of the microchannel, thereby improving inertial particle separation efficiency.
Zhao et al. (Fri,) studied this question.