Raman microspectrometry (RMS) is a useful single particle analysis method that can provide information on the mixing states, molecular species, and chemical functional groups of individual aerosol particles, which are difficult to determine by bulk analysis techniques. On the other hand, drawbacks, such as low Raman cross-section, spatial resolution (∼1 μm), and optical diffraction limit, make the analysis of atmospheric particles in the submicron size range difficult using conventional RMS. This study developed a new strategy to detect individual submicron-size atmospheric particles by combining dark-field (DF) microscopy and surface-enhanced Raman spectroscopy (SERS). The DF technique overcomes optical spatial diffraction limit by contrast enhancement, allowing the visualization of submicron particles. SERS facilitates spectroscopic characterization (obtaining information on molecular fingerprints and mixing states) of trace amounts of analyte by increasing the Raman scattering cross-section at the hot spot. SERS-active silver substrates sputter-coated on a Si wafer efficiently provided a clear background in the dark-field image and uniform hot spots over a large area, which were suitable for single-particle analysis. Various functional groups in individual particles and their heterogeneous mixing states were investigated, demonstrating the potential of this method to provide improved information on submicron atmospheric particles of femtogram-level masses. DF-SERS may elucidate the detailed physicochemical characteristics of individual submicron particles, providing new information on the formation mechanisms and fates of atmospheric particles.