Earthquakes can be destructive, killing many people and causing great financial loss. Costs from earthquakes are increasing rapidly, and thus seismic research is an important tool for reducing loss. Traditional studies of earthquakes have mainly been based on geophysics (seismic waves, geomagnetism, and geoelectricity), hydrogeology, and geogas. However, earthquakes can also release numerous particulates, which can be carried by ascending gas to reach the surface. These particulates provide us with abundant and direct information about earthquake activity and can serve as a new medium for studying earthquakes. In this study, we first find, using a transmission electron microscope (TEM), that ascending gas in the soil of the 2013 Ya'an earthquake region carries many particulates. These particulates are usually 20–600 nm and are composed of amorphous crystal, monocrystalline, and polycrystalline structures. In addition, these particulates contain O, Na, Mg, Al, Si, P, S, Cl, K, Ca, Cr, Fe, Zn, Ba, Hg, Ti, Mn, Ni, Cu, Cd, Sn, Sb, Cs, Pt, and Pb. Compared to the background areas (Yuhe and Hongzi Village and the Fengguanshan inactive fault), particulates carried by ascending gas in the Ya'an earthquake region have many abnormal elements (element associations), including Ti, Mn, Ni, Cd, Sn, Sb, Pt, and Pb. This finding confirms our hypothesis that earthquakes can release particulates carried by ascending gas that reach the surface. In addition, the elements occurring in these particulates indicate that earthquakes can trigger element activation and migration in the form of particulates. As particulates carried by ascending gas in the earthquake region are significantly different from those in the background areas, they can be used to predict earthquakes. The particulates in our study are significantly different from particulates carried by ascending gas from ruptures caused by the Wenchuan earthquake; therefore, we first propose that particulates in ascending gas from earthquake regions may exhibit a close relationship with the magnitudes and hypocentre depths of earthquakes and can be used to predict earthquakes with larger magnitudes and deeper hypocentre depths.