This paper presents a gas source localization algorithm based on infotaxis gradient driving, aimed at detecting pollutant gases in electronic cleanrooms. By combining information entropy and concentration gradient techniques, the algorithm enables robots to accurately locate pollutant sources within cleanroom environments. The study first establishes a turbulence diffusion simulation model for gaseous pollutants using Fluent, providing experimental conditions for subsequent robot-based source localization under various wind directions, wind speeds, and gas source positions. Furthermore, to address the gas source localization problem, an information gradient-driven algorithm is proposed, enabling the identification of gaseous pollutant sources in a turbulent environment without relying on wind direction or wind speed information. Experimental results show that the proposed information gradient-driven algorithm achieves a success rate of 96%, significantly improving the efficiency and accuracy of pollutant gas monitoring, and holds significant application value for pollution control in cleanrooms.