Cleaning wafer surfaces in situ using gaseous reagents facilitates automated cluster tool processing of microelectronic devices. Moreover, to remain within stringent thermal processing budgets for making ultralarge scale integrated (ULSI) circuit devices, it is desirable to perform these dry cleaning processes at low temperatures. We report here the use of attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy to study and optimize room–temperature plasma cleaning of native oxide contaminated Si surfaces. By monitoring the vibrational modes of Si-H, H-SiO, O-H, and C-H in real-time, we develop a hybrid process where an upstream microwave discharge and the gas-flow rate are used to control the neutral atom flux while direct exposure to a radio-frequency plasma is used to control the ion energy flux. To remove SiO2 from the Si surface, a high H2 flow rate (i.e., H flux) and weak RF plasma exposure (i.e., ion bombardment) are required. Hydrocarbon contaminants can be removed easily using the microwave plasma effluent alone. Too large an ion energy flux damages the surface, producing a large Si-H absorption peak which we attribute to formation of a-Si:H. These results demonstrate the power of real–time monitoring for developing and optimizing dry cleaning processes.