A Statistical Modeling Approach for QCM TGA Mass Spectrometer Noise Reduction

The identities and outgassing rates of contaminants associated with a material determine its suitability for space applications. Thermogravimetric analysis (TGA) is one test commonly used for evaluating these material properties. During a TGA, contaminants deposited on quartz crystal microbalances (QCMs) are desorbed through heating while mass spectrometer (MS) data is collected. Three factors contribute to noise and artifacts in the MS data: (a) randomness in QCM outgassing flux, (b) MS measurement noise, and (c) constant chamber background contaminants. We present a two-step noise reduction approach that addresses these sources. First, we use QCM data to determine the number of outgassing species and kinetic parameters governing their desorption. Then, we apply these parameters to fit a linear statistical model to MS data, accounting for variance across the tested discretized mass spectrum. Once the variance is known for each mass bin, we use an adapted N-sigma method to isolate signal from noise. Our approach effectively reduces all three types of noise, improving confidence and efficiency in species identification and enabling MS-based modeling for isothermal outgassing kinetics. Although our analysis relies on the relationship between QCM and MS data, it may be applicable to other test procedures taking MS data concurrently with a measured source of mass flux.