For thrust gas foil bearings integrating Supercritical Carbon Dioxide （S-CO2） as the working medium， a Back Propagation （BP） neural network algorithm is employed to propose a physical property model for S-CO2. To account for the non-idea gas behavior in the bearing， models encompassing gas lubrication with turbulence effects， foil structural mechanics， and the calculation of the average gas film temperature are introduced. Static and dynamic characteristics of the thrust gas foil bearing are studied and contrasted with various lubricant media. The impact of diverse structural parameters on the static and dynamic attributes of the gas foil bearing is analyzed. Results indicate that the physical property model presented in this paper attains high accuracy， boasting a correlation coefficient of 99.997%. The thrust gas foil bearing lubricated with S-CO2 exhibits enhanced load-bearing capacity， with potential for further improvement by adjusting the minimum initial gas film thickness or increasing the film thickness ratio within a suitable range. The dynamic stiffness and damping coefficient of the thrust gas foil bearing utilizing S-CO2 significantly surpass those employing air as the medium， underscoring its superior dynamic characteristics. Furthermore， a reduction in the minimum initial film thickness leads to a rapid increase in the dynamic stiffness coefficients and damping coefficients of the bearing.