In this paper, the spontaneous Rayleigh-Brillouin scattering spectra of air are simulated to study the effect of uncertainties of pressure, temperature, scattering angle and the characteristic parameter uncertainty of the Fabry-Perot interferometer on the accurate measurement of the bulk viscosity. It is found that those uncertainties have an obvious impact on the bulk viscosity measurement deviation and the bulk viscosity can be measured accurately under higher pressures (≥3.0 bar). In order to obtain the accurate bulk viscosity of nitrogen, oxygen and air, the spontaneous Rayleigh-Brillouin scattering spectra are measured with the wavelength of 532 nm under pressure of 4.0-7.0 bar and at temperature from 289.0 K to 400.0 K. The linear relation between the measured bulk viscosity and temperature is established with R 2 being above 0.99 for nitrogen, oxygen and air respectively. By comparison, it is found that our measured bulk viscosities mostly agree with the reported values obtained by spontaneous Rayleigh-Brillouin scattering, coherent Rayleigh-Brillouin scattering, ultrasonic determination or theoretical calculation for nitrogen, oxygen and air within 3σ results at the same temperature. The factors arousing the differences between them are attributed to the obvious measurement error and the measured uncertainty of the bulk viscosity under low pressures and the defects in the theoretical model itself. The empirical formula for calculating the bulk viscosity for air from pure components is proposed and it can match our measured results well.INDEX TERMS Rayleigh-Brillouin scattering, bulk viscosity, empirical formula.

The analysis of thermomolecular pressure difference experiments to obtain information on thermal transport properties is critically examined. It is concluded that the finite difference form of the dusty-gas equation, which has been used to analyze the thermal transpiration phenomenon, is internally inconsistent. A form of the integrated dusty-gas equation developed in this paper, while more difficult to use than the finite difference equation, yields results for the Eucken factor which are in better agreement with direct measurements of this quantity than are obtained with the approximate equations. It is observed, however, that when using the integrated dusty-gas equation there is a temperature variation of the apparent capillary diameter which cannot be explained.

The spontaneous Rayleigh-Brillouin scattering spectra of CO 2 are measured at different temperatures from 278 K to 369 K corresponding to the pressure range over 4-7 bar approximately and compared with the Tenti S6 model. The values of the bulk viscosity of CO 2 at different pressures and temperatures are obtained by the principle of minimum value of χ 2 (normalized chi-square error) at the gigahertz frequency. It shows that the bulk viscosity of CO 2 increases linearly with temperature, with a slope of (5.52±0.21)×10 −8 kgm −1 s −1 k −1. Meanwhile, the differences in the measured bulk viscosity of CO 2 and the ratio of bulk viscosity to shear viscosity between our work and the reported values by spontaneous or coherent Rayleigh-Brillouin scattering and theoretical calculation are analyzed. It is found that the changing of the ratio of bulk viscosity to shear viscosity with the temperature has the same tendency as the theoretical calculation, and the ratios agree with most reported values at the same temperatures, and the function between the ratio of bulk viscosity to shear viscosity and temperature is determined. In order to assess the accuracy of the obtained bulk viscosity in experiment, the bulk viscosity of CO 2 predicted by the obtained relation is used as a known parameter for the theoretical model to retrieve the temperature of CO 2 based on the measured SRBS spectrum under different pressures and temperatures. The absolute errors between the reference temperatures and the retrieved temperatures are less than 2.0 K. This result demonstrates that the obtained linear relationship between the temperature and the bulk viscosity of CO 2 is credible not only under relative higher pressures (4 bar≤ p≤7 bar) but also under lower pressures (p<4 bar).

The results of capillary thermal transpiration measurements for nitrogen, oxygen, carbon monoxide, methane, methyl chloride, hydrogen sulfide, and sulfur dioxide over a (mean) temperature range of 408.6 to 501.3°K are reported. The problem that arises in interpreting the data to obtain thermal transport information is carefully considered. While this difficulty is not completely resolved, our data for the nonpolar gases are combined with our earlier results and used to compute smoothed values of the rotational relaxation numbers and Eucken factors for the temperature range l6S-S00'K.