Samenvatting

Gas-phase diffusion is a widespread process in nature, for example diffusion of pollutants in the atmosphere. A special case of diffusion is the so-called Knudsen diffusion where a gas diffuses in the molecular flow regime. The theory for Knudsen diffusion is well-known therefore, diffusion experiments might be used to calibrate a new high-resolution gas source mass spectrometer. This type of diffusion occurs when the mean free path length of the molecules is larger than the holes it diffuses through. The question that will be answered in this thesis is: How does diffusion in the molecular flow regime change the isotopic composition of a gas? To study this research question experimentally, a diffusion set-up was used. The diffusion set-up is designed for investigating Knudsen diffusion. Subsequent to optimizing the diffusion set-up, fractionation factors and λ values at a mean free path length of 726 µm were obtained. For the residue samples, the fractionation factors obtained were: 1000 ln (34α) = (-31.7 ±1.7 (SE∙t0.95))‰ and 1000 ln (33α)= (-15.8 ±0.6 (SE∙t0.95)) ‰. The λ value, derived from these results is 0.50±0.01(SE∙t0.95). On the diffused side the following fractionation factors were obtained: 1000 ln (34α) = (-29.3 ±2.6(SE∙t0.95))‰ and 1000 ln (33α)= (-15.7 ±0.7(SE∙t0.95)) ‰. The corresponding λ value is: 0.533 ± 0.05(SE∙t0.95). The measured flow rate is (0.225 ±0.08) (SE∙t0.95) mL/h for 100 holes. Diffused gas samples (mL amounts) can be prepared in 1.5h for mass spectrometric calibration purposes. The change in isotopic composition during Knudsen diffusion corresponds to the theoretical prediction (within a 1 𝜎 error), for both the residual- and diffused gas samples at a mean free path length of 726 µm. Increasing the mean free path length resulted in fractionation factors closer to the predicted value. The reason is that the diffusion approaches a pure molecular flow regime at lower starting pressures (higher mean free path lengths). At lower mean free path lengths the occurring flow might be an intermediate of viscous and molecular flow. The results have a large error in the calculated fraction of the gas. Decreasing this error would also result in a better estimation of the fractionation factors. For all mean free path lengths, the observed flow rate is about ten times lower than the expected flow rate. The reason for this deviation is unclear, it might be related to an erroneous estimate of the diameters of the holes.