Evaluating the Experimental Uncertainty in Gas and Vapor Sorption/Adsorption Measurements: Fundamental Considerations and Experimental Design Implications

Abstract

High-quality, accurate, and reliable sorption and adsorption data provide the basis for designing large-scale, energyefficient membrane/adsorption processes for gas and liquid separations and evaluating their techno-economic feasibility. As highlighted by Prof. David Sholl during his plenary lecture at the 2020 NAMS (North American Membrane Society) conference, the lack of uncertainty associated with published sorption/adsorption data represents one of the major roadblocks to progress in separation technologies. In this study, a standard methodology to estimate the uncertainty associated with sorption/adsorption measurements is proposed. A systematic analysis of the experimental uncertainty of gas and vapor sorption/adsorption measurement in polymers using the barometric method is performed in a variety of operative conditions, to individuate which factors contribute the most to the overall uncertainty under different experimental operation modes. This effort should push researchers in the field to adopt a standardized method to estimate the experimental uncertainty, which is expected to make experimental data coming from different laboratories in the world easier to compare. Finally, the validity of the linear error propagation method, generally used to evaluate the uncertainty of sorption/ adsorption data, is demonstrated by a vis-a-vis comparison with the Monte Carlo statistical method. Fundamental aspects and practical implications are highlighted and discussed.

Predicting the solubility of gases, vapors, and supercritical fluids in amorphous polymers from electron density using convolutional neural networks

“…This practice enables straightforward filtering of excessively noisy data once a desired threshold is established. However, confidence intervals are often poorly estimated or omitted, as discussed by Box et al 337 It's also worth noting that temperature or pressure uncertainty, while usually almost negligible, can still be significant. For example, if a sorption isotherm is highly concave towards the sorbed amount axis at high vapor pressures, a minor shift in pressure could result in a substantial change in the predicted value.…”

Predicting the Solubility of Gases, Vapors, and Supercritical Fluids in Amorphous Polymers from Electron Density using Convolutional Neural Networks

“…Hence once the desired threshold is set, filtering out excessively noisy data is straightforward. Though, the confidence interval is often estimated poorly or even omitted; for a discussion, see, for example, Box et al 324 There are also factors with huge indirect impacts. For instance, the sorption isotherm may become extremely concave towards the sorbed solvent amount axis at the high vapor activities of (good and condensable) solvents.…”

Predicting the Solubility of Gases, Vapors, and Supercritical Fluids in Amorphous Polymers from Electron Density using Convolutional Neural Networks

“…for a discussion, see, for example, Box et al 330 There are also factors with huge indirect impacts. For instance, the sorption isotherm may become extremely concave towards the sorbed solvent amount axis at the high vapor activities of (good and condensable) solvents.…”