Diffusion is omnipresent in (bio)chemical processes, such as rectification and liquidliquid extraction, laminar mixing, or generation of stable concentration gradients for cell-based studies in microfluidics. While technical mixing processes proceed on the macroscale, diffusion is the dominant mixing process on the microscale. Since diffusion is often rate-limiting, knowledge of diffusion is fundamental for the design of microfluidic applications,but also to understand industrial processes and environmental phenomena.
However, diffusion data is scarce. While the Dortmund Data Bank contains vapor-liquid equilibria on more than 9000 binary systems, diffusion data is only available for little more than 2000 binary systems. Moreover, the number of available diffusion data decreases one order of magnitude per additional component in the system. As technical systems contain more than two components in general, the lack of data on multicomponent systems is severe. Additionally, most multicomponent diffusion data is on aqueous mixtures only. If diffusion data is not available for a specific system, there are basically three options to generate data: molecular dynamics, engineering models, and experiments. Diffusion coefficients can be fully predictively computed from molecular dynamics (MD). This approach shows promising results for the Maxwell-Stefan (MS) and Fick diffusion coefficients if suitable force-field models are available. Engineering models allow a rapid prediction of diffusion coefficients with moderate uncertainties. Recent Darken-based models have a physically sound basis and improve prediction accuracies of Maxwell-Stefan diffusion coefficients in ideal multicomponent mixtures.
Generally, additional data expressing thermodynamics is required to obtain the predictively relevant Fick diffusion coefficient from the Maxwell-Stefan diffusion coefficient. Still, both MD and engineering models contain large uncertainties. Experimental
multicomponent diffusion data are required if high accuracy is required. This data is also needed for validation and development of theoretical methods. Thus, experimental methods are required that provide accurate and precise values for diffusion coefficients in a short time and with little effort. To date, classical diffusion experiments allow for high accuracy and precision, but classical diffusion experiments are also time-consuming, laborious and require at least nc – 1 experiments for a multicomponent system with nc components. In practice, multiple sets of (nc – 1) experiments are performed. Thus, an efficiently designed diffusion experiment is desirable. For this purpose, this work combines Raman spectroscopy and microfluidics: Raman spectroscopy provides independent concentration information on nc1 species in a mixture from each measurement with high accuracy and high spatial resolution. Consecutive, temporally and spatially resolved Raman measurements result in comprehensive concentration gradient information so that one diffusion experiment is sufficient to determine multicomponent diffusion coefficients. Microfluidics enables short diffusion experiment time (<1 h), since the diffusion time is proportional to the squared diffusion distance. Accordingly, binary diffusion coefficients have successfully been measured using microfluidics. Also, it has repeatedly been hypothesized that microfluidics allows for the measurement of multicomponent diffusion coefficients.