Tackling Multidimensional Multi-polydispersity

I work at DSM, a company active in health, nutrition and materials. Being responsible for the competence field “Molecular characterization of synthetic polymers,” two-dimensional separation techniques and developments are very relevant to my work. To meet increasing customer demands, our materials have become gradually more complex – and for synthetic polymers, that means we are dealing with increasingly complex sets of molecules that are ‘multi-polydisperse.’ To understand polymer properties, we have to gain good knowledge of the exact molecular constitution of these molecular assemblies, which means increasing the resolution of our analytical toolbox. As polydispersity can only be determined by separation-based techniques, the issue of ‘multi-polydispersity’ can only be tackled by coupled separation techniques – and for polymers, that means 2D-LC predominantly.

Nowadays, the main interest in 2D-LC analysis of synthetic polymers is related to new products and trials that come from research. Typically, we are dealing with incidental questions for a very limited number of samples. Here, it is important to quickly generate insights and answers to facilitate the next step in synthesis research. However, this is typically where on-line 2D-LC in its current status fails.

In 2D-GC some kind of universal approach can be used; for example, a long (high plate number) ‘universal’ apolar column in the first dimension and a more polar one in the second dimension. Like all LC techniques, however, 2D-LC relies more on ‘playing around’ with selectivity. Depending on the type of question, the phase systems and the order of the systems in the first and second dimension need to be adapted.

Free choice of the order of the separation is, however, strongly limited, because the solvent from the first dimension often interferes in the second dimension. There is still nothing like a universal interface as in 2D-GC, where the solvent from the first dimension is eliminated. The solvent interference also requires a lot of system optimization when practical parameters are changed, even when the separation order remains the same. In such situations, the traditional off-line approach is still more efficient: semi-preparative fractionation in the first dimension and reinjection (after elimination of the first-dimension solvent and some further sample prep) on the second dimension. However, this also means some quite labor-intensive work, especially for some new polymer systems that require more exotic separation conditions, such as high temperatures, toxic eluents and the use of salts in, for example, SEC. All of this makes fractionation and especially sample prep of the fractions even more complex and labor intensive... Nevertheless, in practice, off-line 2D-LC still remains more efficient – especially since obtained fractions are also available for other types of analysis, such as NMR, IR or DSC.

A universal issue that needs attention is quantification. Analytical tools only become of real value when quantitative analysis is possible and when methods become robust. However, dealing with 2D-data is much more complex than traditional data. For synthetic polymers, hardly any examples of real and full quantitative evaluation of on-line 2D-separations are known, let alone validated. For instance: does the order of separation influence the analysis result or not? And: to what extent do marginal plots from 2D-data (that is: the sum of the obtained 1D-data) meet the results from 1D data, or not?

We need a combined effort of industry and academia to solve the above issues. This is especially true for the world of synthetic polymers, where the commercial interest (the volume) is much lower than in the case of life sciences. Fortunately, I see a gradual recognition of the themes addressed above that are resulting in concrete research projects from groups like Peter Schoenmakers’ at the University of Amsterdam.