High viscosity industrial chromatography for mild food fractionation

Anton Schultze-Jena

Research output: Thesisinternal PhD, WU


Agro-food streams, present in large volumes, contain compounds attractive for food industry, if they are separated from unwanted components. Recovery of such components offers a large potential for industrial applications. A trend towards enriched fractions, rather than purified products, enables sustainable process design via chromatographic separation. Chromatography has the potential to fractionate these agro-food streams at large industrial scale, while maintaining functionality and operating in a sustainable and economical manner. The separation mechanism can be fine-tuned to target specific molecular characteristics. Because the streams to be processed are so large, large equipment is required, which renders the process economically unfeasible. Process economics can be improved by reducing the size of the chromatographic installation, which is directly dependent on the volume to be processed. The stream to be processed can be reduced by increase of concentration, which leads to higher viscosity. The aim of this thesis is to understand the interplay between feed viscosity, mass transfer resistance, pressure drop and eventually productivity and water use of a chromatographic system.

In the design of industrial chromatographic processes, both experiments at lab scale and model calculations are used. Experiments are performed to determine model parameters and to validate the model. Subsequently the model is used to scale-up. Experiments require, in addition to the column, the use of further equipment (connectors, tubing, valves, detectors, etc.), all of which add to extra-column volume. The contribution of the extra-column volume to peak broadening is often neglected assuming that by doing so the column efficiency is underestimated and scaling up results in an oversized system. Chapter 2 addresses method development for measuring the mass transfer resistance in a chromatographic column in such a way that the data can be used to scale up to industrial scale. It was shown, that correction for the extra-column contribution to band broadening must be made in many cases, otherwise the column efficiency is overestimated and subsequently industrial designs under-dimensioned.

In chapter 3, the main hindrance to mass transfer is identified and measured: the intraparticle diffusivity. In industrial applications, generally stationary phase particles of large diameters are used. Large particle sizes facilitate operation at a low pressure drop, but they introduce large characteristic lengths that the molecules have to travers by diffusion. It is difficult to predict intraparticle diffusivity and therefore measurement for each new molecule/stationary phase combination is required. In an attempt to link and quantify pore and molecule characteristics, intraparticle diffusivities, pore diameters, porosities, and tortuosities are measured in various stationary phases with a selection of (relatively) small molecules. Two models from literature that relate stationary phase properties to intraparticle diffusivity, are tested for their predictive quality, but give unsatisfactory results. By modifying one of the models with the addition of the accessible pore fraction, calculated from inverse size exclusion measurements, a good fit of all intraparticle diffusivities for all stationary phases and all molecules is observed. This improved model can serve as a predictive tool for intraparticle diffusivity.

Implications of using a concentrated feed with an increased viscosity on mass transfer resistance inside a chromatographic column is described in chapter 4. Dependence of intraparticle diffusivity on viscosifiers ability, those molecules in the mobile phase which contribute to viscosity, to penetrate pore volume is shown. The ability to penetrate pore volume is expressed by the partition coefficient KD. These results are used to determine the window of operation for viscous feed streams. It is shown that diluting highly viscous feed streams prior to chromatographic separation, should go no further than approximately 2.5 mPa·s. If the feed stream is diluted to lower viscosities, column volume will increase. Diluting feed streams from 8 mPa·s down to around 2.5 mPa·s shows little influence on column volume, but column dimensions change, with a tendency to get narrower and longer as viscosity decreases.

In chapter 5, knowledge gained from working at high viscosities in single column systems is challenged on a multicolumn simulated moving bed SMB system with tomato serum as feed and ion-exclusion as separation mechanism. The aim was obtaining a γ-aminobutyric acid enriched and monomeric sugar strapped fraction. The behavior of the sugars for different feed viscosities is calculated well by the model when the ratio of feed to eluent is used as dilution factor. The behavior of γ-aminobutyric acid in ion-exclusion chromatography is highly concentration dependent and the recovery is not calculated with accuracy. The SMB at the two higher feed viscosities (2.5 and 4 mPa·s) outperforms the SMB at the lower feed viscosity (1 mPa·s) both in terms of water use and productivity.

Finally in chapter 6 the main findings and conclusions are discussed. Further addressed are the use of temperature to reduce viscosity and an assumption used in this thesis, that the required number of theoretical plates remains constant when comparing designs for various feed concentrations. At the end of this chapter an outlook is given on future perspectives for the effort of decreasing chromatographic system size.

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
  • van der Padt, Albert, Promotor
  • Janssen (FPE), Anja, Co-promotor
  • Boon, Floor, Co-promotor
Award date9 Oct 2019
Place of PublicationWageningen
Print ISBNs9789463950305
Publication statusPublished - 9 Oct 2019


Dive into the research topics of 'High viscosity industrial chromatography for mild food fractionation'. Together they form a unique fingerprint.

Cite this