This work was aimed at producing a proof-of-concept temperature-responsive chromatography column prototype. Triply periodic minimal surface (TPMS) -based 4% and 6% virgin agarose sheet structures were manufactured using inverse templating method with sacrificial wax molds. Production methods were streamlined, with build wax dissolution optimised by using proprietary Bioact™ VSO solvent in 70-75 oC region. Large channel size (𝐷ℎ ~1000 μm) structures (outside diameter of 15 mm, 70 mm long, with 50 mm long temperature-controlled region) were repeatedly manufactured. Columns with 𝐷ℎ = 500 μm were able to be fabricated, but with poor reproducibility. Columns with multiple heating jackets were shown to be successfully manufactured.

Agarose cross-linking via addition of 2,3-dibromopropanol yielded structures that displayed excellent thermal and mechanical stability. Addition of temperature-responsive ligands for cation exchange chromatography was achieved via epichlorohydrin activation, followed by amine capping, free-radical immobilisation of 4,4’-azobis(4-cyanovaleric acid), and graft polymerization of poly(NIPAAm-co-tBAAm-co-AAc-co-MBAAm). Cross-linked 6% agarose cubes were functionalized, with the FT-IR spectra displaying the expected peaks. Introduced epoxide densities of 700 ± 70 μmol/g dried support (42 ± 5 μmol/g drained support) were calculated, with predicted ‘pNIPAAm + tBAAm’ co-polymer yields of ca. 4000 μmol/g dried support. Thermoresponsiveness was displayed, with qmax,50 oC/qmax,5 oC ratio of ~1.43 and lactoferrin (LF) binding capacity of qmax,50 oC ~16 mg/mL for cut cubes (for available binding surface area of 3,000 – 6,000 mm2/mL, pores excluded). Predicted binding rates of 35 mg and 89 mg of LF were calculated for 500 μm and 200 μm channel size columns (chromatographic length of 50 mm and channel network diameter of 7 mm) respectively. Temperature-responsive cubes were calculated to have ~17% smaller volumes on average.

Cross-linked TPMS structured displayed shrinking of 38% post- dimethylformamide immersion step, with the final structures not exhibiting thermoresponsive behaviour, most certainly due to diffusion limitations when functionalising of the structure. This volumetric loss is thought to be dictated by the surface area in contact with the solvent. Altering the nature of the hydrogel via changing the degree cross-linking might improve the stability in organic solvents and further research is recommended.

Column holder was designed and produced out of Nylon, stereolithography resin and stainless-steel holder, with the latter allowing for best column fitting. Design and manufacture of custom thermoplastic polyurethane flow distributors lowered channel mixing from 3-8% to 0.40 ± 0.05% with respect to the middle channel when re-circulating the fluids at 2 mL/min for 8 hours. The leak was shown to be independent of the flow direction. Heat transfer experiments were then conducted and oil flowrates of 10.0 ± 0.7 mL/min and 13.3 ± 0.7 mL/min resulted in 12.5 ± 0.1 oC and 11.3 ± 0.1 oC middle channel temperature rise (flowing at 2 mL/min) and displayed heating rates of 0.62 ± 0.04 oC/min and 0.57 ± 0.03 oC/min respectively.

However, high flows resulted in amplified leakage of ~4% whereas 4.5-7.5 mL/min flowrates demonstrated leakage of 0.91-1.29%. A likely reason for this is increased pump suction and potential damage caused to the hydrogel wall structure. Considerable heat losses were displayed, with the metal casing and silicone oil-carrying tubing being the main contributors. System heat losses were quantified, with propagated uncertainties of 17-21% displayed. Calculated losses were within 5-33% (average 22%) of observed losses, with most within the uncertainty range. Despite these significant heat losses, the temperature profile of the middle channel closely followed that of the oil stream exiting the structure which confirms the possibility of heat exchange between the two hydrogel channels.

This work should be taken as a proof-of-concept investigation with further research recommended.