Separation on porous media

Abstract

The separation work in this thesis focuses on adsorption of riboflavin (RB) or Maillard browning colorants from lactose process stream using porous adsorbent materials. The aim was to look at the suitability of using adsorbent resin to decolorise lactose solutions and how it performs compared to activated carbon (AC), which is more commonly used in lactose purification process. The work presents an approach to other separation problems. Adsorbent resin FPX66 was selected and tested to remove RB from whey permeate model solution of 8.5 mg/L (± 1 mg/L) RB in water or in 15% lactose. The RB was adsorbed in stirred and agitated batch vessel at 30 to 60 °C and CAR (mass of resin/mass of RB) ranged from 0.28 to 0.89 g/mg. Adsorption in glass chromatography columns of different packed bed dimensions was undertaken by batch mode (recirculation of feed), continuous mode (feed passes once through column), and at different flow rates. Riboflavin content was estimated using a modified HPLC method or spectrophotometry. Lactose content was monitored using refractometry or a HPLC lactose anomers analysis method. Adsorption data from 21 experiments yielded an equation, which allows predicting the equilibrium RB concentration in the resin and adsorption efficiency from a known CAR. Kinetic and diffusion analysis measured the pseudo-first-order rate constant (k1), pseudo-second-order rate constant (k2), continuous run rate constant (kc), film diffusivity (Df), intra-particle diffusivity (Dp) and full uptake curve diffusivity (D). Results showed that most of the adsorptions in stirred and agitated vessels were predicted better by a pseudo-second-order model while those in the columns fitted better to a slower pseudo-first-order model. Slower adsorption in the columns was attributed to the limitation of resin movement in a packed bed. The faster adsorptions were ahead during the initial stages but most of all experiments reached similar equilibrium stages. Factors that promote higher adsorption and diffusion rates the most were higher CAR, greater resin movement, shorter bed height and higher flow rates. In almost all cases, film diffusivity was the rate-limiting step in the adsorption processes. Film diffusivity in continuous adsorption was greater than in the batch adsorption due to the higher driving force typical of continuous adsorption. For future work, optimisation problem can be set up for selecting between continuous and batch column modes. Regeneration case study showed that resin adsorbed with RB could be regenerated using 20% ethanol solution in downward column flow. Single-pass mode gave far higher regeneration efficiency than batch. Further, polynomial equations allow prediction of the volume of regenerant needed to achieve a certain degree of desorption at flow rates ranging from 9 to 38 mL/min. The resin and AC both performed similarly in decolorising lactose process streams that contain RB such as decalcified whey permeate, but the AC could decolorise streams containing Maillard browning colorants, such as mother liquor, better than the resin. This was attributed to the higher porosity of AC, which suggests its higher adsorption capacity. The systematic and theoretical based approaches used in this study can be used for testing the feasibility of adsorbent media in various applications, not limited to decolorisation of lactose solutions.

Date of Award	2009
Original language	English