Date of Award

Fall 2006

Document Type


Degree Name

Doctor of Philosophy (PhD)


Micro and Nanoscale Systems

First Advisor

James Palmer


Challenges due to surface tension effects for vaporization in microchannels have restricted the implementation of conventional separation processes that involve liquid boiling in a micro-chemical system. However, membrane separations offer a feasible alternative for exploring the advantages of liquid phase separation in micro-chemical systems. The main objective of this research was to evaluate the process intensification effects in a microscale pervaporation process with experiments involving the separation of ethanol/water using a commercially available polymer dehydration membrane.

The microchannels in the microseparator were fabricated by the dry etch process at Louisiana Tech University. Microchannel depths of 20 to 120 μm, with hydraulic diameter in the range of 30 μm to 80 μm were studied. The feed ethanol solution was fed through the microchannels parallel to the direction of the membrane with an effective membrane area of 3 cm2 . A novel membrane module with an in-situ heating unit was developed in order to achieve controlled isothermal conditions. An elevated operating temperature and high vacuum at the permeate region provided the driving force required for separation. A laminar liquid flow regime with Reynolds number in the range of 8 to 180 existed in all experiments. Changes in the feed concentration that occurred as a result of selective removal of water by the membrane were measured at regular intervals using an ABBE Mark II refractometer. The total flux and selectivity as a function of residence time and hydraulic diameter was studied. The resistance-in-series model and Sherwood correlations were used for the prediction of mass transfer coefficients in the boundary layer. The overall mass transfer resistance and boundary layer mass transfer resistance were calculated. The experimental results indicated that the separation performance of the microseparator was independent of changes for the range of hydraulic diameter and residence time tested. The boundary layer resistance calculated was over three orders of magnitude lower than the overall mass transfer resistance, which corroborated that the membrane resistance was the dominant resistance for the transport process.