Advisor:

Dr. Todd A. Sulchek, Ph.D. (Georgia Institute of Technology, Mechanical Engineering)

Thesis Committee:

Dr. Craig R. Forest, PhD. (Georgia Institute of Technology, Mechanical Engineering)
Dr. Wilbur A. Lam, M.D. Ph.D. (Georgia Inst. of Technology | Emory University, Biomedical Engineering)

Dr. Manu Platt, Ph.D. (Georgia Inst. of Technology | Emory University, Biomedical Engineering)
Dr. Fatih Sarioglu (Georgia Inst. of Technology, Electrical and Computer Engineering)

Flow Rate Modulated Periodic Backflush to Improve Dead-End Filtration

Sorting modalities have been developed to achieve high enrichment factors, recovery rates, throughput, and purity. The negative downstream clinical impacts of labelled sorting mechanisms led groups to forego this option in favor of label-free methodologies. No device has yet to produce preferred results for all metrics, but some, especially filters, will realize most at the cost of one or two other metrics. Dead-end filters, specifically, are known for exceptionally large enrichment factors, purity, and throughput at the cost of recovery percentage or yield. Enabling filtration tools to eliminate or minimize these losses can have large impacts in diagnostic and therapeutic markets.

Pulse Width Modulation (PWM), a technique for controlling the proportionality of high to low signal (duty cycle), Pulse Amplitude Modulation (PAM), a technique for controlling the peak amplitude, and periodicity or frequency shifting of a square wave controlling volumetric flow rate and transmembrane pressure are shown to serve as novel techniques to increase recovery percentage in dead-end filtration systems while minimizing throughput tradeoffs. We employ these pulse modulation techniques to periodically backflush dead-end filters during sample processing for both biological and non-biological particulate suspensions and improve yield by interrupting cake formation, reintegrating fouling layers into the bulk of a sample, and improving permeate flux.

This thesis work initially investigates PWM theoretically and experimentally as a proof-of-concept for enabling the use of dead-end systems diagnostically and therapeutically. PWM backflush is then optimized through modulation of amplitude and frequency to minimize the costs to throughput. Finally, we apply the concept to conjugated microsphere recovery, bacterial isolation in a Cystic Fibrosis sorting model, and improve perfusion bioreactor scaffold seeding uniformity.