Includes bibliographical references (p. 181-186). / Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 1997. / Vita. / by Isako Hoshino. / Ph.D.

Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 2000. / Vita. / Includes bibliographical references (p. 243-257). / by Darrell J. Irvine. / Ph.D.

Thesis (Sc. D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 1991. / Vita. / Includes bibliographical references. / by Jennifer Ann Lewis. / Sc.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 1996. / Includes bibliographical references (leaves 161-164). / by Satyavolu Srinivas Papa Rao. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 2010. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student submitted PDF version of thesis. / Includes bibliographical references (p. 99-106). / The throughput and efficiency of membrane separations make polymer filtration membranes an important resource for the pharmaceutical, food and wastewater treatment industries. Nanofiltration (NF) membranes fill an important niche between nonporous reverse osmosis membranes, which have comprehensive solute rejection and low solvent permeability, and porous sieving ultrafiltration membranes. However, challenges in NF membrane design remain outstanding. At the effective pore size of NF membranes (~0.5 nm-2 nm), both electrostatic and steric factors determine membrane selectivity. Most NF membranes are charged under a wide range of environmental conditions and thus preferentially exclude charged solutes. This charge selectivity precludes separation of molecules based solely on size. An additional limitation of NF membranes is the tendency to foul by adsorption of feed components. The purpose of this thesis is to demonstrate control of membrane selectivity in fouling resistant membranes via manipulation of the chemistry of a specific copolymer system, polyacrylonitrile (PAN)-based poly(ethylene oxide) (PEO) graft polymers. Previous work with amphiphilic graft copolymers as membrane materials has included PANg- PEO with an average graft length of 9 (PAN-g-PEO9). PAN-g-PEO9 was shown to have excellent fouling resistance as an antifouling additive in porous ultrafiltration membranes and as a dense selective layer coated onto a support base membrane-a thin-film composite (TFC) NF membrane. The comb morphology of the polymer imposes high interfacial area on the microphase-separated domains, resulting in a bicontinuous structure consisting of a glassy PAN matrix interpenetrated by PEO-filled "nanochannels" that can act as vias for water and small solutes (with a size cutoff of ~0.8 nm). It also presents a PEO brush on the comb surface which acts as a steric barrier to resist irreversible fouling of the membranes. The understanding from previous work on PEO comb NF membranes is that the pore size is determined by the nanochannel's size, i.e. the PEO domain size. Because the graft characteristics (spacing and length) of comb copolymers determine the domain size, it was expected that varying the graft length would allow broad, precise control of the size cutoff of the TFC membranes, a concept demonstrated previously with amphiphilic graft copolymer NF membranes of poly(vinylidene fluoride)-graft-poly(oxyethylene methacrylate) (PVDF-g-POEM). The first aim of this thesis was to tailor the retention properties of PAN-g-PEO TFC NF membranes by modifying the chemistry to tune the electrostatic and steric properties sufficiently to enable complex separations, particularly of solutes with high fouling potential. Comb copolymers incorporating ~40 weight % PEO with side chains varying from 5-40 EO units were synthesized by free radical methods and compared as selective-layer coatings on PAN UF membranes. 3 Membranes incorporating combs with 9 EO units or more were shown to resist irreversible fouling when challenged by a model protein feed solution (bovine serum albumin) for 24 hours. Fouling resistance was found to be compromised, however, upon exposure to acid (pH 2) solution, used to simulate chemical cleaning procedures in industry. Thickness-normalized permeabilities of these PAN-g-PEOn NF membranes exceeded those of commercially available NF membranes by approximately an order of magnitude. A systematic effect of side chain length on permeability was seen when varying temperature, ionic strength, and pressure. Contrary to expectations, the membrane size cutoff (~0.8 nm) for charged rigid molecular probes in deionized water was independent of the comb side chain length. This new finding can be explained by modeling the hydrophilic domains as opposing swollen polymer brushes of uniform density acting as a physical gel. The gel mesh size (distance between chains) is independent of side chain length, and controls the size cutoff in good solvent conditions matching those in which the membrane was equilibrated during fabrication. In poorer solvent conditions, a decrease in the brush height, progressing to complete collapse of the PEO gel, can be expected to create differentiation based on domain size (i.e. side chain length). This is consistent with the finding that retentions of dyes increased with decreasing side chain length in saline solution, as salt is known to reduce PEO-water miscibility. Fluorescently labeled peptides germane to proteomics research were filtered and both chromatographic and size-selective membrane behavior was observed-the first demonstration of size-based nanofiltration of peptides. Based on this finding, two different peptides of molecular weights 1.3kDa and 1.5kDa were fractionated to achieve a six-fold increase in the concentration of the larger peptide relative to the smaller peptide in two filtration steps. The electrostatic selectivity of the PEO comb membranes could also be varied. Terpolymers consisting of PAN-g-PEO with 1-2% charged sulfopropyl acrylate (SPA) or 5% N,Ndimethyl- N-(2-methacryloyloxyethyl-N-(3-sulfopropyl) ammonium betaine (SPE) were synthesized and coated onto PAN base membrane. The divalent salt (Na2SO4) retention of the resulting TFC membranes increased from ~20% for the PAN-g-PEO copolymer to ~45% and 82% for the SPE and SPA terpolymers, respectively. Retention of monovalent NaCl was substantially lower, characteristic of commercial NF membranes. The charged comb membranes did not completely resist fouling by a 1 g/L BSA solution, losing 2% of the initial flux after 24 h exposure. Forming a trilayer TFC, with a layer of PAN-g-PEO coated over a charged terpolymer, reduced membrane fouling compared to the charged layer alone. In summary, the goal of this study was to demonstrate control of membrane selectivity in fouling-resistant PAN-g-PEO NF membranes. An important finding was that the PEO gel created in the hydrophilic domains leads to similar size cutoffs over a wide range of side chain length. To access the desired spectrum of size cutoffs, the quality of solvent for the swollen PEO brush must be reduced. In spite of these limitations, the membrane was shown to have useful fractionating properties as demonstrated with labeled peptides of varying molecular weight. The retention of salts was enhanced by incorporating small amounts of charged monomer into the comb backbone, but at the expense of fouling resistance. / by Nathan Gary Lovell. / Ph.D.

Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 2003. / Vita. / Includes bibliographical references. / Airborne particles fall into one of three size ranges. The nucleation range consists of nanoparticles created from vapor atom collisions. The decisive parameter for particle size and composition is the supercooling of the vapor. The accumulation range, which comprises particles less than 2 micrometers, consists of particles formed from the collision of smaller primary particles from the nucleation range. The composition of agglomerates and coalesced particles is the same as the bulk vapor composition. Coarse particles, the composition of which is determined by a liquid precursor, are greater than 1 micrometer and solidify from droplets whose sizes are controlled by surface, viscous, and inertial forces. The relationship between size and composition of airborne particles could be seen in welding fume, a typical metallurgical aerosol. This analysis was performed with a cascade impactor and energy dispersive spectrometry with both scanning electron microscopy (SEM-EDS) and scanning transmission electron microscopy (STEM-EDS). Other methods for properly characterizing particles were discussed. In the analysis, less than 10% of the mass of fume particles for various types of gas metal arc welding (GMAW) were coarse, while one-third of flux cored arc welding (FCAW) fume particles were coarse. Coarse particles had a composition closer to that of the welding electrode than did fine particles. Primary particles were not homogeneous. Particles larger than the mean free path of the carrier gas had the same composition as that of the vapor, but for particles 20 to 60 nanometers, smaller particles were more enriched in volatile metals than larger particles were. This was explained by the cooling path along the bubble point line of a binary phase diagram. / (cont.) Particles were not necessarily homogenous internally. Because nanoparticles homogenize quickly, they may form in a metastable state, but will not remain in that state. In this analysis, the presence of multiple stable immiscible phases explains this internal heterogeneity. The knowledge contained herein is important for industries that depend on the properties of nanoparticles, and for manufacturing, where industrial hygiene is important because of respirable particle by-products, such as high-energy-density metallurgical processing. / by Neil Travis Jenkins. / Ph.D.

Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 1996. / Includes bibliographical references (leaves 104-105). / by Nicolas Joseph Avril. / S.M.

Thesis (B.S.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 1997. / Includes bibliographical references (leaf 36). / by Christopher Kantner. / B.S.

Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 2004. / Includes bibliographical references. / Magnetic separation is an actively researched field due to its broad applicability to the mineral, chemical, and biological industries. The objective of this work was to design, fabricate, and test systems to study magnetophoresis of particles in suspension. To achieve this goal, two system concepts were developed: an Alternating Field System and a Flow System. Both systems consisted of permanent magnets and miniaturized devices (separation chips), which integrated microfluidic channels with ferromagnetic core elements. The systems produced "sawtooth" magnetic fields that were combined with a long-range magnetic field or pressure- driven flow to bring about migration, focusing and trapping of nonmagnetic particles suspended in ferrofluid. A potential application of such systems is high-resolution, size-based separation of DNA, cellular organelles, viruses, and other like-sized biological entities. The systems were designed using finite element analysis and fabricated using IC/MEMS microfabrication techniques. The fabrication process for the separation chips realized a microfluidic channel and electroplating molds in a single layer of SU-8 photoresist on a glass substrate. Nickel core elements were electroplated into the molds, and a PDMS cover substrate was attached using a novel technique involving contact bond and heat cycling. The systems were tested via experiments using optical fluorescence methods to observe the concentration profiles of polydisperse suspensions of polystyrene beads. / (cont.) Alternating Field System involved simple migration under a long-range magnetic field, focusing under a sawtooth magnetic field, and attempted separation by combining the long-range and sawtooth fields. The most significant findings of the trials were the significant effect of particle- particle interactions and high sensitivity to the core design of the chip. The Flow System trials combined a sawtooth field with flow. The trials demonstrated size-based trapping of particles, where 840 nm beads were trapped earlier along a separation channel, while 510 nm beads were trapped further along. Moreover, the location along the channel at which particles of a given size were trapped was shown to be a function of flow rate. Size-based trapping in magnetic potential wells, as well as flow rate tuning, could form the basis of a high-resolution particle separation system. / by Edward S. Park. / S.M.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Materials Science and Engineering, 2008. / Includes bibliographical references (p. 165-172). / Ge virtual substrates, fabricated using Si1-xGex-.Ge, compositionally graded buffers, enable the epitaxial growth of device-quality GaAs on Si substrates, but monolithic integration of III-V semiconductors with Si CMOS using this platform is hampered by the large thickness of the Si1-xGex graded region. To address this issue, the Silicon on Lattice-engineered Silicon (SOLES) was developed, consisting of a silicon-on-insulator (SOI) structure fabricated on a Ge virtual substrate. Placement of the Si device layer at the surface makes it possible to process this platform similarly to typical SOI wafers, with the added functionality of a buried III-V template which can be used for GaAs device fabrication. This platform was fabricated using a scalable layer transfer technique. AlInGaP LEDs were also demonstrated on a SOLES substrate. In addition, an alternative growth process was investigated for Si1-xGex virtual substrates with lower threading dislocation density (TDD) and thickness. This process, the thermally relaxed ultra-thin (TRUT) buffer process, consists of coherent growth of lattice-mismatched Si1.xGex layers, followed by post-growth annealing. Growth of TRUT buffers over the Si0.5Ge0.5 to Si0.3Ge0.7 alloy range with high strain levels resulted in the nucleation of surface defects which appear to limit the maximum strain rate of compositionally graded buffers. However, application of the TRUT process in the Si0.1Ge0.9 to Ge alloy range resulted in relaxed Ge virtual substrates with a 59% reduction in TDD compared to conventional processes. Lastly, growth of high-quality lattice-matched GaAsyP1.y on Si0.5Ge0.5, Si0.3Geo.7, and Si0.2Ge0.8 virtual substrates was investigated. / (cont.) Adaptation of standard GaAs on Ge processes to this heteroepitaxial system resulted in mostly non-planar growth (similar to typical GaP growth on Si) with only limited regions of planar GaAsyP1-y layers on Si0.2Ge0.8 virtual substrates. Planar growth of GaAsyP1-y on Si0.3Ge0.7 virtual substrates was enabled by minimizing the atmospheric exposure of the Si0.3Ge0.7 as it is transferred between growth reactors, establishing that the GaAsyP1-y growth process on Si1-xGex is strongly affected by atmospheric contaminants. Further minimization of air exposure, through use of Si1-xGex homoepitaxial buffers and growth of Si1-xGex and GaAsyP1-y in a single reactor, is expected to further improve epitaxial quality across the entire lattice-matched GaAsyP1-y/Si1-xGex range, including GaP on Si. / by Carl Lawrence Dohrman. / Ph.D.