UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Evaluation of entropic interaction chromatography media Coad, Bryan R. 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-ubc_2006-199589.pdf [ 12.1MB ]
Metadata
JSON: 831-1.0061138.json
JSON-LD: 831-1.0061138-ld.json
RDF/XML (Pretty): 831-1.0061138-rdf.xml
RDF/JSON: 831-1.0061138-rdf.json
Turtle: 831-1.0061138-turtle.txt
N-Triples: 831-1.0061138-rdf-ntriples.txt
Original Record: 831-1.0061138-source.json
Full Text
831-1.0061138-fulltext.txt
Citation
831-1.0061138.ris

Full Text

E V A L U A T I O N OF ENTROPIC INTERACTION C H R O M A T O G R A P H Y MEDIA by B R Y A N R . C O A D B.Sc., Simon Fraser University, 1997 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES (CHEMISTRY) THE UNIVERSITY OF BRITISH C O L U M B I A AUGUST 2006 © B R Y A N R. COAD, 2006 ABSTRACT Entropic interaction chromatography (EIC) media were synthesized by aqueous atom transfer radical polymerization (ATRP) of AfiV-dimethylacrylamide (DMA) from the surfaces of non-porous and porous matrices. This novel chromatographic medium demonstrated efficient size-based separation of protein mixtures through the entropy change associated with solute partitioning into the layer of polymer attached to the support. Reaction conditions influencing the physical properties of the grafted layer were investigated. The use of "grafting from" initiators on a hydrolyzable ester linkage enabled exhaustive characterization of the grafted layer when cleaved from the surface. Negatively charged sulfate groups on the surface of the support were required for growth of dense polymer brushes and a combination of high surface charge and initiator concentration were necessary for the growth of dense brushes. The molecular weight of the grafts could be controlled by modifying the concentration of D M A used for the polymerization. Low polydispersities for the grafted layer demonstrated that the polymerization was well controlled. Scale up permitted the targeted synthesis of sufficient material grafted with 55.7 kDa grafts of high density (0.164±0.005 chains/nm2) to be packed into a 30 x 1 cm chromatography column. Evaluation of a wide variety of columns synthesized with varying graft molecular weights or densities revealed that by exercising synthetic control over physical properties of the grafted layer, the selectivity curve could be tuned. Reducing the graft density allowed greater partitioning of high molecular weight solutes, extending the linear range of the selectivity curve. Increasing the graft molecular weight also altered selectivity, but more directly affected column capacity by increasing the volume of the grafted layer. Through the use of modeling studies, it was predicted that the shear force could act to reduce the number of conformations available to chains, increasing their rigidity without significantly altering the thickness of the grafted layer. A reduction in protein partitioning predicted by this situation was observed experimentally. The large scale EIC column has potential application towards rapid desalting applications of bimolecular feeds and the ease by which graft properties can be tuned suggests that the synthesis of custom chromatography media may be possible. in TABLE OF CONTENTS Abstract i i Table of Contents iv List of Tables viii List of Figures x List of Symbols xiv List of Abbreviations xviii Acknowledgements xxi Dedication xxiii 1. Introduction 1 1.1. Project Overview 1 1.2. Grafted Polymer Systems and their Applications 4 1.2.1. Free and Grafted Polymers and Their Physical Properties 4 1.2.2. Applications of Grafted Polymer Systems 6 1.2.2.1. Biocompatible Materials 6 1.2.2.2. Grafted Surfaces for Chromatographic Separations 7 1.3. Separating Biological Mixtures on the Basis of Size 9 1.3.1. Traditional SEC 10 1.3.1.1. Overview 10 1.3.1.2. SEC Separation Mechanism and Performance Characterization 10 1.3.2. Entropic Interaction Chromatography (EIC) 15 1.3.2.1. Grafted Surfaces for Entropic Interaction Chromatography: Theory 15 1.3.2.2. EIC in Practice 18 1.4. Synthesis of Grafted Polymer Layers 19 1.4.1. Cerium (IV) Mediated Polymerization 21 1.4.2. Atom Transfer Radical Polymerization 21 1.4.2.1. Overview 21 1.4.2.2. ATRP of Biocompatible Monomers 23 1.4.2.3. Surface Initiated ATRP of Hydrophilic polymers 24 1.4.2.4. Challenges to Tailoring Brush Properties 25 1.5. Characterization of Grafted Polymers 26 1.5.1. Analytical Methods for Probing the Surfaces of Biomaterials 26 1.5.1.1. General Techniques 26 1.5.1.2. Analytical SEC/MALLS for Polymer Characterization 28 1.6. Thesis Overview 30 1.6.1. Rationale for the Proj ect 30 1.6.2. Novelty of the Research and Objectives 31 1.6.3. Thesis Organization 31 2. Synthesis and Characterization of Grafted Latex Particles by Surface Initiated ATRP. 33 2.1. Introduction 33 2.1.1. Synthesis and Characterization of Large Latex Particles 35 2.1.1.1. Emulsion Polymerization of Styrene 35 2.1.1.2. Seeded Growth of PSL 36 2.1.2. Shell Growth Copolymerization: Synthesis of ATRP Initiator Layer 37 2.1.3. Surface Initiated Atom Transfer Radical Polymerization 39 2.2. Experimental 42 2.2.1. Materials and Methods 42 2.2.1.1. Analytical Methods 42 2.2.1.2. Styrene Distillation 43 2.2.1.3. Purification of Potassium Persulfate (KPS) 44 2.2.1.4. Emulsion Polymerization of Styrene 44 2.2.1.5. Dialysis/Cleaning Procedures 45 2.2.1.6. PSL Seeded Growth : 45 2.2.1.7. Reagent Preparation of HEA-C1 46 2.2.1.8. Shell Growth Polymerization 47 2.2.1.9. Characterization of Shell Latex 48 2.2.1.10. Distillation of D M A 49 2.2.1.11. Surface Initiated ATRP Grafting Procedure 50 2.2.1.12. Characterization of Grafted Latex 52 2.3. Results and Discussion 54 2.3.1. Seed Growth Polymerization: Synthesis of Polystyrene Latex (PSL) Particles 54 2.3.1.1. Surfactant-Free Emulsion Polymerization of Styrene 54 2.3.1.2. Characterization of S eed Latex 61 2.3.2. Synthesis and Characterization of ATRP Initiator-Functionalized Latex Particles (Shell latex) 62 2.3.3. Aqueous Surface Initiated ATRP of JV.N-dimethylacrylamide 67 2.3.3.1. Effect of Changing the Monomer Concentration 67 2.3.3.2. Effect of Changing the Cu(II) Concentration 74 2.3.3.3. Image Analysis of Grafted Latex 78 2.3.3.4. Comparisons with Published Research 80 2.3.3.5. Scale-up Reactions 80 2.4. Summary 81 3. Synthesis and Characterization of a High Throughput Entropic Interaction Chromatography Support 83 3.1. Introduction 83 3.1.1. Synthesis and Characterization 84 3.1.1.1. Choice of Matrix 84 3.1.1.2. Estimation of Matrix Surface Area 85 3.1.1.3. Functionalization of the Toyopearl Surface 85 3.2. Experimental 89 3.2.1. Materials and Methods 89 3.2.1.1. Surface Functionalization 89 3.2.1.2. Estimation of Surface Area by 1 2 5I-Labled Bovine Serum Albumin Binding Studies 92 3.2.1.3. Surface-Initiated Atom Transfer Radical Polymerization 94 3.2.1.4. Grafting Reactions with Varying Monomer Concentration and Catalyst Concentration 96 3.2.1.5. Characterization of the Modified Surface and PDMA Grafted Matrices 97 3.2.1.6. Catalyst Solubility and Binding Study 100 3.2.1.7. Kinetics of Hydrolysis 101 3.3. Results and Discussion 101 3.3.1. Estimation of the matrix surface area 102 3.3.2. Surface Modification of Toyopearl Matrix with Charged Sulfate Groups and ATRP Initiators 104 3.3.3. Aqueous ATRP Grafting of PDMA 113 3.3.4. Method Validation for Quantitative Release of Surface Grafted Polymer Chains 115 3.3.5. Effect of Changing the Monomer Concentration 117 3.3.6. Effect of Changing the Surface Charge and Initiator Concentration 127 3.3.7. Effect of Changing Copper Concentrations in the Catalyst Mixture 130 3.3.8. Effect of Cu(II) Concentration 133 3.3.9. Reaction Scale Up 136 3.4. Summary 138 4. Chromatographic Evaluation of EIC matrices 140 4.1. Introduction 140 4.2. Materials and Methods 143 4.2.1. Synthesis of EIC Stationary Phase 143 4.2.2. Choice of Running Conditions 143 4.2.3. EIC Column Packing 143 4.2.3.1. Evaluation of Protein Partitioning and Selectivity Curves 145 4.2.3.2. Determination of Mass Transfer Resistance and Parameters 145 4.3. Results and Discussion 146 4.3.1. Chromatographic Evaluation of Grafted, Nonporous Materials 146 4.3.2. Porous EIC Media 148 4.3.2.1. Protein Desalting Using High-Graft-Density EIC 148 4.3.2.2. Influence of Grafted Polymer Architecture 150 4.3.2.3. Graft Density Effects 151 4.3.2.4. Graft Molecular Weight Effects 154 4.3.2.5. Hydrodynamic Contributions to Partitioning and Band Broadening in EIC Columns 157 4.3.2.6. Moments Analysis 159 4.3.2.7. Shear-Induced Deformation of End-Grafted Polymer Chains 161 4.4. Summary 169 5. Conclusions 171 5.1. Summary and Novelty 171 5.2. Applications of the Research 172 5.3. Outlook 172 References 174 Appendix 1: Statistical Treatment of Data 184 Appendix 2: ATRP Reaction Conditions for Non-Porous Materials 186 Appendix 3: Probes Used for Size Exclusion Chromatography 187 LIST OF TABLES Table 2.1: Comparisons of predicted and experimental latex diameters from surfactant-free emulsion polymerization of styrene using potassium persulfate initiator 56 Table 2.2: Rate of decomposition of potassium persulfate (KPS) 59 Table 2.3: Predicted and experimental latex diameters from synthesis conditions based on empirical stability diagrams 60 Table 2.4: Evaluation of [initiator] and [sulfate] by conductometric titration for batches of shell latex 63 Table 2.5: Comparison of results for [HEA-C1] as determined by N M R and by titration 65 Table 2.6: Summary of initiator and charge concentrations for various shell latex described in this chapter. The graphical symbol is conserved in the figures that follow 67 Table 2.7: The weight average molecular weight, conversion, and fate of polymerized D M A for Shell M10 as a function of monomer concentration 70 Table 2.8: Physical properties of PDMA grafts cleaved from latex with varying monomer concentration for shell L3. The graft density, radius of gyration, and ratio of distance between grafts divided by radius of gyration are shown 73 Table 2.9: Initiation efficiencies for shells containing relatively low (M10 = 2.93 X 10"6 mol/g) and high (L3 = 4.45 X 10"6 mol/g) sulfate concentrations 74 Table 2.10: P D M A polydispersity as a function of catalyst deactivator (Cu(II)) proportion for both surface and solution polymer for shell L3 76 Table 2.11: Properties of grafted latex batches suitable for chromatographic studies. Evaluations of the number average molecular weight, polydispersity, graft density and DIRg are shown over increasing monomer concentration 81 Table 3.1: Summary of analytical results corresponding to incorporation of sulfate charge and ATRP initiators onto Toyopearl matrix 105 Table 3.2: ATR-FTIR peak assignments 109 Table 3.3: Effect of surface charge and initiator concentration on fate of monomer and physical properties of the grafted surface 118 Table 3.4: Effect of activating catalyst (CuCl) concentration on fate of monomer and physical properties of the grafted surface 132 Table 3.5: Effect of changing the concentration of the deactivating catalyst (copper (II)) concentration on fate of monomer and physical properties of the grafted surface 135 Table 3.6: Characterization of the polymerization for one lot of grafted matrix intended for medium scale EIC studies 137 Vlll Table 4.1: Graft and bed properties for a series of EIC columns prepared by surface-initiated ATRP 155 Table 4.2: Geometric and mass transfer parameters determined by moments analysis of elution peaks for pulse injection of K N 0 3 160 Table A1.1: Critical values of t 185 Table A1.2: Critical Q Values 185 Table A2.3: ATRP reaction conditions for shell latex in Chapter 2 186 Table A3.4: Probes used for evaluating size-exclusion properties of EIC columns 187 LIST OF FIGURES Figure 1.1: Example of a calibration curve for SEC. Solute probes elute in a volume between V, and Va inclusive 14 Figure 1.2: Idealized mechanism for ATRP in solution. The initiator reacts with a metal catalyst and generates active radicals. The halogen (X) is transferred to the oxidized-metal complex 22 Figure 2.1: Reaction scheme for the multistage growth of polystyrene latices prepared by surfactant-free emulsion polymerization using the initiator potassium persulfate (KPS) 37 Figure 2.2: Synthesis of shell latex: copolymerization of styrene and 2-(methyl 2'-chloropropionato)ethyl acrylate (HEA-C1) 38 Figure 2.3: Conductivity curve typical of saponified shell latex. End points A and B correspond to equivalence points for -SO3" and -COO" respectively 39 Figure 2.4: Reaction scheme for the synthesis of 2-(methyl 2'-chloropropionato)ethyl acrylate (HEA-C1) 46 Figure 2.5: Structural formula for A^iV-dimethylacrylamide (DMA) 50 Figure 2.6: SEM micrograph of polystyrene latex Lot G. 10 u,m scale shown 57 Figure 2.7: Frequency histogram of counts of measured latex diameter from digital image processing, n =182, mean = 1.053 u.m, standard deviation = 0.053 58 Figure 2.8: Empirical stability map for regions of stable latex produced from seeded growth of PSL. Regions outside of the enclosed areas produced either aggregated latex (for low p,) or bimodal populations (for high values of pi) 61 Figure 2.9: SEM image of 3rd stage seeded-growth latex (Batch B, large scale reaction). 20 um scale shown. 62 Figure 2.10: 'H N M R spectrum from a sample of shell latex, L8. Sample run in CDC1 3, 300 MHz ' H N M R with delay time of 1 second 64 Figure 2.11: SEM image representative of the uniformity of synthesized shell latex. 10 p.m scale shown 66 Figure 2.12: Effect of increasing D M A concentration on resulting graft molecular weight for shells M5 (+), M8 (x), M9 (•), M10 (A) , L3 (•) 68 Figure 2.13: Effect of increasing D M A concentration on resulting graft polydispersity for shells M5 (+), M8 (x), M9 (•), M10 (A) , L3 (•). Solution polydispersity is given as the empty symbol for shells M10 (A), and L3 (•) 69 Figure 2.14: Effect of increasing monomer concentration on graft density for shells M5 (+), M8 (x), M9 (•), M 1 0 ( A ) , and L3 (•) 71 Figure 2.15: The effect of changing Cu(II) proportion on the molecular weight (filled symbols) and polydispersity (empty symbols) for two systems with differing monomer concentrations; 0.16 M (•), 1.56 M ( « ) . Shell was L3 75 Figure 2.16: Effect of catalyst deactivator (Cu(II)) proportion on the graft density for 3 systems with differing monomer concentrations; 0.16 M ( A ) , 0.56 M (•), 0.83 M (•). The shell latex was L3 77 Figure 2.17: S E M image of grafted latex. There appears to be some webbing joining the latex particles. A 5 um scale is shown 79 Figure 3.1: Reactions for incorporating charge and ATRP initiators onto the surface of Toyopearl 87 Figure 3.2: Hydrolysis scheme for liberation of surface-immobilized ATRP initiators 88 Figure 3.3 A and B: Binding isotherms for 1251-Labeled BSA onto A) an amine latex of known surface area (•) and B) Toyopearl ( A ) which had a chemically similar surface. The saturation value for bound protein per gram of latex was extrapolated from equation 3.2 and calculated to be 8.7±0.4 mg/g and 230±8 mg/g respectively 103 Figure 3.4: ATR-FTIR absorbance spectra highlighting characteristic peaks for surface modifications made to Toyopearl. Spectrum of the base matrix (1): Peak 1, O-H stretch (3400 cm"1); Peak 2, C=0 (ester) stretch (1721 cm'1) 108 Figure 3.5 A and B: A) Overlay and enlargement for Peak 1 showing absorbance increase for (2) compared to the base matrix (1) and decreases for (5) and (6). B) Overlay and enlargement for Peak 2 showing absorbance increase for (6) compared to (1) 110 Figure 3.6: Light microscope images of initiator modified matrix taken at 400 X magnification. A) Toyopearl Base Matrix 66±2 u.m; B) Low [initiator] matrix 64±3 pm; C) High [initiator] matrix 67±8 \im 112 Figure 3.7: Schematic representation of surface modifications and grafting of PDMA from a macroporous chromatography matrix 114 Figure 3.8: Spectrum of PDMA grafted matrix. The spectrum contains a new peak at 1632 cm"1 from C=0 (amide) stretching vibrations from the P D M A grafts. The relative decreases in peak 1 and peak 2 are due to the much greater abundance of polymer graft and the decreased ability for ATR-FTIR to probe the surface of the material below the polymer graft 115 Figure 3.9: Amount of PDMA chains released in hydrolyzed samples from 3 hours to 8 weeks 116 Figure 3.10: The effect of monomer concentration on P D M A molecular weight and polydispersity for polymers cleaved from two different matrices. Seed A: high initiator concentration, 2.8 mmol/g ( A ) ; seed C: low initiator concentration, 0.013 mmol/g (•). Seed A was a 3.33 g suspension at 3% solids with [CuCl] = 0.13 M . Seed C was a 2 g suspension at 3% solids with [CuCl] = 0.16 M . For both reactions, [Cu(II)/Cu(I)] = 0.15; Cu(0) molar amount was 8 times Cu(I); H M T E T A molar amount was equal to the sum ofCu(I) and Cu(II) 119 Figure 3.11: The effect of monomer concentration on PDMA graft density for two different seeds. Seed A: low initiator concentration, 0.013 mmol/g (•); seed C: high initiator concentration, 2.8 mmol/g (A) . The experimental details are the same as in Figure 3.10 121 Figure 3.12: The solubility of the copper(II)/HMTETA complex in solutions of increasing D M A concentration (•). The complex consisted of a 1:1 molar mixture of CuCl 2 and H M T E T A (1.5 mmol) complexed in methanol, dried, and reconstituted in water (0.02 M). Approximately 6 mg of initiator seed was added to solutions with no D M A present. Amounts were chosen so that the solution concentration of copper complex was in a slight molar excess to the concentration of sulfate on the matrix. The resulting decrease in complex left in solution is shown for two seeds with approximately the same sulfate concentration (0.2 mmol/g): high initiator seed, 2.77 mmol/g (o); and low initiator seed, 0.133 mmol/g (0) 122 Figure 3.13: Settled bed height of PDMA grafted materials from Series C (Table 3.3). All reactions started with the same amount of initiator functionalized matrix (0.1 g) which was similar in height to tube B. Concentrations of D M A used in grafting were: B) 0.7 M ; C) 2.3 M ; D) 3.9 M ; E) 5.4 M ; F) 7.8 M 124 Figure 3.14: Light microscope images of the grafted matrices shown in Figure 3.13. Images taken at 100 X magnification. A) Initiator (ungrafted) matrix; B-F) grafts from various monomer concentrations. B) [DMA] = 0.7 M , diameter = 71±4 |im; C) [DMA] = 2.3 M , diameter = 133±21 urn; D) [DMA] = 3.9 M , diameter = 180±20 um; E) [DMA] = 5.4 M , diameter = 220±20 urn; F) [DMA] = 7.8 M , diameter = 3 0 5 ± 2 2 u m 125 Figure 3.15: Settled bed heights of medium and high density grafted matrices. E: o^ .ay, = 0.048 chain/nm2, M„ = 162 kDa; F: agraf, = 0.041 chain/nm2, M„ = 225 kDa; G: <jgraf, = 0.11 chain/nm2, M„ = 298 kDa; H: Vgrafl = 0.065, M„ = 377 kDa. E and F had [initiator] = 28 X 10"4 mol/g; G and H had [initiator] = 1.3 x 10" 4 mol/g. All material was grafted with the same ATRP reaction conditions and the same amount of starting matrix (-0.1 g). E and G used [DMA] = 5.4 M ; F and H used [DMA] = 7.8 M . From microscopy and SEC/MALLS, the particle size diameter (and M„) was determined to be E) 220±20 um (162 kDa); F) 305±22 um (225 kDa); G) 166±17 um (298 kDa); H) 330±30 um (377 kDa) 126 Figure 3.16: The effect of surface initiator concentration on measured graft density of P D M A chains cleaved from grafted, modified Toyopearl surfaces with negative surface charge (filled symbols) and without negative surface charges (empty symbols). Reaction conditions were as follows: For (•), [DMA] 1.2±0.4 M and [charge] = 0.18±0.04 mmol/g. For (•) [DMA] = 1.1±0.1 M and no detectable surface charge. For both, [CuCl] = 0.15±0.05 M , [Cu(II)/Cu(I)] = 0.15, Cu(0) molar amount 8 times that of Cu(II), and H M T E T A molar amount equal to the sum of Cu(I) and Cu(II) 128 Figure 3.17: The effect of activating catalyst (Cu(I)Cl) concentration on graft molecular weight (empty symbols) and graft density (filled symbols) for low initiator concentration, 0.0129 mmol/g ( A ) and high initiator concentration, 0.133 mmol/g (•). For reactions for (A) , 2 g of 3% seed suspension was used with [DMA] = 3.88 M . For (•), 14.1 g of 3% seed suspension was used with [DMA] = 1.56 M . For both reactions [Cu(II)/Cu(I)] = 0.15; Cu(0) molar amount was 8 times Cu(I); H M T E T A molar amount was equal to the sum of Cu(I) and Cu(II) 131 Figure 3.18: The effect of Cu(II) to Cu(I) ratio on the graft molecular weight (•) polydispersity (•). For the graft polymerization, 1.67 g of 6% seed matrix was used with [CuCl] = 0.215 M , [DMA] = 1.07 M , [Cu(0)] = 0.26 M , H M T E T A molar amount was equal to the sum of Cu(I) and Cu(II) in each reaction. Samples also included 200 mM NaCl but this had no effect on the properties of the grafted polymer.... 134 Figure 4.1: Calibration curves for nonporous columns. L2Graft (•): M„ = 184 kDa, agraf, = 1.6 X 10"3 chain/nm2; L4Graft (x): M„ = 124 kDa, <jgrafl = 21 X 10"3 chain/nm2; L5Graft (o): M„ = 203 kDa, agraf, = 0.75 X 10"3 chain/nm2; L8Graft (A): M„ = 59 kDa, agraf, = 5.2 X 10"3 chain/nm2 147 Figure 4.2: Chromatogram for removal of K N 0 3 (0.5 g L"1) from BSA (2 g L"') injected as a 100 uL pulse onto the 30 cm HGD EIC column (cr= 164 x 10"3 chains/nm2, M n = 55700 Da, column volume = 23.6 mL) at 0.5 cm min"1. The 10 mM sodium phosphate buffer (pH 7) in the mobile phase contained 300 mM NaCl. Mean elution times for each peak are shown 149 Figure 4.3: The measured selectivity curves for a HGD EIC column (164±5 X 10'3 chains/nm2, D/(Rg} =0.18, M„ = 55700±900 Da, MJM„ = 1.36±0.02) prepared by surface-initiated ATRP, and for a low graft density (LGD) EIC column {Dl(R^j ~ 2, (M„) = 48500 Da) prepared by Ce(IV) initiated grafting chemistry [14] 152 Figure 4.4: Effect of changing graft density in the brush regime for EIC columns of similar graft molecular weight: HGD EIC column (•) (cr= 164 X 10"3 chains/nm2, M n = 55700 Da, column volume = 23.6 mL); L G D EIC column (A) (cr= 1.49 X 10"3 chains/nm2, Mn = 53200 Da, column volume = 1.24 mL). Both columns were operated at u = 0.5 cm min"1 153 Figure 4.5: Effect of graft morphology on the percentage of the total column volume occupied by grafted polymer material (100 [<j>poiyme}}}- The grafts were prepared from 4 sets of EIC media varying in graft molecular weight: (•) 19.1 kDa, (o) 53.2 kDa, (A) 69.3 kDa, and (0) 56.9 kDa 154 Figure 4.6: Retention volume data for probe molecules eluting from near constant a EIC columns (1.24 mL total volume) bearing different graft molecular weights. The solid lines are best-fit sigmoidal functions. The graft density for the columns was 8.58 X 10"4 chains/nm2 and 9.56 x 10"4 chains/nm2 for (•) and (A), respectively 157 Figure 4.7: Measured partition coefficients for BSA in the HGD EIC column described in Figure 2 as a function of u. Error bars represent the standard deviation of duplicate runs 158 Figure 4.8: HETP versus flow rate data for 100 uL pulses of K N 0 3 (0.1 mg/mL) injected onto the HGD EIC column described in Figure 4.2 160 Figure 4.9: Reduced net pressure drop (AP/L) versus u across the 30 mL HGD EIC column described in Figure 4.2 162 Figure 4.10: Predicted normalized velocity profiles for an average solvent velocity (upore^ = 1.8 cm min"1 (yr of 0.039 s"1) in a 1000 A pore bearing an end-grafted PDMA layer. In the model, grafted chain length is specified by the number of freely joined Kuhn segments jV (= 6 iR2^)l 4 > w n e r e m e Kuhn length lK is twice the persistence length / P , estimated from the worm-like chain model to be 5.7 nm for P D M A [25]) and grafting density a is specified by the percentage of pore wall lattice sites bearing a grafted monomer. The Flory parameter Xbo for the polymer-solvent interaction was set at 0.4. All other interactions were assumed athermal. Simulations for solvent hydrodynamics within a pore bearing no grafted polymer are also shown 165 Figure 4.11: Dependence on solvent shear rate of the height H (given by rw an - r b r u s h , where rbrash is the lattice layer in which 4{x) drops below 0.01 #>wan)) and tilt angle 6 (° relative to surface normal) of a P D M A brush (N= 400, 0 = 9.7%, Xbo= 0.4) end-grafted inside a 1000 A cylindrical pore 166 Figure 4.12: Predicted polymer segment density $Vw an - r) profiles for a P D M A brush (N= 400, a = 9.7%, Xbo = 0.4) end-grafted inside a 1000 A cylindrical pore in the absence and presence of a solvent shear rate yr of 0.039 s"1 167 LIST OF SYMBOLS ju\ first moment of an elution peak jU2 second moment of an elution peak a size of a chain segment (Kuhn segment) A2 second virial coefficient c concentration D distance (as in spacing between groups on a surface) D/ axial dispersion coefficient dnldc refractive index increment F* viscous force h Reduced plate height H equilibrium height of the grafted layer hT reduced plate height / ionic strength I scattered@) angular dependance of intensity of scattered light k Boltzmann constant K partition coefficient kd rate of decomposition Km overall mass transfer coefficient / length of a repeat unit L column length m mass M„ number average molecular weight Mr molecular weight of a solute probe Mw weight average molecular weight MW molecular weight MJMy, molecular weight distribution or polydispersity N chain length or number of repeat units NA Avogadro's number Nmeas number of measurements nQ refractive index of the solvent Np number of plates Ns number of seed particles p probability P pressure P(9) scattering function r radius R relative abundance (R^ radius of gyration R(0) excess Rayleigh ratio Rt] Stokes radius t critical value in the Student distribution tR retention time for a solute u superficial velocity V volume Ve elution volume of a solute Vi pore volume Vm volume of solid matrix V0 void volume Vp volume of the stationary phase available for solute partitioning V, total liquid volume W peak width Wh peak width at half peak height zp distance of a particle from the surface AG0 standard Gibbs free energy change AH0 standard enthalpy change AS change in entropy AS0 standard change in entropy AS0 change in conformational entropy £ surface coverage Q number of distinguishable ways of configuring the system /3 effective porosity yr unmodified shear rate X% adsorption energy a column void fraction <f> volume fraction ?] viscosity KV hydraulic permeability coefficient X0 wavelength of incident light in a vacuum X wavelength ju chemical potential v number of degrees of freedom 9 angle pi rate of initiation o* reduced surface coverage cr surface density t, characteristic blob size LIST OF ABBREVIATIONS A F M Atomic force microscopy ATR-FTIR Attenuated total reflectance Fourier transform infrared ATRP Atom transfer radical polymerization BSA Bovine serum albumin D C M Dichloromethane D M A A^A'-dimethylacrylamide EIC Entropic interaction chromatography GPC Gel permeation chromatography HEA 2-hydroxyethylacrylate HEA-C1 2-(methyl 2'-chloropropionato)ethyl acrylate H E M A 2-hydroxyethyl methacrylate HETP Height equivalent of a theoretical plate HGD High graft density HMTETA 1,1,4,7,10,10-hexamethyltriethylenetetamine H M W High molecular weight HT High-throughput KPS Potassium persulfate LGD Low graft density L M W Low molecular weight M A L L S Multi-angle laser-light scattering M E A Methoxyethyl acrylamide M M A Methyl methacrylate NIPAM jty-isopropylacrylamide PDMA Poly(N,N-dimethylacrylamide) PEG poly(ethylene glycol) PHEMA poly(2-hydroxyethyl methacrylate) PMEA Poly(methoxyethyl acrylamide) PNIPAM polyfTV-isopropylacrylamide) PS Polystyrene PSL Polystyrene latex PVP poly(7V-vinylpyrrolidone) RAFT reversible addition-fragmentation chain transfer RI Refractive index RPLC Reverse-phase liquid chromatography SCF Self consistent field SEC Size exclusion chromatography SEM Scanning electron microscopy SSA Specific surface area TFA Trifiouroacetic acid TIR Total internal reflectance UV/Vis Ultra-violet/visible ACKNOWLEDGEMENTS I gratefully acknowledge support from my research supervisor, Donald Brooks. I feel truly privileged to have had the support and guidance offered by him and to share in his wisdom and experience. The same gratitude is extended to my co-supervisor Charles Haynes who has gone beyond the call of duty in helping me and offering support and dedication. I gratefully thank both for trusting me with a challenging project and providing all the resources necessary to complete it. I could not have completed this project without the support and friendship of Jay Kizhakkedathu. His talent, guidance, and support have shaped this work and his dedication to research has been my inspiration to achieve higher goals through my own research. I have had the privilege in participating in Canadian Blood Services' Graduate Fellowship Program and the Strategic Training Program in Transfusion Science at the Centre for Blood Research. The human and financial resources made available to me have been crucial for my success. Portions of this work were directly aided by the help of Brad Steels, Johan Janzen, Oded Kleifeld, Peter Easthope, and Zee Wang. Their contributions helped me elevate this project to one of higher quality. For my colleagues in the Brooks lab past and present: Johan, Rajesh, Raymond, Diane, Yevgeniya, Krishnan, Samuel, Srinivasa, and Irina. I thank them for helping me along the way and for being the greatest group of colleagues that I have known. The decision to partake in this endeavor was largely inspired by my work experience in industry and contacts made through job experience in Canada and Japan through the cooperative education program at Simon Fraser University. I thank all those who mentored me, particularly John Masuhara and Ed Charter at Canadian Inovatech Inc., for demonstrating the value of partnership between academics and industry. I would also like to thank my family, and my friends. Their support has carried me and allowed me to complete this project. Finally, I thank Sarah whose companionship has made the journey seem not so far. for my parents 1. INTRODUCTION 1.1. Project Overview The goal of this research is to develop a more flexible synthesis platform for producing entropic interaction chromatography (EIC) resins, which permit size-based separation of macromolecules through their interaction with an end-grafted polymer layer on the surface of a mechanically stable porous matrix, and for characterizing the mechanical, hydrodynamic, and separation performance of EIC resins as a function of the properties of the grafted polymer layer. An improved understanding of EIC could extend its range of application to meet a number of important but challenging separation problems. For example, human blood plasma is the source material for many life-saving therapeutics, including albumin, intra venous immune globulin, a-antiplasmin, ai-antitrypsin, and a number of coagulation proteins. As it is now thought to contain more than 1500 unique peptides [1;2], human plasma has the potential to yield a number of new protein-based therapeutics. The development of improved technology for resolving the blood plasma proteome without denaturation or alteration of function could provide a powerful basis for analyzing plasma samples, for identifying new drug candidates, and for establishing improved purification protocols at the manufacturing scale. Discovered by Porath and Flodin [3], size-based separations using column chromatography, typically referred to as size-exclusion chromatography (SEC), have found routine laboratory and industrial application for more than 30 years. SEC is currently used in the purification of a wide range of biologies, including plasma proteins, nucleic acids, monoclonal and polyclonal antibodies, and a range of FDA-approved therapeutic proteins produced by recombinant DNA technology [4;5]. In addition, SEC is commonly used for buffer exchange and desalting, and as a product-polishing step for removal of impurities and undesired oligomers or aggregates [6;7]. SEC typically employs a rigid porous bead or a mechanically stable porous gel as the stationary phase. Porous beads of poly(propylene oxide), glycerol (diol) derivatized silica, or crosslinked hydroxylated polymethacrylate are widely used for biological applications [8]. Modern chemistry allows careful control of the pore size distribution within the stationary phase. SEC columns are therefore available packed with resin particles of either monodisperse or polydisperse pore size. The former provide high-resolution separations over a narrow range of solute size. The latter are generally of lower efficiency, but can discriminate on a log-linear basis over as much as four molecular weight decades. The mechanism of size-based separation in SEC columns can be understood using hydraulic arguments, especially when describing separation of protein mixtures. In an ideal SEC column, interactions between the stationary phase and the macromolecular solutes within the pore liquid are steric in nature only, and the elution volume Ve of a solute is given by Ve=K+K,Vp [1.1] where Va is the interstitial (void) volume of the packed bed, Vp is the total volume of the pores within the stationary phase, and Kj is the solute partition coefficient, given by the ratio of the average solute concentration in the stationary-phase pores to that in the interstitial volume. Ki therefore varies from zero for a solute that is fully excluded from the stationary phase to unity for a point-like solute capable of fully and equally accessing all of the pore volume. Thus, all macromolecular solutes elute at a volume between V0 and V0 + Vp. Other strategies for separating complex mixtures of biologies on the basis of size have been proposed [9-12]. Of particular interest is the work of Brooks et al. [13], who pioneered EIC by demonstrating that stationary phases do not require a porous architecture to achieve size-based separations, and Pang et al. [14], who decorated a gigaporous matrix with an end-grafted poly(methoxyethyl acrylamide) (PMEA) layer to evaluate the performance of EIC in a large-bead preparative-scale format. In both cases, the end-grafted polymer layers were synthesized by growing the polymer chains from the base matrix using a cerium (IV)-based surface-initiated polymerization. This chemistry suffers from a number of limitations, including inherently poor control of graft molecular weight and polydispersity [13; 14], that restrict the variety and quality of EIC stationary phases that can be produced by it. Achieving the full potential of EIC therefore requires a new synthesis procedure that permits good control of-graft density, graft molecular weight, graft chemistry, and polydispersity. The development of that chemistry and the characterization of new generation EIC stationary phases prepared using it form the objectives of this thesis. Atom transfer radical polymerization (ATRP) is a powerful and versatile new method for generating a wide range of polymers of defined molecular weight and low polydispersity [15-19]. If the reaction is initiated at a surface, ATRP allows for the synthesis of end-grafted polymer layers of low polydispersity, high graft density and predetermined length [20-22]. ATRP may therefore provide an attractive and versatile approach to synthesize new EIC stationary phases of well-defined properties. Since it is possible to change the graft length and graft density independently using surface initiated ATRP, the technique should allow for a greater understanding of EIC and the ability to synthesize custom EIC matrices for challenging size separation problems. A key goal of this thesis is therefore to define successful strategies for using surface-initiated ATRP to synthesize end-grafted layers of poly(Ar)Ar-dimethylacrylamide) (PDMA) covering a wide range of grafting densities and polymer molecular weights. Context for the work is provided through a review of grafted polymer systems and their applications (section 1.2), size-based separation technology (section 1.3), and current strategies for synthesizing (section 1.4) and characterizing (section 1.5) grafted polymer layers. The chapter then concludes with a more comprehensive statement of the objectives of this work. 1.2. Grafted Polymer Systems and their Applications 1.2.1. Free and Grafted Polymers and Their Physical Properties The term polymer is derived from the Greek words polys, meaning many, and mews, meaning parts. The key feature that distinguishes polymers from other molecules is the repetition of many identical, similar, or complementary molecular subunits (monomers) in these chain molecules. Because polymer synthesis is governed by random assembly from the constituent monomers, polymerization reactions generally create a statistical distribution (often Gaussian) of chain lengths. If more than one monomer is employed, the ensemble of polymer chains will also vary in chemical composition. The amount of data required to characterize a polymer is therefore considerably greater than required for a typical small molecule. Common physical and theoretical parameters used to characterize a homopolymer (single monomer type) include the number of repeating units JVj, the number average Mn (= ^NjMjN, where M is the molar mass of repeating unit i) and weight average Mw i {='Y^NiM2i /iVjM,.) molecular weights, and the polydispersity (= M w / M n ) . The average distribution of segments about a polymer chain's centre of mass in solution is described by its radius of gyration (Rgy Ideally, the chain is free to adopt any conformation with equal probability and the distribution of chain segments can then be predicted using simple statistical arguments embodied in the classic random-walk model [23]. For such an ideal polymer chain, Flory derived a simple but exact expression for (R) where / is the length of each randomly ordered bond in the chain [24]. For real polymers, the finite volume of the segments further reduces the total configurational degrees of freedom of the chain away from that predicted by ideal random-walk statistics. The quality of solvent also influences the solution properties of real polymers. Such solvent effects were considered by Fleer [25] who showed an expression for (RG^ for a freely-jointed polymer chain in solution where a is a parameter that can be determined experimentally, and v takes values between 1/2 in a poor solvent and 3/5 in a good solvent. The average conformational properties of a polymer change considerably when the chain becomes end-grafted to a surface. For an isolated end-grafted polymer molecule, de Gennes postulated two limiting configurations for the chain [26]. In the "mushroom" configuration, the adsorption energy, %s, favouring segment binding to the grafting surface is small (near kT) and the chain is entropically repelled from the surface. The polymer then assumes a hemispheric shape that extends away from the surface an average distance of 2(R). If however j s is sufficiently favourable, the grafted chain will collapse onto the [1.2] [1.3] grafting surface into a flat or "pancake" configuration to maximize enthalpically favourable segment-surface contacts [25]. Other chain conformations are observed when the density of grafted chains is increased to the point where appreciable chain overlap occurs. When the distance d between grafting points is less than 2^i?g^, Alexander and de Gennes [27;28] postulated that the chains become stretched into a "brush" in order to accommodate the volume of the additional segments. Although the transition from the mushroom to the brush configuration is difficult to detect precisely, scaling thermodynamic arguments made by de Gennes [28] predict entrance into the brush regime when cr-f V - >N~6'5 d) b" [1.4] where cr is the grafting density (chains nm'2) and a is the size of the polymer repeating unit. Not surprisingly, there is considerable experimental evidence of significant changes in the hydrodynamic and energetic properties of grafted polymer layers during the transition from the mushroom to the brush regime [29;30]. 1.2.2. Applications of Grafted Polymer Systems 1.2.2.1. Biocompatible Materials Solutes that partition into a grafted polymer layer must occupy volume which is then lost to the grafted chains. As a result, previously accessible chain conformations become inaccessible, resulting in an entropy loss in the grafted layer. When coupled with thermal energy, this negative change in the conformational entropy of the grafted chains ASlrush works to reject the solute from the polymer layer. A loss in solute entropy A.S. is also observed for the partitioning reaction due to the loss of free volume associated with the presence of the grafted polymer chains. Many researchers have attempted to use this property of polymer brushes to engineer biocompatible surfaces that resist protein adsorption [31]. A few of those studies are described here. Sofia et al. suggest that a certain minimum length of graft polymer is required to prevent protein adsorption to a surface [32]. Janzen et al. [33] used (AMsopropylacrylamide) (NIPAM) polymer grafts to investigate the temperature dependent adsorption of plasma proteins onto grafted surfaces, while Yoshikawa et al. [34] used hydrophilic polymer brushes of varying graft densities to evaluate the effect of brush density on the rejection of model protein probes. Hydrophilic polymers such as poly(acrylamide), poly(ethylene glycol) (PEG), poly(vinyl alcohol), and poly(iV-vinylpyrrolidone) (PVP) have been shown to reduce platelet adsorption when compared to the hydrophobic polymers typically used in medical implant devices [35]. PEG has found particularly widespread use in surface modification to improve biocompatibility of medical implant devices. As a result, the thrombogenic properties of PEG surface coatings are well described [36;37] and a number of methods for covalently attaching PEG polymers to functionalized surfaces are available [38-40]. 1.2.2.2. Grafted Surfaces for Chromatographic Separations Polymer graft technology has been applied to media improvement in nearly every mode of chromatography [41]. The technique is particularly popular in adsorption chromatography, where polymer grafts can enhance both static and dynamic binding capacity significantly. Functionalized polymers extending away from the grafted surface have improved access to the mobile phase and to analytes of interest. Rounds et al. exploited this concept to increase the dynamic capacity of silica-based ion exchange resins [42]. Similarly, polymer brush technology has been used to enhance the performance of hydrophobic interaction chromatography, affinity chromatography, chiral chromatography, and other ion-exchange chromatography resins [43;44]. The use of grafted-polymer systems as novel stationary phases for capillary electrophoresis has also been reported [45]. Miller et al. showed that derivatized brushes grafted onto open-tubular capillaries allow for tunable separations by electrochromatography [46]. Other groups have shown how brushes can be used to modify electroosmotic flow and thereby alter elution profiles in capillary zone electrophoresis [47;48]. Application to Size Exclusion Chromatography Matrices Silica and polystyrene based matrices are popular choices for SEC owing to their ease of preparation, including the ability to generate pores of a desired size, and their good mechanical strength. However, application of these SEC matrices to macromolecular solutes, especially proteins, can lead to unwanted solute adsorption to the matrix. Grafting of dextran to porous silica SEC resins has been shown to reduce unwanted protein adsorption [49] by shielding the negatively charged silanol groups of the base silica [50]. Likewise PVP was shown to reduce solute adsorption to a polystyrene-based SEC matrix [51]. Grafted polymer layers have also been used to modify the pore dimensions and therefore change the elution properties of solutes [52]. Based on a set of preliminary studies, Cohen et al. were among the first to suggest that grafted SEC matrices could be used to create a tunable size-based separation material for chromatographic applications by varying the grafted polymer molecular weight and graft density [53]. High Throughput Matrices Recent advances in polymer technology have made available highly porous base matrices that hold up to high flow rates while maintaining structural integrity. Such high-throughput (HT) matrices are valuable for industrial processing as well as for high throughput chromatography arrays. HT matrices (also known as fast-flow matrices) grafted with polymer brushes take advantage of this open rigid support to project a polymer-rich layer into the mobile phase. For example, graft polymerization from the surface of a monolithic column was found to be a good way to improve the ion exchange capacity of soft polymeric monolithic media formed as cryogels [54]. The use of grafted polymer layers has also allowed for HT size-based separations at preparative scales [55]. Fractogel BioSEC media, a product of E. Merck Biosciences, provides good mechanical stability and excellent resolving power in high-throughput separations of proteins and peptides from 5 to 1000 kDa [56]. The new chemistry and methodologies developed in this thesis apply specifically to improving this technology. 1.3. Separating Biological Mixtures on the Basis of Size Differences in size are routinely used to separate complex mixtures containing biological solutes. A variety of effective separation strategies are available when the sizes of the molecules to be separated differs greatly. These include dialysis, the many modes of micro- and ultra-filtration [57-59], as well as various forms of column chromatography. In industry, common size-based separations include cell harvesting by micro filtration [60], product concentration and partial purification by ultrafiltration [61-63], buffer exchange and product desalting by chromatography or ultrafiltration [64], and final product purification or polishing by SEC, where baseline separation can often be achieved for any two solutes differing by more than a factor of 1.8 to 2.0 in molecular weight [6;7]. SEC is therefore widely used by industry for synthetic and protein-based therapeutics, including fractionation and purification of plasma proteins [65] and the separation of recombinant monoclonal antibody (MAb) products from any unwanted oligomeric states of the MAb created during manufacturing [4;5]. 1.3.1. Traditional SEC 1.3.1.1. Overview The first gel filtration materials were dense networks of dextran or agarose fibers that permitted size-based separations but suffered from compressibility issues that limited system throughput. The unoptimized geometry of the gel network meant that long columns had to be used to achieve adequate resolution. As a result, early gel filtration separations took place over the course of several days. Crosslinking the media helped add mechanical stability and structure to the porous network and thereby improve flow characteristics and resolution. Modern crosslinked polysaccharide-based resins (e.g., Sephadex) are based on this technology and remain popular for certain size-based separation applications [66]. Modern SEC media for aqueous applications generally utilize hydrophilic polymer composites. Some of the more popular base matrices are porous beads composed of polyacrylamide, polymethacrylates, or polyvinyl alcohol. The advantage of these materials is that they have a relatively high mechanical strength and a connected pore structure that allows for reasonably high-throughput separations with minimal pressure drop. However, band broadening due to mass transfer limitations resulting from slow rates of solute diffusion in the stationary phase remain a problem, leading to a tradeoff between speed and quality of the separation [67;68]. 1.3.1.2. SEC Separation Mechanism and Performance Characterization Size-based separation by SEC is achieved by the geometry dependent partitioning of macromolecules between a continuous phase and the porous interior of a gel or cross-linked bead. The partitioning process is generally diffusive in nature, and distances of intraparticle diffusion in traditional SEC resins are on the 0.1-micron scale, resulting in fairly long elution times and broad peaks compared with other modes of chromatography. The size of the solute relative to the pores dictates whether, and to what extent, partitioning into the stationary phase will occur, thereby removing the solute temporarily from the mobile-phase flow within the interstitial volume. Large solutes are therefore preferentially rejected from the stationary phase and elute with the column void volume V0. In contrast, the smallest molecule in the mixture, usually the solvent, can sample the entire pore volume of the stationary phase. The average residence volume Vt for a solvent molecule is therefore much larger than V0. A l l solutes are then expected to elute in a volume Ve lying somewhere between V0 and Vt. These concepts may be used to define a working definition for the partition coefficient K{ of solute / where Ve\ is the elution volume of solute /. An experimentally accessible quantity, K\ takes values approaching 0 for solutes that cannot partition into the stationary phase and values approaching 1 for solutes that sample the entire pore volume. Using simple geometric arguments, Ogston [69] predicted for a uniform pore-size SEC resin that K\ (and therefore Ve) is a function of rlR, the ratio of the hydraulic radius of the solute to that of the stationary phase pores. This concept was then refined by Squire [70] to include variations in solute shape and bead pore size and shape, leading to the result o [1.5] [1.6] where a is a constant characterizing the distribution of volume within the column. Size exclusion chromatography can be further explained using thermodynamic arguments [71-73]. The Gibbs free energy change (AG0) for the solute partition process is given by AG0 = AH° -TAS° =-RT\nK [1.7] where K is the partition coefficient, AH0 is the standard enthalpy change, and AS° is the standard entropy change. Properly designed SEC resins interact athermally with partitioning solutes. Solute partitioning is then a function of entropy only K = Qxp(AS°/R) [1.8] where the standard entropy change AS° for the partitioning solute molecule is negative since K < 1. The entropy change AS" for the partitioning of solute / is given by A S ° = i f c l n — [ 1 . 9 ] where k is the Boltzmann constant, Qf o r e is the number of accessible conformations of the solute in the pore volume and, similarly, Q° is the number of conformations available to the solute in the interstitial volume. The geometrically tight confines of pore volume limit the number of conformations available to the solute, so that Qf o r e /Q° is always less than 1, resulting in an entropy change that becomes more unfavourable as the size of the solute molecule increases. Given its thermodynamic basis of separation, the performance of an SEC column is often characterized using the plate model of chromatography. This simple but effective theory treats the chromatographic separation as a series of equilibrium stages or plates. The required number of equilibrium plates Np is a direct measure of column efficiency, as peak capacity increases with Np. For a column of length L, the height equivalent of a theoretical plate (HETP) is given by HETP = — [1.10] NP Any solute eluting from an SEC column as a Gaussian peak can be used to estimate HETP for that column through the central result of plate theory [74;75] HETP = — = r [1.11] N„ ft V " 5.54 where fa is the retention time, given by the first central moment of the peak, and Wh is the peak width at half height and is therefore related to the peak variance. A decrease in HETP reflects a more efficient column. As a lower limit, HETP can be compared to the diameter of a single bead of stationary phase packing (dp). This quantity, called the reduced plate height h, is defined as h = ^ - [1.12] dP A high-performing well-packed column is usually characterized by an h near or less than 2 [76]. Experimentally, the performance of an SEC column is often characterized through measurement of a calibration curve, which plots Ve for a wide range of solutes as a function of an appropriate or convenient measure of solute size, usually the molecular weight (MW) of the solute. As a result, the calibrated column can be used to determine the molecular weight of an unknown solute. When the abscissa is plotted as log(MW), calibration curves are usually sigmoidal (Figure 1.1) with horizontal asymptotes defining solutes that either fully partition or are totally excluded from the stationary phase pore volume. Obviously, molecular weight is not a direct measure of molecular volume. Significant attention has therefore been given to establishing a quantitative correlation between M W and molecular size, most often to the Stokes (hydrodynamic) radius of the solute [77]. One of the best empirical relations was reported by Hagel, who found that Rq = 0.27IMW 0 4 9 8 where Rn is the Stokes radius (in A) [67]. An alternative and equally effective approach is to use multi-angle laser light scattering (MALLS) to measure the radius of gyration (R) of each purified protein directly in the eluent stream and thereby establish a direct correlation between Ve and solute size [78;79]. Probe Molecular Weight Figure 1.1: Example of a calibration curve for SEC. Solute probes elute in a volume between V, and V0 inclusive Probes eluting in Vt (Permeation Limit) Probes eluting in Vo (Exclusion Limit) Characterizing SEC column performance is also achieved through measurement of the selectivity curve, which plots the partition coefficient K\ of a solute against a measure of its size. In practice, the logarithm of a solute's molecular weight is used as the indication of size. As a dimensionless unit, the partition coefficient has the advantage that it allows for comparison of partitioning data between different columns. When evaluating the separation power of a column, it is then sometimes useful to regress the linear region of the selectivity curve to K *a-blogMW [1.13] where a is the intercept, and b is the slope. Columns with large b values permit a greater separation of solutes, while the ratio alb is related to the mean pore size [68]. 1.3.2. Entropic Interaction Chromatography (EIC) 1.3.2.1. Grafted Surfaces for Entropic Interaction Chromatography: Theory Brooks et al. were the first to demonstrate that pores are not required for SEC [13]. That experimental study, which helped pioneer the field of EIC, was driven by a previous theoretical study which used the Flory-Huggins lattice model of polymer-solution thermodynamics to predict a size-dependent exclusion effect for solutes partitioning between a polymer-rich phase and a pure aqueous phase [73]. Although too idealized to be of quantitative value, the model qualitatively captures much of the basic physics of EIC by simply assuming that the grafted polymer layer could be approximately described as a homogeneous (ungrafted) polymer phase. Under the limit of an infinitely-dilute solute, where the presence of the solute does not perturb phase compositions, an expression for the solute partition coefficient K can then be derived from the free energy of mixing the solute with polymer in a common solvent K = ^  = exp[-P 3$(l - l/P2 - Z u + %n)] [1.14] In equation 1.14, <JK is the volume fraction of component /, subscripts 1, 2, and 3 describe the solvent, polymer, and solute components, respectively, and superscipts m and 5 indicate the mobile (pure solvent) and surface (polymer-rich) phases, respectively. Enthalpic contributions to partitioning are captured through the Flory interaction parameter, Xij, which relate the enthalpic cost of exchanging any two components, / and j. In the Flory-Huggins lattice model, the volume fraction for any component / is given by 4 = ^ 7 [ L 1 5 ] where n, is the number of molecules of the ith component and Pt is the ratio of the volume of a molecule of polymer i divided by the volume of a solvent molecule. As was demonstrated experimentally by Brooks et al. [13] using PMEA grafts on polystyrene latex microparticles, equation 1.14 predicts that solute partitioning into the polymer-rich phase decreases with solute size (P3). Moreover, exclusion from the polymer-rich phase is predicted to increase with increasing polymer concentration (i.e., graft density) and molecular weight. Finally, and most importantly, these size-dependent partitioning effects are predicted under conditions where xn = Xv> a n d equation 1.14 thus becomes a purely entropic model. The fundamental basis of EIC is therefore the total entropy change AS*0 associated with solute partitioning from the mobile phase into the grafted polymer layer at the surface of the stationary phase. Two distinct processes contribute to this entropy change. By partitioning into the grafted polymer layer, the solute occupies volume and thereby reduces the total conformational degrees of freedom available to the grafted chains. As a result, a decrease in the conformational entropy of the grafted polymer AS^ is observed. Likewise, the presence of the connected polymer chains reduces the degrees of freedom available to the partitioned solute i, thereby reducing its entropy by an amount AS-. AS°=AScgrafl+AS< [1.16] where AScgraft is the change in conformational entropy of the grafted layer. Quantitative Modeling of EIC Using Self-Consistent Field Theory Quantitative determination of AS" requires a model that both accurately computes the density distribution of end-grated polymer chains away from the grafting surface, and the energy change associated with a particle partitioning into the volume occupied by those chains. Steels et al. [80] and a follow-up study by Pang et al. [14] showed that such a model could be formulated using the self-consistent mean-field (SCF) theory of Scheutjens and Fleer [81;82] which divides a large lattice defining a system into a set of small regions in which properties may be averaged. Their model builds on earlier efforts of Cosgrove et al. [83] and Leermakers et al. [84] who modified the original SCF theory to permit computation of segment density profiles for polymer chains end-grafted to a planar surface which are in excellent agreement with experimentally measured density profiles for grafted polymer layers in aqueous solvents [85]. The numerical SCF model by Steels et al. [80] calculates the energetics of partitioning a particle into a grafted polymer layer. As it deals with an individual particle, it is strictly valid only at dilute solute conditions. However, at equilibrium Kp is independent of solute concentration, making model estimates valid at normal column operating conditions. Pang et al. [14] coupled the self-consistent field model of Steels et al. [80] with a classic equilibrium-dispersion type chromatography model to accurately predict protein elution profiles on wide-pore EIC columns, as well as to analyze the effects of resin properties and operating conditions on column separation efficiency. The resulting SCF-EIC model accurately predicts AS" and its strong dependence on the hydrodynamic volume of the partitioning solute. The dependence of EIC performance on grafting density and chain molecular weight is also captured, with significant changes in column performance predicted for small perturbations in graft properties near the mushroom/brush transition. Columns of grafting density below or near this transition region are predicted to provide efficient separation of protein mixtures over three decades of protein molecular weight. In contrast, high-density wide-pore EIC columns are predicted to offer a potentially attractive means of removing low molecular weight analytes and may therefore find value in desalting applications. 1.3.2.2. EIC in Practice The utility of EIC as a size-based separation method for complex mixtures of macromolecules has been demonstrated in several studies and through the commercial form of this concept: the Fractogel BioSEC column of E. Merck KgaA [86]. Brooks et al. demonstrated that EIC could be performed using columns. packed with non-porous polystyrene shell-modified beads grafted with terminally attached polymers [13]. Grafted PMEA layers were grown from the surface of the beads using a cerium (IV) catalyzed polymerization reaction under conditions similar to those used in the Fractogel BioSEC column. The effect of changing the graft density and molecular weight of PMEA on the separation of protein standards was compared with SCF-EIC model predictions. Results showed that the partition coefficient decreased when the density of the grafted chains on the surface increased, but the relatively poor separation efficiency of this nonporous resin precluded more exhaustive studies. To decrease HETP to values similar to the commercial column, Pang et al. grafted PMEA polymer layers on a mechanically stable highly porous stationary phase, Toyopearl AF-650M, using Ce(IV) surface-initiated polymerization [14]. Two columns, differing primarily in the grafting density of the PMEA layer, were prepared and characterized. The graft densities on both were thought to be at or below the brush transition based on overall mass balance calculations. Selectivity curves for both columns were linear over at least three orders of magnitude of solute molecular weight. As predicted from the EIC-SCF theory for the interaction of protein-like particles with end-grafted polymer chains in a good solvent [14;87], the column displaying the higher chain density excluded proteins more strongly, resulting in lower partition coefficients, faster elution times, and a steeper selectivity curve for proteins with molecular weight less than approximately 100 kDa. The greater throughput of this higher density resin is therefore balanced by its significantly lower resolving power for high molecular weight proteins. 1.4. Synthesis of Grafted Polymer Layers Overview Terminally attached polymers can be attached to surfaces by a number of means usually classified as "grafting to" the surface or "grafting from" the surface. Grafting to the surface can be performed by physiadsorption of a polymer chain. Covalent attachment is also possible by having the appropriate chemical reagents present on the surface and the polymer terminus and reacting them. This method has several advantages. Solutions can be prepared containing very homogenous polymers with regards to their molecular weight and polydispersity, such that when they are grafted to the surface, layers of uniform composition are created. Free polymer prepared in solution can be synthesized by any conventional polymerization technique allowing a wide variety of monomers to be used. The disadvantage of the graft-to method is that by its very self exclusive nature, the first polymer chain grafted will resist the grafting of the second chain at a distance less than twice its radius of gyration. It is therefore very difficult to achieve densely grafted polymer layers by this method. Grafting from the surface requires an activated surface from which chains can be initiated. Use of plasma or glow discharge is an effective means for producing a surface with a dense arrangement of initiators. Another strategy is to incorporate initiators directly onto the surface through chemical reaction. A variety of polymerization methods exist for growing polymer chains from the surface as reviewed by Zhao and Brittain [88]. Many of these methods show great promise for the construction of dense and elaborate polymer layers including brushes. One such method is radical polymerization with reversible addition-fragmentation chain transfer (RAFT). RAFT allows a greater variety of monomers and solvents to be used compared to other types of controlled/living polymerization techniques. Several groups have used RAFT to successfully graft polymer layers from the surfaces of substrates. Yoshikawa et al. used RAFT for the polymerization of poly(2-hydroxyethyl methacrylate) (PHEMA) from a polymer substrate [89]. Plasma discharge was used to create initiators on the surface and the polymerization produced polymer layers in reaction conditions at ambient temperature. Boyes et al. used RAFT and ATRP to synthesize block copolymers estimated to be brushes that conformationally rearranged in response to different solvents [90]. The utility of ATRP for synthesizing uniform polymers by controlled polymerization is further discussed below. 1.4.1. Cerium (IV) Mediated Polymerization Ce(IV) mediated surface initiated polymerization has been the primary method of choice for synthesis of EIC media used in research and commercially. In the reaction, a weak reducing group on the surface of the matrix such as an aldehyde is able to facilitate reduction of Ce(IV) to Ce(III). The radical generated on the initiating surface group will then add to vinylic monomers in solution, propagating the polymerization reaction. At any given moment in the propagation stages, the concentration of radicals remains high. Therefore, the possibility of radical combination by monomolecular or bimolecular means can be a common occurrence the result of which is poor control over the length and distribution of the polymer chains obtained. A serious limitation to growing dense polymer brushes from the surface results from the fact that these radical-radical interactions are more likely to occur at high surface initiator concentration. 1.4.2. Atom Transfer Radical Polymerization 1.4.2.1. Overview Modern atom transfer radical polymerization stems primarily from investigations in 1995 by Wang and Matyjaszewski [15;91]. Since then Matyjaszewski has led the development of ATRP into one of the most important areas of polymer science. ATRP offers many advantages over conventional polymerization techniques [92-94]. As a controlled/living polymerization technique, ATRP is characterized by a linear increase in polymer molecular weight with conversion. A high degree of control allows for very narrow molecular weight distributions. Polymerizations are also considered to be living in that polymerization can be continued in the presence of catalyst when more monomer is added. ATRP is characterized by a chemical equilibrium in which the generation of free radical species (R*) is heavily suppressed. Figure 1.2: Idealized mechanism for ATRP in solution. The initiator reacts with a metal catalyst and generates active radicals. The halogen (X) is transferred to the oxidized-metal complex. X is defined as a halogen or pseudo halogen attached to an ATRP initiator. M" is a transition metal associated with a stabilizing ligand. The forward reaction is characterized by kact, a relatively slow rate of activation whereas the reverse reaction is governed by kdeach a much faster rate of deactivation. Activation results in homolytic cleavage of the R-X bond resulting in a radical production and transfer of the halogen to the metal, which is oxidized to its +1 oxidation state. Once activated, the monomer is converted into polymer by conventional free radical polymerization. Because the rate of deactivation is so much greater than activation, the amount of active radical species present at any given time is very low. This situation prevents side reactions from occurring and limits the amount of radical loss by termination. Both of these factors result in polymer which has a narrow molecular weight distribution (low polydispersity). To function ideally in this capacity, the rate of deactivation must not only be fast but also reversible. ATRP is a very versatile reaction mechanism and amenable to a wide variety of reaction conditions [19]. However, some limitations in reaction conditions do exist. The + X — M t + 1 / Ligand deact R—(Monomer)—X halogen must be a good leaving group and should favour homolytic cleavage of the R-X bond over heterolytic cleavage. Reactions must be carried out in a non-oxidizing atmosphere to prevent oxidation of the transition metal. The ligand must stabilize the metal by facilitating an expanded conformation when the metal is oxidized and must also force the rapid reduction back to its lower oxidation state. Propagation of polymerization occurs as radical addition across the double bond of a vinyl monomer. The specific choice of vinyl monomer is limited in that, as a radical intermediate, the structure must suitably stabilize the radical and also favour the deactivation mechanism over the propagation mechanism. Acrylates, acrylamides, and styrene derivatives are good choices. 1.4.2.2. ATRP of Biocompatible Monomers As discussed above, acrylates and acrylamides are favoured choices for biocompatible materials. ATRP has been very successful for controlled polymerization of methyl acrylate and methacrylates [17;20;95-97]. Of particular note is the preparation of acrylates in aqueous solvents [16;98]. Others have shown successful ATRP of 2-hydroxyethyl acrylate [99] and 2-hydroxyethyl methacrylate [34;46;100]. Initially poor success was reported for ATRP of acrylamides. Accounts in the literature showed that ATRP of A^-dimethylacrylamide (DMA) in solution was not a controlled process by the ATRP mechanism [101]. One report suggested that ATRP of (meth)acrylamides was not possible due to inadequacies of the ligands studied [102]. Huang et al. reported the first example of a surface initiated, metal mediated, living polymerization of acrylamide in 1998 [45]; however, the polymerization mechanism was not investigated. A follow-up report in 1999 investigating the kinetics of this reaction showed that polyacrylamide could be grown via the ATRP mechanism from the surface of porous silica [103]. Recent reports describe continuing work with solution ATRP of acrylamides [104] including the preparation of thermosensitive block copolymers [105]. 1.4.2.3. Surface Initiated ATRP of Hydrophilic polymers ATRP for use of grafting polymer from immobilized initiators is a developing field. Zhao and Brittain have written a review of the synthesis of polymer brushes focusing on linear polymer grafts [88]. A review limited to grafting produced by ATRP was presented by Pyun et al. with emphasis on characterization by atomic force microscopy (AFM) [106]. ATRP has been used to grow grafted polymer layers from a variety of substrates. For example, poly(methylmethacrylate) (PMMA) was grown from the surface of a porous glass filter by ATRP [107]. Silica particles and wafers are common substrates for surface initiated ATRP due to their relative ease of preparation and initiator incorporation. Husseman et al. show ATRP of methylmethacrylate (MMA) from the surface of silicon wafers [108]. This work and the former show good methodology, in that polymers were cleaved from the surfaces for analysis. Matyjaszewski et al. investigated styrene and acrylate polymerization from silicon wafers [20]. Perruchot used mild (20 °C) aqueous conditions for ATRP of methacrylates from silica particles [109]. The kinetics of surface initiated ATRP was investigated separately by Kim et al. with methyl methacrylate and methyl acrylate for ATRP from self-assembled monolayers on gold substrates [110], and also by Pyun et al. for methacrylate and styrene polymerization from colloidal silica surfaces [111]. Miller et al. used ATRP of H E M A inside capillaries to synthesize surfaces for electrochromatography. Recently, the ATRP of M M A and H E M A from silicon gel was reported and showed good results in water at ambient temperatures [100]. Surface initiated ATRP of D M A and related polymers from the surfaces of polystyrene microspheres have been investigated in the Brooks lab. Initiator attachment chemistries using a cleavable linker have allowed for exhaustive characterization of grafted layers showing the densest grafting of PDMA reported [21]. A model of the surface showed that the negatively charged environment on the surface of polystyrene likely contributed to very successful polymerizations of D M A , the characteristics of which were not observed in solution [22]. Further work extended this methodology to work with other acrylamides, namely block copolymer brushes of D M A with methoxyethylacrylamide and N-isopropylacrylamide [112], and other investigations of poly(N-isopropylacrylamide) homopolymer brushes for the purposes of studying protein adsorption [113]. 1.4.2.4. Challenges to Tailoring Brush Properties When designing systems for investigating EIC, experimental considerations for producing large variations in AScbrush should be considered. Furthermore, it would be desirable to minimize the contribution of AS- so that the variation in AScbnish (equation 1.16) can be investigated independently. The relative contribution of AS- to A S 0 can be minimized through the use of a nonporous stationary phase displaying a grafted polymer layer on the particle surface. While good size-based separation can be achieved using this geometry [13], small resin particles are required since the number of theoretical plates scales with the inverse of particle diameter in this case. An alternate strategy is to increase stationary phase surface area by presenting the grafted polymer layer on a giga-porous particle permitting convection through the connected pore architecture. The 1000 A diameter pores of these resins do not provide any sieving effect on their own, but provide a very large specific surface area (m2 g"1) for presenting the grafted polymer layer to enable the size-dependent partitioning effect [14]. Providing variations in AScbrush by design should involve the synthesis of a variety of grafted polymer layers whose properties allow for many different brush conformations. In this context, it would be beneficial to be able to independently vary the density of the brush or the graft molecular weight. Surface initiated ATRP allows excellent control over synthesis of a desired graft molecular weight. If consistent initiation efficiencies can be obtained, it should be possible to measure the effect of changing the graft molecular weight at constant graft density and the effect on the resulting EIC separation. Manipulating the graft density at constant molecular weight is a formidable challenge. The best approach to solving this problem is to grow polymer brushes from a surface attached to a hydrolyzable linkage. The use of limited hydrolysis, providing that chains are cleaved randomly and without preference for slight variation in local environment or graft molecular weight, would be an excellent methodology for addressing this need. The cleavage methodology is also greatly beneficial for characterizing the grafted layer. Strategies for exploiting this function are illustrated below. 1.5. Characterization of Grafted Polymers 1.5.1. Analytical Methods for Probing the Surfaces of Biomaterials 1.5.1.1. General Techniques In recent years, there has been much innovation in the use of advanced or novel technologies to probe the surface of biomaterials. The journal Biomaterials devoted an issue to covering the subject [114]. This included reviews of total internal reflectance (TIR) fluorescence microscopy, attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy, electron spectroscopy for chemical analysis (ESCA), ellipsometry, atomic force microscopy (AFM), and Auger electron spectroscopy. For dynamically probing the surface of a grafted layer in fluid, A F M is a very promising technique. Used for imaging as well as providing quantitative data, A F M is able to probe the surface of the material to a resolution of individual polymer chains. Goodman et al. developed an A F M technique for probing polymer brushes attached to colloidal particles [115]. Upon retraction of the A F M tip from polymer brushes, force curves were used to compute data for the polymer's molecular weight and polydispersity. The morphology of the chains was also seen to have an effect on the measurements. A significant advantage of A F M is the ability to probe surface changes in different solvents. Goodman's A F M technique was applied to temperature sensitive block copolymers and was used to show predicted changes in brush thickness with temperature [112]. Probing techniques that sample only a portion of the surface, however, are not likely to capture chemical modifications made to highly porous materials. One of the best ways to analyze the molecular weight, polydispersity, and radius of gyration of the grafted polymer layer from surface initiated polymerizations of porous materials is to use cleaving procedures for removing the polymer graft from the surface. Published methods have shown that for polymer-grafted silica or porous glass substrates, the surface matrix can be dissolved using hydrofluoric acid [ 103; 107; 111; 116-118]. Choosing initiator attachment chemistries that can be hydrolyzed allows for the polymer graft to be liberated by using sodium hydroxide [21; 119; 120]. Cleaved polymer solutions are then separated using analytical SEC using on-line light scattering coupled to mass detection. This technique is briefly summarized below. 1.5.1.2. Analytical SEC/MALLS for Polymer Characterization Multi-angle laser light scattering (MALLS) is a static light scattering technique used for correlating the size of solute to the intensity of the light scattered at different angles. The principal is based on Rayleigh light scattering in which the incident radiation has a wavelength much larger than the diameter of the solute of interest. Laser light incident on a macromolecule will set up an oscillation of the electric dipole and cause coherent scattering of the light in the plane perpendicular to the plane of the polarized light. The dipole, being a properly of the solute, is a measure of its refractive index. The intensity of scattered light is a function of both the size of the particle and the concentration. The intensity of light scattered at any angle in the plane perpendicular to the incident beam, varies for particles greater than about 10 nm. By examining the scattering intensity at various angles, size information of the solute can be obtained. Individual mass units within a polymer chain cause slight phase shifts in the scattered light and therefore give a relation between these mass units and the centre of mass. Mathematically, the sum of these mass contributions at each distance from the centre of mass is related to the radius of gyration of the macromolecule. Combining theses postulates, it is possible to express the angular dependence of the intensity of scattered light Iscattered(9) as hcaUereAO) * R{0) = K*MrcP{9)[\ - 2A2McP(9)] [1.17] where R(9) is the excess Rayleigh ratio, c is the concentration of the solute, P(9) is the scattering "form" function (relates the angular variation in scattering intensity to \Rg)), and A2 is the second virial coefficient (solute / solvent interaction) [121]. The factor K* is a constant for the solute defined as An n 2„2 / , \2 dn [1.18] where na is the refractive index of the solvent, NA is Avogadro's number, X0 is the wavelength of the incident light in a vacuum, and dnldc is the refractive index increment of the solute. In practice, it is convenient to prefractionate the polymer sample prior to detection by laser light scattering. Because it is possible to measure solute light scattering in an eluent stream, the outlet from a series of analytical SEC columns can be passed through a M A L L S detector. By adding a refractive index (RI) detector in series, simultaneous determination of the concentration can be made as well. In practice, the operator must input constants for the laser wavelength, solvent refractive index, the refractive index increment of all solutes for analysis. From the RI detector, the operator selects the peak which determines the solute concentration. When the operator selects the corresponding light scattering peak, the software computes a linear fit of equation 1.17 resulting in a Debye plot (K*c/R(0) versus sin2(#/2)) and yields values for Mr and (Rg)- In the case of light scattering, the molecular weight determined is a measure of the weight average molecular weight of the polymer (Mw). When the distribution of the light scattering peak is compared to the mass peak from the RI detector, the molecular weight distribution or polydispersity (MJM„) can be calculated. Interpretation of the data allows for a very generous picture of the grafted layer on the surface. For a sampled surface of known area, the mass of the polymer as determined by the RI detector can be related to the density of the grafted chains. 1.6. Thesis Overview 1.6.1. Rationale for the Project Information presented here reveals several deficiencies in the state of the art concerning the synthesis and application of grafted polymer layers applied towards new materials for studying EIC. Ce(IV) mediated, surface initiated polymerization has been used to synthesize EIC matrices from non-porous particles. Preliminary work has proved useful in creating grafted media, albeit of relatively low density, for chromatographic evaluation of EIC columns [122]. The non uniformity of the grafted layer, as seen by large polydispersities and multi-modal distributions, complicates the process of evaluating grafted layers with uniform properties. Recently, Ce(IV) polymerization has been applied to highly porous matrices [14]. A consideration for high throughput separations has been an improvement in the technology. However, the synthesis of the grafted layer involved a non-hydrolyzable covalent graft attachment. As a result, the grafted layers were relatively poorly characterized. Values for molecular weight and density were assumed from mass balance and the polydispersity was unknown. Although brush morphologies have been claimed, there has been no explicit experimental determination of graft properties. Although the variation of the elution parameters was investigated by varying the graft molecular weight at relatively constant graft density, there was no investigation of the variation of the graft density as this parameter could not be measured directly. 1.6.2. Novelty of the Research and Objectives The work described in this thesis seeks to further understand EIC through the application of new polymerization methods and design methodologies. In recent years it has become possible to obtain excellent control over the grafted layer. Using surface initiated ATRP, it is possible to grow chains at extremely high densities with very good control. ATRP has not yet been applied to the grafting of polymer brushes from a highly porous chromatographic matrix. Surface initiated ATRP of D M A from non-porous media is a new field. The work in this thesis demonstrates for the first time that established ATRP reaction conditions for D M A grafting from non-porous media can be modified to apply to porous media and that grafted layers can be successfully synthesized. It will be shown that EIC media of unprecedented density can be synthesized with polymer brushes far into the brush regime. Furthermore, the synthesis methodology allows for cleavage of the grafted layer enabling exhaustive characterization of the grafted polymer layer. The use of graft cleavage also allows for investigation of the mechanism by which the graft density affects EIC separation at constant molecular weight. Furthermore, excellent control over graft molecular weight permits investigations of how molecular weight affects EIC separation at constant graft density. These controls allow for an opportunity to significantly alter AScbrush in a way not before possible. 1.6.3. Thesis Organization Chapter 2 outlines efforts to adapt published methods for D M A grafting by ATRP from 300 to 500 nm diameter polystyrene particles to newly synthesized particles of 2.5 um diameter, suitable for chromatographic evaluation of the EIC properties. Chapter 3 summarizes efforts for functionalizing a highly porous, commercially available chromatography matrix to enable ATRP grafting of D M A from the surface. Chapter 4 shows the chromatographic evaluation of EIC matrices. Both non-porous and porous materials are summarized. Chapter 5 presents the conclusions brought forth from this work and suggests avenues for future study. 2. SYNTHESIS AND CHARACTERIZATION OF GRAFTED L A T E X PARTICLES B Y SURFACE INITIATED ATRP 2.1. Introduction Research conducted previously in the Brooks lab involved the synthesis of polymer chains grown from polystyrene latex (PSL) by cerium (IV) mediated polymerization [122] and led to the conclusion that size exclusion chromatography does not require pores [13]. The primary disadvantage of using this polymerization methodology to grow polymer grafts from the surface is that inhomogeneous graft layers are produced which have a high polydispersity. This lack of control is also evident in the relatively low graft density and relatively high estimated values of DI(R^J (between 1 and 2) indicating loose brush morphologies which may have some mushroom character [119]. Recently, ATRP from surfaces bearing immobilized initiators has matured into a good experimental technique for growing well controlled dense polymer brushes. For example, reports have shown very dense (~ 10"7 mol/m2), well controlled (MJMN < 1.4), polymer grafts of PDMA from PSL far into the brush regime (Dl{R^ = 0.06) [21]. These procedures enabled a high level of control over both grafting density and the ability to select for a desired molecular weight. These ATRP techniques, applied to chromatography media could provide a more sophisticated method by which to study the principals of EIC. While the synthesis of polymer brushes for EIC has been documented for Ce(IV) polymerization techniques on a non-porous matrix [13; 122], there have been no reports of attempts to use ATRP to grow well-controlled, high density polymer brushes from non-porous surfaces for size-exclusion chromatographic applications. Admittedly, EIC columns produced from non-porous materials would be limited to analytical applications due to their low column throughput. For example, polystyrene-based matrices on the order of 3 um, which are relatively easy to prepare by surfactant free seeded polymerizations, would be expected to generate very large backpressures at moderate flow rates. It remains valuable, however, to demonstrate that ATRP can be used to create non-porous EIC media particularly since it enables better understanding of the separation mechanism by minimizing the contribution of AS- . Moreover, application of ATRP technology to nonporous colloidal surfaces advances the understanding of surface initiated ATRP from charged surfaces. Finally, transfer of the knowledge gained provides a stepping stone to the effective use of ATRP on high-throughput chromatography surfaces. PSL is an excellent substrate from which to study surface-initiated ATRP. Monodispersed particles can be conveniently prepared by emulsion polymerization reactions [123] and the synthetic procedures have been well known to the research community for many years. Some applications include the use of latex as surfaces for immunoassays [124], calibration standards [125], and surface coatings [126]. The primary advantage of using latex as a model surface is its high specific surface area. Moreover, suspensions of PSL combine some advantageous features of both solution and surface reactions. For example, rapid mixing of colloidal suspensions helps to provide a high rate of mass transfer to the surface beneficial for promoting surface reactions. The primary disadvantage of latex as a model for surface reactions is that due to the high radius of curvature, particularly for small diameter beads, results may not compare well with flat materials. Furthermore, latices, especially those prepared without added surfactant, necessarily carry surface charges which may not be a desired feature of the surface [127]. For the purposes of studying interfacial ATRP reactions, it is highly desirable that the latex should not contain any surfactant. Although it is possible to remove surfactant from prepared latices with the use of chemical reagents and extensive washing procedures [128], many techniques exist for surface preparation in the absence of surfactant [127; 129-136]. The tradeoff for using the surfactant free technique is that the size of latex is limited to about 1 um [130]. However, seeded growth reactions can be used to build many polystyrene (PS) layers on top of primary particles to produce stable latex up to 3 um or greater [137]. Beaded packings of this size are sufficient for studying the elution properties of small scale EIC columns packed with derivatized media [13]. By following the method of Hritcu, it is possible to synthesize negatively charged polystyrene spheres of approximately 3 um in diameter [122]. Then it is possible to apply to the work of Jayachandran et al. to grow dense PDMA grafts from the latex [21;22]. Work presented in this chapter describes the application of ATRP technology to a beaded model surface for the purposes of creating a working method for producing new and novel EIC stationary phases. 2.1.1. Synthesis and Characterization of Large Latex Particles 2.1.1.1. Emulsion Polymerization of Styrene Goodwin has studied the variables required for successful polymerizations of microspheres (referred hereafter as seed particles) up to about 1000 nm diameter [129;130]. The success of emulsion polymerizations, especially without the aid of a surfactant, is dependent on carefully controlled reagent concentrations, ionic strength, temperature, and stirring conditions. Hansen and Ugelstad proposed a mechanism called homogenous nucleation for the growth of PSL without added surfactant [138;139]. The mechanism proposes that primary particle growth results from an emulsion of unstable particles which, when first created, lack sufficient surface charge to maintain colloidal stability. During the coagulation phase, primary particles coagulate to form aggregates which become stabilized due to accumulation of sufficient surface charge. After initiation, the number of primary particles increases to a maximum and then decreases as particle coagulation dominates. If reaction conditions are such that the number of primary particles increases faster than coagulation, two size distributions of particles will result. This same general theory for homogenous nucleation can be applied to systems with no seed present as in the polymerization of styrene, or to seeded growth polymerization of polystyrene latex to produce larger latex spheres. 2.1.1.2. Seeded Growth of PSL There are many reports for the use of seeded-growth of PSL with and without cross linking for added stability [133; 140-142]. Hritcu adapted the work of Hansen and Ugelstad [138;139;143] for the seeded growth of PSL up to 2.8 um diameter using a positively charged initiator [122]. These methods consider the balance between coagulation causing the size increase of existing particles versus the rate of formation of primary particles. By plotting these variables on independent axes, Hritcu was able to map out areas of stability for successful batches [122]. Hritcu found that the following parameters, adapted from Hansen and Ugelstad, are useful in the synthesis of large PSL microspheres by step growth polymerization. The rate of initiation, p, (L"1 s"1) can be defined by p,=2[I]kDNA [2.1] where / is the ionic strength (mol L" 1), kj is the rate of decomposition of the initiator (s"1); and NA is Avogadro's number (mol"1). The number of seed particles Ns in the reaction mixture (L"1) is Ns=^ [2.2] ms where the subscript s refers to the seed and cs is the seed concentration (g L"1). The value of ms (g) is defined as 4TZT3 m =1.05 '- [2.3] where rs is the radius of the particle (cm) and 1.05 refers to the density of polystyrene (g cm'3). The predicted radius (rcaic) of the growth particle (cm) is [2.4] r x 1/3 ' c + c ' V Cs J where csty is the concentration of styrene in the reaction mixture (g L"1). The general reaction scheme for seeded growth is shown in Figure 2.1. Here, potassium persulfate (KPS) is used for the production of large latex seeds. Styrene K P S ' 7 0 ° C > O + Styrene KPS, 70°C „ pv + s t y p e n e KPS, 70°C First Stage Second Stage V _ y Third Stage PS PS PS 0.9-1um 1.6-2 um 2-2.5um Figure 2.1: Reaction scheme for the multistage growth of polystyrene latices prepared by surfactant-free emulsion polymerization using the initiator potassium persulfate (KPS). 2.1.2. Shell Growth Copolymerization: Synthesis of ATRP Initiator Layer Surface modification of polystyrene latex can be made by polymerizing a functionalized shell on to the 3 r d stage seed particle. The choice of surface modifications is almost limitless i f the desired functional group can be successfully conjugated with a vinylic monomer. In this case, it was desirable to incorporate the ATRP initiator onto the bead surface by a copolymerization with styrene, as in Figure 2.2. The latex bearing the functionalized copolymer is referred to as shell latex. ATRP Initiator Figure 2.2: Synthesis of shell latex: copolymerization of styrene and 2-(methyl 2'-chloropropionato)ethyl acrylate (HEA-C1). Here, HEA-C1 refers to the monomer produced from the conjugation of the acrylic monomer 2-hydroxyethylacrylate (HEA) with 2-chloropropionyl chloride. Through incorporation of this bifunctional monomer onto the surface of latex, a covalent attachment of the ATRP initiator is made to the surface. The ester-linked ATRP initiator is susceptible to hydrolysis which serves two purposes. Polymer chains grown from the ATRP initiator can be cleaved by saponification and the liberated chains can be characterized using laser light scattering and the amount of ATRP initiator incorporated on the surface can also be determined by the same method. A typical titration curve for saponified shell latex is shown in Figure 2.3. 18-a :e 9 o "D c O o . * ' B '' A 0.0 0.1 0.2 Volume of NaOH (mL) — i 0.3 Figure 2.3: Conductivity curve typical of saponified shell latex. End points A and B correspond to equivalence points for -S0 3" and -COO" respectively. According to Stone-Masui [131] separate endpoints for -SO3" and -COO" are detectable from the titration of acidic latices with sodium hydroxide. These correspond to end points, A and B, respectively. Because endpoint B represents the total volume to reach the endpoints of the carboxyl plus the sulfate residues, the volume to reach the sulfate endpoint alone is B minus A. It should be noted that seed latex can also be analyzed by conductometric titrations for determination of the sulfate content. In this case, only end points is observed. 2.1.3. Surface Initiated Atom Transfer Radical Polymerization ATRP has been shown to be a good method for the polymerization of dense hydrophilic grafts from the surface of functionalized PSL by providing a high degree of control over polydispersity and molecular weight [21]. For PSL, Jayachandran et al. hypothesized that the presence of negative surface charge concentrates the positively charged catalyst species immediately next to the surface, compared to the bulk solution. Since the concentration of negative charges at the interface was less than the concentration of initiators, the local concentration of catalyst in the vicinity of initiators was actually in limiting proportions despite a higher concentration of catalyst in the bulk. Under these conditions, when the reaction starts, not all initiators are reacted with Cu(I) catalyst. Once any given chain was initiated and grown long enough to extend outside the surface region, the end competed less for catalyst. In this way, it was suggested that the limiting amounts of and competition for catalyst in the surface region slowed the initiation rate and produced surfaces of very high graft densities [21]. Polymer chains grown on an ester linkage can be saponified, extracted, and analyzed by analytical techniques. To characterize the molecular weight (M„), polydispersity (MJM,) and radius of gyration ((Rg^), chromatographed polymer fractions separated by analytical SEC can be analyzed by laser light scattering combined with mass analysis by refractive index. The detected mass along with a known surface area permits calculation of the distance between grafted chains, D, and the graft density, a. The following equations demonstrate the calculation of these quantities The concentration of PDMA chains in the hydrolysate ([PDMA], in mg mL"1) is found from the injection volume (Vinj, in mL) and the mass of chains injected (mci, in mg). [PDMA] = ^ [2.5] If a representative sample is withdrawn, the sampled amount is directly proportional to the bulk slurry. Therefore, the mass of chains in the bulk (mcg) is equal to the mass of chains in the hydrolysate divided by the mass of hydrolysate (mn), times the mass of the bulk solution (TUB). mcB=^LxmB [2.6] m. The amount of chains can be expressed as cc, the specific concentration of the chains on the surface (in mol g"1) by knowing the chain molecular weight and the reaction weight. It follows that cc= [2.7] Mnmr where mr is the weight of seed used in the reaction (in g). The surface density of chains (pcbam, in chain-nm"2), assuming a uniform distribution across the surface, is calculated from AT c - - A c [2.8] chain 1 q u s s a where 10 1 4 is a conversion factor (nm2 cm"2). The specific surface area (SSA, in cm2g_1) is of PSL and is calculated by SSA = -^— [2.9] 1.05r where 1.05 is the density of bulk PS (in g cm"3) and r is the radius of the latex sphere (cm). It follows that if we assume that each chain occupies a site on a square lattice then the average spacing between surface groups (D, in nm) is D = -r^= [2-10] V ^ chain 2.2. Experimental 2.2.1. Materials and Methods In all experimental and analytical methods given, processed water was used. Purified water was prepared by reverse osmosis and subsequent filtering though a Milli-Q ultrapure water purification system (Millipore). Unless otherwise stated, all chemical reagents were supplied by Sigma-Aldrich. 2.2.1.1. Analytical Methods Evaluation Using the Light Microscope Clean latex samples were diluted appropriately and a single drop was observed at 400 x magnification. The sample batch was inspected to make sure no large aggregates were present. For first or second stage growth reactions, latex particles were barely visible. Any large particles were diagnostic of aggregates or evidence of a possible secondary population. Determination of Solids For the determination of the solid content of latex, small amounts were freeze dried as follows. Approximately 0.3 g of latex was weighed into a test tube of known weight. The sample was kept at -20 °C until frozen and then freeze-dried overnight until the water had sublimed. The final net weight of dry latex solids was determined and the percentage solid (w/w) of the original suspension was calculated. Particle Size Measurements using Dynamic Light Scattering A N4+ particle analyzer (Beckman-Coulter) was used to characterize lots of primary or secondary growth particles. Cleaned samples were diluted appropriately with water, placed in a cuvette and analyzed for 300 seconds at a scattering angle of 90°. Samples that had a low polydispersity were analyzed using the unimodal calculation mode of the instrument. The cut-off range for the instrument was 3000 nm. Latex produced with diameters greater than 2000 nm were subject to instrumental error. Such errors were evident when the measured values were compared to sizing estimates using electron microscopy. Size Determination using Scanning Electron Microscopy (SEM) Samples were diluted to approximately 1% (v/v) so that when dried, individual beads were well separated. Graphite substrates were prepared by polishing them on fine grit abrasion paper and then white paper. A single drop of latex suspension was placed on a substrate and allowed to dry for several hours. Samples were sputtered with gold discharge under vacuum and then transferred to the SEM (Hitachi model S3000N) under ultrahigh vacuum. The beam was set to 20 kV with an aperture of 3. A Robinson backscattering detector was used to collect some images. Images were processed with Northern Eclipse software (Empix Imaging). The size marker from the SEM image was calibrated to the pixel dimension in the digital image. Particles were measured individually, or i f the microspheres did not overlap, an automated counting procedure was used. A comparison of values measured by either method did not differ by more than 2%. 2.2.1.2. Styrene Distillation Styrene was washed to remove the inhibitor and purified prior to use. Styrene was washed 3 times with 1% sodium hydroxide, dried with sodium sulfate and purged with argon. The styrene was purified by vacuum distillation under an argon atmosphere. To prevent spontaneous polymerization, 0.5 g of hydroquinone was added to the distillation flask. The first fraction (10-15 mL) was discarded before collecting the bulk of the distillate. Under a good vacuum, the first portion distilled at 21°C and the bulk fraction distilled in the range of 21-24 °C. 2.2.1.3. Purification of Potassium Persulfate (KPS) Potassium persulfate was purified through recrystallization. Up to 9 grams of crude potassium persulfate was dissolved in 100 mL of warm water (40 °C). The saturated solution was cooled at 22 °C and then at 4 °C until a white flaky precipitate was observed. The precipitate was filtered and washed with small portions of ice cold water. The crystals were dried in a vacuum oven overnight and stored desiccated in a capped vial covered with aluminum foil. 2.2.1.4. Emulsion Polymerization of Styrene For all polymerization reactions, care was taken to make sure that all glassware was clean and free of surfactants. A typical batch of PSL was synthesized as follows. The surfaces of all laboratory apparatus were coated with chromic acid solution, rinsed with water, rinsed with 0.1 M HC1 solution, and finally, with copious amounts of more water. A 4-necked round-bottom flask was immersed in a 70 °C water bath and charged with water and sodium chloride (~ 46 mM). While stirring with a stir paddle attached to an overhead stirrer, the flask was purged with argon through a stopcock for 5 minutes. On another neck joint, a water-jacketed condenser was terminated with a stop-cock which controlled the vacuum from a water aspirator. The argon valve was closed and the system was evacuated for 5 minutes. Argon and vacuum cycles were repeated 5 times. During this time, the styrene was weighed (13.6 % (v/v) in proportion to the aqueous solvent, 450 mL) and purged with bubbling argon for at least 5 minutes. After purging, the flask was bathed in an argon atmosphere and the styrene added. The styrene was stirred for 10 minutes to allow it to come to temperature and to be distributed into insoluble emulsion drops in the aqueous medium. To start the reaction, an aqueous solution of KPS (~ 3 mM) was added from a pressure equalizing dropping funnel attached to the 4 t h neck. Reactions proceeded for 16 hours at 70 °C. To reduce the shear rate at the bottom of the flask, the stir paddle was elevated 10 mm relative to the bottom of the flask and operated at 300 rpm. 2.2.1.5. Dialysis/Cleaning Procedures Latex product was filtered through glass wool to remove aggregates and then transferred to dialysis tubing (Spectra/Por, 5000 molecular weight cutoff, Spectrum Laboratories Inc.) where it was dialyzed against at least five changes of water over the course of one week. The product was transferred to centrifuge tubes and centrifuged at 5000 G for 30 minutes. The supernatant was extracted and discarded. A fresh portion of water was added and the latex was resuspended using a transfer pipette. The wash cycles were repeated four times. After the last centrifuge cycle, the latex was sonicated and stored at 4 °C until needed. 2.2.1.6. PSL Seeded Growth These procedures were the same whether carrying out the first, second, or third stage seeded growth polymerization. Cleaned and sonicated seed was introduced into the reaction flask, diluted to about 3% (w/w) and warmed to 70 °C with stirring at 350 rpm in a water bath. The system was purged with argon. Styrene concentration, fixed at 8.8% (v/v of solution), was added to the solution. The seed was allowed to saturate with styrene for 10 minutes. The system was initiated with the addition of KPS solution (1.14 mM final concentration in solution) which was previously prepared in a dropping funnel. The reaction proceeded for 6 hours. The latex was cleaned by dialysis and characterized after the reaction. 2.2.1.7. Reagent Preparation of HEA-C1 HEA-C1 was synthesized according to a published procedure [21]. The reaction scheme is shown in Figure 2.4. The product, 2-(methyl 2'-chloropropionato)ethyl acrylate is referred to as HEA-C1. o H 2 C = C H - C - 0 - C H 2 - C H 2 - O H + H 3 C - C H - C — C I • " JL. O CI HEA 2-chloropropioyl chloride ATRP Initiator CI N ( C H 2 C H 3 ) 3 ^ H 2 C = C H - C - 0 - C H 2 - C H 2 - 0 - C - C H - C H 3 + HCI DCM o O HEA-CI Figure 2.4: Reaction scheme for the synthesis of 2-(methyl 2'-chloropropionato)ethyl acrylate (HEA-CI). HEA (18.29 g) was weighed into a 3-necked, 500 mL round bottom flask which was immersed in an ice-water bath. Dichloromethane (DCM, 250 mL) was added and the contents were stirred with an overhead stirrer at 350 rpm. Triethylamine (17.53 g) was carefully added. The 2-chloropropionyl chloride, 20 g, was mixed with about 40 mL of DCM'and put in a dropping funnel attached to the flask. The system was purged with argon gas. The steady addition of the reagents to the flask was monitored over the course of an hour. The reaction was left to react for an additional 16 hours by continuous stirring overnight while the water bath warmed to room temperature. The reaction mixture was filtered through filter paper. The filtrate was washed consecutively with saturated NaHCC>3 solution (2 times, 50 mL), 0.1 M HCI (2 times, 50 mL) and water (2 times, 50 mL) in a separating funnel. The organic layer was separated, dried over anhydrous sodium sulfate for five to six hours and filtered. The solvent was removed on a rotary evaporator and then vacuum distilled. Hydroquinone (0.5 g) was added during the distillation to prevent spontaneous polymerization. The low-boiling fraction was collected in a separate chamber of the distillation apparatus. Under vacuum, the product distilled at a temperature above 100 °C. The purified product was characterized by N M R to check the purity and confirm the structure. Proton nuclear magnetic resonance spectra was performed on a Bruker Avance 300 N M R spectrometer using deuterated solvents (CDCI3, Cambridge Isotope Laboratories, 99.8% D) with the solvent peak as a reference. 2.2.1.8. Shell Growth Polymerization The following synthesis follows published reports [21]. The functionalized monomer was incorporated into 3 r d stage growth polystyrene by a seeded shell-growth copolymerization of HEA-C1 and styrene (Figure 2.2). The product contains the surface-immobilized residue of methyl 2-chloropropionate which is referred to as the ATRP initiator. Sonicated seed was introduced into the reaction flask and diluted to 3% (w/w) and warmed to 70 °C with stirring at 350 rpm in a water bath. The system was purged with argon. Styrene was added at either 0.23 or 0.13 g/g of seed. The seed was allowed to equilibrate with styrene for 10 minutes. HEA-C1 (variable amount from 12:1 to 3:1 mol ratio of styrene:HEA-Cl) was then added. After another 5 minutes, the polymerization was initiated with the addition of KPS solution (3.2 mM) scaled in proportion to the weight of latex. The reaction proceeded for 6 hours. Shell latex was filtered, dialyzed and subjected to at least 5 cycles with centrifugation and supernatant with water. 2.2.1.9. Characterization of Shell Latex Conductometric Titrations for Quantification of Sulfate and Initiator Residues To clean shell latex equivalent to 0.4 g of solid, 1.5 mL of 2 M NaOH was added and the sample was gently agitated for 36 hours. The sample was centrifuged and washed 5 times with water. The saponified shell was then acidified by addition of 3 mL 0.1 M HC1, resuspended and left to equilibrate for 5 minutes at room temperature. The sample was then washed 5 times with water until the pH of the supernatant was not significantly different from water. The pellet was resuspended in approximately 10 mL of water and transferred quantitatively to a 20 mL tube with a magnetic stir bar. Procedures for the conductometric titrations were adapted from published methods [144; 145]. The contents of the tube were stirred in a water bath at 25 °C water bath for a 5 minute equilibration period with argon purging. Conductivity was measured with a conductance meter (YSI model 35, 3403 cell with platinum electrode) while titrant (0.01 M NaOH) was delivered at 0.0102 mL/min by a syringe pump (Harvard Instruments). The concentration of initiators, [Initiator] (mmol g"1), was calculated by [Initiator] = ± ^ - 1 [2.11] mshell where B and A refer to the volumes for the end points exemplified in Figure 2.3 (in mL), [NaOH] is the concentration of sodium hydroxide (mol L"1) standardized against potassium phthalate, and msheii is the mass of the shell latex (in g) taken for saponification. The concentration of sulfate ( [-SO3"] ) is calculated by the same equation except using the volume for end point A alone. 'HNMR Determination of Bulk Initiator Content In addition to conductometric titrations, the amount of initiators was quantified by ' H N M R expressed as the total amount present per gram of material. A small amount of freeze-dried shell latex (20 mg) was taken and dissolved in about 1 ml of CDCI3 (Cambridge Isotope Laboratories, 99.8% D) The sample was filtered through a cotton filter and transferred into an N M R analysis tube. Samples were analyzed on a Bruker Avance 300 NMR. In the aliphatic and aromatic regions of the spectrum, peaks at 4.45 ppm correspond to absorbance of the five aliphatic protons in the HEA-CI side chain (-O-CH2-CH2-O- and -CHCI-) whereas the peaks between 6.25 and 7.25 ppm correspond to the five protons on the aromatic ring of styrene. Since the number of protons corresponding to each peak is the same, the peak areas represent the relative abundance of bulk initiator or seed. The amount of initiator calculated from ' H N M R is the total amount of HEA-CI incorporated on the latex. Differences between results by this method and by titration provide an indication of the proportion of initiators that are accessible to aqueous reagents. 2.2.1.10. Distillation of D M A D M A was purified by vacuum distillation. The distillation apparatus was purged with argon for 5 minutes and the D M A and hydroquinone was added. Distillation occurred at 27 °C under a good vacuum. The first 5 mL of the product was collected in a separate chamber and discarded. D M A was stored under argon at -80 °C until required. The structure of D M A is given in Figure 2.5. H 2 C = C H C - N - C H 3 11 1 J O C H 3 DMA Figure 2.5: Structural formula for Af ./V-dimethylacrylamide (DMA) 2.2.1.11. Surface Initiated ATRP Grafting Procedure Surface initiated ATRP from PSL greater than 1 urn diameter has not previously been attempted. Compared to smaller latex, large latex has a much smaller surface area per gram and a smaller radius of curvature. Development work was carried out on small scale batches (0.1 g) to determine the effect of varying the monomer concentration, the surface concentration of ATRP initiators, and the Cu(II) concentration on the resulting graft properties. The ATRP initiator-incorporated shell latex of known solids (0.1 g solids in suspension, 1.34 umol/g initiator concentration) was transferred to a 3-necked round bottom flask and diluted to the appropriate solids (3% w/w) using water. A magnetic stir bar was added, and 2 stopcocks and a stopper were fit to the necks. One stopcock was attached to an argon source and the other to either a vent or a vacuum aspirator. The flask was purged with argon and put into an ice bath. While stirring, the flask was cycled between 5 minutes exposure to vacuum, and backfilling with argon to atmospheric pressure, for a total duration of 2.5 hours. During this time, D M A was thawed and added in excess into a storage vial. Argon was bubbled through the D M A until ready for further use. The catalyst system consisted of the initiator, Cu(I)Cl, an external deactivator, Cu(II)Cl, a zero valent metal for controlling excess deactivator, finely divided copper metal, and the ligand for stabilizing the Cu(I)/Cu(II) redox equilibrium, 1,1,4,7,10,10-hexamethyltriethylenetetamine (HMTETA). These components were weighed into a separate reaction vial as follows: CuCl at 9 times molar proportion to the initiator (0.6 mg), CuCb at 0.15 times molar proportion to CuCl (0.1 mg), HMTETA molar amount was equal to the sum of the CuCl and CuCL; molar amounts (1.6 mg), copper powder was in a 8 times molar proportion to CuCL; (0.5 mg). A non-ionic surfactant used to aid with stabilizing aqueous ATRP reactions, Brij-35, (Pierce Biotechnology Inc.) was calculated to be 0.16% (w/w) proportion of the seed suspension amount (5.3 mg), and was weighed in a separate vial. The degassed seed, purged DMA, catalyst, and surfactant were transferred into an argon-filled glove box. The Brij surfactant was added to the seed and stirred on a magnetic stirrer for at least 5 minutes until it dissolved completely. The desired amount of D M A was added to the catalyst vial. This vial was stirred on a magnetic stir plate until the ionic copper reagents dissolved and complexed with the HMTETA. During this time, the vial was observed to warm slightly indicating an exothermic heat of complexation. Before starting the reaction, the catalyst was allowed to return to room temperature. The reaction was started by adding the calculated weight of seed suspension delivered from a previously calibrated 5 mL pipettor. When the reaction was started, the catalyst vial was capped and left to stir for 24 hours while the reaction proceeded. At the start of the reaction, the latex solution became blue/green and the flask was observed to warm slightly, indicating the start of the polymerization. After 24 hours, the vial was removed from the glove box and exposed to air to ensure that the reaction stopped. Previous studies have shown that the reaction is complete in less than 24 hours [21]. The amount of grafted solid material required to pack chromatography columns (1.2 mL) was 1.5 grams. Successful small-scale recipes were scaled-up 20 times provide enough material for packing and characterization. 2.2.1.12. Characterization of Grafted Latex Analysis of DMA conversion From the vial containing the well mixed reaction mixture, 0.1 g of material was sampled and diluted about 200 times with water. The suspension was centrifuged for 5 minutes and samples of the supernatant were analyzed for D M A conversion using reverse phase liquid chromatography (RPLC) and the polymer formed in solution. The system used was a Hitachi model L-6210 HPLC with an L-4200 UV/Vis detector and a Lichrospher 60 RP-select B reverse phase column from Merck. The mobile phase was 0.1% trifiouroacetic acid (TFA) pumped at a flow rate of 4 mL/min at 22 °C. The 20 uL injections produced a D M A peak at about 6 minutes which was monitored by the detector at X = 235 nm. For every analysis a three point D M A calibration curve was established with samples of known concentrations of D M A . On the Lichrospher column, D M A is of sufficient polarity to be eluted isocratically. Impurities of greater polarity were eluted in the first minute while non-polar impurities remained bound to the column. Peak areas of the replicate analyses differed by less than 1%; furthermore, the baselines during the water injections remained stable and did not show any additional eluted peaks. Recovered D M A in the mass balance represented the portion of monomer not polymerized on the surface or in solution. Graft Hydrolysis The liberation of grafted chains from the surface of the latex support by hydrolysis was accomplished by following published reports [21]. Adapted procedures are given below. About 1 mL of a known amount of grafted latex suspension was sampled into a separate test tube and 10 M sodium hydroxide (67 uL) was added so that the final concentration was ~ 0.67 M . This solution was gently agitated for a period of time until the amount of cleaved chains in solution did not increase (at least 2 weeks). The suspension was centrifuged and the hydrolysate extracted into a separate vial. The pellet was washed with a 1 mL portion of water, centrifuged, and the washings were added to the hydrolysate. Portions of the hydrolysate were filtered through 0.45 um membrane filters (Millipore) into vials prepared for analysis. Evaluation of Liberated Polymer Evaluation of the cleaved chains took place on an analytical SEC system. A pump and autosampler (Waters 2690 separation module) were used to deliver injections of the sample through 2 Ultrahydrogel columns (guard column, Ultrahydrogel linear with bead size 6 - 1 3 um, elution range 103 - 5><106 Da and Ultrahydrogel 120 with bead size 6 um, elution range 150 - 5><103 Da connected in series; from Waters) to the detector modules at 22 °C. The eluent used was an aqueous solution of 0.1 M NaNC>3 filtered 3 times through 0.22 um Millipore membrane filters. The mobile phase was delivered at 0.8 mL/min by the pump. Detection was monitored by in-line M A L L S and RI detectors. The D A W N EOS multi-angle laser light scattering detector (laser A = 690 nm) (Wyatt Technology Corp) was used for determination of molecular weight. Mass determinations were made with a refractive index detector (OPTILAB DSP, Wyatt) at X = 690 nm. Astra software was used to collect and process data from the D A W N and the OPTILAB systems. After selecting the appropriate peak baselines and integration limits, the software algorithm produced a summary report for the peak of interest. For analysis of PDMA, the refractive index increment (dn/dc) used was 0.150 mL/g at 690 nm as previously determined in-house [21]. Data obtained in the summary report included the mass injected, weight average molecular weight (Mw), the molecular weight distribution (MJMn) also known as the polydispersity index, the radius of gyration of the free polymer ((Rg)), and the mass of the sample represented by the peak. Solution polymer contained in the same tubes used for evaluation of D M A recovery was also analyzed on the SEC/MALLS system. The analysis characterized the properties of the solution polymer as well as the total amount of polymer in solution formed along with graft polymer from the mass balance. 2.3. Results and Discussion 2.3.1. Seed Growth Polymerization: Synthesis of Polystyrene Latex (PSL) Particles 2.3.1.1. Surfactant-Free Emulsion Polymerization of Styrene Goodwin's equation was found to be a reasonably good predictor of size of latex produced by surfactant-free emulsion polymerization of styrene [130]. Using an initiator for producing negatively charged latex, KPS, the ionic strength and concentration of initiator remained constant and the monomer concentration and temperature were varied in an attempt to synthesize large latex particles in one step. Comparisons with predicted values are provided in Table 2.1. Goodwin observed that latex sizes limited to approximately 1 um could be obtained at 55 °C [130]. A communication by Tuin et al. showed that large (3.2 um diameter) PSL could be obtained in one step by a surfactant-free technique using KPS [134]. An attempt was made with Lot A to try and replicate this work. Lowering the temperature to 55 °C forced constrained variables outside of the empirical ranges set by the equation of Goodwin. It was not surprising that Lot A differed so significantly from the predicted size. It is likely that the mechanical specifications of the reactor design and the stir paddle are crucial factors that allowed Tuin et al. to achieve results not predicted by Goodwin and not reproduced in this lab. Table 2.1: Comparisons of predicted and experimental latex diameters from surfactant-free emulsion polymerization of styrene using potassium persulfate initiator. Lot 7" Styrene [NaCl] [Initiator] Predicted Experimental Experimental cone. diam. n diam. (st. dev.) | diam. (st. dev.) J °C % (v/v) M mM nm nm nm A 55 18.4 0.050 0.22 2454 833 (360) n/d B 60 18.4 0.057 0.49 1865 1006 (440) n/d C 60 10.2 0.040 2.8 891 899 (410) n/d D 60 13.3 0.060 2.0 1185 1058 (230) n/d E 70 6.62 0.042 3.0 589 1592 (710) 1280(370) * F 70 9.93 0.042 3.0 696 1183 (520) 1200(80) * G 70 13.2 0.042 3.0 783 1581 (540) 1050 (50) H 70 17.6 0.057 2.0 1037 1277 (600) n/d 1 70 13.6 0.046 3.2 793 897 (390) n/d J 70 13.6 0.046 3.2 793 940 (300) n/d K 70 13.6 0.046 3.2 793 1007 (450) 1010 (50) L 70 13.6 0.046 2.8 814 1991 (980) 1580 (140) * M 70 17.6 0.047 2.9 905 1177 (550) 1130 (80) N 70 13.6 0.046 2.8 814 951 (400) n/d 0 70 13.6 0.046 3.2 793 900 (350) n/d n calculated by the method of Goodwin [130]. | as determined by light scattering using the N4+ particle sizer X as determined by SEM imaging * results of these trials were not reproducible n/d results not determined Lot E revealed that particles can be produced that yield large size; however, as the results could not be replicated, it demonstrates that variables that are difficult to control, such as small variations in stirring method, can produce batches of unusual size. As the later lots show, it was relatively easy to prepare consistent batches of 1 um latex. This method was used to prepare, for example, Lot P, a large batch which yielded 193.2 g of latex suspension (23.1% w/w) representing an 80% conversion of styrene. To determine the particle size and uniformity, latex particles were analyzed by SEM and particle size analysis by light scattering. A representative SEM image of PS latex particles is shown in Figure 2.6. Figure 2.6: SEM micrograph of polystyrene latex Lot G. 10 urn scale shown. The histogram of counts of latex sizes is shown in Figure 2.7. BO-S' c of 40' 20 I 1 I 1 I 1 I ' 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 —r 2.0 Diameter of PSL (nm) Figure 2.7: Frequency histogram of counts of measured latex diameter from digital image processing, n =182, mean = 1.053 urn, standard deviation = 0.053. For Lot G, image analysis produced a value for the mean calculated to be 1.053±0.008 um. Samples were also analyzed by light scattering. On average, comparisons between these two measurements resulted in a value 20% lower for SEM (Table 2.1); however, the difference was much greater for this lot. An error in the instrument's unimodal calculation mode was found to be the cause. Because of this, the accuracy of the light scattering method was found to be much lower and less reliable than the SEM. On average the relative standard deviation in the light scattering method was from 5 to 10 times higher than for the SEM method. Although the precision of the light scattering measurements was less than the SEM method, the consistently higher diameter measured by light scattering in solution suggests that differences in measurement is likely caused by the hydrodynamic size of the particle. This is plausible since SEM measurements are made in vacuo and light scattering measurements made in solution. For the purposes of synthesizing seed latex for further reactions, the light scattering method was preferred for speed and convenience while the. SEM method was useful for precise, accurate measurements of the (dehydrated) size of the latex. Seeded Growth Polymerizations Stability maps from Hritcu's work were used to predict synthesis conditions for step-growth polymerization of PS microspheres [122]. Even though Hritcu used a cationic initiator with reaction temperatures at 50 °C, stability maps for the anionic system were suitably applied for reactions at 70 °C. The applicability comes from the descriptor Ns x rs which describes the physical properties of the seed independent of chemical composition. Secondly, equation 2.1 was modified for KPS using the data presented in Table 2.2 [146]. Table 2.2: Rate of decomposition of potassium persulfate (KPS). Temperature kdx~\0 °C s"1 50 0.95 60 3.16 70 23.3 80 91.6 In Hritcu's work, the predicted stability region for first stage growth was (2.12 < pt < 4.24) x 101 6 L" 1 s"! and (2.86 < Nsrs < 5.04) x 108 dm"2. Outside these conditions, the seed was found to be aggregated or creamed (initiation rate too low), or bimodal (many particles produced of small radius). As a first attempt, the means of these ranges were used as starting synthesis conditions. Fixing these experimental variables allowed for the calculation of the expected particle size from Equation 2.4. For seed A (12 g of styrene, 6 g seed PSL, 150 mL total aqueous reaction volume), the first stage growth reaction produced a product free of aggregates or a secondary population (Table 2.3). Seed B (36 g styrene, 18.5 g seed PSL, 450 mL total aqueous reaction volume) also produced viable latex. Seeded polymerizations were stable in the absence of added ionic modifier. Table 2.3: Predicted and experimental latex diameters from synthesis conditions based on empirical stability diagrams. Stage Batch Reaction volume Seed Cone. Styrene Cone. Initiator Cone. Diameter (predicted) Diameter (experimental) mL % (w/w) % (v/v) mM urn um 1 s t A 150 4.51 8.8 1.13 1.4 1.5 B 450 4.11 8.8 1.14 1.4 1.3 A 150 8.33 8.8 1.17 2.0 1.9 B 450 4.32 8.8 1.17 1.8 1.8 3 r d A 150 3.85 8.8 1.38 2.8 2.7 B 450 3.40 8.8 1.38 2.7 2.8 B 1000 3.20 8.8 1.38 2.7 2.4 * A l l size determinations analyzed by light scattering except * as determined by SEM. Subsequent (2 n d and 3 rd) stage growth was aided by modified stability diagrams after the method of Hritcu [122]. For second stage growth: (2.25 < pt < 4.30) x 101 6 L" 1 s"1 and (2.20 < Nsrs < 2.60) x 108 dm"2. For third stage growth: (2.75 < pt < 5.00) x 101 6 L"1 s"' and (0.97 <JV>* < 1.18) x 108 dm'2. The stability map developed is shown in Figure 2.8. T < 1 1 1 1 r 2 3 4 5 p.x 10"1 6(dmV) Figure 2.8: Empirical stability map for regions of stable latex produced from seeded growth of PSL. Regions outside of the enclosed areas produced either aggregated latex (for low pj) or bimodal populations (for high values of pi). The large scale, 3 r d stage growth reaction (80 g of styrene, 32 g seed PSL, 1000 mL total aqueous volume), yielded sufficient material (83 g of PSL) for all further experimental work. 2.3.1.2. Characterization of Seed Latex Particle size analysis by SEM shows a uniform distribution of particles after the 3 r d stage growth. A representative SEM micrograph is shown in Figure 2.9. Figure 2.9: SEM image of 3rd stage seeded-growth latex (Batch B, large scale reaction). 20 urn scale shown. In general, the 3 r d stage growth latex was uniformly distributed. A few large particles were seen but these did not make up a significant proportion of all analyzed beads when size distributions were analyzed. The final size, between 2.4 and 2.7 um depending on the batch, was found to be sufficient as a chromatography matrix, and a 4 t h stage growth reaction was not necessary. Light scattering could not be reliably used for size determination of latex approaching 3 um in diameter because of instrumental limitations. 2.3.2. Synthesis and Characterization of ATRP Initiator-Functionalized Latex Particles (Shell latex) Shell synthesis reactions were deemed successful by maximal incorporation of ATRP initiators. Factors that influenced the copolymerization of styrene and HEA-CI were the ratio of styrene to HEA-CI, the concentration of seed in the starting vessel, and uniformity of stirring. The demand for consistent, large batches of shell product made scaling up of the reaction a priority. The greatest challenge when attempting to adapt these synthetic procedures from published accounts was achieving a large amount of initiator incorporation without aggregation. The concentrations of surface initiator and sulfate groups as determined by conductometric titrations after saponification of shell latex are summarized in Table 2.4. Table 2.4: Evaluation of [initiator] and [sulfate] by conductometric titration for batches of shell latex. Batch Feed Ratio [HEA-CI] (ATRP initiator) [Sulfate] Styrene: HEA-CI mol/g x 10 6 mol/m 2 x 10 7 mol/g x 1 0 6 m o l / m 2 x 1 0 7 M10 12:1 2.7 11 3.3 14 L1 9:1 0.67 2.8 2.3 9.6 L3 9:1 1.6 6.4 1.5 5.6 L8 3:1 3.6 16 1.4 6.0 L9 3:1 2.9 13 1.4 6.4 From the above data the average spacing between initiators was calculated to be approximately 1-3 nm, and 0.8 - 1.3 nm for sulfate groups. Analysis by ] H N M R confirmed the bulk incorporation of ATRP initiators. A representative spectrum is shown in Figure 2.10. 1 A . 0 Figure 2.10: 'H N M R spectrum from a sample of shell latex, L8. Sample run in CDC1 3, 300 MHz 'H N M R with delay time of 1 second. The peaks for the aliphatic protons on HEA-CI at 4.45 ppm and from the aromatic protons on styrene from 6.25 and 7.25 ppm both represent five protons each. The compared peak areas, therefore, reflect a measure of the relative abundance. The aromatic protons were in greater in abundance by 150.7 times the number of aliphatic protons. Since in the bulk of the sample, PS was in a much greater abundance than HEA-CI, the molecular weight of PS (104 g/mol) could be used to represent the molecular weight of the shell latex. The concentration of HEA-CI ([HEA-CI], mol/g) was calculated from 1 [HEA-CI] = 104xi? [2.12] where R is the relative abundance of PS to HEA-CI as measured by N M R peak areas. Comparisons between the *H N M R data and conductometric data for shells L8 and L9 (replicate batches by the same recipe) are given in Table 2.5. Table 2.5: Comparison of results for [HEA-CI] as determined by N M R and by titration. Batch Feed [HEA-CI] [HEA-CI] by 1 H NMR [HEA-CI] by titration HEA-CI accessible to aqueous reagents mol/g x 10 5 mol/g x 10 5 mol/g x 10 5 % of bulk L8 44 6.4 0.36 5.6 L9 44 5.3 0.29 5.5 The difference between the feed concentration and the amount detected by N M R shows that the amount of HEA-CI actually incorporated into either seed was between 12 and 15%. The low incorporation was likely due to coagulation of a portion of latex. The overall yield of the reaction was 65% with the loss evident as aggregates that were filtered out of the product after the reaction. There is a large difference between the value of [HEA-CI] detected by ] H N M R and by conductometric titration. Previous studies have suggested that the accessibility of an analyte to aqueous titrant is decreased if the active groups are buried in the matrix [22; 122]. A styrene to HEA-CI ratio of 3:1 has been shown by calculation to produce a random copolymer [21]. For shell lots L8 and L9, one would expect a random distribution of styrene and HEA-CI in the shell. Results from Jayachandran et al. show that shell synthesis produces an observable size increase as the difference between the hydrodynamic thicknesses of the shell latex compared to the seed as measured by light scattering [21]. The hydrodynamic size is representative of the diffuse network of chains on the surface supported by aqueous solvents and amounts to an increase in diameter between 68 and 110 nm for beads on the order of 500 to 700 nm [21;22]. A SEM micrograph of a typical sample of shell latex is shown in Figure 2.11. Figure 2.11: SEM image representative of the uniformity of synthesized shell latex. 10 u,m scale shown. The hydrodynamic size could not be determined due to particle sedimentation in the dynamic light scattering apparatus and the high vacuum present in the SEM which removes water from the sample. The observed size of shell latex was virtually identical to seed latex as measured by SEM. For the large scale batch B, the parent latex had the same value and error (2.40±0.03 um) as the average size for ten shell latex batches. Given the magnitude of the error from the analysis procedure, it is unlikely that any statistically significant size increase from the mean would have been observed for shell latex using SEM. 2.3.3. Aqueous Surface Initiated ATRP of AyV-dimethylacrylamide In this section, 5 lots of shell material were polymerized and are used for comparisons in the discussion. The ATRP reaction conditions of these materials are reported in Appendix 2. Table 2.6: Summary of initiator and charge concentrations for various shell latex described in this chapter. The graphical symbol is conserved in the figures that follow. Shell Graphical Symbol [Initiator] mol/g x 10 6 [Charge] mol/g x 10 6 L3 • 1.38 4.45 M9 • 1.49 4.03 M8 x 1.61 3.80 M5 + 2.06 3.39 M10 • 2.72 2.93 2.3.3.1. Effect of Changing the Monomer Concentration At constant surface initiator concentration, surface initiated ATRP is expected to give a linear increase in molecular weight of the grafted polymers with increasing monomer concentration [20]. The graph of M„ versus monomer concentration for a series of shell latex with different initiator concentrations is shown in Figure 2.12. 1500000-1000000-CO Q 500000• 0.0 —I— 0.7 [DMA] (M) — i 1.4 Figure 2.12: Effect of increasing D M A concentration on resulting graft molecular weight for shells M5 (+), M8 (X), M9 O) ,M10 (A) , L3 (•). Graft molecular weight increased roughly linearly with monomer concentration for all cases. For shell L3, graft molecular weights in excess of 1250 kDa were observed making these some of the highest molecular weight grafts observed for surface initiated ATRP. The order of the series does not seem to follow trends in either initiator or sulfate content. However, since only small differences in initiator or sulfate surface concentration exist, it would be hard to explain the large difference in observed M„ based exclusively on either of these parameters. i . o H — , — , — , — i — . — i — . — i — i — i — • — i — - — i 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 [DMA] (M) Figure 2.13: Effect of increasing D M A concentration on resulting graft polydispersity for shells M5 (+), M8 (x), M9 (•), M10 (A) , L3 (•). Solution polydispersity is given as the empty symbol for shells M10 (A), and L3 (•). The polydispersities of the polymers are shown in Figure 2.13. Trends in polydispersity indicate that for surface polymerization, the system was well controlled (MJM„ < 1.3) and does not appear to increase with monomer concentration. In some cases, polydispersities as low as 1.1 were observed indicating the highly controlled nature of ATRP experiments. Along with graft polymer, the polymer formed in solution was also characterized. For solution polymer there was a noticeable reduction in control, demonstrated by the higher values of MJM„. The polydispersity of the solution polymer tends to increase slightly with increasing monomer concentration. The reason for the higher polydispersity in solution as compared to graft polymer is because the reaction conditions have been optimized for the surface polymerization. Representative data for solution polymer are shown in Table 2.7. Table 2.7: The weight average molecular weight, conversion, and fate of polymerized D M A for Shell M10 as a function of monomer concentration. Shell [DMA] Graft Mw Solution Mw Conversion Solution polymer * Graft polymer * M kDa kDa % % % M10 0.16 274 190 28.0 99.4 0.6 0.31 570 n/d n/d n/d n/d 0.52 970 716 35.3 99.2 0.8 0.83 1287 978 n/d n/d n/d 1.0 1417 1133 34.7 99.3 0.7 1.6 1614 1343 34.6 99.5 0.5 * calculated as a percentage of the total conversion n/d not determined Along with a small amount surface graft polymerization, nearly all of converted D M A goes to form polymer in solution. The data presented are representative of trends in all of the other shell latex studied. In general, the molecular weight of the solution polymer is from 40 to 85% higher than the graft polymer. Here Mw is used to express molecular weight in order to account for the higher polydispersities of the solution polymer. The reason for the formation of solution polymer may be that transfer of radicals from the surface to solution occurs in the early stages of the polymerization [22]. This argument is supported by the lower molecular weight of solution polymer compared to grafted polymer (Table 2.7). The mechanism for the radical transfer is unknown but it has been suggested that restricted diffusion of monomer at the surface could result in an increased tendency for generated radicals to transfer either to ligand or solvent [21]. These radicals would presumably diffuse rapidly to the bulk solution and begin polymerization in solution. The greater mobility of forming solution polymer permits greater access to solution-associated reagents and may account for the dominance of solution polymer. However, as seen by MJM„, these polymerizations lack the type of control facilitated by surface reactions. When looking at data from all shells, the conversion varies from 20% to 55%. In every case, the vast majority of monomer is converted into solution polymer. The proportion of solution polymer was not affected by the monomer concentration. The variation in graft density is shown as a function of increasing monomer concentration in Figure 2.14. 0.0100' o 0.0010- A + XA —1 1 1 1 1 1 1 1 I 1 ' 1 ' 1 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 [DMA] (M) Figure 2.14: Effect of increasing monomer concentration on graft density for shells M5 (+), M8 (x), M9 (•), M10(A) , and L3 (•). Graft density appeared to increase slightly with increasing graft density with the largest increases observed at lower monomer concentrations. This same trend is not observed in solution polymerization of D M A [101; 102] but has been observed for surface initiated polymerization from PSL shell latex with high initiator content [22]. Figure 2.14 shows that the shell with the lowest initiator content ( A ) seems to have a graft density which is the most invariant to monomer concentration. The two shells with the lowest initiator content, L3 (•) and M9 (•), have the highest graft densities but the graft density does not follow trends in the initiator content for the rest of the series. Graft density, however, is not a function of initiator content alone. It has been suggested that other factors such as sulfate content and hydrophilicity or hydrophobicity of the shell also may affect graft density in surface initiated polymerizations [22]. Kizhakkedathu and Brooks also suggest that the sum total of initiator and sulfate groups might be a determinant of the order of graft density [22]. As Table 2.6 shows, the total of initiator and sulfate groups for all shells studied here are approximately constant and therefore it is unlikely that small differences in this sum could be used to explain the differences in graft density seen in Figure 2.14. Since either sulfate or initiator content could affect the grafting density, competing effects of these two variables could be the cause of the variation in graft density. As noted, the nature of the latex system makes this factor very hard to investigate [22]. Since stable emulsifier-free PSL requires sulfate charges to be present, it is impossible to study initiator and sulfate content independently. For nearly all polymerization reactions studied, graft morphology in the brush regime was observed. Table 2.8 shows the effect of increasing monomer concentration on graft morphology. Table 2.8: Physical properties of PDMA grafts cleaved from latex with varying monomer concentration for shell L3. The graft density, radius of gyration, and ratio of distance between grafts divided by radius of gyration are shown. [DMA] agran fa) Dlfa) M chain/nm 2 x 10 3 nm 0.16 6.07 22.2 0.58 0.31 4.49 28.8 0.52 0.52 14.9 38.7 0.21 0.83 19.1 57.3 0.13 1.0 18.6 61.5 0.12 Graft density generally increased as monomer concentration increased. Since yRgj increases with M„, the net effect of dividing D by fa) is that graft morphologies become more brush-like for increasing monomer concentration since increasing monomer concentration produces high molecular weight grafts. Very dense polymer brushes are favoured when both monomer concentration and graft density are high. With graft M„ in excess of 106 Da and polydispersities for this series below 1.2, these densely grafted systems are some of the most controlled, high molecular weight brushes reported. Initiation efficiency is calculated by the amount of initiators used for polymerization divided by the total initiator content for a given shell latex. The initiation efficiencies for two shell latices are given in Table 2.9. Table 2.9: Initiation efficiencies for shells containing relatively low (M10 = 2.93 X 10"6 mol/g) and high (L3 = 4.45 x 10"6 mol/g) sulfate concentrations. Initiation Efficiency [DMA] M Low [sulfate] shell M10 % High [sulfate] shell L3 % 0.16 0.15 1.6 0.31 0.32 1.2 0.52 0.27 3.9 0.83 0.25 4.9 1.0 0.27 4.8 1.6 0.24 3.3 For all reactions studied, the initiation efficiencies did not exceed 5% and is similar to values reported in the literature [20; 147; 148]. It has been suggested that the relative sulfate content may affect the initiation efficiency of surface initiated polymerizations of D M A form PSL [22]. Sulfate groups attract positively charged catalyst and concentrate it at the surface. A result of this could be higher initiation efficiency. It should be mentioned, however, that the very small increase in sulfate content in these two seeds (1.5 times) would be an unlikely reason to explain the 10 times variation in observed initiation efficiency as the only factor. For these two seeds it should be noted that initiator content changes as well. 2.3.3.2. Effect of Changing the Cu(ID Concentration The effect of increasing the proportion of Cu(II) to Cu(I) is shown in Figure 2.15. Two lots at different monomer concentrations are compared to show the similar trends between data sets. Increasing the Cu(II) concentration is expected to shift the ATRP equilibrium to favour radical dormancy according to the general ATRP reaction scheme shown in Figure 1.2. This will reduce the propagation rate and could potentially improve the control of polymerization. Others have used this method as a way of improving control over the ATRP reaction [20;102;149]. 2.0 - i [Cu(ll)/Cu(l)]x 100 Figure 2.15: The effect of changing Cu(II) proportion on the molecular weight (filled symbols) and polydispersity (empty symbols) for two systems with differing monomer concentrations; 0.16 M (•), 1.56 M (•). Shell was L3. As expected, the MN for grafted polymer decreased as the Cu(II) proportion increased. There appears to be little change in the polydispersity in the case of the surface polymer, as can be seen from Figure 2.15 and Table 2.10. One reason for this may be that with MJMN already less than 1.4 for surface polymer, these polymerizations had little to gain by increasing the Cu(II) concentration. The controlling effect of high Cu(II) to Cu(I) proportion, however, is evident for the solution polymer as seen in Table 2.10. Table 2.10: P D M A polydispersity as a function of catalyst deactivator (Cu(II) ) proportion for both surface and solution polymer for shell L3. low [DMA] high [DMA] [Cu(ll)/Cu(l)] MJMn MJMn x 100 Surface Solution Surface Solution 15 1.17 1.67 1.10 2.84 50 1.36 1.54 1.29 1.95 100 1.30 1.55 1.26 1.61 200 1.37 1.38 1.22 1.48 l o w [DMA] = 0.16M; high [DMA] = 1.6M The Cu(II)/Cu(I) proportion had an interesting effect on the graft density in that the graft density increased through a maximum when the proportion of Cu(II) to Cu(I) was increased from 0.15 to 2 (Figure 2.16). This same trend was observed in three data sets for lots with differing monomer concentrations. 1 ' 1 ' 1 ' 1 ' 1 • 0 50 100 150 200 [Cu(ll)/Cu(l)]x 100 Figure 2.16: Effect of catalyst deactivator (Cu(II) ) proportion on the graft density for 3 systems with differing monomer concentrations; 0.16 M (A) , 0.56 M (•), 0.83 M (•). The shell latex was L3. The observed changes in the graft density were quite small with the maximum density representing an increase of about 0.007 chain/nm2 over the next two highest points at 0.56 and 0.83 M . The change was even smaller for the low monomer case. However, given that this phenomenon was observed for 3 different monomer concentrations some speculation of this phenomenon is warranted. Overall, one would expect that increasing the Cu(II) proportion would cause an increase in radical deactivation. While this effect has been used to great effect in controlling the system polydispersity in the propagation reactions, little is known about how it affects the initiation reactions as must be the case here. In the surface model presented, Jayachandran et al. recognize that the initiation of chains on PSL is a complicated process considering that the surface interface is a region of high surface activity and low solvent content [21]. It is unclear why at the proportion of 0.5 (Cu(II) to Cu(I) ) that a maximum in graft density is observed. It may be that the initiation efficiency is improved because as Cu(II) is increased, a portion becomes more closely associated with sulfate residues because of the higher charge which is likely to bind more to the negative groups, thereby allowing a relatively larger population of Cu(I) complexes to be available to initiators. At higher Cu(II) concentrations, the deactivating effect becomes dominant and shuts down radical initiation for surface initiators thereby lowering the initiation efficiency. Because so many possible factors may influence this phenomenon (i.e. surface hydrophobicity, copper redox equilibrium), more studies should be undertaken in an attempt to better explain the role of Cu(II) in influencing initiation efficiencies in surface-initiated polymerizations of negatively charged surfaces. 2.3.3.3. Image Analysis of Grafted Latex Grafted material was visualized by SEM and a representative example is shown in Figure 2.17. Fuzzy areas between the particles are evident in grafted material and not seen in any of the images of ungrafted samples (cf. Figure 2.11) What looks like "webbing" could be due to interaction of polymer grafted layers causing areas of bridging between the particles. Figure 2.17: SEM image of grafted latex. There appears to be some webbing joining the latex particles. A 5 um scale is shown. Using image analysis, the average latex diameter was estimated at 2.44±0.03 um. The parent shell latex was 2.30±0.04 um implying an apparent increase in radius 0.073±0.026 um following the synthesis (non significant digits included to show the magnitude of the error). This lot had a M„ of 184 kDa and the (Rg) of the free chain in solution was 19.5 nm. Given that the distance between grafts was 25.1 nm, the Dl{Rg) value of 1.3 indicates that the grafted layer would have the brush morphology. From an empirical calculation relating the hydrodynamic thickness of PDMA grafted brushes to their M„ [150], an estimated thickness of the grafted layer was calculated to be 0.285 um. From these estimates it seems clear that SEM imaging in vacuo does not capture the true size increases of grafted PSL, due to a loss of incorporated solvent in the brushy layer which would support a larger, hydrated graft thickness. 2.3.3.4. Comparisons with Published Research The results presented show a fairly consistent picture with results reported by Jayachandran et al. [21] and Kizhakkedathu and Brooks [22] for surface initiated ATRP from PSL, albeit on smaller sized particles. Dense PDMA brushes could be synthesized with good control over molecular weight and polydispersity. When monomer concentration and initiator content were increased an increase in graft density was observed for shell latex of higher initiator content but was relatively invariant for a shell with lower initiator content (as shown in [22]). The dense grafting behaviour likely results from a concentration of catalyst on the surface. It was shown that shell latex shows the same reduced accessibility to aqueous reagents as previously studied although it was not possible to estimate the concentrating effect because the hydrodynamic volume of the shell could not be measured. Future studies could use A F M methods in solution to provide an estimate of this volume [115]. Variation in the sulfate content showed a weak correlation to the initiator efficiency; however, this picture may be complicated by other variables such as initiator concentration. As identified, the nature of surfactant-free PSL prevents investigation in the absence of surface charge [21;22]. The changing density with variation in Cu(II) was not seen in earlier published work; however, in the results above and published studies, the variation in density is very small over large changes in Cu(II) concentration [21]. 2.3.3.5. Scale-up Reactions Based on reaction conditions for the controlled growth of successful batches of PDMA-grafted brushes from PSL, ATRP reactions were scaled up from 0.1 g to 1.5 g for the intended purpose of chromatographic evaluation. The synthesis conditions are listed in Appendix 2 and the graft characterization is listed in Table 2.11. Table 2.11: Properties of grafted latex batches suitable for chromatographic studies. Evaluations of the number average molecular weight, polydispersity, graft density and DIRg are shown over increasing monomer concentration. Batch [DMA] MJMn D / { R S M kDa chain/nm 2 x 10 3 L2Graft 1.0 184 1.26 1.59 1.3 L4Graft 2.0 124 1.34 21.2 0.4 L5Graft 2.0 203 1.29 0.746 1.8 L8Graft 1.0 59 1.37 5.24 1.4 Reactions were scaled up successfully producing more than enough material for column packing and characterization. By choosing a latex suspension of 9% (w/w), sufficient material was grafted without substantially increasing the total reaction volume. The variation in graft molecular weight and graft density provided a wide range of physical properties for chromatographic evaluation (graft M„ between 59 and 184 kDa; D/(RG) between 0.4 and 1.4). A l l materials were in the brush regime with the grafted material from batch L4 being a very dense brush. Chromatographic evaluations of these materials are discussed in Chapter 4. 2.4. Summary A multi-step synthesis procedure was used to produce grafted PSL. The intended use of the media for chromatographic applications required that the latex be of a relatively large size, that polymer grafts could be grown from the surface, and that the grafted layer could be exhaustively characterized. Using seeded growth emulsion polymerization of styrene, polystyrene beads were grown to 2.5 um, which is comparable to some chromatography packings used in analytical SEC columns. Incorporation of a copolymer shell containing initiators on a hydrolyzable linkage allowed for surface grafting of polymer from the surface to take place. Grafting was performed by aqueous atom transfer radical polymerization. Good control over molecular weight and polydispersity were obtained. In nearly every case, dense grafting densities were observed and the grafts were calculated to be true polymer brushes. The sulfate content of the shell latex appeared to influence the initiation efficiency; however, other factors such as the initiator content also play a role. It was not possible to test this assumption independent of other variables because sulfate charge is a physical requirement for colloidal stability of these latices. Salient points concerning aqueous ATRP were learned. The specific chemical environment of the initiating surface is paramount in the production of dense polymer grafts. In future designs of initiating surfaces it would be beneficial to use chemical synthesis to include targeted concentrations of initiators and sulfate. Moreover, attempting polymerization in the absence of charged groups may provide insight into the nature of PDMA grown using surface initiated ATRP. These suggestions were incorporated into design methodologies for the synthesis of high-throughput EIC media outlined in the next chapter. 3. SYNTHESIS AND CHARACTERIZATION OF A HIGH THROUGHPUT ENTROPIC INTERACTION CHROMATOGRAPHY SUPPORT 1 3.1. Introduction With advanced materials and polymerization procedures, there have been many new polymer-based materials produced for use as chromatography packings in recent years. Such materials are characterized by their use over a large range of pH and ionic strength, their size uniformity, and their mechanical robustness [8]. Mechanical strength has been an important consideration for the design of high throughput chromatography media. To achieve high flow rates through a chromatography matrix, the material should be of reasonably high porosity. By supporting a large pore volume, the amount of matrix in the column is reduced allowing for a greater fraction of the total column volume to be devoted to use as a chromatographic partitioning phase. However, proper rigidity is required to prevent matrix collapse under flow pressure. Traditional SEC demands that resins have a defined pore structure which imposes a limitation on the number of conformations available to a solute. For protein separation applications, pore size necessarily has to be on the order of the size of the analytes of interest and the distribution of pore sizes is the criterion for determining the resolution of the column. Therefore high throughput applications for SEC have been limited because a large portion of the column volume is needed to support an architecture of pores. EIC matrices could be a good choice for application of size separation technology to high-throughput separations. Since grafted layers can be supported on a highly porous 1 A version of this chapter will be submitted for publication. B. R. Coad, J. N. Kizhakkedathu, C. A. Haynes, and D. E. Brooks. Synthesis of Novel Size Exclusion Chromatography Support by Surface Initiated Aqueous Atom Transfer Radical Polymerization. Macromolecules. (2006). matrix, high throughput separations are possible so long as there is sufficient mass transfer of the solute to the grafted layer. As seen in Chapter 2, ATRP is a versatile polymerization method capable for growing dense polymer brushes with good control over graft properties. To date, there are no published accounts of applying ATRP to create high-throughput SEC matrices. The work in Chapter 2 and published accounts [21;22] demonstrate that ATRP of D M A by surface initiated ATRP from PSL is understood well enough to permit the design of dense polymer brushes from non-porous supports. However, it was also suggested that the negative surface charge influence polymerization behaviour and it has been suggested that other factors such as surface hydrophobicity / hydrophilicity may play a role as well. The adaptation of surface initiated ATRP technology to porous surfaces should address these design issues. Knowledge gained from work with non-porous supports has also led to the development of excellent strategies for polymer graft characterization. These features should also be applied. The research described in this chapter shows efforts to extend success with grafting dense hydrophilic polymer brushes on non-porous latex particles to a highly porous chromatography support. A novel approach to producing tunable EIC matrices is presented by demonstrating molecular weight and density control of grafted polymers with full characterization of the grafted layer. 3.1.1. Synthesis and Characterization 3.1.1.1. Choice of Matrix The commercially available matrix, Toyopearl AF-Amino-650M (hereafter referred to as Toyopearl) was chosen for modification. The matrix is of large diameter (65 um nominal bead diameter) and is highly porous (1000 A nominal pore diameter). The high flow properties of the matrix make it an ideal choice for attempting to synthesize a chromatographic medium useful for high-throughput applications. The starting material contains amine groups which can be used for surface attachment chemistries. 3.1.1.2. Estimation of Matrix Surface Area In order to obtain accurate graft density estimations, it was necessary to determine the surface area of the Toyopearl matrix. Although there are many methods for determining surface area by probing with small molecules such as nitrogen or mercury, it was desirable to measure the fully hydrated surface area. Furthermore, one should choose a probe of size similar to a randomly-coiled PDMA chain in a good solvent in order to effectively model the physical environment of grafted chains. If one were to probe the surface area with, for example, nitrogen, the surface area reported would be a poor representation of the matrix as used in grafting experiments since the probe would have access to small interstices within the (dehydrated) matrix. This could possibly over represent the surface area by probing areas within the matrix where macromolecular polymer chains could not exist. The probe chosen for this study was bovine serum albumin whose geometry can be approximated by a polygon with equilateral triangular faces of 8 nm long and a thickness of 3 nm [151]. By comparison, the majority of the hydrolyzed polymer grafts in this study had radii of gyrations ranging from 2.5 nm to 9 nm. Therefore the estimated surface area used for calculations of graft density should be representative of the physical environment within the matrix where polymer chains are expected to exist. 3.1.1.3. Functionalization of the Toyopearl Surface The amine groups present on the Toyopearl matrix are not of sufficient density to permit studies for investigating how both high concentrations of sulfate and initiator groups function in the growth of potentially dense polymer grafts by surface initiated ATRP. The first synthesis stage should be to multiply surface functionalities to allow for the creation of many potential derivatization sites. Many charged sulfate groups could then be introduced along with ATRP initiators. Furthermore, it would be desirable to have ATRP initiators incorporated on an ester linkage to allow for degrafting. Ring opening of a hydroxyl-rich lactone could be one way to convert a single amine group into many hydroxyl groups. Modification of surface hydroxyls could then take place through esterification. A synthesis scheme which addresses these issues is shown in Figure 3.1. - N H 2 - N H 2 -NH 2 —NHj - N H 2 1 OH bVo RT MeOH —NH —NH-R NH-R NH-R —NH2 2 o c--CH--CH--CH--CH-CH2-OH OH OH OH OH CH3COCI Triethylamine ) DCM —NH —NH-R' —NH-R' —NH-R' - N H - C -O C-CH-CH-CH-CH-CH2-OAc OAcOAcOAcOAc ^ = R ' -CH3 NaOMe DCM + MeOH -NH— -NH-R -NH-R -NH-R -NH-C-11 O C-CH-CH-CH-CH-CH 2-OH OH OH OH OH -CH3 X 0 limiting NaH THF —NH-— —NH-R' —NH-R" —NH-R" C-CH-CH-CH-CH-CHj—OH OH OH OH O^CHj^SQjNa C 7 ~ —NH-C—CH3 O CI-C-CH-CH3 CI limiting Triethylamine 1 DCM —NH -NH-R" -NH-R"' -NH-R"' —NH-C—CH3 6 O O C-CH-CH-CH-CH-CH 2 -0-C-CH-CH 3 OH OH OH 6-(CH2)-3S03Na CI ure 3.1: Reactions for incorporating charge and ATRP initiators onto the surface of Toyopearl. In this scheme, residual amine groups are protected (product 3) to prevent undesired side reactions. Hydroxyl group are then deprotected (4) to permit further derivatization. Unlike PSL, incorporation of sulfate groups using 1,3-propanesultone is optional and can be controlled by using sub-stoichiometric amounts of reagent. Furthermore, esterification of latent surface hydroxyl groups using 2-chloropropionyl chloride allows for a controlled incorporation of ATRP initiators on labile linkages. Characterization of the surface charge is permitted by conductometric titration of the sulfate groups in a similar fashion as was the case for PSL in Chapter 2. For initiator residues, a different situation exists. Hydrolysis of the ester leaves a surface immobilized hydroxyl group which cannot be quantified by conductometric titrations. However, the acidified hydrolysis product, 2-chloropropionoic acid, can be quantified by RPLC. The hydrolysis scheme is given in Figure 3.2. - R - O - C - C H - C I + NaOH C H 3 o CD Q_ O o "a CD 0) A T R P Initiator — R - O H O + N a O - C - C H - C I C H 3 Figure 3.2: Hydrolysis scheme for liberation of surface-immobilized ATRP initiators. Surface initiated ATRP can be performed by similar procedures to those given in Chapter 2. 3.2. Experimental Many of the experimental and analytical methods used in Chapter 2 have been used here. Changes and additional procedures are listed below. 3.2.1. Materials and Methods 3.2.1.1. Surface Functionalization Conversion of Amine to Hydroxyl Groups The highest conversions for this step were obtained when the seed matrix was fully dehydrated by washing and filtering several times each with water, ethanol, methanol, and acetone on a sintered glass funnel and placing in a vacuum oven at 50°C for 48 hours. A known amount (12.6 g) of the dried matrix was weighed and transferred into a 500 mL, 2-necked round bottom flask made up to be 5% w/v in methanol. The flask was fit with an argon purge port and an overhead stir paddle. 8-gluconolactone (1 g) was added to the stirred reaction flask (175 rpm) in a quantitative mole ratio to the amount of amine groups (0.690 mmol/g, predetermined from pH metric titration). The reaction was allowed to proceed for 16 hours at room temperature. After washing and filtering the product extensively in water, 0.1 M NaHCCh solution, and at least 5 more times with water until the pH was indistinguishable from water on pH indicator paper, the residual amine groups were quantified. Protection and Deprotection Reactions Remaining amine groups on the hydroxyl functionalized matrix were protected with acetyl chloride in the following manner. A known quantity of hydroxyl-derived matrix (12.6 g) was dried as above and charged into a 500 mL round bottom flask in D C M 5% w/v. The flask was placed in an ice-water bath and stirred via an overhead stirrer at 175 rpm. Triethylamine (7.69 mol, which was 1.1 equivalents to acetyl chloride) was added to the flask. The addition of acetyl chloride (5.42 mol) took place over the course of 1 hour by drop-wise addition of the reagent dissolved in 15-20 mL of D C M . A large excess of acetyl chloride, 5 times molar excess in proportion to the sum total of estimated hydroxyl and amine groups, was used to ensure complete protection of all surface active groups in light of the undefined nature of the starting matrix. Information from the manufacturer suggests that there were native hydroxyl groups already present from the matrix synthesis [152]. The pH of the reaction was followed with pH indicator paper and adjusted with triethylamine as needed to keep the pH > 8. After the addition, the contents of the flask were stirred for another 16 hours while ice melted and brought the temperature of the bath to room temperature. The product was filtered and washed with aqueous sodium bicarbonate solution, water and the solids and yield were calculated by assaying the remaining amine groups. Hydroxyl groups were deprotected by reacting with sodium methoxide in the following manner. The dried, protected starting material (12.6 g) was transferred to the round bottomed flask as above but the solvent used was a 1:1 mixture of D C M and methanol. While stirring, a 3 times molar excess in relation to the target OH groups to deprotect (0.090 mol) was added and the reaction proceeded for 16 hours at room temperature. The product was filtered, washed extensively with water, and analyzed for extent of deprotection and solids. The quantity of methyl acetate in the filtrate gave an estimate of the number of active hydroxyl groups present. Incorporation of Charged Groups Charged sulfonic acid groups were incorporated on to the hydroxyl-modified Toyopearl by reactions with either chlorosulfonic acid (99%) or 1,3-propanesultone (> 99%). The matrix (12.6 g) was first washed 2 times with methanol then with D C M and then transferred with D C M into a 3-necked round bottom flask which was placed in an ice bath and was stirred by an overhead stir paddle at 175 rpm. The following reaction is given as an example of a targeted 20% conversion of hydroxyl to sulfate. Triethylamine (4.4 mmol, which was 1.1 molar equivalents of the required amount of chlorosulfonic acid for the target conversion) was added into the flask and chlorosulfonic acid (4 mmol) was diluted with 5 - 1 0 mL of D C M in a dropping funnel. The contents of the dropping funnel were added over the course of an hour and the pH was monitored so that more triethylamine could be added if the pH became less than 8. The reaction was left to stir and warm to room temperature over the course of 16 hours. For the propanesultone pathway, the starting matrix was washed with methanol and acetone, filtered to dryness and dried for 24 hours in a vacuum oven at 50°C. The solid material was transferred to a 3-necked round bottom flask using anhydrous THF and made to a final concentration of about 5% (w/v). The contents were stirred at 175 rpm using an overhead stirrer at room temperature with argon purge. For the same amount of matrix, sodium hydride, 1 molar equivalent to the estimated total amount of hydroxyl groups present (0.020 mol), was carefully added and left to stir and react for 16 hours an argon atmosphere until the evolution of H2 gas ceased under. The quantity of sodium hydride used allowed for complete activation of all of the estimated hydroxyl groups on the surface. The 1,3-propanesultone (0.489 mmol) was dissolved in 15 mL of anhydrous THF and put in a dropping funnel and was added drop-wise over the course of an hour and the reaction was left to react at room temperature for 16 hours. The product was filtered, washed extensively with water, and analyzed for the incorporation of sulfate groups. Incorporation of ATRP initiators The incorporation of ATRP initiators follows the same procedure as the protection reaction except with different proportions of reagent. For example, in a targeted 75% conversion of remaining hydroxyl groups on 13.6 g of starting material, 1.62 g of triethylamine and 1.78 g of 2-chloropropionyl chloride were used. Synthesis of Matrix with Differing Sulfate or Initiator Concentrations Limiting amounts of reagents were used for the incorporation of charge or initiators. The amount of chlorosulfonic acid was varied to produce a desired level of modification. Reactions were carried out targeting from 0% to 30% conversion of hydroxyl groups on the surface. In a similar fashion, initiator groups were varied from 5% to 100% of the remaining surface hydroxyl groups. 3.2.1.2. Estimation of Surface Area by 1 2 5I-Labled Bovine Serum Albumin Binding  Studies Bovine serum albumin (BSA, 98% monomer) was dissolved in 1 mL of 0.1 M sodium phosphate buffer, pH 6.5, to a final concentration of about 6 g/L. The protein was labeled with 125Iodine (GE Healthcare) by using Iodo-beads (Pierce) according to the manufacturer's instructions. A 200 uL aliquot was added to an approximately 14 mL Sephadex G-25 (GE Healthcare) chromatography column which had been pre-equilibrated with buffer. Fractions of 600 uL were collected using a Gilson Micro Fractionator and aliquots were counted on a gamma counter (Cobra II auto-gamma, Perkin Elmer). The 2 or 3 fractions showing the highest counts were pooled and were determined to have 25270 CPM/uL above background. Aliquots of labeled protein were stored frozen at -20 °C until needed. The specific activity of the labeled protein was calculated from the assayed activity and the protein concentration as determined by UV/Vis spectroscopy using a standard curve for BSA previously determined. Spectra were recorded on a Helios Alpha UV/Vis spectrometer (Thermo Electron Corp) at 22 °C. Adsorption experiments were conducted in 0.01 M Tris-HCl buffer, pH 8.00±0.05, at 22 °C. A working stock solution of 1 2 5 I -BSA was prepared by thawing a frozen aliquot of radiolabeled protein, diluting with an unlabeled protein stock until the final protein concentration was about 0.25 g/L and the specific activity was about 1000 dpm/ug of protein. An aliquot of this solution was counted to determine the actual activity of this solution. To determine the protein concentration, samples of the working stock were compared against a calibration curve of absorbance at A = 280 nm versus concentration previously determined for BSA in buffer. To determine the fraction of bound to unbound label, a 50 uL aliquot of working stock protein was added to 950 uL aliquot of BSA carrier solution (5 g/L) and precipitated with 1 mL of 30% trichloroacetic acid solution. After centrifugation, 1 mL of the supernatant was transferred to a separate test tube and both tubes were counted to determine the free/bound ratio from the measured activities and volumes of the precipitate and the supernatant. For the determination of protein binding onto latex of known surface area, 200 uL aliquots of amine-functionalized latex microspheres (Interfacial Dynamics Corporation, supplied as an aqueous suspension of 22 g/L in ultrapure water, radius of 6.19±0.09 x 10"5 cm, 1.4% confidence interval, amine group density of 61.3 ueq/g) were added to predetermined amounts of protein stock solution and diluted to 600 uL in microcentrifuge tubes. The final protein concentration in solution was chosen to range from 0.002 g/L to 0.150 g/L. Samples were mixed and left to incubate at room temperature for 4 hours - a time period previously determined to provide a reasonable estimate of equilibrium binding. After incubation, the samples were centrifuged and the equilibrium protein concentrations were determined from the supernatant activity concentrations and specific activity. These counts were corrected for the proportion of labeled protein as determined earlier. Samples were washed at least 5 times with fresh buffer and centrifuge cycles until the counts in the wash solution were less than 1000 C P M above background. The supernatant was discarded and the bound counts associated with the pellet were determined. For binding studies of 125I-labled BSA onto Toyopearl AF-Amino-650M, conditions were changed from above as follows. The protein stock solution was modified to contain 30 g/L protein with a specific activity of about 14 CPM/ug. The Toyopearl matrix, washed more than 5 times with water and suction filtered on a sintered glass funnel, was adjusted to 25 g/L as determined by freeze-drying. For the adsorption experiment, 200 uL aliquots of well-mixed matrix slurry were added to volumes of working protein solution and diluted to 600 uL total volume. The dilutions for the protein solution were chosen so that the initial protein concentration ranged from 0.250 g/L to 16.4 g/L. The rest of the adsorption procedure was similar to latex case given above. 3.2.1.3. Surface-Initiated Atom Transfer Radical Polymerization Reactions were carried out as in Section 2.2.1.11. Stirring with a magnetic stir bar could not be used because of excessive damage to the matrix. Mixing reactions were performed by gently hand-swirling the mixture, automated mixing by an overhead stir paddle, or automated mixing by end-over-end tube inversion. Use of Brij-35 surfactant was found to be not necessary for these reactions. The following is an example of conditions typical for a large-scale reaction from which the chromatography medium for packing a 30 cm long, 1 cm i.d. column was prepared. An aqueous suspension of modified beads (7.5 g equivalent dry weight) with known solids content, initiator concentration (0.133 mmol/g solid) and sulfate concentration (0.184 mmol/g) was transferred to a 500 mL, round bottom flask and diluted with water to form a suspension of 3% by weight (250 g total weight). The flask was fit with a stopcock and subjected to 10 vacuum/argon cycles. During each vacuum cycle, the stopcock was opened and closed quickly to reduce the pressure in the flask. Dissolved gas was observed to bubble out when the slurry was swirled in the flask. These brief vacuum exposures were continued until the rate of bubbling was greatly reduced (about 10 times). After each vacuum cycle, the atmosphere in the flask was replenished with argon, UHP. The appropriate amounts of catalyst system were weighed separately into a test tube: Cu(I)Cl, Cu(II)Cl2, copper powder, and HMTETA. CuCl was set to 30 times the molar initiator content (3.1 g), CuCb was weighed in 15 mol percent ratio of CuCl (0.62 g), and copper powder (2.35) was 8 times the molar amount of Cu(II)Ci2. For HMTETA, a molar amount of equivalent to the sum of Cu(I) and Cu(II) (8.06 g) was used. D M A (38.11 g), weighed so that its final concentration in the reaction would be 1.6 M , was put in a separate vial and bubbled with argon for 15 minutes. The deoxygenated seed, monomer, and catalyst mixture were all transferred to an argon filled glove box. The monomer was added to the catalyst mixture and shaken until the soluble powders dissolved. When all of the catalyst had dissolved and returned to room temperature, the mixture was quickly added to the seed and the reaction mixture was mixed by overhead stirring at 350 rpm. After 24 hours, the reaction was taken out of the glove box and exposed to air. A small portion of the reaction supernatant was saved for analysis of D M A conversion by RPLC and solution polymer analysis by using SEC/MALLS. The product was filtered on a sintered glass funnel and washed extensively with water, 50 mM EDTA (at least 10 minutes contact time), water, 50 mM NaHS03 (at least 10 minutes contact time), and final water washes, at least 5 times, to remove all traces of NaHSCh. Samples were saved for IR analysis, graft analysis, and solids. During the various washing stages, it was possible to settle out insoluble copper powder while decanting the matrix. Repeated many times, it was possible to recover the majority of copper. During the final wash the product was centrifuged and any product contaminated with copper in the bottom 1 or 2 mL of the centrifuge tube was sacrificed. 3.2.1.4. Grafting Reactions with Varying Monomer Concentration and Catalyst  Concentration Small scale test reactions for investigating ATRP reaction variables were typically performed on 0.1 g (equivalent dry weight) samples of seed slurry adjusted to 3% w/w. Reactions were performed in 15 mL plastic centrifuge tubes which were mixed by end-over-end mixing on a test tube mixer (30 rpm on a Roto-Rack instrument). To investigate the effect of varying the monomer concentration, the molar concentration of D M A in solution was varied from 0.12 M to 7.8 M at constant Cu(I), Cu(II), and Cu(0) proportions. For experiments with variations in Cu(I), the solution concentration was varied from 0.03 to 1.2 M . When the CuCl concentration was changed, CuCb, copper powder, and HMTETA were varied in proportion to Cu(I) as described above. Experiments with Cu(II) concentration were conducted for CuCL; concentrations varying from 0.15 to 2 times the molar amount of Cu(I). When the CuCh concentration was changed, HMTETA was varied proportionally but the amount of copper powder was not increased. This was done so that the true effect of varying the divalent copper could be investigated while minimizing the possibility that copper powder could interfere by changing the concentration of copper species in the redox equilibrium. 3.2.1.5. Characterization of the Modified Surface and PDMA Grafted Matrices pH Metric Titrations Titrations were used to assay the starting amine content, the conversion of amine to hydroxyl, and the protection of amine functions. The concentration of surface amine groups was assayed according to literature supplied by the manufacturer [153] with the following changes. For each trial, 20 mg of solid material was assayed. Instead of using an auto-titrator, the pH was monitored by a placing a pH probe (Accumet), which was previously calibrated at pH 4, 7, and 10, into an incubated (25 °C) 50 mL reactor with the analyte suspended with 10 mL of 0.1 M NaCl solution under constant stirring via a magnetic stir bar. The titrant, 0.01 M hydrochloric acid solution, was added at a known delivery rate of 0.0408 mL/min via a syringe pump. The equivalence point was recorded as the volume of HCI required to titrate the slurry to pH 4.5. The HCI was standardized by sodium carbonate solution with bromocresol green indicator and boiling the solution and retitrating the endpoint to correct for dissolved CO2. Determinations were done in triplicate. The concentration of the amine group i[NH2], in umol/g of solid matrix) could then be calculated from the following equation ^ . ^ J x C x l O O O p ] ] where A was the time in minutes to reach pH 4.5, B delivery constant of the syringe pump (in mL/min), C was the concentration of HCI (in mol/L), and D was the mass of matrix (in g). Evaluation of Deprotection Reactions To evaluate the deprotection of acetyl chloride-protected hydroxyl groups by sodium methoxide, evolved methyl acetate in the filtrate (known volume) was assayed by UV/Vis spectroscopy using a standard curve and evaluating the absorbance of methyl acetate in 50:50 DCM/Methanol at X = 226 nm. Quantification of Surface Charge Conductometric titrations were done on a YSI model 35 conductance meter and 3403 cell with platinum electrode at 25°C monitored while titrant (0.01 M NaOH) was delivered at a flow rate of 0.0102 mL/min delivered by a syringe pump (Harvard Instruments). Prior to freeze drying, the matrix was conditioned by washing/filtration cycles on a sintered glass filter. After washing with water, 0.1 M HC1 was added and left to equilibrate with the matrix for 10 minutes. After filtering, the matrix was washed/filtered at least 5 more times with water after which time the filtrate was checked on pH indicator paper for a pH not significantly different from water. In a typical determination, 20 mg of solid was suspended in 10 mL of deionized water and placed in a reactor in a water bath. The conductance was measured as 0.01 N NaOH (standardized using potassium hydrogen phthalate) was added via the syringe pump. The end point was clearly visible as an increase in slope after the acidic residues had been neutralized. Quantification of Hydrolyzed Initiators To quantify the amount of surface immobilized initiators, 0.1 g of freeze-dried matrix was hydrolyzed by addition of 1 mL of 0.67 M NaOH and gently agitated on a shaking platform mixer for 36 hours. After centrifuging, a 200 uL extract was neutralized with an equivalent amount of 0.67 M HC1. The sample was diluted with mobile phase (0.02 M HC1) as necessary and determinations of 2-chloropropionoic acid were made by RPLC by analyzing peak areas of 20 uL injections against a calibration curve made by injections of solutions of known concentrations for (R)-(+)-2-chloropropionoic acid (99%). Reverse-phase liquid chromatographic analysis was done on a Hitachi model L-6210 HPLC with an L-4200 UV/Vis detector and a Lichrospher 60 RP-select B reverse phase column from Merck. Analysis of 2-chloropropionoic acid was carried out in 0.02 M HC1 mobile phase with a flow rate of 2 ml/min at 22°C and X = 210 nm. Evaluation by ATR-FTIR Matrix modifications were evaluated by ATR-FTIR. Spectra were collected using about 5 mg of freeze-dried powder. After collection, the spectra were adjusted to a common baseline and displayed using a common scale. Identification of peaks and their absorbance magnitude therefore gave absolute differences from the baseline, which, when compared to other peaks in the spectrum, gave ratios which can be interpreted to be semi-quantitative measures of relative abundance. Surface ATR-FTIR spectra were collected on a Nexus 670 FT-IR ESP (Nicolet Instrument Corp.) with a MCT/A liquid nitrogen cooled detector, KBr beamsplitter, and an M k l l Golden Gate Single Reflection attenuated total reflectance accessory (Specac Inc.). The sample stage contained a diamond window and a sapphire anvil on a torque limiting screw set to deliver 80 lbs pressure. Characterization of the Grafted Matrix The grafted matrix was characterized by analyzing the D M A conversion and polymer released by hydrolysis similar to procedures in Chapter 2, Section 2.2.1.12. A known amount of matrix (approximately 0.15 g solids in 1 mL of suspension) was placed in a 15 mL test tube and 0.67 uL of 10 N NaOH was added (final concentration ~ 0.7 M). The suspension was agitated on a shaking platform mixer until no change in the amount released was detected (3 weeks). The supernatant and washings were collected and analyzed by SEC/MALLS. The mass of polymer in solution following cleavage was determined using a calibrated RI detector. Light Microscopy A Zeiss Axioscope 2 Plus microscope and an attached Retiga camera were used for microscopic characterization of the modified matrix. Quantitative measurements were obtained by imaging a measured grid and calibrating for distance. Image analysis was performed using Northern Eclipse software (Empix Imaging). 3.2.1.6. Catalyst Solubility and Binding Study To investigate the affinity of the modified matrix for binding complexed copper salts, test reactions containing copper (II) complexed to HMTETA were incubated with modified matrix. CuCL; (0.2 g) was added to 1 mol equivalent of HMTETA and dissolved with methanol and mixed for 2 hours until completely dissolved. The Cu(II)-ligand complex was rotary evaporated to dryness and evacuated overnight on a high-vacuum line. The residue was reconstituted in 100 mL of water and appropriately diluted to obtain a standard curve on the LlV/Vis spectrometer at X = 340 nm. Identical aliquots (60 uL) of the copper complex solution were then added to aqueous solutions of D M A ranging from 0 to 7 M (1 mL total volume) and the amount of copper complex apparent in solution was quantified. The wavelength was chosen such that the highest concentrated D M A solutions in this experiment would not interfere with the analyte spectrum (highest concentrated D M A solution absorbance at X = 340 nm was 1.1% of the absorbance of the copper complex). To investigate ligand binding to the initiator surface, known amounts of modified matrix (approximately 6 mg) were incubated in solutions containing no D M A and the resulting decrease in copper complex was quantified. 3.2.1.7. Kinetics of Hydrolysis The kinetics of graft hydrolysis was investigated to determine the time required for maximum cleavage of chains from the support. Grafted samples were hydrolyzed in 0.7 M NaOH from a period of hours to months for one lot of grafted material divided into 12 fractions. Hydrolysis was stopped by HCI neutralization followed by supernatant extraction and water washing the seed two times. Time points were taken at 3h, 8h, Id, 3d, and then weekly for 8 weeks. 3.3. Results and Discussion Toyopearl AF-Amino-650M is composed of a proprietary polymer base matrix of glycidyl methacrylate and ethyleneglycoldimethacrylate coupled to an amino group via a short spacer arm [152]. Literature from the manufacturer suggests the chemical composition of the amine-functionalized bead is hydrophilic containing an unspecified quantity of hydroxyl groups in the base polymer and at the amine ligand attachment site. The resin itself is listed as having a 1000 A nominal pore diameter and a nominal 65 um diameter. This versatile matrix has the physical properties of being mechanically strong, resistant to harsh chemical conditions and with its "giga-porous" matrix, provides high flow rates through the porous network. The wide pore diameter produces no size dependent separation of solutes up to 2 x 106 Da in packed beds, as evaluated by size exclusion chromatography (results not shown). 3.3.1. Estimation of the matrix surface area The specific surface adsorption of radio-labeled BSA on latex spheres of known surface area was first investigated. The amine-functionalized latex bead was assumed to be chemically equivalent to that of the starting Toyopearl matrix. Figure 3.3A shows the adsorption behavior of the 125I-labeled probe onto the surface of the latex of known size. The maximum binding to the latex surface was extrapolated from a non-linear least squares fit of the data to an isotherm of the Langmuir type represented by nKL l + KL [3.2] where L is the solution concentration of the protein in equilibrium with the surface, v is the average amount of protein bound at equilibrium per unit area, and Ka is effective association constant. The equation was solved for n, the maximum amount of protein bound at equilibrium per unit area at saturation. Using the known value of the surface area of the latex standard, the maximum packing density of adsorbed protein was calculated to be 1.89±0.07 mg protein/m2. This appears to be a reasonable estimate which is in the range of results for monolayer BSA side-on binding with results found by other researchers [154;155]. The same radio-labeled probe was adsorbed to the Toyopearl matrix under the same binding conditions. A similar non-linear fit to equation 3.2 was performed using the data presented in Figure 3.3B. Assuming the same BSA packing density at saturation, the specific surface area accessible to macromolecules of this size was estimated to be 121±8 m2/g. A Figure 3.3 A and B : Binding isotherms for 1251-Labeled BSA onto A) an amine latex of known surface area (•) and B) Toyopearl ( A ) which had a chemically similar surface. The saturation value for bound protein per gram of latex was extrapolated from equation 3.2 and calculated to be 8.7±0.4 mg/g and 230±8 mg/g respectively. 3.3.2. Surface Modification of Toyopearl Matrix with Charged Sulfate Groups and ATRP Initiators Earlier work suggested that the presence of charged sulfate groups at the interface creates a unique catalyst-rich volume in the aqueous medium due to electrostatic attraction between the surface negative charges and the positively charged catalyst complexes [21]. The increased concentration facilitates the surface initiation and polymerization under aqueous ATRP conditions [22]. Previous work in the Brooks lab has resulted in a model of the surface which has been used to explain the apparent restricted diffusion of solution D M A at the interface, accumulation of catalyst and helps to explain the experimental observations [21]. The goal of the first step was to functionalize the surface with negatively charged sulfate groups and ATRP initiator groups. This also created cleavable ester groups between the grafted polymer chains and surface. Hydroxyl groups provided a convenient avenue for incorporating charges via chlorosulfonic acid or 1,3-propanesultone and adding ATRP initiators through an ester linkage by reaction with 2-chloropropionyl chloride. By taking advantage of the ring opening reaction of 8-gluconolactone [156] with primary amines on the Toyopearl matrix surface the amount of hydroxyl groups on the surface was multiplied 5-fold (product 2 in Figure 3.1). This allowed for much versatility for tuning the surface concentration of initiators or sulfate groups in subsequent reactions against a background rich in hydroxyl groups. Table 3.1 shows the quantification of various analytical parameters chronicling the surface modification of one lot (7.5 g solid) of base matrix (results given are typical for smaller-scale reactions). Table 3.1: Summary of analytical results corresponding to incorporation of sulfate charge and ATRP initiators onto Toyopearl matrix. Product* Analyte Analytical Method Assayed Conclusions amount mmol/g 1 Amine pH-metric titration 0.690 2 Amine pH-metric titration 0.354 51.4% conversion of amine groups 3 Amine pH-metric titration 0.045 6.6% of amine groups remained unprotected 4 Methyl Acetate (MeAc) Concentration by UV/Vis 1.593 Equals the concentration of active hydroxyl groups (1 mol of MeAc produced for every mol of hydroxyl) 5 Sulfonic acid Conductometric titration 0.184 11.5% of hydroxyl groups converted to charged groups. The target was 15% giving a reaction efficiency of 77% 6 2-chloropropionoic acid R P L C 0.133 9.5% of remaining hydroxyl groups converted to A T R P initiators. The target was 73% giving a reaction efficiency of 13%. Product numbers correspond to numbered species in Figure 3.1. Information from the manufacturer suggests that on the base matrix (1) there is at least one hydroxyl group present on the linker for every amine group. Thus the final hydroxyl concentration in the post-lactone reaction product (2) was estimated to be at least 5 moles of hydroxyl for every mole of amine reacted plus one mole of hydroxyl for every mole of amine detected in the initial assay. The average spacing between hydroxyl groups, if one assumes that each occupies a site on a square lattice, was estimated to be 2.9 A. Protection of the residual amine groups (3) was facilitated by reacting with acetyl chloride. Protection of residual amine groups left after the gluconolactone reaction was crucial for several reasons. First, unprotected amines would have produced amide bonds on the initiator functionalized surface and would result in polymer grafts which could not be cleaved. This would make absolute characterization of the grafted polymer chains impossible. Second, for material intended for chromatographic studies, one would want to reduce or eliminate the possibility of enthalpic effects due to protein interaction with underivatized amine groups on the base matrix. Last, one should strive to reduce the possibility of amine groups participating in electron transfer during ATRP. The final degree of protection of amine groups was 93.4% (Table 3.1) and found to be adequate for future steps. Upon reaction with sodium methoxide, each deprotection reaction regenerated hydroxyl groups and produced an equivalent amount of methyl acetate, leaving the amide bond intact (4). By assaying the amount of methyl acetate produced, the quantity of reactive active hydroxyl groups present on the surface was estimated. Following deprotection, the surface was modified through the incorporation of sulfate groups. Two methods were experimented with: the use of chlorosulfonic acid or the use of 1,3-propanesultone. Initial experiments were done via the chlorosulfonic acid route; however, the propanesultone reaction was used for all later lots because the reaction efficiencies were higher, conversions were more consistent, and greater control over the charge modification was observed. The lower consistency of the chlorosulfonic acid reactions can be attributed to the relative instability of the reagent during storage. There was no chemical difference between lots produced by either method either in analytical determination of the sulfonic acid residues by conductometric titration or during subsequent reactions in the surface modifications and polymer grafting. After surface charge incorporation (5), which was an optional step for some materials produced, a fraction of remaining hydroxyl groups was esterified with 2-chloropropionyl chloride to provide the ATRP initiator surface (6). When using limiting quantities of 2-chlorpropionoyl chloride, the incorporation of ATRP initiators via esterification had low reaction efficiency, perhaps due to small amounts of water remaining on the base material. When excess reagent was used, a very high degree of derivatization was measured, representing complete conversion of hydroxyl groups as estimated by mass balance. The quantity of initiators was measured by cleavage of initiator groups by saponification for 1 week of 0.1 g of modified matrix in 1 mL of 0.67 M NaOH. A portion of the extract was neutralized with an equal quantity of 0.67 M HC1, filtered, and analyzed by RPLC against a standard curve of 2-chloropropionoic acid. The surface modifications were also followed by surface attenuated FTIR. Surface analysis techniques provide information about the surface modifications while reducing background interference from signal originating from the polymer base matrix [157]. Figure 3.4 shows representative FTIR absorbance spectra from 4000 to 1500 cm"1 for the base Toyopearl matrix (1). 4000 Peak 1 Peak 2 3500 3000 2500 2000 Wavenumbers (cm-1) 1500 Figure 3.4: ATR-FTIR absorbance spectra highlighting characteristic peaks for surface modifications made to Toyopearl. Spectrum of the base matrix (1): Peak 1, O-H stretch (3400 cm"1); Peak 2, C=0 (ester) stretch (1721 cm"1). In this region it is possible to identify peaks on the base matrix for an O-H stretching vibration at 3400 cm"1 (Peak 1), various C-H stretching vibrations from 2800 to 3000 cm"1 representing alky groups, and a C=0 (ester) stretching vibration at 1721 cm"1 (Peak 2). During surface modifications, changes to the intensity ratio of the C=0 stretch and the OH and were quantified and are listed in Table 3.2. For example, reaction of (1) with gluconolactone produced many OH functions on (2), increasing the absorbance of Peak 1 relative to Peak 2, which remained unchanged. Table 3.2: ATR-FTIR peak assignments. Product Peak Assignment Frequency of Vibration (cm"1) Absorbance Intensity Absorbance Ratio * 1 OH 3400 0.050 0.18 C=0 (ester) 1721 0.279 2 OH 3391 0.057 0.21 C=0 (ester) 1721 0.266 3 OH 3400 0.043 0.19 C=0 (ester) 1723 0.229 4 OH 3368 0.033 0.23 C=0 (ester) 1723 0.146 5 OH 3400 0.037 0.19 C=0 (ester) 1723 0.194 6 OH 3400 0.040 0.18 Low [initiator] C=0 (ester) 1723 0.217 6 OH 3400 0.026 0.09 High [initiator] C=0 (ester) 1723 0.298 | refers to numbered products in Figure 3.1 * refers to the absorbance ratio of C=0 to OH Examples of some of these modifications are shown in Figure 3.5A showing decreases in Peak 1 absorbance as hydroxyl groups were converted. For initiator incorporation, the FTIR spectrum of (6) showed both reduction of Peak 1 and an increase in Peak 2 as shown in Figure 3.5 A and B. Initiator Modified (6)— A Figure 3.5 A and B : A) Overlay and enlargement for Peak 1 showing absorbance increase for (2) compared to the base matrix (1) and decreases for (5) and (6). B) Overlay and enlargement for Peak 2 showing absorbance increase for (6) compared to (1). It was difficult to observe other peaks related to initiator incorporation such as C-Cl vibrations due to the strong background interference in the low wavenumber region of the spectrum from the base matrix. By light microscopy, there was little observable difference between the physical appearance and dimensions of initiator modified matrix. Figure 3.6 shows little physical difference between surface B and the starting material, A . Surface B was synthesized by carrying out at least 5 synthetic steps involving complete dehydration and many hours mixing in stirred reactors and shows a small amount of degradation. The increased amount of broken beads visible on Surface C was due to a supplementary initiator synthesis stage where B was used as the starting material. This extra stage was not normally carried out in the regular synthesis procedure because it is possible to synthesize high initiator content material directly. Unusual reaction conditions and bead damage in this case are noted. Figure 3.6: Light microscope images of initiator modified matrix taken at 400 X magnification. A) Toyopearl Base Matrix 66±2 nm; B) Low [initiator] matrix 64±3 nm; C) High [initiator] matrix 67±8 p.m. 3.3.3. Aqueous ATRP Grafting of PDMA The initial work investigating aqueous ATRP grafting methods for polymer growth from surfaces was developed on functionalized latex particles [21]. Compared to latex particles, the physical properties of Toyopearl differ greatly, having a much higher specific surface area and hydrophilicity. Since Toyopearl does not contain any native negatively charged groups (verified by conductometric titration) a synthetic procedure to incorporate a desired quantity of sulfate residues to enable a comparison between polymer grafting from initiator-functionalized surfaces with and without negative charge was developed. ATRP reaction conditions from published work was used to guide the initial choice of experimental conditions (copper (O/I/II) catalyst system and the ligand, HMTETA) [21;22]. The concentration of charged groups, initiator groups, the monomer concentration, and the concentrations of Cu(I) and Cu(II) in the catalyst system were varied and the resultant graft and solution polymer were analyzed. Polymerizations were completed in 24 hours. The overall reaction from Toyopearl surface modification and polymer grafting is conceptually pictured in Figure 3.7. Polymer grafts were cleaved from the surface by saponification and the molecular weight and the amount of grafted material were determined by quantitative SEC analysis. NaH /1,3-propane sultone THF 2-Chloropropionoyl chloride / Et^ J DCM D-Gluconolactone Modified Surface (amine protected) Porous Matrix Pore Volume - 0 1 - 0 1 - o x - O I - O I - O I - O I - O I - O I - O I - O I - O I - O I I I I I I I \ o o o o o o o 1 1 1 1 1 1 I I I I I I I - 4->:' 0 O O O O O o • I I I I I I / j 0 l O I O I — O I — O I O I O I HMTETA/CuCl/CuCl,/DMA . > 3 ) — 2- > r- Water/22 C 0 = I- V,' I = — C = 0 H3C -C—H CI <D - 0 1 — H — H O I H <D — H - O I I I I 1 1 1 \ 0 H O H f l l H H 1 I w H O H H H M O I — O I — H <D — O I ( PDMA Grafted Surface Figure 3.7: Schematic representation of surface modifications and grafting of PDMA from a macroporous chromatography matrix. PDMA grafted matrices were analyzed by ATR-FTIR. The C=0 (amide) stretching vibration was clearly visible at 1632 cm"1 (Figure 3.8) as a separate adsorption from the C=0 (ester) vibration at 1725 cm"1 on the base material (cf. Figure 3.4). The much decreased OH and C=0 (ester) peaks are a result of a much higher abundance of PDMA at the ATR interface. Diminished peak intensities for OH and C=0 (ester) vibrations are a result of the matrix surface being buried under PDMA and not due to a decrease in their surface concentrations. 4000 3500 3000 2500 2000 Wavenumbers (cm-1) 1500 Figure 3.8: Spectrum of PDMA grafted matrix. The spectrum contains a new peak at 1632 cm"' from C=0 (amide) stretching vibrations from the PDMA grafts. The relative decreases in peak 1 and peak 2 are due to the much greater abundance of polymer graft and the decreased ability for ATR-FTIR to probe the surface of the material below the polymer graft. 3.3.4. Method Validation for Quantitative Release of Surface Grafted Polymer Chains After grafting, a large batch of functionalized matrix that originally weighed 7.5 g had a final grafted weight of 14.4 g. The mass increase, 0.92 g/g relative to the starting material, was compared to the hydrolyzed amount over the 8 week timed hydrolysis. By analytical SEC/MALLS, the graph of mass of chains released at time periods after 8 hours had appeared to reach a plateau value. Figure 3.9 shows a graph of the mass of polymer chains released as a function of hydrolysis time. The region of constant mass was rapidly reached. 0.10- • o> 0.08-OJ E TO C/) 0.06-c to c | 0.02-< -200 0 200 400 600 800 1000 1200 1400 Hydrolysis Time (hours) Figure 3.9: Amount of PDMA chains released in hydrolyzed samples from 3 hours to 8 weeks. The mass of polymer in solution following complete cleavage (8 weeks) agreed with the value estimated by mass balance to 96%. At all hydrolysis time points, the M n , MJM„, and (R) values remained essentially constant, indicating that the ester hydrolysis reaction was independent of grafted chain length and position within the porous matrix. For routine analysis by this method, it was more convenient to perform hydrolysis for 3 weeks. The graph shows that 3 weeks of saponification is sufficient to release 98±2% of the total chains and so the 3 week time period provides a reasonable estimate of the total number of chains present. o c 3 0.00 Over the same 8 week time period, control samples were analyzed to determine the stability of the starting matrix, lactone-modified matrix, and for potential interference from chloropropionoate ion released from hydrolyzed, initiator-modified matrix. The bare Toyopearl and lactone-modified matrix showed insignificant peaks when analyzed revealing no observable degradation. Samples containing chloropropionoate ion showed some baseline interference but this was an insignificant contaminant in routine analysis of PDMA from hydrolyzed matrices. 3.3.5. Effect of Changing the Monomer Concentration The concentration of D M A was varied from 0.8 M to 7.8 M in this set of experiments to understand its effect on the surface polymerization. Three different matrixes with different initiator concentrations on the surface were used. The grafted PDMA chains were cleaved from the surface and analyzed. Results are given in Table 3.3 and Figure 3.10. Table 3.3: Effect of surface charge and initiator concentration on fate of monomer and physical properties of the grafted surface. Series [Charge] [Initiator] [DMA] Converted Monomer Solution polym. Graft polym. Solution Polymer M„ Solution M „ / M „ Graft Polymer M„ Graft M„l M„ Initiation Efficiency Ograft Dl(R,) mol/g x 10* mol/g x 10 4 M % (w/w) % o f Conv. % of Conv. • a Da % cha in /nm 2 A 1.4±0.2 0.13 0.8 4.2 54.3 45.7 6100 1.02 13500 1.20 28.0 0.0179 1.8 2.3 3.7 79.5 20.5 19900 1.23 20300 1.14 21.3 0.0136 1.6 3.9 5.4 78.1 21.9 41300 1.17 80300 1.09 14.5 0.0092 0.9 5.4 7.2 87.9 12.1 44600 1.18 52900 1.21 22.0 0.0141 0.9 7.8 7.2 89.0 11.0 64900 1.21 92300 1.21 16.8 0.0108 0.7 B 1.8±0.3 1.3 0.3 19.5 35.3 64.7 5600 1.28 5200 1.21 18.0 0.119 1.3 1.6 12.3 42.8 57.2 17100 1.31 16000 1.23 16.1 0.107 0.7 3.1 6.0 64.7 35.3 18900 2.97 19300 1.28 9.6 0.0637 0.8 C 2.1±0.3 28 0.8 34.7 19.6 80.4 20400 1.19 23700 1.24 1.5 0.205 0.4 2.3 30.7 37.9 62.1 62400 1.21 74000 1.25 0.7 0.0968 0.3 3.9 19.6 27.3 72.7 112200 1.22 121300 1.28 0.5 0.0740 0.2 5.4 13.3 31.6 68.4 155200 1.26 162100 1.27 0.3 0.0476 0.3 7.8 10.6 28.8 71.2 223200 1.25 225200 1.26 0.3 0.0412 0.2 Experimental Conditions: A) 2.0 g of 3 %(w/w) suspension of functionalized Toyopearl; B) 14.1 g of 3 %(w/w) suspension of functionalized Toyopearl; C) 3.3 g of 3 %(w/w) suspension of functionalized Toyopearl. For all: Cu(I)Cl = 0.14±0.01 M; Cu(II)Cl2 = 0.15 times molar amount of Cu(I)Cl; HMTETA = sum of molar amounts for Cu(I) and Cu(II). 1.50 n 1.25 1.00 250000 - i 200000• TO 150000' (C 100000 i — 50000 —I 1 [ -4 6 [DMA] (M) Figure 3.10: The effect of monomer concentration on P D M A molecular weight and polydispersity for polymers cleaved from two different matrices. Seed A: high initiator concentration, 2.8 mmol/g ( A ) ; seed C: low initiator concentration, 0.013 mmol/g (•). Seed A was a 3.33 g suspension at 3% solids with [CuCl] = 0.13 M . Seed C was a 2 g suspension at 3% solids with [CuCl] = 0.16 M . For both reactions, [Cu(II)/Cu(I)] = 0.15; Cu(0) molar amount was 8 times Cu(I); H M T E T A molar amount was equal to the sum of Cu(I) and Cu(II). The molecular weight of the grafted polymer increases almost linearly with the monomer concentration (Figure 3.10 and Table 3.3) for low initiator and high initiator surface. By changing the monomer concentration the molecular weight of the grafts could be changed from 5 kDa to 225 kDa. The polydispersity of the grafted polymer chains (given in Figure 3.10 and Table 3.3) shows that surface polymerization was well-controlled (less than 1.3 in every case). In all polymerization experiments solution polymer was formed along with graft polymer. Molecular weight properties and amount of solution polymer were measured and is given in Table 3.3. Molecular weight of the solution polymer is slightly lower or similar to that of the graft polymer and indicates that chain transfer reactions occur early on the time scale of the ATRP [22;102]. Conversion of monomer ranged from 4.2% to 34.7% with the highest conversions for the matrix with the high initiator content. The fate of converted monomer (solution or graft polymerization) was also affected by the initiator content. Graft polymerization was favored over solution polymerization when the initiator concentration was high. Despite the high proportion of graft polymerization, series C in particular showed low initiation efficiencies (below 2%). A discussion of this result is resumed in a later section concerning effects of varying the amount of surface charge. As for the graft density, Figure 3.11 shows that the graft density decreases with increasing monomer concentration. The effect is greater for surfaces with high initiator concentration. 0.22-, 0.20-0.18-0.16-0.14-c 0.12-~£ hai 0.10-0.08-0.06-0.04-0.02-0.00-• [DMA] (M) Figure 3.11: The effect of monomer concentration on PDMA graft density for two different seeds. Seed A: low initiator concentration, 0.013 mmol/g (•); seed C: high initiator concentration, 2.8 mmol/g ( A ) . The experimental details are the same as in Figure 3.10. To investigate possible reasons why graft density should decrease with increasing monomer concentration, the solubility of the copper/ligand complex in aqueous solutions of D M A was studied. Figure 3.12 shows that the concentration of the selected test compound, a water soluble copper (II)/HMTETA complex, decreases as the molar concentration of D M A in aqueous solution is increased. It is likely that insoluble portions of complex were spun down during centrifugation leaving decreased amounts of soluble complex in the supernatant. Thus, one might predict that the amount of soluble copper (I)/HMTETA complex in solution available for initiation would decrease at higher monomer concentrations. This can also be seen in Table 3.3 by the decreasing trend in initiation efficiency as [DMA] is increased. 0 . 0 0 1 4 • 0 . 0 0 1 2 "a c TO 0 . 0 0 1 0 -0 . 0 0 0 8 -Q 0 . 0 0 0 6 -E 0 . 0 0 0 4 ' '5 cr W 0 . 0 0 0 2 0 . 0 0 0 0 i 1 1 1 r -1 0 1 i — • — r 2 3 4 5 [DMA] (M) Figure 3.12: The solubility of the copper(II)/HMTETA complex in solutions of increasing D M A concentration (•). The complex consisted of a 1:1 molar mixture of CuCl 2 and H M T E T A (1.5 mmol) complexed in methanol, dried, and reconstituted in water (0.02 M). Approximately 6 mg of initiator seed was added to solutions with no D M A present. Amounts were chosen so that the solution concentration of copper complex was in a slight molar excess to the concentration of sulfate on the matrix. The resulting decrease in complex left in solution is shown for two seeds with approximately the same sulfate concentration (0.2 mmol/g): high initiator seed, 2.77 mmol/g (o); and low initiator seed, 0.133 mmol/g (0). When incubated in these test solutions with no D M A added, samples of the initiator-modified surface containing similar charge concentrations showed different affinities for the copper/ligand complexes. The calculated bound amounts for high initiator and low initiator content respectively were (1.4±0.4) x 10"4 and (0.74±0.02) x 10"4 mol/g. The former material contained an initiator surface concentration 20 times higher than the latter. This observation reveals that there is a weak association of the copper-ligand complex with the surface that is influenced by the initiator concentration. Implications of this association are discussed along with other initiator effects below. The predicted morphology of the grafted layer was determined to be brush-like as calculated by the ratio of distance between the grafts (D) and the radius of gyration. This ratio decreased, implying morphologies were more brush-like as the monomer concentration was increased (Table 3.3). Partly due to the general trend to lower values for D, low values of -D/(i?g) were mostly fueled by the increasing ^Rg) as the graft molecular weight increased. Size changes in the grafted matrix Increasing the concentration of monomer in the ATRP reaction mixtures resulted in large physical changes observed in settled beds heights of the grafted material as shown in Figure 3.13. When imaged by light microscopy, an increase in diameter of the grafted material was seen (Figure 3.14). The large difference in graft density between medium and high initiator concentrations as seen in Table 3.3 is exemplified by large changes in settled bed height as shown in Figure 3.15. B C D Figure 3.13: Settled bed height of PDMA grafted materials from Series C (Table 3.3). All reactions started with the same amount of initiator functionalized matrix (0.1 g) which was similar in height to tube B. Concentrations of DMA used in grafting were: B) 0.7 M; C) 2.3 M; D) 3.9 M; E) 5.4 M; F) 7.8 M. 250 |jm scale +^+++^  Figure 3.14: Light microscope images of the grafted matrices shown in Figure 3.13. Images taken at 100 X magnification. A) Initiator (ungrafted) matrix; B-F) grafts from various monomer concentrations. B) [DMA] = 0.7 M, diameter = 71±4 urn; C) [DMA] = 2.3 M, diameter = 133±21 um; D) [DMA] = 3.9 M, diameter = 180±20 urn; E) [DMA] = 5.4 M, diameter = 220±20 um; F) [DMA] = 7.8 M, diameter = 305±22 um. 12 S~ E F G H Figure 3.15: Settled bed heights of medium and high density grafted matrices. E: crgraf = 0.048 chain/nm2, M„ = 162 kDa; F: = 0.041 chain/nm2, M„ = 225 kDa; G: a%raf = 0.11 chain/nm2, M„ = 298 kDa; H: O J ^ , = 0.065, M„ = ill kDa. E and F had [initiator] = 28 x 10~4 mol/g; G and H had [initiator] = 1.3 x 10'4 mol/g. All material was grafted with the same ATRP reaction conditions and the same amount of starting matrix (-0.1 g). E and G used [DMA] = 5.4 M; F and H used [DMA] = 7.8 M. From microscopy and SEC/MALLS, the particle size diameter (and M„) was determined to be E) 220±20 urn (162 kDa); F) 305±22 u.m (225 kDa); G) 166±17 urn (298 kDa); H) 330±30 urn (377 kDa). 3.3.6. Effect of Changing the Surface Charge and Initiator Concentration Previous studies investigating the ATRP of PDMA from PSL in aqueous solution showed a strong influence of static negative surface charge producing a concentrating effect on positively charged copper catalysts near the surface [21;22]. Previous studies utilized negatively charged polystyrene latices prepared by polymerization of styrene initiated by potassium persulfate. Endemic sulfate groups were omnipresent on the surface and are a fundamental and necessary requirement for synthesis of stable surfactant-free latex. The use of a macroporous base matrix for chromatography studies relieved the requirement for charged groups native to the surface but, with the proper choice of synthesis conditions, it was possible to synthesize sulfate groups in the desired quantities. By this means a library of matrices containing from zero to relatively high surface concentrations of sulfate groups was produced. Along with variations in the surface initiator concentration, the influence of these two manipulated variables on the physical properties of the grafted polymer such as M„, MJM„, conversion etc. was studied over a limited range of surface charge concentration. Figure 3.16 shows the dependence of the polymer graft density on the surface initiator concentration at the surface sulfate concentrations near 1.8 x 10"4 mol/g and monomer concentrations near 1.0 M (see legend to Figure 3.16 and Table 3.3 for values). Also shown are data for a system where no sulfate was added to the surface. • - i — i — i i i 111 1 1 — i — i i i i 11 1 1 — i — i i i 111 1 1—i—I—n 1E-5 1E-4 1E-3 [Initiator] (mol/g) Figure 3.16: The effect of surface initiator concentration on measured graft density of PDMA chains cleaved from grafted, modified Toyopearl surfaces with negative surface charge (filled symbols) and without negative surface charges (empty symbols). Reaction conditions were as follows: For (•), [DMA] 1.2±0.4 M and [charge] = 0.18±0.04 mmol/g. For (•) [DMA] = 1.1±0.1 M and no detectable surface charge. For both, [CuCl] = 0.15±0.05 M , [Cu(II)/Cu(I)] = 0.15, Cu(0) molar amount 8 times that of Cu(II), and H M T E T A molar amount equal to the sum of Cu(I) and Cu(II). It is clear that the graft density increases with increasing surface initiator concentration, and that the presence of negative charges strongly increased the graft density. The strong dependence on surface charge density can be understood by considering formation of the electrical double layer adjacent to any charged surface in an ionic solution. The ions of opposite charge to the surface are known as the counterions and in this case are those residing on the positively charged catalyst; they are attracted to the surface by the surface potential. This potential is roughly proportional to the fixed surface charge density, producing a surface concentration of counterions that depends exponentially on the surface charge density [158]. It is quite plausible that this very strong dependence of the catalyst concentration near the surface initiators on the surface charge produces the effects observed. 0.1 E c "5 nj o 0.01 Table 3.3 shows that increasing the concentration of initiator relative to the concentration of charged groups is associated with the initiation efficiency decreasing to below 2%. When the [charge] to [initiator] ratio greatly exceeds 1, the initiation efficiency varies between 20 to 30%. Notably, for series B and C which have similar charge concentrations, increasing the initiator concentration 20 fold causes the initiation efficiency to decrease from around 20% to 1% but with only a doubling of graft density. Graft density can increase with increasing initiator surface concentration despite a decrease in initiation efficiency simply by mass action. Comparing series B and C, series C has a higher graft density and lower initiation efficiency. This is due to the very high concentration of initiators on surfaces in series C. Even when only 1.5% of all initiators were activated, the density of grafts was still greater than when 18% of all initiators were activated in series B because the original surface initiator concentration in series B was much lower to begin with. As shown in Figure 3.12, surfaces with higher initiator concentrations have a greater affinity for cationic ligands. At constant sulfate concentration, increasing the initiator content by 2.6 mmol/g caused a 2 times increase in graft density. However, Figure 3.16 demonstrates that at roughly constant initiator concentration, adding a small amount of sulfate (0.18 mmol/g) to an otherwise neutral matrix, the graft density increased by 58 times. The much larger effect on graft density when adding sulfate might explain why surfaces of high initiator content, although enriched in surface-associated cationic ligands, have lower initiation efficiencies and therefore lower graft densities. These arguments suggest that to maximize graft density both a high surface initiator concentration and a high negative surface charge density is required. Molecular weight is then controlled by monomer concentration. The graft density at the desired molecular weight can then be tuned by controlled hydrolysis of the dense brushes. Reaction procedures developed here have been designed to allow for such a targeted synthesis. 3.3.7. Effect of Changing Copper Concentrations in the Catalyst Mixture In the ATRP reaction equilibrium, increasing the Cu(I) concentration is expected shift the equilibrium to the right and decrease the dormancy of the active radical generating species. It was of interest to investigate whether this parameter could be used to manipulate the molecular weight or graft density of grafted PDMA chains on the modified chromatography matrix. It was possible to investigate this effect on surfaces with differing initiator concentrations Figure 3.17 shows the dependence of the graft and solution polymer molecular weight and polydispersity on the variation of Cu(I). ra Q 160000 140000-120000-100000• 80000 60000 40000 20000 —I— 0.0 —I— 0.2 —I— 0.4 —I— 0.6 —I— 0.8 —I— 1.0 —I— 1.2 [Cu(l)CI] (M) r-0.16 -0.14 -0.12 -0.10 0 ID ;* -0.08 (chai -0.06 in/nm ' -0.04 -0.02 -0.00 Figure 3.17: The effect of activating catalyst (Cu(I)Cl) concentration on graft molecular weight (empty symbols) and graft density (filled symbols) for low initiator concentration, 0.0129 mmol/g (A) and high initiator concentration, 0.133 mmol/g (•). For reactions for ( A) , 2 g of 3% seed suspension was used with [DMA] = 3.88 M . For (•), 14.1 g of 3% seed suspension was used with [DMA] = 1.56 M . For both reactions [Cu(II)/Cu(I)] = 0.15; Cu(0) molar amount was 8 times Cu(I); H M T E T A molar amount was equal to the sum of Cu(I) and Cu(II). For increasing Cu(I) concentration, the initiation efficiency and therefore the graft density was seen to decrease for the high initiator seed while it stayed relatively constant for the low initiator seed (Table 3.4). Table 3.4: Effect of activating catalyst (CuCl) concentration on fate of monomer and physical properties of the grafted surface. Series [Charge] [Initiator] [CuCl] Conv. DMA Solution polym. Graft polym. Solution Polymer M„ Solution MJM„ Graft Polymer M„ Graft MJM„ Initiation Efficiency Ografl D>(*.) mol/g x 10* mol/g x 10* M % (w/w) %of Conv. %of Conv. Da Da % chain/nm2 A 1.4±0.2 0.1 0.079 12.6 90.4 9.6 116000 1.57 146000 1.15 8.4 0.0054 0.8 0.178 5.4 78.1 21.9 41300 1.17 80300 1.09 14.5 0.0092 0.9 0.323 7.6 91.7 8.3 20000 1.76 32700 1.66 18.7 0.0119 1.3 0.469 3.9 89.2 10.8 18100 1.34 22700 1.17 17.9 0.0114 1.7 0.708 2.4 82.4 17.6 18500 1.33 21500 1.20 19.1 0.0122 1.7 1.168 1.3 80.9 19.1 16500 1.17 12600 1.18 19.1 0.0122 2.3 B 1.8±0.3 1.3 0.032 16.1 29.5 70.5 22400 1.39 18600 1.26 22.6 0.1491 0.5 0.126 12.3 42.8 57.2 17100 1.31 16000 1.23 16.1 0.1067 0.7 0.283 8.5 37.4 62.6 14400 1.36 13400 1.23 14.6 0.0965 0.8 Experimental Conditions: A) 14.1 g of 3 %(w/w) suspension of functionalized Toyopearl; [DMA] = 3.9 M; B) 2.0 g of 3 %(w/w) suspension of functionalized Toyopearl; [DMA] = 1.6 M; For both: Cu(II)Cl2 = 0.15 times molar amount of Cu(I)Cl; HMTETA = sum of molar amounts for Cu(I) and Cu(II). In both cases, the molecular weight dramatically decreased. For seed B, the Cu(I) ratio was increased from 1:7, to 1:30 to 1:69 times. For seed A , an error in the experimental design conserved the concentration of CuCl in solution and not the Cu(I) to initiator ratio. Because series A had very low initiator concentrations the Cu(I) to initiator ratios varied from nearly 1:200 to about 1:3000. The large variation had a dominant effect on lowering the graft molecular weight. The graft density, which one would predict to be low due to the low initiator concentration showed little variance; differences were not considered significant within experimental error. The graft densities were seen to decrease for the higher initiator seed. The graft polydispersities were less than 1.3 in nearly every case except for one outlier (1.66). Overall, the dominating effect of decreasing chain length by increasing Cu(I) serves to increase the ratio of DI(R^J for both seeds. Dense polymer brushes therefore can be synthesized at low Cu(I) to initiator ratios. 3.3.8. Effect of Cu(II) Concentration The addition of copper II (deactivator) along with Cu(I) is expected to give good control to an otherwise uncontrolled ATRP reaction [20;22;102;149]. The effect of varying the added Cu(II) mole proportion from 0.15 to 2 times the Cu(I) amount was investigated and results are shown in and Figure 3.18 and Table 3.5. 20000 -I 1 1 1 1 1 1- i.uu 0 1 2 Cu(ll)/Cu(l) (mol/mol) Figure 3.18: The effect of Cu(II) to Cu(I) ratio on the graft molecular weight (•) polydispersity (•). For the graft polymerization, 1.67 g of 6% seed matrix was used with [CuCl] =0.215 M , [DMA] = 1.07 M , [Cu(0)] = 0.26 M , H M T E T A molar amount was equal to the sum of Cu(I) and Cu(II) in each reaction. Samples also included 200 mM NaCl but this had no effect on the properties of the grafted polymer. Table 3.5: Effect of changing the concentration of the deactivating catalyst (copper (II)) concentration on fate of monomer and physical properties of the grafted surface. [Charge] [Initiator] Cu(ll)/ Cu(l) Converted Monomer Solution polym. Graft polym. Solution Polymer M„ Solution MJM„ Graft Polymer M„ Graft MJM„ Initiation Efficiency Ograft D/(*.) mol/g x 10* mol/g x 10* mol/mol % (w/w) %of Conv. %of Conv. Da Da % chain/nm2 0.55 0.08 0.15 5.6 82.6 17.4 14400 1.37 19200 1.19 11.2 0.00429 3.0 0.5 2.3 77.1 22.9 22900 1.29 13800 1.17 8.2 0.00315 4.3 1 0.5 45.6 54.4 20200 1.27 11400 1.33 4.9 0.00186 6.3 2 0.4 30.7 69.3 49600 1.64 7800 1.43 7.1 0.00271 6.6 Experimental Conditions: 1.67 g of 6 %(w/w) suspension of functionalized Toyopearl; [DMA] = 1.1 M; Cu(I)Cl = 0.22 M; HMTETA = sum of molar amounts for Cu(I) and Cu(II). Increasing the Cu(II) proportion decreases the graft molecular weight - a result which is consistent with shifting the ATRP equilibrium to favor a decrease in the proportion of active radicals. Reactions at low Cu(II) to Cu(I) proportions were well controlled with a MJM„ less than 1.2. There is very little change when increasing the proportion to 0.5. For ratios at 1 or greater, the increase in polydispersity indicates less control over the reaction. Together, the relatively large decreases in molecular weight and invariance in the graft density act to increase the value of Dl(R^j. Although all the grafted polymers in this case would be considered mushrooms, an increasingly brushy character would be helped by lower copper II proportions. When the Cu(II) proportion was fixed to 0.15, many of the polymer systems shown elsewhere in this work produced true polymer brushes when manipulating the reaction parameters to increase the graft density. 3.3.9. Reaction Scale Up Small scale (0.1 g) test reactions were used to scout for optimal reaction parameters required for large batches of media for column packing. To pack a chromatography column of 30 cm long by 1 cm inner diameter, it was estimated that 7.5 grams of seed was needed. For the purposes of evaluating its use an EIC separations medium a dense polymer brush of moderate molecular weight (around 50 kDa) was selected as a target. High graft density was also targeted by synthesizing a seed that had nearly equal proportions of charged and initiating groups. To select for the targeted molecular weight, monomer concentration was chosen to be 1.6 M . A small amount of Cu(II) (0.15 molar proportion to Cu(I)) was chosen to select for a low polydispersity. Details for the reaction parameters are supplied in the experimental section. The results from characterization of the grafted polymer are given in Table 3.6. Table 3.6: Characterization of the polymerization for one lot of grafted matrix intended for medium scale EIC studies. Measured Parameter Value Converted Monomer 45.6 % (w/w) Solution polymerization 24.4 Percent of Conv. Surface polymerization 75.6 Percent of Conv. Graft Polymer M„ 55700 Da MJMn 1.36 Initiation Efficiency 24.8 % Cgraft 0.164 chain/nm2 0.3 The resulting material showed both a high degree of conversion and high initiator efficiency. Most notably, polymer grafted to the matrix caused a weight increase of 192% and a significant increase in the volume of the settled material. The resultant graft length (55700 Da) was very close to the targeted value. The low value of Dlt^Rg^j (0.3) suggests that the morphology of the grafted layer was far into the brush region. The slightly higher value for MJM„ (1.36) may be indicative of the scale of the reaction. Unlike small scale reactions, overhead stirring was used for the polymerization reaction and may have had an effect on the degree of control during polymerization; however, this phenomenon was not investigated further. The flow and elution properties, including demonstrated size separation of solutes based on the principals of EIC, were all acceptable and results of these studies are discussed in Chapter 4. 3.4. Summary An investigation into the influencing effect of surface initiator and sulfate concentration on the aqueous ATRP of D M A to high throughput chromatography matrices was extended from Chapter 2. A desired polymer molecular weight could be selected by changing the monomer concentration. Increased concentration of monomer in the feed produced higher molecular weight grafts at the expense of creating lower density surfaces - an effect of the lower solubility of the catalyst system in monomer-rich feeds. Grafted polymers produced for this study covered a wide range of morphologies from dilute mushrooms to dense brushes. Surfaces with high initiator concentrations have the potential to produce dense brushes because of their greater affinity for catalyst and low distance between initiating sites. The presence of negatively charged sulfate groups is a requirement for having higher graft densities. These principals were used to create a large batch of grafted media used for studies of the chromatographic behavior of EIC columns. The material has been well-characterized and the hydrolyzed grafted polymer molecular weight, (-Rg), and polydispersity have been explicitly determined by SEC/MALLS. From a surface model which considers the surface area accessible to macromolecules of similar size to polymer grafts, a representative surface area which captures the physical environment of grafts was obtained. This permitted calculation of graft density and Dl (R^j values from the known amount of polymer grafted for this highly porous matrix. The resultant material was a densely grafted polymer brush with a targeted graft molecular weight. Because the molecular weight can be controlled by ATRP reaction conditions and graft density can be targeted through synthesis pre-grafting, or through partial hydrolysis, post-grafting it is clear that surface initiated ATRP has been adapted to produce a tunable system that can be applied to studies of the principals of EIC. 4. CHROMATOGRAPHIC EVALUATION OF EIC MATRICES 2 4.1. Introduction Entropic interaction chromatography is a novel and efficient method for separating mixtures of macromolecules on the basis of size [13; 14]. The stationary phase of an EIC column presents a layer of hydrophilic end-grafted linear homopolymer into which macromolecular solutes may partition, resulting in a decrease in the configurational entropy of the grafted polymer chains AScbrush that scales strongly with the hydrodynamic volume of the partitioning solute [13;87]. Akin to traditional size-exclusion chromatography, the partitioning process also decreases the entropy of the solute AS', resulting in a total decrease in entropy of AS 1 0 = AScbrush+AS.. Like AScbrush, AS. becomes more negative with increasing size of the partitioning solute, resulting in a strong inverse dependence of elution time on solute molecular weight. Previously the performance of two preparative EIC columns [14] was reported, each displaying a relatively low density PMEA layer grown on the walls of a commercial matrix bearing surface aldehyde groups (Toyopearl AF-650M). Toyopearl AF-650M is a hydrophilic, dimensionally stable matrix with uniform 1000-A pores that permit intraparticle convective flow but provide no size-based separation of proteins or other macromolecules less than 1,000,000 Da in molecular weight. Each grafted PMEA layer was introduced onto the porous surface of the base resin using a cerium (IV) catalyzed polymerization reaction at conditions similar to those used in a commercial form of EIC, the Fractogel BioSEC column 2 A version of this chapter will be submitted for publication. B. R. Coad, B. M . Steels, J. N . Kizhakkedathu, D. E. Brooks, and C. A. Haynes. The Influence of Grafted Polymer Architecture and Fluid Hydrodynamics on Protein Separation by Entropic Interaction Chromatography. Biotechnology and Bioengineering. 2006. of E. Merck KgaA, and described in previous work [13]. The two columns differed primarily in the grafting density of the PMEA layer. The selectivity curves for both columns were linear over two to four orders of magnitude in solute molecular weight, depending on the graft density. As predicted from a self-consistent field theory for the interaction of protein-like particles with end-grafted polymer chains in a good solvent [80;87], the column displaying the higher chain density excluded proteins more strongly, resulting in lower partition coefficients, faster elution times, and a steeper selectivity curve for proteins with molecular weight less than ca. 30 kDa. The superior throughput of this higher density resin is balanced by its lower resolving power for high molecular-weight proteins, suggesting that a desired set of column performance characteristics can be achieved by tuning the properties of the grafted polymer layer, including the grafting density, average chain length, and the chain chemistry. Although it can be used to produce efficient EIC columns, Ce(IV) catalyzed grafting chemistry has certain limitations. In particular, the chemistry does not readily permit synthesis of dense polymer brushes where the distance D between grafting sites is considerably less than 2(Rg^, where (^Rg^ is the polymer radius of gyration. Precise control of grafted chain length N and polydispersity (= MJMW) is also very difficult [119]. Many of these problems can be addressed through the use of surface-initiated atom transfer radical polymerization, which permits synthesis of a wide range of grafted polymer layers with high graft density, low polydispersity, and desired number-average molecular weight (M n). This method allows the controlled synthesis of polymers from a variety of monomers in both aqueous and organic media under mild conditions. Another advantage of this method is that the end of the polymer chain is capped with an active halide atom that can be reinitiated to form block copolymer brushes with addition of fresh catalyst and monomers. Hence, a variety of micro-structured brushes can in principle be constructed [19;88;112]. ATRP has therefore been used to produce several novel stationary phases for capillary electrophoresis [45;46] and for capillary zone electrophoresis [47]. Here, EIC columns from Chapter 2 and Chapter 3 are evaluated. EIC from nonporous particles is covered briefly to complete the study showing that ATRP can be used to create a functioning EIC medium from PSL. Elution properties and graft properties affecting resolution are discussed. Evaluations of columns synthesized in Chapter 3 are covered in much more detail including moments analysis, shear flow behaviour, and comparisons with SCF theory. Based on combining SCF theory with the classic equilibrium-dispersion model of chromatography, the theory of EIC [14] predicts that high graft density (HGD) brushes will exhibit steep selectivity curves over a narrow distribution of relatively low protein molecular weights, giving them performance characteristics distinct from the low density EIC columns previously synthesized using Ce(IV)-catalyzed grafting chemistry, and potentially making them well suited for desalting and buffer-exchange applications. ATRP chemistry is used to explore the effects of a and M n on protein partitioning and separation performance, while ester hydrolysis is combined with SEC-MALLS and refractive index measurements to characterize the properties of the grafted brush and thereby provide a much more detailed understanding of the influence of brush architecture on protein separation by entropic interaction chromatography. The hydrodynamic response of the grafted polymer layer and its effect on column performance is also explored through estimation of hydraulic permeabilities, first and second central moments analysis, and interpretation of chromatography data using a self-consistent field theory describing the configurations of an end-grafted polymer brush in the absence and presence of solvent flow [159; 160]. Results are used to examine the potential of surface-initiated ATRP for custom design of EIC resins for particular size-based separations. 4.2. Materials and Methods 4.2.1. Synthesis of EIC Stationary Phase Procedures for the synthesis of EIC media from nonporous supports were discussed in Chapter 2, and from porous media in Chapter 3. 4.2.2. Choice of Running Conditions The buffer used for the mobile phase for size-exclusion studies was 10 mM sodium phosphate buffer (pH 7.00±0.05) containing 300 mM NaCl. The mobile phase was chosen to be a good solvent for protein probes used in the study and was filtered twice through 0.22 micron filters (Millipore) to remove particulates prior to use. Injection volumes were chosen to be approximately 0.5% of the column volume. For each injection, the sample loop was filled with a 5 times excess of sample to ensure that the sample buffer had been replaced by a homogenous sample of analyte. 4.2.3. EIC Column Packing A l l EIC columns were packed using a standard balanced density slurry packing technique [161]. A packing tube 15 cm in length was attached via an adaptor to the top of a Pharmacia 10/300 column (1 cm i.d., 33.5 cm length end-to-end). The total volume of the column and the packing tube was 46 mL. A slurry containing 10% excess amount of grafted EIC matrix required to pack the column (26.3 mL) was transferred to a 50 mL centrifuge tube and mobile-phase buffer was added to bring the total volume to 46 mL. The resulting 57% (v/v) suspension was washed ca. 10 times with buffer by centrifuging at 1500 rpm for 5 minutes to settle the larger particles. Matrix fines were siphoned off following each centrifugation. The final slurry was resuspended in mobile-phase buffer and quickly transferred to the packing column, which was then sealed with a top flow adaptor and connected to an A K T A Purifier 10 system (GE Healthcare). Mobile phase was pumped through the column at 10 mL min"1 for 15 minutes to pack the matrix to a height of 30.7 cm under a backpressure of 0.99 MPa. A distribution filter and flow adaptor were then mounted and mechanically lowered to yield a final bed height of 30 cm, representing a 2.3% by volume compression of the original bed. Smaller scale columns were also packed for more limited characterization studies. Omega columns (Upchurch) of length 7.5 cm and inner diameter 0.46 cm were packed by attaching a 10 cm long packing reservoir (same inner diameter) to the column. The bottom of the column assembly was fit with a 2 jam porous frit secured in place with a screw end fitting. Approximately 3 mL of clean EIC matrix was suspended and transferred to the column packing assembly. Packing was facilitated by pumping buffer through the system at 1 mL min"1 for 15 minutes, after which the packing tube was disassembled and the packed bed was leveled to the top of the column. The top frit was set into place and secured with an end fitting, again providing a slight compression of the bed. The quality of packing was evaluated using the method of Bristow [162] by determining HETP for 100 uL (large columns) or 5 uL (small columns) spikes of 0.3% v/v acetone in water injected onto the column at a superficial velocity of 0.5 cm min"1. Acceptable column packing was assumed for small columns with reduced plate height hT (= HETP I'dp, where dp is the diameter of the stationary phase particles) less than 2. On columns prepared by ATRP chemistry, reduced plate heights from 1.5 to 4.0 relative to the diameter of the stationary phase particles were observed. 4.2.3.1. Evaluation of Protein Partitioning and Selectivity Curves A l l chromatography experiments were.performed on an A K T A Purifier 10 equipped with 0.25 mm PEEK tubing of minimum length between the injector and column and between the column and detector to minimize dead-volume and peak-dispersion effects. Large and small column samples were loaded into a 100 uL and 5 uL sample loop, respectively, and (generally) injected onto the column at a superficial linear velocity of 0.5 cm min"1. Each elution chromatogram was monitored by U V absorbance at 280, 254 and 215 nm. Solute standards used to characterize the column included U V active polysaccharides, proteins, polypeptides, small organic molecules, and inorganic salts (see Appendix 3). The molecular weights of polypeptide standards were determined by M A L D I MS. A l l remaining M W standards were obtained commercially and used as supplied. 4.2.3.2. Determination of Mass Transfer Resistance and Parameters Resistances to mass transfer and associated mass transfer parameters were determined by central moments analysis of KNO3 elution bands using the standard one-dimensional equilibrium-dispersion model of column chromatography for a non-binding solute [163; 164]. Duplicate injections (100 uL) of the salt dissolved in buffer (0.1 g/L) were made at superficial velocities from 20 to 110 cm/hr in increments of 10 cm/hr and detected by U V absorbance at X = 215 nm. The pressure drop across the column was measured with the built-in pressure sensor. At the highest flow rates, no physical compression of the bed was observed. The net pressure drop was calculated by subtracting the system pressure at the same flow rates when an unpacked tube offering no significant flow resistance was used in place of the column. 4.3. Results and Discussion 4.3.1. Chromatographic Evaluation of Grafted, Nonporous Materials Grafted PSL lots from Chapter 2 were packed in small 1.2 mL chromatography columns. Packed columns of 7.5 cm length typically generated pressures upwards of 5 MPa at moderate flow rates of 0.1 mL min"1. Media grafted with high molecular weights caused the greatest flow resistance. In some cases, media could not be packed because of pressures exceeding 10 MPa. The low pressure chromatography pumping system and columns were not suitable for continuous operation at these high pressures because of leaks and possible damage to the system. For nonporous media, only a small number of columns could be successfully packed and evaluated by chromatographic methods. For these columns, selectivity curves were generated and are shown in Figure 4.1. The physical properties of these lots are given in Table 2.11 and are summarized in the figure caption. 1.3-1 r r q r - i-1 i t n i | 1 — i ~ r i i m\ r - T , . , . . , T n j , — r ' l ' f m r q r 100 1000 10000 100000 1000000 MW Figure 4.1: Calibration curves for nonporous columns. L2Graft (•): M„= 184 kDa, agraf, = 1.6 x 10" chain/nm2; L4Graft (x): M„ = 124 kDa, agraft = 21 x 10"3 chain/nm2; L5Graft (o): M„ = 203 kDa, agrafi = 0.75 x 10"3 chain/nm2; L8Graft (A): M„ = 59 kDa, agraf, = 5.2 x 10"3 chain/nm2. Data in Figure 4.1 show that PSL-based columns of very different graft properties show different selectivity curves. The two columns with the highest graft densities (L4Graft and L8Graft) were the most selective. However, since graft M„ also changes in this data set, it does not permit an unambiguous, independent analysis of the graft density effect. Nevertheless, it can be concluded that ATRP was successful in creating EIC media and that differences in column selectivity were influenced by the graft properties of the matrix. Further characterization of grafted columns was not possible due to the relatively low pressure restriction of the chromatography system and columns. A more detailed discussion concerning evaluation of EIC media is continued in the next section, where a library of porous, high-throughput stationary phases were synthesized and used to independently investigate graft M„ and density effects. 4.3.2. Porous EIC Media 4.3.2.1. Protein Desalting Using High-Graft-Densitv EIC High-resolution separation of proteins by size exclusion chromatography is generally carried out at relatively low superficial velocities due to the slow diffusivities of the high molecular weight (HMW) analytes and the relatively large diffusional paths within the porous stationary phase matrices. The use of SEC in preparative applications therefore challenges overall process throughput, an important consideration when selecting purification methods, especially for less complex separations such as buffer exchange or the removal of specific salt from a biopolymer preparation. EIC permits convective flow through the giga-pores of the stationary phase, as has been previously shown, allowing rapid protein contact with the grafted-polymer layers providing the sieving mechanism [14]. Because they create steep selectivity curves that effectively exclude proteins of molecular weight > 10 kDa, HGD EIC columns offer a potentially attractive method for buffer exchange or desalting of biopolymer samples. Efficient removal of a salt or buffering agent can be achieved (Figure 4.2) due to the selective partitioning of the low molecular-weight salts into the dense grafted polymer phase. The unwanted salt or buffer is thereby retained and replace with the desired salt or buffer in the mobile phase. 3000 -, 2500 -2000-3 < w 1500-1000 5 0 0 -0 -6.65 17.14 BSA KNO, 10 15 20 V. (mL) i — 25 Figure 4.2: Chromatogram for removal of K N 0 3 (0.5 g L"1) from BSA (2 g L"1) injected as a 100 uL pulse onto the 30 cm HGD EIC column (cr= 164 X 10"3 chains/nm2, Mn = 55700 Da, column volume = 23.6 mL) at 0.5 cm min"1. The 10 mM sodium phosphate buffer (pH 7) in the mobile phase contained 300 mM NaCl. Mean elution times for each peak are shown. ATRP chemistry allows good control of graft properties (see below). However, the architecture, length and operating conditions of the HGD EIC column used to generate the chromatogram reported in Figure 4.2 were not optimized to maximize throughput while preserving a resolution greater than unity. To better address the performance characteristics of this HGD EIC resin, the reported elution data was combined with the plate model of Grushka et al. [165]. For a column containing a large number of plates Np, elution peaks are predicted by plate theory to be Gaussian in shape with the distribution of solute concentration c(Ve) in the eluate given by [164] R-NP(r-K)2 c(7J = c0exp 2v: [4.1] where c0 is the concentration of the solute at the elution peak maximum, V is the volume of mobile phase passed through the column, and Ve is the elution volume of the solute at peak maximum. For a Gaussian elution peak, Np = \6(VJW) . Equation 4.1 therefore predicts the minimum time and column length required to fully remove salt from BSA using this HGD EIC resin are 3.6 min and 13.3 cm, respectively. A reduction in plate height through improved column packing could lower these values further, but even now the HGD EIC column prepared by ATRP chemistry is predicted to achieve a 1.8-fold higher throughput than either of the lower graft density EIC resins shown in the literature [14]. 4.3.2.2. Influence of Grafted Polymer Architecture In a good solvent system, the chain length N and grafting density a are the primary determinants of the conformations and equilibrium properties of grafted chains. As TV and/or a are increased, the distance between chains decreases, ultimately reaching a point where D is less than two times (Rg)- To minimize their free energy, the grafted chains then extend away from the surface in a conformation analogous to the bristles of a brush. Scaling thermodynamics predict that the grafted polymer layer enters the brush regime when [26] f \i -6/5 >N~°'3 [4.2] An understanding of grafted chain conformation can also be obtained by calculating the reduced surface coverage <y = niR^jlL, which gives the ratio of the cross-sectional area of the chain free in solution to the average area S (=11 d) available to chain once it is end-grafted to the surface. The brush regime is therefore entered when a* approaches unity [27;28]. Although precise determination of a was not possible for the two EIC columns produced by Ce(IV) catalyzed grafting chemistry due to the unknown polydispersity of the grafts, both columns are thought to be characterized by values of o* slightly less than unity, indicating that they at most display a very weak brush-like architecture [14]. Surface-initiated ATRP chemistry allows one to synthesize HGD resins that can be used to explore EIC separation performance over a much wider range of graft densities and polymer molecular weights, while at the same time permitting careful control of grafted-chain polydispersity. 4.3.2.3. Graft Density Effects Both experiment and theory show that polymer segment density profiles and chain configurations depend on grafting density [25;26]. Figure 4.3 compares the globular-protein selectivity curve for a HGD EIC column (164±5 x 10"3 chains/nm2, D/(Rg) = 0.3, M n = 55700±900 Da, MJMn = 1.36±0.02) prepared by surface-initiated ATRP to that for a lower density EIC column prepared by Ce(IV) catalyzed grafting chemistry [14]. Average grafted polymer molecular weights in the two columns are similar, although the MJMn of the Ce(IV) catalyzed graft is not known. Despite this uncertainty, increasing a such that a* is well above unity clearly steepens and narrows the linear region of the selectivity curve such that strong differences in partitioning are now observed for analytes of molecular weight < 10 kDa. 1,2-i 1.0-0.8 o:6 0.4-0.2-0.0 100 1000 o L G D EIC Column o HGD EIC Column J 8 8° '©o : o 10000 M W i — i — i * 1 1 1 1 1 1 100000 1000000 Figure 4.3: The measured selectivity curves for a HGD EIC column (164±5 x 10"3 chains/nm2, D/(^Rg) =0.18, M„ = 55700±900 Da, MJM„ = 1.36±0.02) prepared by surface-initiated ATRP, and for a low graft density (LGD) EIC column (Dl{R^j ~ 2, ( M „ ) = 48500 Da) prepared by Ce(IV) initiated grafting chemistry [14]. As was observed for low-density EIC columns, protein-partitioning behavior within the fully developed brush regime is sensitive to changes in a. Increasing Dl(R^j by a factor of 8.9 widens the linear portion of the selectivity curve from an exclusion cutoff of ca. 10 kDa to near 90 kDa (Figure 4.4). I 100 1000 10000 1 ' " " I — 1 100000 TT 1000000 MW Figure 4.4: Effect of changing graft density in the brush regime for EIC columns of similar graft molecular weight: H G D EIC column (•) (cr= 164 x 10"3 chains/nm2, M n = 55700 Da, column volume = 23.6 mL); L G D EIC column (A) (<r= 1.49 X 10"3 chains/nm2, Mn = 53200 Da, column volume = 1.24 mL). Both columns were operated at u = 0.5 cm min"1. The relatively high concentration of grafted chains within the polymer layer when D/(R) < 2 (Figure 4.5) effectively exclude high-molecular-weight proteins (> 100 kDa). Selection of a cr and M n combination within the brush regime therefore allows one to tune the differential partitioning and separation of lower molecular weight proteins and analytes, while grafts characterized by D/(R) values greater than 2, the so-called "mushroom" graft regime [25], are required to extend the linear region of the selectivity curve to include HMW proteins. 1 ' 1 ' 1 T 1! 0 5 10 15. DKR •» Figure 4.5: Effect of graft morphology on the percentage of the total column volume occupied by grafted polymer material (100 {</>pohme^ )• The grafts were prepared from 4 sets of EIC media varying in graft molecular weight: (o) 19.1 kDa, (o) 53.2 kDa, (A) 69.3 kDa, and (0) 56.9 kDa. 4.3.2.4. Graft Molecular Weight Effects Table 4.1 reports polymer graft properties and solute elution data for a large family of 1.24 mL EIC columns prepared by surface-initiated ATRP chemistry. Mn MJMn Cgraft D D , ( R . ) * a V, V0 Da chain/nm 2 x 10 4 nm mL mL mL 19100 1.12 8.58 34.1 7.3 0.068 1.13 0.90 0.24 0.26 2.62 61.8 13 0.021 1.17 0.90 0.26 0.29 2.27 66.3 14 0.018 1.17 0.89 0.29 0.32 2.28 66.2 14 0.018 1.14 0.84 0.30 0.36 53200 1.16 14.9 25.9 3.0 0.40 1.09 0.72 0.37 0.51 13.6 27.2 3.1 0.37 1.07 0.69 0.38 0.55 9.13 33.1 3.8 0.25 1.09 0.68 0.41 0.60 8.23 34.9 4.0 0.22 1.08 0.71 0.37 0.52 5.96 41.0 4.7 0.16 1.09 0.68 0.41 0.61 4.22 48.7 5.6 0.11 1.10 0.71 0.38 0.54 69300 1.29 18.7 23.1 1.6 0.69 1.06 0.64 0.42 0.65 16.6 24.5 1.7 0.62 1.06 0.64 0.41 0.64 15.9 25.1 1.8 0.59 1.08 0.64 0.44 0.68 13.7 27.0 1.9 0.51 1.07 0.64 0.43 0.66 11.5 29.5 2.1 0.43 1.07 0.65 0.41 0.64 9.56 32.3 2.3 0.36 1.08 0.65 0.43 0.66 SEC-MALLS data for hydrolyzed polymer chains were used to determine a, MJMn and D, and to derive an empirical equation relating the (Rg) of PDMA chains to M n (Rg) = 0.01355 x(M„)A [4.3] 155 where (RGJ values are given in nanometers when M n is expressed in Daltons. Equation 4.3 permits estimation of both o* and Dl(R^j for each polymer graft in the series. The results show that surface-initiated ATRP chemistry can be used to produce very low polydispersity EIC resins covering a wide range of chain lengths and grafting densities, such that grafted PDMA layers displaying either mushroom or brush morphologies are now experimentally accessible through appropriate combinations of M n and a (Figure 4.5). In the brush regime, small decreases in T)I{R^ result in significant increases in polymer density within the column (Figure 4.5). Equally important to separation performance, however, is the fact that an increase in Mn results in an increase in the volume Vp (= Vt - V0) of the grafted polymer layer into which proteins and other higher molecular weight solutes can partition, resulting in increased column capacity and overall resolving power (e.g. a decrease in hv). Table 4.1 reports values of VplV0, which represents the volume of the polymer layer relative to that of the void. Here, VQ has been defined as the elution volume of a high molecular weight tracer (blue dextran, M W = 2,000,000 Da) that is completely excluded from the grafted layer. Increasing M n results in a rapid increase in VplV0 and thus an increase in the column capacity. This latter effect is seen in Figure 4.6, which compares solute-retention curves for two EIC columns having similar grafting densities but significantly different grafted-chain M n values. o M= 19100 Da 1.2-a M = 69300 Da 1.0-0.8-0.6 1000 TTI'tl "1 10000 100000 1000000 MW p Figure 4.6: Retention volume data for probe molecules eluting from near constant a EIC columns (1.24 mL total volume) bearing different graft molecular weights. The solid lines are best-fit sigmoidal functions. The graft density for the columns was 8.58 X 10"4 chains/nm2 and 9.56 X 10"4 chains/nm2 for (•) and (A), respectively. Only a modest decrease in elution volume (Ve) with increasing M n is observed in the low-M n column. In contrast, sharp and significant differences in retention volumes are observed in the high-M n column, permitting efficient separation of analytes on the basis of size due to the increased capacity of the stationary phase to accept low molecular weight analytes. 4.3.2.5. Hydrodvnamic Contributions to Partitioning and Band Broadening in EIC The partition coefficient Kt for each solute / was computed from the corresponding elution volume for the solute as Columns [4.4] where Vt is the elution volume of a low molecular weight tracer (KNO3, M W =101 Da) that is assumed to access the entire liquid volume of the grafted polymer layer. In the low density EIC columns produced by Ce(IV) catalyzed grafting chemistry [14], K\ is independent of mobile-phase superficial velocity u over the entire range of normal column operating conditions, which permits accurate modeling of elution chromatograms using a standard equilibrium-dispersion model of column chromatography [14]. In contrast, K\ is a function of u when the EIC column displays a HGD brush. For example, Figure 4.7 reports the observed decrease in K\ with increasing u for BSA partitioning into the HGD EIC column described in Figure 3.1, indicating that solvent flow through the porous matrix may deform the grafted polymer layer and thereby alter equilibrium properties and possibly hydrodynamic and mass-transfer properties within the column. 0134 0.132 0:130 0.128 0.126-0.124 -0.122-0.120-0.118-20' 30: 40; ;so —r— :60 u (cm h'1) 70 80. Figure 4.7: Measured partition coefficients for BSA in the HGD EIC column described in Figure 2 as a function of u. Error bars represent the standard deviation of duplicate runs. 4.3.2.6. Moments Analysis Although the true functional relationship between band broadening and flow velocity remains a matter of significant debate [166], analysis of first and second central moments [163] offers a widely used and effective strategy for interpreting band broadening using an equilibrium-dispersion model of chromatography that explicitly accounts for resistances to solute mass transfer. Applying this concept, Ruthven [163] showed that -2 [4.5] where ju\ and //2 are the first and second central moments of the elution peak, L is the column length, Di (m2 s"1) is the axial dispersion coefficient, u is the superficial velocity of the mobile phase, Km (s"1) is the overall solute mass transfer coefficient, s is the column void fraction measured by first moment analysis of the elution peak of dextran blue (which is completely excluded from the brush phase) to be 0.79±0.03, and /? is the effective porosity of the stationary phase measured by first moment analysis of the elution peak of KNO3 to be 0.72±0.02. The first term on the right hand side of Equation 4.5 accounts for path-dependent analyte dispersion (the path lengths of individual molecules differ in a random fashion due to eddy diffusion and column packing effects), while the second term gives the overall resistance to analyte mass transfer into and from the stationary phase, which is proportional to u and accounts for band broadening due to analyte in the mobile phase moving ahead of analyte in the stationary phase. Equation 4.5 predicts that a plot of u x HETP versus u should yield a straight line provided s, (3, Km and A are independent of u, which is typically the case with mechanically stable stationary phases and was observed for both of the low jUZ2L _ 2D, + 2u-\-e) 1 + ( 1 - 4 * J density EIC columns prepared by Ce(IV) catalyzed chemistry [14]. Figure 4.8 reports HETP (= L/Np) data for the 30 cm HGD EIC column when KNO3 is injected as a 100 uL pulse. 1.8 -1.6-1.4-0.0 '"t • |.........»~--—..-.^^ 1 T j 0 2 4 6 if (xlO'cmV) Figure 4.8: HETP versus flow rate data for 100 uL pulses of K N 0 3 (0.1 mg/mL) injected onto the HGD EIC column described in Figure 4.2. Linear behavior is observed, allowing estimation of values of Km and D\ (Table 4.2) which can then be compared with those obtained for a low-graft-density (LGD) EIC column. Table 4.2: Geometric and mass transfer parameters determined by moments analysis of elution peaks for pulse injection of K N 0 3 . EIC Column Di Km (m 2 s-1) HGD EIC 0.3 0.75(±0.08)x 10"8 0.131±0.007 LGD EIC ~ 2 5.68 x 10"8 0.516 The roughly two-fold higher grafted polymer mass in the HGD EIC column (Figure 4.5) reduces axial dispersion (D{) by a factor of 7.6, presumably because the significantly higher hydraulic resistance created by the dense polymer graft minimizes differences in molecular path length and promotes plug-flow behavior. This potential reduction in band broadening is, however, partially offset by a nearly 4-fold reduction in the rate of solute (KNO3) mass transfer into the stationary phase, which is expected given the higher density of the graft. 4.3.2.7. Shear-Induced Deformation of End-Grafted Polymer Chains Both end-grafted and adsorbed polymer chains are known to undergo shear-induced tilting and elongation in the direction of flow [160]. Previous studies have largely focused on planar surfaces exposed to fully developed laminar flows [167; 168]. At low shear rates, de Gennes scaling rules for the polymer blob formalism in the semi-dilute region hold, so that the characteristic blob size £ is determined solely by the volume fraction of the monomer as predicted from the approximate relation <f> = (a/Q4/3 first derived by de Gennes assuming self-avoiding walk statistics [26]. At sufficiently high shear rates, the polymer layer thickness, and possibly is expected to decrease as the chains tilt more strongly in the direction of flow. Such changes in brush morphology could lead to the observed flow rate dependence of KQSA reported in Figure 4.7 for the HGD EIC column. Two mechanisms are possible: clearing of the pore center leading to an increase in the hydraulic permeability of the pore (which is not well supported by the mass transfer analysis provided above), or nonlinear changes in the brush segment density profile and configurational entropy that alter £"and the partitioning behavior of solutes into the grafted polymer layer. A change in the hydraulic permeability of the column with flow rate should result in a nonlinear dependence of column pressure drop AP on u. For a packed bed of length L through which flows a mobile phase of viscosity rj, the generalized form of Darcy's law is given by AP = — TJU [4.6] KP A linear dependence of AP on u is therefore predicted unless the hydraulic permeability coefficient KP (cm2) of the column changes with u, due for instance to an opening of the pore volume (increasing the void fraction s) as the brush height collapses under increasing shear rates. However, as shown in Figure 4.9, AP depends linearly on u in the high-graft-density EIC columns; the same is true for the lower graft density EIC columns produced by Ce(IV) catalyzed grafting chemistry [14]. 1 -u i — 1 — i — 1 — i — 1 — i — 1 — i — • — i — • — i — ' — i — 1 — i — 1 — i — 1 — i 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 u (cm hr"1) Figure 4.9: Reduced net pressure drop (AP/L) versus u across the 30 mL H G D EIC column described in Figure 4.2. This result is not surprising, as the flows reported in Figure 4.9 are characterized by unmodified shear rates yr (= w/aakoiumn, where <iCoiumn is the column diameter) between 0 and 0.037 s"1, far below the shear rates required to collapse planar grafted polymer layers [169;170]. The previously described SCF model of grafted polymer layers [14;87] was used to explore whether shear-induced deformation of monomer volume-fraction profiles (fi(r) and grafted chain dimensions in the direction normal to the grafting surface could account for the observed dependence of ^ b s a on u (Figure 4.7). Two modifications were made to the model. The first and more straightforward change was to reposition the grafting surface within the cylindrical geometry of the SCF model to lie along the inner wall at r = r w a i i of the cylinder of length z = L. The grafted chains therefore project toward the cylinder centerline, as they would in a stationary phase pore of the EIC matrix. The second modification involved introducing a shear flow in the +z direction by defining a viscous force F* of the form F\r) = r,aupore{r) [4.7] where a is the effective size of each chain segment (Kuhn segment) upon which the viscous force acts. The probability p of a Kuhn segment within the cylindrical lattice moving in the +z direction is therefore increased by an amount proportional to the magnitude of F^ir) such that A + . - / V . = C . ^ [4-8] kT where k is Boltzmann's constant, T is temperature, and C\ is a proportionality constant set such that pz+\ equals 1/3 for the grafted chains in their fully extended configuration. The modeling work presented in the following section was completed by Bradley M . Steels using data provided by the author and is included with his kind permission due to its relevance to this work To account for shorter length grafts, the strategy of Castro et al. [171] and Cohen [172] was followed to segregate the viscous flow profile into two regions. The first region is defined by the polymer-free cylinder from r = 0 to r = H (where H is the equilibrium height of the grafted layer) in which fluid momentum in the axial direction is given by where P is pressure. Solvent velocity profiles within the annulus from r = H\o the pore wall at r = r w a n are estimated by treating the grafted polymer layer as a porous medium and applying the Brinkman equation where Kp(r) is the hydraulic permeability coefficient of the grafted layer, which can be estimated as C 2 (1 - <f>(r))l(/)(r); the constant C2 for PDMA in water has been regressed from Equations 4.6 to 4.10 were combined with the grafted polymer SCF model of Steels et al. [87] to obtain self-consistent solutions for (ftf), u(r), and H as a function of F2, cr, and N, where N is now the number of Kuhn segments used to represent the chain. Figure 4.10 reports solvent velocity profiles inside a 1000 A cylindrical pore as a function of N and cr for a model PDMA graft. 1 d | du I 1 dP [4.9] r dr V dr) rj dz [4.10] column pressure drop data to be 7.2(±0.4) x 10"15 cm2. — 0 — No Graft - < - o = 4.7%, N = 200 —>—a = 9.6%, N = 200 —O—a = 9.6%, N = 400 0.01 -1E-3-* 0,0 o!l ri'z 0*3 0L4 0^ 5 Cr.e{ OJ Q;8 0^ 9 10 •wall., Figure 4.10: Predicted normalized velocity profiles for an average solvent velocity (upor^ = 1.8 cm min"1 {yr of0.039 s"1) in a 1000 A pore bearing an end-grafted PDMA layer. In the model, grafted chain length is specified by the number of freely joined Kuhn segments N(=6 {R^JI l\, where the Kuhn length / K is twice the persistence length / P , estimated from the worm-like chain model to be 5.7 nm for P D M A [25]) and grafting density a is specified by the percentage of pore wall lattice sites bearing a grafted monomer. The Flory parameter %bo for the polymer-solvent interaction was set at 0.4. All other interactions were assumed athermal. Simulations for solvent hydrodynamics within a pore bearing no grafted polymer are also shown. Compared with the case for the unmodified pore, the velocity increases more slowly as r decreases away from rw an due to the hydraulic resistance of the polymer layer. The reduction of the velocity within the grafted layer increases with increasing 7Y or cr, with an increase in a having a stronger effect. The presence of flow within the grafted-layer volume helps to explain the low retention times in EIC columns as compared to traditional SEC columns of equivalent volume operated at the same u. As with membrane chromatography [173], the model predicts that convective transport dominates diffusive transport within the column. For the PDMA graft and range of shear flows used to generate the HETP data reported in Figure 4.8, the model predicts that the response of the grafted chains to increasing flow is to tilt and elongate in the direction of flow, resulting in an increase in the end-to-end distance of each chain with relatively little change in H (Figure 4.11). 80-, 70 60-| 50 <3i 40-5 3 ° " 20-10-0-D a -Q-—a—H —o—0 0.00 0.01 - • — . • _ _o-0.02- 0.03 Shear Rate (s ) 0.04 Figure 4.11: Dependence on solvent shear rate of the height H (given by r w a n - r b n i s h , where r b m s h is the lattice layer in which <f{t) drops below 0.01 ^waii)) and tilt angle 0 (° relative to surface normal) of a PDMA brush (N = 400, a = 9.7%, Xbo = 0.4) end-grafted inside a 1000 A cylindrical pore. In accordance with experiment, the model therefore predicts no significant change in the hydraulic permeability or the overall volume of the grafted polymer layer with increasing solvent flow over the range of u within which the column can be operated without a significant drop in Np. However, it does predict a significant change in the segment density profile c?5(r) within the brush (Figure 4.12). ... , r , , y.—, ^ r f , 0 10 20 30 40 50 60 70 80 vvall Figure 4.12: Predicted polymer segment density $r w a n - r) profiles for a PDMA brush (N= 400, a = 9.7%, Xbo = 0.4) end-grafted inside a 1000 A cylindrical pore in the absence and presence of a solvent shear rate yr of 0.039 s"1. In the absence of flow, $>) shows a maximum as r decreases away from the grafting surface at r = r w a i i , followed by a region of decreasing density that terminates in a smooth tail region. This predicted density profile closely matches those measured by atomic force microscopy and neutron scattering experiments for polymer brushes grafted to planar surfaces [85;174-176], with the exception that the model predicts that grafting inside a cylindrical pore slightly compresses the extended tail region of the brush due to the decreasing solvent volume available to the chains as r approaches zero. In the presence of a solvent flow generated by setting F* to give a yr of 0.039 s"1 (corresponding to the highest linear velocity explored in this study), <fif) is predicted to adopt a more parabolic profile with a much sharper end-point to the brush. The graft therefore balances the elastic deformation force against the viscous force due to the shear flow by decreasing £ and increasing the end-to-end distance of each chain such that ^r) becomes more uniform across the entire volume of the brush. The brush tilts away from normal to the grafting surface at an average angle 0 of 14.7°±0.4°, computed from the average displacement (in the +z,r direction) of the free chain end, resulting in relatively little change in H despite the elongation of the grafted chains. Although the average end-to-end distance of the chains increases under shear, it remains far below the persistence length of the fully stretched chain. Greater stretching of the chains, and therefore an increase in H, in response to the viscous force is prevented by the fact that the loss in chain configurational entropy is proportional to AH2 [25]. As a result, the model predicts that the grafted chains retain considerable configurational entropy in the presence of a relatively weak viscous shear. This leads to a physical picture for the grafted layer under shear of a lawn of tilted and partially ordered (i.e., more rigid) chains whose configurational entropy has decreased relative to the condition of zero flow. The model results for the hydrodynamic response of the grafted polymer layer to solvent flow therefore provide a means of understanding the dependence of K\ on u observed with the HGD EIC column (Figure 4.8) and not observed in the low graft density EIC columns [14]. As noted in the introduction, solute partitioning into the grafted polymer layer of the EIC column is entropically controlled with the total decrease in entropy AS° for a given partitioning process equal to AScbrush+AS., where both AScbrush and AS^ take on negative values. In the presence of flow, the loss in &Scbrmh is reduced because the elongated state of the chains limits their change in configurational entropy due to solute partitioning. This effect on its own would predict that K\ should increase with increasing u, opposite to the observed trend. It is, however, opposed by the more rigid architecture of the stretched polymer layer, which promotes a larger decrease in AS- (the partitioning solute loses more entropy when the flexibility of the grafted chains is decreased) in a manner analogous to the separation mechanism of traditional SEC. In the two LGD EIC columns for which Dl (R^J is approximately 2, these two opposing effects are roughly equal in magnitude, leading to no observed change in K\ with u. The extent of change in AS^ with u becomes dominant in the HGD column due to the close spacing of the stretched chains that severely limits the degrees of freedom available to the partitioning solute. 4.4. Summary First-generation EIC resins prepared by Ce(IV)-initiated radical polymerization provide good size-based separation of protein mixtures over a fairly broad range of protein molecular weights. Application of the DMA-based surface-initiated ATRP chemistry described in this work allows for synthesis of novel giga-porous EIC resins bearing dense polymer brushes of low polydispersity. These HGD EIC matrices completely exclude high molecular weight solutes (> 10 kDa) while permitting effective size-based discrimination of lower molecular weight solutes. If properly optimized, HGD media, which allows flow through the giga-porous stationary phase, could provide an attractive method for preparative-scale desalting of macromolecular products. The objective of this work however was not to identify an optimal HGD-EIC column for desalting and buffer exchange, although it remains an important long-term goal. Instead, refinements offered by surface-initiated ATRP and ester anchoring-bond saponification were exploited to explore the dependence of EIC column performance on the density and average chain length of the end-grafted polymer layer. The results show that the selectivity curve offered by a given EIC column depends on cr and M n in predictable and controllable ways. While HGD EIC resins permit fractionation of low-molecular analytes, media produced with grafted mushrooms permit solute partitioning over a specific range of solute molecular weights. Hydrodynamics are also shown to affect separation performance in a manner that can be explained by a model that subjects a polymer brush grafted inside a cylindrical pore to a radially dependent shear force. The combination of surface-initiated ATRP chemistry, which permits facile control of the properties of the grafted polymer layer, with the model of the EIC separation process therefore offers the potential to rapidly design and synthesize custom EIC resins for highly efficient size-based purification and processing of macromolecular products. 5. CONCLUSIONS 5.1. Summary and Novelty This thesis described the development of a method for synthesizing EIC matrices by surface initiated ATRP. Two approaches were taken. A concept study using non-porous particles showed that ATRP could be adapted to produce EIC media. Application of this research to porous chromatography supports was used to produce high throughput media. The novelty of research consists of the following: • Reaction conditions were used in surface initiated ATRP to grow polymer brushes of a desired Mn from the surfaces of PSL in the range of 2-3 urn. The necessity of charge to facilitate high density grafting for PDMA from surface initiated ATRP was demonstrated. A method for surface modification of a high throughput chromatography matrix to incorporate a desired amount of sulfate charge and initiators to enable high density graft polymerization was developed. • Degrafting strategies to completely characterize EIC media for explicit determinations of M„, MJMn were used. • Evaluation of EIC columns demonstrated new insights into how the properties of the grafted layer affect separation parameters. It was shown that tuning the graft molecular weight and density of an EIC column by changing the ATRP reaction conditions changed the selectivity curve and allowed for different modes of separation to be selected thus supporting the hypothesis of the thesis. 5.2. Applications of the Research Custom chromatography is an emerging field in industrial separation of biomolecules. As the cost of processing and validation increases, large pharmaceutical companies may be more willing to have media specially designed to solve a complex separation problem in a single step rather than validating additional downstream separation stages. As developed here, the synthesis method for producing EIC media is highly versatile in that the slope and linear range of the selectivity curve can be changed relatively easily by modifying the ATRP reaction conditions, opening up new possibilities for high throughput custom size exclusion media. Application of high throughput SEC media is also highly desirable for potential use in multi-dimensional proteomics screening. Use of size-based separations media in one dimension of a multi-dimensional liquid chromatography array could enable screening of native-state proteins separated by orthogonal strategies. This would be valuable for the native-state analysis of the human blood plasma proteome. Size fractionation of blood plasma proteins has been a useful strategy for purification. However, little is known about very small blood plasma proteins and polypeptides. Work described in Chapter 4 showed that the HGD column had a relatively steep selectivity curve in the region < 10 kDa, which would facilitate small protein separations - an area relatively poorly characterized by existing separation strategies. 5.3. Outlook Controlled surface initiated polymerization will continue to be a developing field of polymer chemistry for many years to come because of research into new materials produced from innovative polymerization strategies. New arrays of grafted polymer phases for EIC could be prepared by ATRP of other hydrophilic monomers such as M E A , M M A , and H E M A to name a few. Furthermore, hybrid materials could be synthesized by creating block copolymers. Temperature responsive polymers such as PNIPAM, utilized as grafts for EIC media, may facilitate unique modes of solute separations. Lastly, desalting applications for the HGD media shown in Chapter 4 could be optimized for the preparative-scale. Development of EIC media for commercial applications would almost certainly require that the media be stable in pH ranges comparable to other commercial products (pH 2-12) and therefore efforts would be needed to use synthesis techniques for initiator attachment on a non-labile linkage and retune the ATRP chemistry. However, this presents a challenge since little is known about how the ester linkage influences surface initiated ATRP of D M A . Therefore, mass balance and new analytical methods for probing the surface (AFM) would have to be calibrated with existing EIC media and applied towards non-labile EIC media to gain insight into the properties of the grafted layer. REFERENCES [1] N . L. Anderson and N . G. Anderson, Mol. Cell. Proteomics 1 (2002) 845. [2] E. W. Deutsch, J. K. Eng, H. Zhang, N . L. King, A . I. Nesvizhskii, B. Y . Lin, H. K. Lee, E. C. Y i , R. Ossola, and R. Aebersold, Proteomics 5 (2005) 3497. [3] J. Porath and P. Flodin, Nature (Lond) 183 (1959) 1657. [4] H. Goetz, M . Kuschel, T. Wulff, C. Sauber, C. Miller, S. Fisher, and C. Woodward, J. Biochem. Biophys. Methods 60 (2004) 281. [5] J. Barackman, I. Prado, C. Karunatilake, and K. Furuya, J. Chromatog. A 1043 (2004) 57. [6] K. M . Gooding and F. E. Regnier, HPLC of biological macromolecules. Marcel-Dekker, New York, 1990. [7] H. G. Barth, B. E. Boyes, and C. Jackson, Anal. Chem. 66 (1994) R595-R620. [8] C.-S. Wu, Column Handbook for Size Exclusion Chromatography. Academic Press, New York, 1999. [9] D. van Roosmalen, M . J. Lazzara, L. J. P. van den Broeke, J. T. F. Keurentjes, and D. Blankschtein, Biotechnol. Bioeng. 87 (2004) 695. [10] D. A. Horneman, M . Ottens, J. T. F. Keurentjes, and L. A . M . van der Wielen, J. Chromatog. A 1113 (2006) 130. [11] L. J. P. van den Broeke, D. van Roosmalen, M . P. J. Dohmen-Speelmans, C. H. J. T. Dietz, L. A . M . van der Wielen, and J. T. F. Keurentjes, Biotechnol. Bioeng. 93 (2006) 355. [12] R. D. Haynes, R. J. Meagher, J. I. Won, F. M . Bogdan, and A. E. Barron, Bioconjugate Chemistry 16 (2005) 929. [13] D. E. Brooks, C. A. Haynes, D. Hritcu, B. M . Steels, and W. Miiller, Proc. Natl. Acad. Sci. U . S. A. 97 (2000) 7064. [14] P. Pang, J. Koska, B. R. Coad, D. E. Brooks, and C. A. Haynes, Biotechnol. Bioeng. 90 (2005) 1. [15] J. S. Wang and K. Matyjaszewski, Macromol. 28 (1995) 7901. [16] S. G. Gaynor, J. Qiu, and K. Matyjaszewski, Macromol. 31 (1998) 5951. [17] K. Matyjaszewski, Y . Nakagawa, and C. B. Jasieczek, Macromol. 31 (1998) 1535. [18] X . S. Wang and S. P. Armes, Macromol. 33 (2000) 6640. [19] K. Matyjaszewski and J. H. Xia, Chemical Reviews 101 (2001) 2921. [20] K. Matyjaszewski, P. J. Miller, N . Shukla, B. Immaraporon, A. Gelman, B. B. Luokala, T. M . Siclovan, G. Kickelbick, T. Valiant, H. Hoffmann, and T. Pakula, Macromol. 32(1999) 8716. [21] K. N . Jayachandran, A. Takacs-Cox, and D. E. Brooks, Macromol. 35 (2002) 4247. [22] J. N . Kizhakkedathu and D. E. Brooks, Macromol. 36 (2003) 591. [23] W.Kuhn,Kolloid(1934)2. [24] P. Flory, Principals of Polymer Chemistry. Cornell University Press, Ithaca, NY, 1953. [25] G. J. Fleer, M . A. Cohen Stuart, J. M . H. M . Scheutjens, T. Cosgrove, and B. Vincent, Polymers at Interfaces. Chapman & Hall, London, 1993. [26] P. G. de Gennes, Macromol. 13 (1980) 1069. [27] S. Alexander, J. Phys. (Paris) 38 (1977) 983. [28] P. G. de Gennes, J. Phys. (Paris) 37 (1976) 1443. [29] J. L. Anderson and J. O. Kim, J. Chem. Phys. 86 (1987) 5163. [30] S. T. Milner, Science 251 (1991) 905. [31] J. Satulovsky, M . A. Carignano, and I. Szleifer, Proc. Nat. Acad. Sci. 97 (2000) 9037. [32] S. J. Sofia, V. Premnath, and W. Merrill, Macromol. 31 (1998) 5059. [33] J. Janzen, Y . Le, J. N . Kizhakkedathu, and D. E. Brooks, J. Biomater. Sci. Polymer Edn 15 (2004) 1121. [34] C. Yoshikawa, A. Goto, Y . Tsujii, T. Fukuda, T. Kimura, K. Yamamoto, and A. Kishida, Macromol. (2006). [35] E. Kulik and Y. Ikada, J. Biomed. Mater. Res. 30 (1996) 295. [36] G. R. Llanos and M . V. Sefton, J. Biomed. Mater. Res. 27 (1993) 1383. [37] G. R. Llanos and M . V . Sefton, J. Biomater. Sci. Polymer Edn 4 (1993) 381. [38] W. R. Gombotz, W. Guanghui, T. A . Horbett, and A. S. Hoffman, in J.M.Harris (Editor), Poly(Ethylene Glycol) Chemistry: Biotechnical and Biomedical Applications. Plenum Press, New York, 1992, p. 247. [39] C. Golander, J. N . Herron, K. Lim, P. Claesson, P. Stenius, and J. D. Andrade, in J.M.Harris (Editor), Poly(Ethylene Glycol) Chemistry: Biotechnical and Biomedical Applications. Plenum Press, New York, 1992, p. 221. [40] K. Bergstrom, K. Holmberg, A. Safranj, A. S. Hoffman, M . J. Edgell, A. Kozlowski, B. A. Hovanes, and J. M . Harris, J. Biomed. Mater. Res. 26 (1992) 779. [41] J. H. van Zanten, Macromol. 27 (1994) 6797. [42] M . A. Rounds, W. Kopaciewicz, and F. E. Regnier, J. Chromatogr. 362 (1986) 187. [43] T. Kawai, K. Saito, and W. Lee, J. Chromatog. B 790 (2003) 131. [44] F. Gasparrini, D. Misiti, R. Rompietti, and C. Villani, J. Chromatog. A 1064 (2005) 25. [45] X . Y . Huang, L. J. Doneski, and M . J. Wirth, Anal. Chem. 70 (1998) 4023. [46] M . D. Miller, G. L. Baker, and M . L. Bruening, J. Chromatog. A 1044 (2004) 323. [47] F. C. Leinweber, J. Stein, and M . Otto, Fresenius Journal of Analytical Chemistry 370 (2001)781. [48] A . Feldmann, U. Claussnitzer, and M . Otto, J. Chromatog. B 803 (2004) 149. [49] X . Santarelli, D. Muller, and J. Jozefonvicz, J. Chromatogr. 443 (1988) 55. [50] Y. Cohen, R. S. Faibish, and M . Rovira-Bru, in E.Pefferkora (Editor), Interfacial Phenomena in Chromatography. Marcel Dekker, Inc., New York, 1999, p. 263. [51] J. G. Heffernan and D. C. Sherrington, Journal of Applied Polymer Science 29 (1984) 3013. [52] M . Petro and D. Berek, Chromatographia 37 (1993) 549. [53] Y . Cohen, P. Eisenberg, and M . Chaimberg, J. Colloid Interface Sci. 148 (1992) 579. [54] I. N . Savina, I. Y. Galaev, and B. Mattiasson, J. Chromatog. A 1092 (2005) 199. [55] W. Muller, European Journal of Biochemistry 155 (1986) 213. [56] L . R. Jacob and L. Britsch, in C.-S.Wu (Editor), Column Handbook for Size Exclusion Chromatography. Academic Press, San Diego, 1999, p. 219. [57] W. P. Olson, Process Biochemistry 18 (1983) 29. [58] R. Ghidossi, D. Veyret, and P. Moulin, Chemical Engineering and Processing 45 (2006) 437. [59] J. F. van Staden, Fresenius Journal of Analytical Chemistry 352 (1995) 271. [60] F. Zokaee, T. Kaghazchi, and A. Zare, Process Biochemistry 34 (1999) 803. [61] P. Gatenholm, C. J. Fell, and A. G. Fane, Desalination 70 (1988) 363. [62] R. Ghosh and Z. F. Cui, Journal of Membrane Science 167 (2000) 47. [63] K. Sakhamuru, D. W. Hough, and J. B. Chaudhuri, Biotechnology Progress 16 (2000) 296. [64] L. O. Lindquist and K. W. Williams, Dairy Industries International 38 (1973) 459. [65] J. M . Curling, J. Berglof, L. O. Lindquist, and S. Eriksson, Vox Sanguinis 33 (1977) 97. [66] M . E. Himmel, J. O. Baker, and D. J. Mitchell, Chromatographic Science Series 69 (1995)409. [67] L. Hagel, in M.Potschka and P.L.Dubin (Editors), Strategies in Size Exclusion Chromatography. American Chemical Society, Washington, DC, 1996, p. 211. [68] L. Hagel, in J.-C.Janson and L.Ryden (Editors), Protein Purification: Principles, High-Resolution Methods, and Applications. John Wiley and Sons, New York, 1998, p. 79. [69] A . G. Ogston, Trans. Faraday. Soc. 54 (1958) 1754. [70] P. G. Squire, Arch. Biochem. Biophys. 107 (1964) 471. [71] W. W. Yau, J. J. Kirkland, and D. D. Bly, Modern Size Exclusion Chromatography. Wiley Interscience, New York, 2006. [72] F. Dondi, A. Cavazzini, M . Remelli, A . Felinger, and M . Martin, J. Chromatog. A 943 (2002) 185. [73] D. E. Brooks and W. Miiller, J. Mol. Recognit. 9 (1996) 697. [74] A. J. P. Martin and R. L. M . Synge, Biochemistry Journal 35 (1941) 1358. [75] J. J. van Deemter, F. J. Zuiderweg, and A. Klinkenberg, Chem. Eng. Sci. 5 (1956) 271. [76] P. A. Bristow and J. H. Knox, Chromatographia 10 (1977) 279. [77] M . E. Himmel and P. G. Squire, in P.L.Dubin (Editor), Aqueous Size-Exclusion Chromatography. Elsevier, Amsterdam, 1988, p. 3. [78] J. Wen, T. Arakawa, and J. S. Philo, Anal. Biochem. 240 (1996) 155. [79] X . Guo, M . Condra, K. Kimura, G. Berth, H. Dautzenberg, and P. L. Dubin, Anal. Biochem. 312(2003)33. [80] B. M . Steels, J. Koska, and C. A. Haynes, J. Chromatog. B 743 (2000) 41. [81] J. M . H. M . Scheutjens and G. J. Fleer, J. Phys. Chem. 83 (1979) 1619. [82] J. M . H. M . Scheutjens and G. J. Fleer, J. Phys. Chem. 84 (1980) 178. [83] T. Cosgrove, T. Heath, B. van Lent, F. A . M . Leermakers, and J. M . H. M . Scheutjens, Macromol. 20 (1987) 1692. [84] F. A . M . Leermakers, J. M . H. M . Scheutjens, and J. Lyklema, Biochim. Biophys. Acta 1024 (1990) 139. [85] T. Cosgrove and K. Ryan, Langmuir 6 (1990) 136. [86] Method and Support Material for Gel-Permeation Chromatography. Miiller, W. [WO 94/26379]. 1994. International Patent. 1994. [87] B. M . Steels, F. A. M . Leermakers, and C. A. Haynes, J. Chromatog. B 743 (2000) 31. [88] B. Zhao and W. J. Brittain, Prog. Polym. Sci. 25 (2000) 677. [89] C. Yoshikawa, A. Goto, Y. Tsujii, T. Fukuda, K. Yamamoto, and A. Kishida, Macromol. 38 (2005) 4604. [90] S. G. Boyes, A. M . Granville, M . Baum, B. Akgun, B. K. Mirous, and W. J. Brittain, Surface Science 570 (2004) 1. [91] J. S. Wang and K. Matyjaszewski, J. Am. Chem. Soc. 117 (1995) 5614. [92] M . Freemantle, Chem. Eng. News 80 (2002) 36. [93] K. Matyjaszewski, in K.Matyjaszewski (Editor), Controlled Radical Polymerization. American Chemical Society, Washington, DC, 1998, p. 258. [94] K. Matyjaszewski, in K.Matyjaszewski (Editor), Controlled Radical Polymerization. American Chemical Society, Washington, DC, 1998, p. 2. [95] G. Moineau, M . Minet, P. Dubois, P. Teyssie, T. Senninger, and R. Jerome, Macromol. 32 (1999) 27. [96] S. M . Zhu and D. Y . Yan, Macromol. 33 (2000) 8233. [97] R. M . Johnson, P. S. Corbin, C. Ng, and C. L. Fraser, Macromol. 33 (2000) 7404. [98] T. Nishikawa, M . Kamigaito, and M . Sawamoto, Macromol. 32 (1999) 2204. [99] S. Coca, C. B. Jasieczek, K. L. Beers, and K. Matyjaszewski, J. Polym. Sci. ,Part A: Polym. Chem. 36 (1998) 1417. [100] K. Zhang, H. T. L i , H. W. Zhang, S. Zhao, D. Wang, and J. Y . Wang, Materials Chemistry and Physics 96 (2006) 477. [101] J. T. Rademacher, M . Baum, M . E. Pallack, and W. J. Brittain, Macromol. 33 (2000) 284. [102] M . Teodorescu and K. Matyjaszewski, Macromol. 32 (1999) 4826. [103] X . Huang and M . J. Wirth, Macromol. 32 (1999) 1694. [104] M . Hua, T. Kanek, X . Y. Liu, M . Q. Chen, and M . Akashi, Polymer Journal 37 (2005) 59. [105] X . C. Yin and H. D. H. Stover, Macromol. 38 (2005) 2109. 106] J. Pyun, T. Kowalewski, and K. Matyjaszewski, Macromol. Rapid Commun. 24 (2003) 1043. [107] M . Ejaz, Y . Tsujii, and T. Fukuda, Polymer 42 (2001) 6811. [108] M . Husseman, E. E. Malmstrom, M . McNamara, M . Mate, D. Mecerreyes, D. G. Benoit, J. L. Hedrick, P. Mansky, E. Huang, T. P. Russell, and C. J. Hawker, Macromol. 32 (1999) 1424. 109] C. Perruchot, M . A. Khan, A . Kamitsi, S. P. Amies, T. von Werae, and T. E. Patten, Langmuir 17(2001)4479. [110] J. B. Kim, W. X . Huang, M . D. Miller, G. L. Baker, and M . L. Bruening, J. Polym. Sci. ,Part A: Polym. Chem. 41 (2003) 386. "111] J. Pyun, S. J. Jia, T. Kowalewski, G. D. Patterson, and K. Matyjaszewski, Macromol. 36 (2003) 5094. [112] J. N . Kizhakkedathu, K. R. Kumar, D. Goodman, and D. E. Brooks, Polymer 45 (2004) 7471. [113] J. N . Kizhakkedathu, R. Norris-Jones, and D. E. Brooks, Macromol. 37 (2004) 734. [114] K. K. Chittur, Biomaterials 19 (1998) 301. 115] D. Goodman, J. N . Kizhakkedathu, and D. E. Brooks, Langmuir 20 (2004) 6238. [116] M . Chaimberg and Y. Cohen, Ind. Eng. Chem. Res. 30 (1991) 2534. [117] M . Khan and W. T. S. Huck, Macromol. 36 (2003) 5088. [118] P. Liu, J. Tian, W. M . Liu, and Q. J. Xue, Materials Research Innovations 7 (2003) 105. [119] D. Hritcu, W. Muller, and D. E. Brooks, Macromol. 32 (1999) 565. [120] K. N . Jayachandran and P. R. Chatterji, Eur. Polym. J. 36 (2000) 743. [121] P. J. Wyatt, Analytica Chimica Acta 272 (1993) 1. [122] D. Hritcu. Synthesis and Characterization of Cationic Latex Bearing Grafted Chains. 1-176.1998. Thesis. The University of British Columbia. [123] J. I. Kroschwitz, in J.I.Kroschwitz (Editor), Encyclopedia of polymer science and engineering, Volume 6. Wiley, New York, 1986, p. 14. [124] S. Kerrigan and D. E. Brooks, Bioconjugate Chemistry 9 (1998) 596. 125] M . S. Hazell, R. Jones, P. E. Luffman, and R. F. Sewell, Measurement Science & Technology 11 (2000)227. [126] P. J. Scott, A. Penlidis, and G. L. Rempel, J. Polym. Sci. ,Part A: Polym. Chem. 31 (1993) 403. "127] K. Tauer, R. Deckwer, I. Kuhn, and C. Schellenberg, Colloid Polym. Sci. 277 (1999) 607. [128] M . S. El-Aasser, in G.W.Poehlein, R.H.Ottewill, and J.W.Goodwin (Editors), Science and Techonolgy of Polymer Colloids. Nijhoff, Boston, 1983, p. 422. [129] J. W. Goodwin, J. Hearn, C. C. Ho, and R. H. Ottewill, Br. Polym. J. 5 (1973) 374. [130] J. W. Goodwin, J. Hearn, C. C. Ho, and R. H. Ottewill, Colloid Polym. Sci. 252 (1974)464. [131] J. Stone-Masui and A. Watillon, J. Colloid Interface Sci. 52 (1975) 479. [132] P. J. Feeney, D. H. Napper, and R. G. Gilbert, Macromol. 20 (1987) 2922. [133] A. Eshuis, H. J. Leendertse, and D. Thoenes, Colloid Polym. Sci. 269 (1991) 1086. [134] G. Tuin, A . C. I. A . Peters, A. J. G. van Diemen, and H. N . Stein, J. Colloid Interface Sci. 158 (1993) 508. [135] G. Tuin and H. N . Stein, Langmuir 10 (1994) 1054. [136] I. Kuhn and K. Tauer, Macromol. 28 (1995) 8122. [137] J. Ugelstad, A. Berge, T. Ellingsen, R. Schmid, T. N . Nilsen, P. C. Mork, P. Stenstad, E. Homes, and O. Olsvik, Prog. Polym. Sci. 17 (1992) 87. [138] F. K. Hansen and J. Ugelstad, J. Polym. Sci. ,Part A: Polym. Chem. 17 (1979) 3033. [139] F. K. Hansen and J. Ugelstad, J. Polym. Sci. .Part A: Polym. Chem. 17 (1979) 3069. 140] D. Zou, L. Sun, J. J. Aklonis, and R. Salovey, J. Polym. Sci . , Part A : Polym. Chem. 30 (2002) 1463. 141] D. Z. Zou, L. Q. Sun, J. J. Aklonis, and R. Salovey, J. Polym. Sci. ,Part A: Polym. Chem. 30 (1992) 1463. [142] S.-E. Shim, Y.-J . Cha, J.-M. Byun, and S. Choe, J. of Appl. Polym. Sci. 71 (1999) 2259. [143] F. K. Hansen and J. Ugelstad, J. Polym. Sci. ,Part A : Polym. Chem. 17 (1979) 3047. [144] S. H. Tan, Delftsche U.M. , Delft, 1967, p. 7. [145] G. V. I. Seaman, Latex Handling and Processing (Lecture Series). Princeton, NJ, 1996. [146] K. W. Dixon, in J.Brandrup, E.H.Immergut, and E.A.Gulke (Editors), Polymer Handbook. John Wiley & Sons, New York, 1999, p. 1. [147] B. Zhao and W. J. Brittain, Macromol. 33 (2000) 8813. [148] D. M . Jones, A. A. Brown, and W. T. S. Huck, Langmuir 18 (2002) 1265. [149] T. von Werne and T. E. Patten, J. Am. Chem. Soc. 123 (2001) 7497. [150] J. N . Kizhakkedathu, personal communication, 2006. [151] X . M . He and D. C. Carter, Nature 358 (1992) 209. [152] Chromatographic Process Media Catalog. 2004. Montgomeryville, PA, Tosoh Bioscience. [153] Evaluation Methods for and Specifications of Toyopearl® AF-Amino-650. Version 1.10. 1999. Grove City, OH, Tosoh Bioscience. [154] W. Norde and J. Lyklema, J. Colloid Interface Sci. 66 (1978) 257. [155] H. J. Vanenckevort, D. V . Dass, and A. G. Langdon, J. Colloid Interface Sci. 98 (1984) 138. [156] R. Narain and S. P. Amies, Macromol. 36 (2003) 4675. [157] K. K. Chittur, Biomaterials 19 (1998) 357. [158] P. W. Atkins, Physical Chemistry. W. H. Freeman and Company, New York, 1994. [159] P. G. de Gennes, Scaling concepts in polymer physics. Cornell University Press, Ithaca, NY, 1979. [160] Y . Rabin and S. Alexander, Europhys. Lett. 13 (1990) 49. [161] B. Ravindranath, Principals and Practice of Chromatography. Ellis Horwood Ltd., UK, 1989. [162] P. A. Bristow, Liquid Chromatography in Practice. Wiley Press, UK, 1976. [163] D. M . Ruthven, Principles of Adsorption and Adsorption Processes. John Wiley and Sons, New York, 1984. [164] M . R. Ladisch, Bioseparations Engineering: Principles, Practice and Economics. Wiley Inerscience, New York, 2001. [165] E. Grushka, J. H. Knox, and L. R. Snyder, Journal of Chromatographic Science 13 (1975) 25. [166] J. H. Knox, J. Chromatog. A 960 (2002) 7. [167] V. Kumaran, Macromol. 26 (1993) 2464. [168] S. C. Ji and J. D. Ding, J. Chem. Phys. 123 (2005) 1449041. [169] P. Pincus, Macromol. 9 (1976) 386. [170] R. S. Parnas and Y . Cohen, Macromol. 24 (1991) 4646. [171] R. P. Castro, H. G. Monbouquette, and Y. Cohen, Journal of Membrane Science 179 (2000) 207. [172] Y . Cohen, Macromol. 21 (1988) 494. [173] F. T. Sarfert and M . R. Etzel, J. Chromatog. A 764 (1997) 3. [174] R. Baranowski and M . D. Whitmore, J. Chem. Phys. 103 (1995) 2343. [175] J. B. Field, C. Toprakcioglu, R. C. Ball, H. B. Stanley, L. Dai, W. Barford, J. Penfold, G. Smith, and W. Hamilton, Macromol. 25 (1992) 434. [176] M . C. Guffond, D. R. M . Williams, and E. M . Sevick, Langmuir 13 (1997) 5691. [177] D. P. Shoemaker, C. W. Garland, and J. W. Nibler, Experiments in Physical Chemistry. McGraw-Hill, Inc., New York, 1989. APPENDIX 1: STATISTICAL TREATMENT OF DATA Uncertainty in Replicate Measurements In cases where replicate measurements have been made of the same quantity, the uncertainty, A,was expressed from the Student t distribution. A = fx S • [1.11] \ ^meas In this equation, t is the critical value referenced from a table of critical values, S is the standard deviation of the data and Nmeas is the number of measurements. In the table of critical values (Table A l . l ) , t is chosen from the degrees of freedom v, where v=N- 1, and the probability is chosen as 0.95. Thus, the uncertainty in a mean value is expected to fall within the given range of values 95% of the time. Rejection of Discordant Data During replicate measurements, if a single measurement appeared to differ significantly from the mean, the Critical Q test was applied as a basis for statistical rejection of the data point. \(V )-(V )\ Q |V suspect' V closest / | ^ \2~\ where Vsuspect is the suspected outlier, Vciosest is the value closest to it, Vnig„ and V/ow refer to the highest and lowest values in the series. The value for Q was then compared against a table of critical Q (Qc) values (Table A1.2) for the number of samples in the set. If Q was equal to or larger than Qc, the measurement was rejected. Tables were adapted from published sources [177]. Table A l . l : Critical values of t. v P = .95 1 12.7 2 4.30 3 3.18 4 2.78 5 2.57 6 2.45 7 2.36 8 2.31 9 2.26 10 2.23 15 2.13 20 2.09 30 2.04 CO 1.96 Table A1.2: Critical Q Values. N 3 4 5 6 7 8 9 10 Qc 0.94 0.76 0.64 0.56 0.51 0.47 0.44 0.41 APPENDIX 2: ATRP REACTION CONDITIONS FOR NON-POROUS MATERIALS Table A2.3: ATRP reaction conditions for shell latex in Chapter 2. Lot Graphical Symbol Latex Suspension Suspension Cone. used in text g % (w/w) M5 + 3.0 3.0 M8 X 3.5 3.0 M9 • 3.0 3.0 M10 • 3.5 3.0 L3 • 3.5 3.0 L4 n/a 16.7 9.0 L5 n/a 16.7 9.0 L8 n/a 16.7 9.0 For all reactions: Cu(I) molar amount was 12 times the initiator, Cu(II) molar amount was 0.15 times Cu(I), Cu(0) molar amount was 8 times Cu(II), HMTETA molar amount was the sum of Cu(I) and Cu(II), Brij-35 was 0.16% (w/w) of the latex suspension. APPENDIX 3: PROBES USED FOR SIZE EXCLUSION CHROMATOGRAPHY Table A3.4: Probes used for evaluating size-exclusion properties of EIC columns. Analyte Mw Final Supplier Notes Cone. Da (mg/ml) Potassium Nitrate 101 0.1 S Cytidine 243 0.08 S 99% Synthetic Peptide 5 964 1.0 I Vitamin B12 1355 0.01 S Approx. 99% Synthetic Peptide 3 1419 0.9 I Bacitracin 1423 2.5 S 50000 units/g Synthetic Peptide 1 1800 3.1 I Synthetic Peptide 2 2743 3.2 I Poly-l-lysine.HBr 1 3970 3.0 s Insulin f 5700 2.0 s Bovine Pancreas, 28.3 U S P units/mg Aprotinin 6500 1.4 s Bovine Lung Synthetic Peptide 4 8369 1.0 I Lysozyme 12200 2.3 s Chicken Egg White a-Lactalbumin 14400 1.8 s Type I, > 85% P A G E Myoglobin 17600 1.8 s Horse Skeletal Muscle, 95 -100% Trypsin Inhibitor 40200 3.7 s Type II: Turkey Egg White Analyte Mw Da Final Cone. (mg/ml) Supplier Notes Ovalbumin 43000 2.9 S Chicken Egg White, Grade VII Human Serum Albumin 66300 1.3 S 99%, globulin free Bovine Serum Albumin 67000 3.3 S 98% Monomer Alcohol Dehydrogenase 150000 1.9 S Yeast. For Gel Filtration. Immunoglobulin G 150000 1.1 S human - reagent grade from serum >= 95% P-Amylase 200000 1.6 S Sweet Potato. For Gel Filtration Fibrinogen 341000 1.1 S Human Apoferritin 443000 1.2 S Horse Spleen. Supplied in 50% Glycerc 25 mg/ml Thyroglobulin 669000 2.7 S Blue Dextran 2000000 2.4 S For Gel Filtration Suppliers: I, In-house standard; S, Sigma-Aldrich. t Insulin prepared in 10 mM HC1 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0061138/manifest

Comment

Related Items