Polyelectrolyte Multilayers

Home , Seashell

POLYELECTROLYTE MULTILAYERS:

SIMULATIONS, EXPERIMENTS, AND APPLICATIONS IN

BIOMINERALIZATION

PRITESH ARJUNBHAI PATEL

Submitted in partial fulfillment of the requirements

for the degree of Doctor of Philosophy

Department of Macromolecular Science and Engineering

CASE WESTERN RESERVE UNIVERSITY

January 2008 CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

______

candidate for the Ph.D. degree *.

(signed)______(chair of the committee)

______

(date) ______

*We also certify that written approval has been obtained for any proprietary material contained therein.

iii

To my Father,

Arjun R. Patel

--who inspires me to achieve the impossible--

Table of Contents

Table of Contents...... 1

List of Tables...... 7

List of Figures ...... 8

List of Schemes ...... 16

Acknowledgements...... 18

CHAPTER 1 ...... 21

1 Introduction...... 21

1.1 Layer-by-Layer ‘Dipping’ Assembly (part I) ...... 21

1.1.1 Characterization Techniques:...... 23

1.1.2 Multilayer Structure, Properties and Growth Mechanisms...... 24

1.1.3 Research Significance - I ...... 28

1.2 Polyelectrolyte Spin Assembly (part II) ...... 29

1.2.1 Prior Studies...... 30

1.2.2 Scaling Model for PSA growth...... 32

1.2.3 Research Significance...... 37

1.3 Biomineralization (part III)...... 38

1.3.1 Polyelectrolyte Multilayers for Biomineralization ...... 39

1.3.2 Hydroxyapatite Formation...... 41

1.3.3 Silica Biomineralization...... 44

1.4 Thesis Outline...... 53

CHAPTER 2 ...... 62

2 Molecular Dynamics Simulations of Layer-by-Layer Assembly of Polyelectrolyte ...... 62

2.1 Synopsis ...... 62

2.2 Introduction...... 63

2.3 Model and Simulation Method ...... 63

2.4 Results...... 68

2.4.1 Formation of Multilayers ...... 68

2.4.2 Effect of Charge Density of Polyelectrolytes ...... 71

2.4.3 Effect of Chain Degree of Polymerization...... 75

2.4.4 Theoretical Model of Multilayer Formation...... 78

2.5 Discussion and Conclusion...... 80

CHAPTER 3 ...... 96

3 Effect of Electrostatic and Short Range Interactions on the

Build-Up of Polyelectrolyte Multilayers ...... 96

3.1 Synopsis ...... 96

3.2 Introduction...... 97

3.3 Simulation Methods and Interaction Parameters:...... 97

3.4 Results and Discussions...... 99

3.4.1 Growth of Polymer Surface Coverage...... 99

3.4.2 Distribution of Polymer Density...... 100

3.4.3 Universality of the Film Growth...... 104

3.4.4 Stability of the Growing Film and Chain Exchange...... 105

3.5 Scaling Model...... 107

3.5.1 Average Multilayer Density...... 107

3.5.2 Charge Density Oscillations ...... 109

3.5.3 Multilayer Growth...... 110

3.5.4 Chain Desorption...... 112

3.6 Conclusions...... 114

CHAPTER 4 ...... 125

4 Combined Effect of Spin Speed and Ionic Strength on

Polyelectrolyte Spin Assembly...... 125

4.1 Synopsis ...... 125

4.2 Introduction...... 126

4.3 Experimental Procedure...... 126

4.3.1 Preparation of Multilayers...... 126

4.3.2 Characterization ...... 128

4.3.3 Atomic Force Microscopy (AFM)...... 129

4.4 Results and Discussion...... 131

4.5 Conclusions...... 142

CHAPTER 5 ...... 155

5 Hydroxyapatite and Silica Biomineralization on

Polyelectrolyte Multilayers ...... 155

5.1 Synopsis ...... 155

5.2 Experimental Method...... 157

5.2.1 Hydroxyapatite Formation...... 157

5.2.2 Experimental Methods: Silica Formation ...... 163

5.3 Results and Discussion: Hydroxyapatite formation...... 165

5.3.1 Kinetics of HA growth in Solution ...... 165

5.3.2 Growth of HA on LBL Surfaces...... 169

5.4 Results and Discussion: Silica Formation...... 173

5.5 Conclusions...... 177

CHAPTER 6 ...... 187

6 Silica formation by Poly(ethylene imine)(PEI) in Solution and on Surfaces ...... 187

6.1 Synopsis ...... 187

6.2 Introduction...... 188

6.3 Experimental Procedure...... 188

6.4 Results and Discussion...... 191

6.4.1 Silica formation from PEI in solution...... 191

6.4.2 Effect of PEI and TMOS Concentration on Silica Formation ...... 195

6.4.3 Morphology of Silica ...... 199

6.4.4 Silica formation on Surfaces...... 202

6.5 Conclusions...... 204

CHAPTER 7 ...... 215

7 Silica Formation in Polymer Scaffolds ...... 215

7.1 Synopsis ...... 215

7.2 Introduction...... 216

7.3 Experimental Methods...... 216

7.3.1 Synthesis of Linear Poly(ethyleneimine) (PEI) ...... 216

7.3.2 Electrospinning of PEI/PVP blend...... 217

7.3.3 Silicification and calcination of fibers ...... 218

7.3.4 Preparation of linear PEI foams...... 219

7.3.5 Characterization ...... 220

7.4 Results: Silica formation on electrospun nanofiber scaffolds...... 221

7.5 Results: Silica formation in PEI foams...... 227

7.6 Discussion...... 230

7.7 Conclusions...... 233

CHAPTER 8 ...... 246

8 Conclusions and Future Outlook ...... 246

8.1 Polyelectrolyte Multilayers Simulation (part I) ...... 246

8.1.1 Conclusions...... 246

8.1.2 Future Recommendations...... 248

8.2 Polyelectrolyte Spin Assembly (part II) ...... 253

8.2.1 Conclusions...... 253

8.2.2 Future Recommendations...... 255

8.3 Biomineralization (part III)...... 256

8.3.1 Conclusions...... 256

8.3.2 Future Recommendations...... 259

9 Bibliography ...... 268

List of Tables

Table 3.1: Interaction parameters and system sizes...... 116

Table 3.2: Number of chains added to simulation box during each deposition step .... 117

Table 5.1: Kinetic results of the effect of addition of PGA or PSS to the seeded Hydroxyapatite crystal growth using the constant composition method at total calcium/phosphate =1.67, pH 7.4 and Ionic strength 0.15 M NaCl at 37 °C and pH 7.4...... 178

List of Figures

Figure 2.1: Dependence of the polymer surface coverage (Γσ2) on the number of MD

integration steps for system with Np=32 and f=1 during the first and the second deposition step for duration of 3.0 × 106 integration steps in each deposition step...... 85

Figure 2.2: Evolution of the layer structure during the adsorption of fully charged (f=1) polyelectrolytes, with degree of polymerization Np=32. Each snapshot is taken after the completion of deposition cycles from 1 through 5 with unique color coding for each step being maintained from one snapshot to the next. For example, the blue chain in the 5th step snapshot is polymer adsorbed originally during the 1st step...... 86

Figure 2.3: Density profiles of the negatively ρ-(z) (continuous line) and positively ρ+(z) (dashed line) charged monomers for system with fully charged (f=1) polyelectrolyte chains with Np=32 after completion of (a) 3rd deposition step and (b) 4th deposition step with a duration of 5×105 MD steps each. The density profile of positively (circles) and negatively (triangles) charged counterions are also shown in secondary axis...... 87

Figure 2.4: Topography plots of the film height distribution for the system of fully 5 charged chains with f=1 and degree of polymerization Np= 32 at the end of 5×10 MD steps after the completion of 3rd deposition step (a) and 4th deposition step (b). The insert shows the height distribution of the main plots...... 88

Figure 2.5: Dependence of the surface coverage (Γσ2) on the number of deposition steps

for polyelectrolyte chains of the degree of polymerization Np= 32 with different fraction of charged monomers f=1 (circles) and f=1/2 (triangles) and f=1/3 (squares). The inset shows the dependence of average thickness on number of deposition steps...... 89

Figure 2.6: Density profiles of the fully charged (f=1) (a) and partially charged (f=1/2) (b)

polyelectrolyte chains with Np=32 after completion of 12 deposition cycles with duration of 5×105 MD steps each. Insert show the difference between the corresponding uniaxial monomer densities of positively and negatively charged chains, Δρ(z) = ρ − (z) − ρ + (z) ...... 90

Figure 2.7: Comparison of the density difference of positively and negatively charged

5 monomers, Δρ(z) = ρ − (z) − ρ + (z) for two different lengths of simulation runs. (i) 5×10 MD steps (circles) and (ii) 3× 106 MD steps (continuous line)...... 91

Figure 2.8: Ion pair correlation functions for chains with different fractions of charged f=1(closed symbols) and f=1/2 (open symbols) and degrees of polymerization, Np = 32

(circle), Np =16 (triangle) and Np =8 (square). The vertical reference lines show the peak positions of the perfectly stratified molecular layers of positively and negatively charged ions...... 92

Figure 2.9: Dependence of the surface coverage (Γσ2) on the number of deposition steps for polyelectrolyte chains with different fraction of charged monomers and chain degree of polymerization...... 93

Figure 2.10: Comparison of density difference of positively and negatively charged

monomers, Δρ(z) = ρ − (z) − ρ + (z) for chains degree of polymerization Np= 8 and different fraction of charge monomers (f = 1 and 1/2)...... 94

Figure 2.11: Density profiles of the negatively ρ-(z) (continuous line) and positively ρ+(z) (dashed line) charged monomers for system of fully charged (f=1) polyelectrolyte chains th 5 with Np=32 after completion of 12 deposition step with duration 5×10 MD steps each. The density profile of positively (circles) and negatively (triangles) charged counterions are also shown...... 95

Figure 3.1: Dependence of the surface coverage (σΓ2) on the number of deposition steps for (a) system A (b) system B (c) system C and, (d) system D as specified according to Table 3.1. The closed symbols are for fully charged chains (f=1) and open symbols for

partial charged chains (f=1/2). The degree of polymerization is Np = 32 (circles), Np =16

(triangles) and Np= 8 (squares)...... 118

Figure 3.2: Density profiles of the fully charged (f=1) chains with Np=32 along z-axis after completion of 8 deposition cycles with a duration of 106 MD steps each deposition cycle for (a) system A (b) system B (c) system C and, (d) system D as specified according in Table 3.1. The monomer density profiles of negatively charged chains (continuous line) and positively charged chains (dotted line) are on left axis and the

negative counterions (triangle) and positive counterions (circle) is on right axis...... 119

Figure 3.3: Correlation function between positively and negatively charged monomers inside multilayers formed by fully charged chains, f=1, with degrees of polymerizations

Np= 32 (circles), 16 (triangles) after completion of the eight deposition steps. (a) Systems A (open symbols) and C (filled symbols); (b) Systems B (open symbols) and D (filled symbols)...... 120

Figure 3.4: Uniaxial monomer density difference of positively and negatively charged

chains, Δρ(z) = ρ − (z) − ρ + (z) for (a,b) systemA, (c,d) system B, (c) system C, (e,f) and (g,h) system D (Table 3.1). The film composition is taken after completion of 8th deposition step...... 121

Figure 3.5: Dependence of the overcharging fraction (|ΔQ|/Qads) on the deposition step number for different fraction of charged monomers f=1 (filled symbols) and f=1/2 (open

symbols) and degree of polymerizations Np=32 (circles), 16 (triangles) and 8 (square).

System A with Np=32 (circles), System A with Np=16 (inverted triangles), System B with

Np =32 (triangles), System C with Np=32 (squares), System C with Np =16 (rhombs),

System C with Np =8 and f=1 (filled triangles) and f=1/2 (inverted open triangles). .... 122

Figure 3.6: Time dependence of the polymer surface coverage for System A with fraction of charge monomers f=1/2 and degree of polymerization Np=16 during second, fourth, sixth and eighth deposition steps...... 123

Figure 3.7: Snapshot of the simulation box during the extended simulation run of the eighth deposition step for the System A with Np=16 and f=1/2. Positively charged monomers on the polyelectrolyte chains are colored in red and green. The negatively charged monomers are shown in blue. The green bead chains are polyelectrolytes added during the eighth deposition step while red bead chains are previously adsorbed polyelectrolytes. Neutral beads on the chains are shown in gray and on the surface are shown in black...... 124

Figure 4.1: Absorbance versus wavelength for PSS/PAH (10-2M, pH=3.5) films deposited by PSA using a salt concentration of 0.05 M and spin speed of 5000 rpm for different numbers of bilayers, shown in parenthesis. The inset shows absorbance (λmax=

226 nm) intensity versus number of bilayers for the same assembly. The solid line is a linear regression of the data set after 10th bilayer (closed circles) and the dashed line is the regression of the data up to 8th bilayer (open circles)...... 144

Figure 4.2: Dependence of the absorbance at λmax (226 nm) on the number of deposition cycle from PSS/PAH solutions (c = 10-2 M, pH=3.5) at salt (NaCl) concentration of 0.1 M at 5000 rpm up to 50 bilayers. The Uv-Vis absorbance is measure at 6 mm from the center of the quartz substrate. Solid lines represent linear regressions of the data after 10th bilayer...... 145

Figure 4.3: Dependence of the absorbance at λmax (226 nm) on the number of deposition cycle from PSS/PAH solutions (c = 10-2 M, pH=3.5) at spin-speeds of (a) 3000 rpm (b) 5000 rpm and (c) 6000 rpm and salt concentration of 0 M (closed circles), 0.1 M (open circle), 0.25 M (closed triangle), 0.5 M (open triangle) and, 1 M (square). The UV-Vis absorbance is measure at 6 mm from the center of the quartz substrate. Solid lines represent linear regressions of the data after 10th bilayer...... 146

Figure 4.4: Dependence of the absorbance growth rate on salt concentration for multilayered films prepared with polyelectrolyte spin assembly method at 3000 rpm (squares), 5000 rpm (circles) and 6000 rpm (triangles). The growth rate is measured at 6 mm from the center of substrate. The solid lines are the best fit to Eqn. (4.1)...... 147

Figure 4.5: Dependence of absorbance growth rate on salt concentration at radial distance of 0 mm (circles), 6 mm (triangles) and 8 mm (squares) from the center of the quartz disc for multilayered films prepared with polyelectrolyte spin assembly method at 3000 rpm. The solid lines are the best fit to Eqn. (4.1)...... 148

Figure 4.6: Radial dependence of the absorbance of PSS/PAH film at different salt concentrations: 0 M (filled circles), 0.1 M (filled triangles), 0.5 M (open circles) and 1.0 M (open triangle) at (a) 3000 rpm and (b) 5000 rpm and (c) 6000 rpm. Each point is averaged for the four different azimuthal angles. The dashed lines in Figure (a) are the best fit to the equation (4.4). All measurements are taken after completion of 20 deposition cycles...... 149

Figure 4.7: Dependence of the roughness of PSS/PAH films on salt concentration at

5000 rpm. AFM measurements are taken at 6 mm from the center of the disc after completion of 32 deposition cycles at different salt concentrations: (a) 0 M, (b) 0.1 M, and (c) 1 M...... 150

Figure 4.8: Dependence of the roughness of PSS/PAH films on salt concentration at 6000 rpm. AFM measurements are taken at 6 mm from the center of the disc after completion of 32 deposition cycles at different salt concentrations: (a) 0.1 M, (b) 0.5 M, and (c) 1 M...... 151

Figure 4.9: Peak-valley height distribution functions measured from the AFM images at 5000 rpm and different salt concentrations. The width of the solid line is the Full Width at Height Maximum (FWHM) and the dashed line represents the width of the RMS roughness. Inset shows the FWHM at different salt concentration...... 152

Figure 4.10: Dependence of film roughness on the salt concentration after completion of -2 32 deposition cycles of (PSS/PAH)32 at (c= 10 M, pH=3.5) for 5000 rpm (circles) and 6000 rpm (triangles)...... 153

Figure 4.11: Thickness of the PSS/PAH multilayer coatings at 6mm from the center of the substrate after deposition of 32 bilayers at 6000 rpm...... 154

Figure 5.1: Rate of the addition of Ca/P solutions to maintain constant composition during the growth of HA seed particles without addition of any polymer (closed circle) or with addition of PSS (25 μM) (open circles), PGA (5 μM) (closed triangle), PGA (25 μM) (open triangle) and PGA (25 μM) adsorbed at pH of 5.5 (closed square). The addition rate corresponds to the rate of the consumption of Ca/P during the HA growth by constant composition method (CCM)...... 179

Figure 5.2: Powder WAXD analysis for hydroxyapatite (HA) formed during the constant composition method utilizing two concentrations of PGA (a) 5 μM, (b) 25 μM and (c) HA seed...... 180

Figure 5.3: TEM images of (a) HA seed particles, (b) HA formed during CCM growth study after 60 minutes of 25 μM PGA addition and (c) HA formed during CCM study after 120 minutes of 25μM PGA addition (upturn in Figure 5.1). The horizontal scale

bars in the images are 100 nm...... 181

Figure 5.4: SEM image of hydroxyapatite nucleated on the surface of Si wafer coated

with layer-by-layer of PEI-(PGA-PSS)5-PGA exposed to supersaturated (metastable) Ca/P solution. Constant composition of the solution was maintained during the crystal growth by CCM...... 182

Figure 5.5: SEM image of hydroxyapatite formed in the Simulated Body Fluid (SBF) on

the Si wafer substrate coated with layer-by-layer with PEI-(PGA-PSS)5-PGA. The substrate was soaked in SBF for 6 days in order to deposit thick films of apatite. (Inset) High magnification SEM image of the boundary region of HA and the substrate...... 183

Figure 5.6: SEM images of silica formed on multilayers of (a) PEI-(PSS-PAH)19-PSS-

PLL (b) PEI-(PSS-PLL)20 and (c) control substrate of PEI-(PSS-PAH)20 after exposing the multilayers to hydrolyzed TMOS (113 mM) for 15 minutes...... 184

Figure 5.7: Tapping-mode AFM images of the silica formed on the multilayers of PEI-

(PSS-PLL)20 (a) phase and (b) amplitude trace of the surface showing presence of different sizes of the particles with different mechanical compliance and (c) amplitude trace of silica particle similar to the one observed in Figure 5.6...... 185

Figure 5.8: FTIR-ATR spectra of the silica formed on the multilayers of PEI-(PSS-

PLL)20 after exposure of the multilayers to hydrolyzed TMOS (100 mM). The underlying substrate to the multilayers was KaptonTM film...... 186

Figure 6.1: Silica yield obtained by addition different amount of poly(ethylene imine) (3 wt% aqueous solution) of two different molecular weight into TMOS or TEOS with ratio 1:1 (v/v). Silica yield is calculated based on the total conversion of silicon alkoxide into silica species after excluding the amount of catalyst (PEI) added to the resulting mixture. The vertical line is the initial slope of the linear regression of first three points...... 206

Figure 6.2: Thermo gravimetric analysis of the silica formed by addition of PEI solution to TMOS at ratio 1:1 (total 1.0 ml) (a) solids obtained at varying PEI concentration at TMOS fraction of 1.0 and (b) solids obtained at varying TMOS fraction in ethanol..... 207

Figure 6.3: Inorganic fraction of the solids formed by PEI addition to TMOS. Each

individual point represents the solution concentration of PEI and the TMOS fraction used for the silica formation. The inorganic fractions were calculated from weight % of the solids remaining at 850 °C from TGA curves...... 208

Figure 6.4: The silica yield obtained at various PEI and TMOS concentration. The silica yield was calculated from Eqn. 6.1...... 209

Figure 6.5: SEM images of silica obtained by addition of PEI solution concentration of (a) 1 wt%, (b) 3 wt%, (c) 5 wt% and (d) 10 wt% to pure TMOS at 1:1 v/v...... 210

Figure 6.6: SEM images of silica obtained by addition of 7 wt % PEI solution concentration to TMOS fraction of (a) 0.1, (b) 0.5, (c) 0.9 and (d) 1.0 at 1:1 v/v...... 211

Figure 6.7: Analysis of the silica particle diameter obtained from (a) addition of varying PEI concentration to pure TMOS and (b) addition of 7 wt% PEI solution to varying TMOS fraction in ethanol at 1:1 v/v...... 212

Figure 6.8: Silica formed on to the surface of polyelectrolyte multilayers containing (PEI-PSS)10-PEI layers upon exposure to TMOS for 15 minutes, (a,b) SEM image, scale bar 10 μm and 2 μm, resp., and (c) Contact-mode AFM image...... 213

Figure 6.9: SEM images of the silica formed on the surface by exposure of the single layer of spin-coated PEI on to the quartz surface to TMOS. (a) Low Mw PEI, scale bar 5μm and (b) High Mw PEI, scale bar 10 μm...... 214

Figure 7.1: 1H NMR of (a) poly(2-ethyl-2-oxazoline) and (b) linear PEI obtained from hydrolysis of (a)...... 236

Figure 7.2: SEM images of the fibers of the 50:50 blends of linear PEI and PVP in ethanol electrospun at concentration of (a) 5 wt%, (b) 10 wt% at flow rate of 0.1 ml/h and (c) 10 wt% at flow rate of 0.3 ml/h...... 237

Figure 7.3: SEM images of the silicified electrospun fibers of PEI:PVP (50:50) corresponding to the electrospun fibers of (a) Figure7.2a and (b) Figure 7.2b and (c) Figure 7.2c...... 238

Figure 7.4: Energy Dispersive X-ray (EDX) spectra of (a) PEI:PVP(50:50) electrospun

fibers, (b) after silicification by TMOS and (c) control sample of PVP fibers after TMOS exposure for 10 minutes...... 239

Figure 7.5: TGA analysis of the silicified and non-silicified electrospun fibers of 50:50 blends of linear PEI and PVP (i) PEI:PVP fibers, (ii) silicified electrospun PVP fibers (control), (iii) silicified PEI:PVP fibers after exposure at ca. 40%humidity and (iv) silicified PEI:PVP fibers after exposure at ca. 80% humidity...... 240

Figure 7.6: High resolution SEM images of the silicified fibers after calcination at 600 oC for 1 hour (a) top view (b) cross sectional view...... 241

Figure 7.7: FTIR of the electrospun PEI-PVP fibers (a) before silicification, (b) after silicification and (c) after calcination at 600 °C for 1 h...... 242

Figure 7.8: Density of the linear PEI foams obtained after freeze drying (closed circles) and after silicification (triangle) as the function of initial PEI concentration used to make the aqueous hydrogels. The error bar represents the range of the density values measured for 3 samples. Inset shows the pictures of the composite foams after the silicification as shown by the arrows...... 243

Figure 7.9: TGA analysis of the composite foams obtained by the silicification of the freeze dried linear aqueous PEI solution of (a) 2 wt % (continuous line), (b) 3 wt% (long dash line), (c) 5 wt % (medium dash line) and (d) 10 wt% (at periphery) (dot-dash line)...... 244

Figure 7.10: SEM images of linear PEI foams obtained by freeze drying of aqueous linear PEI solutions of 5 wt% (a,b) as such (c,d) after silicification and (e,f) after calcination at 500 °C for 1 h...... 245

Figure 8.1: (a) Relationship between the chain rigidity kb and the radius of gyration, Rg. 2 2 Squares and solid line indicate the Rg . Circles and dashed line indicate the ratio Ree / 2 Rg , in which Ree is the end to end distance of polymer chains. (adapted from Toshiaki et 234 al ) (b) The relation between the rigidity parameter s2 and the angle correlation of two successive bond vectors (adapted from Miura et al235)...... 265

List of Schemes

Scheme 1.1: Schematic representation of a polyelectrolyte multilayer build-up on a charged planar substrate by layer-by-layer assembly.143 The positively charged planar substrate on immersion in a negatively charged polyelectrolyte solution results in the charge reversal of the substrate by polyelectrolyte adsorption. Rinsing the substrate and immersing in the oppositely charged (positive) polyelectrolyte solution results in bilayer formation. Further build-up of multilayers is achieved by alternating immersion in the oppositely charged polyelectrolyte solutions with intervening rinse steps. (adapted from P.T. Mather and C.J. Lefaux)...... 56

Scheme 1.2: The zone model for the build-up of polyelectrolyte multilayers shows the progressive development of zones during film deposition starting from an adsorbed bilayer pair of polyelectrolytes. Multilayer build-up occurs mainly by the increase in thickness of zone II. The plot at the right depicts a model consisting of individual bilayers forming multilayers with a 50 % relative overlap between the layer-pairs. (Adapted from Decher, G:Science 1997)...... 57

Scheme 1.3: Schematic representation of the adsorbed layer (a) without shear; (b) without shear and with indication of Pincus blob; and (c) under applied shear (adapted from Lefaux, PhD dissertation, 2004).68...... 58

Scheme 1.4: The influence of pH and ionic strength on silica morphology by the sol-gel reaction of tetraethyl orthosilicate (adapted from Iler, 1979, p. 174).144 Formation of colloidal silica is favored under basic conditions and the absence of salt. Under acidic conditions, or when salt is present, gelation is favored that results in more networked and glassy materials...... 59

Scheme 1.5: Chemical structure of units responsible for silica polymerization145 (a) silaffin-1 A from Cylindrotheca Fusiformis (adapted from Kroger et al82) and b) lysine modification introducing permanently positive charges into silaffins from Eucampia zodiacus (adapted from Wenzl et al 146)...... 60

Scheme 1.6: Schematic drawing of the templating mechanism by the phase separation model [(A) to (D)] and scanning electron micrographs of C. wailesii valves during

morphogenesis [(E) to (H)]. (A) The monolayer of polyamine-containing droplets in close-packed arrangement within the SDV guides silica deposition. (B and C) Consecutive segregations of smaller (about 300 nm) droplets open new routes for silica precipitation. (D) Dispersion of 300-nm droplets into 50-nm droplets guides silica deposition. Silica precipitation occurs only within the water phase (white areas). The repeated phase separations produce a hierarchy of self-similar patterns. (E to H) SEM images of morphogenesis at the corresponding stages of development. (Adapted from Kroger et al82) ...... 61

Scheme 7.1: Chemical structures of the polymers involved in hybrid nanofiber formation: (a) poly(2-ethyl-2-oxazoline), (b) poly(vinyl pyrrolidone), and (c) poly(ethylene imine)...... 235

Scheme 8.1: Schematics showing the detail procedure to prepare the PEI/silica composite material by PEI hydrogel silicification by TMOS...... 266

Scheme 8.2: Idealized schematic of the procedure to obtain the silica patterning on the substrate by biomimetic silica formation from PEI by TMOS exposure (a) The layer of precursor polymer poly(2-ethyl-2-oxazoline) (PxOz) mixed with diphenyl iodonium toluene sulfonate (DITS) (photo acid generator) is spin-coated on the substrate (b) the exposure of the substrate to UV light with mask converts the precursor polymer, PxOz to PEI in the unmasked area by the reaction shown in the right (c) the exposure of the film to TMOS result in to the silica formation only on the area where PEI is present and (d) the calcination of the film to remove all the organic material leaves the pattern of the silica that is positive replica of the mask...... 267

Acknowledgements

Throughout the course of my graduate research that resulted into this dissertation,

I have been helped by a lot of people, both directly and indirectly, to whom I would like to express my gratitude. I would like to thank to my advisor, Dr. Patrick T. Mather, for his help, support, guidance, and critical feedback during this entire journey. His creative thinking, organizational abilities, mentoring skills and commitment to teaching have always been a source of inspiration to me. Sincere gratitude goes to my co-advisor, Dr.

Andrey V. Dobrynin, for his help and support, with his quick wit and imagination, to simplify the scientific results which have shaped the simulation work in the dissertation.

I would also like to thank Dr. A. Jon Goldberg for introducing me to biomineralization and dental applications and for his help and support through various discussions.

I would like to express my gratitude to my associate advisors, Dr. David A.

Schiraldi and Dr. Alexander M. Jamieson for useful discussions. Dr. Junhwan Jeon, for his help teaching me various complex algorithms in simulations. Dr. Christophe Lefaux and Dr. Maria Advincula for numerous helpful discussions. All the group members of

Mather and Dobrynin Research Groups, for their constant help. Thanks to all friends that

I have made at Case and UConn for making this long journey enjoyable in everyway.

My deepest appreciation goes to my mother, Hansaben, my brother, Sumit, and my extended family, for always standing besides me in all endeavors. To my love (and wife), Rita, for her constant encouragement, support and unconditional love. Finally, I dedicate this dissertation to my late father, whose dream for me when fulfilled, is not here in body but is surely with me in spirits.

Polyelectrolyte Multilayers:

Simulations, Experiments and Applications in Biomineralization

Abstract

PRITESH ARJUNBHAI PATEL

Polyelectrolyte multilayer formation is achieved by alternate adsorption of oppositely charged polymers in a layer-by-layer (LbL) fashion from dilute polyelectrolyte solutions. This dissertation examines the formation, growth, structure, and morphology of polyelectrolyte multilayers by utilizing molecular dynamics (MD) simulations and polyelectrolyte spin assembly (PSA) experiments employing a spin-coating radial flow.

Application of multilayers as substrates for biomineralization of hydroxyapatite (HA) and silica is also examined.

MD simulations of assembly of flexible polyelectrolytes into multilayers were performed at a charged planar surface from dilute polyelectrolyte solutions. These simulations show that multilayer growth proceeds through surface overcharging, chain intermixing, and a linear increase in polymer surface coverage at each deposition step.

The strong electrostatic attraction between oppositely charged polyelectrolytes at each deposition step is a driving force behind the multilayer growth. Polymer surface coverage and multilayer structure are each strongly influenced by the charge fraction of polyelectrolytes, as well as the strength of electrostatic and short-range interactions.

Experimental results from PSA elucidated the synergistic effect of the spin-speed

and the solution ionic strength on the growth and morphology of multilayers. The growth rate and polymer surface coverage of multilayers shows a non-monotonic dependence on solution ionic strength, first increasing and then decreasing as the solution ionic strength is increased. This is a manifestation of two competing mechanisms responsible for multilayer formation in agreement with Flory-like theory of multilayer formation from polyelectrolyte solutions under flow. At low salt concentrations, the electrostatic interactions control the multilayer assembly process while, at high salt concentrations, the multilayer assembly it is dominated by shear flow.

For applications of multilayers in biomineralization, the possibility of forming

HA and silica from simple synthetic macromolecules and polypeptides that have similar functionality as proteins found in nature was examined. Poly(glutamic acid) was studied for HA formation, while poly(lysine) and poly(ethylene imine) (PEI) were studied for silica formation. Such mineral formation was investigated on the surfaces of multilayers, in solutions, and on the polymer scaffolds formed using electrospinning or freeze-drying techniques.

CHAPTER 1

1 Introduction

1.1 Layer-by-Layer ‘Dipping’ Assembly (part I)

Layer-by-Layer (LbL) deposition of charged molecules is an assembly process

used to produce multilayered thin films with molecular resolution. The LbL deposition

principle, first described by Iler in 19661 for assembly of colloidal particles, was later rediscovered and extended to polyelectrolytes in the early 1990s by Decher.2-5 LbL multilayer films are made by sequential adsorption of alternating charged molecules, one molecular layer at time, with an intervening rinse step. The LbL assembly of charged molecules is a simple and yet versatile technique for the formation of multi-component thin films [for review see 6-9 ]. LbL deposition has distinct advantages over other thin

film fabrication process like the Langmuir-Blodgett (LB) technique,10,11 in which

monolayers with angstrom-level precision in thickness are formed on a surface and

subsequently transferred onto a solid support. Unlike the LB technique that requires

special equipment and has severe limitations with respect to substrate size and topology,

LbL deposition has practically no limitations on the type of charge-bearing species, or the

substrate, allowing fabrication of multilayers with limitless charged species such as synthetic polyelectrolytes, DNA,12 proteins,13 and nanoparticles14 onto various substrates.

LbL assembly has also been extended to multilayer fabrication that utilizes hydrogen

bonding,15 donor/acceptor interactions,16 and metal-ion coordination17 in addition to the

usual electrostatic interaction. Thus, LbL assembly provides a platform for construction

of complex multi-component heterostructured thin films capable of incorporating more

than one type of building block with the desired functionality. This ability has resulted in

utilization of multilayered films for a variety of applications that can be classified in two

major categories:7 (i) tailoring surface interactions, and (ii) fabrication of surface-based devices. Various examples of applications in the first category include corrosion protection,18 antireflective coatings,19 antistatic coatings for electrophoresis,20 surface- induced nucleation of minerals,21 antibacterial22 and antifouling coatings,23 and bio-

sensing,24 among many others. Examples for fabrication of surface-based devices

include nanoreactors,25 photonic devices such as light emitting diodes,26 and barrier or

separation layers,27 among other applications.

LbL multilayer films are assembled by alternating immersion of a charge-bearing

substrate in solutions of oppositely charged polyelectrolytes with a rinse step in between

to remove loosely adsorbed chains. Scheme 1.1 shows such an assembly process,

subsequently referred to as ‘dipping’ assembly. First, a charged substrate is immersed

(‘dipped’) into a dilute aqueous solution of cationic (or anionic) polyelectrolytes until

polyelectrolyte adsorption reaches equilibrium (saturation or desire polymer surface

coverage), resulting in a molecularly adsorbed polyelectrolyte layer. A subsequent

rinsing step in pure solvent is necessary to remove polymers that are not tightly adsorbed

to the substrate. The next step involves the dipping of the substrate into a solution of

oppositely-charged polyelectrolytes, again followed by a rinsing step. Each exposure to dilute solution of polyelectrolytes reverses the surface charge and thus reconstructs the

surface properties, leaving it primed for the next deposition step of oppositely charged

polyelectrolytes. This process is repeated multiple times to maintain film growth until

the desired thickness or the number of ‘bilayers’ is achieved. Here, a bilayer consists of

an adsorbed layer-pair of polyanion and polycation. The key to sustained growth of

multilayers by LbL is charge inversion and subsequent reconstruction of the surface

properties after each deposition step. Experiments have shown a stepwise increase of

multilayer thickness, film mass, and/or surface coverage with each deposition step –

hereby, referred to as growth rate - indicating a steady state regime of the multilayer

assembly. Typically, the thickness increment has been shown to be around 10-60 Å per

each ‘bilayer’ or deposition cycle.

1.1.1 Characterization Techniques:

Various characterization methods have been utilized to deduce the internal

structure and organization, film thickness, and roughness of the multilayer films. UV/Vis

spectroscopy is commonly used to characterize the growth rate of the multilayers when

one or both polyelectrolytes absorb radiation with at wavelength between 190-800 nm.7

For example, UV-Vis absorbance spectroscopy was used to study the growth rate of multilayers formed by oppositely charged polyelectrolyte pairs of poly(styrene sulfonate)/poly(allyl amine)(PSS/PAH) because of absorbance by the distinct π to π* transition of PSS at 226 nm that allows determination of the mass of deposited PSS after each deposition step.6 Besides UV/Vis spectroscopy, several other techniques such as

ellipsometry,28 quartz crystal microbalance (QCM),29 streaming potential measurements

(SPM),28 atomic force microscopy (AFM),30 and x-ray and neutron reflectivity31 have

been used to characterize multilayer growth-rates. For in-situ characterization of

multilayer structure, X-ray and neutron reflectometry31 are most commonly used along

with other techniques like surface plasmon spectroscopy, ellipsometry, in-situ AFM,32 attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR),30 and

Zeta potential measurements.7 The roughness of the polymer coatings at film-air

interfaces have primarily been characterized by AFM, though techniques like neutron

scattering or SPM have also been used.

1.1.2 Multilayer Structure, Properties and Growth Mechanisms

Structural and physical properties of multilayered films assembled by LbL have been studied extensively over the last ten years. Various studies have established that: (i)

multilayers are not stratified into well-defined layers, but are interdiffused and

intermixed;2,31 (ii) chain adsorption is irreversible on the time scale of multilayer

assembly and the counterions do not participate in the charge balance within the

multilayers;33 (iii) layer thickness and molecular organization of adsorbed polymers can be precisely tuned by varying salt concentration, solvent quality, polyelectrolyte charge density, pH of the solutions, etc. with polyelectrolyte charge density and ionic strength being the most influential parameters determining multilayer formation;34 (iv) polymer concentration, molecular weight, and deposition time are less influential parameters;35 and (v) oppositely charged monomers form ion-pairs and adsorption is limited by the electrostatic barrier at the surface.36

Neutron and X-ray reflectivity have been used extensively to deduce the internal

structure of multilayers based by adsorbing deuterated polyelectrolytes at specific

deposition steps. Schmitt et al31 observed for PSS/PAH multilayers that the

polyelectrolyte chains interdigitate (or interpenetrate) one another intimately over the

length scale of 12 Å compared to the layer-pair thickness of 20-30 Å. The internal

structure of the multilayer is believed to be strongly dependent on the charge density of

polyelectrolytes, though very limited experimental evidence exists that deduce the

influence of charge density on internal structure. Schmitt et al31 has indicated that the layers of weak polyelectrolytes are significantly more interpenetrated and tend to form a

1:1 stoichiometic complexes of anionic and cationic pairs. The same group later37 found that ~ 40% of water by volume is present within the multilayers and that inorganic salt counterions are excluded within multilayers and play a minor role in the multilayer assembly. Tarabia et al38 used X-ray and neutron reflectivity measurement to study

internal structure of films formed from deuterated poly(phenylene vinylene) (D-PPV) and

found a similar (~12 ± 3 Å) interfacial thickness between the adjacent polyelectrolytes.

Using streaming potential measurements (SPM) and scanning angle reflectometry (SAR),

Ladam et a l39 have studied the structural properties of multilayers of PSS/PAH using streaming potential measurements (SPM) and scanning angle reflectometry (SAR), and concluded that a symmetrical and constant charge inversion occurs during the multilayer build-up. It was also shown for PSS/PAH multilayers that a regular build up regime is reached after the first six bilayers.

Combination of X-ray and neutron reflectivity, radioactive labeling of counterions,

SPM, Zeta potential measurements, and ellipsometry have established a three zone model of the multilayer films2,7,39, as depicted in Scheme 1.2. Zone I consist of polyelectrolyte

layers in proximity to the substrate and extends to the first few polyelectrolyte layers.

Zone II is the bulk film that contains interpenetrating layers of polyelectrolytes. Although

there is significant intermixing between polyelectrolytes adsorbed during different

deposition steps, some level of stratification is observed as seen in Scheme 1.2.

Polyelectrolytes form a 1:1 stoichoimetric complex in Zone II.39 It is important to note that the interfaces between each zone are not sharp but probably represent the gradual transition between each zone. The local structure in Zone II of multilayers formed by flexible polyelectrolytes is believed to be similar to that of bulk polyelectrolyte complexes formed between similar polymers. Zone III is comprised of one or a few polyelectrolyte layers close to the surface of the film and includes a diffuse double layer of counterions to maintain overall electroneutrality.

Despite the extensive experimental studies of layer-by-layer deposition of charged molecules,2,6,28,31,33,38,40 a theoretical understanding of such systems is lagging behind.41-44

Solis and de la Cruz22 have developed a model of spontaneous equilibrium layering of

mixtures of positively and negatively charged polymers close to a charged wall due to

their chemical incompatibility. Netz and Joanny45 have proposed a scaling model of

multilayer formation in semi-flexible polyelectrolytes. However, this model lacks

intermixing between polyelectrolyte chains in neighboring layers. Castelnovo and

Joanny41 have taken into account the strong interpenetration of polyelectrolyte chains in

consecutive layers by incorporating complex formation between oppositely charged

polyelectrolytes into self-consistent field equations, describing the polymer density

profile in the adsorbed layers. The numerical solutions of the self-consistent field

equations have been recently presented by Wang46 and by Shafir and Andelman.47 These calculations have shown that a sufficiently strong short-range attraction between oppositely charged polymers is essential for the formation of multilayers. The formation of ionic pairs between polyelectrolyte chains forming multilayers was taken into account

by Park et al44 and by Lefaux et al.48 These models show promising results by predicting the correct salt concentration dependence of multilayer growth by sequential adsorption and by spin-coating methods. However, these models neglect strong intermixing between layers by assuming a frozen layer structure after completion of each deposition step. Such assumption can only be justified for description of multilayer assembly by spin-coating where chains do not have sufficient time to diffuse into the film during film assembly. A primary reason for the slow development of theoretical models for LbL is the difficulty associated with experimental verification of the assumptions made in theoretical models.

At this stage, molecular simulations offer a valuable tool that can help in understanding the basic physical ‘mechanisms’ governing LbL assembly.

Monte Carlo simulations of multilayered film assembly from mixtures of oppositely charged polyelectrolytes near charged spherical particles and uniformly charged surfaces were performed by Messina et al.49-51 These papers tested the hypothesis that multilayer formation is an equilibrium process that occurs not only when one proceeds in a step-wise fashion, as done in experiments, but also when oppositely charged polyelectrolytes are added together and the resulting solution is exposed to a charged substrate. It was shown that additional short-range attractive interactions between polyelectrolytes and the surface are required to successfully initiate film growth.

Unfortunately, these simulations were limited to only a few (three) deposition steps; thus, the system was far from reaching a steady-state regime in which film thickness and mass increased linearly with the number of deposition steps as seen in experiments. This shortcoming was addressed in recent molecular dynamics (MD) simulations by

Panchagnula et al52,53 who studied the sequential adsorption of oppositely charged

polyelectrolytes onto a charged spherical particle. These simulations confirmed that layer

build up proceeds through surface overcharging during each deposition step and that the

system reaches a steady state regime after a few deposition steps with non-linear growth

of polymer mass in the aggregate. Despite this steady growth, however, the spherical

symmetry of such a particle precluded formation of well-developed multilayered

structures.

1.1.3 Research Significance - I

The experimental and preliminary simulation studies on the formation, structure, and properties of multilayers provide partial evidence of the basic physical ‘mechanisms’ governing the LbL assembly process. For example, even though it is widely accepted that electrostatic interactions govern the LbL assembly, the relative importance of the short range hydrophobic interactions is not known. None of the studies have answered the most important question regarding successful multilayer growth; are there any universal mechanisms governing LbL multilayer formation? Also, various assumptions made in the theoretical analysis of the multilayer formation could not be tested experimentally. Hence, at this stage, molecular simulations of multilayer formation from a polyelectrolyte solution onto a planar substrate that mimics experimental conditions will offer a useful tool in order to better understand the multilayer assembly process.

A detailed study of multilayer assembly and layer structure including the effects of fraction of charged monomers in the polymer backbone, chain degree of polymerization, electrostatic interactions, and short range interactions is described.

Detailed analysis of the effect of the system parameters on the polymer density

distribution in the growing film; film surface morphology, polymer intermixing, and

interdiffusion of polyelectrolyte chains between layers; and ion pair formation between

oppositely charged macromolecules. Based on the results of the computer simulation a

theoretical model of multilayer assembly is developed. Finally, the universal mechanism

governing multilayer formation is described.

1.2 Polyelectrolyte Spin Assembly (part II)

Despite its advantages, dipping-based LbL for multilayer film assembly is a very

time-consuming process lasting hours with each dipping step requiring between 10 min to

30 min for completion. As an alternative, spin-assisted LbL assembly allows rapid

processing of multilayered films by employing conventional spin-coating methods.48,54-59

With this technique, a spin coater is used to adsorb the polyelectrolyte layers onto a charged substrate by applying excess polyelectrolyte solution before or during the spinning, followed by a rinse consisting of exposure of the spinning substrate to pure water. During the spinning, most of the (excess) solution is expelled out (< 1 s) due to centrifugal forces leaving a film with thickness ~ 1 µm, following which the solution coating more gradually thins and dries over the course of 2-15 s. The whole deposition process of a single polyelectrolyte layer usually takes about 10-15s compared to the conventional 10–30 min for the quiescent adsorption (‘dipping’). This spin-assembly

(SA) or polyelectrolyte spin-assembly (PSA), has been shown to produce more compact and less intermixed layers with lower roughness compared to the films produced by the dipping process.54 Due to faster processing time provided by PSA, free standing films

have been made for a variety of applications.60 This method has been applied to build

multilayered films consisting of synthetic polyelectrolytes,54 nanoparticles,61 and dendrimers,62 among other charged species.

1.2.1 Prior Studies

The mechanism of polyelectrolyte adsorption in PSA is different than the

diffusion-controlled adsorption of polyelectrolytes in the ‘dipping’ technique. In PSA,

the combination of the shear stress due to flow and electrostatic interactions between

polyelectrolytes controls the resulting film structure. Several studies have reported the

effect of parameters like polyelectrolyte molecular weight,57 concentration,58,63 and spin-

speed57,59,64 on the growth rate of multilayers formed by PSA. In a series of papers,57,59,64,65 Lee and Cho studied the effect of such parameters on the growth of

multilayers by PSA. They observed that the growth rate depends weakly (logarithmically)

on polyelectrolyte molecular weight, with higher molecular weight polyelectrolytes

showing a lower growth-rate.57 These findings were attributed to a denser packing of the low molecular weight polyelectrolytes as compared to the high molecular weight polyelectrolytes. The same group reported that the growth rate of multilayers by PSA first increases with polyelectrolyte concentration up to 10 mM (based on repeat unit molar mass) and then saturates at a constant value at higher polyelectrolyte concentrations. In contrast, the multilayer growth rate of the PSA process decreases with increasing the spin-speed.57,59,64 Such PSA growth rate (absorbance per bilayer, Γ ) dependence was fit to an empirical power-law function that included dependence on the polyelectrolyte concentration, c, and spin-speed,ω,[Γ ~ ωαcβ].59 The power-law exponent

for polyelectrolyte concentration,β, was equal to 0.78 in the polymer concentration range

of 1-10 mM. The power-law exponent of the spin-speed,α, was equal to -0.34, a smaller

(negative) value compared to that obtained for the conventional spin-coating of non-

charged polymer solutions (usually, -0.5). This weaker spin-speed dependence was

attributed to the effect of the strong electrostatic force felt by polyelectrolytes during

spin-coating flow, but no fundamental explanation has been given.

In contrast to the effects of polymer concentration, molecular weight, and spin

speed, salt concentration has a strong effect on the multilayer build-up by PSA. For PSA

assembly from polyelectrolyte solutions with no added salt, it was observed that the

growth rate of the multilayer coatings by spin-assembly is higher than that obtained by

the ‘dipping’ technique.66 However, Cho et al67, while studying the multilayer growth by

PSA on patterned surfaces, found that the PSA growth rate of multilayers is lower than

that deposited by the dipping technique in the presence of salt in the concentration range

between 0.1 M and 1.0 M NaCl. They attributed this to strong desorption forces,

specifically centrifugal and shear force, experienced by the polyelectrolytes with

increasing salt concentration due to change in their conformation with adding salt.

However no fundamental explanation or empirical relation for the dependence of growth

rate on spin speed and salt concentration was suggested. Additionally, the same

researchers found that well-defined patterns (“high pattern quality”) were obtained at an

intermediate ionic strength of 0.4 M using PSA. Lefaux et al48 using UV-Vis

spectroscopy and AFM, studied the dependence of multilayer growth rate on the solution

ionic strength at a fixed spin speed of 3000 rpm. It was found that the growth rate of

polymer surface coverage and thickness increased rapidly with salt concentration up to

0.1 M and then reached a constant value at higher ionic strengths. These experimental

results were analyzed within the framework of the Flory-like model of multilayer

assembly that accounts for both the effect of electrostatic interactions between oppositely

charged chains within multilayers and chain deformation by shear flow.

1.2.2 Scaling Model for PSA growth68

A brief description of a scaling model for polyelectrolyte adsorption under flow,

starting from the quiescent ‘dipping’ assembly case of no flow, is described below. A more detailed analysis can be found elsewhere.48,68

1.2.2.1 Dipping Assembly

An adsorbed polyelectrolyte chain is schematically depicted to have a brush-like loop structure (Scheme 1.3a) containing N Kuhn monomers of length b at a charged

surface with a charge number density of σ (mol/m2). The equilibrium structure is determined by the interplay of three main contributions to the chain’s free energy, Fch: (i) electrostatic repulsion energy between the charged monomers within the adsorbed layer,

(ii) loop elasticity arising from the stretching sections of polyelectrolyte chains, and (iii) the binding energy of the charged monomers to the surface by ion pair formation. The energy of the ion pair is equal to -εakBT, where kB is the Boltzmann constant, T is the

absolute temperature, and εa is dimensionless binding energy strength. The three

contributions to the chains free energy, Fch, are combined to yield the following equation,

with each term in the right hand representing these contributions in the order discussed,

F N ⎛ m2l r 2σ H 2 ε ⎞ ch ≈ ⎜ B D + − a ⎟ (1.1) ⎜ 2 ⎟ k BT m ⎝ H mb 2 ⎠

where H is the thickness of the adsorbing layer, lB is the Bjerrum length (defined as the

length scale at which the Coulomb interaction between two elementary charges e in a

dielectric medium of dielectric constant ε is equal to the thermal energy, kBT), rD is the

Debye radius that depends on ionic strength (defined as the characteristic length scale at which the electrostatic interactions between two charges decays with distance), and m is half of the number of monomers in each loop formed by chain sections. The equilibrium thickness of the adsorbed polyelectrolyte layer, H, can be calculated by minimization of the chain free energy (Eq 1.1) with respect to H. Further considering that, at equilibrium, the chemical potential of chains in solution is equal to that of chains in an adsorbed layer gives the number of monomers in each loop.

ε a m ≈ 2 / 3 (1.2) uσr 2 ()D

The next layer is considered to be capable of forming ionic pairs only with the charged

monomers of the previous layer within the thickness on the order of the Debye screening

length rD from the top of the adsorbing layer. Thus, the effective surface charge density of

each new layer, σi+1, includes only the fraction of charges presented by the adsorbed

chains:

r σ 2 / 3r1/ 3 σ ≈ σ m D ≈ i D (1.3) i+1 i H u1/ 3b

For steady state (linear) growth, each newly adsorbed layer completely reconstructs the surface properties so that σi+1 = σi , leading to a surface charge density of the simple form:

rD σ * ≈ 2 (1.4) b lB

ε ε 1 Γ ≈ mσ * ≈ a ∝ a c 2 (1.5) lB rD lB

1.2.2.2 Scaling theory for effect of Shear Rate (Flow):

The shear force due to hydrodynamic drag affects the conformation of adsorbed polyelectrolytes, which in turn affects surface coverage (Γ) and thickness (H). This effect of shear on polyelectrolyte conformation was considered by deformation of the

b 2 chain characterized by Pincus blobs of size D ≈ .69 The Pincus blob size is defined as rD the length scale below which the electrostatic repulsion between neighboring chains is not enough to perturb chain conformation. At length scales larger than D, the size is influenced by electrostatic repulsion between charged monomers. The shear flow will start deforming the Pincus blobs at the time scale comparable to the relaxation time

(Zimm) of the Pincus blobs, given by,

3 ηb3 ⎛ b ⎞ ⎜ ⎟ τ e ≈ ⎜ ⎟ (1.6) k BT ⎝ rD ⎠

A dimensionless group, the Deborah number Dee ≡ γcτ e , was introduced to compare the relaxation time with the characteristic time of shear flow. It was postulated that the external shear flow will influence the brush structure,70,71 until the chain adjusts

their conformations, which results in a local Deborah number of Dee ≡ γcτ e ≈ 1. Also, the shear tilts the “brush” in such a way that the vector sum of (i) the shear force on top

of the brush layer, (ii) elastic force, and (iii) the electrostatic repulsive force are balanced at equilibrium (see Scheme 1.3(b, c)).72,73 The tilting of the brush results in the total brush length to be equal to H 2 + R 2 , where R is the lateral displacement of the end point of the loops from the grafting point (Scheme 1.3b). The balance of the projections of the forces into normal and parallel to the surface directions leads to two equations, defining H, R, as function of the number of monomers per loop m:

H m2l σ *r 2 k T ≈ k T B D (1.7) B mb2 B H 2

R 2 1/ 3 2 / 3 kBT 2 ≈ηγD ≈ ()(ηγ kBT ) (1.8) mb

In Eqn. (1.7) the expression for the blob size D, was obtained from the following

3 condition γτ ≈ γηD / k BT ≈ 1. Eqn. (1.7) yields the same relation between the height of the brush and the number of monomers in a loop, m, as Eqn. (1.8).

Substituting the relations of H/mb2, and R/mb2 from Eqns. 1.7 and 1.8 to chain free energy, we can write the expression of the chain free energy (single chain chemical potential) as a function of the Debye radius, shear rate, and binding energy, yielding Eqn.

(1.9).

2 / 3 F N ⎛ H 2 R 2 ε ⎞ ⎛ r 2 ⎛ηγb3 ⎞ ε ⎞ ch ≈ ⎜ + − a ⎟ ≈ N⎜ D + ⎜ ⎟ − a ⎟ (1.9) k T m ⎜ mb2 mb2 2 ⎟ ⎜ b 2 ⎜ k T ⎟ 2m ⎟ B ⎝ ⎠ ⎝ ⎝ B ⎠ ⎠

Setting Eqn. (1.9) to zero yields a prediction for the number of monomers per half-loop as a function of shear rate and Debye screening length :

2 b ε a ε a m ≈ 2 2 2 / 3 → 2 / 3 (1.10) γ>>r 2 rD + b (γτ 0 ) D (γτ 0 )

3 where factoring leavesτ 0 ≈ ηb /(kBT ) , the relaxation time of a Kuhn monomer, and the arrow indicates the limit of high shear. Analysis of Eqn. (1.10) reveals that the crossover between electrostatic-dominated and shear-dominated regimes occurs when the Debye

1/ 3 length, rD, becomes smaller than b(γτ 0 ) . Beyond this crossover, the number of monomers in a brush strand decreases as the shear rate increases. Using this new expression relating the number of monomers in a strand, m, to shear rate we can obtain dependence of the brush thickness H and polymer surface coverage Γ on the shear rate and salt concentration:

2 1/ 3 r ε b H ≈ bm σ *ur 2 ≈ D a (1.11) ()D 2 2 2 / 3 rD + b ()γτ 0

ε ()r l Γ ≈ σ *m ≈ a D B (1.12) 2 2 2 3 rD + b ()γτ o where b is the Kuhn length, γ is the shear rate, and τo is a “bare” relaxation time for each

Kuhn segment – generally an unknown quantity. The thickness-per-bilayer is predicted to follow a similar form.

Thus, at steady state multilayer growth, the model predicts that the growth rate of polymer surface coverage per bilayer, Γ, can be described by the following equation, in phenomenological form:

c−1/2 Γ≈α (1.13) c−12/3+ βγ

where α and β are fitting parameters, γ is the shear rate and c is the salt concentration of the solution, including both added salt and polyelectrolyte counterions. Eqn. 1.13 predicts two regimes of multilayer build-up by PSA: (i) at low salt concentration, the c-1 term dominates over the shear rate term (in denominator), predicting a square root dependence of surface coverage on salt concentration, Γ ∝ c1/2, and (ii) at high salt concentration, the shear rate term dominates, predicting an inverse square root dependence of surface coverage, Γ ∝ c-1/2.

1.2.3 Research Significance

Using UV-Vis spectroscopy and AFM, Lefaux et al,48 studied the dependence of

PSA multilayer growth rate on the solution ionic strength at a fixed spin-speed of 3000 rpm. It was found that the growth rate of polymer surface coverage and thickness increased rapidly with salt concentration up to 0.1 M and then reached an apparent constant value at higher ionic strengths. Thus, at the studied spin-speed of 3000 rpm, the second regime, where the surface coverage decreases with an increase in ionic strength, was not observed. Furthermore, it is known that polymer surface coverage in PSA exhibits significant radial dependence, due to the fact that shear rate depends on radius in spin-coating flow. The shear rate increases linearly near the surfaces with the distance r from the center of the rotating disk according to expression given by,

γ = ρω2rh / η (1.14) where η is the viscosity of the polyelectrolyte solution, ω is the angular velocity of the rotating disk, and h is the thickness of the ‘thinning’ polymer solution. Such radial

dependence was tested only at 3000 rpm at higher ionic strength of 0.5 M by Lefaux et al,48 giving very little evidence for its validity at lower ionic strength and/or higher spin speed or shear rate. Hence, there was a need to verify the scaling model that predicts the effect of shear rate on the multilayer build-up using PSA.

In part, this dissertation extends the previous study and considers the combined effects of both PSA spin speed and salt concentration on coating growth rate, morphology, thickness, and roughness of multilayer coatings. Additionally, the radial dependence of polymer surface coverage on spin speed and salt concentration was studied. We determined whether or not the observed behavior can be predicted on the basis of a Flory- like theory of multilayer formation from polyelectrolyte solution under shear flow.

1.3 Biomineralization (part III)

Nature excels in the design and synthesis of complex and hierarchical hybrid materials for various functional purposes through biomineralization processes.74-76

Hybrid materials like bone, mollusk seashell, nacre, and silica structures in diatoms and sponges have inspired researchers to mimic their structure and function for various applications.77-79 In such hybrid materials, organic molecules are closely integrated with inorganic moieties for various structural and functional purposes.80,81 In particular, the formation of natural hybrid materials are often regulated by proteins that catalyze mineral formation while spatially directing or templating the growth of hierarchical structures.

Proteins, for example, like silaffins found in various species of diatoms, or bone sialoprotein (BSP) in mammalian bone, regulate mineral growth (silica and hydroxyapatite, respectively) to form species-specific nano-structured composite

materials.82,83 Such functional aspects of the protein associated with mineralization, namely (i) to catalyze mineral formation and (ii) act as templating agent have been extensively studied by various researchers to design biomimetic materials for various applications. Such biomimetic strategies has led to application to develop novel materials such as advanced composites and coatings for medical, chemical, optical and electronic applications.84-86

1.3.1 Polyelectrolyte Multilayers for Biomineralization

It is of high interest to researchers to develop various ‘simple’ synthetic macromolecules that mimic the functional aspect of the proteins involved in the mineralization. In order to understand the mineralization processes in-vitro, various systems has been developed that allow modification of the mineral nucleation sites in a controlled manner to study the growth and morphology of the mineral. One of the convenient ways to study the formation of minerals is to induce mineralization on the solid surfaces where the functional macromolecules are localized. Such mineral formation on surfaces have been long constituted a model system for understanding the role of natural and synthetic macromolecules in the biomineralization processes.87,88 The nucleation of minerals on the surfaces is thought to occurs primarily by one of the two mechanisms (i) template matching or (ii) electrostatic interactions. Template matching, relies on degree of molecular recognition between the underlying film and the nucleating species.87 The underlying substrate mimics a particular plane in the nucleating crystal and leading to a nucleation bounded by the matching plane. Alternatively, electrostatic attraction of the ions to the surfaces followed by either geometrical or stereochemical

matching play a role in the nucleation of biominerals. Accordingly, various type of substrate like Langmuir-Blodgett (LB) films, Langmuir monolayers bearing various functionalities,89,90, a layer of proteins extracted from mollusk shell,91 self-assembled monolayer (SAM) of surfactant molecules84 or simply a layer of adsorbed macromolecules on to crystal surfaces92 have been extensively used to study mineral formation and morphology under carefully designed physicochemical conditions.

Polyelectrolyte multilayers, obtained by layer-by-layer (LbL) assembly of polyelectrolytes, offer a potential route as an alternative to study the biomineralization processes on surfaces. The important advantage of the LbL assembly is that it is very simple, yet versatile technique to incorporate majority of the functional molecules in the multilayers, unlike previous techniques like LB films or SAM. LbL deposition has practically no limitations on the shape of the template, type of charge-bearing species, or the substrate, allowing fabrication of multilayers from variety of macromolecules and substrate. Other major inherent advantage of the multilayers is that majority of the proteins or macromolecules used in the mineral formation are charge-bearing species that can be easily incorporated in the multilayers. For the other neutral (uncharged) species, multilayers could be constructed on the substrate utilizing secondary interactions like hydrogen bonding or hydrophobic interactions. Hence, one of the dissertation objectives is to develop the use of polyelectrolyte multilayers as a model substrate to study the biomineralization process, particularly the formation of hydroxyapatite and silica from simple polypeptides having similar functionality as the proteins that catalyzes such mineral formation.

The background and prior studies related to the hydroxyapatite formation is described in section 1.3.2, while the background and prior studies of silica formation is described in section 1.3.4.

1.3.2 Hydroxyapatite Formation

Calcium Phosphates exist in many forms that include with decreasing order of solubility: amorphous calcium phosphate (ACP), dicalcium phosphate dihydrate

(CaHPO4.2H2O ,DCPD), tricalcium phosphate (Ca3(PO4)2,TCP), octacalcium phosphate

(Ca4H(PO4)3.2.5H2O, OCP) and hydroxyapatite (Ca5(PO4)3OH, HA). Hydroxyapatite is the most stable form of calcium phosphate at normal temperature and pH of 4 and 12, and is the main constituent of bone and teeth in the mammals. In bone, the mineralization of

HA occurs by deposition of carbonated HA crystals in an extra-cellular matrix consisting of type I collagen and a variety of non-collagenous proteins. The major proteins responsible for the mammalian bone formation are considered to be bone sialoprotein

(BSP)93 and osteopontin.94 Both of these proteins contain regions enriched in acidic amino acids, particularly Glutamic acid and Aspartic acid.95,96 Particularly, BSP has a continuous sequence of ten Glutamic amino acids in two regions of the protein sequence, while osteopontin has a continuous sequence of 9 glutamic acid residues.96 Thus, in general, amino acids containing carboxylic acid group has been thought to responsible for the mineralization of calcium phosphates in various systems.

The importance of carboxylic acid side chains containing amino acids, like glutamic and aspartic acid, has been tested by many researchers using model systems in- vitro in solution and surfaces for their ability to induce nucleation and growth of calcium

phosphates. In solution, extensive studies exists for the role of various macromolecules, that affects the growth of the hydroxyapatite from metastable calcium phosphates [ see

92,97,98]. Most of these studies utilize studying the growth of the HA crystals in solution by constant composition method (CCM).99 CCM, allows one to maintain constant activity of all ionic species during the crystal growth of HA seed in the metastable solution of calcium, phosphate, and hydroxyl ions.99,100 Using CCM, the HA crystal growth in solution have been measured in presence of various synthetic macromolecules,98 amino acids containing uncharged polar (carboxylate) side groups,101 and various polypeptides containing carboxylic acid groups like poly(glutamic acid) and poly(aspartic acid).92 Most of the macromolecules binds to the HA surface and inhibits the growth of HA crystal when present in solution. Addadi et al102,103 have shown that acidic proteins containing carboxylates and sulfates, nucleate HA crystal formation when immobilized on the surface while inhibit crystal formation when free in solution by biding to the crystal.104 Hunter et al,95 showed that poly(l-glutamic acid) and poly(d- glutamic acid) has ability to nucleate hydroxyapatite in agarose gel systems from metastable calcium phosphate solutions. Tsortos et al92 have recently studied in detail the adsorption of PGA on to the germanium surfaces and their subsequent ability to nucleate the HA formation when on surfaces. They have concluded that PGA can nucleate the formation of calcium phosphate phases of OCP and HA, simultaneously, from metastable calcium phosphates.

In contrast to the numerous studies of HA mineralization on to various substrates like agarose gel, LB films and monolayers, very limited studies exist for the mineral formation on to the polyelectrolyte multilayers. Ngakam et al21 have shown that the

multilayers of PSS/PAH, induce the growth of calcium phosphates when exposed to calcium/phosphate supersaturated solutions, irrespective of the presence of PSS or PAH in outermost layer. The critical calcium or phosphate concentration to observe nucleation in multilayer films ending with PSS was 5.75 mM (Ca/P =1.0), while critical concentration for PAH ending layer was 6.12 mM. The nucleation was believed to occur by electrostatic attraction of the ionic species (calcium or phosphate ions). However, the calcium/phosphate concentration used in this study ([Ca]>5 mM, Ca/P =1.0) is well above the metastable regime and corresponds to the regime of spontaneous precipitation of calcium phosphates.105 In other recent study (reported after our study on polyelectrolyte multilayers for HA formation was done), Ball et al106 have shown that polyelectrolyte multilayers with poly(l-lysine) as outermost layer and made from either 3,

6 or 8 bilayers in which the inner layer content varies in PGA, are able to initiate calcium phosphate nucleation and growth.

1.3.2.1 Research Significance (HA formation)

While studying the growth of hydroxyapatite seed in the metastable solution

([Ca]=2.0 mM, Ca/P =1.67), we found that PGA does not completely inhibit the growth of the HA crystal, unlike the earlier studies reported in the literature and summarized above. Instead, the PGA only lowers the growth rate up to certain time, then the growth of the HA seed crystal switches to the original growth rate. Hence, in Chapter 5, our studies for the growth of the HA crystal in presence of PGA in solution is reported. Our data indicates that the PGA binds only to certain crystal faces of hydroxyapatite and thus partially covering the hydroxyapatite crystal surface. The covered area by PGA inhibits

the growth of hydroxyapatite while the crystal growth in the uncovered area is still taking place. Also, in Chapter 5, it is reported that polyelectrolyte multilayers of PEI-

(PGA/PAH)5-PGA can successfully induced nucleation of HA and OCP from relatively low metastable solutions ([Ca]=2 mM, Ca/P=1.6). The hydroxyapatite formation on multilayers of PEI-(PGA/PAH)5-PGA by prolonged exposure to simulated body fluid

(SBF) under physiological conditions is also reported.

1.3.3 Silica Biomineralization

In nature, silica occurs in both amorphous and crystalline forms, with biogenic silica (silica from living organisms) constituting a majority of the amorphous phase in nature. A silicate mineral exist in the form of various crystalline polymorphs and generally is non-biogenic. Amorphous silica, one of the main constituent of glass used for centuries, has gained substantial technological significance over the past half century due to ever-emerging applications in catalyst support, water purification, adsorbent and thickening agent, chromatographic separations, and as a filler in composite materials.107

Each of these applications requires tailor-made silica with a specific particulate morphology, in turn demanding specific synthetic approaches rather than utilization of naturally occurring silica with impurities. Conventionally, most amorphous silica is synthesized by a sol-gel process, where silica is obtained from the two step reaction of hydrolysis and condensation of organosilicate precursors. Such processes involve extreme pH, high temperature, using acid/base catalysis, and long times. On other hand, biogenic silica precipitation occurs under benign conditions, generally at higher rates compared to the sol-gel process, and with desired morphological control. Recent findings

provide insight into simpler alternative biomimetic routes for silica formation, creating the potential for a huge technological advancement in silica-based materials. This dissertation aims, in part III, to deduce the role of bio-inspired simple molecules in silica biomineralization and to apply such knowledge to the design and synthesis of hybrid organic/inorganic materials containing polymer and silica.

1.3.3.1 Sol-Gel Silica Synthesis

The synthesis of silica carried out using alkoxysilanes (organosilicate) is generally referred as sol-gel synthesis.107 Tetramethyl orthosilicate (TMOS) or tetraethyl orthosilicate (TEOS) are most commonly used as precursor alkoxysilanes for the sol-gel synthesis of silica. The silica synthesis proceeds via the two step reaction108 of alkoxysilanes like TMOS and TEOS: (i) hydrolysis of the alkoxy (e.g.~Si-OCH3 in

TMOS) functional group to form silanol (Si-OH), and (ii) condensation of the silanol group either with an alkoxy group or another silanol group to form a siloxane (~Si-O-Si~) bond, liberating alcohol or water, respectively. The hydrolysis is generally carried out by acid or base catalysis, and thought to occur by an SN2 mechanism that involves a silicon forming a pentacoordinate intermediate.108,109 The hydrolysis rate is slowest at neutral pH, and increases in either direction of increasing or decreasing the pH from neutral conditions. The condensation or polymerization of the silanol groups can be catalyzed by acid/base or spontaneous over the time, usually requiring hours to days for completion depending on the reaction conditions. Silica polymerization reaction rate has minimum at pH 2; the reaction rate increases with increasing or decreasing pH from 2.107

Silica formation and its morphology can be controlled by varying pH and salt

concentration.107 Scheme 1.4 summarizes such an effect on the formation and morphology of silica particles by the sol-gel reaction of TEOS. The hydrolysis of TEOS, and subsequent condensation, forms oligomeric nano-particles ranging in size from 5 nm to 30 nm and generally referred to as ‘sols’. Such silica sol particles can either condense further or remain in solution depending on the pH of the medium. For acid-catalyzed reactions, the hydrolysis rate is rapid and condensation is the rate-limiting step and usually incomplete. This phenomenon reverses under alkaline conditions where hydrolysis is generally slow and rate-limiting. At alkaline conditions, pH > 7.0, the sol particles generally are negatively charged, preventing them from further condensation and thus the silica sol solution remains stable due to repulsion between particles unless an electrolyte (like NaCl) is used as a flocculent to precipitate the sol particles. The addition of the electrolyte reduces the repulsion of the charged species thus leading to its coagulations driven by Van der Waals attraction. At acidic conditions, pH<7.0, the particles are neutral, resulting in slow condensation of the sol particles over time to form a network of glassy materials. Thus, formation of colloidal silica or ‘sol’ particles occurs under basic conditions while gelation occurs at acidic conditions.

Major applications of silica-based materials have been made possible due to use of the sol-gel reaction on templates formed by organization of various molecules over wide range of length scales. Mesoporous silica, which has a wide-variety of applications, was synthesized using ammonium surfactant as a template.110 Several other templates formed by organic molecules (transcription), including spherical polymer particles or crystals, tri-block copolymers, and “organogelators” have been used to form structured composites [for review see111]. The templating route utilizes specific interactions like

hydrogen bonding and/or electrostatic interactions between the sol-gel precursor and the templating substrate to synthesize silica with prescribed nano, micro, or mesostructure.

Thus, silica can be shaped into numerous structures at varying length scales from nanometer to micron by varying the nature of the templating substrate as well as the sol- gel reaction conditions.

1.3.3.2 Biogenic Silica Formation

Marine organisms like diatoms, sponges and radiolaria112 are well-known for their ability to form silica by silicic acid uptake from the sea environment at very low concentrations, typically 10-70 µM, well below the solubility limit of silicic acid (2 mM)(for review see80,113). Such organisms are responsible for up to 40% of the silica found in nature and play an important role in the natural silicon cycle.114 Diatoms, and sponges, are the most studied organisms for silica formation. Silica formation in diatoms occurs in a specific organelle called the silica deposition vesicle (SDV), during cell wall morphogenesis.115 In the SDV, silica precursors are stored in a concentration ranging from 19-380 mM at slightly acidic pH ~ 5.115 The chemical nature of the storage compounds has not been determined but argued to be partially-precipitated oligomeric species76,115 or organosilicates such as sugar-silicate or silicon-catechol complexes.76 The silica formed in diatoms generally consists of nanoparticles typically 10 nm to 100 nm in diameter.116 The sol-gel reaction at pH ~ 5 generally favors gelation rather than flocculation to form nanoparticles, indicating that the silica formation occurs by mechanisms distinct from those described previously. Also, silica formation in such cases occurs very rapidly compared to the conventional sol-gel reaction.

Silica-mediating proteins, from the diatom species Cylindrotheca fusiformi, have been studied in detail for their ability to form silica from hydrolyzed precursors. The proteins from diatoms were isolated in various fractions by HF acid extraction and were labeled into three main fractions: (i) native silaffins-1 (natSil-1), (ii) native silaffins-2

(natSil-2), and (iii) long chain polyamines117 (Scheme 1.5). Analysis of the amino acid sequences of purified natSil1 revealed seven highly homologous repeating units (R1 to

R7) containing lysine modified residues.76 The functional units responsible for the rapid silica precipitation of all these fractions have been identified as polylysine residues covalently modified by the oligo-N-methyl-propylamine unit that are polyamine structures.82 These modifications have been proven to play a central role in the silica precipitating activity of the native enzyme.82,118 In addition, polyvalent ions such as phosphates are believed to be assisting the ‘aggregation’ or self-assembly of polyamines and promote the condensation of silica into spherical particles.119-121 Catalytic domains in the protein, primarily consisting of polyamines, self-assemble into specific spatial arrangements that direct silica growth in the diatoms.82 The polyamines catalyze the silica formation due to an alternating sequence of protonated and non-protonated amine groups in the polyamine chains, which promotes hydrogen bond formation with the oxygen attached to silicon in the precursor and consequently facilitates ~Si-O-Si~ bond formation.122

The specific patterns of silica observed in various species of diatoms are due to the combination of oil/water phase separation of catalytic polyamine domains induced by silica polymerization and the concomitant creation of fresh polyamine/water interfaces due to the amphiphilic nature of the catalytic domains.82 Scheme 1.6 shows the model of

phase separation/silica polymerization process and compares with the high resolution

SEM images of corresponding stages of morphogenesis in diatoms. The model (Scheme

1.6 - top row) postulates repeated phase separation processes within the SDV, which produces an emulsion of micro-droplets at initial stages, and subsequent breakdown of such droplets into smaller size. The breakdown is thought to occur due to ‘consumption’ of polyamines in the droplets by co-precipitation, and thus the spontaneous rearrangement of unreacted droplets to a smaller size driven by the amphiphilic properties exhibited by methylated long-chain polyamines. The newly created water/polyamine interface continues to precipitate silica that is mediated by polyamines and thus consuming another polyamine fraction. Such processes continue until well defined droplets, ~ 50 nm in diameter, are formed arranged with hexagonal packing.

1.3.3.3 Biomimetic Silica Formation

The above-mentioned findings relating to silica catalysis by proteins in diatoms provide insight into simpler, alternative ‘biomimetic’ routes for silica formation, creating the potential for development of silica under mild conditions and higher rates. Here, the term ‘biomimetic’ will be used in the sense of mimicking the natural silica formation process, but with simple (even synthetic) polymers having identical or similar functionality as the natural catalyst isolated from diatoms. Proteins, like silaffins, natSil1, and natSil2 isolated from marine species of diatoms were found to form silica instantaneously from hydrolyzed precursor tetraalkoxysilanes.123 R5, one of the repeating sequences of the protein natSil1, has been shown to catalyze silica formation under ambient conditions124 with various morphologies in-vitro ranging from spheres to

fibrillar structures depending on the concentration and shear forces applied during silica formation.125

Numerous studies were aimed at identifying small amino acids or small peptide sequences for their catalytic ability to precipitate silica.125-127 It was recognized that the simple oligomers of lysine128 and various polyamines129,130 can precipitate silica from their precursor solutions. Mizutani et al129 reported that homo-polypeptides like polylysine, polyallyamine, and polyarginine precipitate silica from their precursors.

Polylysine has been widely studied for the precipitation of silica due to the initial study signifying the role of lysine residues in the silica precipitation activity.128 Influence of various chemical and physical conditions like concentration of OH- ions and shear on silica morphology was studied in detail by Rodriguez et al.131 Various studies now exist for the silica precipitating ability of polyamines including poly(allylamine),129 polyarginine127,129, polylysine,127,130,132 polyethylene imine,133-135 and polypropylene imine. 133 It has been demonstrated that in the presence of such macromolecules silica polymerization rate increases compared to control samples. Because such a wide variety of polymeric species have been observed to catalyze silica formation, question was raised about the simplicity of silica precipitation being merely due to electrostatic interactions.136 Indeed, electrostatic interactions between these molecules and silica species does play an important role in the silica synthesis,137 but the ability of polyamines to direct (rapid) silica formation and morphology at near acidic pH still cannot be explained by electrostatic interactions. This explains the post-translation modification of lysine residues of the proteins in diatoms that precipitate silica by addition of polyamines structures. The above investigations have opened doors to the controlled synthesis of a

wide range of organic-inorganic functional materials.

The importance of polyamines in silica precipitation has led to the development of various bio-inspired molecules, synthetic macromolecules, polypeptides, block copolypeptides, and small functional molecules containing polyamines that catalyze silica.

The time scale of silica precipitation in the presence of such molecules varies depending on the type of functionality and silica precursor. The majority of these biomimetic silica formations have been studied in the presence of a pre-hydrolyzed precursor, usually hydrolyzed by acid/base catalysis rather than directly from pure alkoxysilanes. On the other hand, poly(ethylene imine) (PEI), linear and branched, have the ability to form silica directly from pure alkoxysilanes such as tetramethyl orthosilicate (TMOS) in the presence of catalytic amounts of water.61,138 Additionally, linear PEI can form fibrous aggregates because of its ability to crystallize in aqueous media above a concentration of

0.5 wt%.139 Such aqueous, linear PEI aggregates have also been shown to induce a rapid hydrolytic condensation of TMOS and instantly create silica of different morphologies depending on the concentration of alkoxysilanes, water, and solvent.135,138 It was also shown that the silica formed on crystalline PEI filaments has a 5 nm to 7 nm core containing PEI fibrils with a 6 nm to 8 nm thick silica shell.140

1.3.3.4 Flocculation of Silica Sol by PEI

Even though recent studies suggests that the PEI can catalyze silica polymerization reactions, similar to the polyamines found in the nature, the polycationic nature of the PEI has ability to flocculate the silica sol particle during the usual sol-gel synthesis reaction. Iler107 has shown that above pH 9, the silica polymerization reaction

produces sol particles, diameter of which depends on the solutions conditions of pH, and temperature and is usually between 2-10 nm. Thus, silica formation from TMOS by aqueous PEI solutions can occur by two competitive processes. In the first hypothesis, the PEI chains merely bridge the already formed silica particles from the instantaneous hydrolysis and condensation of TMOS. It has been shown that PEI can strongly adsorb on to the silica surface at pH > 2.0. Lindquist and Stratton141 have studied such flocculation of the silica sol particles of diameter of ca. 20 nm (LudoxTM) by PEI. They observed that at pH 3-9, the flocculation is governed by electrostatic interactions between negatively charged sol particles and protonated PEI. Above pH 9, where PEI is less protonated, the adsorption occurs by hydrogen bonding interactions of nitrogen atom of

PEI with oxygen of the surface silanol groups and the resulting flocculation occurs by bridging mechanism, where the silica sol particles are bridged by the PEI chains. There exists a critical flocculation concentration (CFC) of PEI below which the flocculation does not occur and a redispersion concentration (RC)142 above which the colloidal particle is coated with polymer and bears the same charge as the polymer and is redispersed. The flocculation regime is between CFC and RC. Lindquist and Stratton141 showed that flocculation regime depends mainly on the pH and ionic strength of the solution and poorly on molecular weight of PEI.

1.3.3.5 Research Significance – Silica Formation

Based on the above mentioned hypothesis for silica formation from PEI, the role of PEI to induce silica formation directly from non-hydrolyzed TMOS was examined. In parallel to the studies by Yuan et al,138 it was observed that branched PEI can directly

catalyze silica formation when added to silica precursor – TMOS. Such unique ability of

PEI to form silica directly from TMOS is then investigated in solutions and surfaces

(Chapter 6 and 7). In solution, the effect of aqueous PEI and TMOS concentration (in ethanol) is studied on the inorganic content, yield, and morphology of the resulting composite. The concentration regimes of PEI and TMOS are well above the redispersion concentration of silica sol particle observed in study by Lindquist et al141 in the flocculation study. The fact that the solid precipitates at such PEI concentration above the reported redispersion concentration, suggests that the silica formation is not just by bridging of sol particles formed due to the usual silica polymerization reaction of TMOS.

Silica formation by PEI is argued to be combined effect of the ability of PEI to catalyze siloxane bond formation and flocculate the silica particles formed during the reaction. To study formation of silica on surfaces, PEI was deposited either into various layers within polyelectrolyte multilayers or with single layer onto the quartz substrate. In both cases, silica formation was observed when the surfaces containing PEI were exposed to TMOS, further confirming that PEI can catalyze the silica formation on surface. Such rapid synthesis of silica networks is further extended in Chapter 7 for silica formation in polymer scaffolds obtained from electrospinning techniques as well as from freeze drying of the crystalline linear PEI aqueous solutions. Rapid silica synthesis on to such scaffolds offer potentials to obtain organic/inorganic hybrid materials for various applications such as hard tissue engineering, dental materials, silica aerogels and nanocomposites, in general.

1.4 Thesis Outline

The thesis consists of three parts as outlined previously in the present chapter.

Part I (Chapter 2 and 3) aims to deduce the basic mechanisms governing the formation of polyelectrolytes multilayers by molecular dynamics simulations. Part II (Chapter 4) aims to study the multilayer formation under the spin-coating flow and compare the multilayer formation with the predictions of Flory-type theory developed for formation of polyelectrolyte multilayer under shear flow. Part III (Chapter 5, 6 and 7) study the formation of minerals (hydroxyapatite and silica) in solution and surfaces.

Chapter 1 (present chapter) outlines the research significance of each part and is a general introduction to all the chapters in the thesis. This chapter is referred frequently in subsequent chapters to serve as the introduction of the content discussed in that particular chapter.

Chapter 2 describes the simulation model and method to study multilayer formation by molecular dynamics simulations similar to the dipping LbL assembly. The chapter describes the effect of fraction of charged monomers in the polymer backbone, and chain degree of polymerization on the multilayer formation and layer structure. The multilayers is characterized by polymer density distribution in growing film, film surface morphology, intermixing and interdiffusion of polyelectrolyte chains between layers, and ion pair formation between oppositely charged macromolecules.

Chapter 3 is continuation of Chapter 2 and extends the analysis of multilayer formation and layer structure to consider the effect of electrostatics and short range interactions. The effect of above mentioned parameters on the multilayer stability is described. Based on the results of the simulations, a theoretical model of multilayer

assembly is developed. Finally, the universal mechanism governing multilayer formation is described.

Chapter 4 presents an experimental study to deduce the effect of spin-speed and salt concentration on the growth rate, morphology, thickness and roughness of the multilayer coatings. Additionally, the radial dependence of the polymer surface coverage at different spin speed and salt concentration is studied. The experimental results are compared with that predicted from Flory-type theory of multilayer formation.

Chapter 5 describes the applications of the polyelectrolyte multilayers to study formation of hydroxyapatite and silica on the surfaces. Hydroxyapatite crystal growth in presence of PGA in solution and surfaces is described by constant composition method.

The silica formation and morphology with poly(l-lysine) localized in to the multilayers is also studied.

Chapter 6 describes the direct hydrolytic condensation of TMOS by PEI in solution and surfaces. In solution, the silica formation is studied by varying aqueous PEI and TMOS concentration to deduce the basic mechanisms governing the silica formation.

The silica formation on the surfaces of the multilayers of PEI/PSS as well as on a single layer of PEI adsorbed on to the surface is described.

Chapter 7 describes the rapid formation of organic/inorganic hybrid materials by silica formation on to the PEI scaffolds. The scaffolds are made by electrospinning or by freeze drying of the aqueous crystalline linear PEI hydrogels. Finally, Chapter 8 summarizes the conclusion and future recommendation of the present thesis.

Rinse

Further Build-Up PolyAnion PolyCation

Rinse

10-60Å

Scheme 1.1: Schematic representation of a polyelectrolyte multilayer build-up on a

charged planar substrate by layer-by-layer assembly.143 The positively charged

planar substrate on immersion in a negatively charged polyelectrolyte solution results in the charge reversal of the substrate by polyelectrolyte adsorption. Rinsing the substrate and immersing in the oppositely charged (positive) polyelectrolyte solution results in bilayer formation. Further build-up of multilayers is achieved by alternating immersion in the oppositely charged polyelectrolyte solutions with intervening rinse steps. (adapted from P.T. Mather and C.J. Lefaux)

Scheme 1.2: The zone model for the build-up of polyelectrolyte multilayers shows

the progressive development of zones during film deposition starting from an adsorbed bilayer pair of polyelectrolytes. Multilayer build-up occurs mainly by the increase in thickness of zone II. The plot at the right depicts a model consisting of individual bilayers forming multilayers with a 50 % relative overlap between the layer-pairs. (Adapted from Decher, G:Science 1997)

a) Η

1 / σ * b) D

F el Fshear c) F elast

Scheme 1.3: Schematic representation of the adsorbed layer (a) without shear; (b) without shear and with indication of Pincus blob; and (c) under applied shear

(adapted from Lefaux, PhD dissertation, 2004).68

Scheme 1.4: The influence of pH and ionic strength on silica morphology by the sol-

gel reaction of tetraethyl orthosilicate (adapted from Iler, 1979, p. 174).144

Formation of colloidal silica is favored under basic conditions and the absence of salt. Under acidic conditions, or when salt is present, gelation is favored that results in more networked and glassy materials.

Eucampia zodiacus (adapted from Wenzl et al 146).

Scheme 1.6: Schematic drawing of the templating mechanism by the phase

separation model [(A) to (D)] and scanning electron micrographs of C. wailesii

valves during morphogenesis [(E) to (H)]. (A) The monolayer of polyamine- containing droplets in close-packed arrangement within the SDV guides silica deposition. (B and C) Consecutive segregations of smaller (about 300 nm) droplets

open new routes for silica precipitation. (D) Dispersion of 300-nm droplets into 50-

nm droplets guides silica deposition. Silica precipitation occurs only within the

water phase (white areas). The repeated phase separations produce a hierarchy of

self-similar patterns. (E to H) SEM images of morphogenesis at the corresponding

stages of development. (Adapted from Kroger et al82)

CHAPTER 2

2 Molecular Dynamics Simulations of Layer-by-Layer

Assembly of Polyelectrolyte*

2.1 Synopsis

In this chapter, Molecular Dynamics (MD) simulations of electrostatic assembly of multilayers of flexible polyelectrolytes at a charged surface are described. The multilayer build-up was achieved through simulatd sequential adsorption of oppositely charged polymers in a layer-by-layer fashion from dilute polyelectrolyte solutions. The steady state multilayer growth proceeds through a charge reversal of the adsorbed polymeric film which leads to a linear increase in the polymer surface coverage after completion of the first few deposition steps. Polymer density distribution in a growing film, surface morphology, intermixing and interdiffusion of polyelectrolyte chains between layers, and ion pair formation between oppositely charged macromolecules are deduced from the simulations. Substantial intermixing between chains adsorbed during different deposition steps is observed. This intermixing is consistent with the observed requirement for several deposition steps to transpire for completion of a single layer.

However, despite chain intermixing, there are almost perfect periodic oscillations of the density difference between monomers belonging to positively and negatively charged macromolecules in the adsorbed film. A detailed study of multilayer formation and layer structure including the effects of fraction of charged monomers on polymer backbone and

* Reproduced with permission from Patel, P. A.; Jeon, J.; Mather, P. T.; Dobrynin, A. V. Langmuir 2005, 21,

chain degree of polymerization is presented in this chapter. This chapter is organized as follows: Section 2.3 describes model and simulation method, section 2.4 describes the results of the MD simulations that includes the formation of multilayers, effect of charge density and degree of polymerization on multilayer structure and theoretical model for the multilayers and section 2.5 presents the discussion and conclusions. The next chapter,

Chapter 3, will consider the effect of electrostatic and short-range interactions on multilayer formation and structure.

2.2 Introduction

Refer Section 1.1 of Chapter 1.

2.3 Model and Simulation Method

The MD simulations of multilayer assembly are performed from dilute polyelectrolyte solutions of chains with degree of polymerizations Np = 32, 16 and 8.

The fraction of charged monomers on each chain is equal to f = 1, 1/2 or 1/3, corresponding to every, every second, and every third monomer carrying a charge. All the combinations of charge fraction f and the chain degree of polymerization, Np, were studied except for the system with fraction of charged monomers f = 1/3 and degree of polymerization Np = 8. The charge distribution for this system (f = 1/3, Np = 8) is asymmetric, resulting in charge sequence effect. Polyelectrolytes are modeled as bead- spring chains consisting of Np monomers of diameter, σ. The connectivity of beads in the chains is maintained by the finite extensible nonlinear elastic (FENE) potential:132

⎛ r 2 ⎞ U (r) = −0.5k R 2 ln⎜1− ⎟ (2.1) FENE s max ⎜ 2 ⎟ ⎝ Rmax ⎠

2 with the spring constant ks = 30k BT σ , where kB is the Boltzmann constant and T is the

absolute temperature, and the maximum bond length being Rmax = 1.5σ . Counterions with diameter σ are explicitly included in our simulations. Electrostatic interaction between any two charged particles bearing charge valences qi and qj, and separated by a distance rij is given by the Coulomb potential:

lB qi q j U Coul (rij ) = k BT (2.2) rij

2 where lB is the Bjerrum length, lB = e εk BT , defined as the length scale at which the

Coulomb interaction between two elementary charges e in a dielectric medium of dielectric constant ε is equal to the thermal energy kBT. For our simulations, the Bjerrum

length was fixed at lB = 1.0σ. All charged particles in our simulations are monovalent ions with valence qi = ±1.

The adsorbing surface was modeled by a periodic hexagonal packed lattice of particles with diameter σ located at z = 0. Every second particle on the lower surface has univalent charge. A similar but uncharged nonselective surface was located in the opposite side of the simulation box to prevent chains from escaping and hence maintain

2-D periodicity in the lateral (x and y) directions. The system size is 20σ × 20.784σ ×

80σ for systems with fraction of charged monomers f = 1 and 1/2. For the systems with f

= 1/3 the box size was enlarged to 20σ × 20.784σ × 160σ to have enough charges in a

system to overcharge a surface at monomer concentration 0.038σ −3 . The Particle-Particle

Particle-Mesh (PPPM) method for the slab-geometry implemented in LAMMPS147 with the sixth order charge interpolation scheme was used to calculate the electrostatic interactions in the system. In this method, the 2-D periodic images of the system are periodically replicated along the z-direction with distance L = 3Lz between their boundaries. This reduces the problem of calculation of the electrostatic interactions in a

2-D periodic system compared to those in a 3D system.

In addition to electrostatic interactions, both charged and uncharged particles in the system interact through a truncated-shifted Lennard-Jones (LJ) potential.

⎧ 12 6 12 6 ⎡⎛σ ⎞ ⎛σ ⎞ ⎛ σ ⎞ ⎛ σ ⎞ ⎤ ⎪4ε ⎢⎜ ⎟ − ⎜ ⎟ − ⎜ ⎟ + ⎜ ⎟ ⎥ for r ≤ r U (r) = LJ ⎜ ⎟ ⎜ ⎟ cut (2.3) LJ ⎨ ⎢⎝ r ⎠ ⎝ r ⎠ ⎝ rcut ⎠ ⎝ rcut ⎠ ⎥ ⎪ ⎣ ⎦ ⎩ 0 for r > rcut

A cutoff distance of rcut = 2.5σ is chosen for the surface particles/polymer-polymer

1/ 6 interaction while a smaller value, rcut = 2 σ , was selected for polymer-counterion, surface particle-counterion, and counterion-counterion interactions for the results

described in the present chapter. The interaction parameter was equal to ε LJ = k BT for all pairwise interactions.

Each bead in the coarse-grained bead-spring model used in our simulations is meant to represent several chemical units of the polymer chain. If we assume that the

Bjerrum length, lB = 1 σ , is equal to the Bjerrum length in aqueous solutions at room

temperature (T=298 K), then lB = 7.14 Å, and the monomer size is equal to σ = 7.14 Å.

This corresponds to approximately 2.9 monomers of poly(styrene sulfonate) sodium salt

(PSS) with monomer size 2.5 Å. This further leads to a chain degree of polymerization for Np=32 to be of the order of 100 monomers.

During each deposition step, the simulations are carried out with a constant number of particles, volume, and temperature (NVT) ensemble. The constant temperature is maintained by coupling the system to a Langevin thermostat. In this case, the equation of motion for the ith particle is

G dv G G G m i ()t = F (t) −ξv (t) + F R (t) (2.4) dt i i i

G G th where vi is the bead velocity, and Fi is the net deterministic force acting on the i bead

G R of mass m. Fi is the stochastic force with zero average value and δ-functional

G G R R correlations Fi ()t Fi (t′ ) =6ξkBTδ(t −t′) . The friction coefficient was set to

1 2 ξ = m τ LJ = 1/τ LJ , where τLJ is the standard LJ time, τ LJ = σ (m ε LJ ) , and m is a particle

mass that was set to 1 σ. The velocity-Verlet algorithm with a time step Δt = 0.01 τLJ was used to integrate the equations of motion (2.4).

Simulations were performed using the following procedure. Counterions from the charged surface were uniformly distributed over the simulation box. Then, M1 negatively charged polyelectrolytes with Np monomers corresponding to monomer concentration of

−3 0.038σ (e.g., M1 = 40 for Np = 32 and f = 1 and 1/2), together with their counterions, were added to the simulation box and simulations continued until completion of the first deposition step. For chains with fraction of charged monomers f=1/3, the number of chains with degree of polymerization Np = 32 was M1 = 80 to maintain the same

concentration of polyelectrolytes since the simulation box size was doubled in the z- direction. After completion of the first simulation run (‘dipping’ step), unadsorbed polyelectrolyte chains were removed (‘rinsing’ step). Between steps, the unadsorbed polyelectrolytes were separated from adsorbed ones using a cluster algorithm148 with a cutoff radius equal to 1.2σ. Additionally, between adsorption steps, the only counterions required to maintain the system electro-neutrality (compensating for the excess charge in the growing polymeric layer) were kept in the simulation box. At the beginning of the second step, the simulation box is refilled with M2 = M1 oppositely charged polyelectrolytes together with their counterions. The concentration of newly added polyelectrolytes is the same as before, 0.038 σ −3. This, then, is followed by another simulation run. These dipping and rinsing simulation steps were repeated twelve times to study multilayer formation and structure.

It is important to optimize the number of integration steps for each simulation run

(deposition step) so that the system reaches saturation of the polymer adsorbed amount ( i.e. approach equilibrium), thereby enabling overall steady state growth. The increase in the number of adsorbed polymers in the growing layers was monitored by plotting the polymer surface coverage Γ, defined as the total number of adsorbed monomers per unit surface area of the charged planar surface, S, as a function of the number of integration

(MD) steps. Figure 2.1 shows the evolution of the polymer surface coverage for the first two deposition steps for fully charged polyelectrolytes with f = 1 of degree of

6 polymerization Np = 32 during the simulation runs with duration 3×10 integration steps.

For both cases there is relatively fast saturation in the adsorption amount (about 90%)

during the first 5×104 integration steps. Hence, the duration of the simulation run for each deposition step were set to 5×105 integration steps, which is about ten times longer than is necessary to achieve a saturation limit. However, as mentioned earlier, the simulation box was doubled in size for simulations of weakly charged polyelectrolyte chains with f = 1/3 and Np = 32 and 16 while keeping the same concentration. In these cases, the duration of each simulation run was increased to 1.5×106 integration steps to allow polymer chains to diffuse through the enlarged simulation box. Our simulation corresponds to Rouse dynamics of a polymer chain for which the chain’s relaxation time

2 increases with the chain degree of polymerization Np as Np . Thus, the selected length of simulation runs is also sufficient for shorter polyelectrolyte chains with degree of polymerization Np = 16 and 8 to reach the steady state regime.

2.4 Results

2.4.1 Formation of Multilayers

Multilayer build-up by MD simulations was achieved by sequential adsorption of oppositely charged polyelectrolytes as envisioned in experiments. Figure 2.2 shows the evolution of the layer build-up during the first five deposition steps for fully charged chains, f = 1, of the degree of polymerization Np= 32. After the first deposition step, the polyelectrolyte chains almost uniformly cover the whole adsorbing surface. The few loops and tails that are formed contribute to the overcharging and the surface charge reversal necessary for reconstruction of surface properties and continuation of the chain adsorption during the next deposition step. Interestingly, after completion of the second

deposition step, the adsorbed chains do not completely cover the substrate, but instead leave islands with high polymer content coexisting with empty regions protruding down to the bare surface. These islands are less pronounced for partially charged chains (f =

1/2) compared to the fully charged chains (f = 1). During the third deposition step, adsorbing polyelectrolyte chains bear the same charge as those being adsorbed during the first deposition step. Thus, these chains adsorb onto the islands formed by oppositely charged polyelectrolyte chains from the second deposition step and refill the empty spots on the surface (see third step in Figure 2.2). The polyelectrolytes added into the simulation box during the fourth deposition cycle have the same charge as those adsorbed during the second deposition step. These chains refill the empty spots being left after completion of the third deposition step and then start formation of the fourth layer on top of the oppositely charged polymers deposited during the third deposition step. Further layer growth proceeds in similar fashion such that two deposition steps of similarly charged polymers are usually required for the completion of a single layer. The described pattern of the layer formation was seen for all 12 deposition steps performed in each simulated system. The polymer surface coverage, Γ, and average thickness of the adsorbed layers increases linearly with each deposition step for all the systems studied (as described later in Figure 2.5). A steady state linear growth regime is generally observed in experiments once the first few layers have been deposited.149

A density profile of positively and negatively charged monomers in the multilayers after completion of the third deposition step is shown in Figure 2.3a. The monomer density of negatively charged chains, ρ-(z), shows two peaks near 1σ and 3σ that correspond to the first and third deposition steps of negatively charged

polyelectrolytes. The only peak in the density profile of positively charged chains, ρ+(z), corresponds to monomers adsorbed during the second deposition step. The first peak near the surface has higher amplitude as compared to the peaks corresponding to the chains adsorbed during the second and third deposition steps. This is due to the high surface charge density in comparison with the value of the surface overcharging achieved after completion of the each deposition step. The large amount of polyelectrolytes adsorbed during the first deposition step is required to compensate for the surface charge as well as to overcharge the surface for subsequent layer build-up. The strong surface charge effect is reminiscent of the one observed in experiments where the initial growth rate of the film is different than the growth rate observed in a steady state regime.

Experiments have shown that the surface charge could influence the layer build-up for as many as the first six deposition steps for dipping assembly150 and up to sixteen deposition steps for spin-assembly.56 Figure 2.3b shows the density profile after completion of four deposition steps. Since polyelectrolytes deposited during the fourth deposition step have the same charge as those adsorbed during the second deposition step, the monomer density, ρ+(z), of positively charged chains increases. However, the majority of newly adsorbed monomers are added at a distance of about 2σ, corresponding to formation of the second layer of positively charged chains with a small amount of monomer being added on the top of the third (negatively) charged layer. Interestingly, the third layer peak observed in Figure 2.3a disappears after the fourth deposition step, giving rise to a slight increase in polymer density near the surface. This could be explained by the filling of the holes that were formed on the surface after the second and third deposition steps as

shown in Figure 2.2

To evaluate the surface topography, the monomer height sorting algorithm was used to select a monomer located at the furthest distance away from the surface covered by (10 × 10) bins in the xy-plane. The two-dimensional plot of this 10 × 10 matrix gives a surface topography that is analogous to Atomic Force Microscopy (AFM) measurements. The average thickness of the layers, , was calculated as the average value of the height distribution. The standard deviation of this distribution was averaged after equilibration to obtain the average value of the surface roughness at each deposition step. Figure 2.4a and 4b show the topography plots collected after the third and the fourth deposition steps. There are significant troughs present, indicating holes formed after the third deposition step. These holes are filled by newly incoming chains during the fourth deposition step. The multilayers rearrange and form a smooth surface after the fourth deposition step is completed. The surface roughness after completion of the third and fourth deposition steps was equal to 1.125σ and 0.974σ, respectively. The insets in

Figures 4a and 4b show the film height distribution function after the third and the fourth deposition steps. The presence of holes in these plots is manifested by the bimodal distribution as it is seen in Figure 2.4a. In Figure 2.4b, the holes are filled during the subsequent deposition step, as suggested by the single peak distribution function indicating a relatively smooth surface.

2.4.2 Effect of Charge Density of Polyelectrolytes

The dependence of polymer surface coverage, Γ, on the fraction of charged monomers on the polymer backbone at different deposition steps is shown in Figure 2.5.

As it follows, from this figure, the steady state regime is reached after completion of just the first few deposition steps for all polyelectrolyte systems with different charge fractions. This is indicated by the linear growth of the polymer surface coverage on the number of deposition steps. The average thickness of the adsorbed layer also increases linearly with the number of deposition steps for all studied systems (see inset in Figure

2.5). These linear dependences are observed despite the fact that two deposition steps are usually required to complete one layer. For partially charged chains with f = 1/2, the growth rate of polymer surface coverage is higher than for the case of fully charged chains. This is also in agreement with experimental observations of thicker layers for partially charged polyelectrolytes compared to very thin layers obtained for the fully charged polyelectrolytes, for which charge regulation was achieved by varying solution pH.151 In the case of partially charged chains, for each adsorbed charge there are extra

1/f-1 monomers added to the adsorbed layer; thus, allowing a larger number of monomer segments to be adsorbed per charged group in the underlying layer. For weakly charged chains with f = 1/3, however, although the surface coverage is higher than for fully charged chains, f = 1, and half charged ones, f = 1/2, the growth rate given by the slope is almost the same as that observed for systems with f = 1/2. The higher initial value of the surface coverage for adsorption of polyelectrolyte chains with f = 1/3 is due to surface effects where more chains are needed for surface charge compensation and overcharging.

After deposition of the first two layers, the growth rate for the systems with f = 1/3 is similar to those observed for polyelectrolytes with f = 1/2.

The distribution of polymer density ρ (z) during different deposition steps for chains with degree of polymerization Np = 32 and fraction of charged monomers f = 1

and f = 1/2 are shown in Figure 2.6a and 2.6b. These distribution functions were averaged separately for each set of adsorbed chains during different deposition steps for the duration of time required for deposition of the final layer. This procedure allows clear inspection of the interpenetration of the layers during the deposition process. There is a significant intermixing between polyelectrolyte chains adsorbed during different deposition cycles, though less so for the outermost layer. Despite such intermixing, a multilayered nature of the adsorbed polymeric film persists. This can be seen in Figure

2.6a inset, which shows the difference between local monomer densities of positively and negatively charged species. This plot clearly indicates the existence of alternating layers with excesses of positively or negatively charged polymeric components. The interpenetration between layers is enhanced for partially charged chains with fraction of charged monomers f = 1/2 as shown in Figure 2.6b. Close inspection of this distribution function reveals the presence of a broader polymer density distribution with lower magnitude.

The dependence of the density distribution function on the duration of the simulation run could potentially reveal whether the observed oscillations are kinetically trapped states or are representative of an equilibrium state. Figure 2.7 compares the density difference between positively and negatively charged chains for short (5×105 MD steps) and long simulation runs (3×106 MD steps) for the 12th deposition step. The density difference profile or the net charge distribution among the layers is remarkably close for both simulations suggesting that multilayers are an equilibrium state.

Interestingly, analysis of the density distribution ρ(z) after long simulation runs shows a

difference in the density profiles of the individual layers. In particular, the sharp peak of the 12th deposition step (outermost layer) seen in Figure 2.6a disappears for longer simulation run. The polyelectrolytes adsorbed during this deposition step diffuse further into the multilayers while keeping the density variations among the layers intact. This suggests that a dynamic exchange of the polyelectrolytes within the layers occurs by preserving the density difference between positively and negatively charged monomers among the layers.

The formation of ionic pairs between oppositely charged groups in the multilayers

152 is characterized by the charge-charge correlation function, g±(r), between positively and negatively charged monomers. This function gives the probability of finding a negatively charged monomer at a distance r from a selected positively charged one.

Figure 2.8 shows the charge-charge correlation function for all studied systems averaged after equilibration. All correlation functions have a maximum at a distance slightly greater than 1σ. This is approximately equal to the distance of the closest approach between monomers forming an ionic pair. The other distances at which such ionic correlations are enhanced—where these correlation functions show secondary and ternary peaks—are equal to 1.95σ and 2.8σ. Vertical reference lines show the peak positions of the charge-charge correlation function for completely stratified molecular layers of polyelectrolyte chains covering the whole surface. The completely stratified molecular layers were created with hexagonally packing of layers and analyzed for the charge- charge correlation function. The oppositely charged polyelectrolyte layers in this stratified film are shifted by a distance of σ/2 with respect to one another along the x-

direction. At least for the first three, these reference peaks are very close to the observed peaks, suggesting (by comparison) some structured layering of ion pairs. Although these peak positions are the same for all our systems, their magnitudes decrease with decreasing fraction of charged monomers on the polymer backbone. Another important feature seen in Figure 2.8 is the effect of the chain’s degree of polymerization Np on ion pair distribution. For fully charged chains (f = 1), there is no effect of chain degree of polymerization on the shape of the charge-charge correlation function. However, for partially charged chains with fraction of charged monomers f = 1/2, the charge-charge correlation function shows weak dependence on the chain degree of polymerization.

Finally, Figure 2.8 gives important information on the relative density of ion pairs for multilayers formed by chains with different fractions of charged monomers. Thus, ion pairs could dictate the formation, stability, interpenetration, and chain dynamics observed in the multilayers.

2.4.3 Effect of Chain Degree of Polymerization

The dependence of the polymer surface coverage, Γ, during the film growth on the chain degree of polymerization is shown in Figure 2.9. For fully charged chains, f =

1, the growth rate increases slightly with the chain length, while partially charged chains yield a weak decrease in growth rate with increasing chain length. This Np dependence can be explained by a combined effect of the discreteness of the net charge that each adsorbed chain delivers to the surface and the finite size effect of the simulation box.

Each adsorbed chain delivers a quantized amount of charge not necessarily optimal for surface overcharging. Shorter chains allow better charge adjustment for the finite

systems than the longer ones. This effect is even more pronounced for partially charged chains because, for each extra charged group, there are 1/f-1 uncharged ones added to the layer, leading to a larger difference in polymer surface coverage after each deposition step (see Figure 2.9). This is illustrated by the large difference in polymer surface coverage for weakly charged chains of monomer fraction f = 1/3 and degrees of polymerizations Np =32 and 16.

The chain degree of polymerization seems to have almost no effect on polymer density oscillations. Figure 2.10 shows the plot of the density difference between positively and negatively charged monomers for chain degree of polymerization Np = 8 and different charge fractions of monomers (f = 1 and 1/2). This remarkable similarity between the density oscillations for chains with different degree of polymerization and same charge fraction of monomers can be seen by comparing Figure 2.10 with Figure

2.6a and 2.6b inset. These density oscillations are also similar for chains with degree of polymerization Np = 16 (not shown). This suggests that the chain degree of polymerization has almost no effect on polymer density oscillations. Shorter chains have high diffusivity between layers as compared with that for longer chains due to the lower amount of ion pairs per chain. This high mobility inside multilayers formed by shorter chains still preserves the layered structure and the symmetric oscillations observed for the systems of longer chains. This is similar to the results obtained for longer simulation runs using chains having degree of polymerization Np = 32 as previously described in the discussion regarding Figure 2.7. The statement that longer simulation runs allowing further chain interdiffusion does not alter the periodic oscillation of the density difference observed within multilayers thereby suggesting an equilibrated structure is thus

reinforced.

Short chains with degree of polymerization Np = 8 show dynamic exchange during the deposition process between adsorbed polyelectrolytes and those in solution.

The frequency of chain exchange increases as the fraction of charged monomers on the polymer backbone decreases. This is another indication of the effect ion pairs have on chain dynamics. There are more ion pairs formed between oppositely charged chains inside multilayered films composed of strongly charged chains than in those of weakly charged ones. This leads to a higher energy barrier for adsorbed chains to overcome in order to escape from the adsorbed layer. This is supported by the fact that there is practically no exchange on the time scale of our simulation runs for fully charged chains with degree of polymerization Np = 32. The number of exchanged chains, however, grows to about 10% of the total number of adsorbed chains during the whole deposition

151 process, even for shorter chains with degree of polymerization Np = 8. Schlenoff et al. have reported a slow exchange of the adsorbed polyelectrolyte chains along with a kinetically reversible nature of the deposited layers. In addition to this observation it was observed that the probability of chain exchange decreases with increasing fraction of charged monomers on the polymer backbone as well as chain degree of polymerization.

It is worthwhile to note that, roughly, for each chain desorbed during the deposition step, an extra chain of the same type is simultaneously added, keeping the net value of layer overcharging nearly constant. This suggests that there is a simple relation between the overcharging and the number of charges adsorbed during each deposition step universal to all studied systems.

2.4.4 Theoretical Model of Multilayer Formation

The distribution of polymeric species inside the multilayered film (see Figures

2.6, 2.7 and 2.10) resembles the layered structure assumed by Castelnovo et al.153

However, this model does not account for strong effect of short-range interactions, and exclusion of counterions from the interior of the multilayered film established in our simulations. In light of the results described herein, the Castelnovo et al. model 153 can be modified to describe the polymer density profile and surface overcharging in the growing film (see Figure 2.11). In the concentrated mixture of positively and negatively charged chains, the oscillations of polymer density, Δρ, with the period, d, is a result of the competition between two opposing effects: polymeric and electrostatic. The excess of the polymeric part of the system free energy per period d due to the one-dimensional density wave along the z-direction of magnitude Δρ with respect to the average polymer density ρ is estimated as154

2 d σ 2 ⎛ dρ(z) ⎞ σ 2 Δρ 2 ΔF ≈ k TS dz ≈ k TS (2.5) pol B ∫ ⎜ ⎟ B 0 ρ(z) ⎝ dz ⎠ ρd

This polymer density wave induces charge density oscillations of smaller magnitude

(fΔρ). The one-dimensional charge density wave formed in a multilayered film can be

viewed as a system of parallel plate capacitors whose plates carry charge Q± ≈ ±efΔρSd , are of area S, and are separated by a distance d. The electrostatic energy of such a parallel plate capacitor is

2 3 U elect ≈ k BTSlB ()fΔρ d . (2.6)

(This estimate is only true in our case since the counterions are excluded from the adsorbed layer.)Thus, the optimal length scale of the density oscillation is obtained by minimizing polymeric and electrostatic contributions with respect to the period of oscillations d. This leads to

1 / 4 2 2 d ≈ ()σ / ρlB f . (2.7)

The period of density oscillations increases155 with decreasing fraction of charged monomers on the polymer backbone as f-1/2. This inverse square-root dependence of the period of density oscillations is in agreement with the 1.32-fold increase of the parameter d seen in our simulations for systems with f = ½, compared to that for systems of fully charged chains (f =1).

In the steady state regime, the magnitude of the density oscillations is controlled by the layer overcharging. During each deposition step, the film overcharging

ΔQads ≈ efΔρdS is obtained by balancing the energy of electrostatic repulsion between

unbalanced charges with the cohesive energy per monomer in a film − k BTε coh and the

interaction energy per monomer in a solution k BTε sol . These parameters ε coh and ε sol will be considered an adjustable parameters in the model. The energy of electrostatic

repulsion per excess charged monomer within a layer of excess charge ΔQads exposed to a solution with the Debye radius rD is estimated as

2 lB fΔρdrD U rep ≈ k BT ≈ k BTl B fΔρdrD . (2.8) rD

The surface overcharging stops growth when the energy of a chain with Np

monomers inside the overcharged region N p fU rep − k BTN pε coh matches (in order of the

magnitude) the chain’s energy in a solution k BTN pε sol . Thus, the magnitude of the polymer density oscillations is estimated to be

Δρ ≈ ε + ε / f 2l r d (2.9) ()coh sol ( B D ).

The magnitude of these polymer density oscillations causes the rate of change in the polymer surface coverage ΔΓ ≈ Δρd (i.e., the increase of the polymer surface coverage per each deposition step) to be proportional to

ΔΓ ≈ ε + ε / f 2l r ∝ f −3 / 2 (2.10) ()coh sol ( B D ) ,

−1/ 2 where rD ≈ ()4πlB fc is substituted for the Debye radius and c is the original monomer concentration in the solution. This expression can be used to estimate the ratio of this parameter for the partially and the fully charged systems,

ΔΓ(0.5)/ ΔΓ(1) ∝ 23/ 2 ≈ 2.8. In these simulations, this value is close to 2.95 for systems comprised of short chains with degree of polymerization Np =8 (see Figure 2.5). The system of the shortest chains is chosen for comparison since it is less susceptible to the finite size effects. This agreement is encouraging; however, to show that this theory is indeed capable of explaining the multilayer formation one has to consider the dependence of the polymer surface coverage rate on the cohesive energy and the Bjerrum length.

This will be further considered in Chapter 3.

2.5 Discussion and Conclusion

The molecular dynamics simulations of multilayers studied in this chapter support

the three-zone model describing the structure of a growing polymeric film. Zone 1 contains the layer in the vicinity of the surface which generally has one type of polyelectrolyte having a charge opposite to that of the surface (see Figures 2.3 and 2.6).

The thickness of this zone is on the order of one molecular layer. Zone 2 contains complexes (symplexes) of polyelectrolytes of opposite charges. The layers in this zone are highly interpenetrated and essentially exhibit 1:1 charge stoichiometry. It should be noted that in spite of the formation of such complexes, the alternating pattern of charge excess is still preserved in the multilayers for all studied systems (see Figure 2.10). The dynamic exchange of polyelectrolytes between multilayers, then, occurs in such a way that the local density distribution is preserved. Zone 3 includes the outermost layer along with the counterions, the latter of which forms an electrical double layer. The overall multilayer charge neutrality is thereby maintained by these counterions. Interestingly, these counterions are excluded from the interior of the growing film, which is in agreement with experimental observations. This outermost polymeric layer evolves during the deposition of oppositely charged polyelectrolytes during the next step by forming complexes with the incoming polyelectrolyte chains.

It is well-known that the local structure of the polyelectrolytes multilayers is similar to that of bulk polyelectrolyte complexes formed between oppositely charged macromolecules.156,157 This is also observed in our simulations where the conformations of the adsorbed chains inside layers are very similar to the structure of polyelectrolyte complexes in bulk solutions. Two different types of conformations were observed, however, with regard to the different types of polyelectrolyte complexes. The complex structure between ionic groups on the charged surface and polyelectrolytes adsorbed

during the first deposition step resembles that of a “ladder-like” complex.158 This is due to the fact that the surface charges are fixed and discrete and conformations of polyelectrolyte chains at the surface are close to rod-like ones that allow formation of well-organized “ionic” bonds in a ladder-like fashion between the charges on the surface and those on the polymer backbone. In contrast, the local structure of polyelectrolytes inside the multilayers and away from the surface is similar to a “scrambled-egg” complex158 structure where chains are more collapsed and intertwined. The transition between these two regimes occurs after the deposition of the first few layers (one to two steps in the case of our MD simulations). McAloney et al.159 have observed such transitions via AFM measurements of the first five layers where surface roughness increased with the number of deposited layers. The multilayers rearrange during the deposition process to form islands and holes that can better accommodate incoming chains. The formation of such islands and holes as seen in our simulations after the second deposition step (see Figure 2.2, 2.3, and 2.4) is due to the conformational rearrangements of the polyelectrolytes during the deposition process. Because chains adsorbed during the first deposition step favor formation of a “scrambled-egg” complex, the resulting partial shrinkage of the chains exposes the surface and leaves uncovered spots, or holes. These holes are filled after two deposition steps by polyelectrolytes of similar charge, completing the formation of densely packed layers as seen in Figure 2.3b.

Menchaca et al160 have observed the appearance of such polyelectrolyte-complex grains by liquid-cell AFM monitoring of the evolution of surface roughness during deposition of the first few layers.

Comparison of the simulation data with that obtained in the case of multilayer

formation on charged spherical particles52,53 shows that the layers formed at charged surfaces show a higher degree of ordering. The reason for this is the increase in available area around the spherical particle for adsorbing chains with each additional deposition step. Thus, the number of deposition cycles required for the completion of a single layer increases, giving rise to a high probability of mistakes in the layer structure. Even in the case of adsorption at a flat surface, considerable intermixing between chains deposited during subsequent steps still occurs. Such intermixing between layers has been neglected in previously proposed theoretical models of multilayer assembly.45,161 As already mentioned, several deposition steps are required to complete the formation of each layer.

Nevertheless, the alternating pattern of surface charge excess is maintained for partially charged chains though the fluctuation amplitude is reduced due to chain interpenetration.

Short-range interactions play an important role in multilayer formation. This observation is in agreement with Kotov’s results162 which described the contribution of hydrophobic interactions between polyelectrolytes and the charged surface, identifying them as the important factor determining the ability of the compounds to self assemble via LbL assembly. Our simulation results are also in agreement with previous molecular simulations of layer formation near charged planar surfaces49 and charged spherical particles50,52,53 revealing the requirement of an extra short-ranged interaction in order to achieve successful polymeric film growth. The importance of the short-range interactions in the multilayer structure and properties will be further addressed in Chapter

In conclusion, molecular dynamics simulations of layer-by-layer deposition of

polyelectrolytes from a dilute solution was performed in order to study the effects of the fraction of charged monomer as well as the chain degree of polymerization on the structure, stability, and mechanism of the multilayer formation. Polyelectrolyte chains in multilayers form “scrambled-egg” complexes of intertwined chains, thus increasing the amount of polymer adsorbed during each deposition step. For all of our simulations, approximately two deposition steps of similarly charged polymers were necessary to complete formation of a single compact layer. The polyelectrolyte chains are not perfectly stratified within the multilayered structure but, instead, there is intermixing between polyelectrolyte chains deposited during different depositions cycles. There are almost perfect periodic oscillations of density difference between positively and negatively charged chains after several deposition steps, despite the high degree of chain intermixing. Weakly charged chains allow significant larger polymer encapsulation within layers than the strongly charged ones.

1.6

nd 1.4 2 Layer 1.2

1.0 st 1 Layer 2 0.8 Γσ

0.6

0.4

0.2

0.0 0 5 10 15 20 25 30 6 MD steps / 10

Figure 2.1: Dependence of the polymer surface coverage (Γσ2) on the number of MD integration steps for system with Np=32 and f=1 during the first and the second deposition step for duration of 3.0 × 106 integration steps in each deposition step.

Planar Charged Surface, 0 MD 1st Step 2nd Step

rd th th 3 Step 4 Step 5 Step

Figure 2.2: Evolution of the layer structure during the adsorption of fully charged

(f=1) polyelectrolytes, with degree of polymerization Np=32. Each snapshot is taken after the completion of deposition cycles from 1 through 5 with unique color coding for each step being maintained from one snapshot to the next. For example, the blue chain in the 5th step snapshot is polymer adsorbed originally during the 1st step.

Error!2.0 0.4 (a)

1.5 0.3 3 σ 3 (z) σ 1.0 0.2 (z)

ρ counterion ρ 0.5 0.1

Error!0.0 0.0 02468 z/σ 2.0 0.4 (b)

1.5 0.3

3 σ 3 (z) σ 1.0 0.2 (z) ρ counterion ρ 0.5 0.1

0.0 0.0 02468

z/σ

Figure 2.3: Density profiles of the negatively ρ-(z) (continuous line) and positively

ρ+(z) (dashed line) charged monomers for system with fully charged (f=1)

rd polyelectrolyte chains with Np=32 after completion of (a) 3 deposition step and (b)

4th deposition step with a duration of 5×105 MD steps each. The density profile of positively (circles) and negatively (triangles) charged counterions are also shown in secondary axis.

0.10 0.08 0.06 0.04 0.0 frequency 0.02 3.0 0.5 0.00 1.0 2.5 1.5 0123456 2.0 2.0 h/σ 2.5 3.0 1.5 h/ 1.0 0.5 0.0 2 4 Y 6 2 8 4 6 8 10 X 0.10 0.08 0.06 0.04 0.0 frequency 0.02 3.0 0.5 0.00 1.0 2.5 0123456 1.5 2.0 2.0 h/σ 2.5 1.5 3.0 h/ 1.0 0.5 0.0 2 4 Y 6 2 8 4 6 8 10 X

Figure 2.4: Topography plots of the film height distribution for the system of fully

5 charged chains with f=1 and degree of polymerization Np= 32 at the end of 5×10

MD steps after the completion of 3rd deposition step (a) and 4th deposition step (b).

The insert shows the height distribution of the main plots.

12 12 10

σ 8 10 6

/ 4 8 2 0 123456789101112

2 Nstep 6 Γσ

0 123456789101112

Nstep

Figure 2.5: Dependence of the surface coverage (Γσ2) on the number of deposition steps for polyelectrolyte chains of the degree of polymerization Np= 32 with different fraction of charged monomers f=1 (circles) and f=1/2 (triangles) and f=1/3 (squares).

The inset shows the dependence of average thickness on number of deposition steps.

(a) 2.0 3

σ )

z 1.0 (

Δρ 0.0 0.8 -1.0 024681012 0.6 3 z/σ

) ) 0.4

z ( ρ 0.2 12 0 1 0.0 8 6 2 ep 4 4 st 6 2 N 8 10 z/σ 12 0 (b)

2.0 3

) 1.0 z 0.8 ( Δρ 0.0

0.6 -1.0 3 024681012 σ z/σ

) 0.4

z ( ρ 0.2 2 1 0.0 8 10 2 6 4 4 p 6 8 2 N ste 10 12 0 z/σ Figure 2.6: Density profiles of the fully charged (f=1) (a) and partially charged

(f=1/2) (b) polyelectrolyte chains with Np=32 after completion of 12 deposition cycles with duration of 5×105 MD steps each. Insert show the difference between the corresponding uniaxial monomer densities of positively and negatively charged

chains, Δρ(z) = ρ − (z) − ρ + (z) .

2.5

2.0 1.5 3

σ 1.0

) z 0.5

Δρ( 0.0 -0.5

-1.0 0246810

z/σ

Figure 2.7: Comparison of the density difference of positively and negatively

charged monomers, Δρ(z) = ρ − (z) − ρ + (z) for two different lengths of simulation runs. (i) 5×105 MD steps (circles) and (ii) 3× 106 MD steps (continuous line).

(r) 15 +,- g

0 0123456

r/σ

Figure 2.8: Ion pair correlation functions for chains with different fractions of charged f=1(closed symbols) and f=1/2 (open symbols) and degrees of polymerization, Np = 32 (circle), Np =16 (triangle) and Np =8 (square). The vertical reference lines show the peak positions of the perfectly stratified molecular layers of positively and negatively charged ions.

14 f=1 Np=32 f=1 Np=16 12 f=1 Np=8 f=1/2 Np=32 10 f=1/2 Np=16 f=1/2 Np=8 f=1/3 Np=32 8 f=1/3 N =16 2 p

Γσ 6

0 123456789101112

Nstep

Figure 2.9: Dependence of the surface coverage (Γσ2) on the number of deposition steps for polyelectrolyte chains with different fraction of charged monomers and chain degree of polymerization.

2.0 2.0 (a) (b) 1.5 Error!1.5 1.0 3 3

σ 1.0

σ 0.5 z) (z) (z) Δρ

Δρ( 0.5 0.0

-0.5 0.0

-1.0 -0.5 02468101214 0 2 4 6 8 10 12 14

z/σ z/σ

Figure 2.10: Comparison of density difference of positively and negatively charged

monomers, Δρ(z) = ρ − (z) − ρ + (z) for chains degree of polymerization Np= 8 and different fraction of charge monomers (f = 1 and 1/2).

1.4

1.2

3 1.0

z) z) 0.8 ρ( 0.6

0.4

0.2

0.0 024681012 z/σ

Figure 2.11: Density profiles of the negatively ρ-(z) (continuous line) and positively

ρ+(z) (dashed line) charged monomers for system of fully charged (f=1)

th polyelectrolyte chains with Np=32 after completion of 12 deposition step with

duration 5×105 MD steps each. The density profile of positively (circles) and negatively (triangles) charged counterions are also shown.

CHAPTER 3

3 Effect of Electrostatic and Short Range Interactions

on the Build-Up of Polyelectrolyte Multilayers*

3.1 Synopsis

In this Chapter, the effect of the strength of electrostatic and short-range interactions on the multilayer assembly of oppositely charged polyelectrolytes at a charged substrate is studied by molecular dynamics simulations. The multilayer build-up was achieved through sequential adsorption of charged polymers in a layer-by-layer fashion from dilute polyelectrolyte solutions as described in Chapter 2. The strong electrostatic attraction between oppositely charged polyelectrolytes at each deposition step was found to be a driving force behind the multilayer growth. The strength of electrostatic and short-range interactions was systematically varied to study their effect on the multilayer build-up, internal structure, intermixing, and ion-pair formation. It is shown that the polymer surface coverage and multilayer structure are each strongly influenced by the strength of electrostatic and short-range interactions. The multilayer systems with additional short-range interactions among monomers (or ‘hydrophobic’ polyelectrolytes or poor solvent conditions) yield a higher surface coverage and polymer density inside the multilayers. On other hand, the multilayer systems with weaker short- range interactions have lower surface coverage and density inside multilayers. The

results are rationalized in the framework of a scaling model that takes into account the

* Reproduced, in part, with permission from Patel, P. A.; Jeon, J.; Mather, P. T.; Dobrynin, A. V. Langmuir 2006, 22,

strength of electrostatic and short-range interactions and charge fraction of polyelectrolytes. Finally, analysis of the multilayer systems showing a stable growth reveals a constant universal value of the 50 % overcharging related to the total number of charges adsorbed during the deposition step. The rest of the chapter is organized as follows; Section 3.3 briefly describes the simulation and interactions parameters. Section

3.4 presents the results and discussion that include the effect of electrostatics and short- range interactions on the growth of polymer surface coverage, growth, stability and universality for the stable multilayer system. Section 3.5 presents scaling model and

Section 3.6 describes the conclusion of the present work.

3.2 Introduction

Refer Section 1.1 of Chapter 1

3.3 Simulation Methods and Interaction Parameters:

The MD simulations were performed using the simulation method described previously in section 2.2 of Chapter 2. Multilayer formation was studied by systematically varing electrostatic and short-range interactions. The electrostatic

2 interactions were varied through the value of the Bjerrum length, lB = e εk BT while the

short-range interactions were varied through Lennard-Jones (LJ) parameter, ε LJ , and

rcut on a truncated-shifted LJ potential were used to control the solvent quality for the polymer backbone, and hydrophilicity or hydrophobicity of the adsorbing surface. For example, the LJ interaction parameter εLJ = 1.0 results in poor solvent conditions due to the increased monomer-monomer LJ interactions. Similarly, the LJ interaction

parameters of εLJ = 0.3 describes the θ-solvent condition for the polymer backbone. The hydrophilicity or the hydrophobicity of the adsorbing surface was modified by the cut off

1/6 distance, rcut . The cut-off distance of rcut = 2 σ modifies the LJ potential in a manner that yields net repulsive interactions between the chain segments (monomers) and the

surface, while a larger rcut = 2.5σ results in an attractive LJ potential that corresponds to a hydrophobic surface. We note that previous simulation results described in Chapter 2 concluded that a short-range interaction between the surface and the oppositely charged chains is required for the successful overcharging of the surface and the multilayer growth. Thus, the interaction between surface and the oppositely charged chains in all the simulations described in this chapter corresponds to the same as described in Chapter

2. However, the interactions between the surface and the same charged monomers as surface have been varied in this chapter to study the effect of short-range interactions on the multilayer growth and structure. Table I summarizes the systematic variation of the short-range and electrostatic interactions parameters used in the simulations study. As summarized in Table 3.1, the polymer-polymer LJ-interaction parameters for the Systems

A and B are close to those for θ-solvent conditions for the polymer backbone, while the

LJ-parameters for the Systems C and D correspond to poor solvent conditions for polymer backbones with a negative value of the monomeric second virial coefficient.

Charged polymers in poor solvent conditions for the polymer backbone are also called hydrophobic polyelectrolytes. In the Systems A and B there is an additional short-range repulsion between positively charged chains and the substrate. Systems A and C have

higher electrostatic interactions, lB = 3.0σ, compared to the system B and D where lB =

1.0σ. It is to be noted that the simulation box size in XY direction is similar for all systems except for System A, which has double surface area in the XY plane. This was increased in order to limit the finite-size effect of the simulation box for System A. The system sizes are summarized in Table 3.2.

3.4 Results and Discussions

3.4.1 Growth of Polymer Surface Coverage

The polymer surface coverage, Γ , was found to increase with each deposition step with a trend that depended on the fraction of charged monomers on the polymer backbone and values of the interaction parameters Figure 3.1 (a-d)). These plots clearly indicate that the strength of the polymer-polymer interactions play an important role in the successful film growth. For the systems with short-range interaction parameter between monomers close to that at the θ-point for the polymer backbone (Systems A and

B) the growth in polymer surface coverage with each step was found to feature a strong

Np-dependence. For partially charged polyelectrolyte chains with the fraction of charged monomers, f = ½, only the longest polymer chains (Np =32) show film growth (see Figure

3.1 (a, b)). A strong effect of short-range repulsion is also observed for fully charged chains but it is less pronounced than for the systems of partially charged polyelectrolytes.

Furthermore, the system with higher electrostatic interaction, lB = 3.0σ, (System A) has a higher growth rate compared to the systems with lower electrostatic interaction, lB = 1.0σ

(System B). Thus, the observed trend in the growth rate (Figure 3.1(a-b)) is the result of competition between short-range repulsion and electrostatic attraction between oppositely

charged chains. The polyelectrolyte adsorption is then governed by the competition between the net chain adsorption energy by ion-pair formation and the short-range repulsion among the adsorbing surface and the polyelectrolyte. For shortest chains with

Np = 8 the magnitude of the chain adsorption energy is four times lower than that for the longest chains with Np = 32, making chain desorption more probable. Similarly, for the partially charged polyelectrolytes, f=1/2, the magnitude of the chain adsorption energy is halved, making polyelectrolyte desorption more probable. Such desorption of the polyelectrolytes lowers the amount of overcharging and eventually causes unstable multilayer growth. This can be seen in Figure 3.1(a,b) where systems A and B, f=1/2 and

Np =16 and 8 do not show any growth of surface coverage.

In comparison to systems A and B, hydrophobic polyelectrolytes (Systems C and

D) form stable films (see Figure 3.1(c, d)). In these cases, the steady state regime was reached after completion of just the first few deposition steps, regardless of charge fraction and degree of polymerization. This is supported by the linear growth of the polymer surface coverage with number of deposition steps. The additional hydrophobic interactions (εLJ = 1.0 vs. εLJ = 0.3) enhance affinity between oppositely charged polyelectrolytes, strengthening chain associations within the multilayers. For partially charged chains with f = 1/2, the growth rate of polymer surface coverage was found to be higher than for the case of fully charged chains. In the case of partially charged chains, for each adsorbed charge there was one extra monomer added to the adsorbed layer that results in a increase in surface coverage, as was revealed in the previous chapter.

3.4.2 Distribution of Polymer Density

100

A density profile of monomers belonging to positively and negatively charged chains in the multilayers is shown in Figure 3.2. All data were collected during the eighth deposition step for chains with Np=32 and f=1. The polymeric systems with weak short-range interactions (Systems A & B, Figure 3.2 (a, b)) show a less pronounced stratification of oppositely charged polyelectrolytes than systems with additional short- range attractive interactions (Systems C & D, Figure 3.2 (c, d)). The monomer density of negatively charged chains for systems A and B, ρ-(z), shows two peaks near 1σ and 3σ.

The well developed peaks in the density profile of positively charged chains, ρ+(z), are located at 2σ and 4σ. The first peak near the surface is clearly larger than the other peaks.

This is due to the high surface charge density of the initial substrate in comparison with the value of the surface overcharging achieved after completion of each deposition step.

A larger number of adsorbed polyelectrolyte chains are required during the first deposition step to compensate for the surface charge while also overcharging the surface for subsequent layer build-up. A similar trend in higher polymer surface coverage of the surface layer is seen for hydrophobic polyelectrolyte systems as well (Systems C & D;

Figure 3.2 (c, d)). However, these systems show more layers and better stratification between positively and negatively polyelectrolytes in comparison with systems A and B.

Thus, additional attraction between polyelectrolyte segments leads to better-organized multilayered films. Such improved stratification between layers leads to the formation of a larger number of well-defined layers after completion of the same number of deposition steps for systems C and D.

The film composition, shown in Figures 3.2, supports the three-zone structure of

101

the multilayer film described in Chapter 1. Zone I contains the layer in the vicinity from the adsorbing surface with excess of molecules carrying a charge opposite to that of the substrate. The thickness of this layer depends on the electrostatic and short-range interactions between polyelectrolyte chains and substrate. Zone II contains complexes of oppositely charged macromolecules. Inside this zone, polyelectrolytes are well intermixed and show 1:1 charge stoichiometry. This zone is thicker for hydrophobic polyelectrolytes that also show sharper boundaries between different layers. The growth of the film occurs by increasing the thickness of the zone II. Detailed analysis of the average density within the multilayers in zone II by scaling arguments is presented below in Section 3.4.1. Zone III includes the outmost layer along with counterions, which neutralize the excess charge in the growing polymeric film. The counterions diffuse further into the polymeric film for the systems with repulsive short-range interactions

(see Figures 3.2 a,b). This correlates well with the lower polymer density inside the film.

On the contrary, hydrophobic systems (Figures 3.2 c,d) feature a counterion density profile that is narrow and has a large peak magnitude located just outside the ridge of the polymeric film. The exclusion of counterions from the film interior is a result of the higher polymer density and relatively lower free volume available for counterions inside multilayered films assembled from hydrophobic polyelectrolytes.

The charge-charge correlation function, p(r), between the oppositely charged monomers for all the systems with Np =32 and f =1 is shown in Figure 3.3 (a,b). The values from this function are proportional to probability of finding a negatively charged monomer at a distance r from a given positively charged one or vice-versa. This function is same as depicted in Figure 2.8 of the previous chapter that deduced the formation of

102

ionic pairs between oppositely charged monomers in the multilayers. Similar to Figure

2.8, all correlation functions have a peak at a distance slightly greater than 1σ. Other secondary peaks at 1.95σ and 2.8σ signify a stratification between the layers as was discussed previously in chapter 2. As seen in Figure 3.3a, the multilayers formed from poor solvent conditions of the polymer or ‘hydrophobic’ polyelectrolytes (System C), has increased magnitude of the peak at 1σ compared to the multilayer system with weak short-range interactions (System A). For the hydrophobic polyelectrolyte multilayer systems (C and D), the magnitude of peak at 1σ is higher for the system C compared to system D. The higher magnitude of the peak at 1σ is due to the higher electrostatic interactions for system C (lB =3.0σ) compared to those of system D (lB =1.0σ). Thus, increasing the electrostatic or short-range attractive interactions leads to increase in the magnitude of the first maximum around 1σ in the charge-charge correlation function.

This corresponds to increased ion-pair formation in such systems. The observed increase in ion-pair formation correlates with the better stratification of the layers observed for hydrophobic polyelectrolyte systems as evident in Figure 3.2 (c-d) compared to Figure

3.2 (a-b). The magnitude of the first peak decrease with decreasing fraction of charged monomers on the polymer backbone consistent with the previous observations in Figure

2.8, further corresponding to lower amount of ion-pair formation. This also corresponds to the observation of comparatively less stratification of the multilayers in Figure 3.2 for partially charged chains. Thus, the claim that ion-pair formation dictates multilayer formation and stability is reinforced.

The density difference between the positively and negatively charged monomers

103

for fully and partially charged polyelectrolytes with degree of polymerization Np=32 for system A through D is shown in Figure 3.4 (a,d). This allows direct observation of the stratification of the oppositely charged layers within the multilayers. The multilayers with weak short-range interactions (Systems A and B) have less pronounced stratification compared to multilayers with additional strong LJ short-range interactions (System C and

D). Interestingly, the period of oscillations, d, of the positive and negative charge along the surface is different for the fully and the partially charged polyelectrolytes for the multilayers, with the later having higher d. This dependence of the period of density oscillations, d, in the multilayered film on the fraction of charged monomers f can be rationalized in the framework of the scaling model presented in Section 3.5.2.

3.4.3 Universality of the Film Growth

To achieve steady state (linear) LbL growth, the layer should be overcharged by the same amount at each step to recreate the surface properties. The steady state is then achieved due to the same amount of polyelectrolytes adsorption at each deposition step

(linear growth) that results in constant value of surface overcharging. Indeed, this surface recreation was observed in all simulations, indicating universality among the systems studied. This universality of the overcharging process for systems showing steady state film growth is shown in Figure 3.5, where we plot the ratio of the absolute value of the layer overcharging, |ΔQ|, to the net charge carried by adsorbed chains at a given deposition step, Qads=f(N(s)-N(s-1)) (where N(s) is the total number of adsorbed monomers after completion of the s-th step), versus the number of deposition step Nstep.

This quantity is relatively independent of the fraction of charged monomers on the

104

polymer backbone as well as the chain degree of polymerization. After several deposition steps, when the processes reach a steady state, this ratio approaches a value of

1/2 for all studied systems. Thus, for multilayer growth with constant overcharging, one charge is needed per each excess ionic charge on the multilayer to compensate for the surface charge while another is needed to recreate the surface properties for the adsorption of the next layer. Note that if this ratio is smaller than 1/2 the film eventually stops growing, while if it is more than 1/2, layer mass will show exponential growth. In both cases, the growth process is unstable. Fluctuations around the saturation value of ½ for |ΔQ|/Qads can be attributed to corresponding fluctuations in the number of adsorbed chains and should decrease with increasing system size. It is important to point out that surface overcharging plays two roles. First, it recreates the surface properties (primarily charge) for the next deposition layer and, second, it prevents the unrestricted growth of adsorbing polymers through electrostatic interactions between excess charges.

3.4.4 Stability of the Growing Film and Chain Exchange

To study film stability and chain exchange during multilayer assembly a longer molecular dynamics simulations of the System A with Np=16 and f=1/2 was performed.

The selection of this system was dictated by the fact that it shows initial film growth that stops after the completion of the eighth deposition step (see Figure 3.1a). Thus, this system demonstrates both the stable film growth at the initial stages of the deposition process and unstable film growth, with saturation in Γσ2 values, at the later stages. The initial configurations for these simulations were the final configurations of the simulation runs after completion of the second, fourth, sixth, and eight deposition steps. These

105

simulations were continued for an additional 4 × 106 MD steps for the second deposition step and 9 × 106 MD steps for all other deposition steps. The time dependence of the polymer surface coverage during these longer simulation runs is shown in Figure 3.6. For the second and fourth deposition steps the polymer surface coverage fluctuates around an average value. The fluctuations increase in magnitude for the fourth deposition step in comparison with those during the second deposition step. Upon close inspection, we observed that this increase in amplitude of fluctuations is due to a large number of the negatively charged loops and chain’s ends dangling into solution after completion of the third deposition step. The fluctuations in number of contacts between these loops and positively charged chains in a solution are responsible for the variations in the polymer surface coverage seen in Figure 3.6.

A qualitatively different picture for the time-dependence of polymer surface coverage is observed during the longer simulations of the sixth and eighth deposition steps. For these deposition steps, the polymer surface coverage not only shows oscillations but also gradually decreases as the simulation runs continue. This decrease is associated with desorption of the negatively charged chains, which were adsorbed during the previous deposition steps. The desorbed negatively charged chains can be seen in

Figure 3.6 (blue beads) that presents a snapshot of the simulation box during the extended simulation run of the eighth deposition step. It is interesting to point out that these negatively charged chains form complexes with positively charged chains in a solution.

These could be either 1:1 or 1:2 complexes. Furthermore, desorption of negatively charged chains occurs dynamically in conjunction with positively charged ones. The double chain desorption process has lower activation barrier (shown below in Section

106

3.5.4) than desorption of a single negatively charged chain. Note that the single chain desorption events are still possible but a desorbing negatively charged chain will immediately form a complex with positively charged one. This happens already during the escape process when part of the negatively charged chain is still buried inside the film.

3.5 Scaling Model

3.5.1 Average Multilayer Density

The average polymer density inside zone II – low for systems A and B and high for systems C and D – is controlled by the fine interplay between fluctuations/ correlation-induced attraction between oppositely charged chains and excluded volume interaction between monomers.163,164 To understand this, a concentrated polymer solution with correlation length, ξ is considered. At length scales smaller than the solution correlation length the chain statistics is unperturbed by fluctuation/correlation- induced attractive interactions resulting in the usual scaling relation between the correlation length ξ and g, the number of constituent beads (monomers) in a correlation

“blob”, ξ ≈ σg 1/ 2 , where σ is the bead diameter. For length scales larger than the correlation length, attractive interactions cause dense packing of the correlation blobs.

The local structure of the melt of blobs resembles that of a concentrated solution with each blob being surrounded by the oppositely charged blobs with higher probability.

This structure of the adsorbed layer is supported by the charge-charge correlation function p(r) between positively and negatively charged monomers shown in Figure 3.3.

This function is proportional to probability of finding a negatively charged monomer at a

107

distance r from a selected positively charged one.

The electrostatic interaction between any two neighboring oppositely charged blobs separated by a distance ξ is of the order of the thermal energy kBT

l f 2 g 2 l f 2 g 3 / 2 − k T B ≈ −k T B ≈ −k T (3.1) B ξ B σ B

This leads to the number of monomers in a blob and its size to be equal to

−2 / 3 −1/ 3 g ≈ ()uf 2 , and ξ ≈ σ (uf 2 ) (3.2)

where u is the ratio of the Bjerrum length lB to the bond length σ, u=lB/σ. With increasing Bjerrum length (and thus increasing the value of the parameter u), the blob size decreases which is manifested in Figure 3.3 as growth and sharpening of the first maximum in the correlation function, p(r).

The correlation blobs inside the film are space-filling, leading to the following expression for the average polymer density:

1/ 3 ρσ 3 ≈ σ 3 g /ξ 3 ≈ ()uf 2 (3.3)

The polymer density inside each layer increases with increasing strength of electrostatic interactions. Note that the scaling analysis presented above can only be applied to describe average film density for the systems A and B for which the parameters of the LJ-potential are close to that for a θ-point. Comparison of the ratio of the average polymer density in the middle of the polymeric film for Systems A and B with lB=3σ ( ρ(1) / ρ(0.5) ≈ 1.48) and lB=1σ, (ρ(1) / ρ(0.5) ≈ 1. 52 ) that are close to the

108

ratio ρ()1 / ρ (0.5 )=22/3=1.59 obtained from the equation (3.3).

For systems C and D the parameters of the LJ-interactions correspond to poor solvent conditions for the polymer backbone. These systems do no show a strong effect of the Bjerrum length on average polymer density such that for both systems it is close to

ρσ3=0.53 (Figure 3.2 (c,d)). This indicates that LJ-interactions rather than fluctuation/correlation induced attractive interactions control the average polymer density inside the polymeric film.

3.5.2 Charge Density Oscillations

The charge density oscillations (in space) of the polymer composition in a concentrated mixture of positively and negatively charged chains (such as our multilayers

(Zone II) see Figures 3.4) are a result of competition between polymeric and electrostatic effects. Applying analysis similar to the one describe in Chapter 2 (see Equation 2.5, 2.6 and 2.7), one obtains the following expression for the period of oscillations d.

−1/ 6 −1/ 3 1/ 4 ⎧σu f , Systems A & B d 2 / l f 2 (3.4) ≈ ()σ ρ B ≈ ⎨ −1/ 4 −1/ 2 ⎩σu f , Systems C & D

Thus, the period of density oscillations inside Zone II increases with decreasing fraction of charged monomers on the polymer backbone as f -1/3 for the systems A&B and as f-1/2 for the systems C&D, and decreases with increasing the strength of the electrostatic interactions, the value of the parameter u. System A which has an increase in the period d by factor 1.3 for system with f =1/2 in comparison with that for a system of fully charged chains (Figure 3.4). This increase is in agreement with predictions of the

109

Equation. 3.4. For systems C and D the average polymer density ρ is a constant and does not depend on f. In this case the period d is proportional to f−1/2. This inverse square-root dependence of the period of density oscillations is in agreement with factor 1.32 increase of the parameter d seen in our simulations for system D with f = 1/2 in comparison with that for system of fully charged chains.

3.5.3 Multilayer Growth

The stepwise rate of change of polymer surface coverage in the multilayer depends on the net overcharging of the surface achieved during the previous deposition step. The surface overcharging during each deposition step can be evaluated by using the following simple scaling arguments. During each deposition step (excluding the initial layer growth where the polymer surface coverage is controlled by the interactions with adsorbing substrate), the growing polymeric film is overcharged by the amount

ΔQ ≈ efΔρSd (excess charge of Zone III). This excess charge is screened by counterions

at the length scale of the order of the Debye screening length rD. The excess charge ΔQ and the neutralizing diffusive layer of counterions can be viewed as a parallel plate capacitor with a gap size of the order of the Debye screening length. The electrostatic energy of such capacitor is equal to:

2 U III lB ()fΔρdS rD 2 ≈ ≈ lB ()fΔρd SrD . (3.5) k BT S

The energy of electrostatic repulsion per excess charged monomer within the overcharged region is estimated as:

110

U m 1 U III ≈ ≈ lB fΔρdrD (3.6) k BT fΔρdS k BT

For a polymer chain with Np monomers the total energy of a chain in this overcharged region is equal to the sum of the repulsive energy fNpUm and chain cohesive energy

− k BTN pε coh that is due interaction between a chain and its surroundings. The cohesive energy depends on the strength of the electrostatic and LJ interactions. For systems A and

B the attraction between oppositely charged chains is controlled by correlation/fluctuation induced attractive interactions and chain cohesive energy is of the

2 2 / 3 order of the thermal energy kBT per each correlation blob, − k BTN p / g ≈ −k BTN p (uf ) .

For systems C and D there are two contributions to the chain cohesive energy. The first one is due to short-range attractive LJ-interactions and another one is due to electrostatic

interactions. The first contribution is proportional to − k BTN pε LJ and the electrostatic contribution is on the order of the energy of electrostatic attraction between oppositely charged monomers separated by a typical distance ( fρ)−1/ 3 ≈ σf −1/ 3 that can be estimated

−1/ 3 4 / 3 as − k BTN p fl B /()fρ ≈ −k BTN puf . Note that all evaluations of the cohesive energy are done on the scaling level up to a numerical prefactor.

Chain’s adsorption that leads to surface overcharging ceases to occur when the energy of a chain with Np monomers inside the overcharged region,

k BTN pU m − k BTN pε coh , becomes comparable (in order of magnitude) to the same chain’s

energy in a solution k BTN pε sol . Note that in the framework of the scaling model of a

2 2 / 3 polyelectrolyte the energy per monomer in dilute solution εsol is on the order of (uf ) .

111

Thus, the stepwise rate of change in the polymer surface coverage ΔΓ ≈ Δρd is equal to:

⎧ 2 2 / 3 ()ε coh + ε sol −3 / 2 −1/ 2 ⎪ (uf ) , Systems A & B ΔΓ ≈ ∝ f u ⎨ (3.7) f 2l r 4 / 3 . B D ⎩⎪ε LJ + uf , Systems C & D

−1/ 2 where rD ≈ ()4πlB fc is the Debye radius and c is the original monomer concentration in the simulation box.

We can use expressions Equation (3.7) to compare the ratio of the rates of change of the polymer surface coverage ΔΓ for Systems B and D with partially and the fully charged chains. For System B this ratio is equal to ΔΓ(0.5) / ΔΓ(1) ∝ 21/ 6 ≈ 1.12 and for

System D - ΔΓ(0.5) / ΔΓ(1) ∝ 23 / 2 ≈ 2.8. In our simulations these values are close to 1.18 and 3 respectively. Thus, simulations show reasonable agreement with this scaling model.

3.5.4 Chain Desorption

There is a simple explanation for why two-chain desorption process is more favorable than a single chain event. For single chain desorption, the activation barrier that chains should overcome to escape from the polymeric film is proportional to the absolute value of the chain’s cohesion energy:

(1) EkTNact≈ B pε coh (3.8)

This activation energy is proportional to the chain’s degree of polymerization because, by desorbing, a chain eliminates all favorable contacts with its surroundings inside the multilayered film. However, by desorbing in pairs, chains form a 1:1 complex whose interior structure is similar to that inside multilayered film. Thus, only monomers

112

located on the surface must break favorable attractive interactions. The activation energy for this favored process is proportional to the number of monomers on the surface of the

complex, Ns, times the absolute value of the cohesive energy per monomer, εcoh:

(2) 2/3 EkTNkTNgact≈≈ B sε coh B(/) p (3.9)

In writing equation (3.9) scaling model described earlier in section 3.5.3 is used.

The activation energy of the two-chain process is lower than the single chain one, Ns

Eqn. 3.9 also explains why a single chain desorption is always accompanied by complexation with an oppositely charged chain in a solution. Instead of eliminating all favorable contacts, the desorbing chain only loses part of them, Ns, by recreating the rest of them through complexation in solution with an oppositely charged chain.

Let us now estimate Np-dependence of the characteristic time-scale for chain desorption. In our simulations we used the Langevin thermostat to control the system temperature. Molecular dynamics simulations with the Langevin thermostat correspond to the Rouse chain’s dynamics.165 In this case, the chain relaxation time in a solution is

2 proportional to Np so that the characteristic time-scale for chain desorption is estimated as:

2 2 / 3 τ des ≈ τ 0 N p exp(()N p / g ) (3.10)

where τ0 is a characteristic monomeric time scale. Thus, the stable multilayer growth shown in Figure 3.1 could be argued to be a result of a slow chain desorption process that only happens for relatively short chains with weak attractive interactions. With increasing number of ionized groups, chain degree of polymerization, and chain hydrophobicity,

113

chain desorption is slowed, favoring the formation of stable multilayered structures.

3.6 Conclusions

A molecular dynamics study of the effect of short-range and electrostatic interactions on sequential multilayer assembly at charged surfaces has been presented in this Chapter. The simulations confirm our hypothesis that surface overcharging is crucial for stable film growth. Furthermore, steady state multilayer growth strongly depends on the strength of the electrostatic and Lennard-Jones interactions. Those systems with LJ- interaction parameters close to θ-conditions for the polymer backbone (Systems A&B) only show stable layer growth for systems with sufficiently strong chain cohesive energy.

This is indicated by the stable film growth for the System A with Np=32 and f=1; 1/2;

Np=16 and f=1 and for the System B with Np=32 and f=1; 1/2. For shorter polymer chains, the activation barrier against chain desorption is sufficiently low to allow frequent chain desorption events. Interestingly, we observed that polyelectrolyte chains desorb in pairs and show that 1:1 complex stoichiometry minimizes the number of favorable ionic and monomer-monomer interactions to be broken during desorption.

In poor solvent (hydrophobic) conditions for the polymer backbone (Systems

C&D), the additional attractive LJ-interactions improve film stability, resulting in steady state multilayer growth for all studied chain lengths. By improving the affinity between polymer chains, the activation barrier against chain desorption is increased. Furthermore, additional affinity between polymer backbones improves layer stratification. Systems C and D witnessed faster growth (steeper slope) than Systems A and B. Irrespective of interactions, partially charges chains were seen to allow higher polymer surface coverage

114

than the fully charged ones. Within the formed multilayers, positively charged monomers are surrounded by negatively charged monomers (see Figures 3.3 a, b). This charge distribution is similar to the charge distribution found in polyelectrolyte complexes and inside core of the diblock polyampholyte micelles.163,166 The average polymer density inside the multilayers was shown to be a result of the fine interplay between electrostatic and short-range interactions, with systems in poor solvent conditions for the polymer backbone (Systems C&D) being found to feature higher average polymer density inside the multilayers.

115

Interaction parameters System A System B System C System D

lB-Bjerrum length 3.0σ 1.0σ 3.0σ 1.0σ

εLJ =0.3 kBT εLJ =0.3 kBT εLJ =1.0 kBT εLJ =1.0 kBT monomer-monomer rcut=2.5σ rcut=2.5σ rcut=2.5σ rcut=2.5σ

surface bead – monomer εLJ =1.0 kBT εLJ =1.0 kBT εLJ =1.0 kBT εLJ =1.0 kBT belonging to positively r =21/6σ r =21/6σ r =2.5σ r =2.5σ charged chains cut cut cut cut

surface bead – monomer εLJ =1.0 kBT εLJ =1.0 kBT εLJ =1.0 kBT εLJ =1.0 kBT belonging to negatively r =2.5σ r =2.5σ r =2.5σ r =2.5σ charged chain cut cut cut cut

monomer-counterion εLJ =1.0 kBT εLJ =1.0 kBT εLJ =1.0 kBT εLJ =1.0 kBT

1/6 1/6 1/6 1/6 counterion-counterion rcut=2 σ rcut=2 σ rcut=2 σ rcut=2 σ

20σ × 20σ × 20σ × Simulation box size 28σ × 29.4σ 20.784σ × 20.784σ × 20.784σ × (x × y × z) × 81σ 81σ 81σ 81σ

Table 3.1: Interaction parameters and system sizes

116

Np System A System B System C System D

32 80 40 40 40

16 160 80 80 80

8 320 160 160 160

Table 3.2: Number of chains added to simulation box during each deposition step

117

10 10 (a) 8 (b) 8

6 2

Γσ 4 Γσ 4

2 2 0 12345678910 0 12345678910 Nstep Nstep 10 10

6 6 2

Γσ Γσ s 4 4

2 2

0 0 12345678910 12345678910 Nstep Nstep Figure 3.1: Dependence of the surface coverage (σΓ2) on the number of deposition steps for (a) system A (b) system B (c) system C and, (d) system D as specified according to Table 3.1. The closed symbols are for fully charged chains (f=1) and

open symbols for partial charged chains (f=1/2). The degree of polymerization is Np

= 32 (circles), Np =16 (triangles) and Np= 8 (squares).

118

2.5 0.12 2.5 0.12 (b) (a) 0.10 0.10 2.0 2.0

0.08 0.08 3 3 3 1.5 3 1.5 σ σ σ σ 0.06 0.06 z) z) (z) (z) ( c c

ρ( ρ 1.0 ρ 1.0 ρ 0.04 0.04

0.5 0.5 0.02 0.02

0.0 0.00 0.0 0.00 02468101214 024681012 z/σ z/σ

2.5 0.12 2.5 0.12 (d) (c) 0.10 0.10 2.0 2.0

0.08 0.08 3 3 3 1.5 3 σ σ 1.5 σ σ z)

z) 0.06 0.06 (z) (z) c ρ( ρ( c ρ 1.0 1.0 ρ 0.04 0.04

0.5 0.5 0.02 0.02

0.0 0.00 0.0 0.00 024681012 024681012 z/σ z/σ

Figure 3.2: Density profiles of the fully charged (f=1) chains with Np=32 along z-axis

after completion of 8 deposition cycles with a duration of 106 MD steps each

deposition cycle for (a) system A (b) system B (c) system C and, (d) system D as

specified according in Table 3.1. The monomer density profiles of negatively

charged chains (continuous line) and positively charged chains (dotted line) are on

left axis and the negative counterions (triangle) and positive counterions (circle) is

on right axis.

119

40 40

(a) (b)

30 30 (r) 20 (r) 20 p p

10 10

0 0 123456 123456 r/ r/σ σ

Figure 3.3: Correlation function between positively and negatively charged

monomers inside multilayers formed by fully charged chains, f=1, with degrees of

polymerizations Np= 32 (circles), 16 (triangles) after completion of the eight deposition steps. (a) Systems A (open symbols) and C (filled symbols); (b) Systems B

(open symbols) and D (filled symbols).

120

3 3 (a) f=1 Np=32 (b) f=1/2 Np=32

2 2 3 3 σ σ z) 1 z) 1 Δρ( Δρ( 0 0

-1 -1 0246810 0246810 z/σ z/σ

3 3 (c) f=1 N =32 (d) f=1/2 Np=32 p 2 2 3 3 σ σ z) z) 1 1 Δρ( Δρ(

0 0

-1 -1 0246810 0246810 z/σ z/σ

3 3 (e) f=1 Np=32 (f) f=1/2 Np=32 2 2 3 1 3 1 σ σ z) z) Δρ( 0 Δρ( 0 -1 -1

-2 -2 0246810 0246810 z/σ z/σ

3 3 (g) f=1 Np=32 (h) f=1/2 Np=32 2 2 3 3

σ σ z) z) 1 1 Δρ( Δρ( 0 0

-1 -1 0246810 0246810 z/σ z/σ

Figure 3.4: Uniaxial monomer density difference of positively and negatively

charged chains, Δρ(z) = ρ − (z) − ρ + (z) for (a,b) systemA, (c,d) system B, (c) system

C, (e,f) and (g,h) system D (Table 3.1). The film composition is taken after completion of 8th deposition step.

121

1.00

0.75

ads 0.50 Q|/Q |Δ

0.25

012345678910 Nstep

Figure 3.5: Dependence of the overcharging fraction (|ΔQ|/Qads) on the deposition step number for different fraction of charged monomers f=1 (filled symbols) and f=1/2 (open symbols) and degree of polymerizations Np=32 (circles), 16 (triangles) and 8 (square). System A with Np=32 (circles), System A with Np=16 (inverted triangles), System B with Np =32 (triangles), System C with Np=32 (squares), System

C with Np =16 (rhombs), System C with Np =8 and f=1 (filled triangles) and f=1/2

(inverted open triangles).

122

3.2

3.0 8th Deposition Step 2.8

2.6

2.4 2

2.2 Γσ th 2.0 6 Deposition Step

1.8 th 1.6 4 Deposition Step

1.4 2nd Deposition Step 1.2 2.0 4.0 6.0 8.0 10.0 MD Steps / 106

123

124

CHAPTER 4

4 Combined Effect of Spin Speed and Ionic Strength on

Polyelectrolyte Spin Assembly*

4.1 Synopsis

Polyelectrolyte spin assembly (PSA) of multilayers is a sequential process featuring adsorption of oppositely charged polyelectrolytes from dilute solutions undergoing spin-coating flow. This chapter reports on the dependence of PSA multilayer growth, morphology, thickness and roughness of poly(sodium-4-styrene-sulfonate) (PSS) and poly(allylamine hydrochloride) (PAH) on solution ionic strength and spin speed. It was observed that at a given spin speed, the PSA coating growth rate (thickness/bilayer) and polymer surface coverage shows a non-monotonic dependence on salt concentration, first increasing and then decreasing with increasing the solution ionic strength. This is argued to be a manifestation of two competing mechanisms responsible for the layer formation. At low salt concentrations the electrostatic interactions control the multilayer assembly process while at high salt concentrations it is dominated by shear flow. This non-monotonic behavior is explained in the framework of a Flory-like theory of multilayer formation from polyelectrolyte solution under shear flow. This theory is described in details in introduction Chapter 1. In particular, the prediction of the Flory- like theory that the PSA growth rate of polymer surface coverage per bilayer,Γ, follows

Equation 1.13 in Chapter 1 (also given below), is tested.

c−1/2 Γ≈α (4.1) c−12/3+ βγ where α and β are fitting parameters, γ is the shear rate and c is the salt concentration of the solution, including both added salt and polyelectrolyte counterions. Additionally, the

PSA process led to multilayer coatings with a radial dependence on thickness at lower spin-speed in the shear dominated regime. On increasing spin speed, such radial dependence subsided eventually leading to uniform coatings by planarization. The surface topography of the multilayered coatings adsorbed at salt concentration less than

0.1 M was flat and featureless for all studied spin speeds. Unique morphological features in the films were formed at salt concentration higher than 0.1 M, the size of which depended on the spin speed and ionic strength. The rest of the chapter is organized as follows: Section 4.3 presents the experimental procedure, Section 4.4 describes the results and discussion followed by Section 4.5 with conclusions.

4.2 Introduction

See Section 1.2 in Chapter 1

4.3 Experimental Procedure

4.3.1 Preparation of Multilayers

Quartz (1” Diameter x 1/16” Thick) disks (Chemglass) were used as the substrates for preparation of multilayered coatings by polyelectrolyte spin assembly. Substrates were treated with freshly prepared piranha solution (70:30 v/v H2SO4:H2O2) at 60 °C for

1 h, followed by sonication for 5 minutes (Caution: Piranha solution is extremely

126

corrosive, special care must be taken during handling of such solution). The substrates were then rinsed multiple times in ultra pure deionized (DI) water (Milli-Q, ρ > 18 MΩ cm) and dried. To produce a charged surface primed for polyelectrolyte adsorption, a layer of low molecular weight poly(ethyleneimine) (PEI) (Aldrich,CAS No. 025987068)

(3 wt% solution in DI water) was deposited on the substrate by spin coating at 5000 rpm for a duration of 15 s (Laurell Technologies Corporation, Model WS-400B-6NPP/LITE).

Such a substrate was then rinsed with DI water three times under the same spinning conditions to remove the loosely adsorbed PEI and obtain a uniform layer of the positively charge on substrate. It was found that the multiple rinsing steps with DI water are necessary to remove the loosely adsorbed PEI from substrate since this irreproducibly affects the initial growth rate of the LbL assembly before the steady growth rate is achieved. For the growth of multilayers using polyelectrolyte spin-assembly, poly(sodium-4-styrene-sulfonate) (PSS) (Mw = 70 kDa) (Aldrich) was used as a polyanion and poly(allylamine hydrochloride) (PAH) (Mw = 15 kDa) (Aldrich) was used as polycation. The concentration of both polyelectrolyte solutions was 0.01 M based on the molar mass of each repeat unit of polymer (206 g/mol and 93.5 g/mol for PSS and

PAH, respectively). Noting that the pKa values for PSS and PAH are ca. 1.0 and 8.5, resp., the pH of each polyelectrolyte solution was adjusted to 3.5 using 0.1 M HCl. Thus both solutions will contain polyelectrolytes that are nearly fully charged.48 The ionic strength of the polyelectrolyte solution (0.01 M) was varied by adding NaCl to different concentration levels between 0 M (no salt) and 1.0 M. In the text we refer different solutions based on their NaCl concentration; however, it is worth noting that the total ionic strength of the polyelectrolyte solutions includes both the counterion concentration

127

of the polyelectrolytes (10-2 M) and the added salt.

Multilayer coatings by PSA on quartz substrate were made for each combination of the spin speed (3000, 5000 and 6000 rpm) and salt concentration (0, 0.05, 0.1, 0.25,

0.5, and 1.0 M) using polyelectrolyte (PSS or PAH) concentration at 10-2 M. Multilayers were constructed by following the procedure reported previously.48 Briefly, the deposition process for each cycle, or bilayer, was performed as follows: (1) several drops of PSS solution are placed on the quartz substrate to completely cover the surface followed by spinning at a prescribed spin speed for 15 s; (2) two washing steps with DI water are accomplished using the same sequence as during the step 1; (3) step 1 is repeated with PAH polyelectrolyte solution and, (4) washing steps with DI water similar to step 2 is repeated. These steps were repeated multiple times until the prescribed number of deposition cycles was reached. It is to be noted that in our experimental procedure, the polyelectrolyte solutions were first deposited on substrate before spinning, in contrast to some reports167 where the solution was deposited on to the already spinning substrate. This was done to ensure that the surface is entirely covered with excess polyelectrolyte solution before spinning the substrate. The delay between the solution deposition on the substrate and the spinning of substrate in this case was typically, 2-3 s.

Such short delay would not result in any adsorption similar to the quiescent dipping case, since for the latter case the deposition generally requires 10 – 20 min to reach equilibrium adsorption.

4.3.2 Characterization

Film growth by PSA on quartz substrates was monitored by measuring the

128

increasing absorbance intensity of the distinct π to π* transition of PSS at 226 nm with

UV-Vis spectrometry (Varian Cary 50 UV-Vis Spectrophotometer). A clean quartz disk with the same thickness was used as a reference for the spectra, while the UV-Vis absorbance spectra were taken every two or four cycles of consecutive deposition of PSS and PAH with a rinse step in between. In contrast to PSS, PAH shows no detectable features in the UV-Vis spectrum (190 nm – 700 nm). Spatially, the UV-Vis spectrometer allowed measurements of the amount of PSS (surface coverage) within the spot of the beam with diameter of 1 mm on the substrate. This feature was exploited to examine potential radial dependence of coating thickness. Thus, spectra were taken at precise radial positions of each disk after first aligning the beam to the disk center and then displacing the disk using a custom built linear translation stage. The growth rates of the multilayered coatings are reported from the PSS absorbance taken at 6 mm from the disk center, unless otherwise noted. The arrangement also allowed measurements of multilayer growth rate at different radial positions on each substrate. The radial dependence of the UV absorbance, each value being a measure of the amount of polymer adsorbed, was investigated after deposition of 20 bilayers of the PSS and PAH for each combination of the spin-rate and solution ionic strength at constant polyelectrolyte concentration.

4.3.3 Atomic Force Microscopy (AFM)

Surface morphology, roughness, and thickness measurements of multilayers were obtained from the height images collected using contact-mode atomic force microscopy

(AFM) (Veeco Digital Instruments Dimension 3100 scanning probe microscope).

129

Measurements were performed on the multilayered coatings on the quartz after deposition of 32 bilayers for each set of spin speed and salt concentrations. Contact mode-AFM can non-invasively probe multilayer surface morphology48,168 owing to the high modulus and toughness of such coatings in the dry state.169,170 The surface roughness was quantified from the AFM images using the imaging software (Veeco

Instruments NanoScope ® II) that calculates the surface roughness by the following equation:

N 1 2 Ra = ∑ (hi − h ) (4.2) N i=1

where Ra is the root mean square roughness, hi is the height value of each pixel, h is the average height of all the pixels and N is number of pixels in the scan area. It is known that roughness measurements depend on several factors such as scan size, image bow or sample tilt, and other correction procedures for tip convolution effects on the acquired image.171 In these studies, all the images were corrected for bow/tilt using a first order flatten function in the image software before the roughness measurements were analyzed.

AFM images were taken as 10 µm × 10 µm scans on the multilayer film on quartz substrate and roughness calculations were then performed over 5 µm × 5 µm areas of the scan. The roughness values were averaged over six randomly selected areas covering the whole image. The same AFM tip and image processing algorithm were used for all of the measurements to allow fair comparison of roughness values across all samples.

AFM-based thickness measurements were performed by first etching a portion of the film using a sharp edge of razor blade that penetrates to the quartz substrate (but no

130

further).48 The etching of the film was done very gently with the razor blade along the lines on the quartz disc (chord) at a precise radial distance from center. Contact-mode

AFM scans of 50 µm × 50 µm were then performed on a region including the bare quartz.

The width of the etched portion of the film was ca. 25 µm. Section analysis of the image was then performed that measured the difference in the z-positions with the line scan using the imaging software (Veeco Instruments NanoScope® II). The thickness was then determined by the average height difference between the areas with and without the multilayered coatings.

4.4 Results and Discussion

The growth rate of the multilayer film deposited by polyelectrolyte spin assembly

(PSA) technique was followed by growth of the PSS absorbance peak at 226 nm using

UV-Vis spectroscopy. Figure 4.1 shows the representative evolution of the absorbance spectra for PSS/PAH multilayers deposited at salt concentration of 0.05 M, and spin rate of 5000 rpm with increasing deposition cycle number. Each absorbance spectrum was measured at a distance of 6 mm from the center of the disk in order to elucidate the effect of shear rate. The absorbance value (the peak amplitude) at 226 nm (Figure 4.1, inset) shows a linear growth after completion of 8 deposition cycles. For the first few layers the absorbance values have different initial slope compared to the later stage of steady-state regime. This is consistent with prior studies, which have revealed that a steady state linear growth regime is usually observed in the experiments after completion of first few deposition cycles for the case of ‘dipping’ assembly40 and PSA.56 The growth rate difference at early stages – in this case a higher growth rate – is believed to result from

131

the difference in surface charge density of the substrate compared to the natural overcharging at steady-state regime.

Overcharging – that is, the overcompensation of surface charge by ca. 50% during adsorption of oppositely charged polyelectrolyte – is observed for steady-state growth as reported for PSA56 and from simulations described in Chapter 3. In experiments, this difference in growth rate has been observed for deposition from two to sixteen deposition steps depending on the substrate and the polyelectrolyte pairs.40,56

Nevertheless, after the linear (regular) growth regime is achieved, the slope does not change for subsequent deposition cycles, even for 50 deposition cycles at 5000 rpm from polyelectrolyte solutions with 0.1 M salt concentration (see Figure 4.2). In order to study the linear steady-state growth rate of the multilayer build-up for each experiment, growth process by completing up to 32 deposition cycles was followed by UV-Vis spectroscopy.

The steady-state growth rate was measured for each sample by linear regression of the absorbance values between the 10th and 32nd deposition steps. This allows comparison of the amount of the polyelectrolyte adsorbed (or surface coverage) per deposition cycles at steady-state for a range of the spin-rate and salt concentrations for the PSA process.

Previously, it has been shown that the linear growth of the adsorbed amount of PSS measured by UV-Vis spectrometry was consistent with the thickness growth measurements obtained by AFM48 and ellipsometry56 for the same polyelectrolyte pair deposited at 3000 rpm.

First, the impact of added salt on PSA growth rate for several spin speeds is considered. Figure 4.3 shows the growth of polymer surface coverage with number of

132

deposition cycles at different salt concentrations for PSS/PAH films spin assembled at (a)

3000 rpm, (b) 5000 rpm and (c) 6000 rpm. The curves follow a similar linear growth pattern as seen previously in Figure 4.1 (inset) where a linear increase in the polymer surface coverage with number of deposition cycles was observed after the completion of

8 deposition cycles. The difference in the initial slopes of all the curves are attributed to difference between the initial surface charge and overcharging achieved after each deposition at given conditions of spin-speed and salt concentration, as was pointed out earlier. The different initial absorbance growth-rate compared to the steady-state growth rate is further evidenced by the non-zero y-intercepts of the regression lines shown in

Figure 4.2. Unlike the initial growth rate which depends on the primer layer, the growth rates at the later stage signifies the effect of deposition conditions and are reproducible for each set of deposition conditions.

The growth rate of the multilayer coatings in the linear steady-state regime shows a non-monotonic dependence on the salt concentration at a given spin rate. Figure 4.4 shows such behavior in the absorbance growth rate for the range of the salt concentrations and spin rate. Here, the growth rate was taken as the slope at steady-state

(late stage) for data sets shown in Figure 4.3. The growth rate first increases rapidly with salt concentration for c < 0.1 M, reaches a peak near 0.25 M salt concentration and then decreases. The rapid increase in the growth rate at low salt concentration is quite similar for all spin rates studied. However, the salt concentration dependence of the growth rate at higher salt concentrations – the decreasing regime - shows strong dependence on the spin-speed. In this salt concentration regime, the growth rate decreases faster for the multilayers built at higher spin speeds. The observed trend in the dependence of growth

133

rate indeed follows the predicted form for polymer surface coverage given by Eqn. 4.1 (in which the salt concentration, c, should include both the concentration of added salt and concentration of polyelectrolyte counterions).

Indeed, the above observation confirms the existence of two competing mechanisms for the multilayer build-up process using PSA: a low salt regime of increasing growth rate and a high salt regime of decreasing growth rate. In the first regime at low salt concentrations, c < 0.1 M, the adsorption of polyelectrolytes is controlled by electrostatic interactions between charged chains. The multilayer assembly in this salt concentration regime is similar to the quiescent LbL assembly. It is known that for dipping (quiescent) LbL assembly, the multilayer growth rate steadily increases with salt concentration. Dubas and Schlenoff172 have shown that for the dipping LbL assembly the growth rate varies linearly with salt concentration, while Decher and

Lvov173 and others174,175 have observed that the growth rate increases as a square root of the salt concentration. The addition of the salt screens the electrostatic repulsion between similarly charged groups, increasing the size of the adsorbed polyelectrolytes. Such charge screening and swelling of polyelectrolytes in multilayers results into additional adsorption of polymer chains to maintain required overcharging of the surface and layer growth.39 Quantitatively, the square root dependence of growth on salt concentration is predicted for quiescent (dipping) assembly at low ionic strength175,176 and using the

Flory-type theory.48 At high ionic strength (c > 0.5 M) Castelnovo and Joanny177 have predicted a linear dependence on salt concentration for quiescent LbL assembly.

The second regime occurs at higher salt concentrations. In this regime the growth

134

rate of the polymer surface coverage decreases with increasing the salt concentration.

According to the Flory-type model for polyelectrolyte adsorption under shear flow,48 and following similar work on polymer brushes178, in this salt concentration range the shear flow created by the rotating disk is capable of deforming the adsorbing chains and work against the electrostatic attraction between oppositely charged groups favoring chain’s adsorption. It is known that polymer surface coverage in PSA exhibits significant radial dependence,63 due to the fact that the shear rate depends on radius in spin-coating flow.

The shear rate increases linearly the distance r from the center of the rotating disk according to approximation close to the surface,48,179

γ = ρω2rh / η (4.3) where η is the viscosity of the polyelectrolyte solution, ω is the angular velocity of the rotating disk, and h is the thickness of the ‘thinning’ polymer solution. Indeed, the growth rate measured further away from the disk center (at 8 mm) for spin speed of 3000 rpm shows even more pronounced effect of shear rate. Figure 4.5 shows absorbance growth rate at various salt concentration for the range of radial distance from the disc center for the spin speed of 3000 rpm. The absorbance growth rate is similar for all radial distances below the salt concentration of 0.1 M. At higher salt concentration above 0.1

M, the growth rate significantly depends on the radius, with increasing the radial distance resulting into decreased growth rate. This is consistent with the previous data of Figure

4.4, where at fixed radial distance, decreasing regime was observed at higher spin speed.

Either increasing spin-speed or the radial distance results into higher shear rate according to Eqn. 4.3. It is to be noted that in previous study at 3000 rpm,48 it was reported that the

135

growth rates of multilayers measured by UV-Vis spectroscopy and AFM, showed a plateau for salt concentrations higher than 0.1 M. Such observations in that study were made at the disc center, whereas those for Figure 4.3 were measured 6 mm from the disc center, a higher shear rate position (Eqn. 4.3) allowing resolution of a shear flow effect at higher salt concentrations.

The radial dependence of the polymer surface coverage in multilayer coatings by

PSA was followed by the PSS absorbance measured after completion of 20 deposition cycles. Such measurements are a good measure of PSA growth rate, though not direct due to the bi-linear nature of multilayer growth shown in Figures 4.1 (inset) and 4.2.

Figure 4.6 shows such radial dependence of polymer surface coverage for multilayers at different salt concentrations and spinning speed of (a) 3000 rpm, (b) 5000 rpm and (c)

6000 rpm. At lower salt concentration, c = 0 M and 0.1 M, and spin-speed of 3000 rpm

(Figure 4.6a) the radial profile is essentially flat suggesting a uniform polyelectrolyte surface coverage. This is another indication of the dominance of the electrostatic interactions in low salt concentration regime. However, the growth rate of polymer surface coverage shows radial dependence for the salt concentration of 0.5 M and 1.0 M

(Figure 4.6a), with a decrease in coverage with increasing radius. This functional form of the radial profile can be fitted by following equation obtained by substituting Eqn. (4.3) in Eqn. (1.1),

Γ Γ≈ 0 (4.4) 1()+ ζω22/3r

where, Γ0 and ζ are fitting parameters. However, the absorbance value at 2 mm from the

136

disk center for the case of 0.5 M and 1.0 M salt concentration in Figure 4a, though reproducible, does not conform to this functional form. This may be due to the combined effect of higher curvature at low radius and fairly large spot size (1 mm) used to measure absorbance in spectrophotometer. Nevertheless, the reasonable fit of the radial data at

3000 rpm gives an indication of shear rate dominance at high salt concentration. Such radial dependence was also observed in our previous study.48 Chiarelli et al63 have studied radial dependence of the multilayer coatings assembled at 3000 rpm from salt free solution for weak polyelectrolyte pairs of PEI and poly[1-[4-(3-carboxy-4- hydroxyphenyl-azo)benzenesulfonamide]-1,2-ethandilyl)(PAZO). They found that the film thickness, measured by ellipsometry, has a weak dependence on radius unlike the uniform coatings observed in our study from salt free polyelectrolyte solutions. This difference might be due to the difference in charge density of the polyelectrolyte chains compared to the fully-charged polyelectrolytes used in our study. As mentioned in

Chapter 1, for PSA, the surface localized shear flow may deform the adsorbed polyelectrolyte and compete with the electrostatic interactions. Thus, partially charged polyelectrolytes can have strong shear-rate dependence compared to fully charged polyelectrolytes. A detailed analysis of the effect of the polyelectrolyte charge density on

PSA multilayer uniformity (radial dependence) is beyond the scope of the present study, but does warrant future attention.

Increasing the spin speed leads to interesting observations on radial dependence.

Figure 4.6b shows effect of salt concentration on the radial dependence of the multilayered coatings at 5000 rpm. At low salt concentration (c = 0 M and 0.1 M), the polymer surface coverage with radius is uniform, further confirming the dominance of

137

the electrostatic interaction in the layer build-up. At higher salt concentration, 0.5 M and

1.0 M, the radial profile is different than that seen for similar salt concentration at 3000 rpm. Compared to Figure 4.6a, for radius, r < 4 mm, the polymer surface coverage is decreased, though not as much for radial distance r > 4 mm. Increasing the spin speed to

6000 rpm further lowers the polymer surface coverage at lower radius (r < 4 mm) producing essentially flat uniform layers at higher salt concentration (Figure 4.6c). There is negligible radial dependence of the polyelectrolyte surface coverage even for the highest studied salt concentration of 1.0 M. Thus, the decrease in polymer surface coverage with increasing spin speed is more pronounced at lower radius. It is believed that this is due to the fact that the shearing flow can stretch the polyelectrolyte chains only to a finite extent. In analogy to the case of polymer solutions180 or tethered polymer chains181 exposed to elongational flow, there should exist a critical deformation rate beyond which the chain has fully extended conformation on surface, thus showing no further effect of shear rate on the coatings. The shear rate in PSA depends on the spin- speed and radius according to Eqn. 4.2. For multilayers assembled at 5000 rpm, such critical shear rate is apparently achieved at radius, r ≈ 4 mm. On increasing the spin speed, the critical shear rate corresponding to full extension of the polyelectrolyte chain, is achieved at lower radial distance (see Eqn. 4.4) as seen for decrease in absorbance at lower radius for films assembled at 6000 rpm (Figure 4.6c). This eventually leads to a planarized coating as observed for higher shear rate and salt concentration. Conventional spin coating method of the uncharged polymers has shown to produce such highly planarized coatings.182-184 Thus, the shear stress affects the adsorbed polyelectrolyte chain’s conformation and thereby affects the surface coverage and growth rate.

138

In light of the effects described above, it is reasoned that the impact of shear stress on polyelectrolyte adsorption may be non-local and impact the surface morphology.

Thus, contact-mode AFM was used to investigate the effect of the spin speed and salt concentration on the surface morphology, roughness and thickness of the multilayered coatings. The data were collected after completion of 32 deposition cycles at the radius of 6 mm from the disc center. The roughness of the multilayer coatings increases monotonically with the salt concentration at a fixed spin speed unlike the non-monotonic behavior observed for the growth rate of these same multilayered coatings. Figure 4.7 shows the surface topography obtained by AFM at different salt concentrations (a) 0 M,

(b) 0.1 M and (c) 1.0 M and at the spin rate of 5000 rpm. The PSA multilayer coatings deposited from salt free solution are very smooth with the root mean square (RMS) roughness value equal to 1.31 nm. As the salt concentration is increased to 0.1 M, the surface roughness increases (Figure 4.7b), with a RMS roughness at this salt concentration tripling 3.94 nm. Further increasing salt concentration to 1.0 M (Figure

4.7c), the RMS roughness increases ~ 10 times, compared to the multilayers assembled from salt free solution, to 10.12 nm. A similar increase in the roughness of multilayered coatings with increasing ionic strength has been observed with AFM studies of dipping assembly185,186 and for the PSA at 3000 rpm.48 In a previous study of dipping LbL assembly,185 wormlike (or “vermiculate”) patterns were observed for those coatings deposited from solutions with 1.0 M NaCl concentration. Those features were ~ 50 nm in height, 200 nm in width and had RMS roughness of ~ 20 nm. In the present study, for

PSA at 5000 rpm and 1.0 M NaCl concentration (Figure 4.7c), the features are ~ 40 nm in height, ~450 nm in width and with much lower RMS roughness of ~ 10 nm. Thus, the

139

spinning of the substrate produces features that are broader and flattened as compared to the films obtained quiescently at the same solution ionic strength.

These characteristic features become less pronounced at higher spin speeds.

Figure 4.8 shows the surface topography obtained by AFM at different salt concentration of (a) 0.1 M, (b) 0.5 M and (c) 1.0M and at the spin speed of 6000 rpm. The multilayered coatings spin-coated at 6000 rpm has lower roughness compared to the coatings obtained at 5000 rpm and similar salt concentration. For example, the multilayered coatings spin- coated at 6000 rpm and salt concentration of 1.0 M have features ~15 nm in height and ~

350 nm in width with mean square roughness of around 7 nm (Figure 4.8c). It is known that increase in the salt concentration increases the characteristic features size.185 Shear tends to flatten out the features, increasingly at higher spin speeds. These characteristic features of the multilayered coatings at high salt concentrations might give an additional clue in understanding the construction of the multilayered coatings under the effect of shear flow. In order to quantify for the AFM measurements, the height-histogram analysis was employed. Figure 4.9 shows the film height distribution function obtained from AFM images taken for the multilayered coatings at different salt concentration at

5000 rpm. The coating roughness increases with increasing salt concentration as indicated by the increase of the width of the height distribution functions. The height distribution function for coatings deposited from a solutions with salt concentration 1.0

M was very broad, indicating a rough coating compared to the narrow distribution for films assembled from salt free solutions (0 M) at the same spin rate. The width of the peak at the half maximum height is shown by the solid lines in the Figure 4.9 and plotted in inset, while the dashed horizontal lines represent the RMS roughness. The inset

140

indicates that the coating roughness increases almost linearly with the salt concentration, which is in agreement with the previous studies for the dipping185 and PSA48 at 3000 rpm.

The dependence of RMS roughness (Eqn. 2) on ionic strength is plotted in Figure

4.10. This figure shows that the roughness of coatings made at a higher spin rate (6000 rpm) is significantly lower than that for coatings assembled at 5000 rpm for all salt concentrations. Furthermore, the RMS roughness increases sharply at the salt concentration above 0.5 M for both spin speeds of 5000 rpm and 6000 rpm, as seen in

Figure 4.10. On comparison, the roughness values of the multilayered coatings obtained by dipping185 and by PSA at 3000 rpm48 are higher than those values achieved for spin speeds of 5000 rpm and 6000 rpm in the present study. Given the findings on growth rate as influenced by the combined effects of added salt and shear flow, it is reasoned that the observed decrease in the coating roughness is a manifestation of shear flow’s impact on chain conformation and larger-scale adsorption structures.

Finally, AFM was used in a profilometric manner to quantify the thickness of the multilayers by PSA to confirm the existence of the two regimes as predicted by our model and as witnessed using UV-Vis absorption measurements (Figures 4.3 and 4.4).

As one example, Figure 4.11 shows AFM-measured thicknesses for multilayered coatings after completion of the 32 deposition cycles from PSS/PAH solutions at 6000 rpm and different salt concentrations. The measured film thickness features a non-monotonic dependence on salt concentration similar to one obtained for polymer surface coverage by

UV-Vis measurements (see Figure 4.4). The thickness increases with increasing salt concentration for the range of salt concentrations below 0.1 M, and then decreases at

141

higher salt concentrations 0.5 M and 1.0 M. The salt concentration dependence of the film thickness is also in a good agreement with the predictions of the scaling model of

Lefaux et al as given by Eqn. (4.1), thus confirming the existence of the two regimes of electrostatic dominance at lower salt concentration and shear dominance at high salt concentration.

4.5 Conclusions

The combined effect of spin speed and salt concentration on the growth, morphology, thickness and roughness of the multilayer coatings made by sequential spin coating of the polyelectrolyte solutions, termed polyelectrolyte spin assembly was studied.

The growth of the multilayered coatings shows a non-monotonic dependence on ionic strength, first increasing and then decreasing with increasing solution ionic strength. This behavior is a manifestation of two competing mechanisms for the multilayer assembly process, electrostatic interactions dominating film growth at low ionic strength and shear flow dominating at high ionic strength. This can be explained in the framework of the

Flory-type theory48 of the multilayer adsorption that takes into account the formation of the ionic pairs between oppositely charged chains and chain deformation in the external shear flow created by a rotating disk. The associated scaling equation fits our experimental data for the multilayer growth of surface coverage and thickness of the multilayer at different spin speeds reasonably well. Additionally, while the multilayered coatings do not showed a radial dependence for the multilayers made at low salt concentrations below 0.1 M, independent of spin speed, a strong radial dependence was observed at higher ionic strengths, with coverage decreasing radially. This radial

142

dependence of polymer surface coverage, in good agreement with our model for PSA, is concluded to be due to the linear increase of the local shear rate with distance (r) from the disk center. However, further increase of the spin rate to 6000 rpm leads to planarization of the film and almost uniform polymer surface coverage over the entire disk and it is argued that the influence of shear is bounded, perhaps by full chain extension.

Topographically, the multilayered coatings are smooth and featureless at low salt concentrations with dimensions of the characteristic features evolving as the ionic strength of the polyelectrolyte solution is increased. The dimensions of these features are less pronounced for the multilayers formed by the spin-assembly than quiescent adsorption at the same salt concentrations. The feature size of the multilayered coatings at high ionic strength, decreases with increasing the spin speed, further confirming the effect of the shear rate on the chain conformations. Thus, the variations in salt concentration and the spin-rate are two interacting parameters that allow control over the growth rate and film thickness during multilayer assembly by spin-coating method.

143

0.6 1.4 (32) 0.5

1.2 (28) 0.4 0.3 (24) 1.0 0.2 (20) 0.1 Absorbance (a.u.) Absorbance 0.8 (16) 0.0 0 5 10 15 20 25 30 35 Number of Bilayers 0.6 (12)

Absorbance (a.u.) 0.4 (6) 0.2

0.0 200 220 240 260 280 Wavelength (nm)

(λmax= 226 nm) intensity versus number of bilayers for the same assembly. The solid line is a linear regression of the data set after 10th bilayer (closed circles) and the dashed line is the regression of the data up to 8th bilayer (open circles).

144

0.1M, 5000 rpm, growth 1.0

0.8

0.6

0.4

Absorbance (a.u.) Absorbance

0.2

0.0 0 1020304050 Number of Bilayers

145

0.6 0.6

0.5 0.5 (a.u.)

(a.u.) 0.4 0.4 max max λ λ 0.3 0.3

0.2 0.2 Absorbance at Absorbance at 0.1 0.1

0.0 0.0 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Number of bilayers Number of bilayers

0.6

0.5 (a.u.) 0.4 max λ 0.3

0.2 Absorbance at 0.1

0.0 0 5 10 15 20 25 30 35 Number of bilayers

Figure 4.3: Dependence of the absorbance at λmax (226 nm) on the number of

deposition cycle from PSS/PAH solutions (c = 10-2 M, pH=3.5) at spin-speeds of (a)

3000 rpm (b) 5000 rpm and (c) 6000 rpm and salt concentration of 0 M (closed

circles), 0.1 M (open circle), 0.25 M (closed triangle), 0.5 M (open triangle) and, 1 M

(square). The UV-Vis absorbance is measure at 6 mm from the center of the quartz

substrate. Solid lines represent linear regressions of the data after 10th bilayer.

146

0.018

0.016

0.014

0.012

0.010

0.008 3000 rpm 5000 rpm 6000 rpm 0.006 Absorbance growth rate (a.u./bilayer)

0.00.20.40.60.81.01.2

Csalt (M)

Figure 4.4: Dependence of the absorbance growth rate on salt concentration for multilayered films prepared with polyelectrolyte spin assembly method at 3000 rpm

(squares), 5000 rpm (circles) and 6000 rpm (triangles). The growth rate is measured at 6 mm from the center of substrate. The solid lines are the best fit to Eqn. (4.1).

147

0.018

0.016

0.014

te (a.u./bilayer) 0.012

0.010

0.008

0.006 Absorbance growth ra

0.00.20.40.60.81.01.2

Csalt (M)

148

0.35 0.35 (a.u.) (a.u.) max 0.30 max 0.30 λ λ

0.25 0.25 Absorbance at Absorbance 0.20 Absorbance at 0.20 (a) (b) 0.15 0.15 02468 02468 Radial Distance (mm) Radial Distance (mm)

0.35 (a.u.)

max 0.30 λ

0.25

at Absorbance 0.20 (c) 0.15 02468 Radial Distance (mm)

Figure 4.6: Radial dependence of the absorbance of PSS/PAH film at different salt concentrations: 0 M (filled circles), 0.1 M (filled triangles), 0.5 M (open circles) and

1.0 M (open triangle) at (a) 3000 rpm and (b) 5000 rpm and (c) 6000 rpm. Each point is averaged for the four different azimuthal angles. The dashed lines in Figure

(a) are the best fit to the equation (4.4). All measurements are taken after completion of 20 deposition cycles.

149

(a) (b)

(c)

Figure 4.7: Dependence of the roughness of PSS/PAH films on salt concentration at

5000 rpm. AFM measurements are taken at 6 mm from the center of the disc after completion of 32 deposition cycles at different salt concentrations: (a) 0 M, (b) 0.1 M, and (c) 1 M.

150

(a) (b)

(c)

Figure 4.8: Dependence of the roughness of PSS/PAH films on salt concentration at

6000 rpm. AFM measurements are taken at 6 mm from the center of the disc after completion of 32 deposition cycles at different salt concentrations: (a) 0.1 M, (b) 0.5

M, and (c) 1 M.

151

18 16 14 (0M) 12 10 8

FWHM (nm) 6 4 2

Counts 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 (0.1 M) c (M) salt (0.5 M) (1.0 M)

0 102030405060 Peak-Valley height (nm)

152

RMS Roughness (nm) 2

0 0.00.20.40.60.81.01.2

csalt (M)

Figure 4.10: Dependence of film roughness on the salt concentration after

-2 completion of 32 deposition cycles of (PSS/PAH)32 at (c= 10 M, pH=3.5) for 5000 rpm (circles) and 6000 rpm (triangles).

153

140

120

100

Thickness (nm) 60

20 0.0 0.2 0.4 0.6 0.8 1.0 1.2

csalt (M)

Figure 4.11: Thickness of the PSS/PAH multilayer coatings at 6mm from the center of the substrate after deposition of 32 bilayers at 6000 rpm. The solid line is the fit to

the Equation(4.2).

154

CHAPTER 5

5 Hydroxyapatite and Silica Biomineralization on

Polyelectrolyte Multilayers

5.1 Synopsis

Mineralization in living organisms is regulated by specific proteins (e.g., bone sialoproteins (BSP) are responsible for hydroxyapatite (HA) formation in bone while silaffin forms silica in diatoms) and a review of this is detailed in Chapter 1. Glutamic acid-rich sequence in proteins have been shown to regulate the nucleation and growth of

HA by BSP while, similarly, lysine-rich sequences and polyamines have a role in silica formation by silaffins. Accordingly, the possibility of forming and directing the growth of HA using simple synthetic polypeptides like poly(glutamic acid) (PGA) and silica by poly-L-lysine (PLL), has been intensively studied recently. One of the convenient ways to study the formation of minerals is to induce mineralization on the solid surfaces where the functional macromolecules are localized. Polyelectrolyte multilayers, obtained by layer-by-layer (LbL) assembly of polyelectrolytes as discussed in previous chapters, offer a potential route as an alternative to study the biomineralization processes on surfaces.

Hence, this chapter investigates the formation of the biominerals like hydroxyapatite and silica from such simple polypeptides in solution and on the surfaces of multilayers.

Limited studies exist for the ability of PGA to control the growth of hydroxyapatite when present in the solution and surfaces. Hence, we first studied HA

155

growth in presence of polypeptide, PGA, solutions of supersaturated (metastable) calcium/phosphate by the Constant Composition Method (CCM). CCM enabled study of

HA crystal growth kinetics at constant composition and, subsequently morphology of the

HA crystallites formed when PGA is present in the solution. Our CCM study indicated that PGA lowers the HA seed particle growth when present in the solution. Furthermore, there is an indication that the PGA binds to specific faces of the growing crystal, thereby directing HA growth in solution. WAXD and TEM were used to deduce the difference in the HA particle grown with or without PGA in solution. Subsequently, it is shown that even though PGA lowers the rate of HA growth in solution, PGA can successfully nucleate hydroxyapatite when localized on or within polyelectrolyte multilayers by layer- by-layer assembly. With regard to silica formation, the influence of synthetic polypeptides, particularly PLL, was investigated. The formation and morphology of silica in the presence of PLL was studied at surfaces of the multilayers, since extensive studies exist already for the silica formation from PLL in solution. The present study indicates that multilayers containing PLL can successfully induce silica formation upon exposure to pre-hydrolyzed silica precursor. The nature of silica formation depends on whether or not PLL is present only on the outermost layers or, instead, localized on every alternating layer within a multilayered coating. The latter arrangement of PLL results in increased particle size and amount of silica formation. Silica formation by polyamine like poly(ethylene imine) (PEI) will be further studied in Chapter 6 and Chapter 7.

5.2 Introduction

See Section 1.3 in Chapter 1

156

5.3 Experimental Method

5.3.1 Hydroxyapatite Formation

5.3.1.1 Materials and Methods

Solid reagent-grade calcium chloride dihydrate, potassium dihydrogen phosphate, sodium chloride and potassium hydroxide were purchased from Sigma Aldrich and used as received. Hydroxyapatite (HA) (5-25 μm, FisherScientific) was used as a reference and as a seed for constant composition method (CCM). Poly(L-glutamic acid) sodium salt (PGA) (Mw=15000-50000 Da), poly(allylamine hydrochloride) (PAH) (Mw=15000

Da) and branched poly(ethyleneimine) (PEI)(Mw = 25000 Da) (all from Sigma-Aldrich) were used as received.

5.3.1.2 Hydroxyapatite Synthesis

Calcium Phosphates exist in many forms that include, with decreasing order of solubility: monocalcium phosphate (MCP), dicalcium phosphate dihydrate (DCPD), tricalcium phosphate (TCP), octacalcium phosphate (OCP) and hydroxyapatite.

Hydroxyapatite is the most stable form of calcium phosphate at normal temperature and pH of 4 and 12. Hydroxyapatite (HA) synthesis was executed according to the experimental method reviewed by Kousopoulos.187 In brief, HA formation was achieved in solution by slow addition of potassium dihydrogen phosphate solution (0.3M) into calcium dichloride solution (0.5M) with a continuous flow of gaseous nitrogen (500 ml,

3-neck flask). The nitrogen purge was employed to avoid the influence of CO2 that otherwise forms trace amount of carbonated apatite.187 The pH of this reaction was

157

constantly monitored and maintained close to 9.5 with metering of 0.5M potassium hydroxide solution into the flask. Typically, our reactions lasted for 1.5 h, after which the precipitates were filtered and washed with DI water, and then resuspended into DI water. The precipitates were then annealed at 60 °C for 12 hours to convert the precursor phases – such as TCP and OCP – to HA as reported in literature.187 Next, the HA powders were dried under vacuum overnight at 70 °C and subsequently characterized using Fourier transform infrared spectroscopy (FTIR) and wide-angle x-ray diffraction

(WAXD) analysis (see section 5.2.1.4 for details). The addition rate of phosphate solution was varied to include 0.5, 1, and 2 ml/min. FTIR analysis showed that the product obtained at slowest addition rate of 0.5 ml/min had the least amount of impurities of other calcium phosphate phases like dicalcium phosphate dehydrate (DCPD) and tricalcium phosphate (TCP). Furthermore, FTIR spectra for HA obtained by the slowest phosphate addition rate (0.5 ml/min) was comparable to commercially available HA (data not shown). The synthesized HA and the HA (5-25 μm, Fisher-Scientific) was used as seed in the subsequent study.

5.3.1.3 Constant Composition Method (CCM)

Constant Composition Method (CCM) allows one to maintain constant activity of all ionic species during the crystal growth of HA seed in the metastable solution of calcium, phosphate, and hydroxyl ions.99,100 During HA formation, protons are released in the solution offering a sensitive means (pH) to monitor the crystal formation according to the reaction,

5CaCl2 + 3KH2PO4 + KOH Æ ½ Ca10(PO4)6(OH)2 + 6HCl + 4KCl (5.1)

158

Here, Eqn. (5.1) represents a simplified reaction scheme for the formation of HA.

Obviously, the ions are dissociated in the solution, and the dissociation constant depends on solution conditions. The ionic equilibrium and the activity of the all ionic species are taken into account while calculating the solution composition that is to be added to maintain the constant ionic activity (relative to the initial conditions) by pH monitoring.

The composition of the solution(s) to be added is determined by ion speciation program developed by Nancollas group (see detail in next paragraph).98,188 According to this program, CCM studies at relatively high supersaturation require simultaneous addition of two solutions, one containing sodium with calcium chloride and other a base (KOH) with phosphate. The addition of these two solutions is triggered by a change in concentration of the active species, a release of proton in this case. The active species was measured through pH measurements using ion-specific electrode (Metro-ohm MP-12 pH meter) after proper calibration. Thus, this method enabled to evaluate the kinetics of the HA formation in solution on HA seed particles, both in presence and absence of the polypeptide, PGA and polyelectrolytes.

The crystal growth experiments were made according to the procedure reported in the literature.99,101,189 The experiments were carried out in a thermostated (37 °C) 3-neck flash of total volume 500 ml. At first, the working solution of 200 ml was prepared by adding of calcium chloride solution (2 mM) and equilibrating at 37 °C. Sodium chloride salt (dried at 60 °C in vacuum for 4 h) was added to the calcium solution, such that the total ionic strength after the addition of phosphate solution would be 0.15 M. Note that sodium chloride is an inert electrolyte in this reaction, added only to minimize the

159

fluctuations in ionic equilibrium during the reaction. After salt addition, well characterized HA seed (20 mg) (either synthesized HA or commercial) and varying amounts of PGA were added and allowed to reach equilibrium for 5 minutes (for PGA adsorption on HA). After 5 minutes, the potassium dihydrogen phosphate solution (200 ml) was added (1.2mM, Ca/P=1.6) and immediately the suspension pH was adjusted to

7.4 by dilute HCl or KOH. The relative supersaturation value respect to HA, σHA, was calculated using the equation,

1/ν 1/ν -1/ν σHA = (IP - Kso ) Kso (5.2)

188 where, IP is the ionic activity product calculated from the speciation program, and Kso

-59 9 9 189 is the solubility product of HA, taken as 2.35×10 mol /L in the calculations and ν is the number of ions in the product, ν=9 for hydroxyapatite. The relative supersaturation with respect to HA for the studied system is 511. At Ca/P ratio of 1.6, the solution with relative supersaturation value of ca. 4800 induces instantaneous precipitation of the calcium phosphates. Indeed, our system is far from the precipitation regime; however, the system still has high relative supersaturation compared other studies where σHA values of 100 or lower were used.92 The use of an automated system is necessitated for experiments at low supersaturation, since single experiments to reasonable yield require

48 h or more.

The crystal growth was then monitored by the change in pH. To each of the necks of the flask was attached a pH meter probe, and a syringe containing the calcium and phosphate solution. The required composition of the two reagents was calculated from the following equations, 101

160

Solution 1: (10 Cat + 2 Cat) CaCl2 and

(0.3 –(20 Cat +30 Pt + 10/5 Cat)) NaCl (5.3)

Solution 2: (10 Pt + 2 Pt) KH2PO4 and

(20 Pt + 10/5 Cat + 2 CK+) KOH (5.4)

- where, the molar ratios of Cat/Pt/OH = 5:3:1, the subscript t means total concentration of the species in the solution where the crystal growth is studied and CK+ is the total concentration of potassium ions in the solution. A drop in pH of 0.01 units from 7.4, resulting from the precipitation of HA, triggered the addition of equal volumes of the two solutions (Eq. 1 and 2), to achieve constant composition of all species in solution. The addition of these reagents, under the condition of constant pH, was monitored to obtain the kinetics of HA crystal growth. The reactions were continued until pH could not be controlled any longer, due to an excessively high reaction rate. The HA suspensions thus produced were immediately filtered (Medium pore size (4-5 μm) filter, Fisher Scientific) and exhaustively washed with DI water. In a typical reaction, 125 mg of HA was synthesized from the 20 mg seed before reaction termination.

5.3.1.4 Characterization

The hydroxyapatite used as a seed for CCM (commercial and synthesized) was analyzed by Fourier transform infrared (FTIR) (Thermo Nicolet) and wide-angle x-ray diffraction (WAXD). FTIR was performed by a potassium bromide (KBr) pellet method.

KBr pellets were made by mixing ~1 mg of sample with 90 mg of FTIR-grade KBr.

FTIR (Thermo Nicolet - Nexus 870) spectra of the prepared KBr pellets were recorded

161

with a range of wave numbers spanning 400 cm-1 to 4000 cm-1 with averaging over 64 scans. For the x-ray diffraction analysis, a Rigaku x-ray diffractometer (SA-HF3 x-ray generator) was used with a Cu Kα radiation source of 1.5418 Å wavelength. RINT software was used to control the driver and monochromator. Voltage, current, and power were set to 30 kV, 30 mA and 0.90 kW, respectively. Samples were scanned at a rate of

0.4°/min using a 0.1 step sampling width and a 1.2 μm sized window. Samples for transmission electron microscopy (TEM) were prepared by dispersing the HA in ethanol using sonication, followed by and drying a drop of the dispersion on carbon-coated copper TEM grids. Electron microscopy was carried out on a JEOL 1200 EX TEM operating at 120 kV.

Scanning Electron Microscopy (SEM) (Philips XL30 Environmental SEM) coupled with Energy Dispersive X-ray (EDX) analysis was used to identify the morphology and elemental composition of the formed solids. Prior to analysis, samples were sputter coated with palladium for 30 s using a current of 45 mA under argon at a pressure of approximately 200 mTorr, yielding a coating thickness ca. 50 Å. Elemental analysis was performed using Energy Dispersive X-ray (EDX) probe attached to the same

SEM described above.

5.3.1.5 Polyelectrolyte Multilayers Construction and Mineralization

The deposition of PGA was performed on piranha-cleaned quartz and silicon (100) substrates using layer-by-layer (LbL) approach with PEI as initial (primer) layer to obtain the charged substrate. Caution: Piranha is a highly corrosive solution (70:30 v/v

H2SO4/H2O2). To achieve the adsorption of the first polyelectrolyte layer and thereby

162

produce a positively charged surface, a few layers (2-3 to ensure proper coating) of low molecular weight poly(ethyleneimine) (PEI) were deposited on the substrate by spin- coating at 5000 rpm for a duration of 10 seconds. Polyelectrolyte Spin Assembly (PSA) was employed for LbL deposition of the polyelectrolytes (see Chapter 4 for detail method). In brief, the multilayers were constructed by repeated deposition cycles that consist of following steps: (i) Deposition of several drops of PGA solution from a syringe in order to wet the whole charged surface followed by substrate spinning at 3000 rpm for 10 seconds (ii) two washing steps with several drops of pure deionized water at same conditions; (iii) Repetition of step (i) for PAH (10-2 M) solution. Thus, each step or

‘bilayer’ formation consists of alternate deposition of the polyanion (PGA) and polycation (PAH) with a rinse step of DI in between the depositions. The quartz or silicon substrate were thus deposited with five bilayers (PEI-(PGA-PAH)5-PGA) terminating with an extra layer of polypeptide (PGA). These substrates were then studied for HA nucleation and growth by CCM or by immersion in simulated body fluid (SBF). The composition of SBF was obtained from the literature.190 SBF was prepared in a 50 mM tris(hydroxymethyl)aminomethane (Tris) buffer with pH adjusted to 7.4 by addition of

1N HCl. The composition of the SBF was as follows (137 mM NaCl, 2.5 mM CaCl2, 1

mM KH2PO4, 4.2 mM NaHCO3, 3 mM KCl, 1.5 mM MgCl2, 0.5 mM Na2SO4).

5.3.2 Experimental Methods: Silica Formation

5.3.2.1 Multilayer Construction

All chemicals were purchased from Sigma-Aldrich and used as received. The deposition of poly-L-lysine by spin-assembly is performed on both quartz (1″ dia) and

163

silicon (100) (1″×1″) substrates. To clean each substrate, the substrate was treated with piranha solution (70:30 v/v H2SO4/H2O2) at 50 °C for 1 h, followed by rinsing and sonication in ultra pure water for 15 min. Multilayer build-up was achieved by the procedure reported in the section 5.2.1.5. The polyelectrolyte solution concentration was

10-2 M, based on repeat unit molar mass. The pH of the PSS solution was adjusted to 3.5 by 0.1M HCl, while the pH of PLL was as such from the prepared solution i.e, 4.6. Three samples of multilayers were prepared for silicification (i) PEI-(PSS-PAH)19-PSS-PLL, (ii)

PEI-(PSS-PLL)20 and (iii) PEI-(PSS-PAH)20 (control). This allowed comparison of the silica formation where PLL is localized either in top layer or every other layer to that with the control sample containing no PLL.

5.3.2.2 Silicification

The silane precursor (1M tetrahydroxysilane) was synthesized by dissolving tetramethyl orthosilicate (TMOS) in 1mM HCl. This product was then added to a sodium phophate citrate buffer to produce a final concentration of 113mM.191 pH buffer used was at pH 7.2. Sodium phosphate (Na2HPO4) and citric acid (C6H8O7) were used to make sodium phophate citrate buffer solution. The samples were silicified by the following procedure: (a) the multilayer surface was wetted by multiple drops of hydrolyzed TMOS solution (freshly prepared), (b) the sample was then placed in the closed humid chamber for more than 12 h, and (c) the silicified samples were then extensively washed with DI water and dried at room temperature in vacuum for 24 hrs.

5.3.2.3 Characterization

164

The silicified samples of multilayers were analyzed using Scanning Electron

Microscopy (SEM) (Jeol USA Inc, MA), Atomic Force Microscopy (AFM) (Asylum

Research MFP-3d, Santa Barbara, CA) and FTIR-Attenuated Total Reflectance (ATR)

(Bruker Optics, MA). All instruments are located at Institute of Materials Science (IMS),

University of Connecticut. AFM measurements were done in the tapping mode with frequency of 0.5-1 Hz. For FTIR-ATR measurements, the multilayers were constructed on the surface of the Kapton film supported by quartz substrate by the procedure described in previous section, prior to the silicification. This was done to ensure that the measurements reflect the true mineral (silica) obtained on the surface of multilayer and not the underlying substrate (quartz or silicon wafer).

5.4 Results and Discussion: Hydroxyapatite formation

5.4.1 Kinetics of HA growth in Solution

The kinetics of HA seed particle growth was studied by CCM in the solution with relative supersaturation, σHA = 511, Ca/P ratio of 1.67 at pH 7.4. Figure 5.1 shows the addition of titrate solutions (Eqn. 5.3 and 5.4) to maintain constant composition in the solution during the HA seed growth with or without addition of PGA in solution.

Syntheses with HA seed, but without the presence of PGA polypeptide, were found to feature very high growth rate (high slope in Fig. 5.1) as evident from the sharp upturn of the curve. The corresponding HA growth rates calculated from the initial slope of the

CCM curves in Fig. 5.1 are given in Table 5.1. Addition of PGA to the solution during

HA growth significantly lowers the growth rate (Fig. 5.1 and Table 5.1). On other hand

165

PSS, an alternate polyanion, have very little effect on the HA growth rate when present in the solution. Thus, the lower growth rate of HA is specific only to PGA, but not for PSS.

More interestingly, the growth rate returns to the original ‘uncontrolled’ growth resembling that of the HA seed particle after certain time (Figure 5.1). The time for this return to the original growth of the HA seed depends on the amount of the PGA present in the solution, since increasing the amount from 5 μM to 25 μM increases the time required for the upturn to take into effect (Fig. 5.1). To explain this observation, it is hypothesize that the PGA is present both adsorbed on the HA seed as well as excess in the solution. When all the PGA in the solution is consumed by binding the newly formed hydroxyapatite on to the seed particle, the growth rate returns to the original growth rate

(upturn) of the seed particle.

The observed lowering of HA production rate is thought to be due to the selective adsorption of PGA onto HA particle. In particular, the adsorption of PGA onto select faces of HA seed particles, has been extensively studied.192-195 Fernandez et al195 suggested that the PGA adsorbs on to the surface with a flat ‘extended’ conformation and that the carboxylic acid of PGA adopts a face down conformation with the calcium ions of HA. Kresak et al193 estimated that about 35 % surface area coverage of HA while

Inoue and Ontaki194 have estimated the range of 20 -80 % surface area coverage by PGA.

Furthermore, the surface coverage of the adsorbed PGA also depends on its conformation during adsorption. It is known that the conformation of the PGA in solution and on the surface of hydroxyapatite depends on the pH.196,197 PGA has α-helix conformation under more acidic conditions (pH < 5.5) and random coil conformation around neutral pH and

166

possibly a β-sheet structure at higher pH.196,198 Fujisawa and Kuboki199 using NMR techniques, have suggested that adsorption of polypeptide produce near β-sheet structure on to the HA surface at near neutral or higher pHs.

Indeed, when the PGA adsorption is made at either pH of 5.5 or 8.8 instead of pH

7.4, the constant composition growth rate is lowered (Table 5.1 and Fig. 5.1). This might be due to the higher surface coverage of the adsorbed polypeptide on to the HA surfaces.

As mentioned earlier, the surface area coverage varies in the range of 20- 80 % for the

PGA adsorbed on to the surface. Thus, PGA, depending of the pH conditions during the adsorption, effects the growth of the HA seed particles. Thus, the lower rate of HA growth is only on the surfaces where PGA is not adsorbed. It might be possible that the polypeptide adsorbs onto the specific crystal face(s), thus allowing the crystal growth only at remaining crystal surfaces which results in a lower overall growth rate. As the new surfaces form due to the HA growth, PGA ‘reserve’ in the solution gets depleted and at a certain point, the rate increases to that corresponding that without the PGA. It is known that a similar phenomenon takes place during enamel formation in human teeth by a protein named amelogenin, which binds to a specific crystal face, resulting in elongated

HA crystallites in enamel.200,201 Such crystallites, while taking longer time to grow, have high strength and hardness, important to enamel’s function.202

Wide-angle X-ray diffraction (WAXD) and TEM analysis of the HA powder was performed to deduce the effect of the PGA on the HA crystal structure. The WAXD patterns are shown in Figure 5.2 for two PGA concentrations and compared with the

WAXD pattern for HA seed. Clearly, the synthesized HA adopts the same crystalline

167

structure as the seed, known to be dipyramidal hexagonal203 with unit cell dimensions: a

= 9.418 Å and c = 6.875 Å. It is recognized that the diffraction patterns might be dominated by the large fraction of HA obtained during the uncontrolled growth phase i.e., upturn in CCM curve in Figure 5.1. Even though this is the case, the crystal structure of the HA is preserved, further confirming that the calcium phosphate nucleated on the HA seed particle is primarily hydroxyapatite and not other crystal forms of calcium phosphates, like DCPD or OCP. However, WAXD analysis alone cannot reveal important information about particle shape after the HA growth in presence of PGA. As mentioned earlier, it was hypothesized that the PGA binds specific crystal phases and the growth occurs on the other phases where PGA is not adsorbed.

To deduce the effect of PGA presence while forming HA on powder morphology,

TEM analysis was performed on those HA particles formed at various growth stages.

Figure 3a shows the TEM images of the HA seed particles used in the CCM study, while

Figure 3b shows the TEM image of the HA particles after 60 minutes of the growth of

HA seed in presence of 25 μM PGA in solution (indicated by arrow in Fig. 5.1). Hence, at this time the growth of the HA seed particles is limited by the PGA adsorption and the

‘uncontrolled’ growth has not yet taken place. Figure 5.3c shows the HA particles from

CCM study of HA seed in presence of 25 μM PGA after the sharp upturn (indicated by * in Fig. 5.1) which signifies that the growth rate was similar to bulk HA seed without PGA adsorption. Comparison of these cases allows direct determination of any morphological changes during the various stages of the HA seed growth by CCM.

The TEM micrograph of the seed HA crystal in Figure 5.3a shows that the

168

crystalline particles are aggregated. A close inspection reveals the presence of elongated crystals ~ 10 nm in diameter and 40 nm in length. The morphology of the HA is in agreement with various prior reports, though the HA particles were formed by different synthetic routes.98,188,203 The HA particles obtained from the CCM process during the growth at the PGA adsorption stage are well-dispersed and more elongated compared to seed particle as shown in Figure 5.3(a). Measurement of the few visible crystallites

(using ImageToolTM analysis) indicates that such elongated particle is ca. 15 nm diameter and ca. 65 nm length. This is in agreement with the hypothesis that PGA binds to specific crystal phase of the HA, while the growth on the other crystal face continues.

This is further supported by the fact that HA crystal has face dependent charge. PGA being a negatively charge at pH 7.4, adsorbs only in the calcium rich phase of the crystal surface and thus allowing the growth on the other crystal phases. In contrast, the morphology of the HA particle after the upturn is similar to the HA seed particles, showing very few elongated crystallites as shown by Figure 5.3(c). Thus, the PGA adsorption modifies the growth and the morphology of the HA seed particle. This is similar to the growth of enamel crystals by amelogenins as mentioned earlier.200,201

However, unlike the ca. 65 nm length of HA particles obtained in this study, natural enamel HA crystals grow to about 1000 nm in length. This explains the long time required to develop the enamel in mammalian teeth.

5.4.2 Growth of HA on LBL Surfaces

In order to study ability of PGA to nucleate HA when localized on the surface within the multilayers, PGA was deposited onto a silicon wafer using the PSA layer-by-

169

layer approach (Chapter 4) and then studied by CCM. PGA was incorporated in layer- by-layer fashion on a substrate with following configuration PEI-(PGA-PAH)5-PGA.

Figure 4 shows SEM image of the substrate after immersion in supersaturated (metastable)

Ca/P solutions while maintaining constant ionic activity by CCM for 4 hours. Here, no

HA seed was used as compared to previous CCM studies. This shows that PGA has ability to nucleate HA crystals when present on surface as opposing to growth inhibition when in solution. By comparison, a control experiment with no PGA surface coating showed negligible HA formation. These results are encouraging to use this biocatalyst in

HA nucleation and growth on the multilayer surfaces. However, a limitation to this approach is that the rate of nucleation and growth is considerably lower. The rate of addition of Ca/P solution by CCM method was very slow (~1 ml/h). The slow rate is due to very few nucleation events on the surface as seen in Figure 5.4. Also, the HA particles are easily washed away when rinsed with DI water (SEM image of washed sample not shown). Hence, the adhesion to the apatite formed to the surface needs to be understood and improved for it to be relevant to nucleate hydroxyapatite for any applications. Close inspection of the SEM image in Figure 5.4 reveal two types of crystals, one with smooth edges and other is very rough. EDX analysis of the rough particles shows that it has Ca/P ratio of ca 1.43 and the smooth particles has the Ca/P ratio of about 1.6. The Ca/P ratio, different for two type of particles, then corresponds to the two phases of calcium phosphate, namely OCP and HA. The two kind of morphology is comparable to the morphology of the particle obtained from the previous study of PGA on germanium surfaces,92 with one of them OCP and other being HA particles.

Thus, even though PGA inhibits the HA seed growth when present in solution, the

170

localization of it on to the surfaces induces the HA nucleation. Hunter et al95,204 has shown that both enantiomer of PGA, namely poly(L-glutamic acid) and poly(D-glutamic acid) nucleate HA growth in the agarose gel system. Nancollas et al92 have also shown such ability of the PGA to crystallize HA from the metastable Ca/P solutions on the layer of PGA adsorbed to the germanium crystal. The nucleation of HA by PGA is thought to be due to a specific spatial arrangement of the carboxylic groups in PGA formed on the surface due to PGA adsorption. The carboxylic acid groups bind the calcium ions and subsequently nucleate growth of HA crystals. This is the reason why bone sialoproteins and osteopontin responsible for nucleating HA have presence of glutamic acid-rich sequence.95,205 Both of the proteins contain regions enriched in acidic amino acids, particularly Glutamic acid and Aspartic acid. Particularly, BSP has a continuous sequence of ten Glutamic amino acids in two regions, while osteopontin has a continuous sequence of 9 glutamic acid residues. Thus, in general, amino acids containing carboxylic acid group are responsible for the HA mineralization.

Although the nucleation of HA on PGA-adsorbed surfaces has been studied previously, none of the studies relate to the ability of the HA to be nucleated when PGA is localized into the polyelectrolyte multilayers with other polycations. Ngankam et al206 have shown that PSS/PAH nucleates calcium phosphates above critical concentration of ca. 5.5 mM (Ca/P=1.0), in polyelectrolyte multilayers of PSS/PAH, irrespective of whether the outermost layer is PSS or PAH. It is to be noted that concentration of 5.5 mM at Ca/P ratio is well above the metastable region105 into the precipitation regime of calcium phosphates. A CCM study conducted by us to study HA nucleation on the

PSS/PAH multilayers from the metastable solution of calcium phosphates ([Ca]=2 mM,

171

Ca/P=1.67) showed no HA growth even after exposure of multilayers to extended period of time of 2 h. Thus, the previous report206 of HA nucleation and growth on PSS/PAH multilayers seems to be result of a higher supersaturation of the calcium phosphate solution used in that nucleation study. Furthermore, various studies have shown that HA formation occurs at relatively high supersaturation upon prolonged exposure of the surfaces with a variety of coatings including polymer scaffolds,207 chitosan,208 block- copolymers of polyethylene oxide (PEO),209 poly(l-lactic acid)210 and other surfaces.211

Most of these studies report apatite formation on the substrate during the prolong exposure of the substrate to simulated body fluid (SBF). SBF is chosen to resemble the physiological body fluid, which is also supersaturated (metastable) with respect to calcium and phosphate.

Thus, in order to compare the morphology of the HA nucleated by prolonged exposure of multilayers to SBF, PEI-(PGA/PAH)5-PGA multilayer substrates were exposed to SBF for 6-12 days. Indeed such long exposure resulted in a HA growth but the morphology as shown in Figure 5.5 was different from that observed with CCM study of Ca/P supersaturated solutions (Figure 5.4). The bare surface in the image at the center in Figure 5.5 is silicon substrate where the HA coating was removed. The figure inset shows a higher magnification SEM image of the substrate in the same region. A close inspection reveals the morphology of HA particles to be spherical – distinct from the flower-like or flat HA particles formed during the CCM study (Figure 5.4). These particles aggregate to form a thin layer of HA onto the surface. Furthermore, EDX analysis of the particles formed indicated that Ca/P ratio was ca. 1.4, lower than the expected 1.66 for the HA. Thus, the HA-like particles nucleated on the multilayers of

172

PEI-(PGA/PAH)5-PGA substrate has spherical morphology as compared to the flat or flower-like morphology of HA obtained when lower supersaturation in CCM is used.

5.5 Results and Discussion: Silica Formation

Unlike surface hydroxyapatite nucleation by PGA, a polyanion, silica polymerization occurs in the presence of polycations like polylysine127,130,132,212 or synthetic polyamines.133-135 Such polycations have been inspired by the catalytic domains found to play a role in diatoms construction.82 Indeed, silica formation mediated by PLL in solution has been studied extensively,128,130,212 though limited reports exist213 for the same on the formation of the surfaces. The present study is aimed at elaborating silica formation from the PLL when localized into the polyelectrolyte multilayer coatings on a substrate.

PLL, when localized into the multilayers, induces silica formation when exposed to the hydrolyzed silica precursor (TEOS) solutions (113 mM) according to the protocol mentioned in the experimental section. Figure 5.6 shows SEM images of silica formed on multilayers of (a) PEI-(PSS/PAH)19-PSS/PLL (b) PEI-(PSS/PLL)20 and (c) control substrate of PEI-(PSS/PAH)20 after exposing the multilayers to hydrolyzed TEOS (113 mM) for 15 minutes. The samples were analyzed by SEM after extensive washing with deionized water and drying in vacuum. PLL surfaces produced distinct silica formation as evident from Figure 5.6(a,b), in contrast to the control that contain no PLL. The sample without PLL was devoid of the characteristic features (Fig. 5.6(c)). In the case where PLL is present in the outmost layer of the multilayers, silica formation occured in the sparse regions as shown by bright spots in Figure 5.6(a). The sparse regions were

173

present at regular intervals. The silica particles formed on such multilayers has a snow- flake or dendrite-like morphology. Such morphology is unique since the silica formed from PLL in solution either has a plate-like or spherical morphology.131 Increasing the

PLL content in the multilayers (every alternating layer), the silica formation on the surface increased (Fig. 5.6(b)). The particles had snow-flake or dendrite-like morphology similar to Figure 5.6(a) but with greatly reduced inter-particle distance compared to the one observed with only outermost PLL layer in the multilayer. Also, the particles growth into larger size (~ 20 μm) compared to ~4 μm particle size for silica formed from multilayers containing only PLL as outermost layer.

Interestingly, the silica particles seen in Figure 5.6(b) did not cover the whole surface even when PLL was present in every alternate layer. This correlates with the observed ‘holes’ during the multilayer formation (or spots in this case) in the simulations study mentioned in Chapters 2 and 3. Also, the availability of PLL on the multilayer surface is limited since protonated amines of PLL, thought to be responsible for silica formation, are occupied in the formation of ion pair with sulfonate ions in polyanion, PSS.

As mentioned in the simulation study of Chapter 2 and 3, ion pair formation is important and governs the multilayer formation. Still, the formation of silica, even though in sparse regions, signifies that the not all the ionic sites of PLL are occupied in the multilayer formation. Furthermore, the observed increase in silica formation with increasing PLL incorporation in every alternating layer signifies that the surface has higher PLL concentration. We postulate that this is due to the interdiffusion of the polyelectrolyte within multilayers, as discussed in detail in Chapter 2 and 3. Thus, when PLL is localized only in the outermost layer, and the innermost layers consist of (PSS/PAH)19,

174

the interdiffusion of PAH to surface reduces the nucleation sites for silica formation. The reduction in the silica formation is postulated due to the non-catalytic ability of PAH to form silica when localized in multilayers. It has been observed (Fig. 5.6(c)) that PAH does not nucleate silica formation when localized in multilayers unlike PLL. On other hand, when PLL is present every alternating layer in (PSS/PLL)20, their interdiffusion essentially does not change the surface composition which is high in PLL and thus increased silica formation is observed(Fig. 5.6(b)). Thus, PLL localized on to the multilayer does not significantly lose the ability to induce silica formation from the pre- hydrolyzed precursor. Interestingly, the multilayers of PEI-(PSS/PAH)20 that does not contain PLL, does not induce the silica formation even though PAH is, itself, a polycation. A recent study has shown that single PAH layer on the surface induces the formation of continuous smooth silica films when such a PAH-containing surface is exposed to silica precursor solution.214 Apparently, such an ability of PAH to form silica is lost when it is localized in the multilayers with PSS. This is in general agreement with various studies, where PLL has higher ability to form silica from silica precursor compared to PAH.130

The silica formed on the PEI-(PSS/PLL)20 multilayers was further studied by tapping-mode AFM as shown in Figure 5.7. AFM revealed two types of morphology, in contrast the one type of snow-flake or dendrite-like particles observed in the SEM images.

The first kind, similar to the one observed in the SEM, has dendritic or snow-flake like appearance. The silica appears to have grown out of a central particle in the dendritic like crystalline structure. The other kind of morphology is particulate type as seen in the phase contrast image of Figure 5.7(a), where different particles of varying diameter are

175

present. Such particles have different mechanical compliance than the underlying substrate and are well dispersed on the surfaces. Such regular arrangement of the particles on the surface represents a brick-mortar type arrangement of particles in the matrix of the multilayer surface. We postulate that the observed dendritic structure in

Figure 5.7(c) grow out of such initially formed particles. The dendritic particles located at regular intervals has average diameter of ~ 20 μm compared to the spherical particles which has an average diameter of in the range of 50-100 nm, similar to the core of the dendritic particles. Furthermore, the surface of the multilayers is very rough as seen in the amplitude trace as seen in Figure 5.7(b). Such roughness arises due to the silica formation on the sparse regions on to the multilayer surface. Contact mechanical stiffness measurements by our colleagues have revealed that the average stiffness of such silica particles is 40 N/m, as compared to ~48 N/m for the non-silicified polyelectrolyte film.215 Such lower values suggest that multilayer organic film is amorphous and highly hydrated.

The formation of silica on multilayers of PEI-(PSS/PLL)20 was further confirmed by FTIR-ATR of silicified films on a Kapton™ substrate as shown in Figure 5.8. Three important absorbance peaks that indicate the presence of silica include the following: --

Si-OH stretching around 950 cm-1(here, at 975 cm-1), --Si-O-Si-- symmetric stretching around 790 -870 cm-1 and --Si-O-- asymmetric stretching at 1060-1090 cm-1 (Figure 5.8).

Silica formation on the multilayers consisting of PLL is clearly evident from the presence of all the absorbance due to silica. Furthermore, the comparison of the spectra in Figure

5.8 with standards reveal (Bruker Optics) reveal the close match of the assigned peaks to hydrated silica. This can be seen in the FTIR-ATR spectra with the presence of broad

176

peak at 3230 cm-1 assigned to –OH groups. This peak represents either the uncondensed silanol group or the –OH group in the water or alcohol liberated due to the condensation reactions of silicic acid. The observed peak of O-H occurs in spite of the drying of silicified fibers for 24 h at 40 °C in vacuum. Thus, the FTIR analysis indicates that the silica is formed on to the multilayers where PLL is localized in the alternating fashion by

LbL assembly. Furthermore, the x-ray analysis (data not shown) of silica formed on the same Kapton™ film revealed that the silica formed on the multilayer surface was amorphous.

5.6 Conclusions

To summarize, it has been demonstrated that the simple polypeptide, PGA, greatly affects HA formation rate and morphology. Constant composition method (CCM) analysis indicated that PGA lowers HA seed growth when present in the solution, while it is able to nucleate the HA particles when localized on the surface. In solution, the lower growth rate results due to the PGA adsorption on to the HA crystal on the specific crystal faces. Furthermore, CCM study indicates that the rate of nucleation and growth of HA was quiet lower in the multilayers compared to the seed HA growth. Similar to the HA formation by PGA localized in the multilayers, silica is formed when PLL is localized in the multilayers. Furthermore, the silica formation depends on the whether PLL is present only on the outermost layers or localized on every alternating layer in the multilayer. The latter arrangement of PLL results into increased particle size and amount of silica formation. Thus, it has been demonstrated that the various polypeptide catalyst when localized on to the multilayers, can successfully mediate formation of the mineral phases.

177

HA growth HA growth

Addition a b Sr. Concentration pH of Rate Rate Polyelectrolyte rate No (μM) Adsorption (10-8 mol. (10-6 mol. (ml/min) g-1.min-1) m-2.min-1)

1 Control 0 -- 0.4334 5.20 1.77

2 PSS 25 7.4 0.3628 4.35 1.48

3 PGA 5 7.4 0.1306 1.57 0.53

4 PGA 25 7.4 0.133 1.59 0.54

5 PGA 25 5.5 0.0405 48.6 0.17

6 PGA 25 8.8 0.0327 39.24 0.13

Table 5.1: Kinetic results of the effect of addition of PGA or PSS to the seeded

Hydroxyapatite crystal growth using the constant composition method at total calcium/phosphate =1.67, pH 7.4 and Ionic strength 0.15 M NaCl at 37 °C and pH

7.4.

a HA growth rate calculated from the addition rate of calcium solution (Eq. 3) as shown in Column 5 b HA growth rate per m2 calculated based on the assumption of surface area of ~ 34 m2/g HA189

178

25 *

10 Volume Added (ml) Volume Added 5

0 0 20406080100120140 time (min)

Figure 5.1: Rate of the addition of Ca/P solutions to maintain constant composition

during the growth of HA seed particles without addition of any polymer (closed

circle) or with addition of PSS (25 μM) (open circles), PGA (5 μM) (closed triangle),

PGA (25 μM) (open triangle) and PGA (25 μM) adsorbed at pH of 5.5 (closed

square). The addition rate corresponds to the rate of the consumption of Ca/P during the HA growth by constant composition method (CCM).

179

(a)

(b) Intensity

(c)

10 20 30 40 50 60

2θ

Figure 5.2: Powder WAXD analysis for hydroxyapatite (HA) formed during the constant composition method utilizing two concentrations of PGA (a) 5 μM, (b) 25

μM and (c) HA seed.

180

(a) (b)

100 nm 100 nm

(c)

100 nm

Figure 5.3: TEM images of (a) HA seed particles, (b) HA formed during CCM growth study after 60 minutes of 25 μM PGA addition and (c) HA formed during

CCM study after 120 minutes of 25μM PGA addition (upturn in Figure 5.1). The horizontal scale bars in the images are 100 nm.

181

Figure 5.4: SEM image of hydroxyapatite nucleated on the surface of Si wafer

coated with layer-by-layer of PEI-(PGA-PSS)5-PGA exposed to supersaturated

(metastable) Ca/P solution. Constant composition of the solution was maintained during the crystal growth by CCM.

182

(a)

Figure 5.5: SEM image of hydroxyapatite formed in the Simulated Body Fluid (SBF) on the Si wafer substrate coated with layer-by-layer with PEI-(PGA-PSS)5-PGA.

The substrate was soaked in SBF for 6 days in order to deposit thick films of apatite.

(Inset) High magnification SEM image of the boundary region of HA and the substrate.

183

(a) (b)

(c)

Figure 5.6: SEM images of silica formed on multilayers of (a) PEI-(PSS-PAH)19-

PSS-PLL (b) PEI-(PSS-PLL)20 and (c) control substrate of PEI-(PSS-PAH)20 after exposing the multilayers to hydrolyzed TMOS (113 mM) for 15 minutes.

184

(a) (b)

(c)

Figure 5.7: Tapping-mode AFM images of the silica formed on the multilayers of

PEI-(PSS-PLL)20 (a) phase and (b) amplitude trace of the surface showing presence of different sizes of the particles with different mechanical compliance and (c) amplitude trace of silica particle similar to the one observed in Figure 5.6.

185

0.35 1069 0.30

0.25 975

0.20 860 1596

Intensity 0.15 3230

0.10

0.05

0.00 4000 3000 2000 1000 Wavelength (cm-1)

Figure 5.8: FTIR-ATR spectra of the silica formed on the multilayers of PEI-(PSS-

PLL)20 after exposure of the multilayers to hydrolyzed TMOS (100 mM). The underlying substrate to the multilayers was KaptonTM film.

186

CHAPTER 6

6 Silica formation by Poly(ethylene imine)(PEI) in

Solution and on Surfaces

6.1 Synopsis

Polyamines, like poly(ethyleneimine) (PEI), are known to affect silica formation from hydrolyzed alkoxysilanes precursor, in two ways. They act to flocculate the negatively charge sol-particles due to its polycationic nature, and catalyzes the formation of siloxane bond to form silica in various biosilicification processes. The latter phenomenon has been discovered recently due to the ability of proteins found in diatoms to precipitate and direct the unique silica morphology. The catalysis occurs owing to the presence of lysine-residues modified with polyamines in proteins found in diatoms

(discussed in Chapter 1). In this chapter, the role of PEI to catalyze silica formation directly from non-hydrolyzed tetramethylorthosilicate (TMOS) was examined. It is observed that aqueous PEI solution rapidly, within seconds, catalyzes silica formation when added to silica precursor –TMOS. Such unique ability of PEI to form silica directly from TMOS was then investigated in solutions and on surfaces. In solution, the effect of aqueous PEI and TMOS concentration (in ethanol) on the inorganic content, yield, and morphology of the resulting composite was studied. The inorganic content of the resulting composites was found to vary with PEI and TMOS fraction (in ethanol) used for silica formation. In particular, the silica yield was highest at an intermediate

187

concentration of 3 wt% PEI. Silica yield increases with decreasing the TMOS fraction at constant PEI concentration used for silica formation. Silica formation by PEI is argued to be the combined effect of the ability of PEI to: (1) catalyze siloxane bond formation and

(2) flocculate the silica particles formed during the reaction.

To study formation of silica on surfaces, PEI was deposited either as layer-by- layer similar to the one described in Chapter 5 or with single layer onto the quartz substrate. In both cases, silica formation was observed when the surfaces were exposed to TMOS, further confirming that PEI can catalyze the silica formation on surface. Silica formation on PEI nanofibers and foams by TMOS exposure will be further discussed in

Chapter 7.

6.2 Introduction

See Section 1.4 in Chapter 1.

6.3 Experimental Procedure

Tetramethyl orthosilicate (TMOS) (CAS 681-84-5) and tetraethyl orthosilicate

(TEOS) (CAS 78-10-4) were used as alkoxysilane precursor for silica formation were purchased from Sigma-Aldrich. Branched poly(ethyleneimine) (PEI) (water free) with two different molecular weights were purchased from Sigma Aldrich and used after dilution to desired concentration with ultra-pure deionized (DI) water (ρ >18 MΩ cm).

Low molecular weight (Mw) PEI (CAS 25987-06-8) has MW = 800 kDa, PDI=1.33 and high Mw (CAS 9002-98-6) has MW = 25000 kDa, PDI = 2.55. Both of the polymers are expected to be highly branched with the ratio of the primary, secondary and tertiary

188

amines in the structures to be close to 1:2:1.

Typically, the silica formation experiments were carried out by addition of 500μl of TMOS/Ethanol mixture to varying concentration of PEI solution in the ratio of 1:1 v/v

(total volume 1.0 ml) in a centrifuge tube (Fisher Scientific), unless otherwise noted in the text. The PEI concentration used in silica formation was varied from 1 wt % to 40 wt% in DI water (1, 3, 5, 10, 20 and 40 wt %), while the TMOS concentration was varied from 0.1 to 1.0 (v/v TMOS/ethanol) (0.1, 0.3, 0.5, 0.7, 0.9 and 1.0). The solid silica precipitates obtained from each combination of the PEI and the TMOS concentration were diluted by DI water to 10 ml and then centrifuged for 1 h at 5000 rpm. The solids were separated from the clear supernatant solution and dried at 50 °C in vacuum for at least 12 h before weighing. The dried precipitates were then analyzed by TA Instruments

Q500 thermo-gravimetric analyzer (TGA) at heating rate of 20 °C/min from room temperature to 1000 °C. The inorganic (silica) fraction of the silicified samples was then measured by calculating the weight percentage remaining at 850 °C in the TGA spectra.

The silica yield was then obtained by following equation,

weight of solids (gC ) × inorganic fraction at 850o ×100(%) Silica Yield (%) = (6.1) [Si] concentration (mol / l ) × V ( l ) × 60( g / mol )

Here, the concentration of TMOS, [Si], is multiplied by the molecular weight of the silica (SiO2) in the denominator. It is to be noted that the weight of the composite obtained after drying is multiplied by the inorganic fraction at 850 °C to obtain the yield using equation (6.1). Thus, any additional condensation of silanol groups that takes place during the heating to 850 °C (in TGA at 20 °C/min) gets inadvertently accounted in the

189

silica conversion.

Silica morphology was analyzed using Scanning Electron Microscopy (SEM)

(Philips XL30 Environmental SEM) after drying at 50°C in vacuum for 12 h. Prior to

SEM analysis, samples were sputter coated with palladium for 30 s using a current of 45 mA under argon at a pressure of approximately 200 mtorr that yields a coating thickness ca. 50 Å. Elemental analysis was performed using Energy Dispersive X-ray (EDX) probe attached to the same SEM described above to confirm the existence of silicon and oxygen.

For the PEI deposition on the surface, silicon wafer (~1 cm × 1 cm) or quartz (1”) disc were used as a substrate. Prior to the deposition, the substrates were cleaned with

Piranha solution (70:30 H2SO4:H2O2) for 1 hour at 60 °C and rinsed with DI water. PEI was deposited on to silicon wafer by two methods: (i) layer-by-layer with alternating deposition of polyanion poly(4-sodium-styrene sulfonate) (PSS) by polyelectrolyte spin assembly (PSA) process or (ii) spin-coating a single layer of 3 wt% PEI solution onto the silicon wafer at 3000 rpm for 40 s. For the latter process, to form a thick coating of the

PEI, up to 4 layers of the PEI solution was spin coated at the same condition. Unlike poly(l-lysine), a single surface PEI layer could be achieved owing to its branched structure that promotes adhesion to various surfaces.

The silicification of the surface was then achieved by exposing the surface to either pure TMOS or hydrolyzed TMOS precursor. The silicic acid precursor solution

(113 mM) or TMOS was then applied drop-wise to wet the whole surface after which the surface was covered using standard microscope cover-slips. After 15 minutes of

190

silicification, samples were rinsed with DI water and dried in the vacuum oven for at least

4 hours at 50 °C. The samples were then analyzed using environmental scanning electron microscopy (ESEM) according to procedure mentioned earlier.

6.4 Results and Discussion

6.4.1 Silica formation from PEI in solution

Aqueous poly(ethylene imine) (PEI) solutions catalyze silica formation instantaneously from alkoxysilane precursor, tetramethylorthosilicate (TMOS). Addition of aqueous solutions of low Mw PEI (branched) with varying concentration from 0.5-40 wt% to TMOS (non-hydrolyzed) in the ratio 1:1 v/v (total volume =1.0 ml) resulted in the formation of white precipitates of silica within 10 seconds. It was observed that water is required for such rapid silica formation from TMOS by PEI; addition of anhydrous PEI to TMOS did not result into silica formation. On the other hand, polycations similar to

PEI, like poly(l-lysine) (PLL) hydrobromide and poly(allylamine) (PAH) hydrochloride does not have ability to cause silica formation directly from TMOS. This was verified by addition of 0.5 wt% of PLL and PAH to the pure TMOS, which resulted into no silica formation even after extended period of 2 h. Such ability of aqueous PEI solutions for silica formation directly from TMOS, is also reported in the recent study by Yuan et al138 with linear PEI, where instantaneous precipitation of the silica on the fibrous crystalline hydrogels of linear PEI was observed when exposed to TMOS. Linear PEI forms continuous fibrous crystalline structure in aqueous solutions resulting into formation of hydrogel. Silicification by exposing the aqueous solution of linear PEI to

191

TMOS/methanol mixture resulted into formation of multiply shaped silica morphology.139,140

Silica formation was further investigated by varying the type of alkoxysilane, molecular weight and the equivalent amount of the PEI used in the reaction. The silica formation was investigated by addition of varying amount of 3 wt% of PEI solution to constant mass of TMOS (3.48 × 10-3 mole). Addition of aqueous PEI solution to TEOS, an alternative sol-gel precursor to TMOS, resulted in negligible silica yield compared to silica yield obtained by PEI addition to TMOS. Figure 6.1 shows the silica yield obtained with increasing amount of PEI (3 wt% solution) to TMOS or TEOS. The solids obtained from the precipitation were dried at 50 °C overnight in vacuum and then the silica yield were calculated based on the total solids expected for 100% conversion of

TMOS or TEOS. It was assumed that all the PEI added to the reaction was incorporated into the solids. This assumption will be further tested by the TGA analysis shown later.

Importantly, the silica yield from TEOS addition is very low and plateaus at ~7 wt% yield at higher amount of PEI addition. The low silica yield with TEOS could be limited by the hydrolysis step in the formation of silica (see Chapter 1). The rate of hydrolysis is lower for TEOS compared to TMOS, which in turn might limit the silica formation.108

The silica yield from TMOS addition increased with the amount of equivalent PEI, irrespective of the molecular weight of the PEI used in the study. Addition of more than

21 mg of equivalent PEI (3 wt% solution) results into a thick paste of the gluey material limiting the further addition of higher amount of PEI in to the reaction. For this case, the silica obtained is around 80 % of the theoretical silica yield.

192

Silica formation from TMOS by aqueous PEI solutions can occur by two competitive processes that we hypothesize. In the first hypothesis, the PEI chains merely bridge the already formed silica particles by instantaneous hydrolysis and condensation reactions of TMOS. It has been shown that PEI can strongly adsorb onto the silica surface at pH > 2.0. Lindquist and Stratton141 have studied such flocculation of the silica sol particles (LudoxTM) by PEI. At pH 3-9, the flocculation is governed by electrostatic interactions between negatively charged sol particles and protonated PEI. Above pH 9, where PEI is less protonated, the adsorption occurs by hydrogen bonding interactions of hydrogen in imine group of PEI with oxygen of the surface silanol groups and the resulting flocculation occurs by bridging mechanism, where the silica sol particles are bridged by the PEI chains. There exists a critical flocculation concentration (CFC) of

PEI below which the flocculation does not occur and a redispersion concentration (RC)142 above which the colloidal particle is coated with polymer and bears the same charge as the polymer and is redispersed. The flocculation regime is between CFC and RC.

Lindquist and Stratton141 showed that the flocculation regime depends mainly on the pH and ionic strength of the solution and poorly on molecular weight of PEI. At the pH of

11, the same for our study, the CFC and RC values for PEI of molecular weight 18,400

Da is 8 mg/l and 158 mg/l. In our study the concentration of PEI employed (10000 mg/l) is well above the redispersion concentration (RC) of PEI. The fact that the solid precipitates at such PEI concentration above the reported redispersion concentration, suggests that the silica formation is not just by bridging of sol particles formed due to the usual silica polymerization reaction of TMOS.

Alternatively, recent studies suggest that the silica polymerization occurs due to

193

siloxane bond formation by nucleophilic substitution catalyzed by the polyamines.

Several mechanisms have been proposed for polyamine-assisted hydrolysis and condensation of the silica from precursor molecules. Sahai and Delak,216,217 while studying hydrolysis and condensation of a model compound trimethylethoxysilane

(TMES) by oligomeric amines at pH 5.0 by 29Si NMR, have shown a direct correlation between the concentration of conjugate base in amines with its ability to catalyze hydrolysis and condensation. They postulated a nucleophile-catalyzed mechanism of hydrolysis by polyamines216 where the conjugated base of the amine attacks silicon atom in organosilicate to form a penta-coordinate intermediate that differs from conventional

108 base-catalyzed SN2 mechanisms. They found that the hydrolysis rates are orders of magnitude higher than the condensation rates. Besides the hydrolysis reaction, PEI assists in the rapid condensation of the hydrolyzed precursor to form siloxane functional groups. According to the proposed mechanism for polyamine assisted condensation,122,218 polyamine chains forms hydrogen bonds with two molecules of precursor acid (Si-OH) per two repeat unit of amine in the catalyst. This facilitates the

Si-O bond formation by stabilizing the transition state by bringing the reacting species together. The hydrogen in each amine group of the polyamines hydrogen bonds with the oxygen of forming silica, integrating the polyamines within the resulting composite material.

It is possible that one or both of the existing hypothesis result into the observed rapid silica formation. If the first hypothesis of silica sol-particle flocculation by bridging mechanisms is true, then increasing the amount of PEI would lead to re-stabilization of the sol particle solution, as is commonly observed for the case of PEI. Ries142,219 has

194

shown that in colloidal particle solutions, excess polyelectrolyte, if present, is coated on the colloid and the colloid particle bears the same charge as the coated polyelectrolyte and thus is redispersed. Such redispersion of the sol-particles was not observed during silica formation or after dilutions of the silica obtained from PEI addition to TMOS.

Alternatively, if silica formation occurs by biocatalytic mechanisms, the resulting silica should contain proportional amount of PEI incorporated in the composite. Studies by

Kroger et al118 has found that silica formed from polyamine catalysis isolated from silaffins has a relative composition of 1.25:1 SiO2/polyamine based on weight. Thus, if the silica formation occurs by the catalytic effect of PEI, the latter should get proportionally incorporated in the formed silica. If the silica formation occurs solely by bridging of sol-particles, then increasing the PEI addition to constant amount of TMOS should not result in proportional increase of the incorporated PEI and should result into saturation of the PEI incorporated in to the silica formed (by bridging mechanism).

To test this above mentioned two competing processes, we systematically varied the PEI and TMOS concentration and studied the resulting solids formed by TGA and

SEM. The TGA analysis was intended to measure the amount of organic (PEI) incorporated in to the solids, while the SEM analysis was pursued to yield the effect of

PEI and TMOS concentration on the resulting morphology of silica. Furthermore, the optimum PEI and TMOS concentration with respect to silica yield was deduced.

6.4.2 Effect of PEI and TMOS Concentration on Silica Formation

Silica formation was studied by addition of PEI concentration (1 wt % to 40 wt%) to the TMOS of varying fraction in ethanol. During each time a fixed volume (500 μl) of

195

PEI solution was added to the same volume of the TMOS/Ethanol mixture (total volume

=1.0 ml). All combinations of PEI and TMOS concentration studied resulted into an instant formation of white silica precipitates within a minute. After drying these precipitates, it was observed that the solids formed from addition of PEI solution of low concentration (1-7 wt %) to TMOS were in the form of a white powder, while the solids obtained from PEI of 20 wt % or higher concentration were tacky due to excess PEI.

The inorganic content of the solid precipitates depends on PEI and TMOS concentration used for silicification, as evident from TGA weight loss curves of Figure

6.2 (a,b). All the TGA curves in Figure 6.2 have weight loss in two temperature range

(regimes). The first weight loss regime occurs at temperatures below 120 °C owing to the removal of moisture adsorbed due to the hygroscopic nature of the PEI incorporated in the solids. The second weight loss regime occurring in temperature range of 250 °C -

600 °C, corresponds to degradation of organic (PEI) molecules and additional condensation of silanol units that liberates water. Figure 6.2(a) shows the TGA curves of solids obtained by addition of PEI of increasing concentration to pure TMOS at the ratio

1:1 (v/v). Increasing the PEI concentration leads to decrease in the inorganic fraction of the solids (residue at 850 °C). Similarly, the weight loss in two regimes increases with increasing PEI concentration used in composite formation. This suggests that the resulting composite incorporates PEI into the solids proportionally to the PEI added during the reaction. Figure 6.2(b) shows TGA curves of the silica obtained by addition of

7 wt % aqueous PEI solutions to varying TMOS fraction in ethanol. The weight loss regimes in Figure 6.2(b) are similar to Figure 6.2(a). Interestingly, the inorganic content

196

of the solids (residue at 850 °C) in Figure 6.2b slightly increases with decreasing TMOS fraction to 0.7 from 1.0 and then decreases sharply with further decreasing the TMOS concentration. This suggests that the silica/PEI ratio (or inorganic fraction) in the resulting composite strongly depends on the TMOS fraction. Thus, the inorganic fractions of the solid precipitates were calculated from TGA curves for all the solids formed by varying PEI and TMOS concentration.

The inorganic fraction strongly depends on the PEI and TMOS concentration used for silica formation. Figure 6.3 shows the inorganic fraction (residue at 850 °C) of the solids obtained from addition of PEI solution (1-40 wt% concentration) to TMOS of varying fraction. Each inorganic fraction is obtained at different PEI concentration. For the lower PEI concentration of 1 wt% in the reactant, the inorganic fraction increases from 70 to 80 wt% as the TMOS fraction is increased from 0.1 to 1.0. On other hand, for

PEI concentration of 20 wt % or higher, the inorganic fraction first increases with increasing TMOS fraction up to 0.7 v/v and then decreases with increasing TMOS fraction. The increase in the inorganic content with decreasing TMOS fraction up to 0.7 is an effect of dilution by ethanol. TMOS and aqueous PEI solutions are initially immiscible, unless the hydrolysis of TMOS occurs. Increasing the dilution of TMOS by ethanol, a mutual solvent to both PEI and TMOS, increases the availability of TMOS for hydrolytic condensation, thus slightly increasing the inorganic content. Further decreasing TMOS fraction, or increasing dilution, ultimately leads to lower inorganic fraction.

Now, let us compare the inorganic content of the composites to the one expected

197

if the PEI catalyzes the silica formation. In silaffins, where the catalytic domains consist of oligo-N-methyl propylamine, the silica formed has a relative composition of 1.25:1

118 SiO2/polyamine based on weight; i.e., 55 wt % of inorganic content. In our studies with PEI, the relative mass composition is expected to be 2:1 SiO2/PEI, assuming that the silica formation from PEI occurs by the same mechanism mentioned above for N-methyl propylamine, due to the difference in repeat unit molar mass (43 g/mol for ethyleneimine versus 71 g/mol for N-methyl propylamine). Accordingly, the silica/PEI composition should be ~ 66 % inorganic. The composites obtained from aqueous PEI solution of 3 -5 wt% at almost all TMOS fraction have the inorganic fraction around value of 66 %

(Figure 6.3). Higher inorganic content than 66 %, observed for PEI concentration of

1wt%, suggests condensation of additional silica by usual sol-gel process. Lower inorganic content, i.e. lower silica/PEI ratio, signifies lower conversion of the TMOS to silica. Thus, the dependence of conversion of the TMOS to silica on the PEI and TMOS concentration would further deduce the mechanisms of the silica formation.

The inorganic fraction of the solids at 850 °C (Figure 6.3) and the weight of the solids obtained during the silicification were used to calculate the percentage conversion of TMOS to silica according to Eqn. 1. The conversion of TMOS is optimum at PEI concentration of 3 wt% at all TMOS fraction. This is evident from Figure 6.4 which shows the combined effect of TMOS and PEI concentration on the inorganic silica conversion. The TMOS conversion sharply increases with increasing PEI concentration up to 3 wt% and then decreases for all studied TMOS fraction. Interestingly, the observed maximum (at 3 wt% PEI concentration) corresponds to the inorganic content of

66 % in the composites (Figure 6.3). This is in accordance with the catalytic effect of

198

PEI as discussed previously. The decrease in the silica yield (or TMOS conversion) with increasing PEI concentration is intriguing, but may be due to limited hydrolysis of TMOS at higher PEI (or lower water) concentration. This is further evident from the inorganic fraction where increasing the PEI leads to lower inorganic fraction (Figure 6.3) and hence lower silica/PEI ratio. Furthermore, the conversion of TMOS to silica also depends on the TMOS fraction (or dilution), with increasing dilution by ethanol resulting into higher conversion. This again is the effect of dilution as discussed previously. The dilution makes the PEI and the TMOS more compatible and thus increasing the chance of hydrolytic condensation.

6.4.3 Morphology of Silica

The silica formed by the direct hydrolytic condensation of TMOS by PEI has spherical morphology, with the particle diameter depending on the concentration of the

PEI and TMOS used for silicification. Figure 6.5(a-d) shows the SEM images of the silica formed by the addition of aqueous PEI solution of concentration from 1 wt% to 10 wt% to pure TMOS at 1:1 v/v. A close inspection of all of the images reveals that the particles vary in diameter, and multiple small particles coagulate to form a large particle.

The coagulations of particles increase with increasing PEI concentration. Such coagulation between particles is particularly evident from the multiple edges that are evident in Figure 6.5(d) for silica obtained from 10 wt% PEI concentration. The presence of edges signifies that the particles were not aggregated after a well-defined spherical morphology was formed, but instead the ‘sticking’ occurs simultaneously during the silica formation and then the formed agglomerate grow by further condensation of the

199

monomeric silica precursor which then forms larger diameter particles. Besides PEI concentration, the morphology of silica particle also depends on the TMOS concentration, with increasing TMOS concentration resulting into increase in the particle diameter.

Figure 6.6 (a-d) shows the SEM images of the silica formed from the addition of 7 wt% aqueous PEI solution to varying TMOS concentration in ethanol at the ratio of 1:1 v/v.

The spherical particles formed at 0.1 v/v TMOS fraction have comparatively smaller diameter compared to the one formed at higher concentration. Such particles are also mono-disperse (Fig. 6.6(a)), and further have low silica/PEI ratio (refer Fig. 6.2). The low silica/PEI ratio is due to lower TMOS concentration in the solution. The increase in the silica particle diameter with increasing TMOS concentration is consequence of increased rate of condensation. Furthermore, the coagulations of particles, as previously seen in Figure 6.5, is also evident at all the TMOS concentration less so for the TMOS of lower concentration. Such coagulation of particles is unique, since the silica particles obtained from usual sol-gel reaction and flocculation by the polyelectrolytes have narrow diameter distribution, unlike the particles with large diameter distribution obtained by

PEI. The coagulation of particles is thought to occur due to the adsorption of PEI onto the silica and further growth of the silica on to the adsorbed surface. When sufficient particles are formed, the coagulation then occurs by bridging of the particles as proposed by Lindquist et al.141

In order to further quantify the diameter distribution, average and standard deviation of particles were measured using imageToolTM software for the total of 30 particles from SEM images of Figure 6.5 and 6.6. Figure 6.7(a,b) shows the measurements of average and standard deviation (error bars) obtained from Figure 6.5

200

and 6.6 respectively. The particle diameter first slightly decreases with increasing PEI concentration, reaching a plateau around the PEI concentration of 3 to 5 wt%, and then increases again with increasing PEI concentration. Interestingly, the plateau concentration of 3 to 5 wt% PEI is same as the PEI concentration where the maximum silica yield is obtained (Fig. 6.4). Also, the aggregation of the silica particles is higher for the silica obtained from the PEI concentration of 5 wt% or higher, as evident from the large standard deviation compared to the silica obtained from PEI of lower concentration.

On other hand, increasing the TMOS concentration at constant PEI concentration, leads to proportional increase in the particle diameter as seen in Figure 6.6(b). This effect is merely due to increase in the TMOS concentration and thus increase reaction rate, thus resulting into larger particle diameter, as discussed previously.

In our study, pH and temperature were both held constant and the particle morphologies achieved were different from that formed from silicic acid polymerization at high pH, suggesting an alternative mechanism than bridging of the silica particles by polymer. It has also been shown that the silica morphology is dictated by the morphology of the PEI in solution. For example, Yuan et al have shown that the silica is formed on the fibrillar PEI in solution, similarly, various block-copolymers or polyamine directed structure has been used to direct the final silica morphology. Accordingly, various models have been proposed for the observed correlation between the formation of silica nano-spheres by polyamine catalysis and the silica found in diatoms. Vrieling et al220 suggested that polyamines not only induce rapid precipitation of silica but also assist in aggregation process by interacting with silica. Brunner and Sumper221 have proposed that the silica formation by polyamines and hence by proteins found in diatoms, occurs

201

by micro-phase separation of polyamines and silica precursor by forming a coacervate

(“liquid precipitate”), which finally hardens by further condensation and silica formation.

This is in agreement between the phase separation mechanism of polyamines proposed by

Kroger and Sumper82 and discussed in detail in Chapter 1. Besides these studies, various other studies 82,122,222-224 have supported the proposed polyamine assisted condensation of the silica precursor.

Thus in our case of silica formation by PEI, it is possible that the phase separation of the aqueous PEI and TMOS by the formation of the polyamine coacervate (or liquid precipitate) direct the formation of silica. This could explain the increase of the silica particle diameter and coagulation with increasing PEI concentration. Furthermore, the pH and the temperature are same to obtain silica by varying concentration of PEI and

TMOS. This signifies that morphological difference in particle diameter is not due to increase TMOS hydrolysis or condensation rate.

6.4.4 Silica formation on Surfaces

In the previous chapter, silica formation on the surfaces was studied by condensation of hydrolyzed TMOS (in 1mM HCl) on to the surface where either poly-l- lysine or PEI was localized in polyelectrolyte multilayers. This section describes the silica formation on surface where PEI is localized, by directly exposing the surface to pure non-hydrolyzed TMOS. Poly-lysine does not catalyze silica formation directly from

TMOS in solution and on surfaces as was verified in previous chapter. In the present study, the PEI was localized on to the surface by two methods: (i) Polyelectrolyte multilayers obtained by alternating deposition of polyanion, poly(4-sodium-styrene

202

sulfonate) (PSS) by polyelectrolyte spin assembly (PSA) process or (ii) spin-coating a single layer of 3 wt% PEI solution onto the silicon wafer at 3000 rpm for 40 s. In both the cases, the silica formation occurs on to the surface, albeit with slightly different morphology than the silica obtained from solutions as described in detail below.

Silica formation occurs on the surfaces where PEI is localized either in a single layer or in polyelectrolyte multilayers, when those surfaces are exposed to TMOS for 15 minutes. Figure 6.8(a,b) shows the SEM images of the silica obtained after exposing the surface coated layer-by-layer of PEI (low Mw) and PSS until 10 bilayers, to pure TMOS for 15 minutes. The surface were PEI is localized by layer-by-layer up to 10 layers

((PEI-PSS)10-PEI) has a sparse regions of the silica formation. Interestingly, the structure of the silica formation is same as those obtained by polylysine as shown in previous chapter. Further, the AFM analysis of such smooth film revealed the presence of small spherical particles of diameter ~50 nm (Fig. 6.8(c)). These particles are closely packed, unlike the silica obtained from the solution or silicification of single PEI layer on surface.

The silica formed on the surface where only PEI is present as a single layer, has porous structure with presence of particles with larger diameter than when localized on surface with LbL process. This is evident in Figure 6.9 (a, b) where the low Mw PEI and high

Mw PEI was localized on to the surface by spin-coating of single layer. Thus, the presence of oppositely charged polyelectrolyte, in this case polyanion PSS, transforms the porous structure silica structure into sparsely nucleated silica. Such sparsh regions of silica formation might be due to sparsh presence of PEI on the multilayer surface. The multilayer has the presence of holes as was described in detail in Chapter 2. In any case, the silica formation occurs on surface by directly exposing the PEI localized surfaces to

203

TMOS. Furthermore, when polycation other than PEI, like polylysine or polyallylamine hydrochloride were localized by LbL process and no silica formation was evident by exposing to TMOS and subsequently analyzing those in SEM (see previous chapter).

Thus PEI can catalyze the direct hydrolytic condensation of TMOS on the surfaces, similar to the mechanisms in solutions.

6.5 Conclusions

Silica formation from the direct hydrolytic condensation of TMOS is studied in the solution and surfaces. The instantaneous formation of the silica is observed when the

PEI aqueous solution is added to TMOS. Such direct hydrolytic condensation is not observed if the TMOS is replaced with other alkoxysilane precursor, TEOS. The hydrolysis rate is the rate determining step for this process. The PEI may act to stabilize the intermediate product silicic acid and then the condensation follows giving the resulting silica according to the various mechanisms of catalysis by polyamines proposed in the literature. The PEI and the TMOS concentration were varied in order to deduce the effect on the resulting silica formed. The inorganic fraction varied depending on the PEI and TMOS concentration, and was higher for the lowest concentration of the PEI used in the study. The silica yield was highest at an intermediate concentration of 3 wt% PEI and increased with decreasing the TMOS fraction. The latter is an effect of dilution where the ethanol a mutual solvent to PEI and TMOS. The silica precipitated from PEI had spherical morphology and that increasing the PEI concentration increases the coagulation of the particles, further increasing the diameter. The coagulations of the particles were more evident from the large diameter distribution of the particles. Finally, silica

204

formation was achieved on to the PEI localized on the surfaces on polyelectrolyte multilayers. The PEI localized on the surface had similar morphology of the particles to that of solutions and single PEI layer gave higher density of the particles compared to the one obtained when PEI was localized onto the multilayers. This simple process of direct hydrolytic condensation of TMOS by PEI can be used for instantaneous silica formation from TMOS.

205

100

60 (excluding PEI)

2 TMOS-LMw TEOS-LMw TMOS-HMw 40 TEOS-HMw

% yield based on SiO % yield 0 0 5 10 15 20 25

Equivalent PEI added (mg)

206

100 PEI concentration 80 1 wt % 3 wt % 5 wt % 60 7 wt %

40 20 wt % Weight Remaining (%) Remaining Weight 20 40 wt % (a)

0 0 200 400 600 800 o 100 Temperature ( C)

TMOS/Ethanol 80 0.9 0.7

60 0.5 0.3 1.0 40

0.1 Weight Remaining (%) 20 (b) 0 0 200 400 600 800 Temperature (oC)

207

100

80 PEI 1 wt%

3 wt% 60 C by TGA (%) o 5 wt%

40 7 wt%

10 wt%

20 20 wt%

Residue at 850 40 wt%

0 0.0 0.2 0.4 0.6 0.8 1.0 TMOS fraction (v/v) in ethanol

Figure 6.3: Inorganic fraction of the solids formed by PEI addition to TMOS. Each individual point represents the solution concentration of PEI and the TMOS fraction used for the silica formation. The inorganic fractions were calculated from weight % of the solids remaining at 850 °C from TGA curves.

208

1-tmos 0.9-tmos 100 0.7-tmos 0.5-tmos 0.3-tmos 0.1-tmos 80

silica yield (%) 40

0 0 5 10 15 20 25 30 35 40

PEI wt %

Figure 6.4: The silica yield obtained at various PEI and TMOS concentration. The silica yield was calculated from Eqn. 6.1.

209

(a) (b)

Figure 6.5: SEM images of silica obtained by addition of PEI solution concentration of (a) 1 wt%, (b) 3 wt%, (c) 5 wt% and (d) 10 wt% to pure TMOS at 1:1 v/v.

210

(a) (b)

Figure 6.6: SEM images of silica obtained by addition of 7 wt % PEI solution concentration to TMOS fraction of (a) 0.1, (b) 0.5, (c) 0.9 and (d) 1.0 at 1:1 v/v.

211

2000

(a) 1500

1000

(nm) Diameter 500

0 024681012 PEI concentration (%)

2000

(b) 1500

1000

500 Particle Diameter (nm) Particle

0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 TMOS concentration (v/v) in Ethnaol

212

(a) (b)

(c)

Figure 6.8: Silica formed on to the surface of polyelectrolyte multilayers containing

(PEI-PSS)10-PEI layers upon exposure to TMOS for 15 minutes, (a,b) SEM image, scale bar 10 μm and 2 μm, resp., and (c) Contact-mode AFM image.

213

(a) (b)

214

CHAPTER 7

7 Silica Formation in Polymer Scaffolds

7.1 Synopsis

Biomimetic formation of silica from synthetic polyamines such as poly(ethylene imine) (PEI), inspired by the proteins found in diatoms and sponges, was investigated in

Chapter 6 as a potential route to silica formation compared to the conventional sol-gel process. In this chapter, silica formation onto scaffolds of linear PEI obtained via electrospinning or freeze drying techniques is investigated. Scaffolds of linear PEI obtained via electrospinning consisted of nanofibers of linear PEI blended with poly(vinyl pyrrolidone) (PVP) in ratio 50:50 (w/w) and is referred as PEI nanofibers.

Similarly, scaffolds of linear PEI obtained via freeze drying of aqueous linear PEI hydrogels consisted of fibrous crystalline PEI in open-cell foam structure and is subsequently referred as PEI foams. The active component in these nanofiber- or foam- based scaffolds, PEI, catalyzed rapid silica formation, within minutes, upon immersion of the scaffolds in the silica precursor, tetramethyl orthosilicate (TMOS). The silica formation in electro-spun nanofibers was then investigated by scanning electron microscopy (SEM), Energy Dispersive X-ray analysis (EDX), thermogravimetric analysis

(TGA) and Fourier-transform infra-red (FTIR) spectroscopy. The silica content in the

PEI/PVP nanofibers could be controlled by pre-treatment of the fibers at different conditions of relative humidity prior to the silicification. Fibers exposed at higher (80%)

215

relative humidity led to higher inorganic (silica) content compared to those exposed to relative dry conditions (<20% relative humidity). Calcination of the fibers indicated that silicification proceeded across the whole fiber cross section that consisted of nano- structured silica. Similarly, silica formation in the PEI foams obtained by freeze drying of linear PEI hydrogels was investigated by varying the initial PEI concentration. The silica formation on such forms by TMOS infiltration was limited by high density of foams at PEI concentration of 10 wt% or above, and by the structural integrity of foams at PEI concentration of 3 wt % or lower. The optimum PEI concentration to obtain the

PEI scaffolds for uniform silica formation was found to be ~5 wt%. The morphology of the polymer/silica foams before and after silicification and calcination was then investigated by SEM. SEM analysis revealed that silica formation in the PEI nanofibers and foams occurred primarily on the linear PEI and not in the voids of either nanofibers or the foams. Thus, for both the cases of PEI nanofibers and foams, PEI plays a central role in the rapid silica formation to obtain organic-inorganic hybrid composites when the fibers or foams are exposed to TMOS.

7.2 Introduction

See section 1.3.3.1, 1.3.3.3 and 1.3.3.4 in Chapter 1

7.3 Experimental Methods

7.3.1 Synthesis of Linear Poly(ethyleneimine) (PEI)

Linear PEI was synthesized from the precursor poly(2-ethyl-2-oxazoline) (PeOz)

(Polyscience, Mw = 500 kDa) according to a previously reported procedure.138 In brief,

216

PeOz was hydrolyzed in a 5 M HCl solution with a 3:1 molar ratio of HCl to acetylethyleneimine units at 90 °C for 12 h (polymer concentration = 0.145 g/ml). A white precipitate formed during the reaction was filtered and washed with acetone, followed by and dissolution in ultra-pure deionized (DI) water (Milli-Q, ρ > 18 MΩ.cm).

The solution was either directly precipitated by the addition of 0.5 M KOH or sealed in a dialysis tube (Spectra/Por membrane, Mw cut-off 3500 Da, Spectrum Labs) and dialyzed against a 14 % aqueous ammonia solution for 3 days, replacing the solution every 24 h.

In both cases, the white precipitate obtained was filtered, washed with excess acetone and

DI water, and dried in vacuum at 40 °C for 24 h.

The conversion of oxazoline (Oz) units in PeOz to imine in PEI was characterized

1 by H NMR spectroscopy (Varian Inova 600 MHz) in CD3OD at room temperature.

Figure 7.1 shows the 1H NMR spectra of (a) PeOz and (b) PEI obtained from hydrolysis of PeOz. The inset shows the chemical structure of each individual polymer along with their respective labels of the peaks expected for each proton. The conversion of the oxazoline (Oz) units in PeOz to imine unit in PEI was calculated by evaluating the Oz peak ( δH = 2.4 ppm, —CH2— (label b, Fig. 7.1a) and δH = 1.1 ppm, —CH3 (label c, Fig.

7.1a)) relative to the ethyleneimine peak ( δH = 2.74 ppm (label a, Fig. 7.1b)).

Additionally, the ethyl protons adjacent to nitrogen in PeOz ( δH = 3.5 ppm, —CH2CH2N-

Oz— (label a, Fig. 7.1a) shift upon conversion to PEI ( δH = 2.74 ppm, —CH2CH2ND—, label a, Fig. 7.1b). The conversion of the oxazoline units to ethyleneimine units was 94

%, irrespective of the precipitation method used.

7.3.2 Electrospinning of PEI/PVP blend

217

Electrospinning (e-spinning) of linear PEI solutions in ethanol spanning 5 wt-%-

15 wt-% in concentration resulted in submicron fibers (here termed “nanofibers”) that quickly absorbed moisture at ambient conditions due to hygroscopic nature of the PEI.

This eventually led to the formation of films by ‘flash welding’ of the fibers. To solve this problem, a second miscible polymer was incorporated. In particular, our linear PEI was solution-blended with poly(vinyl pyrrolidone) (PVP) (Mw 350,000 Da) (Aldrich) or its precursor polymer PeOz.43 Scheme 1 shows the structure of the polymers. Nanofibers stable to humid environments were obtained by e-spinning of linear PEI/PVP (50/50 w/w) blends in 5 wt-%-15 wt-% concentrated solutions in ethanol at a flow rate between 0.1 ml/h – 0.4 ml/h, a voltage gradient 0.8 kV.cm-1 over the 10 cm distance between the tip and the collector, and processing time of 3 h. For e-spinning, polymer solutions of a given concentration were pumped through a stainless steel needle (I.D. 0.26 mm) using a syringe pump (KD Scientific, Model 780100) at a specified flow rate. The needle was held at high electric potential relative to ground using a high voltage power supply (0-20 kV, Ultravolt, Model 30A12-P4) and controlled by a DC power supply (0-6 V, Agilent,

Model E3630A). The samples were collected in a steel (SS type 304) mesh (screen) of square shape with area ~ 25 cm2, wire gauge of 40 × 40 meshes per linear inch, and wire size 0.0075 inch (Custom Filtration Inc, MN) for approximately 3 h.

7.3.3 Silicification and calcination of fibers

Silicification of the resulting nanofibers was accomplished by immersing the fiber mats (w/ mesh) in tetramethylorthosiliciate (TMOS) for 10 minutes followed by rinsing with excess acetone multiple times followed by drying in vacuum at 40 °C for at least 4 h.

218

The effect of water content on silica content in the PEI/PVP fibers was studied by exposing the fiber mats to different controlled humidity conditions for at least 24 h before immersing the fiber mats into the TMOS for silicification. Controlled humidity conditions were achieved by use of a small chamber within which aqueous lithium chloride solutions (0.143 g LiCl/ml for 80% relative humidity and 0.367 g LiCl/ml for

40% relative humidity) or desiccant (for relative humidity <20 %) were placed. For the chamber with desiccant, the relative humidity fluctuated between 10 % and 20 % and hence is reported to be <20 % in the text. The humidity was measured using a digital hygrometer (Traceable® Control Company, TX). The silicified fibers were removed from the mesh after drying for subsequent calcinations and other characterization. Calcination of the silicified nanofibers was then achieved by heating the fiber mats in a crucible to the

600 °C in a furnace for 1 h. The resulting materials were fragile and required careful handling.

7.3.4 Preparation of linear PEI foams

The unique ability of linear PEI to gelate water and form a fibrous crystalline hydrogel, as reported previously,139 was explored to obtain the foams by freeze drying.

To obtain the hydrogels at room temperature, linear PEI was dissolved at various concentrations (1-10 wt% ) in hot water (~60 °C) under constant stirring in a closed vial for 2-3 h. The resulting clear solution (at 60 °C) was then allowed to cool for 2 h in plastic vials to room temperature, which resulted in the gelation of a white crystalline hydrogels of PEI. The melting temperature of crystalline PEI was analyzed using differential scanning calorimeter (DSC) and found to be 47.3 °C, with ΔH =37 J/g, at

219

second heating cycle at 10 °C/min The hydrogels were then freeze-dried to obtain linear

PEI foams. Freezing of the hydrogels was achieved at ca. -70 °C in dry ice/acetone bath for 15 minutes, followed by the sublimation of the ice crystals obtained during the freezing in a dryer (pressure ~ 15-50 × 10-6 bar). The freeze drying is required to preserve the structure of the PEI foams, which otherwise collapses in the drying process at ambient conditions. Bulk densities of foams were calculated from the weight and the physical dimensions of the foams obtained after the freeze drying process. (Caution: PEI foams are extremely moisture sensitive and should always be stored in desiccators.

Adsorption of moisture results into significant shrinkage of the foams).

Silica formation in foams was achieved by vacuum infiltration of TMOS into the foams. For TMOS infiltration, the foams were soaked in the TMOS and continuous vacuum was pulled to remove the air bubbles trapped inside the foam. No visible change in the sample was observed during the infiltration process. After 10 minutes of TMOS infiltration, the sample was soaked in acetone bath for 10 minutes to remove the residual

TMOS, followed by excess (3×) rinse in acetone and finally by DI water. The samples were then dried in a vacuum oven at room temperature for at least 12 h. The silica/PEI composite obtained was then stable against shape change at ambient humid conditions.

7.3.5 Characterization

The electrospun nanofibers and linear PEI foams were analyzed using Scanning

Electron Microscopy (SEM) (Philips XL30 Environmental SEM) before and after silicification. For the nanofiber case, the average and standard deviation (SD) of fiber diameter from SEM images were calculated using ImageToolTM (v3.0) image processing

220

software for total of 50 diameter measurements. Pixel dimensions of the images were calibrated from the scale bar in the image. Calcined fibers were imaged using high resolution SEM (Field-Emission Gun Scanning Electron Microscope Hitachi S4500).

For both cases of SEM measurements, prior to analysis, samples were sputter-coated with palladium for 30 s using a current of 45 mA under argon at a pressure of approximately

200 mTorr, yielding an coating thickness ca. 50 Å. Elemental analysis was performed using Energy Dispersive X-ray (EDX) probe attached to the same SEM described above.

The nanofiber webs and silicified PEI foams were characterized using a TA

Instruments Q500 thermo-gravimetric analyzer (TGA) heating at rate of 20 °C/min from room temperature to 1000 °C. The inorganic (silica) fraction of the silicified samples was measured by calculating the weight percentage remaining at 900 °C in the TGA spectra. Fourier Transform Infra-Red analysis (FTIR) of nanofibers was performed by a potassium bromide (KBr) pellet method. KBr pellets were made by mixing ~1 mg of sample with 90 mg of FTIR-grade KBr. FTIR (Thermo Nicolet - Nexus 870) spectra of the prepared KBr pellets were recorded with a range of wave numbers spanning 400 cm-1 to 4000 cm-1 with averaging over 64 scans.

7.4 Results: Silica formation on electrospun nanofiber scaffolds

Electrospun nanofibers of linear poly(ethylene imine) (PEI) and poly(vinyl pyrrolidone) (PVP) blends from their ethanol solutions were obtained using conditions specified above. Figure 7.2 (a,b) shows the SEM image of nanofibers obtained by e- spinning a PEI/PVP (50/50 w/w) solution in ethanol at a flow rate of 0.1 ml/h and at concentrations of 5 wt% and 10 wt%, respectively. Analysis of Figure 7.2a obtained

221

from the 5 wt% PEI/PVP solution indicates a broad diameter distribution, with an average diameter of 452 nm and standard deviation, SD, of 210 nm. Close inspection of the micrographs reveals that the large standard deviation values are a reflection of a bimodal diameter distribution, with the presence of relatively smaller diameter fibers along with larger fibers (Fig. 7.2a). The origin of such a bimodal distribution in the

PEI/PVP fibers is not known. Increasing the concentration of the PEI/PVP solution to 10 wt% (Fig. 7.2b), while having other parameters the same, resulted in a narrower distribution of fiber diameters and a decrease in the average diameter to 285 nm (SD = 56 nm), compared to the 5 wt% solution. The diameter of fibers could be controlled by changing the concentration and flow rate of the polymer solution. For example, electrospinning fibers with a 10 wt% solution at a flow rate of 0.3 ml/h through an electric field of 0.8 kV cm-1 over a 10 cm tip-to-collector distance resulted in a large average diameter of 1147 nm (SD = 313 nm) (Fig. 7.2c). The fibers which appear as collapsed tubules are in fact a result of merging or welding of the two individual fibers.

Such welding results in the large SD of such nanofibers. Besides the increase in fiber diameter, increasing the solution concentration for e-spinning also increases the web density of the fiber mats as evident in Figure 7.2c. The effect of web density on the silicification and subsequent hybrid fiber formation will be not considered in this paper, but does warrant future attention.

PEI nanofibers, shown in Figure 7.2, were silicified by immersing in TMOS for

10 min, rinsing in excess acetone to remove unreacted TMOS, and finally drying at 40 °C for 4 h under vacuum. The PEI/PVP nanofiber web became detectably stiffened by silicification. Figure 7.3(a,b) shows the SEM images of silicified PEI/PVP fibers

222

corresponding to the non-silicified samples of Figure 7.2(a,b). Interestingly, our silicification process caused adjacent fibers to ‘weld’ at contact points, as evident in the

SEM images shown in Figure 7.3(a,b). Additionally, the fiber diameters increased during silicification, with fibers obtained from the 5 wt% PEI/PVP solutions (Fig. 7.3a), showing a greater increase in diameter than fibers spun from the 10 wt% solution (Fig.

7.3b). For example, the fiber diameters in Figure 7.3a, while having a similar distribution to those in Figure 7.2a, almost tripled to 1383 nm (SD=394 nm). In contrast, the average fiber diameter in Figure 7.3b increased by ca. 100 nm to 390 nm (SD = 96 nm).

Furthermore, the silica formation on the dense nanofiber webs as shown in Figure 7.2c, resulted into formation of porous film (Fig. 7.3c). Due to silica formation in such dense nanofiber webs resulted into bridging of the fibers, though the underlying structures of the fibers is still visible. The observed difference in diameters and subsequent silica formation might be due to a difference in water contents in the fibers before silicification.

The equilibrium water content affects the level of silicification from PEI as was verified by TGA analysis shown later. Nevertheless, rapid silica formation - within minutes - was achieved by immersion of the fibers in pure TMOS.

PEI catalyzes the hydrolysis and condensation of TMOS to form silica during the short exposure of TMOS to electrospun PEI nanofibers for 10 minutes. To confirm this,

EDX spectra of the fibers before and after silicification was collected and compared to a control sample of PVP fibers immersed in TMOS. The PVP fibers were obtained by electrospinning a 10 wt% PVP solution in ethanol using the same conditions described above for PEI/PVP nanofiber processing. After immersion in TMOS, such PVP fibers did not show any presence of elemental silicon in EDX analysis as seen in Figure 7.4 and

223

also showed no change in diameter (data not shown); unlike the silicified PEI/PVP fibers.

In contrast, EDX spectra of silicified PEI/PVP fibers showed a strong peak of elemental silicon present in the EDX spectra (Fig. 7.4). On other hand, the oxygen peak shows very little change upon silicification, despite the fact that silica is forming. The origin of such discrepancy is not exactly known, but may be within experimental error. The elemental analysis showed that the atomic weight percent of silicon was ca. 58 % and that of oxygen to be 42 %, close to the one found in silica (~50 % for oxygen and silicon).

Nevertheless, the absence of silicon peak in the PVP fibers exposed to TMOS signifies further that the silicon peak seen in PEI/PVP silicified fibers is not merely due to TMOS adsorption and subsequent silica formation by sol-gel hydrolysis and condensation.

TGA experiments enabled further quantification of nanofiber silicification, as revealed in Figure 7.5. It was found that the inorganic (silica) yield in the electrospun

PEI/PVP nanofibers requires PEI and varies with water content in the fibers as evidenced by TGA analysis of fibers pre-treated with different levels of humidity before silicification (Fig. 7.5). Consistent with EDX results discussed above, Figure 7.5 reveals that the control PVP fibers after TMOS exposure had a negligible inorganic content of

4.7 wt% at 900 °C. Further, the non-silicified fibers did not show any inorganic content, as was expected. Depending on the pre-treatment of fibers at different relative humidity levels, the silica content was found to vary significantly. For example, fibers exposed to ca. 40% relative humidity prior to silicification yielded only 12 wt% inorganic (silica) content (at 900 °C) compared to 41 wt% for the fibers exposed to 80 % relative humidity prior to silicification. Also, it was found that fibers exposed to 20 % or less relative

224

humidity before silicification yielded a negligible (2.6 wt %) presence of inorganic content.

It was reasoned that the humidity level affects the equilibrium moisture content of the fibers, in turn affecting silicification via the hydrolysis of TMOS by PEI. As evident from Figure 7.5 (curve (i)), a 18 % weight loss of the PEI/PVP fibers occurs at temperatures below 120 °C, roughly indicating the level of water absorption by the fibers.

Upon silicification (curves (iii) and (iv)), the weight loss at 120 °C decreases to 7-8 wt % due to the utilization of water in the hydrolysis. Note that PVP fibers are less hygroscopic than PEI and hence have a weight loss of only 4 % at 120 °C. Water in the fibers gets consumed during hydrolysis and then later gets liberated during condensation of the silanol (Si-OH) groups to form siloxane (-Si-O-Si-) linkages due to PEI catalysis as well as self-condensation of silanol groups at higher temperatures. The latter phenomenon is evident from the significant weight loss observed for the fibers exposed to 80 % relative humidity (curve (iv)) between temperatures of 600 °C and 700 °C. Thus, water content in the fibers is important for rapid silicification by PEI.

Interestingly, silica formation in the nanofibers is not confined to the surface of such fibers but permeates the whole fiber cross-section. This was confirmed by calcination of the silicified PEI/PVP nanofibers at 600 °C for 1 h and imaging the resulting fibers in cross-section with high resolution SEM as shown in Figure 7.6(a,b). It is evident that the calcined fibers consist of silica nano-structures across the whole fiber cross section. The porous structures, as evident in the SEM images, consist of nano- structured silica particles. To our surprise, the fiber structure was preserved, leading to a

225

porous, calcined nanostructure consisting of two length scales: the sub-micron fibers (~

450 nm) and the silica nanoparticles (~ 20 nm) with a relative ratio of silica particle diameter to fiber diameter of about Dsilica/Dfib ~ 0.04 to 0.07. In contrast, calcination of silica obtained from sol-gel condensation is known to result in dense particles,108 rather than the porous structures seen in Figure 7.6.

The formation of silica on electrospun PEI/PVP fibers was further confirmed by

FTIR analysis of fibers before and after silicification and calcination (Figure 7.7). Three important absorbance peaks that indicate the presence of silica include the following: --

Si-OH stretching at 950 cm-1, --Si-O-Si-- symmetric stretching at 790 cm-1 and --Si-O-- asymmetric stretching at 1090 cm-1 (each indicated by vertical reference lines in Figure

7.7). Silica formation in the electrospun fibers is clearly evident from the Si-O-Si vibration band at 790 cm-1, which is otherwise absent in the spectra of PEI/PVP fibers.

Furthermore, a sharp peak at 1090 cm-1, indicating a Si-O vibration appears in the silicified fibers. Interestingly, the Si-O-H stretching mode at 950 cm-1 only has a slight increase in intensity compared to the siloxane (--Si-O-Si--) group at 790 cm-1, implying quite complete condensation during silica polymerization. The increase in intensity of the broad peak around 3400 cm-1 assigned to –OH groups (indicated by * in Figure 7.7), represents either the uncondensed silanol group or the –OH group in the water or alcohol liberated due to the condensation reactions of TMOS. The observed peak of O-H occurs in spite of the drying of silicified fibers for 4 h at 40 °C in vacuum. Other peaks corresponding to various groups in PEI and PVP are also present in the region of 1200-

2000 cm-1 and indicates very little or no change between non-silicified and silicified

LPEI/PVP fibers. After calcination, all such organic peaks disappear, confirming the

226

removal of organic content. The three main absorbance peaks corresponding to silica remain, as indicated by the vertical reference lines in the figures, further confirming the presence of silica. There remains a very weak peak at 950 cm-1 and 3400 cm-1 indicating a slight presence of uncondensed Si-O-H, in spite of calcination at 600 °C for 1 h.

However, the latter peak is very weak and is negligible compared to the other peaks in calcined fibers indicating that most of the material in the calcined fibers is silica.

7.5 Results: Silica formation in PEI foams

PEI foams were obtained by freeze drying of aqueous linear PEI crystalline hydrogels at various PEI concentrations as described in experimental section. The density of the PEI foams obtained after freeze drying (without silicification) depends on the initial PEI concentration of the hydrogels (Figure 7.8). Importantly, freeze drying did not lead to significant shrinkage of the linear PEI gel precursors (ca. 5 – 10 % in diameter and height) and the foam structure was retained (see Figure 7.8 insets). Silicification of the foam was then carried out for 10 minutes by vacuum infiltration of TMOS. The density of the foams increased upon silicification (Figure 7.8). The foams obtained from

3 wt% or lower PEI concentration in the hydrogel did not have structural integrity and further the silicified foams collapsed during the drying process. Thus, the density of such foam is not reported. On the other hand, the foams made from 10 wt % PEI concentration led to the silica formation only on the periphery of the foam (see inset

Figure 7.8). This is due to due to higher density and simultaneous silica formation during the TMOS infiltration that presents ‘wicking’ of the TMOS further inside the foams.

Thus, the TMOS penetration into the foam made from 10 wt% PEI was limited by

227

simultaneous silica formation during the infiltration process. Such reaction induced limitation leads to silica formation only on the periphery of the dense (10 wt% PEI) foam.

There is an upper limit of aqueous linear PEI concentration in hydrogel (10 wt% in this case) where the foams made from freeze drying of linear PEI hydrogels are too dense for the TMOS to penetrate while simultaneously forming silica. On other side, foams made from aqueous linear PEI of 3 wt% or lower concentrations were very fragile and weak and required very careful handling for further silicification. Owing to these limitations, linear PEI made from freeze drying at intermediate PEI concentration of 5 wt % have a nearly uniform silica formation within the foam. The density of such silica/polymer composite foam is 0.34 g/cm3. Such a low density is comparable to the polymer cross- linked silica aerogels that is synthesized from TMOS with base catalyst (NH4OH) that is aged for 48 h and then cross-linked with hexamethylene diisocyanate.225,226

The inorganic content of the foams decreased with increasing PEI concentration

(and hence, density), except for the case of the silicified foams obtained from 10 wt%

PEI concentration. Figure 7.9 shows the TGA analysis curve of the silicified PEI foams obtained after TMOS infiltration for 10 min and subsequent drying. The higher inorganic content of the 10 wt% sample is due to the non-uniformity across the cross-section of the foam. The TGA sample was taken from the periphery where comparatively silica formation was achieved due to the reaction induced diffusion as discussed earlier.

Nevertheless, with the exception of 10 wt% PEI, the decrease in inorganic content with increasing PEI incorporation signifies that PEI is closely integrated in to the composite structure. Also, as evident from Figure 7.9, ca. 7- 18 % weight loss is observed at temperatures below 120 °C, roughly indicating the level of water absorption by the

228

silicified foams. It was noted in the previous section that water is required for the silica formation by PEI from pure TMOS. Thus, freeze drying does not result into completely water-free foams. Water binds to the crystalline PEI to form dihydrate, sesquihydrate or hemihydrate crystalline structure of linear PEI.227 Such crystalline structure involving water molecules is shown to be stable even at high temperatures.228-230 It is believed that the internal water molecules present in the crystalline PEI take part in the silica formation.

The effect of water content and type of crystalline PEI structure on the formation of silica is beyond the scope of this chapter but does warrant future attention.

The morphology of the PEI foams before and after silicification and subsequent calcination at 500 °C for 1 h was analyzed by SEM. Figure 7.10 (a,b) shows the morphology of the linear PEI foams after freeze drying and before silicification. The structure of such foam is highly porous as evident from the SEM images. The images confirm that PEI in the foam has a co-continuous structure that is made of the fibrils with

< 1 μm diameter. The structure of linear PEI is composted of organized crystalline PEI fibrils that joins together to form a larger fibril (< 1 μm diameter) as seen in Figure 7.10

(b) and also reported in previous study.139

The silicification of the PEI foams by TMOS infiltration preserved the underlying linear PEI structure (Figure 7.10 (c,d)) with little formation of secondary silica in the voids (or pores) of the foams. Thus, the silica formation occurs mainly on the fibrous linear PEI in the foam as evident from the presence of fibrillar ‘struts’ similar to that observed in non-silicified foam in Figure 7.10 (a,b). The density of such a composite foam is 0.34 g/cm3 similar to the hybrid silica/polymer aerogels reported in the

229

literature.231 However, the morphology of such silicified foams is different than the spherical particulate morphology observed for the polymer/silica hybrid aerogels.226,231

The silica formation occurs mainly within the linear PEI fibrillar structure in the foam in our case. Furthermore, PEI is closely integrated with silica and silica formation occurs across the whole fibrillar struts similar to the PEI nanofiber case as confirmed by calcination of the silicified foam at 500 °C for 1h. Figure 7.10 (e,f) shows the SEM images of the silica aerogels after calcination. The fibrillar structure of linear PEI is replicated in the morphology of the calcined samples, as evident by the struts of replicated silica fibers in SEM image of Figure 7.10e. Hence, the silica formation occurs mostly on the linear PEI fibrillar structures. The inorganic content of the calcined sample was analyzed by EDX analysis. The presence of silicon and oxygen peaks in the EDX spectra (data not shown) further confirmed that the inorganic content in the calcined foams is silica.

7.6 Discussion

The silica synthesis from alkoxysilanes like TMOS proceeds via two step reaction

– (i) hydrolysis of the alkoxy (--Si-OCH3) functional group to form silanol (Si-OH), and

(ii) condensation of such silanol group either with other alkoxy or silanol group to form siloxane (--Si-O-Si--) group, liberating alcohol or water, respectively.107 The hydrolysis and the condensation to form silica usually require long time by conventional acid/base catalysis. Unlike these reactions, silica formed by polyamines that were isolated from silaffins were extremely rapid.118 Similar to the catalytic ability of polyamines found in silaffins to form silica, silica formation within minutes on electrospun PEI blends or

230

foams was observed in the present study. The SEM-EDX analysis (Fig. 7.4) and FTIR

(Fig. 7.7) confirmed the presence of silica in the nanofibers. Such rapid silica formation in nanofibers occurs due to the unique catalytic ability of PEI to catalyze hydrolysis and condensation reactions of TMOS.

Several mechanisms have been proposed for polyamine-assisted hydrolysis and condensation of the silica from precursor molecules. Sahai and Delak,216,217 while studying a model compound trimethylethoxysilane (TMES) by 29Si NMR at pH 5.0, have postulated a nucleophile-catalyzed mechanism of hydrolysis by polyamines where the conjugated base of the amine attacks the silicon atom of the organosilicate to form a penta-coordinated intermediate. They found that the hydrolysis rates are orders of magnitude higher than the condensation rates. Besides the hydrolysis reaction, PEI assists in the rapid condensation of the hydrolyzed precursor to form siloxane functional groups. According to the proposed mechanism for oligo N-methyl propylamine-assisted condensation,122,218 polyamine chains have the presence of alternating protonated and non-protonated amine groups that forms hydrogen bonds with two molecules of precursor acid (Si-OH) per two repeat unit of amine in the catalyst. This facilitates the

Si-O bond formation by stabilizing the transition state. The hydrogen of each amine group of the polyamines hydrogen bonds with the oxygen of forming silica, integrating the polyamines within the resulting composite material.

In silaffins, where the catalytic domains consist of oligo-N-methyl propylamine,

118 the silica formed has a relative composition of 1.25:1 SiO2/polyamine based on weight; i.e., 55 wt % of inorganic content. In our studies with PEI, the relative mass composition

231

is expected to be 2:1 SiO2/PEI, assuming that the silica formation from PEI occurs by the same mechanism mentioned above for N-methyl propylamine, due to the difference in repeat unit molar mass (43 g/mol for ethyleneimine versus 71 g/mol for N-methyl propylamine). Accordingly, the silica/PEI composition should be ~ 66 % inorganic.

Likewise, in a 50/50 (w/w) blend of PEI/PVP, we expected around 33% inorganic content. Indeed, this simple argument agrees reasonably well with TGA results (Figure 5) where the maximum inorganic content of 41 % for the silicified PEI/PVP nanofibers exposed to 80 % relative humidity prior to silicification. The additional inorganic content observed (7%) might be due to secondary silica formation due to sol-gel synthesis, as described in detail in next paragraph. The lower inorganic content of the PEI/PVP fibers exposed at lower than 80 % relative humidity is clearly due to insufficient water to hydrolyze the TMOS for the silica formation. Thus, water plays an important role in the silica formation by PEI, as was also confirmed by negligible silica content of silicified fibers (Fig. 7.5) pre-treated at anhydrous conditions (<20 % relative humidity).

After the rapid silica formation by PEI in electrospun nanofibers or PEI foams, there exists a possibility of secondary silica formation due to condensation of alkoxysilanes or silanols with the surface silanol groups. However, such an uncatalyzed process would be slow under ambient conditions and can be considered negligible for the

10 min allowed for immersion of the nanofibers in TMOS as was verified by TGA study for the fibers immersed for extended period of time. For the PEI/PVP fibers immersed in

TMOS for 45 min, the inorganic content measured by TGA at 900 °C was 46.3 wt%, confirming the slow silica formation by the secondary sol-gel process (data not shown).

In the case of the PEI/PVP fibers immersed in TMOS for only 1 min, the inorganic

232

content by TGA was 31 wt%., which is slightly lower than the inorganic content of 41 wt% measured after 10 min TMOS immersion of nanofibers (Fig. 7.5), but close to the predicted by PEI catalysis (33 wt%) as was discussed before. Thus, silica forms almost instantaneously in the nanofibers/foams during the first few minutes of TMOS immersion, after which the non-catalyzed condensation of the TMOS or silicic acid proceeds at a lower rate.

7.7 Conclusions

The rapid formation of hybrid (polymer/silica) scaffolds was achieved by silicification of PEI/PVP electrospun nanofibers or PEI foams by immersion in TMOS for 10 minutes. PEI present in the nanofibers catalyzed near-instantaneous formation of silica - within minutes – resulting in a unique composite material. PVP fibers alone did not form silica when exposed to TMOS under similar conditions, confirming that the silica formation is induced by the PEI in the electrospun nanofibers of PEI/PVP blends.

Silica formation was confirmed and characterized by FTIR, SEM-EDX, and TGA analyses. Compositionally, FTIR analysis indicated the presence of siloxane and silanol functional groups consistent with the formation of silica from TMOS while, morphologically, the fiber diameters showed a slight increase after silicification.

Meanwhile, TGA analysis revealed that water is required for silica formation. In particular, relatively dry nanofibers did not afford silica formation, while fibers pre- treated at high relative humidity (80%) did. Furthermore, the highest inorganic content in the nanofibers was about 41 % for the silicified fibers pre-treated at 80% relative humidity. Surprisingly, the inorganic content of the fibers immersed for just 1 minute in

233

TMOS yielded ~ 31 % inorganic content. This finding was in agreement with the proposed mechanisms of the silica formation by polyamines where the formed silica closely interacts with the polyamines and results in 2:1 silica/polyamine by weight.

Calcination of the silicified fiber mats led to porous ceramic nanofibers consisting of porous nano-structured silica particles. Similar to the silica formation in the PEI/PVP nanofibers, silica formation in the linear PEI foams obtained by freeze drying of aqueous

PEI hydrogels was achieved. The density of the silicified foams was comparable to the silica aerogels. Furthermore, the silica formation on to such forms by TMOS infiltration was limited by high density of foams at PEI concentration of 10 wt% or above, and by the structural integrity of foams at PEI concentration of 3 wt % or lower. The optimum

PEI concentration to obtain the PEI scaffolds for uniform silica formation was ~5 wt%.

SEM analysis of the polymer/silica foams before and after silicification, and calcination revealed that silica formation in foams is primarily on the fibrils of linear PEI. To conclude, we have demonstrated that the simple biomimetic route for silica formation from PEI can be successfully used to obtain organic-inorganic hybrid composite materials in the form of nanofibers and foams.

234

N n n N O

(a) (b)

N H n

(c)

Scheme 7.1: Chemical structures of the polymers involved in hybrid nanofiber formation: (a) poly(2-ethyl-2-oxazoline), (b) poly(vinyl pyrrolidone), and (c) poly(ethylene imine).

235

(a) a a c

a c b

(solvent) Intensity

4 3 2 1 0

δ (ppm)

(b)

d b c

4 3 2 1 0 δ (ppm)

Figure 7.1: 1H NMR of (a) poly(2-ethyl-2-oxazoline) and (b) linear PEI obtained from hydrolysis of (a).

236

(a) (b)

(c)

237

(a) (b)

(c)

Figure 7.3: SEM images of the silicified electrospun fibers of PEI:PVP (50:50)

corresponding to the electrospun fibers of (a) Figure7.2a and (b) Figure 7.2b and (c)

Figure 7.2c.

238

(c)

(b)

Intensity

(a)

01234 Binding Energy (keV)

Figure 7.4: Energy Dispersive X-ray (EDX) spectra of (a) PEI:PVP(50:50)

electrospun fibers, (b) after silicification by TMOS and (c) control sample of PVP

fibers after TMOS exposure for 10 minutes. The vertical dotted line indicates the reference peak of the corresponding elements shown in bracket.

239

100

(iv) 40

20 (iii) Weight Remaining (%) Remaining Weight (ii) 0 (i) 0 100 200 300 400 500 600 700 800 900

Temperature (oC) Figure 7.5: TGA analysis of the silicified and non-silicified electrospun fibers of

50:50 blends of linear PEI and PVP (i) PEI:PVP fibers, (ii) silicified electrospun

PVP fibers (control), (iii) silicified PEI:PVP fibers after exposure at ca.

40%humidity and (iv) silicified PEI:PVP fibers after exposure at ca. 80% humidity.

240

(a)

(b)

Figure 7.6: High resolution SEM images of the silicified fibers after calcination at

600 oC for 1 hour (a) top view (b) cross sectional view.

241

(b) Silicified LPEI-PVP fibers

Absorbance (a.u.)

(a) LPEI-PVP fibers

4000 3000 2000 1000

Wavenumber (cm-1)

Figure 7.7: FTIR of the electrospun PEI-PVP fibers (a) before silicification, (b) after silicification and (c) after calcination at 600 °C for 1 h.

242

0.4

0.3 ) 3

0.2 Density (g/cm 0.1

0.0 024681012 LPEI concentration (wt%)

243

100

40 Weight Remaining (%) 20

0 0 200 400 600 800 o Temperature ( C)

Figure 7.9: TGA analysis of the composite foams obtained by the silicification of the freeze dried linear aqueous PEI solution of (a) 2 wt % (continuous line), (b) 3 wt%

(long dash line), (c) 5 wt % (medium dash line) and (d) 10 wt% (at periphery) (dot- dash line).

244

(a) (b)

(e) (f)

Figure 7.10: SEM images of linear PEI foams obtained by freeze drying of aqueous linear PEI solutions of 5 wt% (a,b) as such (c,d) after silicification and (e,f) after calcination at 500 °C for 1 h.

245

CHAPTER 8

8 Conclusions and Future Outlook

8.1 Polyelectrolyte Multilayers Simulation (part I)

8.1.1 Conclusions

Molecular dynamics (MD) simulations of the LbL assembly in a similar fashion to the 'dipping' LbL assembly in experiments were performed to investigate the underlying mechanisms governing the multilayer formation. Chapter 2 and 3 describes in detail the MD simulations of multilayers, evaluating the effect of degree of polymerization and charge fraction of polyelectrolytes, and relative strength of short- range and electrostatic interactions among polyelectrolytes, on multilayer growth, structure and stability. Such simulations revealed important information about the multilayer formation, growth mechanisms, internal structure, interdiffusion and stability and the universality required for the successful multilayer growth. The important conclusions of the MD simulations are briefly summarized below.

The multilayer formation by layer-by-layer assembly occurs by formation of the polyelectrolyte complexes (or symplexes) of the adsorbed polyelectrolytes with incoming layers of the oppositely charge polyelectrolytes. The formation of polyelectrolyte complexes, results into conformation rearrangement of the adsorbed polyelectrolyte in the layer resulting into the formation of islands and holes during the deposition step. The islands and holes are then covered during the deposition of the next layer of similarly

246

charged polyelectrolytes. Thus, the formation of a single compact layer in the multilayers requires two deposition steps of the similar charge polyelectrolytes. In spite of the formation of islands and holes in the surface, the surface coverage and thickness increases linearly with the deposition step in agreement with the experimental observations.

The internal structure of the multilayers supports the three zone model of the multilayer film deduced from the various experimental studies7 and summarized in

Chapter 1. Furthermore, it was observed that within the multilayers, positively charged monomers are surrounded by negatively charged monomers forming ion pairs. Such ion- pair formation is the key to the successful multilayer growth and dictates the stability and dynamics of the polyelectrolyte multilayers. By forming the ion pairs, the oppositely charged polyelectrolyte chains in the multilayers form ‘scrambled-egg’ type complexes of intertwined chains. In spite of the formation of such complexes and intermixing of polyelectrolytes, to our surprise, the density difference between the positively and negatively charged polyelectrolytes had almost perfect periodic oscillations. Such oscillations were not kinetically trapped state of the polyelectrolyte due to the deposition of polyelectrolytes in LbL fashion. The period and the amplitude of such density oscillations is a strong function of the charge fraction of polyelectrolytes, short-range

Lennard-Jones (LJ) interactions and electrostatic interactions among the polyelectrolytes while it depends weakly on the degree of polymerization of the polyelectrolytes used in the multilayer assembly.

The weakly charged chains shows higher growth of surface coverage and

247

thickness with the deposition step compared to the multilayers made from strongly charge polyelectrolytes, for all the studied system. The steady state multilayer growth depends strongly on the strength of electrostatics and short range LJ interactions. For the simulation systems with weak LJ interactions (close to θ-solvent conditions for polymer backbone), only multilayers assembled from polyelectrolytes that have higher Np show the successful multilayer growth. For shorter chains having weak LJ interactions, the chain desorption is frequent due to lower chain cohesive energy, and ultimately results into unstable multilayer growth. By increasing the LJ interactions (corresponding to poor solvent conditions of polymer backbone), the chain desorption is lowered and the multilayer stability is increased. Increasing the LJ interactions also increases the layer stratification, average polymer density inside the multilayer and the growth rate. The theoretical model of the multilayer assembly that describes the effect of charge fraction,

Np, electrostatic and short-range interactions on the average density, oscillations, multilayer growth and stability (desorption) is presented in Chapter 4. Finally, the simulations confirm the hypothesis that surface overcharging is crucial for the stable film growth. This overcharging is universal among the systems that show successful multilayer growth as evident from the constant ratio of the layer overcharging to the number of charge adsorbed during the deposition step.

8.1.2 Future Recommendations

The molecular simulations presented in this thesis are limited to the case of flexible polyelectrolytes in a solvent modeled as a continuum. Modeling the solvent as a continuum eliminates an important effect of the size of the solvent molecules on the

248

packing of polymer chains at the substrate as well as variations in the solution dielectric constant within growing polymeric film. Such effect on the multilayer formation and growth can be considered for the future studies. Furthermore, the dynamics of the polyelectrolyte chains within the multilayers can be deduced from the existing data on the multilayer formation from the flexible polyelectrolytes. Preliminary studies was made to calculate the dynamics of polyelectrolytes adsorb during the 5th deposition step during the

10th deposition step simulation run in the xy-plane and z-direction (normal to surface).

The analysis indicates that the diffusion of polyelectrolytes in multilayers is significantly lowered in the z-direction compared to the x and y direction. The code to analyze the simulation data has to be extended for the dynamical analysis of polyelectrolyte chains adsorbed during 1st deposition step from the entire simulation run of 1st deposition step to

10th deposition step. This will allow enough statistics to interprete the data and make meaningful conclusions on the polyelectrolyte dynamics within the multilayers.

It was observed that the counterion forms a double layer on the top of the multilayer surface (Chapter 3,4). The spatial distribution of the counterions could be studied in order to gauge the spatial distribution of charges on the surfaces of the multilayers. This will be particularly important since it was found in the present thesis that the growth in the zone III of the multilayers proceeds via the formation and holes and islands of polyelectrolytes. This was indirectly observed by experiments of the mineralization processes on multilayers (Chapter 5). Such mineral formation on multilayer, regulated by electrostatic attractions of the counterions, was sparse on the surface, thus providing indirect evidence of the heterogeneous charge distribution on the multilayer surface. The study of the spatial charge distribution on the surface is very

249

challenging experimentally, and hence simulations can help to understand such phenomenon. Furthermore, the mineralization similar to the hydroxyapatite formation on the multilayer surfaces can also be studied via simulations. The simulation of already formed multilayers in the presence of divalent ions in the system should allow to study the distribution of such divalent counterions on the surface and can be thus related to the mineralization that was studied in Chapter 5..

Another effect that was not considered in our simulations is the effect of chemical structure of polyelectrolyte chains such as chain rigidity and effect of flow during the adsorption of polyelectrolytes. The simulation studies of the flow effect on polyelectrolyte adsorption will further allow testing the assumptions made while predicting the growth and formation of multilayers. Molecular simulations of the multilayer assembly under the flow will help to test the important assumptions made to model multilayer growth under the effect of flow (see Chapter 1). The experimental results for multilayers under spin-coating flow, termed as polyelectrolyte spin assembly

(PSA) is described in the next section. I will briefly discuss the implementation of the chain rigidity parameter and flow conditions, in the molecular dynamics simulation model to deduce their respective effects on the multilayer assembly.

8.1.2.1 Influence of the Chain Rigidity

Very limited studies in the literature considers the of the chain rigidity on the growth and the internal structure of the multilayers.232,233 Gong et al232 studied the importance of the backbone rigidity employing DNA as polyanion on the multilayer growth and found that PAH/DNA multilayers showed non-linear growth compared to the

250

linear growth observed for regular PSS/PAH multilayers. Such behavior was ascribed to long range ordering possible for the rigid polyanion (DNA). Hong et al233 constructed the LbL assembly with azobeneze ionenes type rigid backbone with varying spacer in order to obtain internally ordered multilayers though the lack of Bragg peaks in X-ray diffraction of the multilayers did not confirm such ordering. Thus, there is a need to better understand the effect of polymer backbone rigidity on the multilayer assembly.

The rigidity of the polymer backbone can be easily incorporated in to the simulation model of the multilayers described in Chapter 2. A (parabolic) potential between the neighboring monomers can be incorporated to vary the rigidity of the chain backbone. Such potential is given by,234

1 U (θ ) = k ()cosθ − cosθ 2 (8.1) 2 b 0

where θ is the angle formed by two adjacent bonds in the bead-spring model, θ0 is the equilibrium bond angle and kb is the force constant. The relationship between the force constant (i.e. chain rigidity parameter) and the radius of gyration of a bead-spring model of Np=40 is given in Figure 8.1a. Alternatively, the chain rigidity can be introduced by additional interaction potential between the second nearest neighbor segments within the same chains,235

12 6 ⎡⎛ s ⎞ ⎛ s ⎞ ⎤ U (r) = 4ε ⎢⎜ 2 ⎟ − ⎜ 2 ⎟ ⎥ (8.2) ⎣⎢⎝ r ⎠ ⎝ r ⎠ ⎦⎥

where ε is the same as the Lennard-Jones interaction parameter for the LJ- potential. The chain rigidity can be controlled by parameter s2 (see Figure 8.1b).

251

However, the potential described in Eq. 8.2 might cause slight difference in the average bond length, since it is LJ-type potential. However, the effect of such potential in the bond length is predicted to be less than 10 % of the equilibrium bond length value.

8.1.2.2 Influence of the Shear Rate (Flow)

Adsorption of polyelectrolytes on the surfaces in the presence of flow has been studied theoretically, experimentally and by simulations.236-242 However, the formation of multilayers under the effect of flow has studied experimentally Chapter 4 in the present thesis, has not been studied by simulations of the multilayer assembly. Such simulations will allow testing the assumptions made by the Flory-type scaling model for the multilayer construction under the effect of flow (see Section 1.2) and will provide an important information on the internal layer structure, and properties like density distribution and interdiffusion in the multilayers constructed under the flow conditions.

Various models exists has been described previously in the literature that incorporates flow to study single polyelectrolyte adsorption under the flow conditions.

Panwar et al241 has studied the Brownian dynamics of the polyelectrolyte under the flow and has simulated the unidirectional flow by imposing a velocity as a function of the z- direction position in the integration of the equation of the motion (Eqn. 2.4). Hence, depending on the position of the monomer from the surface, the chain experiences an additional force due to the shear flow. Such

FrvvD =−ζ ⎡⎤ − + ' iixi⎣⎦()

vzxi= γ (8.3)

vi '0=

252

D Where Fi is the net deterministic force in the simulation model, that is incorporated in the equation of motion (Eqn. 2.4), and vx is the velocity imposed in the x- direction which is the function of the z-position of the monomer, and vi′ is the velocity due to hydrodynamic drag. Normally, vi′ is zero for the case where the hydrodynamics interactions are not considered. The importance of flow in the DNA adsorption on the surface has also been studied by Larsen’s group237 and Smith et al242. Such studies provide an alternative simulation models to incorporate flow in the simulation studies.

8.2 Polyelectrolyte Spin Assembly (part II)

8.2.1 Conclusions

In this section (Chapter 4), the combined effect of spin speed and salt concentration on the growth, morphology, thickness and roughness of the polyelectrolyte multilayers made by sequential spin coating of the polyelectrolyte solutions, termed polyelectrolyte spin assembly (PSA) was studied. The main research goal of this study was to deduce the shear-dominated regime in the PSA process, where the multilayer growth-rate decreases with increasing salt concentration at high shear rates as predicted by the Flory-type theory of multilayer formation under flow. In this section, It was shown for the first time that for the PSA process, with increasing ionic strength actually results in a decrease in the growth-rate of the multilayers at higher spin-speed and/or higher radial distance (both corresponding to increasing shear rate). This phenomenon is in agreement with the Flory-type theory of polyelectrolyte adsorption under shear that was described in detail in Chapter 1. The growth of the multilayered coatings shows a

253

non-monotonic dependence on ionic strength, first increasing and then decreasing with increasing solution ionic strength. Such behavior is a manifestation of two competing mechanisms for the multilayer assembly process, electrostatic interactions dominating film growth at low ionic strength and shear flow dominating at high ionic strength as predicted by the Flory-type theory. It was shown that the associated scaling equation fits our experimental data for the multilayer growth of surface coverage and thickness of the multilayer at different spin speeds reasonably well.

Additionally, it was deduced that the multilayer coatings made by PSA has a characteristic radial dependence of thickness and surface coverage that depends on the salt concentration and the spin-speed. While the multilayered coatings did not show a radial dependence for the multilayers made at low salt concentrations below 0.1 M, independent of spin speed, a strong radial dependence was observed at higher ionic strengths, with coverage decreasing radially. This radial dependence of polymer surface coverage, in good agreement with our model for PSA, is concluded to be due to the linear increase of the local shear rate with distance (r) from the disk center. However, further increase of the spin rate to 6000 rpm leads to planarization of the film and almost uniform polymer surface coverage over the entire disk and it is argued that the influence of shear is bounded, perhaps by full chain extension. Topographically, the multilayered coatings are smooth and featureless at low salt concentrations with dimensions of the characteristic features evolving as the ionic strength of the polyelectrolyte solution is increased. The dimensions of these features are less pronounced for the multilayers formed by the spin-assembly than quiescent adsorption at the same salt concentrations.

The feature size of the multilayered coatings at high ionic strength decreases with

254

increasing the spin speed, further confirming the effect of the shear rate on the chain conformations while adsorption on the surfaces. Thus, the variations in salt concentration and the spin-rate are two interacting parameters that allow control over the growth rate and film thickness during multilayer assembly by spin-coating method and the effect of both the parameters can be successfully predicted by the Flory-like theory for polyelectrolyte adsorption under flow.

8.2.2 Future Recommendations

One of the important implications of the polyelectrolyte spin assembly process is the radial dependence of the surface coverage and the thickness of the multilayers. The notion of the polyelectrolyte adsorption under the effect of shear flow is interesting and should provide important clue on the effect of shear flow on the polyelectrolyte conformation during the adsorption. Planarization of the multilayer coatings was observed in the present study at shear rate dominating regime at higher salt concentration and spin rate of 6000 rpm. Such planarization has been explained by finite polyelectrolyte chain extension limit under the effect of shear flow. Further studies needs to be done to test this hypothesis. The present study on PSA was limited to the range of spin speed (3000-6000 rpm) and radial distance of 10 cm. PSA growth rate studies at higher radial distance (>10 cm) (see Eqn. 4.3) and lower spin-speed should allow the testing the hypothesis. Such assumption can further be incorporated in the Flory-like theory presented in Chapter 1 to predict the critical radial distance at given spin speed where the planarization (corresponding to full chain extension) of the coatings will occur.

The critical shear rates along the surfaces required to achieve the finite chain extension

255

could be compared to the models in the literature where such regime is predicted in the case of extensional flow like DNA adsorption under flow.237

8.3 Biomineralization (part III)

8.3.1 Conclusions

The main research goal of part 3 in the thesis was to demonstrate the formation of minerals like hydroxyapatite and silica from simple polypeptides or synthetic polyelectrolytes, inspired from the proteins found in nature, in solution and surfaces.

Towards the formation of hydroxyapatite (Chapter 5), it was demonstrated that the simple polypeptide, poly(glutamic acid) (PGA), greatly affects the hydroxyapatite crystal growth rate and morphology when present in solution. The constant composition method (CCM) to study the growth of the hydroxyapatite seed particles revealed that the PGA lowers the rate of growth of hydroxyapatite seed particles, presumably due to binding of PGA to specific crystal faces of the HA. In contrast, the PGA nucleated the HA formation when present on the outermost layer of in the multilayers of PEI-(PGA/PAH)5-PGA surfaces.

This is in agreement with the studies by Tsortas et al that found that the PGA can nucleate hydroxyapatite crystal growth when present on the surfaces of the germanium crystal. Thus, PGA still retains it ability to nucleate the hydroxyapatite (or OCP) formation on the surfaces when localized in the multilayers with other polycation, PAH.

Similar to the HA formation by PGA localized in the multilayers, silica is formed when

PLL is localized in the multilayers. The silica formation and morphology depends on the whether PLL is present only on the outermost layers or localized on every alternating layer in the multilayer. The latter arrangement of PLL results into increased

256

particle size and amount of silica formation.

While studying the simple polypeptide, poly-l-lysine for the silica formation in solution in solution, it was observed that aqueous solution of poly(ethylene imine)

(branched), rapidly induces silica formation when added to the TMOS. Thus, to further understand such rapid silica formation from alkoxysilanes, such as TMOS, silica formation induced by PEI was studied in solution and surfaces (Chapter 6). Silica formation by PEI is argued to be a combined effect of the ability of PEI to catalyze siloxane bond formation (or stabilize the intermediate silicic acid formed by hydrolysis of TMOS) and to flocculate the silica ‘sol’ formed during the subsequent condensation reaction. Toward understanding this unique ability of PEI to instantaneously form silica, the aqueous PEI and TMOS concentration (in ethanol) was varied. The inorganic fraction and the TMOS conversion (hence, silica yield) in the composites formed by addition of aqueous PEI to TMOS, varied depending on the concentration of PEI and

TMOS. The inorganic fraction was maximum for the lowest concentration of the PEI (1 wt %) used in the study, while the overall conversion of TMOS was highest at an intermediate concentration of 3 wt% PEI and increased with decreasing the TMOS fraction. The increase of conversion with decreasing the TMOS fraction is effect of dilution by ethanol, a mutual solvent to PEI and TMOS. The silica precipitated from PEI had spherical morphology and that increasing the PEI concentration increases the coagulation of the particles, further increasing the diameter. Coagulation of the particles was evident from the SEM images of the silica formed from the addition of PEI to TMOS.

Silica formation was achieved on to the PEI localized on the surfaces on polyelectrolyte multilayers. The silica formed from the PEI localized on the surface had spherical

257

morphology, similar to the silica morphology found in solutions. Furthermore, single

PEI layer gave higher density of the particles compared to the one obtained when PEI was localized onto the multilayers.

Linear PEI has an ability to gelate water and form fibrous crystalline hydrogels at room temperature. We thought that such an ability, combined with the PEI’s ability to form silica rapidly and directly from alkoxysilanes, such as TMOS, would open up interesting applications in the field of the hybrid (organic-inorganic) composite materials.

Hence, the silica formation in the linear PEI scaffolds made by either electrospinning

PEI/PVP blend solution (nanofiber webs) or by freeze drying aqueous linear PEI hydrogels (PEI foams) was studied. The rapid formation of hybrid (polymer/silica) scaffolds was achieved by silicification of PEI/PVP electrospun nanofibers or PEI foams by immersion in TMOS for 10 minutes. PEI present in the nanofibers catalyzed near- instantaneous formation of silica - within minutes – resulting in a unique composite material. PVP nanofiber webs alone did not form silica when exposed to TMOS under similar conditions, confirming that the silica formation is induced by the PEI in the electrospun nanofibers of PEI/PVP blends. Further analysis of the nanofiber webs exposed to varying humidity levels revealed that water is required for silica formation, since relatively dry nanofibers did not afford silica formation compared to the fibers pretreated at high relative humidity. The inorganic content of the fibers immersed for just 1 minute in TMOS yielded ~ 31 % inorganic content. This finding was in agreement with the proposed mechanisms of the silica formation by polyamines where the formed silica closely interacts with the polyamines and results in 2:1 silica/polyamine by weight.

Calcination of the silicified fiber mats led to porous ceramic nanofibers consisting of

258

porous nano-structured silica particles. Similar to the silica formation in the PEI/PVP nanofibers, silica formation in the linear PEI foams obtained by freeze drying of aqueous

~5 wt%. The SEM analysis of the polymer/silica foams before and after silicification, and calcination revealed that silica formation in foams is primarily on to the fibrils of linear PEI. This simple biomimetic route to rapid synthesis of hybrid composite nanofibers/foams could have widespread use, including diverse applications of catalysis, tissue engineering, and structural materials including silica aerogels.

8.3.2 Future Recommendations

8.3.2.1 Polyelectrolyte multilayers mimicking nacre structure

In 1970’s, Reiss et al142 and Iler1 have observed an important phenomenon of the charge reversal of the colloidal particle due to excess polyelectrolytes adsorption from solution onto the surfaces of colloidal particles. Such charge reversal in the colloidal particle has being used to construct multilayers on the colloidal particles.53 Our studies on the hydroxyapatite growth by CCM reveals that PGA can selectively adsorb, presumably as monolayer, from solution onto HA seed particles lowering the HA growth rate. Such adsorption of PGA has been extensively studied in the literature.192-195 If we assume that the adsorption of the excess PGA on the HA particles surfaces can lead to

259

charge reversal of the HA particles, we can construct a multilayer assembly on the planar surface utilizing such negatively (over)charged (PGA+HA) particles with an alternating polycation like PAH or PEI. Such overcharging of the HA particles could be measured by zeta-potential of HA particles. The alternating layers of inorganic particles(PGA+HA) and polyelectrolytes (PAH) can thus be constructed by either dipping or spin-coating techniques. The multilayers of the inorganic clay platelets and polyelectrolytes have been constructed before by Kotov’s group and found to resemble a structure of brick- mortar structure of nacre.243,244 The brick-mortar structure similar to the nacre structure has further shown to result into tough multi-component films. The advantages of using the overcharged HA a particle over the clay platelet in the multilayer assembly is the stability of HA against the water adsorption compared to the clay platelets used in the previous study. Thus, such simple construction of multilayers from charged particles and polyelectrolytes can bring us one step closer towards mimicking tough composite structures like nacre found in the Nature.

8.3.2.2 PEI hydrogel silicification

As an alternative to the silica formation on to linear PEI foams obtained by the freeze drying of the aqueous linear hydrogels, the following method to obtain polymer/silica composite material is suggested. This method was developed in parallel to the foam silicification by TMOS infiltration that was reported in Chapter 7.

The detail method for PEI gel silicification is described in Figure 8.2. In brief, the aqueous linear PEI solution at various concentration (1, 3, 5 and 10 wt%) is heated above 60 °C to melt the crystalline PEI (Tm ~ 46 °C ) and then the clear solution is

260

poured in the 10 ml syringe with capped bottom. The solution is then allowed to cool to room temperature to foam a hydrogel for at least 2 h. Linear PEI has ability to gelate the water to form fibrous crystalline hydrogel as reported previously.139 The hydrogel is then slowly infiltrated with equal volume of pure TMOS (Figure 8.2), by applying slight pressure from the top of the syringe if needed. After TMOS infiltration, silica formation was allowed for 10 minutes, after which the gel was washed with extensive (3 x) volume of first acetone and then DI water. Acetone dissolves TMOS but not PEI and thus the hydrogel structure was not disturbed. However, special caution was taken for hydrogels of low PEI concentration to not to remove wash-out the silicified hydrogels during washing. The silicified gels where then freeze dried to obtain the silica aerogels. The freezing of the sample was made at -70 °C in dry ice/acetone bath for 15 minutes.

One of the main challenges in the gel silicification process is step 3 in the Figure

8.2, namely TMOS infiltration by displacing the water from the hydrogels. TMOS and water are immiscible liquids and hence to ensure complete displacement of the water in the pockets of the fibrous crystalline structure of the hydrogels, TMOS should not form channels in the hydrogels but displacement of the water should occur through whole meniscus. This in turn depends on the pressure applied to obtain the flow through hydrogel during the displacement process. Also, there is simultaneous silica formation on to the fibrils of linear PEI which changes the nature of the hydrogels. Similarly, secondary silica formation in the bulk ‘pockets’ of the hydrogels rather than the surfaces of the fibrous crystalline PEI, would add to the pressure drop to the flow and is highly undesired.

261

Preliminarily studies of the gel silicification by TMOS were carried out in the hydrogels obtained from linear PEI at concentration range of 3 - 10 wt% PEI according to the procedure shown in Figure 8.2. Most of the composite materials (or aerogels) obtained after freeze drying of the silicified gel had defects present in their internal structure. The defects like formation of channels where the bulk TMOS passed through the hydrogels or uneven silica formation (mostly higher inorganic content at the periphery) resulted into irreproducibility among the samples. However, it was still possible to obtain a few samples that had uniform silica formation, similar to the samples obtained by the foam silicification (Chapter 7). The key to TMOS infiltration process

(step 3 in Figure 8.2) seems to be the control of the pressure applied to displace the water by TMOS. The constant pressure or constant flow conditions should allow overcoming the experimental challenge of the defects and should be considered for further studies.

8.3.2.3 Biomimetic Silica Patterning

Silica nanostructures patterning has recently created a lot of interest due to its applications in sensors and photonic materials.245 More recently, biomimetic patterning of silica from various functional molecules, both synthetic and natural proteins, localized spatially have been studied by various researchers. 246,247 Stone and co-workers245 were first to report formation of silica micropatterns (of nanoparticles) using holographic two- photon induced photopolymerization to generate patterns of R5 peptide units (from protein found in diatoms, see Chapter 1 for details). The deposition of silica onto polypeptide rich areas by polycondensation of silicic acid led to the holographic patterns for photonic applications. Following this, Coffman et al246 used patterning of poly-l-

262

lysine to generate micropatterns of silica, though the patterns had dimensions of ~25 μm compared to the few hundred nanometer level resolution of the conventional photolithography. Most of these studies, however, have demonstrated the formation of silica of the dimensions in microns (~ 10 -25 μm in diameter), well above the limit of 100 nm resolution achieved by conventional photolithographic techniques. Hence, based on the results of rapid silica formation obtained by the PEI reported in Chapter 6 and 7 combined with conventional photolithographic techniques, a procedure for nano- patterning of silica is proposed, and is described in detail below.

Figure 8.3 shows the proposed scheme for the silica formation that combines the chemistry to obtain linear PEI from the hydrolysis of poly(oxazolines) by UV-activated photoacid generator (PAG) followed by the rapid silica formation in the exposed area where PEI is synthesized by TMOS exposure of the substrate. Let us consider the step- by-step procedure to obtain such patterns. First, a solution of the poly(2-ethyl-2- oxazoline) (PxOz), precursor to PEI, mixed with a photoacid generator based on iodonium borate salt248,249 is spin-coated to desired thickness on a substrate (Step 1). The selection of iodonium borate salt PAGs is based on the predicted miscibility of ethanol solution of PxOz and high photo-efficiency to generate the required acid molecules that catalyzes the conversion of oxazoline (Oz) units of PxOz to imine units of PEI. The chemistry of the photoacid generation reaction is crucial to the success of the patterning.

First, the miscibility of the photoacid generators with PxOz has not been studied experimentally and needs to be verified. Such problem is anticipated but could be easily solved by trying wide-variety of the PAGs available commercially. Secondly, the chemistry of the acid-hydrolysis of PxOz to PEI by the photoacid generated by UV-

263

exposure is very challenging and has to be further confirmed by experiments.

After the spin-coating, the thin film is brought under UV light with a patterned mask as shown in step 2. Only the UV-exposed area of the thin film generates the required acid by the photo-activation of the PAGs.248 The generated acid catalyzes the

PEI formation according to reaction shown in the right column of Figure 8.3. After the desired conversion has reached (the UV-exposure time have to be optimized), the substrate is exposed to pure TMOS solution. The exposure leads to silica formation only where the PEI is present on the substrate, thus ‘developing’ patterns of the composites similar to the procedure in the photolithographic techniques. The PxOz rich surface does not form silica. Calcination of such thin films with patterns of composite materials leads to the formation of desired silica patterns.

264

(a)

(b)

Figure 8.1: (a) Relationship between the chain rigidity kb and the radius of gyration,

2 Rg. Squares and solid line indicate the Rg . Circles and dashed line indicate the ratio

2 2 Ree / Rg , in which Ree is the end to end distance of polymer chains. (adapted from

234 Toshiaki et al ) (b) The relation between the rigidity parameter s2 and the angle correlation of two successive bond vectors (adapted from Miura et al235).

265

(1) (2) (3) (4) (5)

Acetone, TMOS

LPEI LPEI TMOS Silicification Washing aqueous Hydrogel (Acetone,DI solution * Freezing at -70 oC in acetone/dry ice bath for 15 minutes water)

Scheme 8.1: Schematics showing the detail procedure to prepare the PEI/silica composite material by PEI hydrogel silicification by TMOS.

266

Spin-coated layer

(a) N n +

O poly(oxazoline) Photoacid generator (PAG) Substrate UV light exposure PAG + h ν (b) N n N Mask H - n O O

(-) O

TMOS exposure (c) Polymer/Silica Composite

Calcination (d) Patterned Silica

267

9 Bibliography

(1) Iler, R. K. Journal of Colloid and Interface Science 1966, 21, 569-594. (2) Decher, G. Science 1997, 277, 1232-1237. (3) Decher, G.; Hong, J. D. Berichte der Bunsen-Gesellschaft 1991, 95, 1430- 4. (4) Decher, G.; Hong, J. D. Makromolekulare Chemie, Macromolecular Symposia 1991, 46, 321-7. (5) Decher, G.; Hong, J. D. Ber. Bunsen-Ges. Phys. Chem. 1991, 95, 1430- 1434. (6) Decher, G. In The Polymeric Materials Encyclopedia: Synthesis, Properties and Applications; Slasmone, J. C., Ed.; CRC Press: Boca Raton, 1996. (7) Decher, G. In Multilayer Thin Films:Sequential Assembly of Nanocomposite Materials; Decher, G., Schlenoff, J. B., Eds.; Wiley-VCH: New York, 2003, p 1-46. (8) Hammond, P. T. Current Opinion in Colloid & Interface Science 1999, 6, 430-442. (9) Hammond, P. T. Advanced Materials 2004, 16, 1271-1293. (10) Blodgett, K. B. Journal of the American Chemical Society 1934, 56, 495. (11) Zasadzinski, J. A.; Viswanathan, R.; Madsen, L.; Garnaes, J.; Schwartz, D. K. Science 1994, 263, 1726-33. (12) Sukhorukov, G. B.; Mohwald, H.; Decher, G.; Lvov, Y. M. Thin Solid Films 1996, 220-223. (13) Rusling, J. F. In Protein Architecture: Interfacing Molecular Assemblies and Immobilization Biotechnolgoy; Marcel Dekker: New York, 2000. (14) Caruso, F.; Caruso, R. A.; Mohwald, H. Science 1998, 282, 1111-1113. (15) Hao, E.; Lian, T. Chemistry of Materials 2000, 12, 3392-3396. (16) Shimazaki, Y.; Mitsuishi, M.; Ito, S.; Yamamoto, M. Langmuir 1997, 13, 1385-1387. (17) Krass, H.; Papastavrou, G.; Kurth, D. G. Chemistry of Materials 2003, 15, 196-203. (18) Farhat, T. R.; Schlenoff, J. B. Electrochemical and Solid-State Letters 2002, 5, B13-B15. (19) Hattori, H. Advanced Materials 2001, 13, 51-54. (20) Graul, T. W.; Schlenoff, J. B. Analytical Chemistry 1999, 71, 4007-4013. (21) Ngankam, P. A.; Lavalle, P.; Voege, J. C.; Szyk, L.; Decher, G.; Schaaf,

268

P.; Cuisinier, F. J. G. J. Am. Chem. Soc. 2000, 122, 8998-9005. (22) Rudra, J. S.; Dave, K.; Haynie, D. T. Journal of Biomaterials Science, Polymer Edition 2006, 17, 1301-1315. (23) Muller, M.; Rieser, T.; Lunkwitz, K.; Meier-Haack, J. Macromolecular Rapid Communications 1999, 20, 607-611. (24) Sun, Y.; Zhang, X.; Sun, C.; Wang, B.; Shen, J. Macromolecular Chemistry and Physics 1996, 197, 147-153. (25) Joly, S.; Kane, R.; Radzilowski, L.; Wang, T.; Wu, A.; Cohen, R. E.; Thomas, E. L.; Rubner, M. F. Langmuir 2000, 16, 1354-1359. (26) Fou, A. C.; Onitsuka, O.; Ferreira, M.; Rubner, M. F. Journal of Applied physics 1996, 79, 7501-7509. (27) Dai, J.; Jensen, A. W.; Mohanty, D. K.; Erndt, J.; Bruening, M. L. Langmuir 2001, 17, 931-937. (28) Schwarz, S.; Eichhorn, K. J.; Wischerhoff, E.; Laschewsky, A. Colloids and Surfaces, A: Physicochemical and Engineering Aspects 1999, 159, 491-501. (29) Advincula, R. C.; Baba, A.; Kaneko, F. Polymeric Materials Science and Engineering 1999, 81, 95-96. (30) Caruso, F.; Furlong, D. N.; Ariga, K.; Ichinose, I.; Kunitake, T. Langmuir 1998, 14, 4559-4565. (31) Schmitt, J.; Gruenewald, T.; Decher, G.; Pershan, P. S.; Kjaer, K.; Loesche, M. Macromolecules 1993, 26, 7058-7063. (32) Lobo, R. F. M.; Pereira-Da-Silva, M. A.; Raposo, M.; Faria, R. M.; Oliveira, O. N., Jr. Nanotechnology 1999, 10, 389-393. (33) Schlenoff, J. B.; Ly, H.; Li, M. Journal of the American Chemical Society 1998, 120, 7626-7634. (34) Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153-8160. (35) Decher, G.; Schmitt, J. Colloid and Polymer Science 1992, 89, 160. (36) Lowack, K.; Helm, C. A. Macromolecules 1998, 31, 823-833. (37) Loesche, M.; Schmitt, J.; Decher, G.; Bouwman, W. G.; Kjaer, K. Macromolecules 1998, 31, 8893-8906. (38) Tarabia, M.; Hong, H.; Davidov, D.; Kirstein, S.; Steitz, R.; Neumann, R.; Avny, Y. Journal of Applied Physics 1998, 83, 725-732. (39) Ladam, G.; Schaad, P.; Voegel, J. C.; Schaaf, P.; Decher, G.; Cuisinier, F. Langmuir 2000, 16, 1249-1255. (40) Schlenoff, J. B.; Dubas, S. T. Macromolecules 2001, 34, 592-598. (41) Castelnovo, M.; Joanny, J.-F. Langmuir 2000, 16, 7524-7532. (42) Joanny, J.-F.; Castelnovo, M.; Netz, R. Journal of Physics: Condensed

269

Matter 2000, 12, A1-A7. (43) Solis, F. J.; de la Cruz, M. O. Journal of Chemical Physics 1999, 110, 11517-11522. (44) Park, S. Y.; Rubner, M. F.; Mayes, A. M. Langmuir 2002, 18, 9600-9604. (45) Netz, R.; Joanny, J. F. Macromolecules 1999, 32, 9013-9025. (46) Wang, Q. J. Phys. Chem. B 2006, 110, 5825-5828. (47) Shafir, A.; Andelman, D. European Physical Journal E: Soft Matter 2006, 19, 155-162. (48) Lefaux, C. J.; Zimberlin, J. A.; Dobrynin, A. V.; Mather, P. T. Journal of Polymer Science, Part B: Polymer Physics 2004, 42, 3654-3666. (49) Messina, R. Macromolecules 2004, 37, 621-629. (50) Messina, R.; Holm, C.; Kremer, K. Langmuir 2003, 19, 4473-4482. (51) Messina, R.; Holm, C.; Kremer, K. Journal of Polymer Science part B: polymer physics 2004, 42, 3557-3570. (52) Panchagnula, V.; Jeon, J.; Dobrynin, A. V. Physical Review Letters 2004, 93, 037801. (53) Panchagnula, V.; Jeon, J.; Rusling, J. F.; Dobrynin, A. V. Langmuir 2004, 21, 1118-1125. (54) Cho, J.; Char, K.; Hong, J. D.; Lee, K. B. Advanced Materials 2001, 13, 1076-+. (55) Chiarelli, P. A.; Johal, M. S.; Casson, J. L.; Roberts, J. B.; Robinson, J. M.; Wang, H. L. Advanced Materials 2001, 13, 1167-+. (56) Chiarelli, P. A.; Johal, M. S.; Holmes, D. J.; Casson, J. L.; Robinson, J. M.; Wang, H.-L. Langmuir 2002, 18, 168-173. (57) Lee, S.-S.; Lee, K.-B.; Hong, J.-D. Langmuir 2003, 19, 7592-7596. (58) An, M.; Hong, J.-D. Thin Solid Films 2006, 500, 74-77. (59) Cho, J.; Lee, S.-H.; Kang, H.; Char, K.; Koo, J.; Seung, B. H.; Lee, K.-B. Polymer 2003, 44, 5455-5459. (60) C. Jiang, S. M. V. V. T. Advanced Materials 2004, 16, 157-161. (61) Sohn, B.-H.; Kim, T.-H.; Char, K. Langmuir 2002, 18, 7770-7772. (62) Chiarelli, P. A.; Johal, M. S.; Holmes, D. J.; Casson, J. L.; Robinson, J. M.; Wang, H. L. Langmuir 2002, 18, 168-173. (63) Chiarelli, P. A.; Johal, M. S.; Casson, J. L.; Roberts, J. B.; Robinson, J. M.; Wang, H.-L. Advanced Materials 2001, 13, 1167-1171. (64) Lee, S.-S.; Hong, J.-D.; Kim, C. H.; Kim, K.; Koo, J. P.; Lee, K.-B. Macromolecules 2001, 34, 5358-5360.

270

(65) Fou, A. C.; Onitsuka, O.; Ferreira, M.; Rubner, M. F.; Hsieh, B. R. Journal of Applied Physics 1996, 79, 7501-7509. (66) Cho, J.; Char, K.; Hong, J.-D.; Lee, K.-B. Advanced Materials 2001, 13, 1076-1078. (67) Cho, J.; Jang, H.; Yeom, B.; Kim, H.; Kim, R.; Kim, S.; Char, K.; Caruso, F. Langmuir 2006, 22, 1356-1364. (68) Lefaux, C., PhD Thesis, University of Connecticut, 2004. (69) Rubinstein, M.; Colby, R. H. Polymer Physics; Oxford University Press: New York, 2003. (70) Harden, J. L.; Cates, M. E. Physical Review E 1996, 53, 3782. (71) Rabin, Y.; Alexander, S. Europhysics Letters 1990, 13, 49-54. (72) Harden, J. L.; Cates, M. E. Physical Review E 1996, 53, 3782-3787. (73) Rabin, Y.; Alexander, S. Europhysics letters 1990, 13, 49-54. (74) Lowenstam, H. A. Science 1981, 211, 1126-31. (75) Mann, S. Angewandte Chemie, International Edition 2000, 39, 3392-3406. (76) Bauerlein, E. Angewandte Chemie, International Edition 2003, 42, 614- 641. (77) Dujardin, E.; Mann, S. Advanced Engineering Materials 2002, 4, 461-474. (78) Estroff, L. A.; Hamilton, A. D. Chemistry of Materials 2001, 13, 3227- 3235. (79) Belcher, A. M.; Hansma, P. K.; Stucky, G. D.; Morse, D. E. Acta Materialia 1998, 46, 733-736. (80) Mann, S.; Webb, J.; Williams, R. J. P. Biomineralization: Chemical and Biochemical Perspectives; VCH: Weinheim, 1989. (81) Lowenstam, H. A.; Weiner, S. On Biomineralization; Oxford University Press: New York, 1989. (82) Kroger, N.; Deutzmann, R.; Sumper, M. Science 1999, 286, 1129-1132. (83) Hildebrand, M.; Wetherbee, R. In Silicon Biomineralization: Biology- Biochemistry-Molecular Biology-Biotechnology; Muller, W. E. G., Ed.; Springer: 2003; Vol. 33, p 11-51. (84) Aizenberg, J.; Black, A. J.; Whitesides, G. M. Nature 1999, 398, 495-498. (85) Bunker, B. C.; Rieke, P. C.; Tarasevich, B. J.; Campbell, A. A.; Fryxell, G. E.; Graff, G. L.; Song, L.; Liu, J.; Virden, J. W.; McVay, G. L. Science 1994, 264, 48-55. (86) Mann, S.; Archibald, D. D.; Didymus, J. M.; Douglas, T.; Heywood, B. R.; Meldrum, F. C.; Reeves, N. J. Science 1993, 261, 1286-92. (87) Mann, S. Nature 1988, 332, 119-24.

271

(88) Heywood, B. R.; Mann, S. Advanced Materials 1994, 6, 9-20. (89) Mann, S.; Heywood, B. R.; Rajam, S.; Walker, J. B. A. J. Phys. D: Appl. Phys. 1991, 24, 154-64. (90) Mann, S.; Didymus, J. M.; Sanderson, N. P.; Heywood, B. R.;Journal of the Chemical Society, Faraday Transactions 1990, 86, 1873-80. (91) Weiner, S.; Addadi, L. Journal of Materials Chemistry 1997, 7, 689-702. (92) Tsortos, A.; Nancollas, G. H. Journal of Colloid and Interface Science 2002, 250, 159-167. (93) Fisher, L. W.; Whitson, S. W.; Avioli, L. V.; Termine, J. D. The Journal of biological chemistry 1983, 258, 12723-7. (94) Franzen, A.; Heinegard, D. Biochemical journal 1985, 232, 715-24. (95) Hunter, G. K. Biochemical Journal 1994, 302, 175. (96) Hunter, G. K.; Goldberg, H. A. Proceedings of the National Academy of Sciences of the United States of America 1993, 90, 8562-5. (97) Tsortos, A.; Ohki, S.; Zieba, A.; Baier, R. E.; Nancollas, G. H. Journal of Colloid and Interface Science 1996, 177, 257-62. (98) Tsortos, A., PhD Dissertation, University of Buffalo, 1994. (99) Tomson, M. B.; Nancollas, G. H. Science 1978, 200, 1059-60. (100) Koutsoukos, P.; Amjad, Z.; Tomson, M. B.; Nancollas, G. H. Journal of the American Chemical Society 1980, 102, 1553-7. (101) Koutsopoulos, S.; Dalas, E. Langmuir 2001, 17, 1074-1079. (102) Addadi, L.; Weiner, S. Proc Natl Acad Sci 1985, 82, 4110-4. (103) Addadi, L.; Moradian, J.; Shay, E.; Maroudas, N. G.; Weiner, S. Proc. Natl. Acad. Sci. 1987, 84, 2732-6. (104) Lahav, M.; Addadi, L.; Leiserowitz, L. Proc. Natl. Acad. Sci. 1987, 84, 4737-8. (105) Fleisch, H.; Neuman, W. F. Journal of the American Chemical Society 1960, 82, 996-7. (106) Ball, V.; Michel, M.; Boulmedais, F.; Hemmerle, J.; Haikel, Y.; Schaaf, P.; Voegel, J. C. Crystal Growth & Design 2006, 6, 327-334. (107) Iler, R. K. The Chemistry of Silica; Wiley-Interscience: New York, 1979. (108) Brinker, C.; Scherer, G. Sol-Gel Science: The Physics and Chemistry of Sol-Gel Processing, 1990. (109) Brinker, C. J. Journal of Non-Crystalline Solids 1988, 100, 31. (110) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.; Vartuli, J. C.; Beck, J. S. Nature 1992, 359, 710-12.

272

(111) Caruso, R. A.; Antonietti, M. Chemistry of Materials 2001, 13, 3272-3282. (112) Iler, R. K. In Chemistry of Silica 1979, p 734-739. (113) Martin-Jezequel, V.; Hildebrand, M.; Brzezinski, M. A. Journal of Phycology 2000, 36, 821-840. (114) Nelson, D. M.; Treguer, P.; Brzezinski, M. A.; Leynaert, A.; Queguiner, B. Global Biogeochemical Cycles 1995, 9, 359-72. (115) Hildebrand, M.; Wetherbee, R. In Silicon Biomineralization: Biology- Biochemistry-Molecular Biology-Biotechnology; Muller, W. E. G., Ed.; Springer: 2003; Vol. 33, p 11-51. (116) Noll, F.; Sumper, M.; Hampp, N. Nano Letters 2002, 2, 91-95. (117) Poulsen, N.; Sumper, M.; Kroeger, N. Proceedings of the National Academy of Sciences of the United States of America 2003, 100, 12075-12080. (118) Kroger, N.; Deutzmann, R.; Bergsdorf, C.; Sumper, M. Proceedings of the National Academy of Sciences of the United States of America 2000, 97, 14133-14138. (119) Brunner, E.; Lutz, K.; Sumper, M. Physical Chemistry Chemical Physics 2004, 6, 854-857. (120) Kroeger, N.; Lorenz, S.; Brunner, E.; Sumper, M. Science 2002, 298, 584- 586. (121) Sumper, M.; Lorenz, S.; Brunner, E. Angewandte Chemie, International Edition 2003, 42, 5192-5195. (122) Pohnert, G. Angewandte Chemie, International Edition 2002, 41, 3167- 3169. (123) Cha, J. N.; Shimizu, K.; Zhou, Y.; Christiansen, S. C.; Chmelka, B. F.; Stucky, G. D.; Morse, D. E. Proceedings of the National Academy of Sciences of the United States of America 1999, 96, 361-365. (124) Knecht Marc, R.; Wright David, W. Chemical communications 2003, 3038-9. (125) Naik, R. R.; Whitlock, P. W.; Rodriguez, F.; Brott, L. L.; Glawe, D. D.; Clarson, S. J.; Stone, M. O. Chemical Communications 2003, 238-239. (126) Coradin, T.; Livage, J. Colloids and Surfaces, B: Biointerfaces 2001, 21, 329-336. (127) Coradin, T.; Durupthy, O.; Livage, J. Langmuir 2002, 18, 2331-2336. (128) Belton, D.; Paine, G.; Patwardhan, S. V.; Perry, C. C. Journal of Materials Chemistry 2004, 14, 2231-2241. (129) Mizutani, T.; Nagase, H.; Ogoshi, H. Chemistry Letters 1998, 133-134. (130) Patwardhan, S. V.; Clarson, S. J. Silicon Chemistry 2002, 1, 207-214. (131) Rodriguez, F.; Glawe, D. D.; Naik, R. R.; Hallinan, K. P.; Stone, M. O. Biomacromolecules 2004, 5, 261-265. 273

(132) Monte Carlo and Molecular Dynamics Simulations in Polymer Science; Binder, K., Ed.; Oxford University Press, New York, 1995. (133) Menzel, H.; Horstmann, S.; Behrens, P.; Baernreuther, P.; Krueger, I.; Jahns, M. Chemical Communications 2003, 2994-2995. (134) Jin, R.-H.; Yuan, J.-J. Chemical Communications 2005, 1399-1401. (135) Jin, R.-H.; Yuan, J.-J. Macromolecular Chemistry and Physics 2005, 206, 2160-2170. (136) Coradin, T.; Lopez, P. J. ChemBioChem 2003, 4, 251-259. (137) Patwardhan, S. V.; Clarson, S. J.; Perry, C. C. Chemical Communications 2005, 2005, 1113. (138) Yuan, J.-J.; Jin, R.-H. Advanced Materials 2005, 17, 885-888. (139) Yuan, J.-J.; Jin, R.-H. Langmuir 2005, 21, 3136-3145. (140) Yuan, J.-J.; Zhu, P.-X.; Fukazawa, N.; Jin, R.-H. Advanced Functional Materials 2006, 16, 2205-2212. (141) Lindquist, G. M.; Stratton, R. A. Journal of Colloid and Interface Science 1976, 55, 45-59. (142) Ries, H. E., Jr. Nature 1970, 226, 72. (143) Lefaux, C. J.; Mather, P. T., personal communications (144) Iler, R. K. In Chemistry of Silica 1979, p 174. (145) Sumper, M.; Brunner, E. Advanced Functional Materials 2006, 16, 17. (146) Wenzl, S.; Deutzmann, R.; Hett, R.; Hochmuth, E.; Sumper, M. Angewandte Chemie International Edition 2004, 43, 5933. (147) Plimpton, S. Journal of Computational Physics 1995, 117, 1-19. (148) A chain is considered to belong to a cluster if it has at least one monomer within distance 21/6σ from any monomer belonging to a chain forming the cluster. The cluster analysis was performed by analyzing the matrix of distances between all monomers in the system. (149) Schlenoff, J. B.; Dubas, S. T. Macromolecules 2000, 34, 592-598. (150) Lvov, Y.; Ariga, K.; Onda, M.; Ichinose, I.; Kunitake, T. Colloids and Surfaces A: Physicochemical and Engineering Aspects 1999, 146, 337-346. (151) Schlenoff, J. B.; Ly, H.; Li, M. Journal of American Chemical Society 1998, 120, 7626-7634. (152) Rapaport, D. C.; Cambridge University Press: New York, 1995. (153) Castelnovo, M.; Joanny, J. F. Langmuir 2000, 16, 7524-7532. (154) Grosberg, A. Y.; Khokhlov, A. R. Statistical Physics of Macromolecules; AIP Press, 1994.

274

(155) Solis, F. J.; Cruz, M. O. d. l. Journal of Chemical Physics 1999, 110, 11517-11522. (156) Farhat, T.; Yassin, G.; Dubas, S. T.; Schlenoff, J. B. Langmuir 1999, 15, 6621-6623. (157) Yoo, D.; Shiratori, S. S.; Rubner, N. F. Macromolecules 1998, 31, 4309- 4318. (158) Michaels, A. S.; Miekka, G. Journal of Physical Chemistry 1961, 65, 1765. (159) McAloney, R.; Sinyor, M.; Dudnik, V.; Cynthia, M. Langmuir 2001, 17, 6655-6663. (160) Menchaca, J. L.; Jachimska, B.; Cuisinier, F.; Perez, E. Colloids and Surfaces A: Physicochemical and Engineering Aspects 2003, 222, 185-194. (161) Park, S. Y.; Rubner, M. F.; Mayes, A. M. Langmuir 2002, 18, 9600-9604. (162) Kotov, N. A. Nanostructured Materials 1999, 12, 789-796. (163) Dobrynin, A. V.; Colby, R. H.; Rubinstein, M. Journal of Polymer Science Part B: Polymer Physics 2004, 42, 3513. (164) Grosberg, A.; Nguyen, T.; Shklovskii, B. Reviews of Modern Physics 2002, 74, 329. (165) Doi, M.; Edwards, S. F. The theory of poymer dynamics; Clarendon Press: Oxford, 1989. (166) Borue, V. Y.; Erukhimovich, I. Y. Macromolecules 1990, 23, 3625. (167) Johal, M. S.; Casson, J. L.; Chiarelli, P. A.; Liu, D.-G.; Shaw, J. A.; Robinson, J. M.; Wang, H.-L. Langmuir 2003, 19, 8876. (168) Menchaca, J. L.; Jachimska, B.; Cuisinier, F.; Perez, E. Colloids and Surfaces A: Physicochemical and Engineering Aspects 2003, 222, 185. (169) Richert, L.; Engler, A. J.; Discher, D. E.; Picart, C. Biomacromolecules 2004, 5, 1908. (170) Picart, C.; Senger, B.; Sengupta, K.; Dubreuil, F.; Colloids and Surfaces A: Physicochemical and Engineering Aspects 2007, 303, 30. (171) Kiely, J. D.; Bonnell, D. A. Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures 1997, 15, 1483-1493. (172) Dubas, S. T.; SCHLENOFF, J. B. Macromolecules 1999, 32, 8153-8160. (173) Lvov, Y. M.; Decher, G. Kristallografiya 1994, 39, 696-716. (174) Ruths, J.; Essler, F.; Decher, G.; Riegler, H. Langmuir 2000, 16, 8871- 8878. (175) Steitz, R.; Leiner, V.; Siebrecht, R.; v. Klitzing, R. Colloids and Surfaces A: Physicochemical and Engineering Aspects 2000, 163, 63. (176) Losche, M.; Schmitt, J.; Decher, G.; Bouwman, W. G.; Kjaer, K.

275

Macromolecules 1998, 31, 8893-8906. (177) Castelnovo, M.; Joanny, J. F. Langmuir 2000, 16, 7524-7532. (178) Harden, J. L.; Cates, M. E. Physical Review E: 1996, 53, 3782-7. (179) Alfred, G. E.; Francis, T. B.; Leslie, G. P. Journal of Applied Physics 1958, 29, 858-862. (180) Larson, R. G. Journal of Rheology 2005, 49, 1-70. (181) Perkins, T. T.; Smith, D. E.; Larson, R. G.; Chu, S. Science 1995, 268, 83-7. (182) Stillwagon, L. E.; Larson, R. G.; Taylor, G. N. Journal of the Electrochemical Society 1987, 134, 2030-2037. (183) Stillwagon, L. E.; Larson, R. G. Journal of Applied Physics 1988, 63, 5251-5258. (184) Stillwagon, L. E.; Larson, R. G. Physics of Fluids a-Fluid Dynamics 1990, 2, 1937-1944. (185) McAloney, R. A.; Sinyor, M.; Dudnik, V.; Goh, M. C. Langmuir 2001, 17, 6655-6663. (186) Dubas, S. T.; Schlenoff, J. B. Langmuir 2001, 17, 7725-7727. (187) Koutsopoulos, S. Journal of Biomedical Materials Research 2002, 62, 600-612. (188) Zawacki, S. J., PhD thesis, University of Bufallo, 1988. (189) Koutsopoulos, S. Langmuir 2001, 17, 8092-8097. (190) Oyane, A.; Kim, H.-M.; Furuya, T.; Kokubo, T.; Miyazaki, T.; Nakamura, T. Journal of Biomedical Materials Research, Part A 2003, 65A, 188-195. (191) Brett, L. L.; Naik, R. R.; Pikas, D. J.; Kirkpatrlck, S. M.; Tomlin, D. W.; Whitlock, P. W.; Clarsen, S. J.; Stone, M. O. Nature 2001, 413, 291-293. (192) Tsortos, A.; Nancollas, G. H. Journal of Colloid and Interface Science 1999, 209, 109-115. (193) Kresak, M.; Moreno, E. C.; Zahradnik, R. T.; Hay, D. I. Journal of Colloid and Interface Science 1977, 59, 283-92. (194) Inoue, S.; Ohtaki, N. Journal of Chromatography 1990, 515, 193-204. (195) Fernandez, V. L.; Reimer, J. A.; Denn, M. M. Journal of the American Chemical Society 1992, 114, 9634-42. (196) Nagasawa, M.; Holtzer, A. Journal of the American Chemical Society 1964, 86, 538-43. (197) Olander, D. S.; Holtzer, A. Journal of the American Chemical Society 1968, 90, 4549-60. (198) Achilles, T.; Nancollas, G. H. Journal of Colloid and Interface Science 276

2002, 250, 159-168. (199) Fujisawa, R.; Kuboki, Y. European Journal of Oral Sciences 1998, 106, 249-253. (200) Wang, L.; Guan, X.; Du, C.; Moradian-Oldak, J.; Nancollas, G. H. Journal of Physical Chemistry C 2007, 111, 6398-6404. (201) Iijima, M.; Moriwaki, Y.; Wen, H. B.; Fincham, A. G.; Moradian-Oldak, J. Journal of Dental Research 2002, 81, 69-73. (202) Moradian-Oldak, J. Matrix Biology 2001, 20, 293-305. (203) Ebrahimpour, A.; Johnsson, M.; Richardson, C. F.; Nancollas, G. H. Journal of Colloid and Interface Science 1993, 159, 158-63. (204) Hunter, G. K.; Haushka, P. V.; Poole, A. R.; Rosenberg, L. C.; Goldberg, H. A. Biochemical Journal 1996, 317, 59-64. (205) Tye, C. E.; Rattray, K. R.; Warner, K. J.; Gordon, J. A. R.; Sodek, J.; Hunter, G. K.; Goldberg, H. A. Journal of Biological Chemistry 2003, 278, 7949-7955. (206) Ngankam, P. A.; Lavalle, P.; Voegel, J. C.; Szyk, L.; Decher, G.; Schaaf, P.; Cuisinier, F. J. G. Journal of the American Chemical Society 2000, 122, 8998-9005. (207) Murphy, W. L.; Mooney, D. J. Journal of the American Chemical Society 2002, 124, 1910-1917. (208) Yamaguchi, I.; Tokuchi, K.; Fukuzaki, H.; Koyama, Y.; Takakuda, K.; Monma, H.; Tanaka, J. Journal of Biomedical Materials Research 2000, 55, 20-27. (209) Li, P.; Bakker, D.; van Blitterswijk, C. A. Journal of Biomedical Materials Research 1997, 34, 79-86. (210) Chen, J.; Chu, B.; Hsiao, B. S. Journal of Biomedical Materials Research, Part A 2006, 79A, 307-317. (211) Oliveira, A. L.; Mano, J. F.; Reis, R. L. Current Opinion in Solid State & Materials Science 2003, 7, 309-318. (212) Patwardhan, S. V.; Mukherjee, N.; Clarson, S. J. Journal of Inorganic and Organometallic Polymers 2002, 11, 193-198. (213) Butler, R. T.; Ferrell, N. J.; Hansford, D. J. Applied Surface Science 2006, 252, 7337. (214) Pogula, S. D.; Patwardhan, S. V.; Perry, C. C.; Gillespie, J. W.; Yarlagadda, S.; Kiick, K. L. Langmuir 2007, 23, 6677-6683. (215) Advincula, M. C.; Patel, P. A.; Mather, P. T.; Goldberg, A. J. In Material Reserch Society; MRS Publications: Boston, 2006. (216) Delak Katya, M.; Sahai, N. The journal of physical chemistry. B, 2006, 110, 17819-29. (217) Delak, K. M.; Sahai, N. Chemistry of Materials 2005, 17, 3221-3227.

277

(218) Kroger, N.; Sumper, M. In Biomineralization; Baeuerlein, E., Ed.; Wiley- VCH: Weinheim, 2000, p 151-170. (219) Ries, H. E., Jr. Journal of Applied Polymer Science 1971, 15, 2023. (220) Vrieling, E. G.; Beelen, T. P. M.; van Santen, R. A.; Gieskes, W. W. C. Angewandte Chemie 2002, 114, 1613. (221) Brunne, E.; Lut, K.; Sumpe, M. Physical Chemistry Chemical Physics 2004, 6, 854. (222) Pouget, E.; Dujardin, E.; Cavalier, A.; Moreac, A.; Valery, C.; Marchi- Artzner, V.; Weiss, T.; Renault, A.; Paternostre, M.; Artzner, F. Nature Materials 2007, advanced online publication. (223) Sumper, M. Angewandte Chemie, International Edition 2004, 43, 2251- 2254. (224) Sumper, M.; Kroeger, N. Journal of Materials Chemistry 2004, 14, 2059- 2065. (225) Katti, A.; Shimpi, N.; Roy, S.; Lu, H.; Fabrizio, E. F.; Dass, A.; Capadona, L. A.; Leventis, N. Chemistry of Materials 2006, 18, 285-296. (226) Leventis, N.; Palczer, A.; McCorkle, L.; Zhang, G.; Sotiriou-Leventis, C. J. Journal of Sol-Gel Science and Technology 2005, 35, 99-105. (227) Chatani, Y.; Tadokoro, H.; Saegusa, T.; Ikeda, H. Macromolecules 1981, 14, 315-321. (228) Hashida, T.; Tashiro, K.; Inaki, Y. Polymer 2003, 44, 1721-1724. (229) Hashida, T.; Tashiro, K.; Inaki, Y. Journal of Polymer Science, Part B: Polymer Physics 2003, 41, 2937-2948. (230) Hashida, T.; Tashiro, K.; Aoshima, S.; Inaki, Y. Macromolecules 2002, 35, 4330-4336. (231) Fidalgo, A.; Farinha, J. P. S.; Martinho, J. M. G.; Rosa, M. E.; Ilharco, L. M. Chemistry of Materials 2007, 19, 2603-2609. (232) Gong, H.; Garcia-Turiel, J.; Vasilev, K.; Vinogradova, O. I. Langmuir 2005, 21, 7545-7550. (233) Hong, J. D.; Jung, B. D.; Kim, C. H.; Kim, K. Macromolecules 2000, 33, 7905-7911. (234) Toshiaki, M.; Masuhiro, M. Physical Review E 2007, 75, 031804. (235) Miura, T.; Kishi, R.; Mikami, M.; Tanabe, Y. Physical Review E 2001, 63, 061807. (236) Dobrynin, A. V.; Rubinstein, M. Progress in Polymer Science 2005, 30, 1049-1118. (237) Hur, J. S. Journal of Rheology 2000, 44, 713. (238) Hoda, N.; Kumar, S. Journal of Rheology 2007, 51, 799-820. 278

(239) Panwar, A. S.; Kumar, S. J. Journal of Chemical Physics 2005, 122, 154902/1-154902/12. (240) Filippova, N. L. Journal of Colloid and Interface Science 1999, 211, 336- 354. (241) Panwar Ajay, S.; Kumar, S. J The Journal of chemical physics 2005, 122, 154902. (242) Smith, D. E.; Babcock, H. P.; Chu, S. Science 1999, 283, 1724-1727. (243) Podsiadlo, P.; Liu, Z.; Paterson, D.; Messersmith, P.; thinsp; B; Kotov, N.; thinsp; A Advanced Materials 2007, 19, 949. (244) Tang, Z.; Kotov, N. A.; Magonov, S.; Ozturk, B. Nature Materials 2003, 2, 413. (245) Brott, L. L.; Naik, R. R.; Pikas, D. J.; Kirkpatrick, S. M.; Tomlin, D. W.; Whitlock, P. W.; Clarson, S. J.; Stone, M. O. Nature 2001, 413, 291-3. (246) Coffman, E. A.; Melechko, A. V.; Allison, D. P.; Simpson, M. L.; Doktycz, M. J. Langmuir 2004, 20, 8431-8436. (247) Kim, D. J.; Lee, K.-B.; Lee, T. G.; Shon, H. K.; Kim, W.-J.; Paik, H.-j.; Choi, I. S. Small 2005, 1, 992-996. (248) Feng, K.; Zang, H.; Martin, D.; Marino, H. L.; Neckers, D. C. J. Journal of Polymer Science, Part A: Polymer Chemistry 1998, 36, 1667-1677. (249) Gu, H.; Ren, K.; Grinevich, O.; Malpert, J. H.; Neckers, D. C. Journal of organic chemistry 2001, 66, 4161-4.

279