INCORPORATION of LESS TOXIC ANTIFOULING COMPOUNDS INTO SILICONE COATINGS to STUDY THEIR RELEASE BEHAVIORS a Dissertation Present

Home , Polydimethylsiloxane, Sodium benzoate, Tannic acid

INCORPORATION OF LESS TOXIC ANTIFOULING COMPOUNDS INTO

SILICONE COATINGS TO STUDY THEIR RELEASE BEHAVIORS

A Dissertation

Presented to

The Graduate Faculty of The University of Akron

In Partial Fulfillment

of the Requirement for the Degree

Doctor of Philosophy

Abdulhadi Abdullah Al-Juhni

August, 2006

INCORPORATION OF LESS TOXIC ANTIFOULING COMPOUNDS INTO

SILICONE COATINGS TO STUDY THEIR RELEASE BEHAVIORS

Abdulhadi Abdullah Al-Juhni

Dissertation

Approved: Accepted:

Advisor Department Chair Dr. Bi-min Zhang Newby Dr. Lu-Kwang Ju

Committee Member Dean of the College Dr. George G. Chase Dr. George K. Haritos

Committee Member Dean of the Graduate School Dr. Lu-Kwang Ju Dr. George R. Newkome

Committee Member Date Dr. Gerald W. Young

Committee Member Dr. Teresa J. Cutright

ABSTRACT

Biofouling has always caused serious problems, including increased fuel

consumption and maintenance costs for vessels, to the naval industry. The use of toxic

antifouling compounds to combat the organisms attached and accumulated on the surface

of submerged structures has been common. However, the current ban on the application

of conventional tin-based antifouling compounds has accelerated the research to seek for less toxic alternatives.

In this study, four much less-toxic antifouling compounds (sodium benzoate, benzoic acid, capsaicin, and tannic acid), as compared to tin-based antifoulants, were incorporated into two types of silicone coatings (Sylgard® 184 and RTV11) by applying a

solvent-blending technique. These compounds were proven to be effective against some

bacterial growth and attachments; however, a systematic study on the miscibility and the release of these compounds from the coatings is lacking. Therefore, the focus of the current study is to correlate the miscibility of the compounds in the coatings and their release rates in water, for the purpose of controlling the release of the compounds.

It was found that benzoic acid and capsaicin formed large crystals inside the coating; whereas sodium benzoate and tannic acid were able to form small aggregates inside the coatings. The magnitude of the leaching of the four compounds was in the iii order of: benzoic acid > capsaicin > sodium benzoate > tannic acid. The solvent-assisted

blending technique was adequate for the cases of sodium benzoate and tannic acid

whereas it was not suitable for benzoic acid and capsaicin. Sodium benzoate/Sylgard®

184 coating was then selected as the model system to obtain the miscibility-release relationship. The preparation conditions were found to have important effects on the morphological structure and final distribution of sodium benzoate in silicone, hence the leaching. The minimum average aggregate size obtained was ~ 3 µm, which had resulted in the lowest value for the steady leaching rate of ~ 0.1 µg/cm2/day. Empirical correlations were obtained between the aggregate size as well as the matrix loading of sodium benzoate and the leaching rate. It was found that increasing the aggregate size had a sharp effect on the increase of the leaching rate, whereas increasing the matrix loading (up to 5 wt. %) had a mild effect on the leaching rate. The current study did show that the solvent-assisted blending technique can be an efficient approach for constructing the miscibility-release correlations.

Both thermodynamic analysis and experimental observations showed that sodium benzoate has limited solubility in the Sylgard® 184 coating. This, combined with the

mass transfer analysis of the leaching, led us to confirm that the release mechanism of the

monolithic sodium benzoate/silicone coatings generated via the solvent-assisted blending

technique is mainly by the diffusion of the compound through water-filled pores and

constricted channels within the matrix, not through the continuum of the polymer phase.

ACKNOWLEDGEMENTS

I would like to thank my advisor Dr. Bi-min Zhang Newby, for consistently and

patiently providing me with continues guidance, informative discussion, and educational help throughout my PhD research. I would like also to thank my committee members;

Dr. Teresa Cutright, Dr. George Chase, Dr. Lu-Kwang Ju, and Dr. Gerald Young, for

their informative discussion and valuable comments and suggestion. I would like also to

thank Dr. Cutright and her research group for assisting me in bacterial attachment study.

I would like also to thank Dr. Sung-Hwan Choi for the AFM scanning. Thanks are also

due to all members of my research group, for being helpful during my entire study.

I want to express my deep thanks to the Ministry of Higher Education, Saudi Arabia,

for granting me a full scholarship to get my PhD degree, without their financial support it

would be difficult to accomplish this study. Appreciations are also due to the Department

of Chemical and Biomolecular Engineering, the Ohio Sea Grant (Project: R/MB-2) and

the Ohio Board of Regents, for partial financial support of my research project.

Finally, I would like to express my deep gratitude to my wife Fatimah, being patient

here with me and supporting me to accomplish my objective, and to my parents, far away

and being patient waiting for me.

TABLE OF CONTENTS

Page

LIST OF TABLES………………………………………………………………………...x

LIST OF FIGURES……………………………………………………………………...xii

CHAPTER

I. INTRODUCTION………………………………….…………………………...... 1

1.1 Introduction…...…………………………………………………………………1

1.2 Importance and scope of the study…………….………………………………...5

1.3 Objectives…………………………………….…………………………...... 6

1.4 Dissertation outline…………………………...………………………………….7

II. LITERATURE REVIEW………………………………….…………………………8

2.1 Biofouling and biofilms formation…..…………………………………………..8

2.2 Biofouling controls…..…………………………………………………………10

2.2.1 Conventional toxic antifouling coatings……………………………...11

2.2.2 Silicone foul-release coatings………….……………………………..12

2.2.3 Less-toxic antifoulants…………………….………………………….16

2.3 Antifoulant-matrix miscibility………………………………………………….23

2.4 Modeling of the antifoulants release from polymeric matrices…….…………..33

2.4.1 Active compounds with high solubility in the matrix………………...35

vi 2.4.2 Active compounds with low solubility in the matrix………..………..38

2.4.3 Leaching behaviors of marine antifouling paints……………………..41

III. EXPERIMENTAL APPROACH………..………………………………………….46

3.1 Materials…….………………………………………………………………….46

3.2 Sample preparation……………………………………………………………..48

3.3 Sample processing……………………………………………………………...53

3.4 Characterization techniques…………………………………………………….57

3.4.1 Contact angles technique.…………………………………………….57

3.4.2 The stress - strain technique………….……………………………….59

3.4.3 The JKR technique…………………………………………………....60

3.4.4 Optical microscopy…………………………………………………...61

3.4.5 Scanning probe microscopy………….……………………………….63

3.4.6 High performance liquid chromatography (HPLC)…………………..64

IV. RESULTS AND DISCUSSION FOR SODIUM BENZOATE– BASED COATINGS……………………………………………………………….66

4.1 Effect of sodium benzoate on surface and bulk properties of silicones….…….66

4.1.1 Effect on wettability…………………………………………………..67

4.1.2 Effect on elastic modulus……………………………………………..71

4.2 Miscibility of NaB in silicones……...………………………………………….73

4.2.1 Effect of composition of the mixed solvent…………………………..74

4.2.2 Effect of solvent/polymer ratio……...………………………………..78

4.2.3 Effect of NaB matrix loading……….………………………………...78

4.3 Thermodynamic analysis for the miscibility study ….………………………...83

4.3.1 Prediction by the Flory-Huggins theory……………………………....83

vii 4.3.2 Modification of the Flory-Huggins theory to include electrostatic contribution and concentration-dependent interaction parameters…..………………………………………………………...86

4.3.3 Comparison between the theoretical miscibility trends and the experimental morphology trends……………………………………...95

4.4 Leaching evaluation……...…………………………………………………....100

4.4.1 Effect of composition of the mixed solvent………………………....100

4.4.2 Effect of solvent/polymer ratio……………………………………...102

4.4.3 Effect of NaB matrix loading……...………………………………...102

4.4.4 Effect of type of the silicone matrix………………………………....105

4.4.5 Empirical correlations for the leaching rate of NaB from Sylgard® 184…………………………………………………..108

4.4.6 Effects of continuous stirring and water replacement ………..……..111

4.5 Mass transfer analysis for the leaching study………..………………………..114

4.5.1 Simplified mass transfer model……………………………………...114

4.5.2 Limitation of the simplified mass transfer model…………………...130

4.6 Bacterial attachment evaluations...... ……………………………………...138

V. RESULTS AND DISCUSSION FOR BENZOIC ACID AND CAPSAICIN- BASED COATINGS…...... 144

5.1 Effects of the compounds on coating’s properties…………...……………….144

5.1.1 Effect of benzoic acid……………………………...... 144

5.1.2 Effect of capsaicin…………………………………………………...147

5.2 Miscibility of the compounds in silicones…….……………………………....150

5.2.1 Miscibility of benzoic acid…………………………...... 150

5.2.2 Miscibility of capsaicin……………………………………………...154

5.3 Leaching evaluation…………………………………………………………..157

viii 5.3.1 Leaching of benzoic acid……………………………...... 157

5.3.2 Leaching of capsaicin……………………………………………….159

5.4 Bacterial attachment evaluations for capsaicin-RTV11 coatings.……………163

5.4.1 Effect of immersion in water on coating’s properties………….…...163

5.4.2 Bacterial attachment evaluations……………………………………169

VI. RESULTS AND DISCUSSION FOR TANNIC ACID-BASED COATINGS…...171

6.1 Effect of tannic acid on coating’s properties……………………………….....171

6.2 Miscibility of tannic acid in silicones……………………………………...... 173

6.3 Leaching evaluation……..…………………………………………………….176

VII. CONCLUSIONS AND RECOMMENDATIONS.……………………………...... 179

7.1 Conclusions….….…………………………………………...... 179

7.2 Recommendations for future work…………………….…………………….182

REFERENCES…….…………………………………………………………………...187

APPENDICES………………………………………………………………………….196

APPENDIX A MATLAB FILE FOR SOLVING THE GENERAL MASS TRANSFER MODEL (EQUATION 4.19)……………………..197

APPENDIX B SUGGESTION OF A MORE REALISTIC MASS TRANSFER MODEL – APPLICATION OF THE AVERAGE VOLUME THEORY……………………………...201

LIST OF TABLES

Table Page

3.1 The compositions of the two types of silicones (RTV11 and Sylgard® 184) used in the current study. All percentage shown here are in mass basis …….…...... 48

3.2 The actual combinations of (antifoulant/solvent/polymer) mixtures used in the current study to incorporate the antifoulant into the bulk of the polymer matrix by the solvent-blending technique…………………………………………………49

3.3 The detailed concentrations for NaB/ Sylgard®184 samples prepared at different conditions. All concentration shown here are in mass basis. (Abbreviation: W: water, A: acetone, S: solvent (i.e. water + acetone), P: silicone polymer base)…..51

3.4 The detailed compositions for the samples that were subjected to leaching experiments in the current study. All concentration shown here are in mass basis. (Abbreviation: W: water, A: acetone, S: solvent, P: silicone polymer base, NaB: sodium benzoate, BA: benzoic acid, TA: tannic acid)………………..55

4.1 Static water contact angles of NaB-entrapped RTV11 films. The (solvent: polymer) ratio and the (water: acetone) ratio were respectively (20: 80) and (50: 50) by mass………………………………………………………………...... 69

4.2 Aggregate size distribution of 1 wt.% of sodium benzoate inside the (99 wt.%) bulk of Sylgard® 184 matrix, for samples prepared at different water/acetone ( W/A) mass ratios. The solvent/polymer ratio was fixed at 20/80 by mass……………………………………………………………………...76

4.3 Aggregate size distribution of 1 wt.% of sodium benzoate inside the (99 wt.%) bulk of Sylgard® 184 matrix, for samples prepared at different solvent/polymer (S/P) mass ratios. The water/acetone ratio was fixed at 50/50 by mass……………………………………………………………………...80

4.4 Aggregate size distribution of sodium benzoate inside the bulk of Sylgard® 184 matrix, for samples prepared at different NaB matrix loading (wt% NaB in the matrix). The water/acetone ratio was fixed at 50/50 by mass. The solvent/polymer ratio was fixed at 20/80 by mass……………………………82

x 4.5 Physical parameters of relevance importance to the miscibility of NaB/PDMS. V and δ are the molar volume and the solubility parameter of the material, respectively. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS*…..84

4.6 Data analysis for the results shown in Figure 4.21. The apparent diffusion ® coefficient (DA) for NaB/Sylgard system was obtained by fitting the experimental leaching data to equation 4.17. The solvent/polymer ratio was 20/80, which was fixed for all samples. (Abbreviation: W: water, A: acetone).……………………132

5.1 Static water contact angles of BA-entrapped Sylgard® coatings compared to that of the controlled BA-free SylgardTM coatings. The (solvent: polymer) ratio was (20: 80) by mass………………………………………………...... 146

5.2 Static water contact angles of BA-entrapped RTV11 coatings compared to that of the controlled BA-free RTV11 coatings. The (solvent: polymer) ratio was (20: 80) by mass……………………………………………………………..146

5.3 Elastic modulus of BA-entrapped Sylgard®184 films. The (solvent: polymer) ratio was (20: 80) by mass………………………………………………………..146

5.4 Elastic modulus of capsaicin-entrapped RTV11 films. The (solvent: polymer) ratio was (20: 80) by mass………………………………………………………..149

5.5 Physical parameters of relevance importance to the miscibility of BA/PDMS. V and δ are the molar volume and the solubility parameter of the material, respectively. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS*…153

5.6 Physical parameters of relevance importance to the miscibility of capsaicin /PDMS. V and δ are the molar volume and the solubility parameter of the material, respectively. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS*………………………………………………………………………156

6.1 Physical parameters of relevance importance to the miscibility of tannic acid/PDMS system. δ is the solubility parameter of the material. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS………………..175

LIST OF FIGURES

Figure Page

1.1 Classical examples for controlled drug-technologies. (a) membrane-reservoir. (b) microencapsulation. (c) monolithic coatings…………………………………….4

3.1 A schematic diagrams for the sample preparation steps used for incorporating Sodium benzoate into the silicone polymer coating……………………………….50

3.2 A simplified sketch for the contact angle concept. (a) the static contact angle: a liquid drop being in equilibrium with a solid substrate in air. γLV, γSV and γSL are respectively the surface energies at the liquid/vapor, solid/vapor, and solid/ liquid interfaces, and θ is the equilibrium (static) contact angle. (b) the advancing contact angle (θa). (c) the receding contact angle (θa)………………....58

3.3 A simplified sketch for the set-up of the JKR apparatus…………………………..62

4.1 Water contact angles of NaB-entrapped Sylgard®184 films (advancing: U, receding: , and static: {). The (solvent: polymer) ratio and the (water: acetone) ratio were respectively (20: 80) and (50: 50) by mass, which were fixed at all concentrations. Error for each data point (average over 12 measurements) is presented by the vertical line………………………………………………………68

4.2 Elastic modulus variations of NaB - entrapped Sylgard® 184 coating. The (solvent: polymer) ratio and the (water: acetone) ratio were respectively (20: 80) and (50: 50) by mass, which were fixed at all concentrations. Error for each data point (average over 6 measurements) is presented by the vertical line………………………………………………………………………...72

4.3 Optical microscope (bright field) images (570 µm x 480 µm) of the bulk morphology of 1 wt % NaB/Sylgard® 184 matrix, prepared at different water/acetone ratios, keeping the solvent/polymer ratio fixed at 20/80. (a) 20/80 water/acetone; (b) 30/70 water/acetone; (c) 50/50 water/acetone; (d) 80/20 water/acetone; (e) 90/10 water/acetone; (f) 100% water. All the values are based on weight………………………………………………………...75

xii 4.4 Guidelines chart for preparing NaB-incorporated Sylgard® 184 coatings at different set of conditions. The saturation line “ — ” represents the maximum solubility of NaB in water (0.555 g NaB per 1 g of water). Above this line, NaB is not soluble in the mixed solvent (water + acetone), and hence the bulk entrapment method will not be feasible at this particular set of conditions. (Abbreviations: S: solvent; P: polymer). All values are in mass basis…………...77

4.5 Optical microscope (bright field) images (570 µm x 480 µm) of the bulk morphology of 1 wt % NaB/Sylgard® 184 matrix, prepared at different solvent/polymer ratios, keeping the water/acetone ratio fixed at 50/50. (a) 10/90 solvent/polymer; (b) 20/80 solvent/polymer; (c) 30/70 solvent/polymer; (d) 40/60 solvent/polymer; (e) 50/50 solvent/polymer. All the values are based on weight…………………………………………………………………………..79

4.6 Optical microscope (bright field) images (570 µm x 480 µm) of the bulk morphology of NaB/ Sylgard® 184 matrix of different NaB concentrations, keeping the solvent/polymer ratio and the water/acetone ratio fixed at 20/80 and 50/50, respectively. (a) 0.5 wt% NaB/ Sylgard® 184; (b) 1 wt% NaB/ Sylgard® 184; (c) 2 wt% NaB/ Sylgard® 184; (d) 3 wt% NaB/ Sylgard® 184; (e) 4 wt% NaB/ Sylgard® 184; (f) 5 wt% NaB/ Sylgard® 184; All the values are based on weight………………………………………………………………..81

4.7 The miscibility trends for NaB/acetone/water/PDMS mixtures, for all possible conditions, as predicted by the original Flory-Huggins (FH) model (equation 4.2). (a) Effect of the water/acetone (W/A) ratio; parameters fixed: 1 wt% NaB/polymer. (b) Effect of the NaB matrix loading; parameters fixed: 50/50 water/acetone. All the values shown in the legends are based on weight……………………………..87

4.8 The miscibility trends for NaB/acetone/water/PDMS mixtures for all possible conditions, as predicted by the new model (equation 4.9). (a) Effect of the water/acetone (W/A) ratio; parameters fixed: 1 wt% NaB/polymer. (b) Effect of the NaB matrix loading; parameters fixed: 50/50 water/acetone. All the values shown in the legends are based on weight………………………………………...94

4.9 The free energy of mixing (∆Gmix/nkT) for NaB/acetone/water/PDMS mixtures prepared at different conditions, as predicted by two models: the Flory-Huggins (F-H) model (equation 4.2), and the new model (equation 4.9). (a) Effect of the water/acetone ratio, parameters fixed: 20/80 solvent/polymer and 1 wt% NaB/polymer. (b) Effect of the solvent/polymer ratio, parameters fixed: 50/50 water/acetone and 1 wt% NaB/polymer. (c) Effect of NaB matrix loading, parameters fixed: 50/50 water/acetone and 20/80 solvent/polymer. The solvent is defined here as water + acetone + NaB. All the values are based on weight. The preparation conditions described here correspond to the actual conditions for the morphology experiments performed in the current study. The insert in each plot represent the corresponding experimental morphology trend ………….96

xiii 4.10 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at different water/acetone ratios, keeping the other conditions unchanged (20/80 solvent/polymer, and 1 wt% NaB/matrix). All the values are based on weight. Solid lines are for showing the trends. (Abbreviations: W: water; A: acetone). Error for each data point (average over 2 batches) is presented by the vertical line………………………………….101

4.11 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at different solvent/polymer ratios, keeping the other conditions unchanged (50/50 water/acetone, and 1 wt% NaB/matrix). All the values are based on weight. (Abbreviations: S: solvent; P: polymer). Error for each data point (average over 2 batches) is presented by the vertical line…...... 103

4.12 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at different wt% NaB/matrix, keeping the other conditions unchanged (50/50 water/acetone, and 20/80 solvent/polymer). All the values are based on weight. Error for each data point (average over 2 batches) is presented by the vertical line…………………………………………104

4.13 Cumulative leaching of NaB from its incorporated silicone coating: Sylgard® 184 (U) or RTV11 (▲). The common solvent used was 50/50 water/acetone by weight. The initial concentration of NaB in both coatings was kept constant at 1 wt%., and the solvent/polymer ratios was kept constant at 20/80 by weight for both combinations………………………………………..106

4.14 Empirical correlations for the leaching rate of NaB from Sylgard® 184. (a) Effect of the aggregate size (d, the arithmetic mean size), parameter fixed: 1 wt% NaB/matrix. (b) Effect of the NaB matrix loading, parameter fixed: d ~ 3 - 4 µm. The insert in (b) is for enlarging the scale of the y-axis. The symbols (■) and (O) represent the initial and the steady leaching rates, respectively….....110

4.15 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at the base case conditions ((1 wt% NaB/Sylgard® 184; 50/50 water/acetone; and 20/80 solvent/polymer)), and the leaching were measured at three different conditions: under constant stirring (□), replacing water daily (∆), and at static conditions (O). Error for each data point (average over 3 batches) is presented by the vertical line…………………………………………113

4.16 A simplified sketch (not to scale) for the mass release of antifouling compounds from polymer paint (water-insoluble matrix). In this case, the compound is soluble in the matrix and is initially loaded in excess of its solubility limit in the matrix. The dissolved zone means that the compound is already absorbed by the polymer phase…………………………………………………………...... 115

xiv 4.17 A simplified sketch (not to scale) for the leaching of antifouling compounds from polymer paint (water-insoluble matrix). In this case, the compound is insoluble in the matrix. (Figure re-drawn from Caprari et al. (1990), with slight modification)………………………………………………………………116

4.18 A simplified sketch (not to scale) of a polymer coating incorporated with AF compound, and immersed in water. The purpose of the sketch is to show the meaning of the axial distance x that was used in equation 4.12………………….117

4.19 Parametric sensitivity analysis for the general mass transfer model (equation 4.19). The dimensionless surface concentration (Ψ|ζ=0 = C/CO |x=0) is plotted against 2 the dimensionless time (τ = DA t / L ), for different values of the dimensionless parameter Bm (Bm = k L / DA). The trends were generated by solving equation 4.19 numerically………………………………………………………………….122

4.20 Simplified sketch (not to scale) for the possible leaching mechanisms of NaB from Sylgard® 184 coating: (a) Perfect packing of the particles; (b) Complete ruptures of the thin membranes; (c) The existence of initial porosity of the matrix, which is mainly composed of constricted narrowed channels. The first column represents the coatings initially before immersion, the second column after some time t1 > 0, and the third column after some time t2 > t1………………..…………………….126

4.21 Fitting of the cumulative leaching data for NaB/Sylgard® 184 coatings to the simplified mass transfer model (equation 4.17). (a) Samples were prepared at different NaB matrix loading, keeping the water/acetone ratio and the solvent/polymer ratio fixed at 50/50 and 20/80, respectively. (b) Samples were prepared at different water/acetone (W/A) ratios, keeping the NaB matrix loading and the solvent/polymer ratio fixed at 1 wt% NaB/matrix and 20/80 ratio, respectively. Points are experimental data and solid lines are linear fitting of the model.…………………………………………………………………………….132

4.22 Optical microscope images of the bacterial attachment study for NaB-Sylgard® 184 coating. (a, c) 1 wt% sodium benzoate-blended Sylgard® 184. (b, d) control Sylgard® 184. The coatings were immersed in water containing Lake Erie bacteria for 2 weeks (a, b) and 4 weeks (c, d). Image size is (285 µm x 215 µm)………………………………………………...... 139

4.23 Antibacterial performance of 1 wt% NaB-incorporated Sylgard® 184 coatings, compared to control SylgardTM 184 samples. The % reduction was defined as [(1-A/B) 100), where A and B refer to the area coverage of NaB-containing coatings and NaB-free coatings, respectively……………………………………140

xv 4.24 Optical microscopic (reflection bright field) images of bacterial attachment on controlled RTV11 coatings after the coatings were immersed in water containing Lake Erie bacteria for 28 days. (a) Half of the coatings surfaces were physically cleaned by scotch tape, and overall pictures [image size: (2850 x 2150) µm] were taken showing the cleaned area (right side of picture a) and the un-cleaned area (left side of picture a). Pictures (b) and (c) are the magnifications [image size: (285 x 215) µm] of the two area indicated in picture (a).………………………………..141

5.1 Water contact angles of capsaicin-entrapped RTV11 films (advancing: U, receding: , and static: {). The (solvent: polymer) ratio was (20: 80) by mass, which was fixed at all concentrations. Error for each data point (average over 12 measurements) is presented by the vertical line………………………………148

5.2 Optical microscope (bright field) images of resulting BA distribution in the bulk of Sylgard® 184 matrix when different solvent were used to mix BA with Sylgard® 184: (a) toluene, (b) acetone, (c) acetonitrile, and (d) ether. The concentration of BA in the matrix was fixed at 1 wt%, and the (solvent: polymer) ratio was fixed at (20: 80) by mass. The image size is 2850 µm x 2400 µm for (a), and 1140 µm x 960 µm for (b)-(d)………………… ……………………151

5.3 Optical microscope (transmission bright field) image of resulting capsaicin distribution in the bulk of Sylgard® 184 base material. Toluene was used as the common solvent. The concentration of capsaicin in the matrix was 1 wt%, and the (solvent: polymer) ratio was (20: 80) by mass. The image size is (570 x 480) µm…………………………………………………………………...155

5.4 Cumulative leaching of BA from its incorporated silicone coating: Sylgard® 184 (open symbols) or RTV11 (filled symbols). The common solvent used for BA/silicones were acetone (squares) and Toluene (circle). The initial concentration of BA in all coatings was kept constant at 1 wt%., and the solvent/polymer ratios were kept constant at 20/80 by weight for all combinations………………….….158

5.5 Capsaicin cumulative mass per area (Q, in µg/cm2) released from 1 wt % capsaicin-incorporated RTV11 sheet (total mass and surface area of the sheet were respectively 6.55 g and 114 cm2), plotted against time of immersion in DI water. Toluene was used as the common solvent to mix capsaicin with RTV11, and the (solvent: polymer) ratio was (20: 80) by mass……………………...…...160

5.6 Effect of the mixing order on the capsaicin cumulative leaching from 1 wt % capsaicin-incorporated RTV11 sheet (total mass and surface area of the sheet were respectively 1.64 g and 38 cm2). Ethanol was used as the common solvent, and the (solvent: polymer) ratio was kept constant at (20: 80) by mass. The open squares correspond to the conditions of mixing capsaicin/ethanol solution with the RTV11 base and drying off the solvent before adding the catalyst. The filled squares correspond to the same conditions of the open square data, except that the capsaicin/ethanol solution was mixed after adding the catalyst………………….162 xvi 5.7 Effect of water immersion time on the wettability of RTV11 films in terms of the static contact angles taken for RTV11 immersed in Lake Erie (Δ) and deionized water (▲). The advancing (□) and receding () contact angles taken for RTV11 immersed in deionized water are also presented. Error for each data point (average over 12 measurements) is presented by the vertical line………………………....165

5.8 Surface topographic images of Capsaicn-RTV11 coatings. (a) as-prepared controlled RTV11 film, surface roughness: 6.7 nm; (b) as-prepared RTV11 film containing 1 wt % capsaicin, surface roughness: 12.3 nm; (c) controlled RTV11 film after 14 days of immersion in DI water, surface roughness: 10.5 nm; (d) RTV11 film containing 1 wt% capsaicin after 14 days of immersion in DI water, surface roughness: 88.0 nm. The images (scan size: 80 µm x 80 µm; z-scale: 400 nm) were generated using scanning probe microscopy with the non-contact mode at a scan rate of 0.20 Hz……………………………………………………………166

5.9 Effect of water type and immersion time on the elastic modulus of RTV11 films immersed in different types of water samples (sterilized Lake Erie water: Δ, enriched Lake Erie water: ○, and sterilized Lake Erie water with 20 ppm capsaicin: □). Error for each data point (average over 6 measurements) is presented by the vertical line……………………………………………………………………….168

5.10 Optical microscopic (reflection bright field) images of bacterial attachment for capsaicn-RTV11 coatings. (a) 1 wt% Capsaicin/RTV11, (b) control RTV11. The coatings were immersed in water containing Lake Erie bacteria for 14 days. The size for both images is (285 x 215) µm…………………………...... 170

6.1 Static water contact angles of TA-entrapped silicone films compared to that of the controlled TA-free silicones. The concentration of TA in the matrix was fixed at 1 wt% for both combinations. The (solvent: polymer) ratio was fixed at (20: 80) by mass (acetone was the solvent for both combinations). Error for each data point (average over 12 measurements) is presented by the vertical line…………...... 172

6.2 Optical microscope (transmission bright field) image of resulting TA distribution in the bulk of Sylgard® 184 matrix: (a) 1 wt% TA/Polymer; (b) 4 wt% TA/Polymer. Acetone was used as the common solvent, and the solvent/polymer ratio was 20/80 by mass. The image size is (570 x 480) µm……………………174

6.3 Cumulative leaching of TA from its incorporated silicone coating: Sylgard® 184 (∆) or RTV11 (▲). The common solvent used was acetone for both combinations. The initial concentration of TA in both coatings was kept constant at 1 wt%., and the solvent/polymer ratio was kept constant at 20/80 by weight for both combinations. Error for each data point (average over 2 batches) is presented by the vertical line……………………………………………………..177

xvii

CHAPTER I

INTRODUCTION

1.1 Introduction

Marine biofouling is the accumulation and growth of marine organisms on manmade structures immersed in natural water (Evans & Clarkson, 1993; Davis &

Williamson, 1995). Biofouling has resulted in many problems for the marine industry

(Abdul Azis et. al., 2001). The layer of attached organisms on ship’ hulls decreases ship speed and increases fuel consumption by approximately 30 % (Stupak et. al., 2003). As an example, the US Navy estimates that biofouling costs over $150 million annually in excess fuel consumption and cleaning costs for naval vessels. Biofouling is also directly related to the bio-corrosion and other related problems (Malik et. al., 1996; Melo & Bott,

1997; Wood and Marsh, 1999).

Four steps are involved in biofouling formation (Al-Ahmed et. al., 2000;

Steinberg et. al., 1997). Once a surface is immersed in water, it is immediately covered with a thin “conditioning film” consisting mainly of different proteins and dissolved organic matters. It is followed by the adhesion of single floating bacteria. Once

1 firmly attached, bacteria start to generate extracellular polymeric substances (EPS),

which connect the cells in between each others as well as connecting to the surface. The

cells plus EPS is what we called the “biofilm”. Finally, the biomass continues to grow, with some large macro fouling organisms attach to, while other cells detach from the film. The first step, conditioning films adsorption, is very fast and occurs within hours of immersion, whereas the later step, production of EPS, is very slow and needs weeks or perhaps months to complete.

Antifouling paints have long been the most effective method to prevent biofouling, where biocides or heavy metal compounds such as TBTO (Tributyltin oxide) are released from the coatings and inhibit organism’s attachment. TBT compounds are the most effective compounds for biofouling prevention. Unfortunately, they are also the most toxic compounds against non-target marine organisms (Axiak et. al., 2000;

Haslbeck et. al., 1996). As a result, the International Maritime Organization (IMO) banned the application of TBT compounds on 2003, and the entire removal of TBT coatings by year 2008 will be required worldwide (Champ, 2000).

Recognizing the harmful effects and the already-started bans for TBT compounds, considerable efforts are being committed worldwide to find new non-toxic or less-toxic antifouling alternatives. Possibilities which have been considered include dissolution of adhesive substance by various enzymes, biogenic agents, smart polymer coatings, polymer coatings with defined surface-microstructures (Jelvestam et. al, 2003), foul- release coatings (Brady, 1999, 2000), natural product antifoulants (NPAs) (Hellio et. al.,

2 2001, Ponasik et. al. 1998), and less-toxic and commercially available preservatives, pesticides, drugs, and insecticides.

Upon the use of the less toxic antifouling compounds, the selection of a less toxic antifouling compound is the first step toward biofouling control. The next important step is to control its release from the coating. Control-release technologies that were originally developed for drug-release are expected to be useful for controlling the release of antifouling compounds. Three drug-release technologies are widely used: membrane- reservoir, microencapsulation, and monolithic coatings (Figure 1.1). In a membrane- reservoir device, the active compound is highly concentrated into a middle layer of the polymer coating, which is surrounded by a second layer of another polymer coating. In microencapsulation, the active compound is encapsulated into microcapsules, which are distributed in a polymer coating made of a material differed from that of the capsules. In a monolithic coating, the active compound is distributed within the layer of the polymer coating. Depending on the solubility and loading of the compound in the monolithic coating, the compound can be molecularly dissolved in the polymer phase or it can be uniformly dispersed as a separate droplet phase (as shown in Figure 1.1c). The monolithic coating technology is the most widely used due to its simplicity. Several experimental approaches exist to prepare monolithic coatings; the solvent casting technique and the compressing molding technique (Fan and Singh, 1989) are two common ones. In the current study, monolithic coatings prepared using a solvent-assisted blending technique will be focused.

(a)

(b)

(c)

Figure 1.1 Classical examples for controlled drug-technologies. (a) membrane- reservoir. (b) microencapsulation. (c) monolithic coatings.

4 1.2 Importance and scope of the study

Many non-toxic or less-toxic antifouling (AF) compounds have been identified in

the literature. Some of these compounds are directly blended with different polymer

coatings and tested for biofouling prevention. Although direct blending is a simple

method, in most cases, the release rate of AF compound will be very fast and hence the

service life of the coating will be very short. Different methods have been suggested in

literature to extend the service life of the coating, either by chemically attaching the

compound to the coating surface or by employing sophisticated controlled release

techniques such as the micro-encapsulation method or the membrane reservoir method.

However, these methods are highly complicated and, in some cases, are challenging for

certain compounds. In the current study, we are proposing a simple experimental

approach (dissolving the compound in a solvent or in a blend of solvents and then

homogenizing the solution with the polymer) to generate monolithic coatings that are

expected to result in the desired slow-controlled release of the AF compound. This

experimental approach is termed as the “solvent-assisted blending technique”. For the

current study, a systematic combination of the compound/solvents/polymer will be

carried out, and a systematic evaluation of the miscibility and leaching behavior of the

compound from the coatings will be conducted. The main focus of the current study is to obtain the relationship between the miscibility of the compounds in the coating matrix and their leaching in water, for the purpose of controlling the release rate and maintaining the antifouling performance of the coatings for their long term service.

5 1.3 Objectives

The primary objective is to obtain the relationship between the miscibility of less- toxic antifouling compounds in a monolithic, hydrophobic polymer coating and their release in water. A secondary simultaneous objective is to evaluate the possibility of applying the solvent-assisted blending technique as a simple preparation method for the purpose of controlling the release of four antifouling compounds. Specifically, the study aims to:

(1) Evaluate the feasibility of the solvent-assisted blending technique for

incorporating four less-toxic antifouling (AF) compounds (sodium benzoate,

benzoic acid, capsaicin, and tannic acid) into two silicone matrix coatings

(Sylgard® 184, and RTV11) by examining experimentally the morphological

structure of the compounds in the matrix and their leaching into Deionized (DI)

water.

(2) One of the combinations (sodium benzoate in Sylgard® 184) is selected as the

model system to perform more detailed investigations on the miscibility-leaching

relationship, as listed below:

(a) Compare, experimentally, the role of varying the preparation conditions

on the miscibility and the leaching of the compound, and obtain

correlations between the miscibility and the release rate.

(b) Perform thermodynamic analysis on the miscibility of various coating

combinations to explain the effect of the preparation conditions on the

distribution and morphology of the compound inside the coating matrix.

6 (c) Perform a mass transfer modeling on the leaching to explain the

experimental data and propose a suitable leaching mechanism of the

monolithic sodium benzoate/silicone coating generated using the solvent-

assisted blending method.

1.4 Dissertation outline

The outline of the dissertation is as follow. In chapter 2, the dissertation starts with providing backgrounds and literature review about various issues of relevance to the current study, including biofouling and its control, foul-release coatings and less toxic antifoulants, antifoulant-coating miscibility, and antifoulant leaching modeling. The experimental approach is described in chapter 3. The results and discussion for sodium benzoate-based coatings are presented in chapter 4. The results and discussion for benzoic acid and capsaicin-based coatings are presented in chapter 5, and the results and discussion for tannic acid-based coatings are presented in chapter 6. Finally, in chapter 7, the dissertation concludes with summarizing the key results and providing recommendations for future works.

CHAPTER II

LITERATURE REVIEW

This chapter provides the necessary background and literature review about the

subject of the current study. In section 2.1, background about biofouling and biofilm

formation is provided. Current and proposed solutions to combat biofouling are reviewed

in section 2.2, including conventional toxic antifouling coatings, nontoxic foul-release

coatings, and less/non toxic antifouling compounds. The miscibility of additives and

polymeric matrices is reviewed in section 2.3. The mechanisms and mathematical

models describing the release of the active compounds (i.e. antifoulants) from the matrix into water are reviewed in section 2.4.

2.1 Biofouling and biofilms formation

Four steps are involved in biofouling (Al-Ahmed et. al., 2000; Steinberg et. al.,

1997). Once a surface is immersed in water, it is immediately covered with a thin

“conditioning film” consisting mainly of different proteins and dissolved organic matters.

Then, it is followed by the adhesion of single floating bacteria. Once firmly attached, bacteria start to generate extracellular polymeric substances (EPS), which connect the cells in between each others as well as connecting to the surface. The

8 cells and EPS form the “biofilm”. The biomass continues to grow, and in the meantime

some large macrofoulers attach while some adhered cells detach from the film (due to

shear force of the bulk phase, or due to partial cohesive failure of the biofilms). The first

step, conditioning films adsorption occurs within hours of immersion, whereas the later step, production of EPS needs weeks or maybe months to finish. The conditioning step

cannot be avoided for any kind of surface immersed in the natural water. Therefore, an effective strategy to minimize biofilm formation and the subsequent macro-foulers attachment is to prevent or reduce bacterial adhesion.

The attachment of bacteria to solid surfaces is a complicated process with many variables contributing to the final result. These variables can be subdivided into properties of the substrates (e.g.: surface hydrophobicity and roughness), properties of the bacterium, and properties of the liquid media (e.g.: flow velocity and temperature). A concise review on the effects of these variables on bacterial adhesion was written by An and Friedman (1998).

Marshall and his co-worker (Marshall et al. 1971) are the first to describe that bacterial adhesion to solid surfaces as a two phase process: an initial, instantaneous, and reversible physical phase (phase 1), and a time-dependent and irreversible molecular and cellular phase (phase 2). This classification is largely accepted by the majority of researchers. In the first phase of adhesion, floating bacteria move to a material surface by means of physical forces, such as Brownian motion, van der Waals attraction forces, gravitational forces, electrostatic forces, and hydrophobic interactions. These physical

9 forces are further classified into long-range and short-range interactions (Dankert et al.

1986). The long-range interactions are nonspecific interactions, which are effective when

the distance between the bacterium and the substrate is greater than 150 nm. Short-range

interactions become effective when the distance between the bacterium and the substrate

is less than 5 nm. These short-range interactions include chemical bonds (such as

hydrogen bonding), ionic and dipole interactions, and hydrophobic interactions (An and

Friedman, 1998; Gottenbos et al., 2002). The short-interactions forces from phase 1

provide the suitable ground for the second phase of adhesion to occur, where here

molecular reactions between specific bacterial surface structure and substratum surface

become predominant. The active sites of the bacterium surface that are responsible for

firm adhesions are those polymeric surface structures that contain adhesins as parts of their structures, which include capsules, fimbriae (or pili), and fibrillae. At the irreversible stage of adhesion, bacteria start to produce slime extracellular polymeric substances (EPS), which are exopolymers composed of mainly polysaccharides. EPS can be regarded as external adhesives that connect the bacteria between each others as well as

connecting the bacteria to the substrate.

2.2 Biofouling controls

Biofouling has caused serious problems to the maritime industry including

enormous economic loss due to considerable increase in ship’s fuel consumption and

maintenance costs, and damages and harmful effects resulted from biocorrosion of the

immersed infrastructures. Consequently, the prevention or minimization of biofouling is

10 a necessity. This section highlights the current and future trends of biofouling controls

technologies. Existing biofouling controls technologies are reviewed in subsection 2.2.1,

together with their main drawback of being toxic to the environment. Subsequently, two

environmentally-friendly alternatives are reviewed: foul-release coatings (section 2.2.2), and less/non toxic antifoulants (section 2.2.3)

2.2.1 Conventional toxic antifouling coatings

Antifouling paints have long been the most effective method to prevent biofouling, where biocides or heavy metal compounds such as TBTO (Tributyltin oxide) are released from the coatings and inhibit microorganism’s attachment. The mechanism of action for these compounds is simple and straightforward: the compound is released from the coating and just kills the attached organisms. TBT compounds are found to be the most effective compounds for biofouling prevention. Unfortunately, they are also the most toxic compounds against non-target marine organisms (Silverman and Aubart,

2006, Abdul Azis et al., 2003). Also, these compounds are not biodegradable in water, thus they are accumulated in water and possess a serous environmental hazard. As a result, the International Maritime Organization (IMO) banned the application of TBT compounds on 2003, and the entire removal of TBT coatings by year 2008 will be required worldwide (Yebra et al., 2004; Silverman and Aubart, 2006).

In the marine industry, the antifouling compounds are being incorporated into two main classes of marine coatings: insoluble and self-polishing coatings (Omae, 2003;

11 Yebra et al., 2004). For the former, the compound is incorporated into a polymer matrix

that is water-insoluble and does not erode considerably over time in water. For the later,

the matrix erodes slowly over time in water. Both classes of matrices are used for

incorporating TBT compounds, and it is found that the TBT-self polishing coatings are having better long-term antifouling performance than that of the TBT-insoluble coatings, mainly due to the fact that TBT-self polishing coatings are having a slower and controlled mass release. However, due to the extreme toxicity of TBT compounds, the ban for using TBT compounds includes not only the TBT-insoluble coatings but also the TBT- self polishing coatings.

2.2.2 Silicone foul-release coatings

Silicones belong to the water-insoluble matrices class, and have unique properties that make them distinguished as foul-release (FR) coatings (Brady & Singer, 2000;

Estarlich et. al., 2000; Watermann et. al., 1997; Wynne et. al., 2000). They consist

mostly of Polydimethylsiloxane (PDMS), which has methyl (–CH3) side chains results in

its low surface energy ( 20 - 24 mJ/m2) and a flexible, inorganic -Si-O backbone linkage

leads to its extremely low elastic modulus (~ 1 MPa), both are essential to the extreme

low adhesion of silicone coatings. Thus, biofilms can be easily removed from the surface

by simple mechanical cleaning or during vessel movement. Marine coatings industries

have recently shown much interest in silicone coatings as a non-toxic alternative. In

general, two classes of PDMS are available commercially with distinct curing chemistry.

The first one is the hydrosilation cured PDMS, such as Sylgard® 184. The second class is

12 the condensation cured PDMS, such as RTV11. This subsection reviews some of the

recent works done for investigating silicone coatings as a nontoxic alternative for marine

biofouling control.

Wynne et al. (2000) evaluated two types of PDMS coatings (RTV11, and an in- house unfilled hydrosilation cured PDMS). They evaluated the surface hydrophobicity and roughness of the coatings, and the mass loss of the coatings upon immersion in water. They found that the hydrosilation cured PDMS was fairly stable in water; with the mass loss was less than 0.2 % after 50 days of immersion. RTV11, on the other hand, was not stable in water, with a mass loss of about 0.8 % after 50 days of immersion.

However, the bulk moduli of the two coatings in water were not evaluated. The two coatings were also evaluated for barnacle adhesion. The barnacle adhesion strength to the hydrosilation cured PDMS and to the RTV11 coatings were found to be about 0.5 and

0.78 kg/cm2, respectively. In another publication of the same research group (Bullock et

al., 1999), the surface properties of RTV11 (with different catalyst concentration) were

investigated in more details, and a mechanism for the mass loss of RTV11 in water was

proposed.

Arce et al. (2003) performed a comparative study for the microelastic properties

of RTV11 and IntersleeskTM elastomers (IntersleekTM is the trade name of a PDMS-based coating, which is already of being used as a foul release coating in some vessels). The measurements were done by using AFM and other techniques, which gave valuable information about the local structural and mechanical properties of the as-prepared

13 coatings. However, the behavior of the coatings in water and their antifouling

performance were not evaluated.

The nature and failure mechanism of bioadhesive bonding between barnacles

and two types of PDMS coatings (RTV11 and RTV1556) were investigated by Berglin

and Gatenholm (1999). Analysis of the fracture surfaces indicated that the failure

mechanism was a cohesive failure within the PDMS coatings. As a control, they

compared their results with PMMA coatings, and they found that the failure mechanism of PMMA coatings was more complex than the failure mechanism of PDMS coatings.

Also, the surface energies of the coatings were calculated from contact angles data, and the values were 23.3 mJ/m2 and 22.4 mJ/m2 for RTV11 and RTV1556 coatings,

respectively.

Edwards et al. (1994) evaluated the hydrophobicity and antifouling performance

of some room-temperature-vulcanizing PDMS and polydimethyldiphenylsiloxane

(PDMDPS) coatings. The authors selected these two classes of silicones to relate their

hydrophobicity to their antifouling performances. PDMDPS showed higher

hydrophobicity (contact angle ~ 123) and better antifouling performance than PDMS

(contact angle ~ 112). Incorporation of silicone oils into the coatings was also

investigated in the range of 0 — 20 %. It was concluded that silicone oil had only a

significant positive effect for enhancing the antifouling performance if its concentration

was sufficiently high (between 10 to 20 %), and the enhancement here was speculated to

be by the formation of a surface film.

14 Estarllich et al. (2000) studied the change in surface properties of some PDMS

coatings (RTV11, RTV160, and RTV655) and flourosilicones as a function of immersion

time in different types of water. Bacteria and microalgae attachment tests were also

performed, where it was found that the early attachment was least on RTV11 and greatest on flourosilicones.

Barnacle release mechanism for two silicone coatings (a single layer coating consists of Sylgard® 184, and a duplex coatings consist of RTV11 as a top coat and Silgan® J-501 as the bond coat) were studied by Singer et al. (2000). The results suggested that the coatings with lower modulus and thicker thickness had a better foul release performance.

This was confirmed by the theoretical fracture-mechanical analysis of Brady and co- authors (Brady and Singer, 2000; Brady, 2001).

The two classes of foul release coatings (fluorinated and silicone coatings) have shown to be partially effective and both have advantages and limitations. Brady and

Aronson (2003) synthesized a new foul release coating that combined the best features of the two classes of materials. After optimizing the preparation conditions, the new material (an elastomeric fluorinated polyurethane coating) was found to be effective as foul release coating with the desired surface and mechanical properties. It should be noticed that this new product is a polyurethane-based coating, not a silicone-based coating.

Stein et al. (2002) evaluated systematically model silicone coatings (both condensation-cured and hydrosilylation-cured types) with controlled molecular

15 architectures to determine the effect of compositional variables such as filler loading and

cross-linking density on Pseudobarnacle attachment strengths. The results suggested that there was a trade-off between mechanical integrity of the coatings and foul release performance.

2.2.3 Less-toxic antifoulants

Scientists have searched for non-toxic or less toxic antifoulants, to replace the harmful toxic antifoulants such as TBT compounds. Less-toxic antifoulants include

Natural Products Antifoulants (NPAs), food-preservatives, antibacterial drugs, pesticides, insecticides, and other related compounds.

NPAs are natural compounds extracted from plants and marine species. The discovery of the potential of NPAs came from the observation that the biomass of foulers is usually higher on non-toxic artificial substrates than on the surface of living objects.

Natural defenses against biofouling involve secondary organic exometabolites: phenolic or polyphenolic and halogenorgnic compounds, terpenes, heterocyclic compounds, and other compounds (Railkin, 2004). These compounds are easily bio-degradable, thus making the antifouling process ecologically safe. More than 100 potential NPAs have been identified nowadays (Clare, 1996; Fustian, 2004). However, most of them have not yet been systematically incorporated into suitable coatings and evaluated. In addition, there is a substantial lack of information on the mechanisms of actions on how these

NPAs work against marine fouling species.

16 The structure of any NPA is usually complex with many functional groups, and usually difficult to be commercialized. Efforts are being made to examine simplified structures of a lot of NPAs for antifouling activity and then synthesizing them. The resulted synthesized compound, no longer now called NPAs, could have potentials as antifoulants, antibacterial drugs, and food-preservatives. Based on this strategy, different analogues compounds have been identified, synthesized, and tested for antifouling activity. The synthesized compounds with proved antifouling activity and much less toxicity (compared to heavy metal compounds) include benzoic acid, nicotinic acid, picolinic acid, 2-furyl-n-pentyl ketone, 3-acetyl-2,5-dimethyl furan, and their derivatives

(Stupak et. al., 2003; Clare, 1996; Sundberg et. al., 1997; Railkin, 2004). Again, most evaluations are concerned with the biological side, and a detailed study on the compatibility, miscibility, leaching and material properties of the antifoulant/polymer systems is still lacking.

Understanding the mechanism of action for the antifouling compounds is necessary. For toxic compounds such as TBT and heavy metals, the mechanism of action is acute toxicity, which results simply in killing the attached microorganisms. For less or nontoxic compounds, the identification and experimental validation of the mode of action is rather complicated (Railkin, 2004). In general, two antifouling mechanisms are suggested for the effectiveness of the non-toxic compounds: repellency and chemical anti-adhesive (Sundberg et. al., 1997, Railkin, 2004). As emphasized by Railkin (2004), the following exact definition should be applied to assign the non-toxic repellency mechanism for a certain antifouling compound: “Repellents are cues inducing a negative

17 motor response, taxis, or kinesis in organisms at a certain stage of development, which

causes them to move away from the source of these cues” (Railkin, 2004). This precise

definition differentiates the true repellent property (a non-toxic mode of action) from other toxic properties of the antifouling material. For this definition to be applied, special experimental behavioral tests should be performed. For example, the repellent effect can be evaluated using the special test developed by Railkin (1995), where here chemotactic chambers made of Plexiglas and measuring 36 x 40 x 80 mm can be used, with each chamber consists of 3 sections, separated by 0.92-micro nucleopore filters. The microorganism is placed in the middle section; the antifoulant solution is filled in one side and a reference solution (e.g. sea water) is filled in the other side. However, in most studies, the true repellent function of various natural and synthetic antifoulants is hypothetically assumed rather than experimentally verified (Railkin, 2004). Among a few studies that followed the standard behavioral tests, benzoic acid and tannic acid were proved to have a true repellent mechanism as a non-toxic mode of action. (Mitchell and

Kirchman, 1984; Railkin et. al., 1993; Railkin, 1995, Railkin and Dobretsov, 1994). In addition to the repellency mechanism, the non-toxic antifouling compounds can also exhibit the chemical anti-adhesive mechanism, where molecules of the antifouling compound exist freely in water and act as catalytic inhibitors for the biochemical reactions involved in cell adhesion. The word “chemical” is used here to distinguish this mechanism from the physical anti-adhesive mechanism, which usually refers to the easy- release silicone coatings. The chemical anti-adhesive mechanism, a terminology used by

Railkin (2004), appears to be the same as described by Sundberg and his co-workers

(1997) but in different terminology: “the blocking of the attachment sites mechanism”,

18 where Sundberg et al. (1997) hypothesized that the molecules of the compound exist freely in water could bind to the attachment sites on the microorganism cell wall and thus prevent cell adhesion. Again, special experimental behavioral tests should be performed in order to confirm the chemical anti-adhesive mechanism, such as the standard tests developed by Ina et. al. (1989) or Hellio et. al. (2000). Among a few studies that applied these controlled tests, benzoic acid was also proved to have chemical anti-adhesive properties (Railkin et. al., 1993; Railkin, 1995, Railkin and Dobretsov, 1994).

Among the environmentally benign NPAs, capsaicin, a stable monoclinic crystalline alkaloid extracted from chili peppers, is being considered as a potential.

Capsaicin, or 8-methyl-N-vanillyl-6-nonenamide, and its analogues have been used as active ingredients in medicinal products, and are found to be effective in inhibiting bacterial growth (Jones et. al, 1997; Tsuchiya, 2001; Cichewicz & Thorpe, 1996; Molina-

Torres et. al., 1999). The studies done indicate that both capsaicin and its structural analogs, all containing the phenolic ring and the amide group but with different hydrocarbon tails, have similar inhibitory capability towards bacteria, suggesting that the phenolic ring and the amide group are responsible for the antibacterial activity of capsaicin. It was found that the presence of capsaicin in an aqueous solution reduced the bacterial attachment on a coating (Xu et al., 2005). In addition, capsaicin is significantly less toxic (EC50, the concentration to kill 50% of bacterial population, is ~ 5 to 20 ppm towards various bacteria [Xu et al., 2005]) as compared to organotin compounds (EC50 <

0.01 ppb). Furthermore, capsaicin is biodegradable in the marine environment, and approved by the EPA as an active ingredient for insect, bug and animal repellents.

19 Therefore, it is an attractive candidate to be utilized for generating environmentally benign coating alternatives. For the past 10 years, there exist a lot of patents that document the potential of using capsaicin-based marine antifouling coatings (e.g. Yao et. al., 2004; Hatch, 2003; Veech, 1997; Watts, 1995). Based on our knowledge, only two journal papers (Shi et. al., 2004; Yu et. al. 2004) exist that report the incorporation of capsaicin into polymer coatings for inhibiting biofilm formation, where a good antifouling performance was observed after the exposure to marine environment.

However, based on our knowledge, the incorporation of capsaicin into a foul-release coating has not been published or patented.

Benzoic acid and its sodium salt (sodium benzoate) are most common safe food preservatives and antimicrobial agents. Benzoic acid and sodium benzoate are classified in the United State as Generally Recognized as Safe (GRAS) and their use in food is permitted up to the maximum level of 0.1 % (Sagoo et. al. 2002). From Microtox studies, the toxicity, in term of the concentration that kills 50% of bacterial population or

EC50, of benzoic acid and sodium benzoate are determined to be ~ 7 ppm and ~ 560 ppm

(Haque et al., 2005), respectively, which are three to five orders of magnitude lower than

that of organotin compounds. Recognizing also the commercial availability and

cheapness of benzoic acid and sodium benzoate, they are attractive candidates to be

incorporated into coatings as environmentally benign alternatives.

Previous works (Lueck, 1980) for evaluating the antibacterial behavior of benzoic

acid have concluded that, at different points of the citric acid cycle, benzoic acid

20 deactivates the enzymes that control acetic acid metabolism and oxidative phosphorilation in yeast and bacteria. In a recent review on antibacterial mechanisms

(Chapman, 2003), benzoic acid is believed to act by interfering with the ability of the cell membrane to maintain a suitable pH level, which consequently leads to acidification of the cell interior and widespread disruption of the metabolism process (Eklund, 1985). As mentioned before (Railkin, 2004), the mode of action of benzoic acid has been identified experimentally by standard behavioral tests, where benzoic acid is proved to exhibit both non-toxic modes of actions (repellency and chemical anti-adhesive). This finding highlights the possibility of benzoic acid to be effective against broad spectra of micro and macro fouling species. In addition, field studies had shown that benzoic acid when added to vinyl-rosin coating was effective in inhibiting different species of both micro and macro foulers (Railkin, 1995), making it a very attractive non-toxic or less toxic antifouling candidate. However, due to the extremely fast leaching of the compound, the effectiveness of the coating with benzoic acid incorporated was only for a short period of time (1 month). Therefore, for further utilization of benzoic acid, understanding the causes of the fast leaching is essential and seeking techniques to incorporate benzoic acid into a coating to control its leaching is also important.

The salt form of benzoic acid, such as sodium benzoate (NaB) is even more environmentally benign as compared to benzoic acid. Static biological assays have also proved that NaB showed a narcotic (non-toxic) effect on the investigated microorganisms

(Vetere et. al., 1999). In addition to sodium benzoate, other different benzoic acid salts

(calcium benzoate and aluminum benzoate) have also been tested by biological assays,

21 and antifouling effectiveness similar to NaB were observed (Vetere et. al., 1999). The

mode of action for sodium benzoate is not precisely identified as for the case of benzoic

acid, but the above study could indicate that the benzoate anion freely floating in water is

important for the antifouling of NaB, and that the antifouling mechanism of NaB might

be similar in some aspect as that of benzoic acid (Vetere et. al., 1999, Stupak et. al.,

2003). Nevertheless, field studies had shown that sodium benzoate when incorporated

into a marine paint was successful for inhibiting marine microorganism attachments

(Stupak et. al., 2003). However, based on our knowledge, the incorporation of sodium

benzoate or benzoic acid into a foul-release coating has not been published or patented.

Tannic acid or tannins are naturally accruing polyphenolic compounds of high

molecular weight in the range of 500 to 3000 g/mol. They are important in industry,

food, and environmental science (Ho, 1992). The anticorrosive properties of tannins

were known at least 50 years ago, and several tannin-based products exist in the market

for corrosion treatment application. The antifouling properties of tannic acid and its

derivatives – when exist in solution or entrapped into a coating - were also reported in

several studies (e.g. Perez et al., 2006; Stupak et. al. 2003; Lau and Qian, 2000).

However, up to our knowledge, the incorporation of tannic acid or its derivatives in

silicone coatings has not been reported. In addition, tannic acid is significantly less toxic

than heavy metals-based compounds [the EC50 of tannic acid is about 118 ppm (Xu et al.,

2005)]. As mentioned before, Railkin (2004) emphasized that tannic acid was one of a few compounds that were verified by standard experimental behavioral tests to have true repellent mode of action.

22 2.3 Antifoulant-matrix miscibility

The miscibility of the antifouling compound with the polymer matrix plays an

important role for its distribution within the polymer coating, and consequently on its

leaching into water. In this section, the thermodynamics of mixing between the compound and the polymer matrix is reviewed. One fundamental thermodynamic function that describes the miscibility of a multicomponent system is the free energy of mixing (∆Gmix). As thermodynamics states, the systems becomes more heterogeneous as

∆Gmix increases, and becomes spontaneously homogeneous when ∆Gmix is less than zero.

The challenge is to find or derive the suitable and accurate thermodynamic model to

calculate ∆Gmix for a particular polymer system. The incorporation of an antifouling

compound into a polymer matrix using a solvent or blends of solvents will result in

having a ternary or quaternary system. As a result, interactions between every two

components in the system exist, which might be important. Thus, thermodynamic

information about such a system is required, since the available and widely used binary

models may not be adequate. In the following paragraphs, a literature survey for the

thermodynamic modeling of the miscibility of multicomponent system is presented.

Boom et. al. (1994) applied the Flory-Huggins equation for binary system and

extended it for their quaternary system (solvent + nonsolvent + Polymer base +

Macromolecular additive) where the nonsolvent is nonsolvent for the polymer base but it

is a good solvent for the additive. Their starting point for the derivation is based on the

following equation, which is an application of the Flory-Huggins theory:

23 ∆Gmix / RT = n1 ln Φ1 + n2 ln Φ2 + n3 ln Φ3 + n4 ln Φ4

+ n1 Φ2 χ12 + n1 Φ3 χ13 + n1 Φ4 χ14 + n2 Φ3 χ23

+ n2 Φ4 χ24 + n3 Φ4 χ34 (2.1)

where R and T are respectively the ideal gas constant and the absolute temperature, ni and

Φi are respectively the number of moles and the volume fraction of component i, and χij is

the binary interaction parameter for the component pair i-j. However, in their

calculations, they did not measure or calculate the χij parameters. Instead, they assumed

some constant values for the χij parameters (either 1.0, 1.5, 0.5, or -1.0).

Venkataraman and Hajra (1999) developed a complicated model to describe the solid-liquid equilibrium for their quaternary system. They derived the excess free energy function for the liquid mixture in terms of 37 parameters pertaining to six of the constituent binaries, four ternaries and the quaternary interactions of the system. Their

model is basically an extension of the four-parameter model available for binary system.

The four-parameter model for binary system is:

E ∆G / RT = y1 y2 [ a1 y1 + a2 y2 + y1 y2 ( a3 y1 + a4 y2)] (2.2)

E where ∆G is the excess free energy and a1 ─ a4 are the binary constants. Consequently, the partial excess property for component 1 and 2 are:

E E RT lnγ1 = ∆G - y2 (∂∆G / ∂ y2) (2.3)

24 E E RT lnγ2 = ∆G + (1-y2 ) (∂∆G / ∂ y2) (2.4)

where γ1 and γ2 are the activity coefficients of component 1 and 2, respectively.

Gander et. al. (1995) proposed a different approach for predicting the miscibility

between the polymer and the incorporated compound. Experimentally, they incorporated

a macromolecular additive into PLLA polymer using different solvents, where they first

dissolved the additive into water and then they mixed the aqueous solution with the

polymer/solvent mixture. Theoretically, they hypothesized that the maximum entrapment quality (i.e. maximum additive/polymer miscibility) would be obtained if:

| ∆E1 + ∆E2 | = minimum (2.5)

where ∆E1 is the interaction energy between the solvent (S) and the aqueous

additive/water solution (B), and ∆E2 is the interaction energy between the solvent (S) and the polymer (P), and given by:

∆E1 (J/mol) = ─ 2 VS (δd,S δd,B + δp,S δp,B) ─ (ES EB + CS CB ) (2.6)

∆E2 (J/mol) = ─ 2 VS (δd,S δd,P + δp,S δp,P) ─ (ES EP + CS CP ) (2.7)

where Ei and Ci are the Drago’s constants (Drago et. al., 1993) for component i. In their

analysis, however, they assumed that the properties of the aqueous additive/water

solution are equivalent to the properties of pure water, which might not be correct for

25 concentrated aqueous solutions. In addition, the applicability of the model is limited to systems of known Ei and Ci constants, which are unavailable for most polymeric materials.

Nesterov and Lipatov (1999) used slightly a different form of Flory-Huggins equation. In their analysis for the ternary system, the change in the free energy of mixing three components was expressed per unit volume, and given by:

∆Gmix / V = (Φ1 / V1) ln Φ1 + (Φ2 / V2) ln Φ2 + (Φ3 / V3) ln Φ3

+ Φ1 Φ2 χ12 + Φ1 Φ3 χ13 + Φ2 Φ3 χ23 (2.8)

where V is the mixture volume, and Vi is the molar volume of component i. Based on the above equation, they defined the overall interaction parameter χ123 by the following approximate relation:

χ123 ~ Φ1 Φ2 χ12 + Φ1 Φ3 χ13 + Φ2 Φ3 χ23 (2.9)

Consequently, they implied that a positive value for χ123 is an indication for an immiscible system, whereas a negative value is an indication for a miscible system.

Equations 2.1 and 2.8 are basically the same “the Flory-Huggins equation”, but in different forms and unit basis. Another more convenient form of the Flory-Huggins

26 equation is to express ∆Gmix in dimensionless form and per unit segment of the polymer

chain. For a ternary system, this becomes:

∆Gmix / nkT = Φ1 ln Φ1 + (Φ2 / R21) ln Φ2 + (Φ3 / R31) ln Φ3

+ Φ1 Φ2 χ12 + Φ1 Φ3 χ13 + (Φ2 / R21) Φ3 χ23 (2.10)

where here “∆Gmix / nkT” is the dimensionless free energy of mixing per unit segment or

site, k and T are respectively the Boltzman constant and temperature, n is the total

number of sites in the lattice model, and Ri1 is the size ratio of molecule i to that of

molecule 1, given by:

Ri1 = Vi / V1 (2.11)

In order to apply equation 2.11 correctly, component 1 should be chosen to be the one with the smallest molecular size. In equation (2.11), χij is the interaction parameter

between component i and j in the system, given by:

2 χij = (Vi/RT) (δi – δj) (2.12)

Similarly, the dimensionless free energy of mixing per unit segment for a quaternary

system is given by:

∆Gmix / nkT = Φ1 ln Φ1 + (Φ2 / R21) ln Φ2 + (Φ3 / R31) ln Φ3 + (Φ4 / R41) ln Φ4

+ Φ1 Φ2 χ12 + Φ1 Φ3 χ13 + Φ1 Φ4 χ14 + (Φ2 / R21) Φ3 χ23 +

(Φ2 / R21) Φ4 χ24 + (Φ3 / R31) Φ4 χ34 (2.13)

27 As a rough approximation, the 4 components system (equation 2.13) could be

treated as a binary system, by lumping together three components of the system together

in one component. For illustration purpose only, we will consider the example of our

interest in the current study, the mixing of Sodium benzoate (NaB) with PDMS using a

blend of solvents (water+acetone), where sodium benzoate is first dissolved in the solvent

blend and then the solution is mixed with PDMS. For binary system, equation (2.13) is

reduced to:

∆Gmix / nkT = Φ1 ln Φ1 + (Φ2 / R21 ) ln Φ2 + Φ1 Φ2 χ12 (2.14)

where the subscript 2 refers to PDMS, and the subscript 1 refers to the mixture of water,

acetone and NaB. Here, it is assume that:

δ1 = Φaδa + Φw δw+ ΦN δN (2.15a)

V1 = yaVa + ywVw + yNVN (2.15b)

where yi is the component molar fraction, the subscript a, w, and N refer to acetone, water, and NaB, respectively. The above rough approximations could be reasonable only

if NaB is present in a concentration much below the saturation limit in the solvent, and

that the two solvent components (i.e. water and acetone) are highly miscible with each

other under any ratio.

28 Most of the above models are based on the Flory-Huggins theory. The Flory-

Huggins theory was originally derived to describe the miscibility of polymer-solvent and

polymer-polymer mixtures (Flory, 1953). Although still widely used due to its

simplicity, it is now well-known that the Flory-Huggins theory has many limitations,

which make its prediction weak for many situations. One weakness of the Flory-

Huggins theory came from the original cubic lattice model that was used to derive the mixing functions (entropy and enthalpy of mixing) of a solvent (or a polymer) with another polymer. According to the Flory-Huggins theory, for a binary solvent-polymer system, the lattice is subdivided into n number of sites, each site is filled by either a solvent molecule or a segment of polymer chain (the polymer chain is divided into N number of segments, each segment is assigned to be equivalent in length to the size of the solvent molecule). Similarly, for a binary polymer-polymer system, each site is filled by either a segment of the chain of the first polymer or a segment of the chain of the second polymer. In other words, no sites are allowed to be empty because the polymers are assumed to be simple incompressible fluids and hence there is no volume change upon mixing, which is not a true assumption for a polymeric material due to its huge size. The weakness of the Flory- Huggins theory can be further understood by elaborating more on

the binary interaction parameter, χ12. According to the Flory-Huggins theory, χ12 is always greater than zero, a smaller χ12 indicates a higher chance that the binary solvent-

polymer system would be miscible, and a value of χ12 < 0.5 is the Flory-Huggins criterion for a solvent/polymer system to be completely miscible. However, it is now believed that

χ12 is qualitatively described in the following more general form:

29 χ12 = χ12, d + χ12, fv + χ12, sp (2.16)

where χ12, d accounts for the dispersion interactions, χ12, fv accounts for the interactions resulted from the free volume effect, and χ12, sp accounts for the specific interactions such as acid-base interactions, Hydrogen-bonding, and self-association of the compound.

However, the “Flory-Huggins” theory considers only the dispersion interactions and totally neglects the other two types of interactions, which might result in large errors in estimating the values of χ12. For example, neglecting the free volume effect is based upon the assumption that no volume changes upon mixing, which is rarely satisfied for mixing large macromolecules with solvents of much smaller size.

The incorporation of fillers such as antifouling compounds into polymers will result in having composite materials. The thermodynamic models shown above were originally derived to describe the miscibility of polymer-solvent and polymer-polymer mixtures. Thus, it might not be appropriate to use any of them directly for polymer- fillers composites, where the polymer is “flexible and soft” fluid whereas the fillers are

“rigid” solid particles and in many cases inorganic such as CaCO3 and clay particles. In a series of paper, Balazs and her co-workers (Balazs et al., 1998; Lyatskaya and Balazs,

1998; Ginzburg and Balazs, 1999) emphasized this point, and alternatively developed a new approach for modeling the equilibrium morphology of hybrid systems such as polymer-particle composite. They did not use the conventional Flory-Huggins theory in their modeling. Instead, they applied the Onsager free energy model (Lyatskaya and

Balazs, 1998), the self-consistent field (SCF) theory (Balazs et al., 1998), and the density

30 functional theory (Ginzburg and Balazs, 1999). Although they considered only clay

fillers in all their models, they proposed that their models could be extended for any kind

of solid fillers. One limitation of the models developed by Balazs and her co-authors is

that the models become highly complicated and hence it has to be combined with

numerical simulations.

With the same objective to describe the phase-diagram of polymer-particle

system, Schaink and Smith (1996) developed a new analytical model for the description

of the phase behavior of soft polymer and hard (solid) particle in the present of a solvent.

In their model, Schaink and Smith combined the Flory-Huggins theory (which was used

for polymer/solvent mixture) with the Carnahan-Starling equation of state (which was

used to describe the entropy of the particles), and accounted for the loss of total entropy of the system due to particle-polymer contacts. Although they assumed that the particles were spherical, they proposed that their model could be extended to particles of any geometry. The analytical model of Schaink and Smith was remarkably simple mathematically – compared to the models of Balazs and her co-authors – yet it captured most of the physics of the ternary polymer/solvent/particle system. It should be noticed that the terminology “homogeneous phase and phase separation” differs in the literature when applied to polymer-solvent system and polymer-particle system. For polymer- solvent system, a homogeneous phase means that the polymer is molecularly dissolved in the solvent, whereas in polymer-particle system a homogeneous phase means that the particles are uniformly dispersed in the polymer matrix and no aggregates of particles form.

31 To summarize on the applicability of the above models for our systems, the

following distinction has to be made. The models described for equations (2-1) through

(2.15) are for systems of which the incorporated compound is in dissolved state. On the

other hand, the model of Balazs and her co-authors and the model of Schaink and Smith

are for systems of which the incorporated compound is in dispersed (i.e. solid) state. In

other words, the compounds could be incorporated into the polymer matrix by means of

three methods: mixing the solid particles directly with polymer matrix without using any solvent, dissolving the particles in a solvent at a concentration above the particle/solvent

solubility and then blending with matrix, or dissolving the particles in a solvent at a

concentration below the particle/solvent solubility and then blending with matrix. The

models described for equations (2-1) through (2.15) are applicable only for the third

method whereas the model of Balazs and her co-authors and the model of Schaink and

Smith are applicable only for the first and second methods.

All the above models presented so far are considering only thermodynamic

equilibrium of the system. However, in the process of incorporating antifouling

compounds into polymers by the solvent blending technique, the composition of the

system is continually changing due to evaporation of the solvent, which is resulting in the

nucleation and growth of new phase (solid particles) as the solvent becomes more

concentrated, until the solvent is completely dried off. This dynamic drying step might

strongly affect the final morphology of the composites, but all the above models did not

consider this effect. This process has some similarity with the process of polymer

membrane formation by the dry-casting technique. Several theoretical models exist in

32 the literature that consider both the effect of thermodynamics and mass transfer in order to study the effect of drying on the morphology of the membranes (e.g. Dabral et al.,

2002; Altinkaya and Ozbas, 2004). However, these models are applicable only as long as the system is in one phase. Once phase separation or precipitation occurs, the models are no longer valid. The dynamic effect might be also important even for systems in which the particle is incorporated initially in its dispersed (i.e. solid) state. Balazs (2000) emphasizes that, for a system of two fluids (i.e. polymer + solvent) and a solid particle; there is a dynamic coupling between the fluid-fluid phase separation and fluid-particle wetting phenomena, which significantly affects the kinetic behavior and the final morphology of the system. The dynamic effect becomes more complicated if one of the components (i.e. the solvent) is continually evaporating.

2.4 Modeling of the antifoulants release from polymeric matrices

To model and understand the release mechanism of antifouling compounds from polymeric coatings, it is necessary to consult the literature that discussed the controlled drug release delivery systems. This is the subject of sections 2.4.1 and 2.4.2. In sections

2.4.1 and 2.4.2, the “drug” nomenclature in the original models written by drug release researchers is replaced - by the reviewer – by the “antifoulant’ nomenclature, since both of the drug and antifoulant are some kinds of active compounds whose release are desired at a controlled manner, thus they should show some similarities. On the other side, there exist several modeling efforts - from the perspective of antifouling coatings researchers -

aiming to model the antifoulant release without necessarily consulting the research done

33 for controlled drug release delivery systems, and most of these models are empirical or semi-empirical. A review of this modeling effort is the subject of section 2.4.3.

The mass release mechanism of an antifoulant from a polymer matrix immersed in water depends largely on the solubility of the antifoulant in the matrix. If it has a high solubility, the release primarily follows a diffusion-dissolution mechanism, which takes place in the continuum of the polymer phase. If the solubility is very low, the mechanism is controlled by the channeling/pores formation, where the pores and channels are formed progressively due to water diffusion within the aggregate phase and/or at the polymer/aggregate interface and dissolution of the particles and generation of more empty space (voids), and therefore the media of diffusion here is water that fills the pores, not the continuum of the polymer phase. If the solubility is intermediate, both mechanisms contribute to the overall release and the medium of diffusion are both in the polymer phase and the water-filled pores phase.

Sections 2.4.1 and 2.4.2 review the mathematical modeling efforts published for the case of very high and very low antifoulant/polymer miscibility, respectively. For both sections 2.4.1 and 2.4.2, the review of the modeling efforts is strictly referred to the situations where the following conditions are always satisfied: the polymer matrix is monolithic; the polymer matrix is neither soluble nor swell-able in water; and the active compound is initially loaded in the matrix at a concentration in excess of its solubility limit in the matrix. Also, unless specified, the geometry of the matrix is always in rectangular coordinate.

34 2.4.1 Active compounds with high solubility in the matrix

The Higuchi model (Higuchi, 1963) is the first physical model that describes the

release of the dispersed drug incorporated into a polymer matrix. It is based on pseudo-

steady state analysis, where Higuchi assumed that the concentration gradient is linear in the dissolved zone layer, and assumed that diffusion is the controlling step whereas

dissolution was considered very rapid and therefore its effect was neglected. It was also

assumed that the concentration of the compound is zero at the surface of the matrix (i.e.

perfect sink condition). Therefore, Higuchi was able to come up with the following

simplified analytical expression (Higuchi, 1963):

1/2 Q = [D (2CO ¬ CS) Cst] (2.17)

where Q is the cumulative mass of active compound released (per unit area) at time t, CO is the initial concentration of the active compound in the matrix, CS is the solubility of the

active compound in the matrix, and D is the diffusion coefficient of the active compound

in the matrix. The mass released is described by equation (2.17) as long as there is some un-dissolved compound in the polymer matrix. The mass flux (F) is obtained by differentiating equation (2.17):

1/2 F = [D(2CO ¬ Cs) Cs] / 2t (2.18)

Also, according to Higuchi model, the time (tf) required for the entire compound to be dissolved in the matrix is given by:

35 2 tf = [(CO ¬ CS / 2) L ] / (2 CS D) (2.19)

and the corresponding cumulative mass released (Mf) at tf is given by:

Mf / M0 = 1 - (CS / CO ) / 2 (2.20)

where M0 is the initial mass of the active compound in the matrix. Although the Higuchi

model is based on many serious assumptions that make it oversimplified, its simplicity and clear physical picture make it the standard first step for many researchers to build on a more realistic model.

Paul and McSpadden (1976) modified the Higuchi model where they did not apply the pseudo steady state approximation and did not assume that the driving force is linear in the diffusion zone. Instead, they applied the differential form of Fick’s second

law of diffusion in the diffusion zone, with appropriate boundary and initial conditions, and came up with an exact analytical solution as:

2 1/2 M / M0 = 2 [1 - CS / CO ] β exp(β ) (D t / L) (2.21)

where β is given by:

1/2 2 (π) β exp(β ) erf(β) = CS / (CO - CS) (2.22)

36 where M is the cumulative mass released at time t. Consequently, the time required for

the entire compound to dissolve inside the matrix, tf, and the corresponding releasing

mass at this time, Mf, are given by (Bunge, 1999):

2 2 tf = L / (4 β D) (2.23)

2 Mf / M0 = [1 - (CS / CO )] exp(β ) (2.24)

Although the model of Paul and McSpadden (1976) is more accurate than the Higuchi

model, Paul and McSpadden still neglected the effect of the dissolution step resistance and also assumed the perfect sink condition.

Another different modeling approach is based on first order kinetics, and resulted in (Donbrow and Friedman, 1975):

ln(M0 ¬ M) = ln(MO) ¬ kf t (2.25)

where kf is the rate constant. However, this model is empirical. Upon differentiating

equation (2.25) and dividing both sides by the surface area of the coating (ac), the flux (F)

becomes:

F = (MO / ac) kf exp (¬ kf t) (2.26)

37 2.4.2 Active compounds with low solubility in the matrix

If the solubility of the active compound in the matrix is very low, then the media

of diffusion is the liquid that fills the pores within the matrix, and hence diffusion through

the continuum of the polymer phase could be neglected. Therefore, the polymer matrix

could be visualized as a porous membrane. Known to be the first mathematical model

published in the literature to describe the drug mass release from porous polymeric

matrix, Higuchi (1963) applied the pseudo-steady state approach to come with:

1/2 M = A [ (Db ε / Ө) Cbs (2Ca0 ─ ε Cbs) t ] (2.27)

where A is the total surface area of exposure; Db is the diffusion coefficient of the

compound in the pore-filled fluid (i.e. water in our systems); ε and Ө are respectively the

matrix porosity and tortuosity factor; Ca0 is the initial concentration of the compound in the matrix; and Cbs is the solubility of the compound in water. However, equation 2.27 is

derived based on pseudo-steady state assumption, and on the assumption that the

compound aggregates are quite small comparing to diffusion distance and are evenly

distributed in the matrix. Also, the pseudo-steady state assumption only holds if (Ca0 >>

ε Cbs). Another concern for equation 3.1 is the porosity ε, which is actually not constant

but is a function of time as more particles of the active compound are dissolved. For

simplicity, to account for the time functionality of ε, one may assume that [ε ~ Ca0 / ρa],

where ρa is the density of the active compound, provided that the initial porosity could be neglected, and if the initial volume fraction of the compound in the matrix is high.

38 Desai et al (1966) slightly modified equation (2.27) in order to account for the

drug binding phenomena, and came with:

1/2 M = A { (Db ε / Ө) Cbs [2Ca0 ─ [ε + Keq(1─ ε)] Cbs] t ] (2.28)

where Keq is the equilibrium partition coefficient of the active compound in the matrix,

which is defined as the ratio of the concentration of the compound in the polymer phase to its concentration in the fluid inside the pores. Hence, if Keq is zero, equation (2.28) is

reduced back to equation (2.27).

Miller and Peppas (1982) applied the moving boundary approach to describe the

release from a polymer with continuous porosity formation. In their analysis, the

compound dissolves in the fluid medium (e.g. water) at the interface which moves into

the pore as the dissolution progresses. After defining the position of the interface by [2α

1/2 (Db t / Ө) ], they derived analytical expression for the mass release rate (dM/dt) as:

2/3 1/2 (dM/dt) = [ε A Cbs / erf (α)] (Db / Ө π t) (2.29)

where α is obtained from:

1/2 2 π α exp (α ) erf (α) = Cbs / (ρa ─ Cbs) (2.30)

The model of Miller and Peppas (1982) is much more accurate than the Higuchi model.

However, it is still limited to the case when (Ca0 > ε Cbs), and in their model they

neglected the kinetic of the dissolution step assuming that it is very rapid.

39 To account for the kinetic of the dissolution step, Peppas (1983) modified the above

model (Miller and Peppas, 1982), keeping the other assumptions the same. He described

the process mathematically by:

2 2 (∂C / ∂t) = Deff (∂ C / ∂x ) + (2kd / rp) (Cbs – C) (2.31)

where kd is the dissolution constant, rp is the radius of the pore, and Cbs is the saturation

concentration of the active compound at the pore wall-dissolution medium interface.

Peppas (1983) solved the above differential equations for a special case for obtaining the release rate (dM/dt) at long time t from a slab of thickness (i.e. pore length) δ:

2/3 1/2 1/2 (dM/dt) = ε Cbs (2kd Deff / rp) tanh [(δ/2) (2kd / Deff rp) (2.32)

However, in his model, Peppas assumed a constant value for rp, but in reality it is a

function of immersion time, t.

The effect of the pore-size distributions on the mass release of the active compound

was neglected in all the above models. Very limited works (e.g. Saltzman and Peppas,

1987) considered this effect, by applying the percolations theory approach.

For a system with a finite sink where the bulk solution concentration is not zero,

Gerstl et al. (1998) applied a kinetic model (originally derived by Cobby et al (1974)) to

describe the mass release of the active compound loaded into a spherical matrix:

40 1/2 1/2 2 1/2 3 M/M0 = 3K t ─ 3(K t ) + (K t ) (2.33)

In equation (2.33), K is the release rate constant whose expression is also dependent on

the geometry of the matrix. For a spherical matrix, K is defined by (Gerstl et al., 1998):

½ K = (1/ Ca0 r) [Dm (2 Ca0 ─ Cbs) Cbs] (2.34)

where r is the radius of the matrix (which is a microsphers here), Ca0 and Cbs are as

defined before, and Dm is the diffusion coefficient of the compound in the matrix

(including the effect of the porosity and tortuosity).

2.4.3 Leaching behaviors of marine antifouling paints

With respect to their behaviors in seawater, marine antifouling paints can be classified into three categories: insoluble matrix, soluble matrix, and self-polishing matrix. This section considers only the insoluble matrix category, summarizing the mathematical modeling efforts specifically conducted for quantifying the leaching behaviors of insoluble antifouling paints in water. The models summarized here are from the perspective of antifouling paints researchers, not from the perspective of controlled drug delivery researchers. Although most of the models presented here are empirical or semi-empirical, they are useful since they are supported by real field leaching experiments in seawater. In the following discussion, the pigment (a nomenclature

41 frequently used by antifouling paints researchers) is always referred to the antifouling

compound.

Insoluble matrix coatings are those that do not erode over time in water, and

sometimes called continuous contact or contact leaching coatings, because of the leaching

mechanism involved. Marson (1969) was the first who mathematically modeled the

antifoulant leaching from insoluble matrix paints. In order to come up with his model, he

postulated the following leaching mechanism, with an assumption that the antifoulant

particles (Cu2O in his study) were spherical of equal size and uniformly distributed in multi layers in the matrix (a rubber resin in his study). First, when the coating is initially immersed in water, the pigment particles at the surface layer of the film dissolve forming a saturated solution of the pigment at the pigment/leachate interface. Then, the saturated solution diffuses outward through the diffusion layer in contact with the coating surface.

When a particle dissolves in seawater to reveal the thin binder membrane separating it from the un-dissolved particles, water diffuses into the thin membrane and dissolves some of the un-dissolved particles. Consequently, the resulting osmotic pressure ruptures the membrane and the pores become interconnected. After many simplifications, Marson was able to come up with a simple analytical equation that shows the dependency of the

leaching rate on the parameters of the coating and of the leaching media:

F = B υ / [1 + Cd/P) (2.35)

where F is the leaching rate, B and C are constants, υ is the pigment volume fraction, d is

the thickness of the matrix, and P is the fraction of interconnected holes that depends on υ

42 and the particle size distribution. P could also be defined as the packing factor (a parameter defining the number of voids between interconnecting pigment particles).

However, the Marson model is empirical and not predictive since it does not include the dependency of the leaching rate on time. Later on, Marson (1974) modified his original model to account for the dependency of the number of interconnecting holes on the initial Cu2O loading. However, his modified model is still empirical and not time- dependent. It should be noticed that, since Cu2O is insoluble in the rubber matrix, the leaching mechanism is similar to the one described for drug release from porous matrix

(section 2.5.2).

De La Court and De Vries (1973) modified the Marson model to account for the shape and distribution of the antifoulant particles in the matrix. However, they applied a pseudo steady state assumption in their analysis.

Monaghan et al. (1978) derived an empirical relation to quantify the leaching of organotin compounds form insoluble matrix coatings. However, it appears that their model is empirical and maybe applicable only for specific organotin compounds.

One disadvantage of the Marson model is that it is not predictive because the independent variable is the film thickness. Caprari et al (1990) re-derived and modified the Marson model to let the immersion time to be the independent variable, hence their model turned to be predictive (Caprari et al , 1990):

43 F = B υ / [1 + C e n/ε) (2.36)

where e is the thickness of the leached layer, and n is the number of leached layers. Here,

n is a function of the immersion time (t), and given by the solution of the quadratic

equation (Caprari et al , 1990):

2 2 2 0 = (Cdpe / 2KBθ) n + [dpe / KB + Cdpe / 2KBθ] n + [dpe / KB ─ t] (2.37)

-6 where dp is the pigment density, K is a constant (K = 10 Mp / Mb, where Mp and Mb are respectively the molecular weight of the pigment and the bulk fluid), and ε is the porosity, given by the following empirical equation (Caprari et al , 1990):

ε = exp [─ (1 ─ υ)2 / A υ] (2.38)

where A is a constant. The Caprari model is much better than the original Marson model.

However, it is still a semi-empirical relation.

Vasishtha et al. (1995) performed a leaching experiment to study the leaching of Sea

Nine 211 from VYHH matrix (a commercial polymer from Union Carbide, which is a

copolymer of vinyl chloride and vinyl acetate). They varied the initial loading of Sea

Nine 211 in VYHH from 20-35 wt%, and prepared the coatings on cylindrical rods. In their mathematical analysis, they selected the model of Crank (1975) for describing the release form cylindrical matrix coating:

44 1/2 2 1/2 M / M0 = 4 (π) [D t / a ] (2.39)

where a is the radius of the cylinder. By fitting the leaching data to the above equation,

they calculated the diffusion coefficient, D, to be 1.7 x 10-14 and 3.8 x 10-13 cm2/s at 20

wt% and 35 wt% loading, respectively. However, the model that they selected was strictly derived by Crank for a special case when the compound is initially loaded in the

matrix at a concentration well below the compound/matrix solubility, thus the compound

is always molecularly dissolved in the polymer phase. However, in their work, Vasishtha

et al. (1995) did not justify if the solubility of Sea Nine 211 in VYHH is greater than 35 wt%, in order to use the Crank equation. In fact, it is very rare to have an antifouling compound with such a high solubility in the matrix.

CHAPTER III

EXPERIMENTAL APPROACH

This chapter summarizes the experimental approach followed in the current study to prepare, characterize and evaluate different systems of antifoulants-incorporated silicone coatings. Four antifoulants were considered: sodium benzoate, benzoic acid, capsaicin, and tannic acid. These antifoulants were incorporated into two types of silicone coatings: Sylgard® 184 and RTV11 coatings. The chapter starts with listing all the materials used in the current study. Then, the procedures for preparing the samples are described in detail. Next, the experimental protocol for processing and evaluating the prepared samples are explained. Finally, the principles behind the characterization techniques used in the current study are briefly described.

3.1 Materials

RTV11 and Sylgard®184 were purchased from GE and Dow Corning, respectively. RTV11 came from the manufacturer in two parts: a polymer base (a hydroxy-terminated PDMS) and a catalyst (Dibutyltin dilaurate or DBT). Included in the polymer base are the cross-linking agent (ethyl silicate 40) and the filler (calcium

46 carbonate). Sylgard®184 also came in two parts: a polymer base containing vinyl-

terminated-PDMS and a curing agent having SiH-terminated-PDMS and the Pt-catalyst.

Table 3.1 summarizes the compositions of the two types of silicones.

Sodium benzoate (99% pure, in powder form) was purchased from Sigma Aldrich

Chemical Inc. and used as received. It has a molecular weight of 144.1 g/mol and a

melting point of above 300 o C. It is highly soluble in water (0.555 g / 1 g water) but has a limited solubility in most organic solvents.

Benzoic acid (99% pure) was purchased from Sigma Aldrich Chemical Inc. and used as received. It has a molecular weight of 122.1 g/mol and a melting point of 122 o

C. It is highly soluble in most organic solvents and has a limited solubility in water. The solubility of benzoic acid in water is about 3.4 mg/ml (Farrell and Sirkar, 1999)

Capsaicin is a naturally extracted mixture containing approximately 65% capsaicin and 35% dihydrocapsaicin. It was purchased from Sigma Aldrich Chemical

Inc. and used as received. It has a molecular weight of 305.4 g/mol and a melting point of 63.5 o C. It is highly soluble in most organic solvents and has a very low solubility in

water (about 60 ppm).

Tannic acid was purchased from Sigma Aldrich Chemical Inc. and used as

received. According to the data sheet for the commercial product obtained from Sigma

47 Table 3.1 The compositions of the two types of silicones (RTV11 and Sylgard® 184) used in the current study. All percentage shown here are in mass basis.

Component RTV11 Sylgard® 184

Polymer base OH-PDMS (66.4 %) * Vinyl-PDMS (90.9 %)

Curing agent ES40 (1.6 %) * SiH-PDMS (9.1 %)

Inorganic Filler CaCO3 (32 %) * -

Catalyst DBT (0.5 %) ** Pt (included in the curing agent)

* comes in one part (part 1). ** per 99.5 % of part 1

Aldrich, the molecular weight is specified as 1700 g/mol. It has a melting point of 210 0

C. It is highly soluble in water (1 g/ 0.35 ml water) and also has good solubility in some organic solvents such as acetone and ethanol.

The organic solvents used in the current study (toluene, acetone, acetonitrile, ethanol and ether) were purchased from VWR and used as received. Microscopic glass slides (pre-cleaned, size: 25 x 75 x 1 mm) were obtained from VWR and used as received.

3.2 Sample preparation

The antifoulants (capsaicin, benzoic acid, sodium benzoate, and tannic acid) were incorporated into two polymeric silicone coatings (Sylgard® 184 and RTV11) by the 48

Table 3.2 The actual combinations of (antifoulant/solvent/polymer) mixtures used in the current study to incorporate the antifoulant into the bulk of the polymer matrix by the solvent-blending technique.

Antifoulant Silicone matrix Solvents

Sylgard®184, water/acetone blends Sodium benzoate RTV11 (different ratios)

Acetone, Sylgard®184, toluene, Benzoic acid RTV11 acetonitrile, ether

Toluene, Capsaicin RTV11 Ethanol

Sylgard®184, Tannic acid Acetone RTV11

solvent-blending technique. Briefly, in order to obtain homogeneous antifoulant/silicone

blending by the solvent-blending method, the antifoulant was first dissolved in a solvent

or blend of solvents, and then the solution was homogenized with the polymer base, then

the curing agent was added to the mixture after drying off the solvent used. Table 3.2

summarizes the actual combinations of materials (antifoulant/solvent/matrix mixtures)

used in the current study.

49 For incorporating sodium benzoate (NaB) into silicones, the following procedure

was followed, which is shown schematically in Figure 3.1, and the detailed

concentrations of the samples prepared are summarized in Table 3.3. NaB is much more

inorganic in nature than the other antifoulants, and consequently has low solubility in

organic solvents. On the other side, it has a high solubility in water (0.555 g of NaB per

1 g of water). Therefore, it was first dissolved in de-ionized water at a certain concentration (the exact concentration is shown in Table 3.3). Next, a water-miscible organic solvent, acetone, was added to the aqueous phase at a certain water/acetone mass ratio (the exact ratio is shown in Table 3.3) while maintaining the complete solubility of

NaB in the mixed-solvent. Then, the NaB/(water + acetone) solution was homogenized

NaB Acetone

polymer Curing base solvent agent Water Water + NaB mixing drying mixing

Figure 3.1 A schematic diagrams for the sample preparation steps used for incorporating sodium benzoate into the silicone polymer coating.

Table 3.3 The detailed concentrations for NaB/ Sylgard®184 samples prepared at different conditions. All concentration shown here are in mass basis. (Abbreviation: W: water, A: acetone, S: solvent (i.e. water + acetone), P: silicone polymer base).

Sample wt% NaB/W W/A ratio Wt% NaB/S S/P ratio wt% NaB/P #

1 7.77 50/50 4.0 20/80 1

2a 16.67 50/50 9.1 10/90 1

2b 4.61 50/50 2.4 30/70 1

3a 17.39 20/80 4.0 20/80 1

3b 12.31 30/70 4.0 20/80 1

3c 9.52 40/60 4.0 20/80 1

3d 5.00 80/20 4.0 20/80 1

3e 4.47 90/10 4.0 20/80 1

3f 4.04 100 % W 4.0 20/80 1

4a 3.94 50/50 2.0 20/80 0.5

4b 15.09 50/50 8.2 20/80 2

4c 22.02 50/50 12.4 20/80 3

4d 28.57 50/50 16.7 20/80 4

4e 34.78 50/50 21.1 20/80 5

51 with the silicone polymer base at different solvent/polymer mass ratios (the exact ratio is

shown in Table 3.3) with rigorously mixing. Then, the mixture was heated for 4 hours at

140 oC (to remove the solvent), then left inside the fume-hood at ambient conditions for 2

days, then under vacuum at room temperature for 20 min to completely remove all the

mixed solvent. This procedure produced silicone coatings with different wt%

NaB/polymer (i.e. after drying off the solvent), which were prepared at different

solvent/polymer mass ratios, and different water/acetone mass ratios. The detailed concentrations for all the NaB/silicone samples prepared in the current study are presented in Table 3.3. After that, the samples (after drying off the solvent) were cured and processed as follow. The DBT catalyst (for RTV11 case) or the curing agent (for

Sylgard®184 case) was added according to the manufacturer prescribed ratio (0.5 wt% of

DBT/RTV11, and 1:10 mass ratio for Sylgard® 184). Finally, each mixture (6.0 g) was

poured into a 9 cm in diameter polystyrene Petri dish, and left to cure at ambient

conditions for 1 week. This produced sheets of film thickness of about 900 µm. For producing thinner films, the above mixtures (~ 0.3 g) were spread out on glass substrates

(on surface area of about 1.2 x 5 cm) by means of the Doctor-Blade method, which resulted in film thickness of about 500 µm. This experimental procedure was similar to the procedure used for incorporating zosteric acid into silicone coatings using different solvent blends (Barrios et al., 2005).

For incorporating capsaicin, benzoic acid, and tannic acid into silicones, the following procedure was followed. Since these antifoulants have high solubility (10 to

30 wt. %) in most organic solvents that are miscible with silicone, they were first directly

52 dissolved in the organic solvents at a concentration of 4 wt.%. Next, 1.5 g of the 4 wt. %

antifoulant/solvent solution was homogenized with 6 g of the silicone polymer base with

rigorously mixing. In other words, a solvent/polymer mass ratio of 20/80 was used, which was fixed for the case of capsaicin, benzoic acid, and tannic acid throughout the study. This procedure produced a matrix with 1 wt% of antifoulant entrapped in the silicone matrix after drying off the solvent. Then, all mixtures were left inside the fume- hood at ambient conditions for 2 days, then under vacuum at room temperature for 20 min, to remove the solvent. After that, the samples (after drying off the solvent) were cured and processed exactly the same way as described above for NaB/silicone coatings.

3.3 Sample processing

After making sure that the samples prepared were apparently cured, the samples were preceded for further experimentations and processing, as follow.

First, the surface and bulk properties of the samples were evaluated by a variety of techniques. The wettability of the samples was evaluated in terms of measuring the water contact angles. The bulk property of the samples was evaluated by measuring the elastic modulus, which was done either by the JKR technique and/or the stress-strain technique.

The surface morphology of the samples was examined by Optical Microscopy techniques. The details of the surface morphology and surface roughness were evaluated by SPM technique. SPM was employed here for two set of samples only: RTV11

53 samples, and 1 wt% capsaicin/RTV11 samples, and the SPM scans were done here for

both as-prepared samples and for samples after being immersed in DI water for 14 days.

Second, the miscibility of the antifouling compounds with the silicone matrix was described through the analysis of the aggregate size of the antifouling compounds in the matrix. One advantage of Sylgard®184, compared to RTV1, is that it is highly transparent so any material entrapped could be observed easily by the Optical Microscopy.

Therefore, Optical microscopic images were taken for the bulk of Sylgard®184 contained

the incorporated compounds, after drying off the solvent. The aggregate size was

analyzed by Scion Image Software

Third, the leaching of the incorporated compounds from the silicone coatings into

water was evaluated, as follow (the detailed compositions for the samples that were

subjected to leaching experiments in the current study are summarized in table 3.4). For

benzoic acid, tannic acid, and sodium benzoate - incorporated silicones, large sheets

(total mass: ~ 6.0 g; size: ~0.09 cm x 64 cm2) of silicones with the incorporated

compounds were immersed in large glass beakers each containing 300 ml de-ionized (DI)

water. To determine the amount leached out, the conductivities of the solutions at

different time intervals were measured and their corresponding concentrations were

entrapolated via standard calibration curves. For capsaicin/RTV11 system, large sheet

(total mass: ~ 6.0 g; size: ~0.09 cm x 64 cm2) of RTV11 with the incorporated compound

was immersed in large glass beakers containing 500 ml DI water. Then, the

Table 3.4 The detailed compositions for the samples that were subjected to leaching experiments in the current study. All concentration shown here are in mass basis. (Abbreviation: W: water, A: acetone, S: solvent, P: silicone polymer base, NaB: sodium benzoate, BA: benzoic acid, TA: tannic acid).

Sample S/P wt% Compound Polymer Solvent # ratio compound/P

® 1 NaB Sylgard 184 50/50 W/A 20/80 1

2a NaB Sylgard®184 50/50 W/A 10/90 1

2b NaB Sylgard®184 50/50 W/A 30/70 1

3a NaB Sylgard®184 20/80 W/A 20/80 1

3b NaB Sylgard®184 30/70 W/A 20/80 1

3c NaB Sylgard®184 40/60 W/A 20/80 1

3f NaB Sylgard®184 80/20 W/A 20/80 1

3g NaB Sylgard®184 90/10 W/A 20/80 1

3h NaB Sylgard®184 100 % W 20/80 1

4a NaB Sylgard®184 50/50 W/A 20/80 0.5

4b NaB Sylgard®184 50/50 W/A 20/80 2

4c NaB Sylgard®184 50/50 W/A 20/80 3

4d NaB Sylgard®184 50/50 W/A 20/80 4

4e NaB Sylgard®184 50/50 W/A 20/80 5

5 NaB RTV11 50/50 W/A 20/80 1

55 Table 3.4 The detailed compositions for the samples that were subjected to leaching experiments in the current study (Continued)

Sample S/P wt% Compound Polymer Solvent # ratio compound/P 6a BA Sylgard®184 Toluene 20/80 1

6b BA Sylgard®184 Acetone 20/80 1

6c BA RTV11 Acetone 20/80 1

7a Capsaicin RTV11 Toluene 20/80 1

7b Capsaicin RTV11 Ethanol 20/80 1

8a TA Sylgard®184 Acetone 20/80 1

8b TA RTV11 Acetone 20/80 1

concentrations of capsaicin in the water bath at different time intervals were evaluated by

HPLCE technique where the unknown concentrations were interpolated via standard

calibration curve.

Fourth, bacterial attachment in fresh water containing enriched bacteria isolated

from Lake Erie was conducted to assess the coating’s antibacterial performance of

capsaicin-treated coatings and sodium benzoate-treated coatings. The bacterial water solutions were prepared and provided by the research group of Dr. T. Cutright, Civil

Engineering Department, and the detailed procedure for culturing the bacteria was described elsewhere (Xu et al., 2005; Xu, 2004). The coatings prepared on glass slides were placed in amber bottles half-filled with the water having the isolated bacteria. Care was taken to ensure all the coatings were arranged facing the bottom of the bottle to avoid 56 the settlement of foreign species and organic matter. Coatings were observed at periodic

intervals up to one month. At each time interval, a coating was removed, dipped gently in fresh de-ionized water several times to remove loosely attached objects, and after drying to remove the DI water, they were subjected to optical microscope observation

immediately.

3.4 Characterization Techniques

3.4.1 Contact Angle technique

Contact angle is a quick and easy method to evaluate the wettability of a solid

surface (Chan, 1994). It is also an indirect quantative method for measuring the surface

energy of a solid surface. By comparing the contact angles of two surfaces, higher contact angle indicates lower wettability and higher hydrophobicity of the surface, which

implies lower surface energy. The physics behind contact angle phenomena is best

described through considering the situation of a liquid drop being in equilibrium with a

solid substrate in air, which is frequently described mathematically through the famous

Young’s equation:

γLV cos θ = γSV ─ γSL (3.1)

where γLV, γSV and γSL are respectively the surface energies at the liquid/vapor,

solid/vapor, and solid/liquid interfaces, and θ is the equilibrium contact angle (as shown

57 in Figure 3.2a). Therefore, the Young’s equation relates the contact angle to the surface

energy of the substrate. Experimentally, the contact angles can be measured by applying the sessile drop method where both dynamic and static contact angles can be measured.

For dynamic angles, the advancing and receding angles are achieved by the addition and removal, respectively, of water from the drops formed on the coating surface, whereas for static angles the water drop is placed on the coating surface and let to equilibrate without external force (as shown in Figure 3.2b).

γLV (a) air

θ γSL γSV Substrate

advancing receding (b) (c)

θ θ a r Substrate Substrate

Figure 3.2 A simplified sketch for the contact angle concept. (a) the static contact angle: a liquid drop being in equilibrium with a solid substrate in air. γLV, γSV and γSL are respectively the surface energies at the liquid/vapor, solid/vapor, and solid/liquid interfaces, and θ is the equilibrium (static) contact angle. (b) the advancing contact angle (θa). (c) the receding contact angle (θa).

58 In the current study, the contact angle technique was applied to evaluate the surface wettability of the coatings. Deionized water was the probe liquid, and both dynamic and static contact angles were measured via the sessile drop method. Images of the drops were captured using the Dazzle DVC (Digital Video Creator) and its software, and data were processed using the Scion Image Software.

It should be noticed that both advancing and receding contact angles slightly depend on the rate of fluid injection (for advancing) and removal (for receding). For advancing angle, it slightly increases as the rate of injection increases. For receding angle, it slightly decreases as the rate of removal increases. Therefore, for our experiments, the rate of injection was gradually increased at three rates and the corresponding advancing contact angles were measured and averaged all together.

Similarly, the rate of removal was gradually decreased at three rates and the corresponding receding contact angles were measured and averaged all together.

3.4.2 The stress - strain technique

The stress – strain technique is a common and widely used technique for measuring the elastic modulus of most types of materials, including both soft and hard materials. According to this method, the elastic modulus (E) is defined as ratio of the stress applied on a sample over the strain resulted in the shape of the sample.

Experimentally, the stress is usually varied at a certain range, and at each time the elongation in the sample length is measured. Strain is defined as [(L─L0)/L0)], where L

59 and L0 are the initial length and the elongated length of the sample, respectively. Stress is defined as [mg/(w x t)], where m is the mass of the load used, g is the acceleration of gravity, w and t are respectively the width and thickness of the sample. Hence, E is obtained as the slope of the linear plot of stress v.s strain data.

For the current study, the stress – strain technique was applied to measure the elastic modulus of antifoulant-blended RTV11 coatings, due to their opaque nature. A small rectangular sheet (length ~ 25 mm, width ~ 6 mm, and thickness ~ 1 mm) of coating was vertically hung in air, its elongation under a particular weight was measured from the magnified images captured using the goniometer video system and the Dazzle

DVC and its software. Stress was varied gradually up to ~ 0.13 MPa.

3.4.3 The JKR technique

The JKR technique is a specialized technique for measuring the elastic modulus of soft materials. It is also a direct quantative method for measuring their surface energies. Its name came on behalf of the three scientists (Johnson-Kendall-Roberts) whose originally established the theoretical framework for this method in 1971 (Johnson et al., 1971), and sometimes called contact mechanics technique. The detailed theory behind the JKR technique can be found elsewhere (Chaudhury and Whitesides, 1991;

Johnson et al., 1971). Briefly, a soft elastic lens is brought into contact with an elastic surface, and the deformation of the contact zone (normally a circular area) under a certain load can be related to the elastic modulus of the system, thus the modulus of the coating.

60 In the current study, the JKR method was used to measure the elastic modulus of

Sylgard® 184 coatings and antifoulant-incorporated Sylgard® 184 coatings. The modulus

of antifoulant-incorporated Sylgard®184 coatings could also be measured via the stress-strain technique. However, the JKR technique was selected here because of the extensive usage of the JKR technique for evaluating the properties of Sylgard®184, a highly transparent coating, and the values measured in this study could be compared to the reported values. The elastic modulus of control Sylgard®184 coating was also

measured by the stress-strain method to compare between the accuracy of the two

methods. For the JKR method, the procedures developed by Chaudhury and Whitesides

(1991) were followed in the current study. Briefly, as shown schematically in figure 3.3,

the radii of contact areas for 8 to 10 different compression loads were measured,

and the contact radius vs. load was plotted to extrapolate the modulus of the system

from the slope of the plot. With the known modulus of the lens and assuming the

materials were perfect elastic, the modulus of the coating was deduced from the modulus

of the system.

3.4.4 Optical Microscopy

Optical microscopy is a quick nondestructive tool for providing useful

information about the overall morphology – in two dimensions – of the sample surface.

The principle behind it is that it employs a visible light. Because optical microscope requires a visible light source, it has a limitation that it can not scan very small area of less than tens of microns, therefore localized morphology information in nanometer scale

Optical Microscope

Glass slide

Hemisphere lens

Sample

Analytical balance

Moving

stage

Figure 3.3 A simplified sketch for the set-up of the JKR apparatus.

can not be obtained by this technique. Also, only images can be produced by this

technique and no quantative data can be generated directly. The basic components of

optical microscope are the light source, the condenser (to condense the light), the

objective, and the eyepiece. Common types of Optical microscopy involve reflected light

microscopy (for non-transparent samples) and transmitted light microscopy (for transparent samples). Transmitted light microscopy can be further classified into: brightfield microscopy, phase contrast microscopy, polarized light microscopy, and differential interface contrast microscopy. 62 In the current study, variations in the morphology of the AF compounds- incorporated silicone films were observed using an optical microscope (Model IX-70,

Olympus) having video and still image capturing capabilities.

3.4.5 Scanning Probe Microscopy

Scanning probe microscopy (SPM) involves several types, such as scanning tunneling microscopy (STM), atomic force microscopy (AFM), lateral force microscopy

(LFM), and magnetic force microscopy (MFM). The basic components of SPM are the laser diode, the piezoelectric scanner, the cantilever and tip probe, and the position sensitive photo detector. The tip is usually made of silicon or silicon nitrile. AFM was the technique used in the current study. In AFM, a force probe is applied, which detects the van der Waals interaction force between the probe tip and the surface, to scan over the surface of a sample. AFM has the advantages over the optical microscopy technique is that it can provide quantitative information about the localized surface topography of the samples both in two dimensions and three dimensions, and it can also provide quantative information about the surface roughness (Magonov and Reneker, 1997). The resolution of AFM is very high, close to the atomic level, where a surface area of as small as 1µm x 1µm can be scanned with a high image quality. However, opposite to optical microscopy technique, AFM has a limitation that it is not suitable for scanning large surface area (e.g. > 100µm x 100µm) to provide overall morphology of the sample surface. AFM is commonly operated in either a contact mode (where repulsive force is

63 used) or a non-contact mode (where attractive force is used). The non-contact mode is generally preferred because it usually gives better resolution than the contact mode.

In the current study, the details of the coating surface were examined with the non-contact mode AFM (Metrology 2000, Molecular Imaging), where a Si3N4 cantilever with a spring constant of 21 – 78 N/m was used. All the AFM images presented in this study have a scan size of 80 µm x 80 µm and obtained with a scan rate of 0.20 Hz.

3.4.6 High Performance Liquid Chromatography (HPLC)

HPLC is a popular method of analysis. It has many applications such as separation, identification, purification, and quantification of various compounds. The basic components of the HPLC set-up are the solvent reservoir (where the mobile phase comes from), the pump, the injection port (where the samples are injected into the mobile phase), the column (where the stationary phase is placed), the detector, and the waste reservoir. The basic principle behind HPLC is that certain compounds have different migration rates given a particular column and mobile phase. Thus, the chromatographer can separate compounds from each other, and the degree of separation is mostly determined by the selection of the mobile phase and the stationary phase. The mobile phase is the solvent being continually applied to the column, and acts as a carrier for the sample solution. As a sample solution flows through the column with the mobile phase, the components of that solution migrate and separate according to the non-covalent interactions of the compound and the mobile phase with the stationary phase. For

64 example, those samples which have stronger interactions with the stationary phase than with the mobile phase will elute from the column slower and therefore will have a longer retention time, whereas the reverse is also true (Schoeff and Williams, 1993).

In the current study, HPLC technique was employed to determine the unknown bulk concentrations of capsaicin in water, as part of the leaching experiment. The solutions of unknown capsaicin concentrations were subjected to HPLC analysis (Model

LC-10AT from Shimadzu with the symmetry C-18L column from Walters) using a mixture of acetonitrile + DI water (50:50 vol/vol) with a pH value of 2.1 as the mobile phase. 10 µL of the solution was injected into the column and flown at a constant rate of

1 ml/min. In order to obtain the calibration curve, a set of standard solutions with concentrations in the range of 13 – 5000 ppm of the purchased capsaicin were prepared using the same mobile phase as the solvent.

CHAPTER IV

RESULTS AND DISSCUSSION FOR SODIUM BENZOATE-BASED COATINGS

In this chapter, the results obtained for sodium benzoates (NaB) - incorporated

silicone coatings are presented and discussed. The effect of incorporating the compound on the surface and bulk properties of the coatings is presented in section 4.1. The effect of varying the preparation conditions on the bulk morphology/miscibility of NaB-based coatings is discussed in section 4.2. Theoretical thermodynamic analysis for the miscibility study is presented in section 4.3. The effect of varying the preparation conditions on leaching of the compound in water is presented in section 4.4. Theoretical mass transfer analysis for the leaching study is presented in section 4.5. The antibacterial performance for the NaB-incorporated coating is presented in section 4.6.

4.1 Effect of sodium benzoate on surface and bulk properties of silicones

Sodium benzoate (NaB) was incorporated into two types of silicones (Sylgard®

184 and RTV11). For Sylgard® 184 coating, the concentration of NaB in the matrix was varied from 0 to 5 wt%. For RTV11 coating, only one concentration was prepared (1 wt% NaB in the matrix). For all of the samples prepared, it was observed that the NaB- blended coatings were cured similarly as the NaB-free coatings alone. Further examining

66 the effect of incorporated compound on the surface and bulk properties of the silicone

coatings qualitatively confirmed this observation, as to be discussed in sections 4.1.1 and

4.1.2.

4.1.1 Effect on wettability

The wettability of the coatings was investigated in terms of measuring the water

contact angles. Sylgard® 184 coatings of various amounts of NaB (0 - 2 wt %) were subjected to contact angle measurements, and the results are shown in Figure 4.1. For control samples of NaB - free Sylgard® 184 coatings, the advancing, static, and receding

contact angles were measured to be 110°, 106° and 80°, respectively. As seen in Figure

4.1, the incorporation of NaB in Sylgard® 184 (up to 2 wt. %) had little effect on the wettability of Sylgard® 184 coating. This could suggest that most of NaB molecules

were entrapped inside the bulk of Sylgard® 184 rather than aggregated to the surface.

Otherwise, the contact angle was expected to decrease considerably due to the fact that

NaB has much higher surface energy than silicones (the surface energy for different types

of silicones is in the range of 20 - 24 mJ/m2), and the contact angle hysteresis (difference

between the advancing and receding angles) should increase due to the in-homogeneity of

aggregates if presented on the surface.

In addition, the contact angles of NaB-free RTV11 coating and 1 wt%

NaB/RTV11 coating were also measured, and the results are shown in Table 4.1. For

control samples of NaB - free RTV11 coatings, the static contact angle was 101°,

120

110

100

90 Contact angles Contact 80

70 00.511.522.5 wt% NaB in Sylgard® 184

Figure 4.1 Water contact angles of NaB-entrapped Sylgard®184 films (advancing: U, receding: , and static: {). The (solvent: polymer) ratio and the (water: acetone) ratio were respectively (20: 80) and (50: 50) by mass, which were fixed at all concentrations. Error for each data point (average over 12 measurements) is presented by the vertical line.

68 Table 4.1 Static water contact angles of NaB-entrapped RTV11 films. The (solvent: polymer) ratio and the (water: acetone) ratio were respectively (20: 80) and (50: 50) by mass.

Coating Static Contact angles

Control RTV11 101.2 ± 0.5

1 wt% NaB/RTV11 100.7 ± 0.7

and maintained around this value for the 1 wt% NaB-blended RTV11 coating. The

indifferent contact angle values of 1 wt% NaB-blended RTV11 as those of pure RTV11 could give us the indication that most of NaB molecules were entrapped inside the bulk of RTV11 rather than aggregated to the surface for the same reasoning mentioned above for NaB- Sylgard® 184. To summarize, the indifferent contact angle values of NaB

blended coatings as those of pure silicone is the first indication that the NaB-blended

silicone coatings could be cured.

It could be of interest also to notice here the slight difference in the wettability of pure (i.e. NaB-free) Sylgard® 184 and pure RTV11 coatings. By comparing the static contact angle data for control samples of NaB-free coatings, it could be suggested that

Sylgard® 184 silicone has slightly a higher hydrophobicity than RTV11 silicone. The

same conclusion could be drawn from the surface energy data for the two matrices, as

follow. From literature, the surface energy of RTV11 is 23.3 mJ/m2 (Berglin and

69 Gatenholm, 1999), which is slightly higher than the surface energy of Sylgard® 184 silicone (~ 20 mJ/m2).

Another interesting observation is the effect of the curing temperature on the contact angle hysteresis for the two matrices. During the initial stage of trying different schemes to prepare the samples, it was observed that the curing temperature had a strong effect on the contact angles hysteresis for Sylgard® 184 coating but had no effect on the

contact angles hysteresis for RTV11 coatings. Both control samples of RTV11 and

Sylgard® 184 were cured at two temperatures: 25° C and at 100° C. For RTV11, the temperature had no effect on the advancing (103 °) and receding (95°) contact angles.

For Sylgard® 184, however, the advancing contact angle was about 110° at both

temperatures while the receding contact angle increased significantly from about 80° at

the lower temperature to about 95° at the higher temperature. In other words, for

Sylgard® 184, the contact angle hysteresis decreased from 30° at the lower curing

temperature to 15° at the higher curing temperature. Although it is not intended in the

current study to explore this effect in details, it could be anticipated that there were more

un-cross linked chains aggregated on the surface of Sylgard® 184 at the lower

temperature compared to the higher temperature, which would result in increasing the

contact angle hysteresis. For the purpose of enhancing the antifouling performance of the

silicone coating by means of minimizing its contact angle hysteresis, it might be better technically to process Sylgard® 184 at higher temperatures, although this might not be an

easy task practically. However, for the current study henceforth, all the results associated

with Sylgard® 184 were corresponding to samples cured at room temperature.

70 4.1.2 Effect on elastic modulus

The elastic modulus of a coating is a good measure of its bulk properties. Thus,

Sylgard® 184 coatings of various amounts of NaB (0 - 2 wt %) were subjected to elastic

modulus measurements, and the results are shown in Figure 4.2. For control NaB-free

Sylgard® 184 coating, the elastic modulus was 0.95 MPa, which was in agreement with

the literature value reported (Eddington et. al., 2003). As shown in Figure 4.2, after

incorporating NaB into Sylgard® 184, it was observed that the elastic modulus increased

slightly from those of control value. This slight variation of the modulus could be

attributed to the final distribution and aggregate size of the compound inside the bulk of

the matrix. As to be seen in the next section, sodium benzoate has a uniform distribution

with small aggregate size about 3 µm. Consequently, Sodium benzoate could behave

here as a fine reinforced filler that resulted in increasing the elastic modulus.

Nevertheless, the elastic modulus measurements for the above systems indicated that the

low content of NaB was insignificant in affecting the bulk properties of the silicone

coatings. Also, it confirmed that NaB had little effect on the curing behaviors of

Sylgard® 184 coatings, as the bulk modulus was expected to drop significantly for the

uncured, liquid like coating. To summarize, the incorporation of NaB into Sylgard® 184 coatings did not considerably affect the surface and bulk properties of the Sylgard® 184 coatings, suggesting that the foul-release property of Sylgard® 184 likely be retained.

3.0

2.4

1.8

1.2

0.6 Elastic Modulus (MPa)Elastic Modulus

0.0 00.511.522.5 wt% NaB in Sylgard® 184

Figure 4.2 Elastic modulus variations of NaB - entrapped Sylgard® 184 coating. The (solvent: polymer) ratio and the (water: acetone) ratio were respectively (20: 80) and (50: 50) by mass, which were fixed at all concentrations. Error for each data point (average over 6 measurements) is presented by the vertical line.

72 The elastic modulus of NaB-free RTV11 coating was also measured, and found to

be 1.56 MPa. This value is in good agreement with the literature value reported for

RTV11 (Kohl & Bolstes, 2001). The higher modulus of RTV11 as compared to that of

Sylgard® 184 is due to the fact that RTV11 has a high content of enforced fillers (32 wt%

® CaCO3) whereas Sylgard 184 do not have. For 1 wt% NaB/RTV11 coating, the elastic

modulus was not measured, but at this low concentration the modulus do not expected to differ two much from the corresponding value of NaB-free RTV11 coating. It was physically observed that the 1 wt% NaB-RTV11 coating was cured. This is supported by the fact that RTV11 is much more resistant to poisoning than Sylgard® 184 (the results

obtained for the other compound, capsaicin; support the last statement, as to be discussed

in Chapter 5).

4.2 Miscibility of NaB in silicones

After confirming that the coating was cured, it is necessary to analyze the

miscibility of the compound with the matrix, because it plays an important role for

compound distribution within the silicone coatings, and consequently on its leaching into

water. The miscibility was related to the aggregate size and distribution of the entrapped

antifouling compound. One advantage of Sylgard® 184, compared to RTV1, is that it is

highly transparent so any material entrapped could be observed easily by Optical

Microscopy. Therefore, to roughly examine the antifoulant/silicones miscibility; optical

microscopic images were taken for the bulk of Sylgard® 184 contained the incorporated

compound after drying off the solvents. For NaB/ Sylgard® 184 system, the preparation

73 conditions were systematically varied to investigate their effects on the morphological

structures. The parameters varied were: solvent composition (acetone/water ratio), solvent/polymer ratio, and concentration of NaB in the matrix (i.e. after drying off the solvent). To facilitate the comparisons, we selected the base case conditions to be: 50/50 water/acetone ratio, 20/80 solvent/polymer ratio, and 1 wt% NaB in the matrix, with all ratios in mass basis. Thus, when varying any one parameter, the other parameters were fixed at the base case conditions.

4.2.1 Effect of composition of the mixed solvent

The water/acetone ratio was increased from 20/80 water/acetone to 100 % water, keeping the other conditions unchanged (20/80 solvent/polymer, and 1 wt%

NaB/Polymer). The results are shown in Figures 4.3 and Table 4.2. For the samples prepared with 20-50 %, the particles had a narrow size distribution and there were no aggregate observed with a size greater than 15 µm. However, for the samples prepared with the solvent of 90/10 water/acetone and 100% water, the particles had a wide size distribution and more than 22% of the aggregates had sizes greater than 15 µm, indicating non-uniform NaB dispersion in the matrix. It is clear that adding acetone will help to uniformly disperse NaB in the matrix with smaller aggregates. However, it is not useful to decrease the water/acetone ratio less than a certain liming ratio, because below this ratio NaB will not be soluble in the mixed solvent and hence the aggregate size will not be fine and uniform. From literature, the maximum solubility of NaB in water is 0.555 g/g water, and NaB is practically insoluble in acetone (Bustamante et al., 2000).

(a) (b)

(e) (f)

Figure 4.3 Optical microscope (bright field) images (570 µm x 480 µm) of the bulk morphology of 1 wt % NaB/Sylgard® 184 matrix, prepared at different water/acetone ratios, keeping the solvent/polymer ratio fixed at 20/80. (a) 20/80 water/acetone; (b) 30/70 water/acetone; (c) 50/50 water/acetone; (d) 80/20 water/acetone; (e) 90/10 water/acetone; (f) 100% water. All the values are based on weight.

Table 4.2 Aggregate size distribution of 1 wt.% of sodium benzoate inside the (99 wt.%) bulk of Sylgard® 184 matrix, for samples prepared at different water/acetone ( W/A) mass ratios. The solvent/polymer ratio was fixed at 20/80 by mass.

Arithmetic Quadratic 0-5 µm 5-10 µm 10-15 µm > 15 µm mean mean Sample size range size range size range size range size size * * * * (µm) (µm) ** *** 20/80 2.0 (68.7) 7.2 (28.8) 11.1 (2.5) (0) 3.7 4.5 W/A 30/70 2.1 (74.2) 6.8 (24.2) 12.2 (1.6) (0) 3.4 4.1 W/A 40/60 2.0 (69.8) 7.3 (27.4) 12.2 (2.8) (0) 3.7 4.6 W/A 50/50 2.1 (65.8) 7.1 (30.1) 11.5 (4.1) (0) 4.0 4.8 W/A 80/20 2.5 (54.2) 7.5 (26.2) 12.7 (9.3) 19.6 (9.3) 6.4 8.3 W/A 90/10 2.6 (57.1) 7.1 (14.3) 12.3 (6.0) 54.6 (22.6) 15.6 26.4 W/A 100% 2.7 (62.8) 6.9 (5.1) 13.1 (3.8) 55.0 (28.2) 18.1 29.4 W

* The number outside the parenthesis is the average aggregate size evaluated in that particular size range. The number inside the parenthesis is the percentage of number of aggregate at that particular size range to the total number of aggregates. ** The arithmetic mean size is defined by summing the sizes of all aggregates and dividing by the total number of aggregates. 2 1/2 *** The quadratic mean size is defied as: (∑ di ni/ntot) , where di and ni are, respectively, the average size and the number of aggregates at that particular size range, and ntot is the total number of aggregates.

76 Based on this fact, the limiting acetone/water ratios for different sets of conditions are calculated and presented in Figure 4.4, which serves as a practical guide for preparing the samples.

2.0 1 wt% NaB/P; 20/80 S/P saturation line 1.5 5 wt% NaB/P; 20/80 S/P 7 wt% NaB/P; 20/80 S/P

1.0

0.5

mass NaB / mass water 0.0 0% 20% 40% 60% 80% 100% water/(water + acetone) mass %

Figure 4.4 Guidelines chart for preparing NaB-incorporated Sylgard® 184 coatings at different set of conditions. The saturation line “ — ” represents the maximum solubility of NaB in water (0.555 g NaB per 1 g of water). Above this line, NaB is not soluble in the mixed solvent (water + acetone), and hence the bulk entrapment method will not be feasible at this particular set of conditions. (Abbreviations: S: solvent; P: polymer). All values are in mass basis.

77 4.2.2 Effect of solvent/polymer ratio

The solvent/polymer ratio was increased from 10/90 to 50/50 solvent/polymer,

keeping the other conditions unchanged (50/50 water/acetone, and 1 wt% NaB/Polymer).

As shown in Figures 4.5 and Table 4.3, the solvent/polymer ratio of 20/80 was the

optimum ratio that resulted in the minimum aggregate size with the most uniform and

narrower distribution. Decreasing the solvent/polymer ratio to 10/90 resulted in having 8

% of the aggregates with sizes greater than 15 µm. On the other side, increasing the

solvent/polymer ratio to 30/70 – 50/50 resulted in having 14 - 19 % of the aggregates

with sizes greater than 15 µm.

4.2.3 Effect of NaB matrix loading

The wt% NaB/polymer was varied from 0.5 wt% to 5 wt%, keeping the other

conditions unchanged, and results are shown in Figure 4.6 and Table 4.4. While the size

of the aggregates (~ 3 - 4 µm) and the observed narrow size distribution were similar for

all concentrations, the number of aggregates increased and hence the distance between

aggregates decreased as the concentration increased. For all the samples of different NaB

matrix loadings, there were no aggregates observed with sizes greater than 15 µm. The aggregate size remained at its minimum value because the solvent/polymer ratio and water/acetone ratio were fixed at their optimum values (20/80 solvent/polymer and 50/50 water/acetone), and at these ratios, NaB was still soluble in the mixed solvent even for the samples of highest NaB matrix loading (5 wt% NaB/matrix) prepared.

(a) (b)

(e)

Figure 4.5 Optical microscope (bright field) images (570 µm x 480 µm) of the bulk morphology of 1 wt % NaB/Sylgard® 184 matrix, prepared at different solvent/polymer ratios, keeping the water/acetone ratio fixed at 50/50. (a) 10/90 solvent/polymer; (b) 20/80 solvent/polymer; (c) 30/70 solvent/polymer; (d) 40/60 solvent/polymer; (e) 50/50 solvent/polymer. All the values are based on weight.

Table 4.3 Aggregate size distribution of 1 wt.% of sodium benzoate inside the (99 wt.%) bulk of Sylgard® 184 matrix, for samples prepared at different solvent/polymer (S/P) mass ratios. The water/acetone ratio was fixed at 50/50 by mass.

Arithmetic Quadratic 0-5 µm 5-10 µm 10-15 µm > 15 µm mean mean Sample size range size range size range size range size size * * * * (µm) (µm) ** *** 10/90 3.1 (42.5) 6.9 (37.0) 12.3 (12.7) 18.6 (8.1) 6.9 8.3 S/P 20/80 2.1 (65.8) 7.1 (30.1) 11.5 (4.1) (0) 4.0 4.8 S/P 30/70 2.8 (58.9) 7.3 (19.8) 12.2 (7) 33.3 (14.3) 8.7 13.6 S/P 40/60 2.8 (42.6) 7.2 (29.5) 12.3 (8.2) 31.0 (19.7) 10.4 14.8 S/P 50/50 3.2 (38.4) 7.3 (36.6) 12.1 (11.6) 39.2 (13.4) 10.5 15.7 S/P

(a) (b)

(e) (f)

Figure 4.6 Optical microscope (bright field) images (570 µm x 480 µm) of the bulk morphology of NaB/ Sylgard® 184 matrix of different NaB concentrations, keeping the solvent/polymer ratio and the water/acetone ratio fixed at 20/80 and 50/50, respectively. (a) 0.5 wt% NaB/ Sylgard® 184; (b) 1 wt% NaB/ Sylgard® 184; (c) 2 wt% NaB/ Sylgard® 184; (d) 3 wt% NaB/ Sylgard® 184; (e) 4 wt% NaB/ Sylgard® 184; (f) 5 wt% NaB/ Sylgard® 184; All the values are based on weight.

Table 4.4 Aggregate size distribution of sodium benzoate inside the bulk of Sylgard® 184 matrix, for samples prepared at different NaB matrix loading (wt% NaB in the matrix). The water/acetone ratio was fixed at 50/50 by mass. The solvent/polymer ratio was fixed at 20/80 by mass.

Arithmetic Quadratic 0-5 µm 5-10 µm 10-15 µm > 15 µm mean mean Sample size range size range size range size range size size * * * * (µm) (µm) ** *** 0.5 wt 2.4 (67.1) 7.1 (28.8) 12.5 (3.2) (0) 4.1 4.8 % 1 wt % 2.1 (65.8) 7.1 (30.1) 11.5 (4.1) (0) 4.0 4.8

2 wt% 2.2 (70.4) 6.5 (27.7) 11.8 (1.9) (0) 3.6 4.2

3 wt % 2.2 (70.8) 6.8 (27.0) 12.1 (2.2) (0) 3.6 4.4

4 wt % 1.7 (71.9) 6.7 (24.7) 11.8 (3.4) (0) 3.3 4.2

5 wt % 2.0 (62.4) 6.7 (32.6) 11.7 (5.1) (0) 4.0 4.9

82 4.3 Thermodynamic analysis for the miscibility study

4.3.1 Prediction by the Flory-Huggins theory

The miscibility of a substance (1) with a polymer (2) can be roughly predicted by the substance-polymer interaction parameter, χ12. According to the “Flory-Huggins”

theory, χ12 is given by (Flory, 1953):

2 χ12 = (V1/RT) (δ1 – δ2) (4.1)

where V1 is the molar volume of the smaller specie, the substance in our case, R and T

are respectively the ideal gas constant and the temperature, and δ1 and δ2 are respectively

the solubility parameters of the substance and the polymer. A smaller χ12 indicates a higher chance that the system would be miscible, and a value of χ12 < 0.5 is the Flory-

Huggins criterion for a solvent/polymer system to be completely miscible. Table 4.5 summarizes the χ12 values of various substances and silicone (i.e. PDMS). First, the χ12 values for NaB/PDMS system is 17.6, indicating that NaB is likely not miscible with

PDMS, as observed experimentally. For the two solvents used in NaB/PDMS system, it is clear that water is highly immiscible with PDMS (χ12 = 8.0) while acetone has some

miscibility with PDMS (χ12 ~ 0.9), hence the size of sodium benzoate aggregates

increased as the water content (in the mixed solvent of water/acetone) increased, as

observed experimentally.

83 Table 4.5 Physical parameters of relevance importance to the miscibility of NaB/PDMS. V and δ are the molar volume and the solubility parameter of the material, respectively. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS*.

Boiling V δ ∆δ12 Material 3 ½ ½ point (cm /mol) (MPa) (MPa) χ12 (oC) Water 18.2 47.9 a 33.0 8.000 100

Acetone 74.0 20.3 a 5.4 0.871 56

NaB 100.1 35.8 b 20.9 17.648 -

a, b Values obtained from (Rodriguez, 1989) and (Bustamante et. al., 2000), respectively * For PDMS, the literature value of its solubility parameter (14.9 MPa½, from: Rodriguez, 1989) was used

The interaction parameter calculated by equation 4.1 provides a rough prediction for

the miscibility of the mixture. However, the effect of composition of the mixture is not

explicitly accounted for by this method. In a more general form, the miscibility of the

mixture can be predicted by calculating the free energy of mixing (∆Gmix) for the system

at all compositions. As thermodynamics states, the system becomes more heterogeneous

as ∆Gmix increases, and becomes spontaneously homogeneous when ∆Gmix is less than zero. NaB/water/acetone/PDMS is a quaternary system. According to the “Flory-

Huggins” theory, the free energy of mixing for a quaternary system is given by ((Flory,

1953) :

84 ∆Gmix / nkT = Φ1 ln Φ1 + (Φ2 / R21) ln Φ2 + (Φ3 / R31) ln Φ3 + (Φ4 / R41) ln Φ4

+ Φ1 Φ2 χ12 + Φ1 Φ3 χ13 + Φ1 Φ4 χ14 + (Φ2 / R21) Φ3 χ23

+ (Φ2 / R21) Φ4 χ24 + (Φ3 / R31) Φ4 χ34 (4.2)

where “∆Gmix / nkT” in equation (4.2) is the dimensionless free energy of mixing per unit segment or per site of the lattice model, k and T are respectively the Boltzman constant and temperature, n is the total number of sites in the lattice model, Φi is the volume fraction of component i, and Ri1 is the size ratio of molecule i to that of molecule 1, given by:

Ri1 = Vi / V1 (4.3)

where Vi and V1 are the molar volume of component i and component 1, respectively.

Component 1 should be chosen to be the one with the smallest molecular size. For our system, the following subscripts were used: 1, water; 2, acetone; 3, NaB; and 4, PDMS.

In equation (4.2), the first four terms in the right-hand side represent the configurational entropy of mixing whereas the last five terms in the right-hand side represent the enthalpic terms. In equation (4.2), χij is the interaction parameter between component i and j in the system, given by:

2 χij = (Vi/RT) (δi – δj) (4.4)

where δi and δj are respectively the solubility parameters of component i and j.

85 The free energy of mixing of the NaB/acetone/water/PDMS mixtures - predicted by equation (4.2) - is shown in Figure 4.7 for samples prepared at different preparation conditions. First of all, ∆Gmix is greater than zero, indicating that NaB-PDMS system does not spontaneously mix. As shown in figure 4.7, with other parameters fixed, ∆Gmix

increases as the water content in the mixed solvent increases. Also, under a particular

ratio of water/acetone (other than 100% acetone), ∆Gmix increases as the amount of

solvent in the system increases. The increase of ∆Gmix simply means that the mixture

becomes less miscible, which will lead to increased aggregate size. Also, as shown in

Figure 4.9, increasing the wt% NaB/polymer from 1 wt% to 5 wt%, with other

parameters fixed, does not considerably affect ∆Gmix, which means that the aggregate size will no be affected considerably. In summary, the values of ∆Gmix predicted by the

Flory-Huggins model (equation 4.2) suggest that the water/acetone ratio and the

solvent/polymer ratio have more profound effect on ∆Gmix (and hence on the aggregate

size) than the NaB/polymer ratio, as long as NaB is soluble in the mixed solvent.

4.3.2 Modification of the Flory-Huggins theory to include electrostatic contribution and concentration-dependent interaction parameters

Although the Flory-Huggins theory is the most well-known and widely used

theory in polymer thermodynamics, it is now well-believed that it has certain weakness.

One source of error in the F-H thermodynamic model is in the magnitude of interaction

parameter, χij. It is now believed that χij can be qualitatively described in a more general

form as:

χij= χij, d + χij, fv + χij, sp (4.5)

86 1.8 50/50 W/A 20/80 W/A (a) 30/70 W/A

FH 1.5 40/60 W/A 80/20 W/A 1.2 90/10 W/A 100 % W / nkT) nkT) /

mix 0.9 G ∆ ( 0.6

0.3

0.0 0.0 0.2 0.4 0.6

(Φ1 + Φ2)

1.0 (b)

FH 0.8

0.6 / nkT) nkT) / mix

G 0.4 ∆ ( 1 wt% NaB/Polymer 0.2 5 wt% NaB/Polymer 0.5 wt% NaB/Polymer 3 wt% NaB/Polymer 0.0 0.0 0.2 0.4 0.6

(Φ1 + Φ2)

Figure 4.7 The miscibility trends for NaB/acetone/water/PDMS mixtures, for all possible conditions, as predicted by the original Flory-Huggins (FH) model (equation 4.2). (a) Effect of the water/acetone (W/A) ratio; parameters fixed: 1 wt% NaB/polymer. (b) Effect of the NaB matrix loading; parameters fixed: 50/50 water/acetone. All the values shown in the legends are based on weight.

87 where χij, d accounts for the dispersion interactions, χij, fv accounts for the interactions

resulted from the free volume effect, and χij, sp accounts for the specific interactions such

as acid-base interactions or H-bonding. However, the “Flory-Huggins” theory only considers the dispersion interactions and totally neglects the other two types of

interactions, which might result in errors in estimating the values of χij and consequently

some errors could result in the value of ∆Gmix. For example, neglecting the free volume

effect is based upon the assumption that there is no volume change upon mixing, which is

rarely satisfied for mixing large macromolecules with solvents or additives of much

smaller size. Specific interactions could exist in our system and affect the value of the

free energy of mixing. One specific interaction that could be identified here is the

electrostatic interactions, because our system is in fact containing electrolyte solution

(NaB, which dissociates into salt ions in water). Other specific interactions are the polar

interactions for water/acetone and NaB/water. Another source of error is the assumption

that the interaction parameters are not functions of concentrations. All these factors

together could affect the miscibility prediction. In the current subsection, we are

modifying the Flory-Huggins thermodynamic model (equation 4.2) to include

electrostatic contribution term to the total free energy and concentration-dependent

interaction parameters.

The NaB aqueous solution is an electrolyte solution. Thus, when mixed with

polymers, specific interaction like electrostatic interaction will exist that could change the

phase behavior of the mixture. In this section, we present a simple lattice thermodynamic

model that accounts for this electrostatic interaction. The model was originally

88 developed by Hino et al. (1998) for a ternary (water, salt, and polymer) system. We will

extend the equation to a quaternary (water, salt, polymer, and organic solvent) system,

and then apply the resulting model for our particular NaB/water/acetone/PDMS

quaternary system.

Considered a ternary system consisting of water (1), salt (2), and polymer (3),

Hino et al. (1998) proposed the free energy of the mixture as:

FH DH ∆Gmix / nkT = ∆G / nkT + ∆G / nkT (4.6)

where the (∆GFH / nkT) term is the Flory-Huggins expression for the free energy of mixing, given by an expression similar to equation (4.2), and the (∆GDH / nkT) term is the

electrostatic contribution to the free energy of mixing, given by the Pitzer extension

(Pitzer, 1973) of the Debye-Huckel function. Specifically, (∆G DH / nkT) is given by

(Hino et al., 1998):

DH 1/2 ∆G / nkT = Φw (Mw/1000) [— A (4 I / b) ln (1 + b I )] (4.7)

where Φw and Mw are, respectively, the volume fraction and molecular weight of water,

A and b are constants (A = 0.392, b = 1.2), and I is the ionic strength of the mixture,

expressed as:

I = 0.5 [Φsalt (1000/Mw) / (1 — Φw)] │zM zX│ (4.8)

89 where Φsalt is the volume fraction of the salt, and zM and zX are, respectively, the valences

of the cation and anion. In the above model (equation 4.6-4.8), several assumptions are

employed. The polymer is assumed to be neutral. The ternary system is assumed to be

incompressible. It is also assumed that the salt is completely dissociated into ions.

In the following, we extend the above model (equation 4.6-4.8) to our quaternary

system consisting of water (1), acetone (2), NaB (3), and PDMS (4). The complete

equation is:

1/2 ∆Gmix / nkT = Φ1 (W/1000) [— A (4 I / b) ln (1 + b I )]

+ Φ1 ln Φ1 + (Φ2 / R21) ln Φ2 + (Φ3 / R31) ln Φ3 + (Φ4 / R41) ln Φ4

* + Φ1 Φ2 χ12 + Φ1 Φ3 χ13 + Φ1 Φ4 χ14 + (Φ2 / R21) Φ3 χ23

* + (Φ2 / R21) Φ4 χ24 + (Φ3 / R31) Φ4 χ34 (4.9)

Another new feature of the new model (equation 4.9) is the introduction of the

* concentration-dependent interaction parameters for water-acetone pair, χ12 , and acetone-

* PDMS pair, χ24 . These two parameters are obtained from the literature for experimental

data collected specifically for these two particular systems. Therefore, they are expected

to give better values than the values predicted by equation (4.1). The reason why specifically the above pairs (water/acetone, and acetone/PDMS) are considered

concentration-dependent is because acetone is highly miscible in water and acetone has

some miscibility in PDMS. On the other hand, the interaction parameter for water-

PDMS system (χ14) is considered constant and calculated by equation (4.1). This is

90 because water is highly immiscible with PDMS, and for this situation this assumption is

reasonable (Yilmaz and McHugh, 1998). Similarly, the interaction parameters for NaB-

water system (χ13), NaB-acetone system (χ23), and NaB-PDMS system (χ34) are considered constants and calculated by equation (4.1). Experimentally, we incorporated

NaB in PDMS up to only 5 wt% NaB/PDMS, and therefore Ф3 was always very small in

equation (4.9). Consequently, the terms Φ1 Φ3 χ13, (Φ2 / R21) Φ3 χ23, and (Φ3 / R31) Φ4 χ34

are negligible in equation (4.9), and hence the assumption of constant (χ13 , χ23 , and χ34)

will not strongly affect the results of equation (4.9).

For acetone-water system, the experimental correlation for the interaction

parameter is (Yilmaz and McHugh, 1986):

* χ12 = 0.661 + 0.417 / (1 — 0.755 Ф2) (4.10)

and for acetone-PDMS system the experimental correlation is (Singh et. al, 1998):

* 2 χ24 = 16.0 — 34.9 (1 — Ф2) + 20.7 (1— Ф2) (4.11)

It should be noticed that the correlation given here for acetone-PDMS was obtained for a set of data in the range of 0 — 20 volume % acetone. Fortunately, this is the range of our experimental conditions for preparing the samples. Therefore, this correlation can be safely used in our model in this range.

91 To criticize the validity of equation (4.1) for acetone-PDMS system and acetone-

water system, the following can be mentioned. The experimental value for acetone-

PDMS interaction parameter is in the range of 1.3 to 1.9 ((Singh et. al, 1998), indicating

that acetone is not a good solvent for PDMS. To verify this experimentally, we did a

simple experiment of dissolving a small amount of PDMS in acetone (1 wt%

PDMS/acetone mixture), and it had been observed that the mixture was not miscible even

after two days of continuous stirring. However, the prediction of the interaction

parameter by equation (4.1) gives a value of 0.87, which is lower than the measured value, mainly because of the reason described by equation (4.5). Similarly, the experimental value for the water-acetone interaction parameter is in the range of 1 to 2

(Yilmaz and McHugh, 1986), whereas it is 5.5 when predicted by equation (4.1). Again, this predicted value was higher than the measured value, since water and acetone are highly miscible with each other, and the reason for the disagreement is also as described by equation (4.5). Therefore, it is always better, whenever possible, to use experimental values for the interaction parameters.

In summary, it should be noticed that the new model (equation 4.9) as well as the conventional Flory-Huggins model (equation 4.2) contain no adjustable parameters. Two

* * of the binary interactions, χ12 and χ24 , are concentration-dependents and obtained form

empirical correlations, and the others are concentration-independent and calculated by

equation (4.1). In summary, in equation (4.9), the first term of the right hand side is the

electrostatic contribution term (which is always negative). The remaining terms are

* Flory-Huggins terms but with two concentration-dependent interaction parameters (χ12

92 * and χ12 ). The second to the fifth terms of the right-hand side are the entropic terms

(always negative), and the sixth to the eleventh terms are the enthalpic terms (always

positive). Therefore, compared to equation (4.2), the free energy of mixing could be

negative if the favorable electrostatic term is high enough to overcome the unfavorable

enthalpic terms.

Equation (4.9) is evaluated for our particular system and compared to the results

obtained by the conventional Flory-Huggins model (equation 4.2), and the results are

shown in Figure 4.8. Compared to the trends obtained by equation (4.2), the change in

the trends is observed, but does not change the earlier conclusions that the mixtures are

immiscible because the free energy of mixing is positive at all the conditions. This is

expected because the unfavorable enthalpic interaction between water and PDMS is too

large to be compensated by the favorable electrostatic interaction. The observed

difference between the trends obtained by the new model and the Flory-Huggins model is

due mainly to the effect of introducing the concentration-dependent interaction

* * parameters, χ12 and χ24 , and not because of the introduction of the electrostatic term.

Nevertheless, the new model (equation 4.9) is useful because it isolates most of the

possible interactions existing in the system and thus gives us practical guidelines to tailor

the system. For example, we could slightly modify the polymer from being strongly

hydrophobic to be partially hydrophobic. In this case, the unfavorable enthalpic terms

will be reduced and the system will be more miscible.

50/50 W/A 1.8 20/80 W/A (a) 30/70 W/A 1.5 40/60 W/A 80/20 W/A 1.2 90/10 W/A / nkT / 100 % W

mix 0.9 G ∆ 0.6

0.3

0.0 0.0 0.2 0.4 0.6

(Φ1 + Φ2)

1.0 (b)

0.8

/ nkT / 0.6 mix G

∆ 0.4

1 wt% NaB/Polymer 0.2 5 wt% NaB/Polymer 0.5 wt% NaB/Polymer 3 wt% NaB/Polymer 0.0 0.0 0.2 0.4 0.6

(Φ1 + Φ2)

Figure 4.8 The miscibility trends for NaB/acetone/water/PDMS mixtures for all possible conditions, as predicted by the new model (equation 4.9). (a) Effect of the water/acetone (W/A) ratio; parameters fixed: 1 wt% NaB/polymer. (b) Effect of the NaB matrix loading; parameters fixed: 50/50 water/acetone. All the values shown in the legends are based on weight

94 4.3.3 Comparison between the theoretical miscibility trends and the experimental morphology trends

In this subsection, we are making comparison between the experimental

morphology trends (aggregate size of NaB as function of preparation conditions) and the

theoretical miscibility trends (∆Gmix as function of preparation conditions), in an attempts to explain the role of thermodynamics on the aggregate size and also to compare between accuracy of the Flory-Huggins model and the new model. To do this, ∆Gmix calculated by both models is plotted as function of one preparation parameter, keeping the other parameters fixed. At the same time, the insert in the same plot shows the experimentally measured aggregate size as a function of this preparation parameter. The results are shown in Figure 4.9. As can be seen from Figure 4.9, both models do capture qualitatively most of the important effects of the preparation conditions on the aggregate size of NaB in PDMS. In addition, both models correctly predict that NaB matrix loading has no strong effect on the aggregate size, as long as NaB is soluble in the mixed solvent. To compare between the accuracy of the two models, the following can be mentioned. Experimentally, the aggregate size increased strongly as the solvent/polymer ratio increased from 20/80 to 30/70, but after the ratio of 30/70 the increase was little

(8.7, 10.4, and 10.5 µm for samples prepared at 30/70, 40/60, and 50/50 solvent/polymer ratio, respectively). The trend for ∆Gmix calculated by the new model in this

solvent/polymer ratios range is more similar to the experimental morphology trend than

the trend of ∆Gmix calculated by the Flory- Huggins model (see Figure 4.9b). Therefore,

the new model is shown to be more accurate than the Flory-Huggins model.

95 20 (a) 0.9 15

10 d (um) 5

0.7 0 0 25 50 75 100 water/(water+acetone) wt % / nkT

mix 0.5 G ∆ 0.3 F-H model new model

0.1 0% 20% 40% 60% 80% 100% water/(water+acetone) wt%

0.9 10 (b)

d (um) 5 0.7 0 0204060

/ nkT / solvent/(solvent+polymer) wt %

mix 0.5 G ∆ 0.3 F-H model new model

0.1 0% 10% 20% 30% 40% 50% solvent/(solvent+polymer) wt%

Figure 4.9 The free energy of mixing (∆Gmix/nkT) for NaB/acetone/water/PDMS mixtures prepared at different conditions, as predicted by two models: the Flory-Huggins (F-H) model (equation 4.2), and the new model (equation 4.9). (a) Effect of the water/acetone ratio, parameters fixed: 20/80 solvent/polymer and 1 wt% NaB/polymer. (b) Effect of the solvent/polymer ratio, parameters fixed: 50/50 water/acetone and 1 wt% NaB/polymer. All the values are based on weight. The preparation conditions described here correspond to the actual conditions for the morphology experiments performed in the current study. The insert in each plot represents the corresponding experimental morphology trend.

9 (c) 0.9 6

d (um) 3

0.7 0 0246 wt % NaB/Polymer nkT /

mix 0.5

G ∆ 0.3 F-H model new model 0.1 0% 1% 2% 3% 4% 5%

wt% NaB/Polymer

Figure 4.9 The free energy of mixing (∆Gmix/nkT) for NaB/acetone/water/PDMS mixtures prepared at different conditions, as predicted by two models: the Flory-Huggins (F-H) model (equation 4.2), and the new model (equation 4.9). (c) Effect of NaB matrix loading, parameters fixed: 50/50 water/acetone and 20/80 solvent/polymer. The solvent is defined here as water + acetone + NaB. All the values are based on weight. The preparation conditions described here correspond to the actual conditions for the morphology experiments performed in the current study. The insert in each plot represents the corresponding experimental morphology trend. (continued)

However, both the Flory-Huggins model and the new model could not explain

two experimental observations. First, the aggregate size at the smallest experimental

solvent/polymer ratio used (10/90) was 6.9 µm, which was larger than the corresponding

aggregate size (4.0 µm) at 20/80 solvent/polymer. However, ∆Gmix at the (10/90) ratio is lower than ∆Gmix at the (20/80) solvent/polymer ratio. Second, experimentally, the

97 water/acetone ratio in the range of 20/80 to 50/50 water/acetone did not affect

considerably the aggregate size (the aggregate size was in the range 3-4 µm), but it affected the aggregate size sharply after the ratio of 80/20 water/acetone (the aggregate

size was 15.6 for samples prepared with 90/10 water/acetone ratio). However,

theoretically, ∆Gmix increases almost linearly as the water/acetone ratio increases. The

reasons will be discussed in the next paragraph. Before that, it should be noticed that the

comparison made in Figure 4.9 between the experimental morphology curves and the theoretical miscibility curves is qualitative because the y-axis is not the same. Our objective here is to prove that as ∆Gmix for any set of preparation conditions increases,

the aggregate size increases, which were true for most of the sets of the preparation

conditions.

The difference between the predictions and the above two observations could be

explained as follow. The values of ∆Gmix shown in Figure 4.9 and predicted by equations

(4.2) and (4.9) are the description for a situation of thermodynamic mixing of an isolated

system. That is, the NaB/water/acetone/PDMS system is placed inside a closed vial that

is perfectly insulated, the mixture is well-mixed, and then the mixing is stopped and

enough time is allowed for the system to equilibrate. ∆Gmix shown in Figure 4.9 is,

hence, of what is described above. Under the situation described above, the total

composition of the entire system is presumably the same, in other words, water and

acetone do not evaporate in this isolated system. However, this was not the case for our

NaB/PDMS morphology experiments. That is, in the process of incorporating NaB into

PDMS by the solvent blending technique and during the process of heating and drying the system, the composition of the system was continually changing due to evaporation of

98 water and acetone because the system here was open to the environment. This resulted in the nucleation and growth of new solid phase (solid NaB particles) as the solvent became more concentrated, until the solvent was completely dried off. This dynamic drying step would strongly affect the final morphology of the NaB/PDMS composites, but equation

4.2 and equation 4.9 do not consider such a dynamic effect. In order to account for this dynamic effect, a mass transfer model has to be combined with the thermodynamic model. The mass transfer model would need to account for the phase change

(vaporization) of acetone and water, and to account for the phase change (solidification) of NaB, whereas PDMS can be considered as an inert material (i.e. no phase change for

PDMS) at all time. As more water vaporizing, the NaB volume fraction will be closer to its solubility limit in water, and hence start to precipitate out. Once precipitation occurs, the unfavorable enthalpic terms (in equations 4.2 and 4.9) will increase and the favorable entropic terms (in equations 4.2 and 4.9) will decrease, both will result in the enhancement the phase separation of the system. It has been shown that when mixing solid particles with solvent and polymer, the entropy-driven phase separation is possible even for athermal (i.e. zero χij) system (Schaink and Smith, 1996). Since water and acetone are not good solvents for PDMS, the system can not form a homogeneous phase.

Instead, a two phase or even a multi-phase system is expected, and the development for a multi-phase transport phenomena model for this system and combining it with the thermodynamic model is too complicated and beyond the scope of the current study.

Nevertheless, the rough prediction based on the thermodynamic models presented here did qualitatively describe most of the features of our system, and still useful as a preliminary guide for selecting the components of the quaternary system.

99 4.4 Leaching evaluation

The NaB molecules incorporated into the coatings should leach out in order for

them to antifoul. Therefore, the NaB-entrapped silicone coatings were subjected to

leaching studies in static cells. The effects of different preparation conditions on the

leaching behaviors are presented in the following subsections. The parameters varied

were: solvent composition (acetone/water ratio), solvent/polymer ratio, wt% NaB in the

matrix, and type of the matrix. We selected the base case conditions to be: 50/50

water/acetone ratio, 20/80 solvent/polymer ratio, 1 wt% sodium benzoate in the matrix,

and Sylgard® 184 as the base-case matrix. Thus, when varying any one parameter, the

other parameters were fixed at the base case conditions.

4.4.1 Effect of composition of the mixed solvent

NaB-incorporated Sylgard® 184 samples prepared at different water/acetone

ratios, whose bulk morphology are shown in Figure 4.3, were subjected to leaching

evaluations, and the results are shown in Figure 4.10. For all these samples, the

solvent/polymer ratio was fixed at 20/80, and the wt. % NaB in the coating was fixed at 1

wt. %. As shown in Figure 4.10, the general trend, regardless water/acetone ratio used,

was observed. Sodium benzoate leached out in two stages, a first fast stage occurred in

the initial few days and followed with a second steady stage having a much slower rate.

This leaching behavior is not unique to the sodium benzoate/Sylgard® 184 coating. In

100

500

400

) 40 2 ) 300 2 20/80 W/A 20 50/50 W/A 200 Q (µg/cm 80/20 W/A 90/10 W/A Q (µg/cm 0 100% W 0 5 10 15 20 100 time (day)

0 0 5 10 15 20 time (day)

Figure 4.10 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at different water/acetone ratios, keeping the other conditions unchanged (20/80 solvent/polymer, and 1 wt% NaB/matrix). All the values are based on weight. Solid lines are for showing the trends. (Abbreviations: W: water; A: acetone). Error for each data point (average over 2 batches) is presented by the vertical line.

101 fact, it is a characteristic of hydrophobic monolithic coatings ((Fan and Singh, 1989),

which sodium benzoate/Sylgard® 184 coating belongs to. Back to Figure 4.10, the cumulative leaching amount increased as the water/acetone ratio increased, and this increase was very sharp at the water/acetone ratio higher than 80/20. This is directly related to the size of the aggregates, as to be shown later.

4.4.2 Effect of solvent/polymer ratio

NaB-incorporated Sylgard® 184 samples prepared at different solvent/polymer

ratios, whose bulk morphology are shown in Figure 4.4, were subjected to leaching

evaluations, and the results are shown in Figure 4.11. For all these samples, the

water/acetone ratio was fixed at 50/50, and the wt. % NaB in the coating was fixed at 1

wt. %. As shown in Figure 4.11, at any time interval, the cumulative amount leached out

was highest for samples prepared at 30/70 solvent/polymer ratio, lowest for samples

prepared at 20/80 ratio, and intermediate for samples prepared at 10/90 solvent/polymer

ratio. This is also directly related to the size of the aggregates, as to be shown later. This

also confirmed that the solvent/polymer ratio of 20/80 was the optimum solvent/polymer ratio that resulted in minimum aggregate size and consequently in minimum leaching.

4.4.3 Effect of NaB matrix loading

To explore the effect of NaB loading in the coating, Sylgard® 184 sheets of

different concentrations (form 0.5 to 5 wt% NaB in Sylgard® 184) were prepared and

102

100

80 ) 2

60 10/90 S/P

40 20/80 S/P

Q (µg/cm 30/70 S/P 20

0 5 10 15 20

time (day)

Figure 4.11 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at different solvent/polymer ratios, keeping the other conditions unchanged (50/50 water/acetone, and 1 wt% NaB/matrix). All the values are based on weight. (Abbreviations: S: solvent; P: polymer). Error for each data point (average over 2 batches) is presented by the vertical line.

103

1 wt%

) 60 2 2 wt%

3 wt%

40 4 wt%

5 wt%

Q (µg/cm 20 0.5 wt%

0 6 12 18 24 30

t (day)

Figure 4.12 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at different wt% NaB/matrix, keeping the other conditions unchanged (50/50 water/acetone, and 20/80 solvent/polymer). All the values are based on weight. Error for each data point (average over 2 batches) is presented by the vertical line.

104 subjected to leaching study, and the results are shown in Figure 4.12. For all the above samples, the water/acetone ratio and the polymer/solvent ratio were fixed at 50/50 and

20/80, respectively. As shown in Figure 4.12, the cumulative amount leached out increased as wt% of NaB in the matrix increased. However, the increase was quite gradual rather than to be very sharp. This is also related to the aggregate size, as it is shown in the previous section that the water/acetone ratio and the solvent/polymer ratio had stronger effect on the aggregate size than the NaB/polymer ratio, as long as NaB was soluble in the mixed solvent.

4.4.4 Effect of type of the silicone matrix

To explore this effect, NaB was incorporated into Sylgard® 184 and RTV11 coatings at the same preparation conditions (50/50 water/acetone, 20/80 solvent/polymer, and 1 wt% NaB in the matrix), and subjected to leaching evaluations, and the results are shown in Figure 4.13. For sodium benzoate / Sylgard® 184 coatings, slow leaching was observed compared to RTV11, with the leached percentages to its initial mass of 6.7, 7.3 and 9.8 %, respectively, after 1 week, 1 month, and 4 months being in water. Changing the coating carrier to RTV11 did considerably affect the leaching. For 1 wt% sodium benzoate/RTV11 after 1 week, 1 month and 4 months of immersion, it was found that about 16%, 31% and 88% of the initial mass, respectively, had leached out. As compared to 1 wt% sodium benzoate / Sylgard® 184 coatings prepared under the same conditions, the leaching rate (slope from the curve) for sodium benzoate from RTV11 was about 23 times and 3 times higher during the slow and fast leaching periods, respectively.

105

250

200 ) 2 150

100 Q (µg/cm 50

0 0 6 12 18 24 30 time (day)

Figure 4.13 Cumulative leaching of NaB from its incorporated silicone coating: Sylgard® 184 (U) or RTV11 (▲). The common solvent used was 50/50 water/acetone by weight. The initial concentration of NaB in both coatings was kept constant at 1 wt%., and the solvent/polymer ratios was kept constant at 20/80 by weight for both combinations.

106 Therefore, the effect of changing the matrix type on NaB leaching was clearly

observed, where the leaching of NaB from RTV11 was higher than that from Sylgard®

184. This could be attributed to the following reasons. First, partial degradation and

erosion of the RTV11 matrix in water, which were confirmed experimentally by Bullock et. al. (1999) and Brady (2000), could be contributing and facilitating the leaching.

Wynne et al. (2000) evaluated two types of PDMS coatings (RTV11, and an in-house

unfilled hydrosilation cured PDMS), and they found that the hydrosilation cured PDMS

was very stable in water, whereas RTV11 was not stable in water with a mass loss of

about 0.8 % after 30 days of immersion. This mass loss from RTV11 was attributed to

the leaching of the RTV11 filler (RTV11 has a high content of inorganic fillers including

CaCO3 (32 wt.%)), and was also attributed to the continuous loss of small amounts of

RTV11 constituents, such as Me2SiO, other than CaCO3 (Bullock et. al., 1999). Second,

as mentioned before, RTV11 has a high content of inorganic fillers including CaCO3 (32 wt. %), which has a pore volume of 0.1─ 0.8 cm3/g (Wypych, 1999). At this high filler

content, filler agglomeration, which is a result of incomplete dispersion or flocculation, is

highly possible, which could lead to “voids” in-between the filler particles and at the

filler/polymer interface (Wel and Adaned, 1999). Thus, these voids allow water

molecules to seep into the silicone matrix more easily through these empty spaces and

carry the dissolved antifoulants molecules with them as they leave the coating. Third, the

reported solubility of water in PDMS is 7000 ppm and 700 ppm for filled and unfilled

silicone, respectively (Banerjee et al., 1997), which implies that the water uptake of

RTV11 is higher than that of Sylgard® 18 matrix. The finding of the current study on the

effect of the matrix type on NaB leaching was supported by the leaching results obtained

107 for incorporating another compound (tannic acid) into Sylgard® 184 and RTV11 coatings

at the same preparation conditions, where the same matrix type effect was observed: the

leaching of tannic acid from RTV11 was higher than that from Sylgard® 184 (the results for tannic acid are to be discussed in chapter 6). The matrix type effect observed in the current study was also in agreement with the data reported by Barrios (2004) for the leaching of zosteric acid form Sylgard® 184 and RTV11 coatings, where he reported that

the leaching of zosteric acid from RTV11 was about 10 times higher than the leaching

from Sylgard® 184 (Barrios (2004)).

4.4.5 Empirical correlations for the leaching rate of NaB from Sylgard® 184

Apart from the specific properties of the coating carrier, the preparation

conditions may also affect the morphological structures and the distribution of the

antifouling compound in the matrix, which can play important roles in leaching when carriers are immiscible with the antifoulant compounds. To explore this effect in more

details, the preparations conditions were systematically varied for a particular

combination (NaB and Sylgard® 184). For each set of conditions, the bulk morphology was analyzed before immersing the coating in water (results are shown in section 4.2).

Then, the cumulative leaching in water was measured (results are shown in sections 4.4.1

– 4.4.3). Consequently, the leaching rate was obtained as the slope of the cumulative leaching curve for each set of preparation conditions. The slope for the cumulative leaching data in the 0-2 day immersion period is defined as the initial leaching rate. The slope for the cumulative leaching data in the 14-30 day immersion period is defined as

108 the steady leaching rate. Then, attempt was made to correlate the leaching rate to the

preparation conditions, and the results are shown in Figure 4.14. Two factors of the preparation conditions were considered here: the size of the NaB aggregate and the NaB matrix loading. For the first factor, the NaB matrix loading was fixed at 1 wt%

NaB/matrix, whereas the aggregates size was varied because of varying the water/acetone

ratio and the solvent/polymer ratio. For the second factor, the NaB matrix loading was

varied from 0.5 wt% to 5 wt%, whereas the aggregate size was basically fixed at its

minimum narrower distribution by fixing the water/acetone ratio and the solvent/polymer

ratio at their optimum ratios (50/50 water/acetone and 20/80 solvent/polymer).

As shown in Figure 4.14a, the leaching rate sharply increased as the NaB

aggregate size increased and linear correlations were obtained. For the initial leaching

rate, the correlation obtained was: F (µg/cm2-day) = 3.54 t (day) – 6.30, with R2 value of

0.97. For the steady leaching rate, the correlation obtained was: F (µg/cm2-day) = 0.49 t

(day) – 2.16, with R2 value of 0.91. On the other side, as shown in Figure 4.14b, the

leaching rate gradually increased as the NaB matrix loading increased, and linear

correlations were also obtained. For the initial leaching rate, the correlation obtained

was: F (µg/cm2-day) = 1.75 t (day) + 5.59, with R2 value of 0.91. For the steady leaching rate, the correlation obtained was: F (µg/cm2-day) = 0.02 t (day) + 0.12, with R2 value of

0.79. In summary, by comparing Figure 4.14a to Figure 4.14b at the same scale for the y-

axis, it could be concluded that the increase of the NaB matrix loading (up to 5 wt %) had

a smooth “gradual” effect on the increase of the leaching rate; whereas the increase of the

NaB aggregate size had a sharp effect on the increase of the leaching rate.

109 60 (a) /day) 2 45

15 Leaching rate (µg/cm 0 0 5 10 15 20 aggregate size (µm)

60 1.0

/day) (b) 2 45 0.5

30 0.0 012345

(µg/cm rate Leaching 0 0.0 1.0 2.0 3.0 4.0 5.0

® wt% NaB in Sylgard 184

Figure 4.14 Empirical correlations for the leaching rate of NaB from Sylgard® 184. (a) Effect of the aggregate size (d, the arithmetic mean size), parameter fixed: 1 wt% NaB/matrix. (b) Effect of the NaB matrix loading, parameter fixed: d ~ 3 - 4 µm. The insert in (b) is for enlarging the scale of the y-axis. The symbols (■) and (O) represent the initial and the steady leaching rates, respectively.

110 4.4.6 Effects of continuous stirring and water replacement

All the leaching experiments done so far were performed in static cells. Each

coating sample was immersed in an-unstirred water bath of a constant volume and the concentration of the water bulk was monitored as a function of time. Two questions arise

here. First, is there a build-up of concentration in the water volume, which may result in

lowering the leaching considerably? Second, since the water bath was not continuously

stirred, is it possible that there is a strong establishment of boundary layer close to the

coating surface, which may also lower the leaching considerably?

To answer the above two questions, other sets of leaching experiments were also

performed at three different conditions. All the samples considered here were the ones

that were prepared at the base case conditions (1 wt% NaB/Sylgard® 184; 50/50

water/acetone; and 20/80 solvent/polymer). Each set of samples consisted of three

coatings. In the first set, the samples were immersed in constant volume baths but with

continuous constant stirring. In the second set, the samples were immersed in un-stirred

water baths but with replacing the water daily. In the third set, which was the control

static cell set done here for comparison; the samples were immersed in un-stirred water

baths and without replacing the water. The cumulative leaching was measured as a

function of time up to 19 days. The results are shown in Figure 4.15. It was observed

that the continuous stirring and replacing the water bath did not considerably affect the

cumulative leaching of NaB from Sylgard® 184. For example, after 19 days of

immersion, the cumulative leaching was 29.3 µg/cm2, 31.6 µg/cm2, and 34.1 µg/cm2 for

111 samples corresponded to static conditions, water replacement conditions, and continuous stirring conditions, respectively. Based on the mass balance analysis, about 5.1 %, 5.4 %, and 5.9 % of the NaB original mass had leached out after 19 days of immersion for samples corresponded to static conditions, water replacement conditions, and continuous stirring conditions, respectively.

A final note is made here about the experimental method used to evaluate the leaching of NaB. The amount of NaB leached out into the solution was determined via conductivity measurements. The conductivity meter used in the leaching experiments is of high precision and accuracy. It gives readings in two digits, and its accuracy is specified by the manufacturer as ± 0.4%, and it has the ability to detect as small as 0.1 ppm of dissolved matters. For example, for one of the NaB/Sylgard® 184 samples prepared at the base-case conditions, the concentration of NaB in solution was 0.83 ppm,

6.59 ppm, and 10.01 ppm after 1 hr, 1 day, and 1 week of immersion, respectively. The change in concentrations here was higher than the range of the error of the equipment.

Therefore, we believe that the increase in conductivity of the solution is because of the real change of NaB concentration in solution, which is resulted from leaching of NaB from the coating.

112

60 static condition stirring condition 45 water replacement condition ) 2

Q (µg/cm 15

0 0 5 10 15 20 time (day)

Figure 4.15 Cumulative leaching (Q) of NaB from Sylgard® 184 matrix immersed in water. The samples were prepared at the base case conditions ((1 wt% NaB/Sylgard® 184; 50/50 water/acetone; and 20/80 solvent/polymer)), and the leaching were measured at three different conditions: under constant stirring (□), replacing water daily (∆) , and at static conditions (O). Error for each data point (average over 3 batches) is presented by the vertical line.

113 4.5 Mass transfer analysis for the leaching study

4.5.1 Simplified mass transfer model

The previous subsection (section 4.4.5) provides two empirical correlations for

the effect of NaB aggregate size and NaB matrix loading on the leaching rates. Here, we

are trying to provide more fundamental mass transfer analysis for the leaching process.

The antifoulant release mechanism can be classified according to the solubility of the

compound in the polymer phase. As shown in Figure 4.16, if it has a high solubility and

initially loaded in excess of its solubility limit, the release primarily follows a diffusion-

dissolution mechanism, which takes place in the continuum of the polymer phase

(Cardarelli, 1980; Fan and Sigh, 1989). On the other hand, as shown in Figure 4.17, if

the solubility is very low, the media of dissolution and diffusion is water that fills the

porosity of the matrix, not the continuum of the polymer phase (Cardarelli, 1980; Fan and

Sigh, 1989). Since NaB is not soluble in silicones (the χ12 value for sodium

benzoate/PDMS is 17.6), most likely the release is by the porosity formation mechanism.

In this case, the release process can be described mathematically based on the basics of

diffusion in porous media as follow. The concentration, C, of the compound in the water-

filled pores at time t and axial distance x can be described by (see Figure 4.18):

2 2 (∂C / ∂t) = DA (∂ C / ∂x ) (4.12) where DA is the apparent diffusion coefficient of the compound in the water-filled pores,

given by:

114

Impermeable substrate Un-dissolved zone

Dissolved zone

Water bulk

Boundary layer

Figure 4.16 A simplified sketch (not to scale) for the mass release of antifouling compounds from polymer paint (water-insoluble matrix). In this case, the compound is soluble in the matrix and is initially loaded in excess of its solubility limit in the matrix. The dissolved zone means that the compound is already absorbed by the polymer phase.

115

Impermeable substrate

Unleached matrix

Exhausted matrix

Water bulk

Boundary layer

Leaching holes

Figure 4.17 A simplified sketch (not to scale) for the leaching of antifouling compounds from polymer paint (water-insoluble matrix). In this case, the compound is insoluble in the matrix. (Figure re-drawn from Caprari et al. (1990), with slight modification).

116

Water Boundary Impermeable bulk Layer object Polymer coating of thickness L

x=δ x=0 x=L x

Figure 4.18 A simplified sketch (not to scale) of a polymer coating incorporated with AF compound, and immersed in water. The purpose of the sketch is to show the meaning of the axial distance x that was used in equation 4.12

117 DA = D ε / θ (4.13)

where D is the molecular diffusion coefficient of the compound in water, and ε and θ are,

respectively, the porosity and tortuosity of the matrix. The porosity here is the initial

porosity of the matrix (generated by the preparation conditions) plus the empty spaces

generated progressively with time when the compound is released out. The tortuosity is the deviation of the diffusion path from ideality, where for ideal case the diffusion path is straight and hence θ is equal to unity, whereas for nonideal case θ is greater than one (Fan and Sigh, 1989; Welty et al., 1984). In other words, the tortuosity is a measure that describes how much the diffusion path is zigzagging. The initial and boundary conditions associated with equation (4.12) are:

C = CO at t = 0, all x (4.14a)

DA (∂C / ∂x) = k (C — Cb), at x = 0 (4.14b)

(∂C / ∂x) = 0, at x = L (4.14c)

where CO is the initial concentration of the compound in the matrix, L is the thickness of

the matrix, Cb is the bulk concentration of the compound in the water bath, x=0 is the coating surface -water bath interface, and k is the external mass transfer coeffieicent

(which is associated with the convective bulk flow). For simplicity, Cb can be assumed

zero at all time, and then equation (4.14b) is replaced by:

DA (∂C / ∂x) = k C, at x = 0 (4.15)

118 An approximate analytical solution was obtained for the above PDE by Gurny et al. (1982) for a special case of applying the boundary condition of:

C = 0 at x=0 (4.16)

and the solution is:

1/2 Q = 2 CO (DA t/ π) (4.17)

Due to its simplicity, equation (4.17) is desired to analyze our experimental leaching data.

However, its applicability for our system has to be tested. Equation (4.17) is originally derived for a special case called “perfect sink conditions”, which is based on two assumptions. First, the volume of the water bulk is very big “i.e. infinite volume” and consequently the bulk concentration is almost zero at all time. Second, the water bulk is perfectly and continually stirred so that the bulk concentration is not a function of distance and also there is no boundary layer exists near the coating surface, consequently the surface concentration is equal to the bulk concentration. Based on these two assumptions, the boundary condition (C=0 at x=0) is applicable, which will lead to the simple solution as shown in equation (4.17). To justify the above assumptions for our experimental setup used to study the leaching of NaB, the following could be mentioned.

First, our system was not an infinite volume, but it was experimentally observed that the bulk concentration of NaB in the water bath was very low. For example, after 1 month of immersion of 0.5 wt% NaB/matrix and 5 wt% NaB/matrix sheets, the bulk concentrations

119 were 8 ppm and 23 ppm, respectively. These concentrations are close enough to zero to

justify the first assumption. Second, our system was not continually and perfectly stirred,

but at each time of measurement the water bath was gently stirred to get an average bulk

concentration. This gentle stirring could be enough to satisfy the second assumption.

The difference between perfect stirring and gentle stirring is elaborated more on the next

paragraph.

Equations (4.15) is a general form whereas equation (4.16) is a special form, both

of them are boundary conditions for the same system of equations. Equation (4.15) is

applicable for any degree of stirring whereas equation (4.16) is originally assumed for

perfect stirring. It is of interest to examine mathematically the difference between the

two situations. To do so, the PDE with the general boundary condition (equation 4.15)

has to be solved, and then the situations can be identified where the general boundary

condition (equation 4.15) can be reduced to the special boundary condition (equation

4.16). This procedure is called “model sensitivity analysis or model parametric study”, and we apply it here for our system as follow. Equations (4.13) to (4.15) are rewritten in

dimensionless forms by defining the following dimensionless variables:

2 ζ= x / L; Ψ = C / CO; τ = t DA / L (4.18)

Hence, the PDE with its boundary conditions in normalized form are:

(∂ Ψ / ∂ τ) = (∂2 Ψ / ∂ ζ 2) (4.19a)

Ψ = 1 at τ = 0, all ζ (4.19b)

120 (∂ Ψ / ∂ ζ) = Bm Ψ, at ζ = 0 (4.19c)

(∂ Ψ / ∂ ζ) = 0, at ζ = 1 (4.19d)

where the dimensionless parameter Bm is the Biot number for mass transfer, and defined

as: (Bm = k L / DA ). Equations (4.19a) to (4.19d) are difficult to solve analytically.

Therefore, we solved it numerically by Orthogonal Collocation technique. Details about

the theory and application of the Orthogonal Collocation technique can be found

elsewhere (Ruthven, 1984; Rice and Do, 1994), and for solving the problem a computer

program in MATLAB was written (the program is shown in the appendix). The

simulations are performed here for all range of Bm, from a very small value (almost zero)

to a very large value (almost infinity). These two limits have physical meanings. If Bm goes to infinity (i.e. k L >> DA), it means that the mass transfer process is diffusion-

limited and there is large bulk convection due to high bulk velocity, and hence there will

be no boundary layer at the surface at all because of having perfect mixing or because

that the coated object (e.g. a moving ship) is moving with very high speed, and consequently the surface concentration is equal to zero [i.e. the problem is reduced to the

special case of equation (4.16)]. If Bm goes to zero ((i.e. k L << DA), it means that the

mass transfer process is bulk convection-limited and the coated object and the water bath

are perfectly stagnant and there is a strong boundary layer established, and in this case the

surface concentration is high. This physical meaning is understood more clearly with

referring to the simulations trends shown in Figure 4.19. In Figure 4.19, the normalized

surface concentration is plotted as a function of the normalized time, for different values

of the normalized parameter Bm. As can be seen in the figure, the surface concentration

121

0.9

0.8

0.7

0.6 Bm = 1 Ψ|ζ=0 0.5

0.4

0.3 Bm = 2

0.2 Bm = 10

Bm = 20 0.1

Bm = 100 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Figure 4.19 Parametric sensitivity analysis for the general mass transfer model (equation 4.19). The dimensionless surface concentration (Ψ|ζ=0 = C/CO |x=0) is plotted against the 2 dimensionless time (τ = DA t / L ), for different values of the dimensionless parameter Bm (Bm = k L / DA). The trends were generated by solving equation 4.19 numerically.

122 becomes close to zero when the value of Bm is greater than 100, which implies that the

applicability of equation (4.16) could be valid if Bm is greater than 100.

Bm is a function of k, the external mass transfer coefficient. In standard tests

available in the literature (Court and Vries, 1973), k is determined experimentally by

connecting the coated object to a motor thus the object is rotating with different velocity

(this is to simulate a ship moving with different speeds), and hence k can be calculated by

correlating with the angular velocity. We do not have this kind of setup and therefore we

do not know the exact value of k and Bm for our particular system, but for our

experimental setup where we have gentle stirring at each time of leaching measurement,

we assume that this gentle mixing is enough for Bm to be greater than 100 and therefore the validity of equation (4.17) is assumed. The estimation for the value of Bm for our particular experimental setup is elaborated more on the next paragraph.

The value of Bm for our particular NaB system is estimated as follow. The external mass transfer coefficient, k, can be estimated by the correlation (Welty et al., 1984;

Skelland, 1985):

Sh = k l / D = 0.664 (Re)1/2 (Sc)1/3 (4.20)

where Sh, Re, and Sc are the Sherwood number, Reynolds number, and Schmidt number,

respectively, and the latter two are defined as:

123 Re = ρ v l / µ; (4.21a)

Sc = (µ/ρ) / D (4.21b)

where ρ and µ are, respectively, the density and viscosity of water (1 g/cm3 and 0.01 g/cm/s). l is the characteristic length of the system, which is an arbitrary number that describes the length of the surface of the object on which the boundary layer is established, and for our case it is chosen as the diameter of the Petri dish from which the samples are molded (i.e. l ~ 9 cm). D is the molecular diffusion of NaB in water, and its experimental value is not available in the literature. Alternatively, D is estimated here by

Nernst-Haskell equation (Poling et al., 2001):

2 D = (RT/F ) (1/Z+ + 1/Z-) / (1/λ+ + λ-) (4.22)

where R and T are, respectively, the universal gas constant and absolute temperature, Z+ and Z- are, respectively, the valences of cation and anion, λ+ and λ- are, respectively, the

limiting ionic conductances of cation and anion, and F is the Faraday constant. By

evaluation of equation (4.22), a value of 1.0 x 10-5 cm2/s for D is obtained. This value is

in the same order of magnitude for the experimental D values (Poling et al., 2001) of

different salts (KCl, KNO3, and NaCl) in water, indicating that the prediction by equation

(4.22) is reasonable. In equation (4.21a), v is the velocity of the bulk water bath which is

assumed to be very small but not zero because the bath is gently stirred at each time of

measurement; thus a value of 0.001 cm/s is assumed here for v. Consequently, by

evaluations in equation (4.20), a value of about 6 x 10-6 cm/s for k is obtained. Later on it

124 -11 2 will be shown that the apparent diffusion coefficient, DA, is in the order of 10 cm /s

-11 (see Table 4.6). Thus, from the relation Bm = k L / DA, with L ~ 1 mm and DA ~ 10

2 cm /s, a value of about 600 for Bm is obtained. As demonstrated in Figure 4.19, this

value for Bm is high enough for the surface concentration to be close to zero and hence the validity of equation 4.17 could be justified. This estimated high value for Bm is not because the bulk velocity is high, but because the parameter (ε/θ) is very small, as to be discussed more on the next paragraph.

In order to understand the physical meaning of the porosity that we are talking about and its relation to the aggregate size and the matrix loading, we need to elaborate more on the leaching mechanism. During the immersion of NaB-incorporated Sylgard®

184 coatings, there is more than one possible mechanism for the compound to release.

First, in the most ideal case, the aggregates are spherical and uniformly distributed throughout the matrix, and perfectly packed such that the particles are smoothly touching each others (Figure 4.20a). In this case, a sharp boundary (Caprari et al. (1990) between the dissolved zone and the un-dissolved zone of the matrix will be established during immersion, and the location of this boundary is moving inward into the matrix as immersion time proceeds. The second possibility is that all the particles are not touching each other initially (Figure 4.20b). In this case, a sharp boundary between the dissolved zone and the un-dissolved zone will not be observed. Instead, the following mechanism is hypothesized. Initially, the particles are separated from each others because they are encapsulated in thin membranes of binder (the polymer phase). The particles at the surface layer of the film are first dissolved, forming a saturated solution that diffuses

125

(a)

(b)

(c)

Figure 4.20 Simplified sketch (not to scale) for the possible leaching mechanisms of NaB from Sylgard® 184 coating: (a) Perfect packing of the particles; (b) Complete ruptures of the thin membranes; (c) The existence of initial porosity of the matrix, which is mainly composed of constricted narrowed channels. The first column represents the coatings initially before immersion, the second column after some time t1 > 0, and the third column after some time t2 > t1.

126 outward through the diffusion layer in contact with the coating surface. When such a particle dissolves in water and being removed, a cavity separated by thin membranes from other cavities and from the un-dissolved particles is resulted. Water diffuses into the cavity and through the thin membrane to dissolve some of the un-dissolved particles.

Consequently, the resulting osmotic pressure ruptures the membrane and the cavities become interconnected. This mechanism (the second mechanism) was first postulated by

Marson (1969). The third mechanism is that the matrix has an initial porosity generated by the preparation condition (Figure 4.20c). Therefore, it could be possible for water to diffuse through this initial spacing even though the particles are not initially connected, and then the porosity is increasing with time as more particles are dissolved. For the third mechanism, it is possible that the initial porosity is very small, because it might be composed mainly of constricted channels of very narrow spacing which spread out throughout the matrix and connect the particles between each others. Water diffuses through these channels from the dissolved particles region to the un-dissolved particles region to dissolve the un-dissolved NaB particles, and then carries the dissolved NaB molecules back out toward the surface layer of the coating. These constricted channels exist initially could also be progressively increased during time of immersion. It is possible that the width of these channels is very narrow (<< 1 µm), which is much smaller than the diameter of particles, and it is also possible that the length of these channels is very tortuous (i.e. θ >> 1); therefore the diffusion process slows down considerably during the slow leaching stage. In addition, Sylgard® 184 does absorb water to approximately 0.1 wt% water/polymer (Banerjee et. al., 1997). The water absorbed by the polymer phase may create molecular level pores which may be responsible for

127 specifically increased tortuosity (Miller and Peppas, 1982). Also, Miller et al. (1983)

have shown experimental evidences from the porosomitery analysis that very small pores,

in the order of 100 Å, are present in their system, which is likely possible for our system.

Therefore, for the above reasons, the value of (ε/θ) for our system is likely in the range of

10-4. For our model system where the maximum loading prepared was 5 wt%

NaB/Sylgard® 184, the third mechanism is likely more suitable for our system than the

first and the second mechanism.

For the third mechanism, a fundamental question arises: is the capillary rise effect capable of enhancing the flow of water through these constricted channels? The answer would be no if the surface of the walls of these channels is solely PDMS – a purely hydrophobic surface – and nothing else deposited on the surface. However, for our system, we speculate that some NaB could exist on the surface of these channels, which will turn the walls to be hydrophilic and hence enhance the water flow. The origins of the NaB exist on the surface walls are that the channels are distributed initially at the whole matrix including the NaB aggregates, and NaB prefers water over PDMS.

The answer to the above question could be further understood more clearly through the

Young – Laplace equation:

∆P = 2 γ cos φ / r (4.23)

where ∆P is the hydrostatic pressure drop through the capillary, γ is the surface tension of the liquid, r is the radius of the capillary, and φ is the contact angle of the liquid-capillary

128 wall interface (Adamson, 1990). If the wall is purely hydrophobic (i.e. purely PDMS), φ

is equal to or greater than 900, which implies that the pressure drop is zero or negative

and hence water flow will not be possible. If there is some NaB exist on the wall, which

is our speculation, φ will be less than 900 and hence the pressure drop will be positive and

consequently water flow will be possible through the capillary.

The above proposed mechanism is based on the assumption that NaB is totally

insoluble in the polymer phase. To check for the validity of this assumption, the

solubility of NaB in PDMS was estimated by the following equation, which was

suggested by Zaikov et al. (1988) for the estimation of electrolytes solubilities in

hydrophobic polymers:

2 — ln ФS = 1 + (VS / RT) (δS — δP) (4.24) where ФS is the volume fraction of the solute (i.e. NaB in our case), VS is the molar volume of the solute, and δS and δP are the solubility parameters of the solute and

polymer, respectively. By evaluation in equation (4.24), the solubility of NaB in PDMS

is 8 x 10-7 vol. %, which is close enough to zero to justify the above assumption.

In summary, in all the above possible mechanisms, the key point is that NaB is

insoluble in the polymer phase; therefore NaB diffusion is taking place through pores

(empty space) within the matrix, not through the continuum of the polymer phase. For

this case, as the porosity of the matrix increases it will be easier for water to diffuse into

the matrix and carry with it the dissolved compound, and hence the cumulative release of

the incorporated compound will increase.

129 4.5.2 Limitation of the simplified mass transfer model

The simplified model applied in section 4.5.1 has the advantage that it has explicit

analytical solution for the cumulative leaching as function of time:

1/2 Q = 2 CO (DA t/ π) (4.17)

However, it has several limitations. Before listing the limitations, the simplified model is first applied to fit our experimental data, followed by the extraction of the model parameters, then more elaborations is made on the physical meaning of these extracted

parameters, as follow.

The experimental cumulative leaching data of the NaB/Sylgard® 184 samples

prepared at different set of conditions were fitted to equation (4.17) by plotting the Q data

against t1/2 at the early leaching time period (0 – 4 days), and the results are shown in

Figure 4.21 and Table 4.6. For the fitting analysis, only the sets of preparation conditions

that resulted in the aggregate size smaller than 15 µm were considered here, because these were the optimized preparation conditions that resulted in samples of fine and

uniform dispersion. As shown in Figure 4.21 and Table 4.6, in most cases, a good linear fit was observed, which could suggest that diffusion is the rate limiting step for

NaB/Sylgard® 184 system. From the slope of the linear fit, the apparent diffusion

® coefficient (DA) for NaB/Sylgard system for each set of preparation conditions was

calculated, and the results are summarized in Table 4.6. For all the samples, the apparent

130 60 1 wt% 2 wt% (a) 3 wt%

) 4 wt% 2 40 5 wt% 0.5 wt%

Q (µg/cm 20

0 0.0 0.5 1.0 1.5 2.0 t1/2 (day1/2)

40 20/80 W/A 30/70 W/A (b) 40/60 W/A

) 30

2 50/50 W/A

Q (µg/cm 10

0 0.0 0.5 1.0 1.5 2.0 time1/2 (day1/2)

Figure 4.21 Fitting of the cumulative leaching data for NaB/Sylgard® 184 coatings to the simplified mass transfer model (equation 4.17). (a) Samples were prepared at different NaB matrix loading, keeping the water/acetone ratio and the solvent/polymer ratio fixed at 50/50 and 20/80, respectively. (b) Samples were prepared at different water/acetone (W/A) ratios, keeping the NaB matrix loading and the solvent/polymer ratio fixed at 1 wt% NaB/matrix and 20/80 ratio, respectively. Points are experimental data and solid lines are linear fitting of the model.

131

Table 4.6 Data analysis for the results shown in Figure 4.21. The apparent diffusion ® coefficient (DA) for NaB/Sylgard system was obtained by fitting the experimental leaching data to equation 4.17. The solvent/polymer ratio was 20/80, which was fixed for all samples. (Abbreviation: W: water, A: acetone).

NaB Slope of the 2 Sample D (cm /sec) Solvent matrix linear fitting value A # 2 1/2 11 loading (µg/cm /day ) x 10

1 20/80 W/A 1 wt % 11.71 0.98 1.1

2 30/70 W/A 1 wt % 11.46 0.99 1.1

3 40/60 W/A 1 wt % 12.76 0.98 1.3

4 50/50 W/A 1 wt % 12.57 0.96 1.3

5 50/50 W/A 0.5 wt % 8.02 0.99 2.1

6 50/50 W/A 2 wt % 15.03 0.90 0.5

7 50/50 W/A 3 wt % 18.31 0.85 0.3

8 50/50 W/A 4 wt % 20.16 0.93 0.2

9 50/50 W/A 5 wt % 24.03 0.89 0.2

132 diffusion coefficient was estimated to be in the order of 10-11 cm2/s. The diffusion

coefficient value is in the same order of magnitude for all these samples because these

samples were shown previously in section 4.2 to have similar fine and uniform size

distribution, with no observed aggregate with size greater than 15 µm. However, this value for the diffusion coefficient is extremely small, close to the typical values known in solid-solid diffusion. However, it should be noticed that we define DA as the apparent

diffusion coefficient of NaB in the water-filled pores, not as the molecular diffusion of

NaB in the continuum of the polymer phase, because we believe that NaB has limited

solubility in the polymer phase. Therefore, the value of DA estimated here could be

reasonable only if the porosity of the matrix is extremely small, where in this case the

physical meaning is that there is a large geometrical obstruction for NaB to diffuse

through the matrix.

Next, by applying the relation associated with the simplified model (DA = D ε / θ),

-11 2 -5 2 with considering DA to be ~ 10 cm /s and D to be ~ 10 cm /s (D is the molecular

diffusion of NaB in water, which is estimated here by an independent theory (equation

4.22) as described in section 4.5.1), this implies that the value of (ε/θ) is in the range of

10-6. This suggests that the system has very small porosity (ε) and very high tortuosity

(θ). However, we should here retain the physical meaning of the tortuosity. The

tortuosity is the ratio of the actual diffusion length to the ideal diffusion length.

Therefore, in most nonideal cases, the tortuosity most likely cannot be more than 10, and

in many examples in the drug-release literature it was assumed a value of 3 to 5.

Although many examples in the drug-release literature have reported a tortuosity value

133 above 1000, these values appear unrealistic and have no physical meanings. Thus, in the

most nonideal case, let’s assume a value of 10 for the tortuosity of our system. This

would predict an extremely low value of the porosity in the order of 10-5. This predicted

value for the porosity is an indication for the limitations of the simplified model, because

if the porosity is almost zero, NaB will not leach out due to the fact that NaB has limited

solubility in the polymer phase and has to find some porous space to get through.

One clear limitation for the simplified model is the incomplete description of the

porosity, and considering it as a time independent. From its physical meaning, the

porosity is the empty volume inside the coating over the total volume. The empty

volume is any volume available inside the coating other than the continuum of the

polymer phase itself and the un-dissolved solid NaB particles phase. Therefore, it can be

qualitatively described as:

ε = ε0 + εn + εw

where ε0 is the initial porosity of the coating, εn is the empty spaces generated

progressively with time when the compound is released out, and εw is the porosity due to

water absorption by the polymer (as to be described in the next paragraphs). ε0 is

constant whereas εn and εw are time-dependent. However, the simplified model lumped all these types together as a single constant value for the porosity, which is a clear

limitation of the model.

134 The initial porosity, ε0, is the porosity generated by the preparation procedure,

which is already exist before immersing the coating in water. For our model system,

NaB/Sylgard® 184, this could originate from various sources, such as the existence of air bubbles. Another source is because that packing of NaB is not expected to be perfect or ideal. As a result, in-perfect dispersion is possible, which might lead to tiny “voids” in- between the NaB particles and at the NaB/polymer interface. The magnitude for this initial porosity is not expected to be zero; otherwise it would be very hard for NaB to leach out. Nevertheless, the value of ε0 should still be very small, as an estimation, ε0 for

our model system could be in the order of 0.0001 - 0.01.

The porosity generated by the release of NaB, εN , is a function of immersion time.

Approximately, it could be expressed as:

εN = (mt / ρn ) / (m0 / ρp)

where mt is the mass of NaB released at time t, ρn is the density of NaB, mp is the initial

® total mass of the sample (6.0 g for our experiments), and ρp is the density of Sylgard

184. For our leaching data for the model system prepared at the base-case conditions, εN

was varied from 0 initially to 0.0004 after 1 month of immersion.

The porosity due to water absorption by the polymer, εw , may also be important and

contribute to the total porosity. Sylgard® 184 does absorb water to approximately 0.1

wt% water/polymer (Banerjee et. al., 1997). This value could suggest that εw is in the

135 range of 0.001. The water absorbed by the polymer phase may create molecular level pores (Miller and Peppas, 1982). Also, Miller et al. (1983) have shown experimental evidences from the porosomitery analysis that very small pores, in the order of 100 Å, are present in their system (a water-soluble drug incorporated into a hydrophobic, monolithic coating), which is likely possible for our system. εw is time-dependent because it depends

on the kinetics of water absorption and diffusion in Sylgard® 184. The diffusivity of

water in silicone should play a role here. The diffusivity data of water in silicone

membrane can be obtained from literature. The literature value for the diffusivity of

water in silicone membrane was reported to be 8.6 x 10-7 cm2/s at 25o C (Banerjee et. al.,

1997). However, the simplified model totally neglected the existence of εw and the role

of water absorption and diffusion in silicone membranes. This is a clear limitation of the

simplified model.

Summing up the components of the total porosity (ε = ε0 + εn + εw ), it is expected

that, based on physical meaning, the magnitude of ε to be in the order of 0.01. However,

-5 the value of ε predicted by the simplified model is in the order of 10 . This discrepancy

is a clear indication for the limitations of the simplified model.

To summarize, the limitations of the simplified model are listed briefly. The first major limitation is that it treats the coating system as one homogeneous phase whereas in fact the coating system is a heterogeneous (multi-phase) system. To account for the heterogeneity of the system, the simplified model applies the parameter (ε and θ), where ε and θ are the porosity and tortuosity of the system, respectively. Although the

136 application of these parameters partially accounts for the heterogeneity of the system, two

problems arise here. First, the porosity is, in real situations, a function of time as more

active compound is released from the system, but the simplified model treats the porosity

as constant. Second, the tortuosity becomes at the end of the analysis almost as an empirical parameter because it cannot be measured independently, and in many examples in the drug-release literature it is either assumed a certain value or determined by fitting the release data to the model. Although many examples in the drug-release literature have reported a tortuosity value above 1000, these values appear unrealistic and have no physical meanings. The second limitation is the over-simplification of treating the

problem as a fixed-boundary problem. However, the problem is a moving boundary

problem because as more pores and channels are generated with time the boundary

condition inside the coating is moving inward and its location is not known in prior. The third limitation is that it does not explicitly include the resistance to mass transfer in the boundary layer surrounding the coating. The fourth limitation of the model is that it does not explicitly account for the role of the diffusivity of water through the coating. In other words, the diffusion coefficient of water in silicone membrane is absent in the simplified model. The fifth limitation of the model is that it does not account for the dissolution step resistance to mass transfer. These limitations can be avoided by re-deriving the mass

balance equations for each phase in the coating (the polymer phase, the solid NaB

particles phase, and the water-filled pores phase) separately, as shown in Appendix B.

137 4.6 Bacterial attachment evaluations

To evaluate the coatings antibacterial behaviors, bacterial attachment studies using fresh water containing indigenous enriched microbial consortium isolated from

Lake Erie water were performed. Pure silicone coatings and 1 wt% NaB containing silicone coatings were submerged in the above water at periodic intervals up to one month. Some representative biofilm morphologies are shown in Figure 4.22 for pure

Sylgard® 184 coatings and 1 wt% NaB-blended Sylgard® 184 coatings, for samples

prepared at the base-case conditions. As shown in Figure 4.22, the bacteria can be easily

identified and differentiated, and it can be concluded that a clear reduction of bacterial

attachment was achieved for 1 wt% NaB/Sylgard® 184 coatings compared to Sylgard®

184 alone. The bacterial attachment images for NaB/Sylgard® 184 systems were further

quantified by counting the pixels to approximate the area coverage, and hence the %

reduction (% reduction = (1-A/B) 100, where A and B refer to the area coverage for NaB-

containing coatings and NaB-free coatings, respectively). As shown in Figure 4.23, for a

particular period of immersion, an average of 45 – 55% reduction in bacterial coverage

was achieved for 1 wt% NaB/Sylgard® 184 coatings as compared to Sylgard®184 alone.

For NaB-containing RTV11, the morphology was hard to be observed and the bacteria were difficult to be identified from the pictures directly. To differentiate the bacteria from other objects, the RTV11 coating surfaces were physically cleaned by scotch tape, and pictures were taken before and after cleaning. For control RTV11 coatings, the surface became very clean after applying scotch tapes, indicating that the

138

( a) (b)

Figures 4.22 Optical microscope images of the bacterial attachment study for NaB- Sylgard® 184 coating. (a, c) 1 wt% sodium benzoate-blended Sylgard® 184. (b, d) control Sylgard® 184. The coatings were immersed in water containing Lake Erie bacteria for 2 weeks (a, b) and 4 weeks (c, d). Image size is (285 µm x 215 µm).

139

100

% reduction 25

1234

time (week)

Figure 4.23: Antibacterial performance of 1 wt% NaB-incorporated Sylgard® 184 coatings, compared to control Sylgard® 184 samples. The % reduction was defined as [(1-A/B) 100), where A and B refer to the area coverage of NaB-containing coatings and NaB-free coatings, respectively.

140 (a)

(b) (c)

Figure 4.24 Optical microscopic (reflection bright field) images of bacterial attachment on controlled RTV11 coatings after the coatings were immersed in water containing Lake Erie bacteria for 28 days. (a) Half of the coatings surfaces were physically cleaned by scotch tape, and overall pictures [image size: (2850 x 2150) µm] were taken showing the cleaned area (right side of picture a) and the un-cleaned area (left side of picture a). Pictures (b) and (c) are the magnifications [image size: (285 x 215) µm] of the two area indicated in picture (a).

141 removed objects were likely bacterial biofilm (Figure 4.24). For NaB-RTV11 coatings,

however, the morphology did not change much after applying the scotch tape, suggesting

that the irregularities seen were parts of the coatings surface and were not bacterial

biofilm. The irregularities seen in/on NaB-RTV11 coatings were mostly holes likely

generated by NaB leaching, and the reason that they were much larger than the holes seen

in/on NaB-Sylgard® 184 coatings was the much faster leaching of NaB from RTV11 than from Sylgard® 184.

By comparing control samples of NaB-free RTV11 and NaB free-Sylgard® 184, it was observed that RTV11 had a higher tendency for biofilm formation than that of

Sylgard® 184, attributing to the fact that RTV11 has a slightly higher surface energy, bulk

modulus, and surface roughness than those of Sylgard® 184. Another reason for this

observation is that RTV11 experienced an increase in surface roughness upon immersion in water due to slow surface erosion. Our scanning probe microscopy verified the increase in the surface roughness value (Rq, with a scan size of 80 µm x 80 µm) from 6.9

nm for as prepared RTV11 films to 11.8 nm for 14 day water-aged RTV11 films. For

Sylgard® 184, on the other hand, a previous study (Barrios et al., 2005) verified that

Sylgard® 184 was fairly stable in water with no observed increase in surface roughness.

To summarize, a clear reduction in bacterial attachment on the NaB-treated

Sylgard® 184 coatings was observed, which suggested that NaB could be effective in

inhibiting bacterial attachment when entrapped into Sylgard® 184 coating. Also, the attachment study demonstrated that the antibacterial performance of NaB/Sylgard® 184

142 coating was better than that of NaB/RTV11 coating. In additions, by referring to the leaching data presented previously, it was estimated here that the bacterial water solution had a bulk concentration of NaB of about 1 ppm after 1 month of immersion. This concentration is much below the EC50 of NaB, which is about 560 ppm towards various types of bacteria (Haque et al., 2005, Xu et al, 2005). Therefore, the reduction in bacterial attachment observed here for NaB treated surfaces was most likely not because the bacteria were simply died off, but because of a nontoxic mode of action of the compound. This emphasizes the benefit of NaB to be a nontoxic antifoulant.

143

CHAPTER V

RESULTS AND DISSCUSSION FOR BENZOIC ACID AND

CAPSAICIN– BASED COATINGS

In this chapter, the results obtained for benzoic acid (BA) and capsaicin - incorporated silicone coatings are presented and discussed. The effect of incorporating the compounds on the surface and bulk properties of the coatings is presented in section

5.1. The miscibility/bulk morphology of the compounds with the polymer matrix is discussed in section 5.2. The leaching of the compounds in water is presented in section

5.3.

5.1 Effect of the compounds on coating’s properties

5.1.1 Effect of benzoic acid

Benzoic acid (BA) was incorporated into two types of silicones (Sylgard® 184 and

RTV11), and the concentration prepared was fixed at 1 wt% BA/matrix. For both combinations, it was observed that the BA-blended coatings were cured similarly as the

BA-free coatings alone. Further examining the effect of the incorporated compound on

144 the wettability and bulk modulus of the silicone coatings qualitatively confirmed this

observation. The wettability was examined in terms of measuring the water contact

angles and the results are presented in Tables 5.1 and 5.2 for BA- Sylgard® 184 and BA-

RTV11 coatings, respectively. As shown in the tables, the low content of BA in the

matrix (1 wt %) did not considerably affect the wettability for both types of silicones.

Similarly, the bulk modulus for 1 wt% BA/ Sylgard®184 coatings was measured

and compared to the respected value for the controlled BA-free Sylgard® 184 coatings

(Table 5.3). It was observed that the modulus of 1 wt% benzoic acid/ Sylgard® 184 matrix was slightly lower than the value of the control. This slight variation of the modulus could be attributed to the final distribution of the compound inside the bulk of the matrix. As to be seen in the next section, benzoic acid had an irregular distribution of large crystal (to be discussed in details later) inside the coatings, which could result in irregularity and void spaces in the matrix and hence the bulk modulus could decrease.

Nevertheless, the elastic modulus still confirmed that the coating was cured. If it was not cured, the modulus would dramatically dropped, much below the value presented in

Table 5.3. For 1 wt% BA/RTV11 coating, the elastic modulus was not measured, but at this low concentration the modulus do not expect to differ two much from the corresponding value of BA-free RTV11 coating, especially with knowing the fact that

RTV11 is much more resistant to poisoning than Sylgard® 184.

145

Table 5.1 Static water contact angles of BA-entrapped Sylgard® 184 coatings compared to that of the controlled BA-free Sylgard® 184 coatings. The (solvent: polymer) ratio was (20: 80) by mass.

Coatings Static contact angle

Control Sylgard® 184 106.1 ± 0.6

1 wt% BA/ Sylgard® 184 104.9 ± 1.8

Table 5.2 Static water contact angles of BA-entrapped RTV11 coatings compared to that of the controlled BA-free RTV11 coatings. The (solvent: polymer) ratio was (20: 80) by mass.

Coatings Static contact angle

Control RTV11 101.2 ± 0.5

1 wt% BA/ RTV11 98.7 ± 1.0

TABLE 5.3 Elastic modulus of BA-entrapped Sylgard® 184 films. The (solvent: polymer) ratio was (20: 80) by mass.

Coatings Elastic modulus (MPa)

Control Sylgard® 184 0.95 ± 0.20

1 wt% BA/ Sylgard® 184 0.52 ± 0.02

146 5.1.2 Effect of Capsaicin

Capsaicin was incorporated into RTV11 coatings, up to a concentration of 1 wt%

in the matrix, using toluene as the common solvent. It was observed physically that the capsaicin-blended RTV11 coatings were cured similarly as the capsaicin-free RTV11 coatings. Further examining the surface wettability, surface roughness, and elastic modulus of the coatings confirmed this observation, as discussed below. The wettability of the coatings was evaluated in terms of measuring the water contact angles. The

contact angles for RTV11 films containing various concentrations of capsaicin (0.1 – 1

wt %) were measured, and the results are shown in Figure 5.1. RTV11 surfaces without

capsaicin had advancing, static, and receding contact angles of 103°, 100°, and 95°,

respectively. As shown in Figure 5.1, the advancing and static contact angles were

almost unaffected by the addition of capsaicin. The receding contact angles decreased

slightly as capsaicin concentration increased, a value of 87° for 1 wt% capsaicin was

observed. The indifference in the advancing and static contact angles between both

controlled RTV11 and capsaicin-incorporated RTV11 samples suggested that most

capsaicin molecules were entrapped inside the bulk of the polymer matrix rather than

aggregated to the surface. Otherwise, the advancing and the static contact angle are

expected to drop significantly due to the fact that capsaicin has a higher surface energy (~

45 mJ/m2) than PDMS (~ 20-24 mJ/m2). The surface energy of capsaicin reported here

was our predicted value, predicted by the group-contribution method following the

procedures of van Krevelen (1990).

147

120

110

100

90 Contact angles Contact

70 0 0.3 0.6 0.9 1.2 wt% Caps in RTV11

Figure 5.1 Water contact angles of capsaicin-entrapped RTV11 films (advancing: U, receding: , and static: {). The (solvent: polymer) ratio was (20: 80) by mass, which was fixed at all concentrations. Error for each data point (average over 12 measurements) is presented by the vertical line.

148 The elastic modulus of control RTV11 coatings and 1 wt % capsaicin/RTV11 coatings were measured, and the results are shown in Table 5.4. The elastic modulus for

RTV11 film was measured to be 1.56 MPa, which is consistent with the literature value reported (Kohl & Bolstes, 2001). As shown in Table 5.4, the indifference in the elastic modulus for the capsaicin free-RTV11 and capsaicin-blended RTV11 could indicate that the low content of capsaicin (up to 1 wt %) was insignificant in affecting the curing behaviors and bulk properties of RTV11, as the elastic modulus is expected to drop significantly for the uncured coating. Based on this finding and comparing it with the curing behaviors of the other compounds/matrices combinations that were tried in the current study, and combined also with previous findings for other combinations tried in our research group [zosteric acid in Sylgard® 184 and RTV11 matrices (Barrios, 2005), and capsaicin in Sylgard® 184 matrix (Jaggari, 2003)], it is apparent to us now that

RTV11 matrix is much more resistant to poisoning than Sylgard® 184 matrix, because all the above combinations were cured except capsaicin/ Sylgard® 184. This information is useful to the coating formulation industry.

Table 5.4 Elastic modulus of capsaicin-entrapped RTV11 films. The (solvent: polymer) ratio was (20: 80) by mass.

Coating Elastic modulus (MPa)

Control RTV11 1.56 ± 0.07

1 wt% Caps/RTV11 1.57 ± 0.11

149 5.2 Miscibility of the compounds in silicones

5.2.1 Miscibility of benzoic acid

For benzoic acid-incorporated Sylgard® 184 coatings, the preparation conditions were systematically varied to examine their effects on the distribution and morphological

structures of benzoic acid inside the coating matrix. The coatings were prepared using

four different organic solvents (toluene, acetone, acetonitrile, and di-ethyl ether), whereas

keeping the other preparation conditions unchanged (1 wt% benzoic acid in the coating,

and 20: 80 solvent: polymer mass ratio). As shown in Figure 5.2, large crystals were

observed inside the polymer matrix for all types of solvents used. Toluene was found to

result in the largest crystals (~ 600 ─ 1000 µm), and di-ethyl ether produced the smallest ones (50 ─ 100 µm), whereas acetone and acetonitrile resulted in crystals somewhere in between (~ 200 ─ 500 µm). The estimated size here refers to the average length (the largest dimension) of the crystal. The number densities of the benzoic acid crystals were

(1 – 2) /mm2 and (7 – 8) /mm2 using toluene and acetone, respectively. For the case of

using toluene as the solvent, it is possible that occasionally, some large crystals could

span the entire thickness of the coating. From Figure 5.2, it is also clear that ether

resulted in a much uniform distribution of smallest crystals (the number density was (90 –

95)/mm2) compared to the other solvents.

150

(a) (b)

Figure 5.2 Optical microscope (bright field) images of resulting BA distribution in the bulk of Sylgard® 184 matrix when different solvent were used to mix BA with Sylgard® 184. (a) toluene, (b) acetone, (c) acetonitrile, and (d) ether. The concentration of BA in the matrix was fixed at 1 wt%, and the (solvent: polymer) ratio was fixed at (20: 80) by mass. The image size is 2850 µm x 2400 µm for (a), and 1140 µm x 960 µm for (b)-(d).

151 The effect of solvent type on bulk distribution of benzoic acid in the coating could

be understood by considering four factors: the chemical nature of the solvent, its boiling

point (or alternatively the volatility), benzoic acid solubility in the solvent and solvent-

matrix miscibility. The four solvent selected here are representative of different chemical

classes: toluene (an aromatic), acetone (a ketone), acetonitrile (a nitrile), and di-ethyl

ether (an ether). The physical properties relating to the miscibility of these solvents with

both benzoic acid and silicones are summarized in table 5.5. From literature, the

experimentally measured solubility of benzoic acid in toluene, acetone, acetonitrile, and ether are around 10 wt.%, 32 wt.%, 14 wt.%, and 33 wt.%, respectively (Perlovich et al.,

2003). The normal boiling point of toluene, acetone, acetonitrile, and ether are 111°C,

56°C, 82°C, and 35°C, respectively. By considering the differences in solubility parameters of silicone and the solvent, ∆δ, the high benzoic acid solubility in ether, and the low boiling point of ether, it is not surprising that the smallest crystals were resulted when ether (∆δ = 0.2 MPa1/2) was used as the solvent. On the other hand, toluene has a

higher value of ∆δ (3.3 MPa1/2) compared to ether, and it has the highest boiling point

and lowest benzoic acid solubility among the four solvent. Therefore, the largest crystals

were formed when toluene was used as the solvent.

The miscibility could be also predicted through the calculations of the interaction

parameter, χ12. The interaction parameters between BA and the four solvents used with silicone are predicted by the same equation described in chapter 4 (equation 4.1):

2 χ12 = (V1/RT) (δ1 – δ2) (5.1)

152 Table 5.5 Physical parameters of relevance importance to the miscibility of BA/PDMS. V and δ are the molar volume and the solubility parameter of the material, respectively. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS*.

Boiling V δ ∆δ12 Material 3 ½ ½ point (cm /mol) (MPa) (MPa) χ12 (oC) Toluene 106.8 18.2 a 3.3 0.469 110.6

Acetone 74.0 20.3 a 5.4 0.871 56

Acetonitrile 52.6 24.6 a 9.7 1.998 81.5

Ether 105 15.1 a 0.2 0.002 34.6

BA 92.5 22.9 b 8.0 2.389 -

a, b Values obtained from (Rodriguez, 1989) and (Bustamante et al., 2000), respectively * For PDMS, the literature value of its solubility parameter (14.9 MPa½, from: Rodriguez, 1989) was used

and the results are presented in Table 5.5. First, the χ12 values for benzoic acid/PDMS

mixture is 2.389, indicating that BA is likely not miscible with PDMS, as observed experimentally. For the four solvents used in benzoic acid/PDMS systems, it is clear that ether is most miscible with PDMS (χ12 = 0.002), thus the domains of (ether + benzoic

acid) would likely be the smallest as ether being completely evaporated and resulting in

the smallest benzoic acid crystals as compared to other solvents. Experimentally, the

miscibility of PDMS in the four solvents used was tested in the current study by

dissolving PDMS in the four solvents used. It was observed that both ether and toluene

were very good solvents for PDMS; up to a concentration of 20 wt% PDMS/ether and 20

wt% PDMS/toluene were prepared, and excellent dissolution was observed. This 153 observation is agreed with the χ12 values predicted for ether and toluene, where both values are less than the critical value (χ12 < 0.5) defined by the Flory-Huggins theory for a polymer solution system to be completely miscible. On the other hand, it was observed that acetone and acetonitrile were not good solvents for PDMS compared to toluene and ether. This observation is also agreed with the χ12 values predicted, which are 0.87 and

2.00 for acetone and acetonitrile, respectively.

5.2.2 Miscibility of capsaicin

Although direct blending of capsaicin with Sylgard® 184 resulted in uncured coating (Jaggari, 2003), the blend is still useful to roughly examine the miscibility of capsaicin with RTV11, because Sylgard® 184 is transparent whereas RTV11 is not.

Thus, capsaicin was blended with Sylgard® 184 base by the simple blending technique using toluene as the common solvent following the same mass ratios and conditions as we did with RTV11, but without adding the curing agent. The mixtures were kept inside the air hood for 4 days to dry off the solvent at room temperature, and then optical microscope imaging was taken. As shown in Figure 5.3, the resulted mixture did show the formation of large capsaicin crystals (~ 30 — 50 µm), confirming that capsaicin is not miscible with silicones.

The miscibility of capsaicin with silicones can be predicted by calculating the interaction parameter, χ12, which can be calculated by applying equation 5.1. However, equation 5.1 can not be used directly without knowing the solubility parameter of

154

Figure 5.3 Optical microscope (transmission bright field) image of resulting capsaicin distribution in the bulk of Sylgard® 184 base material. Toluene was used as the common solvent. The concentration of capsaicin in the matrix was 1 wt%, and the (solvent: polymer) ratio was (20: 80) by mass. The image size is (570 x 480) µm.

capsaicin, δ2, which is not available in the literature. Alternatively, we predicted the

solubility parameter of capsaicin by the group contribution method, according to the

relation:

δ = (∑ Fi) / (M/ρ) (5.2)

where Fi is the molar attraction constant of group i in capsaicin structure, and M and ρ are

respectively the molecular weight (305.4 g/mol) and density (1.15 g/cm3) of capsaicin.

Two sets of Fis were used (Hoy’s and van Krevelen’s, both found in van Krevelen

155 (1972)), and the solubility parameter of capsaicin was 22.44 MPa1/2 and 25.77 MPa1/2 by applying Hoy’s and van Krevelen’s data, respectively. Using the average of the two estimated solubility parameters of capsaicin into equation (6.1), the χ12 value for

capsaicin/PDMS system was estimated to be 9.092, indicating that capsaicin is

immiscible with PDMS. In Table 5.6, the properties of the solvent used (toluene) are also

summarized with the properties of capsaicin for comparison. Therefore, even when

capsaicin was mixed with silicone using a common miscible solvent (toluene, χ12 =

0.469), it phased separated from silicone upon removal of the solvent, and resulted in the

formation of large capsaicin aggregates/crystals, as observed experimentally.

Table 5.6 Physical parameters of relevance importance to the miscibility of capsaicin/PDMS. V and δ are the molar volume and the solubility parameter of the material, respectively. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS*.

V δ ∆δ12 Material 3 ½ ½ χ12 (cm /mol) (MPa) (MPa)

Toluene 106.8 18.2 a 3.3 0.469

Capsaicin 265.6 24.1 b 9.2 9.092

a, Values obtained from (Rodriguez, 1989). b` Value predicted in the current study by the group-contribution method (see text). * For PDMS, the literature value of its solubility parameter (14.9 MPa½, from: Rodriguez, 1989) was used.

156 5.3 Leaching evaluation

5.3.1 Leaching of benzoic acid

The BA-entrapped silicone coatings were subjected to leaching studies in static

cells. The effects of different preparation conditions on the leaching behaviors are

presented in the following subsections. The parameters varied were: solvent type

(acetone v.s. toluene), and matrix type (RTV11 v.s. Sylgard® 184). For all the above combinations, the solvent/polymer ratio was fixed at 20/80 by mass, and the wt. % BA in

the matrix was fixed at 1 wt %.

The cumulative leaching of benzoic acid from silicones is shown in Figure 5.4.

Benzoic acid had shown the highest leaching amongst all the antifoulants used in the

current study, regardless of the matrix (Sylgard® 184 or RTV11) or the solvent (acetone

or toluene) used. For benzoic acid/Sylgard® 184 coatings, prepared using acetone as the

solvent, after 1 week and 1 month of immersion, about 73 and 85 % of the initial benzoic

acid content, respectively, had leached out from the coating. When toluene was used as

the solvent or RTV11 as the carrier (acetone as the solvent), benzoic acid completely depleted from the coating in about one month. The leaching rate (slope from the curve) for BA from Sylgard® 184 coating were 83 µg/cm2/day and 2.4 µg/cm2/day during the

slow and fast leaching periods, respectively. Similarly, the leaching rate (slope from the curve) for BA from RTV11 coating were 86 µg/cm2/day and 5.9 µg/cm2/day during the

slow and fast leaching periods, respectively.

157

500 ) 2 g/cm

µ 400

300

200

Cumulative leaching, Q ( Q leaching, Cumulative 100

0 0 5 10 15 20 25 30 time (day)

Figure 5.4 Cumulative leaching of BA from its incorporated silicone coating: Sylgard® 184 (open symbols) or RTV11 (filled symbols). The common solvent used for BA/silicones were acetone (squares) and Toluene (circle). The initial concentration of BA in all coatings was kept constant at 1 wt%., and the solvent/polymer ratios were kept constant at 20/80 by weight for all combinations.

158 The crystal formation behavior of benzoic acid, as illustrated using various

solvents, could likely be the primary reason for its considerably high leaching. It is

possible in some occasions that some large crystals could span the entire thickness of the

coating, and therefore there will be little mass transfer resistance for BA to leach out.

The crystal formation and consequently high leaching of BA could be the main reason for

the relatively short period of antifouling effectiveness reported in the literature (Railkin,

1995), even though in that study a vinyl-rosin coating was used. Therefore, bacterial

attachment studies were not performed and not recommended in the current study for

BA/silicone coatings, because of three reasons. First, the bulk concentration of the

bacterial solution at the initial days of immersion will be most likely much above the

EC50 of BA, which is about 7 ppm towards various types of bacteria (Haque et al., 2005), and consequently the coating will effectively inhibit bacterial attachment but by a toxic mechanism. Second, after one month of immersion, most of the compound will leach out from the coating and hence the coating will not be effective for reducing the bacterial attachment for a longer time. Third, a considerable increase in surface roughness will be expected for BA/silicone coatings during immersion in water because of the high leaching rate of the compound, a factor that will accelerate bacterial attachment after

longer time.

5.3.2 Leaching of capsaicin

The capsaicin entrapped RTV11 coating (using toluene as the common solvent)

was subjected to leaching studies in static cells, and the results are shown in Figure 5.5.

159 250

200 ) 2 150

100 Q (µg/cm 50

0 0 5 10 15 20 25 30 immersion time (day)

Figure 5.5 Capsaicin cumulative mass per area (Q, in µg/cm2) released from 1 wt % capsaicin-incorporated RTV11 sheet (total mass and surface area of the sheet were respectively 6.55 g and 114 cm2), plotted against time of immersion in DI water. Toluene was used as the common solvent to mix capsaicin with RTV11, and the (solvent: polymer) ratio was (20: 80) by mass.

As shown in Figure 5.5, capsaicin leached out rapidly from RTV11 within the first 7 days, and then slowed down as time proceeded. The cumulative mass leached out after

the first and fourth weeks were about 161 and 198 µg/cm2, respectively. Approximately

35 % of capsaicin original mass had leached after one month of immersion. The leaching

rate (slope from the curve) for capsaicin from RTV11 coating were 75 µg/cm2/day and

1.6 µg/cm2/day during the slow and fast leaching periods, respectively. By extrapolation,

it would take approximately 8 months for capsaicin to leach out completely.

160 Capsaicin was also blended with RTV11 using ethanol as the solvent, and the

corresponding leaching data are shown in Figure 5.6. For this particular system, two

cases were considered: capsaicin was homogenized with RTV11 base polymer before (or

after) mixing with DBT catalyst. The objective here was to investigate the effect of

mixing order during the preparation conditions on the leaching behavior. To accurately

compare the mass flux, the surface area and thickness of the coatings were consistent for

both cases. As shown in Figure 5.6, at any particular time, the cumulative leaching was

higher if capsaicin was mixed afar adding the catalyst to the polymer base. This experiment did show the importance of mixing the antifoulant/solvent mixture to the polymer base before adding the curing agent in order to get a more homogeneous coating and consequently a more controllable leaching, although for the case of capsaicin its

leaching was just slowed down slightly. To summarize, the above experiments did also

show that the capsaicin leaching from RTV11 was relatively high, regardless of using

toluene or ethanol as the solvent, or regardless of the mixing order.

The relatively high leaching of capsaicin from RTV11 could be the results of the following reasons. First, capsaicin is immiscible in silicones as discussed above (χ12 =

9.09). Second, partial degradation and erosion of the RTV11 matrix in water, which has

been confirmed experimentally (Bullock et. al., 1999), could contribute and facilitate the

leaching. Third, RTV11 has a high content of inorganic fillers (32 wt% CaCO3), which could lead to presence of “voids” in-between the filler particles and at the filler/polymer interface and allow water molecules to seep into the silicone matrix through these empty spaces and carry the dissolved capsaicin molecules with them as they leave the coatings.

161 200

150 ) 2

100 Q (µg/cm 50

0 0 1020304050 time (day)

Figure 5.6 Effect of the mixing order on the capsaicin cumulative leaching from 1 wt % capsaicin-incorporated RTV11 sheet (total mass and surface area of the sheet were respectively 1.64 g and 38 cm2). Ethanol was used as the common solvent, and the (solvent: polymer) ratio was kept constant at (20: 80) by mass. The open squares correspond to the conditions of mixing capsaicin/ethanol solution with the RTV11 base and drying off the solvent before adding the catalyst. The filled squares correspond to the same conditions of the open square data, except that the capsaicin/ethanol solution was mixed after adding the catalyst.

The fourth reason that could be the large crystals/ aggregates formed inside the bulk of the coating matrix, contribute to the fast leaching, which was also observed for the other compound, benzoic acid, incorporated into Sylgard® 184, where it was found that the leaching of benzoic acid was extremely high from both Sylgard® 184 and RTV11 matrices (section 5.3.1), much higher than the leaching of capsaicin from RTV11 matrix.

162 5.4 Bacterial attachment evaluations for capsaicin-RTV11 coatings

5.4.1 Effect of immersion in water on coating’s properties

Before evaluating the antibacterial performance of the capsaicin - incorporated

RTV11 coating, it is worthwhile to examine the effects of the water type and the immersion time on the coating properties. This is important in order to confirm that the difference in bacterial attachment – if exist – is due solely to capsaicin leaching and not due to a change in the coating properties. This factor was investigated by immersing control samples of RTV11 coatings in different types of water and evaluating the surface and bulk properties of the coatings, as discussed below.

Control capsaicin-free RTV11 coatings were immersed in two types of water

(sterilized DI water and enriched LE water) for up to 14 days to study the effect of water type and immersion time on the wettability of RTV11. For coatings in sterilized DI water, both static and dynamic water contact angles were taken; while for coatings in enriched LE water, only static contact angles were taken. The results are presented in

Figure 5.7. The static contact angles almost remained constant at a value around 100° for the 14 day immersion period, irrespective to the type of water used. The advancing contact angles also remained almost constant at a value around 103°. In general, the static contact angles resemble the advancing contact angles with slightly lower values

(Adamson, 1990). The receding contact angles, however, showed a gradual decrease, down to a value of 80° at the end of 14 day period. Consequently, the contact angles

163 hysteresis (difference between the advancing and receding angles) increased from 8°

initially to 23° after 14 days of immersion. The increase in contact angles hysteresis is

possibly due to slow surface erosion, which would result in a slight increase in surface

roughness. Surface erosion could be the result of a continuous small mass loss of fillers

such as CaCO3 from RTV11, and the micro-pit formation on RTV11 surfaces upon

immersion in water (Bullock et. al., 1999). Our scanning probe microscopy also verified

the increase in the surface roughness value (Rq, with a scan size of 80 µm x 80 µm) from

6.9 nm for as prepared RTV11 films (Figure 5.8(a)) to 11.8 nm for 14 day water-aged

RTV11 films (Figure 5.8(b)).

In addition, the wettability and surface roughness for water-aged 1 wt% capsicin/RTV11 coatings were also evaluated. For 1 wt% capsaicin-blended RTV11 coatings immersed in DI water for 14 days, the advancing contact angles increased to a value of 109°, whereas the receding contact angle decreased to a value of 79°. As a result, the contact angles hysteresis increased from 16° initially to 30° after 14 days of immersion, indicating an increase in surface roughness. The increase in surface roughness for 1 wt % capsaicin-incorporated RTV11 coatings was confirmed by scanning probe microscopy, where it was observed that the surface roughness increased considerably from ~ 12 nm for as prepared capsaicin-treated RTV11 films (Figure 5.8(c)) to ~ 88 nm after 14 days of immersion (Figure 5.8(d)). The increase in surface roughness here is mainly due to the high leaching of capsaicin, where it is possible that capsaicin could leave behind irregular surface geometries (mostly holes) when it leached out, which would cause the surface roughness to increase significantly. By refereeing to the

164

110

100

contact angles contact 80

70 03691215 immersion time (day)

Figure 5.7 Effect of water immersion time on the wettability of RTV11 films in terms of the static contact angles taken for RTV11 immersed in Lake Erie (Δ) and deionized water (▲). The advancing (□) and receding () contact angles taken for RTV11 immersed in deionized water are also presented. Error for each data point (average over 12 measurements) is presented by the vertical line.

165

(a) (b)

Figure 5.8 Surface topographic images of Capsaicn-RTV11 coatings. (a) as-prepared controlled RTV11 film, surface roughness: 6.7 nm; (b) as-prepared RTV11 film containing 1 wt % capsaicin, surface roughness: 12.3 nm; (c) controlled RTV11 film after 14 days of immersion in DI water, surface roughness: 10.5 nm; (d) RTV11 film containing 1 wt% capsaicin after 14 days of immersion in DI water, surface roughness: 88.0 nm. The images (scan size: 80 µm x 80 µm; z-scale: 400 nm) were generated using scanning probe microscopy with the non-contact mode at a scan rate of 0.20 Hz.

166 leaching data, about 30 % of capsaicin original mass had leached after 14 days of

immersion. This amount is high enough for the surface roughness to increase

considerably.

A 42 day immersion period was used to study the effects of water immersion time

as well as bacteria and dissolved capsaicin on the bulk modulus of RTV11. As-prepared

RTV11 coatings were immersed in three different waters (sterilized DI water, enriched

LE water with bacteria, and sterilized LE-20 ppm water). As shown in figure 5.9, the bulk modulus slightly decreased from 1.56 MPa to 1.37 MPa after the 42 days of

immersion in all three waters, with no observed effect from the water type. This

indicates that bacterial attachment or dissolved capsaicin has minimum contribution to

the variation of the elastic modulus of RTV11. Two factors may contribute to the slight

decrease in bulk modulus. First is a slow leaching of CaCO3 filler from the bulk of the

RTV11 coating. Second reason could be a continuous loss of small amounts of RTV11

constituents other than CaCO3 (Bullock et. al., 1999). Previously, Wynne et. al. (2000) performed a quantative mass loss experiment for RTV11 coatings immersed in DI water, where they proved that the mass loss of RTV11 was about 0.8 wt% after 30 days of immersion. This is in accordance with the general conclusion made by Brady (2000) that hydrosilylation-cured PDMS – such as Sylgard® 184 - are stable in water whereas polycondensation – cured PDMS – such as RTV11 - are not stable in water.

In summary, it can be summarized from this subsection that the type of water will

not have effect on the properties of the coatings and consequently will not have effect on

167

3.0 ) 2.4

1.8

1.2

0.6

Elastic Modulus (MPa Modulus Elastic 0.0 0 1020304050 immersion time (day)

Figure 5.9 Effect of water type and immersion time on the elastic modulus of RTV11 films immersed in different types of water samples (sterilized Lake Erie water: Δ, enriched Lake Erie water: ○, and sterilized Lake Erie water with 20 ppm capsaicin: □). Error for each data point (average over 6 measurements) is presented by the vertical line.

the antibacterial performance of the coatings. However, immersion in water will cause

the surface roughness for capsaicin-blended coatings to increase considerably. This considerable increase in surface roughness could enhance bacterial attachment, unless the

amount of capsaicin leached out is high enough to inhibit bacterial attachment, as to be

discussed in the next section. 168 5.4.2 Bacterial attachment evaluations

In order to evaluate the coatings antibacterial behaviors, bacterial attachment

studies using fresh waters containing indigenous enriched microbial consortium isolated

from Lake Erie water were performed. Some representative bacterial attachment images are presented in Figure 5.10. As shown in the figure, much less bacteria were attached to

capsaicin-blended RTV11 coating as compared to RTV11 coating alone. By defining the

% reduction in bacterial coverage to be (1-A/B) 100, where A and B refer to the area

coverage for capsaicin-blended RTV11 coating and control RTV11 coating, respectively, the % reduction was estimated to be (58 ± 11) %. However, based on the leaching data shown previously, the concentration of capsaicin in solution for the immersion period shown in Figure 5.10 was approximately 4-6 ppm. This concentration was very close to

the EC50 of capsaicin [~ 5 to 20 ppm towards various bacteria (Xu et al. 2005)].

Therefore, we are not sure if the reduction of bacterial attachment on the capsaicin-

treated surface shown here is because of the bacteria simply died off or because of the

non-toxic mode of action of capsaicin.

In summary, the clear reduction in bacterial population on the capsaicin-treated

RTV11 coating suggests that capsaicin can effectively inhibit the attachment of bacteria

we tested. However, the antibacterial effectiveness of the capsaicin-treated RTV11

coating is likely short lived due to the relatively high leaching of capsaicin coupled with

the dramatic increases in surface roughness of the coatings when immersed in water.

169

(a) (b)

Figure 5.10 Optical microscopic (reflection bright field) images of bacterial attachment for capsaicn-RTV11 coatings. (a) 1 wt% Capsaicin/RTV11, (b) control RTV11. The coatings were immersed in water containing Lake Erie bacteria for 14 days. The size for both images is (285 x 215) µm.

170

CHAPTER VI

RESULTS AND DISSCUSSION FOR TANNIC ACID – BASED COATINGS

In this chapter, the results obtained for tannic acid (TA) - incorporated silicone

coatings are presented and discussed. The effect of incorporating the compound on the

coating’s properties is presented in section 6.1. The miscibility/bulk morphology of

tannic acid with the polymer matrix is discussed in section 6.2. The leaching of tannic

acid in water is presented in section 6.3.

6.1 Effect of tannic acid on coating’s properties

Tannic acid (TA) was incorporated into two types of silicones (Sylgard® 184 and

RTV11), up to a concentration of 4 wt% TA/matrix. For both combinations, it was observed that the TA-blended coatings were cured similarly as the TA-free coatings alone. Further examining the effect of incorporated antifoulant on the wettability of the silicone coatings qualitatively confirmed this observation. The wettability was examined in terms of measuring the water contact angles and the results are presented in Figure 6.1.

As shown in the figure, the low content of TA in the matrix did not considerably affect the wettability for both types of silicones. For the TA/silicone coatings, the elastic modulus was not measured, but at the low TA matrix loading used in the current study 171

110

105

100

contact angles 95

4 4 1 8 18 1 rd ard1 a lg lg RTV y y S /S TA/RTV11 TA

Figure 6.1 Static water contact angles of TA-entrapped silicone films compared to that of the controlled TA-free silicones. The concentration of TA in the matrix was fixed at 1 wt% for both combinations. The (solvent: polymer) ratio was fixed at (20: 80) by mass (acetone was the solvent for both combinations). Error for each data point (average over 12 measurements) is presented by the vertical line.

172 the modulus do not expect to differ two much from the corresponding value of that of

control silicone coatings.

6.2 Miscibility of tannic acid in silicones

To roughly examine TA/silicones miscibility; optical microscopic images were taken for the bulk of Sylgard® 184 contained the incorporated compound after drying off

the solvent. The solvent used here was acetone, and the solvent/polymer ratio was 20/80

by mass. As shown in Figure 6.2, small aggregates (~ 1 — 3 µm) were distributed

uniformly throughout Sylgard® 184 matrix. The resulted aggregate size here was

considerably small, much smaller than the crystal size of benzoic acid and capsaicin, and

comparable to the minimum aggregate size obtained for sodium benzoate in Sylgard®

184. This could be attributed to the following reason. Acetone is a good solvent for

tannic acid. Also, acetone is quickly dried off, and has some miscibility with silicones.

However, as shown in figure 6.2, phase separation was observed clearly for the tannic

acid/ Sylgard® 184 system. This implies that, despite the excellent and fine distribution

of the compound inside the matrix, tannic acid is not soluble in the polymer phase. Also,

Figure 6.2 demonstrates the effect of the compound matrix loading on the aggregate size.

While increasing TA matrix loading from 1 wt% TA/polymer to 4 wt% TA/polymer had

resulted in increasing the number of aggregates, the aggregate size did not change

considerably. This result had some similarity with the effect of NaB matrix loading on

the aggregate size (Figure 4.6 in Chapter 4), and therefore they supported each others.

173 (a) (b)

Figure 6.2 Optical microscope (transmission bright field) image of resulting TA distribution in the bulk of Sylgard® 184 matrix: (a) 1 wt% TA/Polymer; (b) 4 wt% TA/Polymer. Acetone was used as the common solvent, and the solvent/polymer ratio was 20/80 by mass. The image size is (570 x 480) µm.

The miscibility of tannic acid with silicones can be predicted by calculating the interaction parameter, χ12, which can be calculated according to the same previous

equation discussed before in chapter 4:

2 χ12 = (V1/RT) (δ1 – δ2) (6.1)

However, equation 6.1 can not be used directly without knowing the solubility parameter

of tannic acid, δ2, which is not available in the literature. Alternatively, the solubility

parameter of tannic acid was predicted here by the group contribution method, according

to the relation:

174 1/2 δ = [(∑ Ecoh, i) / (∑ Vi)] (6.2)

where Ecoh, i and Vi are respectively the molar cohesion energy and molar volume of

group i in the structure of tannic acid. The values used here for Ecoh, i and Vi were obtained from Fedors’ table (found in van Krevelen (1990)). Consequently, the predicted value for the solubility parameter of tannic acid was 36.60 MPa1/2. By evaluating the

predicted solubility parameter of tannic acid into equation 7.1, the χ12 value for tannic

acid/PDMS system was estimated to be 14.06, indicating that tannic acid was not soluble

in the polymer phase, as observed experimentally. In Table 6.1, the properties of the solvent used (acetone) are also summarized with the properties of tannic acid and PDMS for comparison.

Table 6.1 Physical parameters of relevance importance to the miscibility of tannic acid/PDMS system. δ is the solubility parameter of the material. ∆δ12 is the difference in solubility parameters between the material and PDMS. χ12 is the interaction parameter between the material and PDMS.

∆ ∆δ12 Material ½ ½ (MPa) (MPa) χ12

Tannic acid 36.6 a 21.7 14.06

Acetone 20.3 b 5.4 0.871

PDMS 14.9 b - -

a` Value predicted in the current study by the group-contribution method (see text). b, Values obtained from (Rodriguez, 1989).

175 6.3 Leaching evaluation

The tannic acid-entrapped silicone coatings were subjected to leaching studies in static cells. In this experiment, only one preparation condition parameter was varied, keeping the other parameters unchanged. The parameter varied was the matrix type

(RTV11 v.s. Sylgard® 184). For all the above combinations, the solvent/polymer ratio was fixed at 20/80 by mass, and the wt. % TA in the matrix was fixed at 1 wt%, and the solvent type was fixed (acetone). The cumulative leaching of tannic acid from both matrices is shown in Figure 6.3. For tannic acid/Sylgard® 184 coatings, a very slow leaching was observed. About 0.14% and 0.17 % of the initial tannic acid content had leached out from Sylgard® 184 coating after 1 week and 1 month of immersion, respectively. Changing the coating carrier to RTV11 did considerably affect the leaching. For 1 wt% tannic acid/RTV11 after 1 week and 1 month of immersion, it was found that about 1.3% and 1.6% of the initial mass, respectively, had leached out. For the two combinations, a general behavior was observed: tannic acid leached out fast in the initial few days, and continued to leach out but with a much slower rate. The leaching rates for tannic acid form Sylgard® 184 were 0.15 µg/cm2/day and 0.01

µg/cm2/day at the initial leaching stage and the final leaching stage, respectively. For tannic acid/RTV11 system, on the other hand, the initial and final leaching rates were 2.0

µg/cm2/day and 0.10 µg/cm2/day, respectively.

Compared to the other antifouling compounds investigated in the current study, tannic acid had shown to have the slowest leaching behavior from both matrices. At least

176 20

15 ) 2

Q (µg/cm 5

0 0 6 12 18 24 30 time (day)

Figure 6.3 Cumulative leaching of TA from its incorporated silicone coating: Sylgard® 184 (∆) or RTV11 (▲). The common solvent used was acetone for both combinations. The initial concentration of TA in both coatings was kept constant at 1 wt%., and the solvent/polymer ratio was kept constant at 20/80 by weight for both combinations. Error for each data point (average over 2 batches) is presented by the vertical line.

two reasons could be attributed here to explain this observation. First, tannic acid had

shown the formation of fine distribution of very small aggregates (~ 1 — 3 µm) inside the silicone matrix, and we have enough evidence now that the leaching will decrease if the aggregate size decreases. Second, tannic acid has the largest molecular weight amongst 177 all the compounds investigated in the current study (the molecular weight of tannic acid

is 1700 g/mol, which is about 12 times higher than the molecular weight of sodium

benzoate), and it is well-known that in general the heavier molecule diffuses slower than

the lighter molecule.

The above results clearly showed the effect of the matrix type on the leaching of

tannic acid, where the leaching of tannic acid from RTV11 was higher than the leaching from Sylgard® 184. This experimental result supports the previous finding for sodium

benzoate/silicones regarding the effect of the matrix type (section 4.4.4), where it was observed that the leaching of sodium benzoate from RTV11 coating was higher than the leaching from Sylgard® 184. Therefore, the reasons for the observed matrix type on the

leaching of tannic acid are similar to what we discussed before in section 4.4.4 and need

not to be repeated.

In summary, it can be evident from the leaching data presented here that leaching of

TA from silicone coatings into water is very slow. This could be an advantage for TA-

incorporated coatings. For example, in the case if TA/silicone coatings are to be

immersed in bacterial water solution for evaluating the antibacterial performance, the

bulk water concentration of TA will be certainly much below the EC50 of TA [~ 118 ppm

(Xu et al., 2005)], and hence bacteria will not die, and hence if a reduction in bacterial

attachment observed most likely it will be by a non-toxic mode of action. However,

bacterial attachment was not performed here for TA/silicone coatings; because this is

beyond the scope of the current study.

178

CHAPTER VII

CONCLUSIONS AND RECOMMENDATIONS

7.1 Conclusions

As stated in the objectives section of this dissertation, the major objective is to

arrive at the connection between the miscibility/distribution of less-toxic antifouling

compounds in a monolithic, hydrophobic polymer coating and their leaching in water. A

minor simultaneous objective is to assess the applicability of applying the solvent-

assisted blending technique as a straightforward incorporation method for the purpose of

controlling the release of four less-toxic antifoulants.

The secondary objective was achieved first. Four significantly less toxic compounds (sodium benzoate, benzoic acid, capsaicin, and tannic acid), as compared to tin-based antifoulants, were incorporated into two types of silicone coatings (Sylgard®

184 and RTV11). The incorporation was achieved by the solvent-assisted blending technique. The feasibility of this technique was assessed by examining experimentally the morphological structure of the compounds in the matrix and their leaching into

Deionized (DI) water. The solvent-assisted blending technique was found to be useful for the cases of sodium benzoate and tannic acid, evidenced by the even dispersion of the small and uniform aggregates of the compounds inside the coating, which had shown the 179 advantage of controlling the release. On the other hand, the solvent-assisted blending technique was not suitable for the cases of capsaicin and benzoic acid, owing to the observation that these two compounds had shown the tendency of forming large crystals inside the coatings, regardless the different solvents used to control the crystal size. The formation of large crystals was the main reason for the fast leaching of benzoic acid and capsaicin observed from the coating carriers, thus truncated their usage as antifoulants to be incorporated into a coating by the solvent-assisted blending technique.

Based on the main findings obtained for the secondary objective, the toxicity of the compounds, and the costs of the compounds, sodium benzoate-incorporated Sylgard®

184 coating was selected in the current study as the model system. Such a model system was used to determine, based on detailed experimental observations and theoretical analysis, the miscibility-leaching relationship. Experimentally, sodium benzoate was found to exhibit slow and controllable leaching by tuning the preparation conditions (the solvent composition, solvent/polymer ratio, and compound/polymer ratio). A fine and uniform aggregate size distribution (~ 3 µm) was obtained at a 20/80 solvent/polymer mass ratio and at a solvent composition of 50/50 water/acetone mass ratio, which had resulted in the lowest value for the steady leaching rate of about 0.1 µg/cm2/day.

Empirical correlations between the effects of the aggregate size and the matrix loading of

sodium benzoate and its leaching rate were obtained. It was concluded that increasing

the aggregate size had a sharp effect on increasing the leaching rate, whereas increasing

the matrix loading (up to 5 wt. %) had a mild effect on the leaching rate. Moreover, as a

supplementary corollary to the study, 1 wt% sodium benzoate/Sylgard® 184 coatings with

180 fine and uniform aggregate distribution exhibited enhanced antibacterial behaviors as

compared to Sylgard® 184 coatings alone. This suggested that sodium benzoate could be

an environmental friendly alternative to the currently used toxic biocides in antifouling

applications, and highlighted the benefit of applying the solvent assisted blending

technique as the incorporation method.

Theoretical thermodynamic analysis was performed for predicting the miscibility

of sodium benzoate with Sylgard® 184 matrix using acetone/water blends as the solvent.

The quaternary Flory-Huggins model was extended to include the electrostatic

contribution and the concentration-dependent interaction parameters. Comparison was

made between the theoretical miscibility trends and the experimental morphology trends.

The extended Flory-Huggins model was found to be more accurate than the original

Flory-Huggins model, and both did predict that the system was not miscible, as observed

experimentally. Both the original Flory-Huggins model and the extended Flory-Huggins

model also captured qualitatively most of the important effects of the preparation

conditions on the aggregate size of NaB in Sylgard® 184 matrix, and the limitations of

both models to accurately predict all the effects of all the preparation conditions were

likely attributed to the total ignorance of the dynamic drying. Nevertheless, the rough

prediction based on the thermodynamic models did qualitatively describe most of the features of our system, and still useful as as a preliminary guide for selecting the

components of the coating system.

181 Theoretical mass transfer analysis was performed for the leaching of sodium

benzoate from Sylgard® 184 matrix, in an attempt to elucidate on the leaching

mechanism. Based on the analysis, and combing it with the results from the miscibility

and thermodynamic study, the following mechanism was proposed: “Sodium benzoate is

insoluble in the polymer phase; therefore, the diffusion of the compound is taking place

through pores (empty space) filled with water within the matrix, not through the

continuum of the polymer phase”. For our experimental conditions, where the highest

matrix loading was only 5 wt%, the small particles (~3 µm in size) uniformly distributed

in the matrix may not necessarily be connected to each other. Instead, constricted

channels of very narrow spacing could spread out throughout the matrix and connect the

particles between each others to allow for water to diffuse through these channels and

dissolve the particles. In this case, the porosity would be very small and the diffusion

path would be very tortuous. This would slow down the leaching process extremely after

a longer period, unless the capillary rise is capable of enhancing the flow of water

through these constricted channels, which is our speculation.

7.2 Recommendations for future work

The current study concluded that sodium benzoate is the most attractive

antifoulant among the four antifoulants investigated in the current study. The followings

(recommendations 1-8) are recommended as a continuation of the current study to

182 investigate more on the miscibility, leaching and antifouling properties of sodium

benzoate-based coatings:

(1) It is necessary to perform experimental microscopic analysis for the change in the

morphological structure and porosity of the coatings during immersion in water.

Scanning Electron Microscopy (SEM) is proposed and applied in the literature as

a powerful technique for the microscopic study of antifouling paints (Fay et al.,

2005), where the sample is usually cross-sectioned along the thickness of the

coating. The data obtained by SEM will give valuable visualization of the actual

diffusion front, and will significantly improve/verify our understanding of the

leaching mechanism. Optical Microscopy could be also useful here.

(2) It is also recommended to combine SEM experiments with independent

experiments to measure directly the porosity, both experiments have to be

performed in parallel. Standard methods exist in the literature for measuring the

porosity experimentally (e.g. Gurny and Peppas, 1982, Miller et al. 1983).

(3) In the current study, the maximum loading of sodium benzoate in Sylgard® 184

matrix prepared by the solvent blending technique was 5 wt%. This would result

in a very small porosity, which could cause the leaching rate to be extremely low

after longer time. Two recommendations are provided here to eliminate this

concern. First, it is recommended to increase the loading to above 5 wt% and

possibly up to 20 wt.%. This will certainly result in the formation of

agglomerates “large aggregates”, but this is not a crucial problem because

standard procedures in the polymers/fillers compounding industry are available to 183 divide the agglomerates into small aggregates. In this case the leaching will be

assured to be continuous until most of the compound is leached out, and the

leaching mechanism will be similar to mechanism (a) and mechanism (b)

described in Figure 4.20. The second recommendation is, if we want to keep the

matrix loading below 5 wt%, is that the initial porosity has to be increased.

Techniques exist in the drug delivery literature (e.g. Miller and Peppas, 1983) to

generate initial porosity of the matrix, which proved to be effective for controlling

the release. The initial porosity that we seek for our system may not have to be

high; a value of 0.1 could be satisfactory for the initial porosity.

(4) The above recommendations are made for the combination sodium

benzoate/silicone coating, where the leaching process is governed by the property

that sodium benzoate is insoluble in the polymer phase. It is recommended also

to incorporate sodium benzoate into a polymer matrix in which the compound has

some solubility on it. In this case, the release mechanism will be totally different

from what we proposed for the system of the current study, because compound

diffusion through the continuum of the polymer phase will take place and govern

the release. One advantage for applying recommendation (4) is that it is possible

to incorporate low matrix loading (< 5 wt. %) and at the same time to assure that

the compound will leach out continually for a long period of time.

(5) The thermodynamic model applied in the current study is useful for roughly

predicting the miscibility of the system. However, it does not account for the

dynamic effect (solvent vaporization and the solidification of sodium benzoate),

184 which appears to be crucial. Further theoretical work is needed to combine the

dynamic effect with the thermodynamic model.

(6) Mass transfer analysis for the leaching study is also recommended if

recommendations 3 and 4 are to be applied. The existing drug release models

described in chapter 2 are expected to be adequate for this purpose, and the

selection of a model depends on the concentration of sodium benzoate in the

matrix and on the compound/polymer solubility (as described in details in section

2.4.1 and section 2.4.2).

(7) It is also recommended to incorporate sodium benzoate into real marine coatings,

which belong to the class “self-polishing polymers”. In this case, the leaching

mechanism will be totally different from silicone coatings, because here both

matrix erosion and compound diffusion contribute to the release of the compound.

Mass transfer analysis for the leaching study is also recommended in parallel.

The drug release models described in section 2.4.1 and section 2.4.2 are not

adequate for this purpose. Instead, the model developed by Kill et al. (2002) for

the analysis of self-polishing antifouling coatings will be useful.

(8) The current study assessed the antifouling performance of sodium

benzoate/silicone coatings by conducting bacterial attachment studies. It is

recommended to asses the antifouling performance of the coatings by using the

common fouling organisms, such as algae, tubeworms and diatoms.

For benzoic acid and capsaicin, the current study proved that the leaching rates of the compounds were high, due to large crystal formation of the compounds inside the

185 matrix. In order to utilize any of them as an effective antifoulants, alternative

incorporation methods other than the method described in the current study are required

in order to control the release of the compounds. Sundberg et. al. (1997) described two methods to achieve constant slow release rate for a long period of service life. They applied these two methods for controlling the release of Sea-nine 211 (a commercial antifoulant). The first method, the “reservoir membrane” method, was a two-layer coating: the base layer composed of a highly concentrated Sea-nine 211 homogenouesd with a plasticizer, and a top layer composed of RTV silicone. The second method was by microencapsulating the active compound before dispersing it in the polymer coating. By applying these methods, they were able to slow down and control the release rate of Sea- nine 211 considerably. Recently, Xing et. al. (2004) reported an efficient experimental procedure for producing capsaicin microcapsules. It will be useful to investigate the possibility of applying these two methods in order to control the release of benzoic acid and capsaicin.

Finally, a previous work in our research group (Barrios et al., 2005) proved that zosteric acid - another nontoxic antifouling compound – incorporated into silicone coatings was effective in inhibiting bacterial attachment. Zosteric acid has shown some similarity with sodium benzoate regarding the miscibility and leaching behavior from silicone coatings, and continues to be attractive nontoxic antifouling compound.

Therefore, the recommendations listed for sodium benzoate (recommendations 1-8) are also recommended for zosteric acid.

186

REFERENCES

Abdul Azis P K, Al-Tisan I, Sasikumar N (2001) Biofouling potential and environmental factors of seawater at a desalination plant intake. Desalination 135: 69-82

Abdul Azis P K, Al-Tisan I, Sasikumar N, Al-Daili M, Green T N, Ba-mardouf K, Al- Qahtani S A, Al-Sabai K (2003) Marine macrofouling: a review of control technology in the context of an on-line experiment in the turbine condenser water box of Al-Jubail Phase-I power/MSF plants. Desalination 154: 277-290

Adamson A W, Gast A P (1990) Physical Chemistry of Surfaces. John Whily & Sons, New York

Al-Ahmed M, Abdul-Aleem F A, Mutairi A, Ubaisy A (2000) Biofouling in RO membrane systems. Part 1: Fundamentals and control Desalination 132: 173-179

Al-Juhni A A (2001) Modeling and simulation of a combined isomerization reactor and pressure swing adsorption unit. MS thesis, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia

Altinkaya S A, Ozbas B (2004) J Membr Sci 230: 71-89

An Y H, Friedman R J (1998) J Biomed Mater Res 43: 338

Arce F T, Avci R, Beech I B, Cooksey K E, Cooksey B W (2003) Microelastic properties of minimally adhesive surfaces: A comparative study of RTV11™ and Intersleek elastomers™. J Chem Phys 119: 1671-1682

Axiak V, Vella A J, Agius D, Bonnici P, Cassar G, Cassone R, Chircop P, Micallef D, Mintoff B, Sammut M (2000) Evaluation of environmental levels and biological impact of TBT in Malta (central Mediterranean). Sci Tot Environ 258: 89-97

Balazs A C, Singh C, Zhulina E (1998) Modeling the interactions between polymers and clay surfaces through self-consistent field theory. Macromolecules 31: 8370-8381

Banerjee S, Asrey R, Saxena C, Vyas K, Bhattacharya A (1997) Kinetics of Diffusion of Water and DMMP through PDMS membrane using coated quartz piezoelectronic sensor. J Appl Polymer Sci 65: 1789-1794

187 Barrios C A, Xu Q, Cutright T, Newby B-m (2005) Incorporating zosteric acid into silicone coatings to achieve its slow release while reducing fresh water bacterial attachment. Colloids and Surfaces B 41: 83-93

Barrios C A (2004) Incorporating zosteric acid into silicone coatings to deter fresh water bacteria attachment. MS thesis, University of Akron

Berglin M, Gatenholm P (1999) The nature of bioadhesive bonding between barnacles and fouling-release silicone coatings. J Adhesion Sci Technol 13: 713-727

Boom R M, Boomgaard Th Vav den, Smolders C A (1994) Equilibrium thermodynamics of a quaternary membrane-forming system with two polymers. 1. Calculation. Macromolecules 27: 2034-2040

Bullock S, Johnston E E, Willson T, Gatenholm P, Wynne K J (1999) Surface science of a filled polydimethylsiloxane-based alkoxysilane-cured elastomer: RTV11. J Colloid Interface Sci 210: 18-36

Bunge A L (1999) Release rates from topical formulations containing drugs in suspension. J Controlled Release 52: 141-148

Bustamante P, Pena M A, Barra J (2000) The modified extended Hansen method to determine partial solubility parameters of drugs containing single hydrogen bonding group and their sodium derivatives. International J Pharmaceutics 194: 117-124

Brady R F, (1999) Properties which influence marine fouling resistance in polymers containing silicon and fluorine. Prog Org Coating 35: 31-35

Brady R F (2000) Clean hulls without poisons: devising and testing nontoxic marine coatings. J Coatings Technol 72: 45-56

Brady R F, (2001) A fracture mechanical analysis of fouling release from nontoxic antifouling coatings. Prog Org Coating 43: 188-192

Brady R F, Aronson C L (2003) Elastomeric fluorinated polyurethane coatings for nontoxic fouling control. Biofouling 19: 59-62

Brady R F, Singer I L (2000) Mechanical factors favoring release from fouling release coatings. Biofouling 15: 73-81

Caprari J J, Meda J F, Damia M, Slutzky O (1990) A mathematical model for leaching in insoluble matrix films. Ind Eng Chem Res 29: 2129-2133

Champ M (2000) A review of organotin regulatory strategies, pending actions, related costs and benefits. Sci Tot Environ 258: 21-71

188 Chan C-M (1994) Polymer Surface Modification and Characterization. Hanser Publishers, New York, 35 – 61

Chapman J S (2003) Biocide resistance mechanisms. Int Biodet Biodeg 51: 133–138

Chase G G (2002) Transport Phenomena in Porous Media. in: Fluid Flow Handbook, Chapter 21, J. Saleh editor, McGraw-Hill, New York, 2002

Chaudhury M K, Whitesides G M (1991) Direct measurement of interfacial interactions between semispherical lenses and flat sheets of poly(dimethylsiloxane) and their chemical derivatives. Langmuir 7: 1013-1025

Cichewicz R, Thorpe P A (1996) The antimicrobial properties of chili peppers (Capsicum species) and their uses in Mayan medicine. J Ethnopharmacology 52: 61-70

Clare A S (1996) Marine natural product antifoulants: status and potential. Biofouling 9: 211-229

Cobby J, Mayersohn M, Walker G C (1974) Influence of shape factors on kinetics of drug release from matrix tablets I: Theoretical. J Pharm Sci 63: 725-737

Dabral M, Francis L F, Seriven L E (2002) AIChE J 48: 25-37

Dankert J, Hogt A H, Feijen J (1986) CRC Crit Rev Biocompat 2: 219

Davis A, Williamson P (1995) Marine Biofouling: a sticky problem. NERC News

De La Court F H, De Vries H J (1973) The leaching mechanism of cuprous oxide from antifouling paints. J Oil Col Chem Assoc 56: 388-395

Donbrow M, Friedman M (1975) Timed release from polymeric films containing drugs and kinetics of drug release. J Pharm Sci 64: 76-80

Drago R S, Dadmun A P, Vogel G C (1993) Addition of new donors to the E and C model. Inorg Chem 32: 2473-2479

Eklund T (1985) The effect of sorbic acid and esters of p-hydroxybenzoic acid on the proton motive force in Escherichia coli membrane vesicles. J General Microbiol 131: 73- 76

Eddington D T, Crone W C, Beebe D J (2003) Development of process protocols to fine tune Polydimethylsiloxane material properties. 7th International Conference on Miniaturized Chemical and Biochemical Analysis Systems, October 5-9, 2003, Squaw Valley, California, USA, 1089-1092

189 Edwards D P, Nevell T G, Plunkett B A, Ochiltree B C (1994) Resistance to marine fouling of elastomeric coatings of some poly(dimethylsiloxane) and poly(dimethyldiphenylsiloxanes). Inter Biodet Biodeg 39: 349-359

Estarlich F F, Lewey S A, Nevell T G, Thorpe A A, Tsibouklis J, Upton A C (2000) The surface properties of some silicone and fluorosilicone coating materials immersed in seawater. Biofouling 16: 263-275

Evans L V, Clarkson N (1993) Antifouling strategies in the marine environment. J Appl Bacteriol Symp Suppl 74: 119S-12S

Fan L T, Singh S K (1989) Controlled Release: A quantitative Treatment. Springer-Field, New York, 17

Fay F et al. (2005) SEM and EDX: Tow powerful techniques for the study of antifouling coatings. Prog Org Coat 54: 216-223

Farrell S, Sirkar K (1999) A mathematical model of an aqueous – organic partition – based controlled release system using microporous membranes. J Controlled Release 61: 345-360

Flory P J (1953) Principles of polymer chemistry. Cornell University Press, Ithaca

Foldes E (1998) Study of the effects influencing additive migration in polymers. Die Angewandte Makromolekulare Chemie 261: 65-76

Galliano A, Bistac S, Schultz J (2003) Adhesion and friction of PDMS networks: molecular weight effects. J Colloid Interface Sci 265: 372-379

Gander B, Nguyen V P, Nam-Tran Ho, Merkle H P (1995) A new thermodynamic model to predict protein encapsulation efficiency in Poly(lactide) microspheres. J Phys Chem 99: 16144-16148

Gerstl Z, Nasser A, Mingelgrin U (1998) Controlled Release of pesticides into water from clay-polymer formulations. J Agric Food Chem 46: 3803-3809

Ginzburg V, Balazs A C (1999) Calculating phase diagrams of polymer-platelet mixtures using density functional theory: Implications for polymer/clay composites. Macromolecules 32 5681: 5688

Gottenbos B, Busscher H J, van der Mei H C, Nieuwenhuis P (2002) J Mater Sci: Mater Med 13: 717

Gurny R, Doelker E, Peppas N A (1982) Modeling of sustained release of water-soluble drugs from porous hydrophobic polymers. Biomaterials 3: 27-32

190 Haque H, Cutright T J, Newby B-m (2005) Effectiveness of sodium benzoate and benzoic acid as freshwater low toxicity antifoulants when dispersed in solution and entrapped in silicone coatings. Biofouling 21: 109-119

Haslbeck E., Kavanagh C J, Shin H W, Banta W C, Song P, Loeb G I (1996) Minimum effective release rate of antifoulants: measurement of the effect of TBT and zosteric acid on hard fouling. Biofouling 10: 175-186

Hatch E B (2003) Plastic anti-fouling film for boat hull protection. United States Patent No. 6, 607, 826

Hellio C, De La Broise D, Dufosse L, Le Gal Y, Bourgougnon N (2001) Inhibition of marine bacteria by extracts of macroalgae: potential use for environmentally friendly antifouling paints. Mar Environ Res 52: 231-247

Higuchi T, (1963) Theoretical analysis of rate of release of solid drugs dispersed in solid matrices. J Pharm Sci 52: 1145-1149

Hino T, Prausnitz J M (1998) Lattice thermodynamics for aqueous salt-polymer two- phase system. J Appl Polym Sci 265: 372-379

Jaggari K R (2003) Incorporation of capsaicin into silicone coatings for enhanced antifouling performance. MS thesis, University of Akron

Jelvestam M, Edrud S, Petronis S, Gatenholm P (2003) Biomimetic materials with tailored surface micro-architecture for prevention of marine biofouling. Surf Interface Anal 35: 168–173

Jones N, Shabib S, Sherman P M (1997) Capsaicin as an inhibitor of the growth of the gastric pathogen Helicobacter pylori. FEMS Microbiol Lett 146: 223-227

Johnson K L, Kendall K, Roberts A D (1971) Surface energy and the contact of elastic solids. Proc R Soc Lond A324: 301-313

Kohl J G, Bolstes R N (2001) A study on the elastic modulus of silicone duplex or bi- layer coatings using micro-indentation. Prog Org Coatings 41: 135-141 van Krevelen (1990) Properties of Polymers. Elsevier, New York, 189-224

Kurian A L, Starks A N (2002) HPLC analysis of capsaicinoids extracted from whole orange habanero chilli peppers. J Food Sci 67: 957-962

Lambert J W, Sum A K (2006) Molecular dynamics study of the properties of capsaicin in an 1-octanol/water system. J Phys Chem B 110: 2351-2357.

191 Lueck E (1980) Antimicrobial food additives: Characteristics, Uses, Effects. Springer- Verlag, Berlin, 1980, 210-217

Lyatskaya L, Balazs A N (1998) Modeling the phase behavior of polymer-clay composites. Macromolecules 31: 6676-6680

Magonov S N, Reneker D H (1997) Characterization of polymer surfaces with atomic force microscopy. Annu Rev Mater Sci 27: 175-222

Malik A U, Prakash T L, Andijani I (1996) Failure evaluation in desalination plants - Some case studies. Desalination 105: 283-295

Marshall K C, Stout R, Mitchell R (1971) J Gen Microbiol 68: 337

Marson F (1969) Antifouling paints. I. Theoretical approach to leaching of soluble pigments from insoluble paint vehicles. J Appl Chem 19: 93-99

Melo L F, Bott T R (1997) Biofouling in water systems. Exp. Thermal. Fluid Sci. 14: 375-381

Milczewska K, Voelkel A, Jeczalik J (2003) The use of Flory-Huggins parameters to characterization of polymer/filler interactions. Macromol Symp 194: 305-311.

Miller E S, Peppas N A (1983) Diffusional release of water-soluble bioactive agents from ethylene-vinyl acetate copolymers. Chem Eng Commun 22: 303-315

Miller E S, Peppas N A, Winslow D N (1983) Morphological changes of EVAc-based controlled delivery systems during release of water-soluble solutes. J Membr Sci 13

Molina-Torres J, Garcia-Chavez A, Ramirez-Chavez E (1999) Antimicrobial properties of alkamides present in flavouring plants traditionally used in Mesoamerica: affinin and capsaicin. J Ethnopharmacology 64: 241-248

Monaghan C P, Kulkarni V H, Good M L (1978) Release mechanisms of organotin toxicants from coating surfaces: a leaching model for antifouling coatings. ACS Symp Ser 82: 359-371

Omae I Kill S, Dam-Johansen K (2003) Organotin antifouling paints and their alternatives. Appl Organometal Chem 17: 81-105

Paul D R, S K McSpadden (1976) Diffusional release of a solute from a polymer matrix. J Membr Sci 1: 33-48

Peppas N A (1983) A model of dissolution-controlled solute release from porous drug delivery polymeric system. J Biomed Mat Res 17: 1079-1087

192 Perry D L, Philips S L (1995) Handbook of Inorganic Compounds. CRC Press, Boca Raton

Pitzer K S (1973) Thermodynamics of electrolytes. I. Theoretical basics and general equations. J Phys Chem 77: 268-277

Poling B E, Prausnitz J M, Connell J P (2001) The properties of gases and liquids. McGraw-Hill, New York, 11.43-11.48

Ponasik J, Conova S, Kinghorn D, Kinney W A, Rittschof D, Ganem B (1998) Pseudoceratidine, a marine natural product with antifouling activity: synthetic and biological studies. Tetrahedron 54: 6977-6986

Polishchuk A Y, Zaikov G E (1997) Multicomponent transport in polymer systems for controlled release. Overseas Publishers Association, Amsterdam, 34-42

Railkin A (2004) Marine biofouling: Colonization processes and defenses. CRC Press, Boca Raton, 194-225

Rice R G, Do D D (1994) Applied mathematics and modeling for chemical engineers. Wiley

Rhine W D, Hsieh D, Langer R (1980) Polymers for sustained macromolecules release: procedures to fabricate reproducible delivery systems and control release kinetics. J Pharm Sci 69: 265-270

Rodriguez F (1989) Principles of polymer systems. Hemisphere Publishing Corp, New York, 29

Ruthven D M (1984) Principles of adsorption and adsorption processes. Wiley- Interscience, New York.

Sagoo S K, Board R, Roller S (2002) Chitosan potentiates the antimicrobial action of sodium benzoate on spoilage yeasts. Letters Appl Microbiol 34: 168-172

Schaink H M, Smit J A (1996) Toward an integrated analytical description of demixing in ternary solutions of nonideal uncharged lattice polymers, hard particles, and solvent. Macromolecules 29: 1711–1720

Schoeff M S, Williams R H (1993) Principles of Laboratory Instruments. Mosby, St. Louis

Scher (1999) Controlled-Release Delivery Systems for Pesticides. 142 Singer I L, Kohl J G, Patterson M (2000). Mechanical aspects of silicone coatings for hard faulant control. Biofouling 16: 301: 309

193 Singh A, Freeman B D, Pinnau I (1998) Pure and mixed gas acetone/nitrogen permeation properties of polydimethylsiloxane [PDMS]. J Polym Sci B 36: 289–301

Shi H, Wang L (2004) Preliminary study on nontoxic capsaicin antifouling paint. Huagong Xinxing Cailiao 32: 54-56

Skelland A (1985) Diffusional mass transfer. Kriger Publishing Co, Malabar, 110-117

Stein et al. (2002) Structure-property relationships of silicone biofouling release coatings. GE Global Research, Technical Information Series

Steinberg P, Schneider R, Kjelleberg S (1997) Chemical defenses of seaweeds against microbial colonization. Biodegradation 8: 211-220

Stupak M E, Garcia M T, Perez M C (2003) Non-toxic alternative compounds for marine antifouling paints. International Biodeterioration & Biodegradation 52: 49–52

Sundberg D, Vasishtha N, Zimmerman R C, Smith C M (1997) Selection, design and delivery of environmentally benign antifouling agents. Nav Res Rev XLIX: 51-59

Tsuchiya H (2001) Biphasic membrane effects of capsaicin, an active component in Capsicum species. J Ethnopharmacology 75: 295-299

Vasishtha N, Sundberg D, Rittschof D (1995) Evaluation of release rates and control of biofouling using monolithic coatings containing an isothiazolone. Biofouling 9: 1-16

Veech R L (1997) Biodegradable noisogenic agents for control of invertebrate pests. United States Patent No. 5, 629, 045

Vetere V, Perez M, Garcia M, Deya M, Stupak M, del Amo B (1999) A non-toxic antifouling compound for marine paints. Surf Coat Int 12: 586-589

Venkataraman L N, Hajra J P (1999) Thermodynamic modeling of quaternary Al-Ga-In- As system. Mater Sci Eng B 60: 143-148

Watermann B, Berger H D, Sonnichsen H, Willemsen P (1997) Performance and effectiveness of non-stick coatings in seawater. Biofouling 11: 101-118

Watts J L (1995) Anti-fouling coating composition containing capsaicin. United States Patent No. 5, 397, 385

Weaver K M, Awde D B (1986) Rapid high performance liquid chromatographic method for the determination of very low capsaicin levels. J Chromatography 367: 438-442 van der Wel G K, Adan O C G (1999) Moisture in organic coatings - a review. Prog Org Coat 37: 1-14 194 Welty J R, Wicks C E, Wilson R E (1984) Fundamentals of Momentum, heat and mass transfer. John Wiley & Sons Inc, New York, 595-600

Wood T S, Marsh T G (1999) Biofouling of wastewater treatment plants by the freshwater bryozoan, Plumatella vaihiriae (Hastings, 1929). Water Res 33: 609-614

Willis M S, Tosun I, Choo W, Chase G G, Desai F (1991) A Dispersed Multiphase Theory and its Application to Filtration. in: Advances in Porous Media, Volume 1, M.Y. Corapcioglu ed., Elsevier, Amsterdam,

Wynne K J, Swain G W, Fox R B, Bullock S, Uilk J (2000) Two silicone nontoxic fouling release coatings: hydrosilation cured PDMS and CaCO3 filled, ethoxysiloxane cured RTV11. Biofouling 16: 277-288

Wypych G (1999) Handbook of fillers. Transcontinental Printing Inc, Toronto, 48

Xu Q-W, Barrios C A, Cutright T J, Zhang Newby B-m (2005) Assessment of antifouling effectiveness of two natural product antifoulants by attachment study with freshwater bacteria. Environ Sci & Pollut Res 12: 278-284

Xu Q-W, Barrios C A, Cutright T J, Zhang Newby B-m (2005) Evaluation of toxicity of capsaicin and zosteric acid and their potential applications as antifoulants. Environ Toxicology 5: 467-474

Xu Q-W (2004) Evaluating the potential of zosteric acid and capsaicin for use as natural product antifoulants. MS thesis, University of Akron

Yao B, Ran Q, Chen W (2004) Capsaicin auto-polishing antifouling coating and its preparation process. Chinese Patent No. 1,477, 166

Yebra D M, Kill S, Dam-Johansen K (2004) Antifouling technology - past, present and future steps towards efficient and environmentally friendly antifouling coatings. Prog Org Coatings 50: 75-104

Yilmaz L, McHugh A J (1986) Analysis of Nonsolvent-solvent-polymer phase diagrams and their relevance to membrane formation modeling. J Appl Polym Sci 31: 997-1018

Yu L, Zhang Z, Xu H (2004) A novel copolymer containing capsaicin moiety and its application in environmentally-friendly antifouling paint. Xiyou Jinshu Cailiao Yu Gongcheng 33: 79-82

Zaikov G E, Iordanskii A P, Markin V S (1988) Diffusion of electrolytes in polymers. Utrecht, The Netherlands, 206-209

195

APPENDICES

196

APPENDIX A

MATLAB FILE FOR SOLVING THE GENERAL MASS TRANSFER MODEL (EQUATION 4.19)

% program written by Abdulhadi AL-Juhni (2006) % file name: pores.m % this file is to calculate the concentration profiles of the compound % inside % the coating, considering the boundary condition of non-zero surface % concentration %************************************************************

The m.file: pores.m

% M=20; no. of nodes % ISOTHERMAL % with external film resistance % no dissulotion % this case is when the AF compound has zero solubility in the % matrix,and % Co << Cs(sol. of A in water) % therfore the mass transfer mechanism is by channelling/pore formation function yd=pores(t,y); %data******************************* bm = 5; % bm = kl/D; dimensionless no % %************************************************************

% call the m file: matrixab20.m (to get the coefficients of A and B matrix) matrixab20;

%matrix size [m,n]=size(y); yd=zeros(m,n);

%assign dummy variables; [h] denote the conc. of the AF compound at the 20 internal nodes; 197 h2=y(1); h3=y(2); h4=y(3); h5=y(4); h6=y(5); h7=y(6); h8=y(7); h9=y(8); h10=y(9); h11=y(10); h12=y(11); h13=y(12); h14=y(13); h15=y(14); h16=y(15); h17=y(16); h18=y(17); h19=y(18); h20=y(19); h21=y(20);

% boundary conditions for mass balance eq of component A (the key point: derive external nodes as function of internal nodes): hin = [ h2; h3; h4; h5; h6; h7; h8; h9; h10; h11; h12; h13; h14; h15; h16; h17; h18; h19; h20; h21]; % internal nodes a1= AX(1,1) - bm; a2= AX(22,1); b1= AX(1,22); b2= AX(22,22); phi1= - AX(1,2:21)*hin; phi2= - AX(22,2:21)*hin; xx = [a1, b1 a2, b2];

zz = [phi1 phi2];

hh = xx\zz; h1 = hh(1); h22 = hh(2);

h = [ h1; h2; h3; h4; h5; h6; h7; h8; h9; h10; h11; h12; h13; h14; h15; h16; h17; h18; h19; h20; h21; h22]; % mass fraction at all nodes

%ODE

%conc. of component A in the diffusion length

198 yd(1)= (BX(2,:)*h); yd(2)= (BX(3,:)*h); yd(3)= (BX(4,:)*h); yd(4)= (BX(5,:)*h); yd(5)= (BX(6,:)*h); yd(6)= (BX(7,:)*h); yd(7)= (BX(8,:)*h); yd(8)= (BX(9,:)*h); yd(9)= (BX(10,:)*h); yd(10)=(BX(11,:)*h); yd(11)= (BX(12,:)*h); yd(12)= (BX(13,:)*h); yd(13)= (BX(14,:)*h); yd(14)= (BX(15,:)*h); yd(15)= (BX(16,:)*h); yd(16)= (BX(17,:)*h); yd(17)= (BX(18,:)*h); yd(18)= (BX(19,:)*h); yd(19)= (BX(20,:)*h); yd(20)= (BX(21,:)*h);

%********************************************************************** %********************************************************************** **************************************

The main file to plot surface conc=f(time)

[T,Y] = ode23s('pores',[0 1],[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]);

199 matrixab20; m=size(Y); bm = 5; hin = Y; a1= AX(1,1) - bm; a2= AX(22,1); b1= AX(1,22); b2= AX(22,22); phi1= - AX(1,2:21)*hin'; phi2= - AX(22,2:21)*hin'; xx = [a1, b1 a2, b2];

zz = [phi1 phi2];

hh = xx\zz; cc = hh';

c5 = cc(:,1); t5 = T; plot (t5,c5)

200

APPENDIX B

SUGGESTION OF A MORE REALISTIC MASS TRANSFER MODEL – APPLICATION OF THE AVERAGE VOLUME THEORY

To overcome the limitations of the simplified model, we propose here a more realistic model based on the average-volume quantum theory. The major thrust of the new model is that it treats the coating system as multiphase system by applying the average volume quantum theory, where the necessary equations are written for each phase rigorously, thus eliminating the main concern of the old simplified model which treats the coating system as a “lumped” one phase. Therefore, the new model removes the uncertainty arises from applying the “lumped” tortuosity factor in the old simplified model.

The new model also treats the problem as a moving boundary problem. The new model also explicitly accounts for the resistance to mass transfer from the dissolution step and from the boundary layer. Also, the new model clearly includes the role of water diffusion in silicone membranes. In addition, the new model also considers the porosity to be time-dependent.

The formulation of the new model shown below follows the procedure outlined in

Chase (2002) and Willis et al. (1991), where details about the average-volume quantum 201 theory and its application can be found, and the nomenclature used here are kept the same

as in the mentioned references. In our new model, the coating system is considered to be heterogeneous and comprising of three distinct phases: the α-phase, the β-phase, and the

γ-phase, where they represent the polymer phase, the particle (i.e. solid NaB) phase, and

the porous phase (i.e. the water-filled pores and channels), respectively. Both the α-phase

and the γ-phase are considered to be multi-components systems whereas the β-phase is a

single component system. The following subscripts are used for the components: 1:

water, 2: NaB, and 3: polymer. Hence, the mass continuity equations for the three phases

are:

α α α ∂ (ε ρ ) / ∂t + Em = 0 (A.1a)

β β β ∂ (ε ρ ) / ∂t + Gm = 0 (A.1b)

γ γ γ γ ∂ (ε ρ ) / ∂t + Em + Gm = 0 (A.1c)

where εi is the volume fraction of phase i, ρi is the total concentration (in mass per

i volume of phase i), and Em is the total mass exchange between phase i and its adjacent

i phase. Gm is the total mass produced (or consumed) in phase i due to phase change. The

i β physical meaning for Gm is that Gm is the mass lost in the NaB solid particle phase due

to dissolution of the particles, which is simultaneously added to the porous phase as an

γ aqueous water solution mass, Gm . Similarly, the mass absorbed into the polymer phase

comes from the porous phase. Hence, the following relations are applied:

α γ Em = — Em (A.2a)

γ β Gm = — Gm (A.2b)

202 The chemical species balances for the α-phase are:

α α α α α ∂ (ε ρ1 ) / ∂t + ∂ (ε J1x ) / ∂x + E1 = 0 (A.3a)

α α α α α ∂ (ε ρ2 ) / ∂t + ∂ (ε J2x ) / ∂x + E2 = 0 (A.3b)

α α α α α ∂ (ε ρ3 ) / ∂t + ∂ (ε J3x ) / ∂x + E3 = 0 (A.3c)

α α where ρk is the concentration of component k (in mass per volume of phase α), Jkx is the

α mass flux – in axial direction - of component k in phase α, and Ek is the mass exchange

– of component k – between phase α and its adjacent phase. The following relation is

applied:

α α α Ei = Em ρi (A.4)

For the β-phase, the chemical species balance is not needed because the β-phase is

a single component system (only contains solid NaB). For the γ-phase, the chemical

species balances are:

γ γ γ γ γ ∂ (ε ρ1 ) / ∂t + ∂ (ε J1x ) / ∂x + E1 = 0 (A.5a)

γ γ γ γ γ γ ∂ (ε ρ2 ) / ∂t + ∂ (ε J2x ) / ∂x + E2 + G2 = 0 (A.5b)

γ γ γ where ρk is the concentration of component k (in mass per volume of phase ), Jkx is the

γ mass flux – in axial direction - of component k in phase γ, Ek is the mass exchange – of

γ component k – between phase γ and its adjacent phase, and G2 is the mass produced of

component 2 in phase γ due to phase change. The following relations are applied:

203 γ γ γ E1 = Em ρ1 (A.6a)

γ γ γ γ γ E2 + G2 = (Em + Gm ) ρ2 (A.6b)

The constitutive relations are:

α α α J1x = — D1 ∂ (ρ1 ) / ∂x (A.7a)

α α α J2x = — D2 ∂ (ρ2 ) / ∂x (A.7b)

α α α J3x = — D3 ∂ (ρ3 ) / ∂x (A.7c)

γ γ γ J1x = — D1 ∂ (ρ1 ) / ∂x (A.7d)

γ γ γ J2x = — D2 ∂ (ρ2 ) / ∂x (A.7e)

β β γ Gm = — kS (ρ — ρ2 ) (A.7f)

α γ α γ α Em = k1 (ρ1 — ρ1 ) + k2 (ρ2 — ρ2 ) (A.7g)

where here the constititve relations for the mass flux (equations A.7a through A.7e) are

β obtained by applying Fick’s second law, the constitutive relation for Gm (equation A.7f)

is expressed in terms of resistance to mass transfer due to dissolution of NaB particles,

α and the constitutive relation for Em (equation A.7g) is expressed in terms of resistance to

mass transfer due to absorption of the aqueous NaB water solution into the polymer

i phase. In the above equations, Dk is the diffusion coefficient of component k in phase i,

and kS is the dissolution coefficient of NaB in water. k1 and k2 are the mass exchange

coefficients of component 1 and 2, respectively, which are related to the mass exchange between the α-phase and the γ-phase. Hence, as shown in equation (A.7a), the inclusion of the diffusivity data of water through silicone membranes is explicitly considered in the new model.

204 The jump mass balance at the moving boundary (i.e. at x=h) is obtained by equating

the physical definition of the mass flux to the definition that is based on Fick’s first law.

Hence, this is expressed as:

γ γ γ β β γ γ — ε D2 ∂ (ρ2 ) / ∂x = (ε ρ — ε ρ2 ) (dh / dt) (A.8)

The following condition is always applied:

εα + εβ + εγ = 1 (4.9)

Equations (A.1 – A.9) describe the mass transport phenomena inside the matrix.

It is possible also to add to this set of equations an equation that describes the resistance

to mass transfer in the boundary layer region. The equation that describes the boundary

layer region, δ, is:

δ water 2 δ 2 (∂C2 / ∂t) = D2 (∂ C2 / ∂x ) (A.10)

The above new model explicitly incorporates the diffusivity of water in silicone

α membrane (D1 , which appears in equation A.7a). The diffusivity data of water in

silicone membrane can be obtained from literature. The literature value for the

diffusivity of water in silicone membrane was reported to be 8.6 x 10-7 cm2/s at 25o C

(Banerjee et. al., 1997). This temperature is the same temperature of our leaching

experiments, and therefore this reported diffusivity data can be safely used in the model.

205 Equations (A.1 – A.10) have to be solved simultaneously. Most likely analytical solution is challenging for this set of equations. Therefore, numerical solution is recommended here. Due to time limitation of the current study, it was not possible to develop the numerical solution code for the above set of equations, and it is recommended to be accomplished in future work.

206