An AFM Study of the Interactions Between Colloidal Particles

An AFM study of the interactions between colloidal particles

vorgelegt von Liset A. C. Lüderitz M. Sc. aus Havanna - Kuba

von der Fakultät II - Mathematik und Naturwissenschaften der Technischen Universität Berlin zur Erlangung des akademischen Grades

Doktor der Naturwissenschaften Dr. rer. nat.

genehmigte Dissertation

Promotionsausschuss: Vorsitzender: Prof. Dr. R. Schomäcker, TU Berlin Berichter: Prof. Dr. R. von Klitzing, TU Berlin Berichter: Prof. Dr. G. Papastavrou, Universität Bayreuth Tag der wissenschaftlichen Aussprache: 14.09.2012

Berlin 2012 D 83 Abstract

This research project is focused on the study of the interaction forces between two colloidal particles. The interaction between colloidal particles may differ from the interaction between macroscopic bodies. The interactions are measured across different electrolytes: CsCl, KCl, NaCl and LiCl using Colloidal Probe Atomic Force Microscope (CP-AFM) techniques. The resulting forces may be different depending on the electrolyte solution used, which is known as ion specificity. In this study no ion specificity effect is observed at long range for the adsorption of counterions to the silica surface but a slight tendency for Cs+ to be more adsorbed at the surface than the other counterions is present at 10−3 M. Deviations from the DLVO theory at small separations (non-DLVO forces) are reported in this work. Short range attractions at 10−4 M ionic strength were measured whereas at 10−3 M short range repulsions are present. An explanation of the non-DLVO forces is given based on the hydration forces. A further chapter studies the interaction forces between silicon oxide surfaces in the presence of surfactant solutions. Based on the qualitative and quantitative analysis of these interaction forces the correlation with the structure of the aggregates on the surfaces is analysed. A colloidal probe atomic force microscope (AFM) was used to measure the forces between two colloidal silica particles and between a colloidal particle and a silicon wafer in the presence of hexadecyltrimethylammonium bromide (CTAB) at concentrations between 0.005 mM and 1.2 mM. Different interaction forces were obtained for the silica particle–silica particle system when compared to those for the silica particle–silicon wafer system for the same studied concentration. This indicates that the silica particles and the silicon wafer have different aggregate morphologies on their surfaces. The point of zero charge (pzc) was obtained at 0.05 mM CTAB concentration for the silica particles and at 0.3 mM for the silica particle– silicon wafer system. This indicates a higher charge at the silicon wafer than at the silica particles. The observed long range attractions are explained by nanobubbles present at the silicon oxide surfaces and/or by attractive electrostatic interactions between the surfaces, induced by oppositely charged patches at the opposing Si oxide surfaces. In order to analyze the role of the nanobubbles on the hydrophobic interactions hydrophilic silicon wafers were studied against aqueous solutions of CTAB at concentrations between 0.05 mM and 1 mM (CMC). AFM studies show that nanobubbles are formed at concentrations up to 0.4 mM. From 0.5 mM upward, no bubbles are detected. This is interpreted as the formation of hydrophobic domains of surfactant aggregates, becoming hydrophilic at about 0.5 mM. The high contact angle of the nanobubbles (140-150◦ through water) in comparison to the macroscopic contact angle indicates that the nanobubbles are located on the surfactant domains. A combined imaging and colloidal probe AFM study serves to highlight the surfactant patches adsorbed at the surface via nanobubbles. The nanobubbles have a diameter between 30 and 60 nm (after tip deconvolution), depending on the surfactant concentration. This corresponds to a Laplace pressure of about 30 atm. The presence of the nanobubbles is correlated with force measurements between a silica probe and a silicon wafer surface. The study is a contribution to a better understanding of the short range attraction between hydrophilic surfaces exposed to a surfactant solution. The substrate properties hydrophobicity and roughness influence the morphology and size of the nanobubbles. Nanobubbles with a contact angle through water of 132◦ and a Laplace pressure of 18 atm were visualized at the interface of a hydrophobically modified silicon wafer exposed to water and surfactant solutions. An increase in surfactant concentration has an impact on the morphology of the nanobubbles, they were flattened at the surface with increasing surfactant concentration.

3 To my father and my grandparents Contents

List of Figures8

List of Tables 13

List of Symbols 14

Acknowledgments 16

1. Introduction and Literature Review 17 1.1. Colloidal Particles...... 17 1.2. Non-DLVO Forces...... 23 1.2.1. Hydration Forces...... 23 1.2.2. Hydrophobic Interactions...... 26 1.2.3. Structural Forces...... 27 1.3. Surfactants...... 28 1.3.1. Classiﬁcation...... 28 1.3.2. Surfactants at Interfaces...... 31 1.4. Nanobubbles...... 32

Bibliography 34

2. Techniques 36 2.1. Atomic Force Microscopy...... 36 2.2. Scanning Electron Microscopy...... 40 2.3. Zeta Potential...... 41

Bibliography 45

3. Force Measurements between Colloidal Particles across Aqueous Electrolytes 46 3.1. Introduction...... 46 3.2. Experimental Section...... 47 3.2.1. Materials...... 47 3.2.2. Preparation and Methods...... 48 3.2.3. Simulations...... 48 3.3. Results...... 49 3.3.1. Eﬀect of Ionic Strength: 10−4 M and 10−3 M...... 49 3.3.2. Eﬀect of pH...... 55

5 Contents

3.3.3. Interactions through Water...... 56 3.4. Discussion...... 58 3.4.1. Eﬀect of Ionic Strength...... 58 3.4.2. Eﬀect of pH...... 62 3.5. Conclusions...... 63

Bibliography 64

4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions 66 4.1. Introduction...... 66 4.2. Experimental Section...... 68 4.2.1. Materials...... 68 4.2.2. Preparation and Methods...... 68 4.2.3. Simulations...... 69 4.3. Results...... 69 4.3.1. Interaction forces between two silica particles (system I)...... 69 4.3.2. Interaction forces between a silica particle and a silicon wafer (system II)...... 71 4.3.3. Point of zero charge...... 74 4.4. Discussion...... 76 4.4.1. Interaction between two silica particles (system I)...... 76 4.4.2. Interaction between a silica particle and a silicon wafer (system II) 78 4.4.3. Comparison between the system silica particle–silica particle (I) and the system silica particle–silicon wafer (II)...... 80 4.4.4. Non DLVO forces...... 81 4.5. Conclusions...... 82

Bibliography 84

5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC 86 5.1. Introduction...... 86 5.2. Experimental Section...... 88 5.2.1. Materials...... 88 5.2.2. Preparation and Methods...... 88 5.2.3. Simulations...... 89 5.3. Results...... 89 5.4. Discussion...... 95 5.4.1. Nanobubbles...... 95 5.4.2. Correlation with Force Curves...... 98 5.5. Conclusions...... 101

Bibliography 102

6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer 104

6 Contents

6.1. Introduction...... 104 6.2. Experimental Section...... 105 6.2.1. Materials...... 105 6.2.2. Preparation and Methods...... 105 6.2.3. Simulations...... 106 6.3. Results...... 106 6.4. Discussion...... 111 6.5. Conclusions...... 114

Bibliography 115

7. Conclusions and Future Work 117 7.1. General Conclusions...... 117 7.2. Future Work...... 119

Bibliography 121

A. Appendix 122

7 List of Figures

1.1. Scheme of the DLVO theory: (a) Surfaces repel strongly, small colloidal particles remain stable; (b) Surfaces are at equilibrium at secondary minimum if it is deep enough, colloids remain kinetically stable; (c) Surfaces come into secondary minimum, colloids coagulate slowly; (d) The critical coagulation concentration, surfaces may remain in secondary minimum or adhere, colloids coagulate rapidly; (e) Surfaces and colloids coalesce rapidly [6]...... 19 1.2. Helmholtz double layer model ...... 20 1.3. Gouy-Chapmann double layer model ...... 20 1.4. Stern model of the double layer ...... 21 1.5. Forces measured between mica surfaces in LiCl solutions at pH 5.4 [8] ... 24 1.6. Force measured between mica surfaces in 1.4×10−3M NaCl solution at pH 5.7. The full line corresponds to the charge regulation model, the dashed line is the constant potential ψ = 138mV boundary condition [8]...... 25 1.7. Force measured between silica surfaces in 5×10−6M CPC and 0.1M NaCl. The interaction was measured in gassed (filled circles) and degassed (open circles) solutions. The gassed solution was measured prior to the degassed solution (A). The measured order was reversed (B) [32]...... 27 1.8. Force measured in SDS micellar solution and microemulsion confined between two drops of perfluorooctane. The oil in water microemulsion consists of: 2 wt% oil phase (tetradecane), 5.5 wt% surfactant(SDS), 5.5 wt% cosurfactant(pentanol) in water [36]...... 28 1.9. Aggregates formed by surfactants ...... 30 1.10. Models for the two step and the four region model [38] ...... 32 1.11. (a) Normalized Raman integrated intensities as a function of CTAB bulk concentration (b) Adsorption isotherm obtained after subtraction of the bulk contribution and conversion of the Raman integrated intensities into adsorbed amounts [38] ...... 32 1.12. AFM image of bubbles on mica surface in water in tapping mode, with normal contact cantilever of spring constant equal to 0.38 N/m. Image size 1 × 1 µm [43] ...... 33

2.1. A representation of the MFP-3D used during experiments [5] ...... 37 2.2. The raw data for InvOLS determination. The y axes represents the deﬂec- tion of the cantilever in volts. The x axes (Zsnsr) is the piezo position. .. 38 2.3. Representation of a force measurement between two particles ...... 40

8 List of Figures

2.4. Instrumentation of a scanning electron microscope [16] ...... 41 2.5. Double layer of a particle ...... 42 2.6. Scheme of the Laser Doppler Velocimetry (LDV) [17] ...... 43

3.1. Forces between a pair of colloidal silica particles across different aqueous electrolyte solutions; LiCl, NaCl, KCl and CsCl, at a fixed electrolyte concentration of 10−4 M and pH=5.8; Hamaker constant A= 8.5 × 10−21 J. The continuous lines correspond to constant charge and the discontinuous ones to constant potential...... 49 3.2. Forces between a pair of colloidal silica particles across different aqueous electrolyte solutions; LiCl, NaCl, KCl and CsCl, at a fixed electrolyte concentration of 10−3 M and pH=5.8; Hamaker constant A= 8.5 × 10−21 J. The continuous lines correspond to constant charge and the discontinuous ones to constant potential...... 51 3.3. Monte Carlo simulation with an explicit surface charge description of the experimental data at 1 mM ionic strength and pH=5.8. The calculations were performed by Christophe Labbez...... 53 3.4. Monte Carlo simulation with an implicit surface charge description of the experimental data at 1 mM ionic strength and pH=5.8. The calculations were performed by Christophe Labbez...... 54 3.5. Forces between a pair of colloidal silica particles across different aqueous electrolyte solutions; LiCl, NaCl, KCl and CsCl, at a fixed electrolyte concentration of 10−4 M and pH=4. Hamaker constant A= 8.5 × 10−21 J. The continuous lines correspond to constant charge and the discontinuous ones to constant potential...... 55 3.6. Monte Carlo simulation with an explicit surface charge description for the interaction curve between a pair of colloidal particles across NaCl aqueous electrolyte solution at a fixed ionic strength of 1 mM and pH=4. The calculations were performed by Christophe Labbez...... 56 3.7. Monte Carlo simulation with an implicit surface charge description for the interaction curve between a pair of colloidal particles across NaCl aqueous electrolyte solution at a fixed ionic strength of 1 mM and pH=4. The calculations were performed by Christophe Labbez...... 57 3.8. Forces between a pair of colloidal silica particles in milli-Q water at pH=4. Hamaker constant A= 8.5 × 10−21 J ...... 58 3.9. Sketch of the adsorption of the cation lithium at a) 10−4 M before and after approaching and b) 10−3 M ionic strength before and after approaching 61

4.1. Forces between a pair of colloidal silica particles (system I) across CTAB surfactant solution, from 0 to 0.1 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J. DLVO_CC (constant charge) and DLVO_CP (constant potential) ﬁts are shown for 0.1 mM surfactant concentration .. 70

9 List of Figures

4.2. Forces between a pair of colloidal silica particles (system I) across CTAB surfactant solutions from 0.1 to 0.5 mM surfactant concentration. Hama- ker constant A= 8.5×10−21 J. Constant charge and constant potential fits are shown for 0.2 mM (DLVO_CC), 0.3 mM (DLVO_CC - overlaps fit at 0.2 mM), 0.4 (DLVO_CP) and 0.5 mM (DLVO_CC and DLVO_CP) surfactant concentration ...... 72 4.3. Forces between a pair of colloidal silica particles (system I) across CTAB surfactant solutions from 0.5 mM to 1.2 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J. Constant charge and constant potential fits are shown for 0.8 and 1 mM surfactant concentration...... 73 4.4. Forces between a colloidal silica particle and a silicon wafer (system II) across CTAB surfactant solutions at 0.005 and 0.05 mM surfactant concentration. Forces between two colloidal silica particles (system I) at 0.05 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J. .... 74 4.5. Forces between a colloidal silica particle and a silicon wafer (system II) across CTAB surfactant solutions from 0.3 to 0.8 mM surfactant concentration. Hamaker constant A= 8.5×10−21 J. DLVO_CC (constant charge) and DLVO_CP (constant potential) fits are shown for 0.4 mM surfactant concentration. For the DLVO_CC fits shown at 0.5 mM and 0.8 mM the plane of charge was set 4 nm away from each surface...... 75 4.6. Forces between a colloidal silica particle and a silicon wafer (system II) across 1 mM CTAB surfactant solution. Hamaker constant A= 8.5×10−21 J. The experimental curve was offset 8 nm under the assumption that micelles/patchy bilayers are adsorbed at the surface. DLVO_CC (constant charge) and DLVO_CP (constant potential) fits are shown for the experimental curve and the shifted curve. Data taken from chapter5...... 76 4.7. Possible surfactant morphologies depending on the concentration...... 79 4.8. Interaction forces between two silica particles (system I) and between a particle and a silicon wafer (system II) at 0.4 mM surfactant concentration. The Debye length is 15.2 nm for both cases...... 80 4.9. AFM tapping mode of a silicon wafer at 0.3 mM surfactant concentration. 82

5.1. AFM tapping mode of a silicon oxide surface in water. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.265 V, setpoint ratio: 0.26, scan rate: 0.5 Hz, drive frequency: 6.91 KHz ...... 89 5.2. AFM tapping mode of a silicon oxide surface at 0.05 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.430 V, setpoint ratio: 0.43, scan rate: 0.5 Hz, drive frequency: 6.34 KHz ...... 91 5.3. AFM tapping mode of a silicon oxide surface at 0.3 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.190 V, setpoint ratio: 0.19, scan rate: 0.5 Hz, drive frequency: 6.16 KHz ...... 92

10 List of Figures

5.4. AFM tapping mode of a silicon oxide surface at 0.4 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.215 V, setpoint ratio: 0.22, scan rate: 0.5 Hz, drive frequency: 6.03 KHz ...... 93 5.5. AFM tapping mode of a silicon oxide surface at 0.5 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.234 V, setpoint ratio: 0.23, scan rate: 0.5 Hz, drive frequency: 7.1 KHz ...... 94 5.6. AFM tapping mode of a silicon oxide surface at 0.8 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.210 V, setpoint ratio: 0.21, scan rate: 0.5 Hz, drive frequency: 7.1 KHz ...... 95 5.7. Schematic picture of nanobubbles (not to scale) seated on the hydrophobic tails of the surfactant molecules ...... 96 5.8. Force curves between a silica particle and a silicon oxide surface in the presence of 0.4 and 0.5 mM CTAB concentration ...... 99 5.9. Force curves between a silica particle and a silicon oxide surface in the presence of 1mM CTAB concentration ...... 100

6.1. AFM tapping mode of a hydrophobically modified silicon wafer in water. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.292 V, setpoint ratio: 0.29, scan rate: 0.5 Hz, drive frequency: 6.27 KHz ...... 107 6.2. AFM tapping mode of a hydrophobically modified silicon wafer in 0.1 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 5.1 KHz ...... 108 6.3. AFM tapping mode of a hydrophobically modified silicon wafer in 0.3 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 5.6 KHz ...... 108 6.4. AFM tapping mode of a hydrophobically modified silicon wafer in 0.4 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 5.6 KHz ...... 109 6.5. AFM tapping mode of a hydrophobically modified silicon wafer in 0.8 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 4.8 KHz ...... 109 6.6. AFM tapping mode of a hydrophobically modified silicon wafer in 1.2 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 4.9 KHz ...... 110

11 List of Figures

6.7. AFM tapping mode of a hydrophobically modified silicon wafer in 1.2 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 4.9 KHz ...... 110 6.8. Forces between a pair of hydrophobically modified silica particles in water and from 0.03 to 1.2 mM surfactant concentration at pH=5.8, Hamaker constant A= 8.5 × 10−21 J. The continuous lines correspond to constant charge. DLVO_CC (constant charge) fits are shown for 0.03 mM, 0.3 mM, 0.4 mM and 1 mM surfactant concentration ...... 111 6.9. Interaction force between a tip and a bubble in 0.01 mM CTAB concentration113

7.1. Scanning electron microscopy of a modiﬁed cantilever ...... 120 7.2. Scanning electron microscopy of a modiﬁed cantilever ...... 120

A.1. Scanning electron microscopy of silica particles ...... 122 A.2. Scanning electron microscopy of a magnetic actuated cantilever ...... 122 A.3. Scanning electron microscopy of a magnetic actuated cantilever ...... 123 A.4. Schematic cross section of a nanobubble ...... 123 A.5. Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.05 mM CTAB concentration (see figure 5.2) ...... 124 A.6. Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.3 mM CTAB concentration (see figure 5.3) ...... 125 A.7. Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.4 mM CTAB concentration (see figure 5.4) ...... 126 A.8. Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.5 mM CTAB concentration (see figure 5.5) ...... 127 A.9. Amplitude-distance data of nanobubbles on a modified silicon wafer immersed in water (see figure 6.1) ...... 128 A.10.Amplitude-distance data of nanobubbles on a modified silicon wafer at 0.1 mM CTAB concentration (see figure 6.2) ...... 129 A.11.Amplitude-distance data of nanobubbles on a modified silicon wafer at 0.3 mM CTAB concentration (see figure 6.3) ...... 130 A.12.Amplitude-distance data of nanobubbles on a modified silicon wafer at 0.4 mM CTAB concentration (see figure 6.4) ...... 131 A.13.Amplitude-distance data of nanobubbles on a modified silicon wafer at 0.8 mM CTAB concentration (see figure 6.5) ...... 132 A.14.Amplitude-distance data of nanobubbles on a modified silicon wafer at 1.2 mM CTAB concentration (see figure 6.6) ...... 133

12 List of Tables

1.1. Geometrical relations of diﬀerent aggregates; V, A, gmax and g refer to the complete spherical aggregate, unit length of a cylinder or unit area of a bilayer [39]...... 31

3.1. Results of simulations of direct force measurements by the Poisson-Boltz- mann theory...... 52 3.2. Size of the counterions, taken from [3]...... 62

4.1. Results for simulations of direct force measurements by the Poisson-Boltz- mann theory. ∗ From ﬁtting with the plane of origin of charge taken at 4 nm from each surface. ∗∗ Assuming that the surfaces were not in contact (micelles/bilayer adsorption)...... 71

5.1. Parameters of the nanobubbles obtained by ﬁtting the cross section to an arc of a circle. The small size of the nanobubbles at 0.4 mM CTAB concentration introduces more errors in the ﬁtting and in the obtained parameters, but still the parameters of the nanobubbles are shown for comparison...... 97 5.2. Parameters of the nanobubbles obtained after tip deconvolution, Rtip =15 nm±5 nm...... 97

13 List of symbols

1. A: Hamaker constant 2. a: Eﬀective area per head group

3. A2: Second virial coeﬃcient

4. c0: Concentration

5. ci: Local ion density 6. D: Distance between the surfaces 7. E: Elastic modulus 8. F (D): Interaction force 9. f(ka): Henrys function

10. FH : Hydration force 11. g: Aggregation number 12. K: Optical constant 13. k: Cantilever spring constant

14. kB: Boltzmann constant 15. l: Cantilever length

16. lo: Length of the hydrocarbon chains 17. M: Molecular weight

18. Ns: Packing parameter 19. P (θ): Angular dependence of the sample scattering intensity

20. Rc: Curvature radius

21. Reff : Eﬀective radius

22. Rθ: Ratio scattered light to incident light of the sample 23. T : Temperature 24. t, h: Cantilever thickness

14 List of symbols

25. UE: Electrophoretic mobility

26. vo: Volume of the hydrophobic part 27. W : Work 28. W (D): Interaction free energy 29. W (D)str: Interaction energy of the structural force 30. w, b: Cantilever width

31. Wa(D): Van der Waals interaction free energy Hphb 32. Wa (D): Hydrophobic interaction energy 33. z: Zeta potential

34. zc: Cantilever deﬂection 35. ∆P : Laplace pressure 36. ∆V : Voltage measured by photodiode 37. : Dielectric constant

38. 0: Permittivity in vacuum 39. η: Viscosity of the medium 40. θ: Contact angle

41. λD: Debye length 42. ν: Resonant frequency 43. ρe: Local electric charge density 44. σ: Surface tension

45. σp: The oscillatory period 46. ψ(x): Potential at a certain distance from the surface

47. ψ0: Surface potential 48. ω: Angular frequency

15 Acknowledgments

The author wishes to thank the DFG via SPP 1273 "Kolloidverfahrenstechnik (KL- 1165-11)" for financial support. Thanks to Ulrich Gernert for the scanning electron microscopy analysis of the spheres and the cantilevers. Prof. Schäfer and Mattias Böcker are acknowledged for the donation of micropipettes to glue particles to the glass slides. I would also like to thank the Stranski laboratory staff, workshop staff and staff secretary for all the help over the years. My special thanks to my colleagues Yan Zeng and Cagri Üzum who introduced me to the AFM measurements and provided valuable suggestions. I would like to acknowledge Prof. Dr. Georg Papastavrou for helpful suggestions in the first experiments with particles. Prof. Dr. Gerhard Findenegg and Prof. Dr. Vincent Craig are acknowledged for helpful discussions. Dr. Christophe Labbez is acknowledged for the debates and simulations of the experiments with silica particles. My thanks to the members of the group of Prof. Dr. Regine v. Klitzing for the helpful discussions and the pleasant moments. I would like to thank my colleagues Bhuvnesh Bharti, Heiko Fauser and Adrian Carl for the pleasent atmosphere in the office. A special appreciation goes to Prof. Dr. Regine v. Klitzing for all the help and fruitful discussions. My thanks also to Prof. Dr. R. Schomäcker for his support in the final stage of my PhD. I would like to thank my husband, Stephan, for his company and love, for giving me the feeling that this is possible and for the care for the family during these years. My thanks to my parents-in-law who helped with child care during my long hours in the lab. Thanks to all my friends who cheered me up when it was necessary and for helping with child care. Also thanks to my sweet children Laetitia and Leonardo for making me forget my worries. I would like to thank my brother Reynel for all the emotional support and above all I would like to thank a wonderful woman, my mother Gladys, for making me who I am today. 1. Introduction and Literature Review

1.1. Colloidal Particles

The deﬁnition of a colloid was given by Thomas Graham in 1861 and was based on the size of the material. A colloid is a material with at least one dimension in the nanometer scale, from 1 to 1000 nm [1]. When colloids are dispersed in a continuous medium, colloidal dispersions are formed. Several types of colloidal dispersions are found in daily life [2].

1. Colloidal sols: small solid particles are dispersed in a liquid (e.g. paint, ink, muddy water) 2. Emulsions: small droplets dispersed in a liquid medium (e.g. drug delivery) 3. Foam: a gas is dispersed in a liquid medium (e.g. vacuoles, ﬁre extinguishers) 4. Aerosol: small solid or liquid particles dispersed in a gas medium (e.g. volcanic smoke, clouds, hair spray) 5. Solid suspension: Solid particles dispersed in a solid medium (e.g. wood) 6. Porous materials: liquid particles dispersed in a solid medium (e.g. oil reservoirs) 7. Solid foam: a gas dispersed in a solid medium (e.g. zeolites)

Colloidal pigments were used to write records of Egyptian Pharaohs [2]. Faraday prepared colloidal sols based on gold, which can still be seen in the British Museum in Lon- don [2]. There is an increased interest in the understanding of the interactions within colloidal dispersions. Many of the biological molecules are in the colloidal range. These molecules perform complex reactions inside our body [2], furthermore technical processes, like mineral ﬂotation, include colloidal particles. The stability of butter, milk, and other emulsions are of great importance in the food industry. Interfacial eﬀects dominate the colloidal systems [1]. In 1945 Derjaguin-Landau and Verwey-Overbeek developed the DLVO theory. The DLVO theory is based on some assumptions [3]:

1. Inﬁnite ﬂat solid surface 2. Uniform surface charge density 3. Constant surface electric potential

17 1. Introduction and Literature Review

4. The concentration proﬁle of both counter ions and surface charge determining ions is constant 5. No chemical reaction between the particles and the solvent

The DLVO theory only takes two kinds of forces into account to explain the stability of colloids in a suspension: van der Waals and electrostatic double layer forces. The van der Waals forces are the sum of the interactions between atomic and molecular dipoles in the particles or between different particles [4]. The van der Waals interaction free energy between two flat surfaces can be defined as [4]:

A 1 W (D) = − (1.1) a 12π D2

(Wa(D): The van der Waals interaction free energy between two ﬂat surfaces; A: Hamaker constant; D: Distance between the surfaces). The interaction energy between to ﬂat surfaces can be related to the forces between two curved surfaces of radius R using the Derjaguin approximation [2,4]

F (D) = 2πReff W (D) (1.2)

1 1 1 = + (1.3) Reff R1 R2

(F(D): Interaction force; Reff : Eﬀective radius; R1 and R2: Curvature radius of the sphere 1 and sphere 2 respectively; W(D): Interaction free energy) then

AR 1 F (D) = − (1.4) 12 D2 The Hamaker constant A depends on the polarisability, permanent dipole moment and ionization energy of the interacting molecules. The values are in the order of (0.4 − 40) × 10−20 J[4] and can be negative or positive. For two colloidal particles of equal sign in a medium the van der Waals interactions are always attractive; these systems will have a positive Hamaker constant. While between diﬀerent bodies in a medium it can be attractive or repulsive (A negative). The Hamaker constant will be negative if the Hamaker constant of the medium are intermediate between those of the two interacting particles [4–6]. The other force described by the DLVO theory is the electrostatic double layer force, which arises due to the overlap of the double layers. When a solid surface is placed in a polar medium, charges will develop at the surface due to the dissociation of surface groups present at the surface. The surface charges produce an electric ﬁeld which will

18 1. Introduction and Literature Review attract the counterions. The layer of surface charge and counterions is called double layer [1]. Figure 1.1 is a schematic representation of the DLVO theory [6]. The interactions depend on the electrolyte concentration and surface charge density. A strong long range repulsion is obtained for highly charged surfaces in dilute electrolyte (case a). At higher electrolyte concentrations, also a secondary minimum is present, normally of about 3 nm (see inset ﬁgure 1.1).

Figure 1.1.: Scheme of the DLVO theory: (a) Surfaces repel strongly, small colloidal particles remain stable; (b) Surfaces are at equilibrium at secondary minimum if it is deep enough, colloids remain kinetically stable; (c) Surfaces come into secondary minimum, colloids coagulate slowly; (d) The critical coagulation concentration, surfaces may remain in secondary minimum or adhere, colloids coagulate rapidly; (e) Surfaces and colloids coalesce rapidly [6].

For the colloidal particles the thermodynamic equilibrium state may be with the particles in contact in the primary minimum, but the energy barrier may be too high for the particles to overcome during a reasonable time period. When this happens, the particles will either come to the weaker secondary minimum or remain totally dispersed in the solution (kinetically stabilized colloid). For low surface charge surfaces the energy barrier is low (case c) and slow aggregation (coagulation or ﬂocculation) occurs. At concentrations

19 1. Introduction and Literature Review greater than the critical coagulation concentration, the particles will coagulate quickly and the colloid will be unstable (case d). At zero surface charge only van der Waals attraction will be present in the system (case e) [6]. Several models have been proposed to explain the electric double layer. Helmholtz proposed a model in 1879, which consists of a monolayer of ﬁxed charges able to neutralize the charges present at the surface [7] (see ﬁgure 1.2). σ

ψ 0

ψ=0

Figure 1.2.: Helmholtz double layer model

ψ 0

σ Potential ψ s (x)

ψ=φ=0

Figure 1.3.: Gouy-Chapmann double layer model

Around 1910 Gouy and Chapman developed a model with a diﬀuse layer of counterions. A high concentration of counterions at the surface is found under equilibrium conditions.

20 1. Introduction and Literature Review

The counterions concentration decreases moving away from the surface. The potential in the solution ψ(x) is also varying from the surface value ψ0(x = 0) to zero far away from the surface [7] (see figure 1.3). In 1924 Stern improved the Gouy-Chapman model of the double layer with the addition of an inner layer. In this new model, the double layer consists of an inner and an outer layer. The inner layer is a monolayer of counterions, but in contrast to the Helmholtz layer, this inner layer does not produce neutralization of the surface charge. The number of ions in this layer is given by the Langmuir adsorption isotherm. The counterions are not considered as point charges anymore, their proximity to the surface and to each other depends on the hydrated radii. The ion specificity effect is also considered in this model [7] (see figure 1.4).

ψ0

Stern layer ψ δ ψ (x)

δ δ x

Figure 1.4.: Stern model of the double layer

The Poisson-Boltzmann theory makes several assumptions for the description of the electric double layer [1,6]:

• The ions are considered point charges • The ions in solutions have a continuous charged distribution, the surface charge is considered homogeneous and smeared out • The non-coulombic interactions are not taken into account • The solvent is taken as a continuous medium and the permittivity is assumed to be constant • The surfaces are taken as smooth in the molecular scale • Image forces between the ions and the surfaces are not considered

The Poisson Boltzmann theory allows the determination of the electric potential ψ near a planar surface [1]. The potential and the distribution of ions varies with the distance normal to the surface x. The charge density and the electric potential are related by the Poisson equation [1,2] in the following way:

21 1. Introduction and Literature Review

2 2 2 2 ∂ ψ ∂ ψ ∂ ψ ρe ∇ ψ = 2 + 2 + 2 = − (1.5) ∂x ∂y ∂z 0

3 (ρe: local electric charge density in C/m , : dielectric permittivity, 0: permittivity in vacuum, = 1). The local ion density can be calculated by the Boltzmann equation as follows:

0 −Wi/kB T ci = ci e (1.6)

(Wi: work required to bring an ion from a distance far away from the surface to a distance closer to the surface). Assuming that only electric work is done and that only 1:1 electrolyte is present in the system, we can deﬁne W + and W − as the electric work required to bring a cation and anion respectively to a place with a potential ψ [1]

W + = eψ (1.7) for a cation and

W − = −eψ (1.8) for an anion Using the Boltzmann equation (equation 1.6) we can rewrite the local cation and anion density as follows:

+ −eψ/kB T c = c0e (1.9) for a cation and

− eψ/kB T c = c0e (1.10)

(c0: bulk salt concentration) for an anion. The local charge density ρe can be expressed as:

−eψ(x,y,z) eψ(x,y,z) + − k T k T ρe = e(c − c ) = c0e(e B − e B ) (1.11)

The charge density can be substituted in the Poisson equation 1.5

22 1. Introduction and Literature Review

eψ(x,y,z) −eψ(x,y,z) 2 c0e ∇ ψ = (e kB T − e kB T ) (1.12) 0

ﬁnally the Poisson-Boltzmann equation (see equation 1.12) is obtained. This equation is solved numerically though for simple geometries like a planar surface it can be solved analytically [1,2]. For a planar surface of low potential eψ << kBT at room temperature ψ ≤ kBT the linearized Poisson-Boltzmann equation can be used [1,2]

−κx κx ψ(x) = C1e + C2e (1.13) where

s 2c e2 κ = 0 (1.14) 0kBT

C1 and C2 are constant deﬁned by the boundary conditions. At the surface, the potential is equal to the surface potential ψ(x = 0) = ψ0 and at distances far away from the potential tends to zero ψ(x → ∞) = 0. Now the potential can be expressed as

−κx ψ = ψ0e (1.15)

−1 ◦ The Debye length is given by λD = κ . Using the water parameters at 25 C, the Debye length for a monovalent salt can be expressed as [1]:

3.04A˚ λD = √ (1.16) c0

mol (c0: concentration in L ).

1.2. Non-DLVO Forces

1.2.1. Hydration Forces

Figure 1.5 shows the interaction between mica surfaces in a LiCl solution at pH 5.4 [8]. Pashley’s [8] experimental data show that for solutions up to 10−2M of LiCl, the mica surfaces always come into the primary minimum, but at a certain concentration (6 × 10−2M) hydration forces are present in the system. For mica surfaces immersed in NaCl solution, the hydration force appears at a similar concentration 10−2M, whereas for mica surfaces in solutions of KCl and CsCl, the hydration forces are already present at 10−4M and 4 × 10−5M respectively. Between hydrophilic surfaces the hydration force

23 1. Introduction and Literature Review

Figure 1.5.: Forces measured between mica surfaces in LiCl solutions at pH 5.4 [8] is exponentially repulsive. The hydration forces are short range forces which avoid that surfaces come into contact due to van der Waals attractions [6]. The hydration forces observed in mica were related to the exchange of ions present in solution like Li+, Na+, Cs+,H+ with the K+ ions at the mica surface. The concentration at which hydration forces appeared at the mica surface was speciﬁc for each ion. The larger the hydration shell of the cation the higher the concentration needed to replace the H+ ions from the mica surface. The interaction of mica surfaces in water are DLVO like, no hydration forces were observed [8]. The mechanism behind the hydration forces was related to the adsorbed layer of counterions. To produce hydration forces, the counterions have to be bound to the mica surface in a speciﬁc way and should not to be desorbed upon the approach of the other surface [8]. Pashley [8] also obtained an interesting phenomenon working with mica surfaces at concentrations where the hydration force is still not present. The jump into contact appears at a separation larger than that predicted by the DLVO theory. The jump-in distance increases with increasing concentration of hydrated counterions close to the mica surface. The explanation given was the overlap of the counterions in the compressed double layer and on the surfaces, which expels the cations and replaces them by H+ ions.

24 1. Introduction and Literature Review

This causes a reduction in the surface potential and increases the jump-in distance. The ﬁnite ion size was also given as explanation of the reduced repulsion observed (see ﬁgure 1.6).

Figure 1.6.: Force measured between mica surfaces in 1.4×10−3M NaCl solution at pH 5.7. The full line corresponds to the charge regulation model, the dashed line is the constant potential ψ = 138mV boundary condition [8].

Hydration forces have also been observed in other materials. Evidence for hydration at the silica surface have been given by several authors [9–12], but the origin of the hydration forces are still controversial. Traditionally, it was accepted that the hydration forces at the silica surface were associated with the presence of a structured layer of water molecules. The overlap of the hydrated layer upon the approach of the surfaces will produce the short range repulsive hydration force [13]. A more recent interpretation was given by Vigil et al. [14]. They reported that the short range repulsion is steric in origin due to the overlap of polysilicic acid chains or silica gel layers present at the silica surface in aqueous medium. The hydration forces for several surfaces, e.g. silica, mica, lipid bilayer, can be ﬁtted as in [13]:

D F (D) = C exp(− ) (1.17) H H λ

D D FH (D) = C1exp(− ) + C2exp(− ) (1.18) λ1 λ2

(FH : short range force, D: separation distance between the surfaces, λ: decay length, CH : a hydration constant).

25 1. Introduction and Literature Review

Repulsive hydration forces have been reported for mica and silica surface immersed in 1:1 electrolytes with decay lengths of about 1 nm. Their effective range is about 3-5 nm [6] although hydration forces between a silica sphere and a silica plate in NaCl solutions have been observed for distances up to 15 nm [15]. In equation 1.18 two decay lengths are used for a better fit of the experimental data. Subramanian [16] cited values of λ1 and λ2 for mica surfaces in 1:1 electrolyte solutions in the range of 0.17-0.3 nm and 0.6- 1.2 respectively. Valle-Delgado et al. [13] reported short range repulsion between silica surfaces in 10−2M NaCl at different pH values: 9, 7, 5 and 3. Dishon et al. [17] observed short range repulsion between silica surfaces immersed in 10−3M of different electrolyte solutions: CsCl, NaCl and KCl. Other non-DLVO forces will be present at the surfaces when hydrophobic surfaces, polymers, or surfactants are included in the interaction.

1.2.2. Hydrophobic Interactions

Water does not wet hydrophobic surfaces because these kind of surfaces cannot bind the water by ionic or hydrogen bonds [18]. Hydrophobic interactions are attractive strong interactions between hydrophobic objects or nonpolar molecules. These interactions play an important role in biology, since they determine the conformation of proteins and the structure of biological membranes [19]. The measured hydrophobic forces are in some cases long range and decay exponentially with a decay length of 1-2 nm in the range of 0-10 nm, and then more gradually further out [18]. Some contradictory results are found in the literature. On one hand a decrease in magnitude of the hydrophobic attraction is reported with decreasing hydrophobicity of surfaces modiﬁed by silanes [20]. On the other hand the opposite behaviour is found for Langmuir-Blodget monolayers, the hydrophobic attraction increases with decreasing hydrophobicity [21]. Hato et al. [21] argued that the hydrophobic forces between macroscopic hydrophobic surfaces in aqueous solution have a range of 15-20 nm (short range interactions) and that the long range attraction (for D ≥ 20 nm) observed between hydrophobic surfaces in some experiments has no hydrophobic origin. Meyer et al. [22] reported that the long range hydrophobic interaction observed in hydrophobic surfaces prepared by LB-deposition of DODAB is due to the interaction between patchy bilayers. Several theories have been proposed to explain the long range hydrophobic interactions: electrostatic charges or correlated dipole-dipole [23–25], water structure [26, 27], phase metastability [28], bridging nanobubbles at hydrophobic surfaces [29–31]. The hydropho- Hph bic interaction energy Wa (D) can be deﬁned as [4]

Hphb −D/λ1 −D/λ2 Wa (D) = −2γ1e − 2γ2e (1.19)

−2 Where λ1 = 1-3 nm and γ1 = 10-50 mJm describes the short range hydrophobic interaction and λ2 and γ2 varies signiﬁcantly. The interaction between silica surfaces in 5 × 10−6 M CPC and 0.1 M NaCl is represented in ﬁgure 1.7[32]. Craig et al. [32] obtained a decrease in the long range hydrophobic attraction when the solution was

26 1. Introduction and Literature Review degassed. The magnitude of the attraction at small distance is very similar for a gas and degassed CPC solution. Although many models have been proposed to explain the hydration and the hydrophobic forces, more experiments are still necessary to clarify the mechanism behind these two important forces.

Figure 1.7.: Force measured between silica surfaces in 5×10−6M CPC and 0.1M NaCl. The interaction was measured in gassed (ﬁlled circles) and degassed (open circles) solutions. The gassed solution was measured prior to the degassed solution (A). The measured order was reversed (B) [32].

1.2.3. Structural Forces

Structural forces are important forces arising in conﬁned liquids. They are oscillatory changing from attraction to repulsion with distance and have a periodicity σp equal to the diameter of the liquid molecule [33]. The conﬁnement forces the liquid in the gap

27 1. Introduction and Literature Review to order in few layers, which are energetically or entropic favourable (energy minimum). Upon approach of another surface the order is disrupted and the layers are squeezed out successively from the gap with decreasing distance, giving rise to structural forces [33]. The structural forces can be deﬁned as [33]:

str −D/σ W (D) = W0cos(2πD/σp)e (1.20)

str (W (D) : interaction energy of the structural force, W0: interaction energy at distance, D = 0, σp: oscillatory period) Structural forces are not restricted to spherical molecules, these forces have been measured in colloidal dispersions confined in rigid and soft walls [34, 35]. Recently, Tabor et al. [36] published measured structural forces in SDS micellar solutions and microemulsions confined between two oil droplets (see figure 1.8)

Figure 1.8.: Force measured in SDS micellar solution and microemulsion conﬁned between two drops of perﬂuorooctane. The oil in water microemulsion consists of: 2 wt% oil phase (tetradecane), 5.5 wt% surfactant(SDS), 5.5 wt% cosurfactant(pentanol) in water [36].

Some other non-DLVO forces (not discussed here) may be present in colloidal systems, e.g. depletion and steric forces.

1.3. Surfactants

1.3.1. Classiﬁcation

Surfactants are used in many industries, e.g. chemical, cosmetic, food and pharmaceu- tical. They are also found in our body like the pulmonary surfactant. A surfactant is an amphiphilic molecule composed of a hydrophilic and a hydrophobic part [37]. If such molecules are dissolved in a polar solvent like water, its presence will disrupt the water structure breaking the hydrogen bonds between the water molecules (the hydrophilic

28 1. Introduction and Literature Review group have strong attraction for the polar solvent, whereas the hydrophobic group has little attraction for the polar solvent). The free energy of the system will increase and the surfactant will be expelled to the air water interface. Its hydrophobic groups will be facing the air (which is also hydrophobic) to minimize the contact with water, the hydrophilic groups will stay in the aqueous phase. The air–water interface becomes covered with a monolayer of surfactant and the surface tension of water is decreased. The surfactants are classiﬁed as anionic, cationic, amphoteric (zwitterionic), and non-ionic surfactants [1]:

• Anionic surfactants have hydrophilic negative charged groups in aqueous solvents. Sodiumdodecylsulfate and sodiumdodecanoate are examples of them. • Cationic surfactants carry a positive charge in their hydrophilic groups. Ammonium bromide surfactants like CTAB (C16H33N(CH3Br)) belong to this category. • Non-ionic surfactants do not carry any charge. They have polar groups which can interact with water. Examples of these surfactants are alkylethylene oxide or

sugar surfactants. Alkylethylene oxide surfactants are represented as Cnc Ene ; nc and ne indicate the number of carbon atoms in the alkyl chain and the number of ethylene oxide units in the hydrophilic head respectively. C12H25(0CH2CH2)6OH will be written as C12E6. The sugar surfactants are also known as alkylglycosides or alkylpolyglycosides. They consist of a hydrophilic head group (mono or oligo- sucrose, glucose or sorbitol) and a hydrophobic alkyl chain. • Zwitterionic surfactants carry a positive and a negative charge. Phosphatidyl- choline is an example of amphoteric surfactant.

As explained before, the addition of surfactants to water will decrease the surface tension of water due to the tendency of surfactant to be adsorbed at the interface. At a certain concentration the surface tension remains constant. This concentration is called the critical micelle concentration [1] and is a characteristic property of each surfactant. Above this concentration the surfactant aggregates spontaneously building micelles. When negative particles or solid surfaces are present in a solution of a cationic surfactant at concentrations well below the CMC, the surfactant will be adsorbed to the solid–liquid interface due to electrostatic interaction with the surface rendering the surface hydrophobic. A further increase of the concentration will produce surface aggregates or so called admicelles due to hydrophobic interactions. They resemble micelles formed in the bulk at a higher concentration [38]. Above the CMC several structures may be found on the surface: cylinders, micelles, bilayers, inverted micelles. Examples of the surfactant structures are given in ﬁgure 1.9. Israelachvili et al. [39] proposed the concept of molecular packing parameter. The type of aggregates (spherical micelle, bilayer, cylinder) depends on the packing parameter in the following way:

Ns = vo/alo (1.21)

29 1. Introduction and Literature Review

Spherical Micelle

Cylindrical or rod- like micelle Vesicle or Liposome

Inverted micelles Bilayer

Figure 1.9.: Aggregates formed by surfactants

(vo: volume of the hydrophobic part, lo: length of the hydrocarbon chains, a: eﬀective area per head group). For spherical micelles with a core radius R, made up of g molecules, the volume of the core can be expressed as [39]:

3 V = gvo = 4πR /3 (1.22) the surface area of the core A can be calculated as follows:

A = ga = 4π/R2 (1.23) therefore

R = 3vo/a (1.24)

30 1. Introduction and Literature Review assuming that there is no empty space between the hydrocarbon chains forming the micelle core. The core radius R will be equal to the length of the hydrocarbon chain lo, then

0 ≤ vo/alo ≤ 1/3 (1.25) for spherical micelles. The packing parameter for the other aggregates is given in table 1.1 Variable Sphere Cylinder Bilayer Volume of core 4πR3/3 πR2 2R V = gvo Surface area of core 4πR2 2πR 2 A = ga Area per molecule a 3vo/R 2vo/R vo/R Packing parameter vo/alo ≤ 1/3 vo/alo ≤ 1/2 vo/alo ≤ 1 vo/alo 3 2 Largest aggregation 4πlo/3vo πlo/vo 2lo/vo number gmax 3 2 Aggregation gmax(3v0/al0) gmax(2v0/al0) gmax(v0/al0) number g

Table 1.1.: Geometrical relations of diﬀerent aggregates; V, A, gmax and g refer to the complete spherical aggregate, unit length of a cylinder or unit area of a bilayer [39].

1.3.2. Surfactants at Interfaces

The adsorption of a solute to the solid–liquid interface increases the surface concentration. When the interaction is favourable the concentration of the solute at the surface will exceed the concentration of the solute in the bulk. This is known as surface excess [40]. The adsorption of ionic surfactant to a hydrophilic surface can be described using two principal models; the two step model and the four region model [38]. Electrostatic, hydrophobic, or a combination of both interactions determine the structure of the aggregates at the surface (see ﬁgure 1.10). At the lowest concentrations (region I), only electrostatic interactions between the positive surfactant head and the negative surface will determine the adsorption. At the highest concentrations (around the CMC), aggregates are formed at the solid–liquid interface, due to hydrophobic interactions (region IV). At intermediate concentrations (region II and III), the two models diﬀer. In the two step model only few isolated molecules are bound electrostatically to the surfaces, which then nucleate the formation of aggregates(admicelles). In the four region model stronger adsorption occurs at low concentration, which leads to the formation of hemimicelles before a second layer is attached via hydrophobic interactions.

31 1. Introduction and Literature Review

The CTAB adsorption isotherm at the silica surface is represented in ﬁgure 1.11.A double plateau can be distinguished in the curve. The adsorption isotherm resembles the two step model of ﬁgure 1.10[38]. Velegol et al.[41] detected rod-like aggregates by AFM at the silica surface with a peak to peak distance of 10 nm for concentration close to the CMC and above. Spherical aggregates (full micelle or half micelle on monolayer) have also been reported [42]. The low concentration region remains an area of investigation, since little information about the adsorbed layer at low surfactant concentration is available.

Figure 1.10.: Models for the two step and the four region model [38]

Figure 1.11.: (a) Normalized Raman integrated intensities as a function of CTAB bulk concentration (b) Adsorption isotherm obtained after subtraction of the bulk contribution and conversion of the Raman integrated intensities into adsorbed amounts [38]

1.4. Nanobubbles

Lou et al. [43] obtained one of the ﬁrst images of nanobubbles at a solid surface immersed in a liquid (ﬁgure 1.12). Nanobubbles can be induced on a hydrophobic surface using the solvent exchange method [43] or just immersing the hydrophobic surface in water [44]. Other methods of nanobubbles production are electrolysis and temperature change [45]. Nanobubbles are characterized by their high stability, several mechanisms have been proposed to explain this property. Ducker [46] explained the high stability of the nanobubbles based on contaminants, which will be present at the gas–liquid interface even in pure solutions. The contaminants will be adsorbed at the gas–liquid interface reducing the surface tension, and therefore the Laplace pressure. Another mechanism was

32 1. Introduction and Literature Review proposed by Brenner et al. [47]; the gas inside the nanobubbles diffuses out, but due to the dynamic equilibrium, the gas molecules return to the nanobubbles and as a result the nanobubbles will be stable. It is assumed that due to the small size of the nanobubbles, the gas leaving the nanobubbles will not collide with the returning gas. This kind of gas is called a Knudsen gas. There will be a circulating flow in the liquid near the gas–liquid interface, which is responsible for the returning of the gas molecules to the substrate and to the nanobubbles. The required energy is small and the substrate can supply the necessary thermal energy to drive this flow over the time scale that nanobubbles are observed [45].

Figure 1.12.: AFM image of bubbles on mica surface in water in tapping mode, with normal contact cantilever of spring constant equal to 0.38 N/m. Image size 1 × 1 µm [43]

33 Bibliography

[1] H-J. Butt, K. Graf, and M. Kappl. Physics and Chemistry of interface. Wiley-VCH, 2003. [2] D. F. Evans and H. Wennerström. The colloidal domain: Where Physics, Chemistry, Biology and Technology meet. VCH publishers, 1994. [3] G. Cao. Nanostructures and Nanomaterials: Synthesis, Properties and Applications. Imperial College Press, 2004. [4] W. Briscoe. Colloid Science: Principles, methods and applications. Ed. by T Cos- grove. John Wiley and Sons Ltd, 2010, p. 343. [5] H-J. Butt and M. Kappl. Surface and Interfacial forces. Wiley-VCH, 2010. [6] J. Israelachvili. Intermolecular and Surfaces Forces. 2nd. Edition. Academic Press, 1991. [7] C. J. Berg. An Introduction to Interfaces and Colloids: The Bridge to Nanoscience. World Scientiﬁc Publishing Co. Ptc. Ltd, 2010. [8] R. M. Pashley. In: Journal of Colloid and Interface Sci. 83 (1981), pp. 531–546. [9] J.-P. Chapel. In: Langmuir 10 (1994), pp. 4237–4243. [10] A. Anderson and W. R. Ashurst. In: Langmuir 25 (2009), pp. 11549–11554. [11] V. V. Yaminski, B. W. Ninham, and R. M. Pashley. In: Langmuir 14 (1998), pp. 3223–3235. [12] B. C. Donose, I. U. Vakarakelski, and K. Higashitani. In: Langmuir 21 (2005), pp. 1834–1839. [13] J. J. Valle-Delgado et al. In: J. Chem. Phys. 123 (2005), pp. 034708–1. [14] G. Vigil et al. In: J. Colloid Interface Sci. 165 (1994), pp. 367–385. [15] W. A. Ducker, T. J. Senden, and R. M. Pashley. In: Nature 353 (1991), p. 239. [16] V. Subramanian. “Eﬀects of Long-chain Surfactants, Short-chain Alcohols and Hy- drolizable Cations on the Hydrophobic and Hydration Forces”. PhD thesis. 1998. [17] M. Dishon, O. Zohar, and U. Sivan. In: Langmuir 25 (2009), pp. 2831–2836. [18] M. Ruths and J. N. Israelachvili. “Nanotribology and Nanomechanics: An Intro- duction”. In: ed. by Bharat Bushan. Springer-Verlag, 2008. Chap. 9, pp. 417–497. [19] J. Israelachvili and R. Pashley. In: Nature 300 (1982), pp. 341–342. [20] Y. I. Rabinovich and R.-H. Yoon. In: Langmuir 10 (1994), pp. 1903–1909.

34 Bibliography

[21] M. Hato. In: J. Phys. Chem. 100 (1996), pp. 18530–18538. [22] E. E. Meyer et al. In: PNAS 102 (2005), pp. 6839–6842. [23] Y-H. Tsao, D. F. Evans, and H. Wennerstroem. In: Langmuir 9 (1993), pp. 779– 785. [24] Y. I. Rabinovich, D. A. Guzonas, and R. H. Yoon. In: Langmuir 9 (1993), pp. 1168– 1170. [25] R. Podgornik. In: Chem. Phys. Lett. 156 (1989), pp. 71.–75. [26] L. R. Pratt and D. Chandler. In: J. Chem. Phys. 67 (1977), pp. 3683–3704. [27] J. Ch. Erikkson, S. Ljunggren, and P.M. Claesson. In: J. Chem. Soc. Faraday Trans. 2 85 (1989), pp. 163–176. [28] H. K. Christenson and Per M. Claesson. In: Science 239 (1988), p. 390. [29] V. V. Yaminski and B. Ninham. In: Langmuir 9 (1993), pp. 3618–3624. [30] A. Carambassis et al. In: Phys. Rev. Lett. 80 (1998), pp. 5367–5360. [31] J. W. G. Tyrrell and P. Attard. In: Phys. Rev. Lett. 87 (2001), pp. 176104–1– 176104–4. [32] V. S. J. Craig, B. W. Ninham, and R. M. Pashley. In: Langmuir 15 (1999), pp. 1562– 1569. [33] M. Ruths and J. N. Israelachvili. Handbook of Nanotechnology. Ed. by B. Bhushan. 2nd. Edition. Springer, 2007, p. 860. [34] S. Klapp et al. In: Phys. Rev. Lett. 100 (2008), p. 118303. [35] Y. Zeng and R. v. Klitzing. In: Soft Matter 7 (2011), p. 5329. [36] R. F. Tabor et al. In: J. Phys. Chem. 2 (2011), pp. 434–437. [37] M. J. Rosen. Surfactants and interfacial phenomena. Ed. by Inc Wiley & Sons. 3rd. 2004. [38] E. Tyrode, M. W. Rutland, and C. D. Bain. In: J. Am. Chem. Soc. 130 (2008), pp. 17434–17445. [39] R. Nagarajan. In: Langmuir 18 (2002), pp. 31–38. [40] R. Atkin et al. In: Adv. Colloid Interface Sci. 103 (2003), pp. 219–304. [41] S. B. Velegol et al. In: Langmuir 16 (2000), pp. 2548–2556. [42] W. A. Ducker and E. J. Wanless. In: Langmuir 15 (1999), pp. 160–168. [43] Shi-Tao. Lou et al. In: J. Vac. Sci. Technol. B 18 (2000), pp. 2573–2575. [44] N. Ishida et al. In: Langmuir 16 (2000), pp. 6377–6380. [45] V S. J. Craig. In: Physics 70 (2011). doi: 10.1103/Physics.4.70. [46] W. A. Ducker. In: Langmuir 25 (2009), pp. 8907–8910. [47] M. P. Brenner and D. Lohse. In: Phys. Rev. Lett. 101 (2008), p. 214505.

35 2. Techniques

2.1. Atomic Force Microscopy

Atomic Force Microscopy (AFM) was developed by Binning, Quate and Gerber in 1985 to measure forces as small as 10−18N[1]. An atomic force microscope can be used to provide high resolution topographical analysis of conducting or non-conducting surfaces [2]. Force measurements can also be performed with this equipment. The technique is well described elsewhere [2–4]. The sample is placed on the scanner and a cantilever is mounted on the AFM head. A piezoelectric positioner is used to bring the cantilever to the surface. The deflection of the cantilever is held at a defined constant value by means of feedback. The deflection of the cantilever due to the interactions with the surface is monitored by a photosensitive detector. The MFP-3D Asylum Research AFM, mounted in an inverted optical microscope (Olympus IX71), was used to perform the scanning and the force measurements experiments. The MFP-3D provides an Igor Pro software extension for the runs and analysis of the experiments. A more detailed description of the apparatus is given below [5]. Figure 2.1 shows a MFP-3D atomic force microscope. A top view camera is included, which allows to position the laser spot on the cantilever. This apparatus has x, y, z piezo stages. The z piezo is seated on the head and is used to move and oscillate the cantilever. The x and y piezos are seated on the base of the MFP-3D and move the sample in the corresponding directions. A position sensitive detector is used to measure the deflection of the cantilever due to the interaction with the surface. The detector consist of a photodiode. The deflection of the cantilever produces changes in the position of the laser beam reflected from the cantilever to the detector. The position of the reflected laser beam on the detector is determined by the angle of the deflected cantilever. The photodiode has 4 segments: A, B, C and D. The voltage generated in each segment is proportional to the amount of light hitting the segment. The deflection signal can be written as the difference between the two segments placed on the top minus the to segments placed on the bottom (see photodetector in figure 2.3).

Deflection = Vtop − Vbottom = (VA + VB) − (VC + VD) (2.1)

The lateral signal can be calculated as the diﬀerence between the two left segment minus the segments in the right side.

36 2. Techniques

Figure 2.1.: A representation of the MFP-3D used during experiments [5]

Lateral = Vleft − Vright = (VA + VC ) − (VB + VD) (2.2)

At the beginning of the experiment, the laser beam is aligned to hit the center of the photodiode, so that the deflection and the lateral deflection is close to zero. If Vtop ≥ Vbottom, then the interactions between cantilever and surface are repulsive. When Vtop ≤ Vbottom, then the interactions between cantilever and surface are attractive. The inverse of the optical level sensitivity (InvOLS) is used to convert the cantilever deflection from volts to meter. The InvOLS is useful when force measurements are performed, it can be determined from the constant compliance region performing a force curve against a hard surface (see figure 2.2) The InvOLS is assumed to be the "zero distance", but in some cases, when using highly deformable surfaces or when layered structures cause strong repulsive forces, the constant compliance region does not represent the "zero separation distance" [6]. The Hookes law can be used to transform the measured deflection into force [6]:

F = k × zc (2.3)

(k: spring constant of the cantilever, zc: cantilever deﬂection).

zc = InvOLS × ∆V (2.4)

(∆V : voltage measured by the photodiode). The distance between tip and surface D, can be calculated as follows:

37 2. Techniques

Constant 0 compliance region -1 ]V[ noitcelfeD ]V[ -2

-3

-4V

-2 -4 -6 -8 -10µm Zsnsr

Figure 2.2.: The raw data for InvOLS determination. The y axes represents the deﬂection of the cantilever in volts. The x axes (Zsnsr) is the piezo position.

D = zc + zp (2.5)

(zp: position of the piezo normal to the surface). There are several methods for the determination of the spring constant [7,8]. Cleveland et al. [9] proposed a method for the determination of the spring constant of the cantilever based on the attachment of a known mass to the end of the cantilever and measuring the change in resonant frequency. The spring constant can be obtained from the geometry of the cantilever [9]:

Et3w k = (2.6) 4l3

(E: elastic modulus; t, w, l: thickness, width and length of the cantilever respectively). For a rectangular cantilever, when a mass M is added to the end, then the resonant frequency can be calculated as:

ω 1 r k ν = = (2.7) 2π 2π M + m∗

∗ (m ≈ 0.24mb; mb: mass of the cantilever). The mass of the cantilever mb can be obtained from the equation below:

38 2. Techniques

mb = ρωtl (2.8)

(ρ: represents the mass of the material). When a mass is added then the resonant frequency takes the form:

t E υ ≈ ( )1/2 (2.9) 0 2πl2 ρ and ﬁnally

M = k(2πν)−2 − m∗ (2.10)

The intercept gives the eﬀective mass and from the slope of the equation 2.10 the spring constant can be calculated. In the Sader method [10] the spring constant can be calculated from the dimensions of the cantilever. For rectangular cantilevers the expression takes the form:

2 k = MeρcbhLωvac (2.11)

(ωvac: fundamental radial resonance frequency of the cantilever in vacuum; h, b, l: thickness, width, and length of the cantilever respectively; ρc: density of the cantilever; Me = 0.2427 for L/b>5 - the normalized eﬀective mass). The method used to measure the spring constant in this research was proposed by Butt et al. [11]. Using the equipartition theorem:

1 1 k T = kx2 (2.12) 2 B 2

(kB: Boltzmann constant), the solution of the equation 2.12 is:

k T k = B (2.13) hx2i

Then the power spectral density (PSD) of the distance x is fitted to the theoretical one for a simple harmonic oscillator and from there the spring constant can be obtained [5]. The surface can be scanned in different modes: contact, tapping, and non-contact mode [5]. The wavelength of the cantilever used is 860 nm. The forces are measured between two colloidal particles. One particle is glued to the cantilever and the other one to a glass slide using a micromanipulator with a mounted micropipette [12](see figure 2.3)

39 2. Techniques

Laser Photodetector

A B

o C D z

4.8 µm silica particle

Figure 2.3.: Representation of a force measurement between two particles

2.2. Scanning Electron Microscopy

A Hitachi S-4000 scanning electron microscope (SEM) was used with a cold field emitter (resolution 2 nm) to obtain images of the silica particles. The accelerating voltage was 20 kV and the beam current 5.15 pA. The image mode was a secondary electron image. The principles of a SEM are discussed below, see figure 2.4 for a representation of a scanning electron microscope. The SEM can image and analyze bulk samples [13]. The electrons coming from an electron gun have a typical energy of 2-40 kV. The electron beam is demagnified into a probe of electrons [14]. The probe of electrons with a diameter of 1- 10 nm carrying a current of 10−9 − 10−12 A is focused onto the surface and moved across the surface in parallel lines [13, 15]. The interaction of the electrons with the surface produces several phenomena, among them the emission of secondary electrons with an energy of 2-5 eV, and high energy backscattered electrons. The limit between secondary electrons and backscattered electrons is drawn at 50 eV. The secondary electrons are emitted from the sample and generated by inelastic collisions to high energy levels, so that the excited electrons can overcome the work function before a deceleration to the Fermi level occurs [13]. The backscattered electrons are electrons from the incident beam, which interact with atoms in the sample and are backscattered again. The intensity of both emissions, secondary and backscattered electrons, is sensitive to the angle at which the incident beam contact the surface. The emissions are collected by the detectors and amplified. The resulting signal is used to control the brightness in a cathode ray tube (CRT) [16]. The CRT scan is synchronized with the beam scan, which allows the signals to be transferred point to point and a map of the scanned area can be displayed. The scanning electron microscopy image is a magnification of the topography of the sample, secondary or backscattered images can be obtained. The contrast of a backscattered

40 2. Techniques

SEM image depends on the intensity of the emitted backscattered electrons. When heavy atoms are present in the sample, more backscattered electrons will be produced and a brighter contrast is obtained. Therefore, local variations in average atomic number vary the contrast of the image [15]. The interaction of the electrons with the sample produces other emissions: X ray photons, Auger electrons, and perhaps light [14]. The spectrum of the x-radiation can be used for quantitative chemical microanalysis. Auger electrons are emitted from atomic layer close to the surface and give information about the surface chemistry.

Figure 2.4.: Instrumentation of a scanning electron microscope [16]

2.3. Zeta Potential

A Malvern zetasizer nano ZS with a 633 nm red laser was used to measure the zeta potentials. Other parameters like particle size (only for monodispersed samples) and molecular weight can be measured with this instrument [17]. To determine the size of the particles, it is necessary to measure the Brownian motion of the particles in the sample using Dynamic Light Scattering (DLS), also known as Photocorrelation Spectroscopy (PCS). Small particles will move quickly and bigger particles will move slower. The particles are illuminated with a laser and the intensity ﬂuctuations of the scattered light are analyzed. If a small particle is hit by a light source the particle will scatter the

41 2. Techniques

Diffuse layer potential

negative charged particle

Ions strongly bound to particle ------

Slipping plane ------Zeta------potential

Figure 2.5.: Double layer of a particle light in all directions. If many particles are present in the system a speckle pattern will be formed which consist of bright and dark areas. The bright areas are regions, where the light scattered by the particles has the same phase and interferes constructively to form a bright patch. The dark areas are regions, where the phase additions are mutually destructive and cancel each other out. The Stokes-Einstein equation relates the size of the particle with its speed due to Brownian motion. Since the particles move, the intensity appears to fluctuate. The zetasizer measures the rate of the intensity fluctuation and from there calculates the size of the particles [17]. When charged particles are present in a medium, an electric double layer will be developed. The double layer consists of ions, which are firmly bound to the surface (Stern layer) and ions, which are loosely bound (diffuse layer) to the surface (see figure 2.5). The slipping plane describes a boundary in the diffuse layer. Any ions within this boundary will move together with the particle. The potential at the slipping plane is called zeta potential [17]. The zeta potential gives an indication of the stability of the sample. Low positive or negative zeta potential values (below 30 mV) are considered as non-stable systems (steric stabilization is not considered). The electrophoretic mobility of the particles is obtained from an electrophoresis experiment performed on the sample and measuring the particles velocity using Laser Doppler Velocimetry (LDV). Electrophoresis is the movement of charged dispersed particles relative to the liquid (dispersant) under the influence of an applied electric field [17]. When an electric field is applied to the dispersion, the charged particles will move to the electrode of opposite charge. The velocity of the particles depends on the following parameters [17]

• Strength of electric ﬁeld

42 2. Techniques

Figure 2.6.: Scheme of the Laser Doppler Velocimetry (LDV) [17]

• Dielectric constant of the medium • Viscosity of the medium • Zeta potential

The electrophoretic mobility is deﬁned as the velocity of a particle in an electric ﬁeld. From the Henry equation the zeta potential of the particle can be obtained [17]:

2zf(ka) U = (2.14) E 3η

(z: zeta potential; UE: electrophoretic mobility; : dielectric constant; η viscosity; f(ka): Henrys function). The Henry function (f(ka)) takes the values 1.5 or 1.0, (f(ka)) is 1.5 for aqueous media and moderate electrolyte concentration and is also known as Smolu- chowski approximation. To fit to the Smoluchowski model the particles have to be larger than 0.2 µm dispersed in 10−3M. For small particles in low dielectric constant medium and for non-aqueous measurements, f(ka) is 1.0 and the Huckel approximation can be used for the calculation of the zeta potential. The Laser Doppler Velocimetry measures the velocity of the particles during the electrophoresis. The scattered light has an angle of 17◦ and is combined with the reference beam, producing a fluctuating intensity signal. The rate of fluctuation is proportional to the speed of the particles (see figure 2.6). Electroosmosis can also occur in the measuring cell. The true electrophoretic mobility is measured at the stationary layer. The stationary layer is a point in the cell where the electroosmotic flow is zero [17]. The molecular weight of a sample can be obtained from the static light scattering. The values obtained at one angle are not so accurate for high molecular weight polymers, due to the non-isotropic scattering profiles; the intensity depends on the angle of observation, but for small particles the sample scattering becomes isotropic and the angle dependence is minimized. Therefore, the molecular weight of small proteins and polymers can be measured using this method [17, 18]. The particles

43 2. Techniques are illuminated by a light source. The particles scatter the light in all directions and the time-averaged intensity of scattered light is measured. In that way the molecular weight and the second virial coefficient A2 can be obtained. A2 describes the interaction strength between the particles and the solvent. When A2 > 0 the dispersion is stable, for A2 < 0 the particles aggregate, and for A2 = 0 the particle–solvent interaction strength equals the molecule–molecule interaction strength, the solvent can then be defined as a theta solvent. To determine the molecular weight, measurements of a sample at different concentrations have to be performed. Then the Rayleigh equation can be applied [17]:

KC 1 = ( + 2A2C)P (θ) (2.15) Rθ M

(Rθ: ratio of scattered light to incident light of the sample; M: molecular weight; A2: second virial coeﬃcient; C: concentration; P (θ): angular dependence of the sample scattering intensity; K: optical constant). When the dispersed particles are much smaller than the incident light then P (θ) is 1. This type of scattering is known as Rayleigh scattering. Then equation 2.15 takes the form:

KC 1 = ( + 2A2C) (2.16) Rθ M

44 Bibliography

[1] G. Binning, C. F. Quate, and C. Gerber. In: Phys. Rev. Lett. 56 (1986), pp. 930– 933. [2] J. Ralston et al. In: Pure Appl. Chem. 77 (2005), pp. 2149–2170. [3] P. West and A. Ross. An Introduction to Atomic Force Microscopy Modes. Santa Clara, CA: Paciﬁc Nanotechnology, Inc., 2006. [4] W. A. Ducker, T. J. Senden, and R. M. Pashley. In: Langmuir 8 (1992), pp. 1831– 1836. [5] MFP-3D manual. url: https://support.asylumresearch.com/forum/content. php?43-MFP-Manual-Version-04-08-Released. [6] H-J. Butt, B. Capella, and M. Kappl. In: Surface Science Reports 59 (2005), pp. 1– 152. [7] T. Senden and W. A. Ducker. In: Langmuir 10 (1994), pp. 1003–1004. [8] J. E. Sader et al. In: Rev. Sci. Instrum. 66 (1995), pp. 3789–3798. [9] J. P. Cleveland et al. In: Rev. Sci. Instr. 64 (1993), pp. 403–405. [10] J. E. Sader, J. W. M. Chon, and P. Mulvaney. In: Rev. Sci. Intr. 70 (1999), pp. 3967– 3969. [11] H-J. Butt and M. Jaschke. In: Nanotechnology 6 (1995), pp. 1–7. [12] G. Toikka, R. A. Hayes, and J. Ralston. In: Langmuir 12 (1996), pp. 3783–3788. [13] L. Reimer. Scanning Electron Microscopy; Physics of Image Formation and Micro- analysis. 2nd. edition. Springer-Verlag, 1998. [14] K. D. Vernon-Parry. In: III-Vs Review 13 (2000), pp. 40–44. [15] A. Putnis. Introduction to Mineral Sciences. Cambridge University Press, 1992. [16] G. W. Kammlott. In: Surface Science 25 (1971), pp. 120–146. [17] Zetasizer Nano Series User Manual. url: http://www.biophysics.bioc.cam. ac.uk/files/Zetasizer_Nano_user_manual_Man0317-1.1.pdf. [18] Molecular weight measurements with the Zetasizer Nano system. url: http://www. malvern.com/common/downloads/campaign/MRK528-01.pdf.

45 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes using CP-AFM

3.1. Introduction

Dispersions of colloidal particles find applications in many areas like pharmaceutics, cosmetics, food industry and others. The DLVO theory remains the starting point to describe the interaction between colloidal particles assuming only two types of forces, repulsive electrostatic forces and attractive van der Waals forces [1]. At short separation some deviations from the theory are reported [2–4]. These deviations are called non- DLVO forces and are present in water and electrolyte solutions. Several theories have been developed to explain the origin of non-DLVO forces at small separation distances between colloidal particles. Chapel [2] proposed that the hydration forces are caused by creation of a hydrogen bonding network at the silanol level, and that this force produces the short range repulsion. This repulsion is reduced in the presence of any salt, and the more hydrated the ion is, the weaker this force will be. Some authors have proposed that the short range repulsion is due to polysilicic acid chains protruding from the silica surface [5–7]. Others related the short range repulsion in the presence of electrolytes to the dehydration of counterions; well hydrated counterions will produce short range repulsion of larger extent [8,9]. According to the model of Torrie et al. [10], small counterions with high affinity for water accommodate better on the structured water layer and large ions like Cs+ prefer to reside outside the hydration layer producing short range repulsion of longer extent. Higashitani’s model [11] explains the difference of the adsorbed layer at the silica surface using highly and poorly hydrated counterions. The cation cesium is less hydrated and can be closely packed at the silica surface. The adsorbed layer composed of the poorly hydrated cesium cations will be thin but strong adsorbed to the silica surface. The adsorbed layer composed of hydrated lithium cations will be thick. The larger the hydration shell, the more unstable are the layers and they can be easily pressed out. The Higashitani’s model contradicts the model proposed by Torrie but is in good agreement with the mechanism explained by Pashley [12] for hydration forces. For hydration forces to be present, the counterions have to be bound to the surface in a specific way and they should not be desorbed upon the approach of the other surface [12]. Until now the interactions at short distances are still discussed controversially and not so well understood. The interaction between colloidal particles may be different depending

46 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes on the electrolyte solution used, which is known as ion specificity. Usually the Hofmeister series is observed for the adsorption of ions to the silica surface [13–15]. Their rheological and zeta potential results at high ionic strength show that the binding of cations to the silica surface is according to the sequence Cs+ >K+ >Na+ >Li+, since with increasing hydration shell, the ions prefer to stay in water. The interaction between particles can be measured using several techniques. Franks [14] investigated the interaction of suspension of two types of silica particles, amorphous and quartz, in different electrolytes using rheology and zeta potential techniques. He proved that less hydrated ions adsorb in larger amount to the silica surface than well hydrated ions. Pashley and Israelachvili [16] demonstrated the presence of short range repulsion between mica surfaces across K+ ions with the SFA. Chapel [2] investigated the influences of ion size on hydration forces for silica surfaces with the SFA. The strongest hydration force was obtained for silica surfaces immersed in pure water. The more hydrated cation produces the weaker force. It might be that the cation Li+ competes with the hydroxyl groups to order the water around the silica surface. Vakarelski et al. [11] studied the adhesive force between a silica particle and a mica surface in electrolyte solution using CP-AFM. A strong adhesion was found for highly hydrated ions (Li+, Na+). Borkovec et al. [5] proved the validity of the Derjaguin approximation between two colloidal particles by the colloidal probe technique across KCl electrolyte solution. Dishon et al. [17] studied the effect of different salts on the force between a silica particle and a silica surface using CP-AFM, but LiCl was not investigated. They obtained that the tendency for the adsorption at the silica surface grew monotonically with the bare ion size according to the sequence Cs+ >K+ >Na+, which is in good correlation with the Higashitani’s model. The addition of salt beyond neutralization led to excess cation condensation and charge reversal in the presence of monovalent ions. To our knowledge, there is no systematic study about the interaction between two silica particles (in the colloidal range) across different electrolytes. The aim of the current work is to study the role of the non-DLVO forces between two silica particles at low ionic strength across different electrolytes, LiCl, NaCl, KCl, and CsCl and to clarify the contradictions between different theoretical models (Chapel [2], Higashitani [11], Torrie [10]).

3.2. Experimental Section

3.2.1. Materials

Dry nonporous silica particles 4.74 µm mean diameter were purchased from Bangs Lab- oratories. The particles were resuspended in pure water (milli-Q water) and centrifuged three times before use. The surface topography of the silica particles was investigated with a MFP-3D by tapping mode in air. The roughness was calculated from images of 5 particles by ﬂattening third order polynomials. Electrolytes solution of LiCl, NaCl, KCl, and CsCl were prepared at diﬀerent pH and ionic strength. The pH was adjusted by

47 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes addition of LiOH, NaOH, KOH, CsOH or HCl. All the chemicals used were of analytical grade quality.

3.2.2. Preparation and Methods

Preparation

A colloidal particle was glued to the end of a tip-less AFM cantilever (CSC12, µ-mach, Lithonia) with a nominal spring constant of 0.03 N/m. Another particle was glued to a glass slide (Menzel-Gläser, Germany) using an optical microscope and a micromanipulator. Colloidal probes and glass slides with attached particles were cleaned with ethanol and water and placed 20 minutes in an air plasma cleaner (Diener electronic. Femto timer).

Methods

The force measurements between the two silica particles [18] were performed using a MFP-3D Asylum Research mounted in an inverted optical microscope (Olympus IX71). This technique is well described elsewhere [19–21]. In brief, a cantilever with the colloidal probe is mounted on the AFM head and the glass slide with the attached particle is placed on the scanner [18]. The two opposing particles are optically aligned. A laser is pointed at the end of the cantilever. The cantilever moves in the z-direction and the deﬂection of the cantilever while approaching the surfaces is registered by a photosensitive detector. The spring constant is determined using the thermal noise method; the typical value is 0.03 N/m. During the measurements, an inverse microscope placed in the AFM was used to check, if the particle was still attached to the cantilever. For the analysis, only approach curves are shown. The velocity of the approach was 600 nm/s. All cantilevers were plasma cleaned before use. The measurements with MPF-3D were obtained at room temperature at 1 atm.

3.2.3. Simulations

The simulations are based on the DLVO theory (DLVO Fitting.ipf procedure written by McKee [22] based on the algorithm proposed by Chan [23]). The DLVO theory only takes two kinds of forces into account to explain the stability of colloids in a suspension: electrostatic double layer and van der Waals forces. The electrostatic interactions are calculated solving numerically the non-linear Poisson-Boltzmann equation for two identically charged solids in water using two boundary conditions, constant charge and constant potential. The van der Waals forces are the sum of the interactions between atomic and molecular dipoles in the particles or between diﬀerent particles [24]. Ac- cording to Borkovec [5], the Derjaguin approximation is still valid for the particle size of 4.74 µm. A water dielectric constant of =78 is used and a non-retarded Hamaker

48 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

(si/water/si) constant=8.5×10−21 J[25] for the calculation of the van der Waals interaction is assumed.

3.3. Results

3.3.1. Eﬀect of Ionic Strength: 10−4 M and 10−3 M

The roughness of the silica particles is RMS=2.0 nm which correlates well with the values reported in literature. Figure 3.1 shows force curves against separation for diﬀerent salts: NaCl, KCl, LiCl, and CsCl at constant pH=5.8 measured between two silica particles of 4.74 µm in diameter.

0.1 NaCl, ö =-24mV, ë=30.03 nm 8 KCl, ö=-20 mV, ë=30.3 nm 0.1 7 LiCl, ö=-20mV, ë=30.3 nm 6 CsCl, ö= -18mV, ë=30.3 nm 6 4

5 2 4 0.01 ðR [mN/m] 6 3

F/2 4

2 2 0.001 0 5 10 15 20 25 30 Separation, D [nm] 0.01 ðR [mN/m] 8

F/2 7 6 5 4

0.001 0 20 40 60 80 100

Separation, D [nm]

Figure 3.1.: Forces between a pair of colloidal silica particles across diﬀerent aqueous electrolyte solutions; LiCl, NaCl, KCl and CsCl, at a ﬁxed electrolyte concentration of 10−4 M and pH=5.8; Hamaker constant A= 8.5×10−21 J. The continuous lines correspond to constant charge and the discontinuous ones to constant potential.

49 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

From zeta potential measurements it is known that at this pH the silica is negatively charged [26]. The theoretical ionic strength was 10−4 M given by the added salt. The DLVO fitting procedure for the experimental curves is described in 3.2.3. The respective average potentials (based on different pairs of silica particles), ϕ and ionic strength, I are shown in table 3.1. At least 50 forces curves were taken during the AFM measurements for a pair of silica spheres and reproducibility of the curves was observed. An 20% error due to spring constant determination is assumed during the measurements. The slopes of the force curves are the same for all the experimental curves (see figure 3.1) and corresponds to a Debye length k−1 of 30 nm and an ionic strength I of 10−4 M. At distances larger than 50 nm the two boundary conditions constant charge and constant potential simulate well the experimental curves. At smaller distances the experimental curves are located between these two boundary conditions. Repulsion was observed at separation >10 nm and is similar for all the studied electrolytes within the experimental errors (see table 3.1). At distances smaller than 10 nm attraction is observed. The attraction decreases in the order of Li+ >Na+ >K+ >Cs+. Figure 3.2 represents force curves versus particle distance for the same salts at 10−3 M ionic strength and constant pH=5.8 (not adjusted normal water pH). The Debye length k−1 is 10.1 nm for NaCl and KCl and 9.59 nm for the other two salts which is close to the ideal value of 10 nm for an ionic strength of 10−3 M. As for the lower ionic strength, the experimental curves are well simulated at large distances (>20 nm) with the constant charge and constant potential model using the same parameters as in figure 3.1 and only repulsion is observed at larger separations. For a given distance the forces are similar at long range, but a slightly lower repulsion is observed for Cs+. That is related to a slight decrease in the simulated surface potential Cs+ (15 mV) (see figure 3.2). A negative sign is assumed for the simulated potentials. Interestingly no short range attraction is observed at this ionic strength, only short range repulsion. The short range repulsion decreases in the order Li+ >Na+ >K+ >Cs+. Comparing the two ionic strengths, 10−4 M and 10−3 M, one can conclude that the surface potential decreases with increasing salt concentration. No ion specific effect is observed for distances > 10 nm, but at 10−3 M ionic strength a slight tendency of Cs+ to be more adsorbed at the surface is observed. Short range attraction is seen at 10−4 M, the attraction decreases in the order of Li+ >Na+ >K+ >Cs+ for 10−4 M. In contrast to this, no short range attraction is observed for 10−3 M. The experimental curves are between the constant charge and constant regulation boundary conditions. Monte Carlo simulations were performed for two silica surfaces immersed in 1 mM electrolyte solution. Two ion diameters were used for the calculation: 3.5 Åand 5 Å; and two different surface descriptions: smeared out surface charge (implicit) and explicit sites (explicit). The parameters used for the calculations were taken from reference [27]. The pK of the sites was fixed at 7.7. The calculations were performed by Christophe Labbez (see figures 3.3 and 3.4). The Monte Carlo simulations are in good agreement with the experimental data. The best agreement is found with explicit sites.

50 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

0.1 8 NaCl, ö=-24 mV, ë=10.1 nm KCl, ö =-20mV, ë=10.1nm 0.1 7 LiCl, ö=-16 mV, ë=9.59 nm 6 CsCl, ö=-15 mV, ë=9.59 nm

5 4 0.01

3 ðR [mN/m] F/2 0.001 2 0 2 4 6 8 10 Separation, D [nm]

ðR [mN/m] 0.01

F/2 8 7 6 5 4

0.001 0 10 20 30 40 Separation, D [nm]

Figure 3.2.: Forces between a pair of colloidal silica particles across diﬀerent aqueous electrolyte solutions; LiCl, NaCl, KCl and CsCl, at a ﬁxed electrolyte concentration of 10−3 M and pH=5.8; Hamaker constant A= 8.5×10−21 J. The continuous lines correspond to constant charge and the discontinuous ones to constant potential.

51 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

Theoretical Avg. Diﬀuse layer Debye Zeta Ionic Strength Layer Potential Length Potential Salt I (mM) pH ϕ (mV) k−1(nm) ζ (mV) 0.20 3.97 -24.00±0 21.40 -2.00 0.10 5.80 -28.00±5.7 30.03 -70.00 1.00 5.80 -28.00±10.0 10.10 -66.00 NaCl 1.00 3.88 0.00 2.00 0.20 3.96 6.50 0.10 5.80 -21.00±1.0 30.03 -70.00 1.00 5.80 -17.00±1.4 9.59 -66.00 LiCl 1.00 4.09 -15.00±0 9.59 11.40 0.20 4.05 -2.00 0.10 5.80 -21.00±2.0 30.03 -67.00 1.00 5.80 -20.00±2.8 10.10 -66.00 KCl 1.00 4.03 0.00 0.00 0.20 3.99 -1.00 0.10 5.80 -19.00±1.4 30.03 -67.00 1.00 5.80 -14.00±0.7 9.59 -67.00 CsCl 1.00 4.01 0.00 -1.58

Table 3.1.: Results of simulations of direct force measurements by the Poisson-Boltzmann theory

52 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

0.25 GCMC_1mM_Exp_Dm_3.5 GCMC_1mM_Exp_Dm_5 LiCl 0.20 CsCl KCl NaCl

0.15

F/R [mN/m] 0.10

0.05

0.00

0 20 40 60 80 100

Separation, D[nm]

Figure 3.3.: Monte Carlo simulation with an explicit surface charge description of the experimental data at 1 mM ionic strength and pH=5.8. The calculations were performed by Christophe Labbez.

53 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

0.25 GCMC_1mM_Imp_Dm_3.5 GCMC_1mM_Imp_Dm_5 0.20 LiCl CsCl KCl NaCl 0.15

0.10 F/R [mN/m]

0.05

0.00

-0.05

0 20 40 60 80 100

Separation, D[nm]

Figure 3.4.: Monte Carlo simulation with an implicit surface charge description of the experimental data at 1 mM ionic strength and pH=5.8. The calculations were performed by Christophe Labbez.

54 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

3.3.2. Eﬀect of pH

Figure 3.5 shows interactions between two silica particles at an adjusted pH=4 across the electrolytes solutions previously mentioned. The background electrolyte concentration was 10−4 M.

0.1 9 8 0.2 mM NaCl pH=3.97, ë=21.4nm, ö=-24mV 0.2 mM CsCl pH=3.99, 0.1 7 0.2 mM KCl pH=4.05 6 0.2 mM LiCl pH=3.96

5 4 0.01

3 ðR [mN/m) F/2

2 0.001 0 5 10 15 20 Separation, D [nm]

0.01 9

ðR [mN/m] 8 7 F/2 6 5

0.001 0 20 40 60 80 Separation, D [nm]

Figure 3.5.: Forces between a pair of colloidal silica particles across diﬀerent aqueous electrolyte solutions; LiCl, NaCl, KCl and CsCl, at a ﬁxed electrolyte concentration of 10−4 M and pH=4. Hamaker constant A= 8.5 × 10−21 J. The continuous lines correspond to constant charge and the discontinuous ones to constant potential.

No repulsion between the silica particles in presence of CsCl, KCl, and LiCl could be observed. The interactions in presence of NaCl electrolyte solution remain repulsive even at short distances. The obtained potential and Debye length is 24 mV and 21.5 nm respectively. The potential is assumed to be still negative at this pH in presence of NaCl, but a slight additional decrease of the pH causes the collapse of the double layer and no repulsion is seen anymore (data not shown). The decrease of the pH to 4 at an electrolyte concentration of 10−4 M causes a variation of the ionic strength from 1×10−4 M to 2 × 10−4 M due to the presence of more protons in the solution. Surprisingly, no

55 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

0.2 GCMC_1mM_Exp_Dm_3.5_pH=3.88 GCMC_1mM_Exp_Dm_5_pH=3.88' NaCl_pH= 3.88'

0.1

0.0 F/R [mN/m]

-0.1

-0.2

0 20 40 60 80 100

Separation, D [nm]

Figure 3.6.: Monte Carlo simulation with an explicit surface charge description for the interaction curve between a pair of colloidal particles across NaCl aqueous electrolyte solution at a fixed ionic strength of 1 mM and pH=4. The calculations were performed by Christophe Labbez. short range attraction is measured for NaCl at pH=4, instead short range repulsion is observed. Is this measured short range repulsion due to hydrogen bonds at the silanol groups? To answer this question, measurements through water were performed. At 10−3 M ionic strength and constant pH=4, attractive interactions were measured for all the salts. A representation of the interaction in the presence of NaCl is given in figure 3.6. The simulations performed with these conditions correlate well with the experimental data (see figure 3.6, 3.7).

3.3.3. Interactions through Water

The interaction through water at constant pH=4 is shown in ﬁgure 3.8. The obtained potential was 28 mV and the Debye length was 30.3 nm, which is in good agreement with the Debye length for an ionic strength 10−4 M, i.e. pH=4. The simulated potential is

56 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

GCMC_1mM_Imp_Dm_3.5_pH=3.88 0.2 GCMC_1mM_Imp_Dm_5_pH=3.88 NaCl_pH=3.88

0.1

0.0 F/R[mN/m]

-0.1

-0.2

0 20 40 60 80 100

Separation, D [nm]

Figure 3.7.: Monte Carlo simulation with an implicit surface charge description for the interaction curve between a pair of colloidal particles across NaCl aqueous electrolyte solution at a ﬁxed ionic strength of 1 mM and pH=4. The calculations were performed by Christophe Labbez.

57 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes assumed to be negative from zeta potential measurements. The interactions are repulsive at larger distances, but for distances smaller than 20 nm only attraction is observed.

0.1 9 8 water, pH=4, ö= -28 mV, ë=30.3 nm 0.1 7 6 5 0.01 4 ðR [mN/m] 3 F/2 0.001 2 0 10 20 30 Separation, D [nm]

0.01 9 8 7 6 ðR [mN/m] 5

F/2 4

0.001 0 20 40 60 80 100 Separation, D [nm]

Figure 3.8.: Forces between a pair of colloidal silica particles in milli-Q water at pH=4. Hamaker constant A= 8.5 × 10−21 J

As can be seen from table 3.1, the fitted diffuse layer potentials do not correlate with the measured zeta potentials. At pH=5.8 the zeta potentials are always larger than the fitted potentials for the studied ionic strengths.

3.4. Discussion

3.4.1. Eﬀect of Ionic Strength

Long Range Interaction

At pH=5.8 and an ionic strength of 10−4 M, repulsion is observed for all salts at separations larger than 10 nm due to the overlap of the double layer and the resulting osmotic

58 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes pressure of the counterions (see figure 3.1). The obtained Debye length of 30.3 nm coincides with the expected Debye length of 30 nm for an ionic strength of 10−4 M. The fitted potentials are similar for all the studied salts (see table 3.1). Some authors reported that the adsorption to the silica surface follows the Hofmeister series [13–15] Cs+ >K+ >Na+, since with increasing hydration shell the ions prefer to stay in water. No ion specificity was observed in this experiments at long range, but at 10−3 M a slight tendency for Cs+ to be adsorbed at the silica surface is observed (figure 3.2). The obtained Debye length of around 10 nm coincides with the expected theoretical value of 9.6 nm. The Debye length at 10−3 M is lower than at 10−4 M meaning that the range of the interaction decreases with increasing electrolyte concentration (see table 3.1). Repulsion is observed during the whole range at this electrolyte concentration. The simulated potentials do not agree with the zeta potentials. During the measurements the counterions are forced to con- dense or to bind to the surface as the distance decreases following a constant regulation interaction between the two surfaces. During zeta potential measurements the ions build a deformed ion cloud, which will result in less screening and a higher effective potential. Another reason may be the difference in preparation methods. For force measurements, the silica particles are subjected to plasma cleaning treatment for a couple of minutes. That may change the surface chemistry. For zeta potential measurements a suspension of silica particles was diluted in the desired electrolyte solution. Some authors [25] related the discrepancy between zeta and fitted potentials to the roughness of the spheres. They obtained larger values of zeta potentials compared to the values of fitted potentials at 10−3 M ionic strength, since under this condition the Debye length and the roughness are similar in magnitude. We can exclude the roughness as a reason for the discrepancy between zeta and diffuse layer potentials, because a disagreement between the two values is also observed at 10−4 M.

Short Range Interaction

The reason for short range interaction is controversially discussed in the literature. While many authors report that hydration forces or short range repulsive steric forces dominate van der Waals forces [4,5] others mention that van der Waals forces are measurable [2, 25, 28]. Figure 3.1 shows attractions at distances lower than 10 nm in the presence of all salts for 10−4 M and pH=5.8. The short range attraction decreases in the sequence Li+ >Na+ >K+ >Cs+. The inverse Hofmeister series is observed with Li+ adsorbing more to the negative silica surface than Cs+. The results correlate well with rheological studies performed by other authors [29]. They report the same adsorption sequence of ions to the silica surface in concentrated solutions and interpret their results according to the hypernetted chain model of Torrie et al. [10]. Chapel [2] cited a small van der Waals jump for LiCl at low ionic strength. Higashitani et al. [11] reported a strong adhesive force between a silica colloid and mica in electrolyte solutions of highly hydrated ions. According to the model of Higashitani et al. [11] the cation cesium is less hydrated and can be closely packed at the silica surface. The adsorbed layer composed of the poorly hydrated cesium cations will be thin but strongly adsorbed to the silica surface.

59 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

The adsorbed layer composed of hydrated lithium cations will be thick. The larger the hydration shell, the more unstable are the layers and they can be easily pressed out. The observed short range attraction might have several possible explanations: van der Waals forces, depletion forces, or steric forces. If the short range attractions are due to van der Waals forces, stronger attraction will be observed with increasing ionic strength due to electrostatic screening. The opposite occurs though and therefore it could be assumed that hydration layer expulsion may dominate the interactions. It might be possible that the Li+ ions are not closely packed to the silica surface due to its huge hydration shell so that they can easily be desorbed from the surfaces and be replaced by the protons from bulk solution; no hydration forces will then be observed. Less hydrated ions will be strongly bound to the surface and will remain there under the approach of the other surface giving rise to repulsive hydration forces. The statement was proposed by Pashley [12] to explain the short range interaction of mica surfaces in the presence of electrolyte solutions and is also consistent with Higashitani’s model [11](see figure 3.9 a). The fact that the attraction increases with decreasing ion size (or increasing hydration shell) may mean that at smaller distances the Li+ ions are partially expelled from the slit pore, which leads to the largest attraction due to depletion forces. With ionic strength increasing up to 10−3 M the attraction observed for Li+and Na+ is reduced (see figure 3.2), meaning that the hydration layer is more stable and repulsive hydration forces may dominate the interactions. Interestingly, the largest short range repulsion occurs for Li+ at this ionic strength. Maybe the ions were already partially dehydrated and more closely packed at the surface with increasing ionic strength. Now the adsorbed layer cannot be squeezed out from the silica surface (see figure 3.9 b). Dishon et al. [17] also observed additional short range repulsion at small distance in the force curves measured between a silica sphere and a silicon wafer at 10−3 M. Higashitani et al. [11] proved that the adhesion force decreases with increasing electrolyte concentration. A slight attraction was detected for Cs+ at 10−3 M ionic strength, which could be interpreted as van der Waals forces. Pashley [12] demonstrated that repulsive hydration forces are not present when the surface is rich in protons as counterions. The same result was obtained for the interaction between a pair of silica particles in milli-Q water at pH=4 (see figure 3.8), where attraction is only observed at short range. The results shown in figure 3.5 diverge from Pashley’s statement [12]. Although the force curve in the presence of NaCl was performed at pH=4, where the silica surface has to be covered with protons, the short range attraction (compared to the force curve of NaCl at pH=5.8; figure 3.1) disappears. Instead short range repulsion is observed. This repulsion may be due to a synergetic effect between protons and cations. A gel layer composed of polysilicic acid tails may be present at the silica surface [7] and influence the interactions. The thickness of this gel layer was estimated to be in the range of 1 nm to 4.4 nm [7, 30]. The counterions can penetrate the gel layer depending on their hydration shell [31] (see also table 3.2) and produce a collapse of the gel layer due to electrostatic screening [30]. Tadros et al. [31] argued that the diffuse layer potential of the silica surface will not be high if penetration of counterions inside the pores of the gel occurs. The last statement may also explain the difference in zeta and diffuse layer potential observed. It may be that

60 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

Silica a) Silica

Silica

b) Silica

Silica

Water

Figure 3.9.: Sketch of the adsorption of the cation lithium at a) 10−4 M before and after approaching and b) 10−3 M ionic strength before and after approaching

61 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes

Cation Bare radius(Å) Hydrated radius(Å) Li+ 0.60 3.82 Na+ 0.95 3.58 K+ 1.33 3.31 C+ 1.69 3.29 TMA+ 2.56-3.47 N/A

Table 3.2.: Size of the counterions, taken from [3] during the force measurements the ions are forced to penetrate the gel layer. Another argument supporting the theory of the gel layer is the soft contact wall obtained in some experimental curves (see ﬁgure 3.2). The aging of silica in water may increase the grow of the hair layer [32].

3.4.2. Eﬀect of pH

Long Range Interaction

Figure 3.5 shows the interaction curves between the silica surfaces across different electrolytes at pH=4. Repulsion is still present for NaCl which gave a negative surface potential of the silica surface at pH=5.8. It is assumed that the silica surface is still negatively charged. In case of Li+,K+ and Cs+ the silica surface is already neutralized and no repulsion occurs anymore. At pH=4 the silica surface is close to its isoelectric point, which leads to a reduction of long range electrostatic repulsion. The obtained Debye length of 21.4 nm coincides with the ideal one of 21.4 nm for 2 × 10−4 M ionic strength and is smaller than at pH=5.8 (see figure 3.1) due to the addition of protons to the solution. If the electrolyte concentration is increased to 10−3 M, the double layer collapse remains at pH=4 for all the electrolyte solutions (figure 3.6). Zeta potential experiments performed under the same conditions confirm the last statement.

Short Range Interaction

The pronounced attraction seen for NaCl at pH=5.8 (figure 3.1) vanishes at low pH (see figure 3.5). The repulsion may be due to the protons and cations adsorbed at the silica surface. If polysilicic acid chains are present at the silica surface, the presence of more protons will collapse the gel layer at the silica surface and no steric stabilization will be present. A slight decrease in pH causes attraction during the whole range due to the neutralization of the silica particles and the possible collapse of the polysilicic acid chains. Measurements in water (figure 3.8) show a large attraction at smaller distances. The attraction may be related to the collapse of the gel layer due to electrostatic screening and van der Waals forces. By increasing the electrolyte concentration to 10−3 M ionic

62 3. Force Measurements between Colloidal Particles across Aqueous Electrolytes strength, only attractive interactions are observed, due to the neutralization of the silica surfaces (see ﬁgure 3.6).

3.5. Conclusions

63 Bibliography

[1] D. F. Evans and H. Wennerström. The colloidal domain: Where Physics, Chemistry, Biology and Technology meet. VCH publishers, 1994. [2] J.-P. Chapel. In: Langmuir 10 (1994), pp. 4237–4243. [3] M. Colic et al. In: Langmuir 13 (1997), pp. 3129–3135. [4] J. J. Valle-Delgado et al. In: J. Chem. Phys. 123 (2005), pp. 034708–1. [5] S. Rentsch et al. In: Phys. Chem. Chem. Phys. 8 (2006), pp. 2531–2538. [6] J. J. Adler, Y. I. Rabinovich, and B. M. Mougdil. In: J. Colloid Interface Sci. 237 (2001), pp. 249–258. [7] G. Vigil et al. In: J. Colloid Interface Sci. 165 (1994), pp. 367–385. [8] B. V. Velamakanni et al. In: Langmuir 6 (1990), p. 1323. [9] J. Israelachvili. Intermolecular and Surfaces Forces. 2nd. Edition. Academic Press, 1991. [10] G. M. Torrie, P. G. Kusalik, and G. N. Patey. In: J. Chem. Phys. 91 (1989), p. 6367. [11] I. U. Vakarelski, K. Ishimura, and K. Higashitani. In: J. Colloid Interface Sci. 227 (2000), pp. 111–118. [12] R. M. Pashley. In: Journal of Colloid and Interface Sci. 83 (1981), pp. 531–546. [13] D. F. Parsons et al. In: Langmuir 26 (2010), pp. 3323–3328. [14] G. V. Franks. In: J. Colloid Interface Sci. 249 (2002), pp. 44–51. [15] M. Boström et al. In: Adv. Colloid Interface Sci. 123-126 (2006), pp. 5–15. [16] Israelachvili. J. N. and R. M. Pashley. In: Nature 306 (1983), p. 249. [17] M. Dishon, O. Zohar, and U. Sivan. In: Langmuir 25 (2009), pp. 2831–2836. [18] G. Toikka, R. A. Hayes, and J. Ralston. In: Langmuir 12 (1996), pp. 3783–3788. [19] P. West and A. Ross. An Introduction to Atomic Force Microscopy Modes. Santa Clara, CA: Paciﬁc Nanotechnology, Inc., 2006. [20] J. Ralston et al. In: Pure Appl. Chem. 77 (2005), pp. 2149–2170. [21] W. A. Ducker, T. J. Senden, and R. M. Pashley. In: Langmuir 8 (1992), pp. 1831– 1836. [22] McKee. DLVO ﬁtting. http://goo.gl/2Rh8c. June 2011.

64 Bibliography

[23] D. Y. C. Chan, R. M. Pashley, and L. R. White. In: J. Colloid Interface Sci 77 (1980), p. 283. [24] W. Briscoe. Colloid Science: Principles, methods and applications. Ed. by T Cos- grove. John Wiley and Sons Ltd, 2010, p. 343. [25] P. G. Hartley, I. Larson, and P. J. Scales. In: Langmuir 13 (1997), pp. 2207–2214. [26] G. Toikka and R. A. Hayes. In: J. Colloid Interface Sci. 191 (1997), pp. 102–109. [27] C. Labbez et al. In: Langmuir 25 (2009), pp. 7209–7213. [28] I. U. Vakarelski and K. Higashitani. In: J. Colloid Interface Sci 242 (2001), pp. 110– 120. [29] M. Colic, M. L. Fisher, and G. V. Franks. In: Langmuir 14 (1998), pp. 6107–6112. [30] A-C. Johnsson. “On the electrolyte induced Aggregation of Concentrated Silica Dispersions: An Experimental Investigation Using the Electrospray Technique”. PhD thesis. 2011. [31] J. Lyklema and Th. F. Tadros. In: J. Electroanalytical Chem. 17 (1968), pp. 267– 275. [32] M. Skarba. “Interactions of colloidal particles with simple elctrolytes and Polyelec- trolytes”. PhD thesis. 2008.

65 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions below the CMC

4.1. Introduction

The morphologies, functions, and applications of surfactants are diverse. There are natural surfactants such as the phospholipid protein, a pulmonary surfactant [1] used to reduce the surface tension in the lung. In addition synthetic surfactants are used in industry, including textile, cleaning products, cosmetics, food, and others. Depend- ing on the concentration the surfactant can act as a destabilizing or stabilizing agent. Therefore, it is important to understand the interaction forces between two surfaces in presence of the surfactant. Related to that, the morphology of the surfactant aggregates on the surfaces has a dominant effect. The morphology of the surfactant aggregates in the bulk depends on the critical packing parameter [2,3]. Therefore, different aggregate morphologies like spherical, cylindrical, globular, oblate micelles, single and multiwalled vesicles, microtubules, bilayers, lamellar phases, and inverted structures can be found depending on the size ratio between head group and hydrophobic tail of the surfactant [3]. It is known that hexadecyltrimethylammonium bromide (CTAB) forms spherical micelles in the bulk at the critical micelle concentration (CMC) [4–6]. Additional parameters, like surface charge, head groups and interaction between the hydrophobic tail, define the structure of the adsorbate on surfaces [4,7]. Several studies of the adsorption of surfactant to a hydrophilic surface, like mica and silica, have been performed [6–9]. Tyrode et al. [8] mention the two limit cases for adsorbing charged ionic surfactant at an oppositely charged hydrophilic surface below the CMC. For low surface charge systems, where the electrostatic interaction between head groups and surface is weak, no monomer adsorption at low concentration takes place. At a certain concentration, the critical surfactant aggregation concentration (csac), aggregates start to adsorbe to the surface. No monolayer formation is present in this system, since a monolayer with aliphatic chains facing towards the aqueous solution would be entropically unfavoured. Due to hydrophobic interactions additional surfactant molecules (with the hydrophilic groups facing out to the water) will start to adsorbe before monolayer coverage is reached. In case of high surface charge, monolayer formation is favoured due to strong electrostatic attraction between the surfactant head groups and the oppositely charged surface. Subramanian et al. [4] studied the effect of the counterion on the shape of the adsorbed aggregates on the

66 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions silica surface with AFM and demonstrated that it is possible to change the morphology of the CTAB from spherical to cylindrical by changing the counterion. With Cl− ions for example spherical micelles are obtained whereas with Br− ions cylinder aggregates on the silica surface are reported. Velegol et al. [9] obtained similar results studying the adsorbed layer at concentrations from 0.9 mM to 10 mM. Other authors [10, 11] reported the presence of micelles on the silica surface above the CMC studied with optical reflectometry and surface force apparatus respectively. Stiernstedt et al. [2] measured the surface force between silica particles across tetradecyltrimethylammonium bromide (TTAB) with a bimorph surface force apparatus. The obtained adsorbed layer thickness is 4 nm. Rutland et al. [11] could not conclude if a patchy bilayer or flattened micelles were present on the silica surface close to the CMC. They observed a strong dependency of the adsorption of CTAB molecules on the surface charge of the particles. Raman scattering and sum frequency spectroscopy showed that the thickness of the CTAB layer on silica particles is about 3 nm. Surface neutralization is obtained around 0.1 mM surfactant concentration. Neither of the two principal models mentioned in the literature ("The two step model" and "The four region model") can explain the adsorption to the silica surface [8]. The surface forces in the presence of surfactants have been investigated by several authors. Parker et al. [12] studied the interactions between glass particles with a surface force apparatus across CTAB solutions of different concentrations. They observed attractive interactions between two silica surfaces for distances at about 20 nm. The attractive forces could not be explained by van der Waals forces and they were present after the charge reversal and at higher concentrations. Several researchers try to explain the long range attraction observed between hydrophobic surfaces. Craig et al. [13] reported long range hydrophobic attraction of about 40 nm for cetylpyridimium chloride (CPC) in 100 mM NaCl adsorbed to silica surfaces. Carambassis et al. [14] associate the long range attraction between hydrophobic surfaces with the presence of bubbles. In a further work Craig et al. [15] obtained a slightly less attractive hydrophobic force for adsorbed CPC layers on the silica surface in 100 mM NaCl when the surfactant solution was degassed. Although the influence of dissolved gas on the hydrophobic interaction was suspected, they could not prove that nanobubbles may be responsible for the observed attractions. Kekicheff et al. [16] reported the correlation between the prefactor of the long range hydrophobic interaction for silica particles in CTAB solutions and the ionic strength of the solutions, supporting the hypothesis that the long range attraction between hydrophobic surfaces may have electrostatic origin. In the work of Yaminski et al. [17] microcavitation between the adsorbed patches of CTAB on the surface is considered to explain the large range hydrophobic interactions. Pashley et al. [18] studied the phenomenon of cavitation in CTAB monolayer. No evidence of cavitation was observed for this system since the contact angle was less than 90 degrees. So far, no systematic study of the interaction forces in the presence of cetyltrimethylam- monium bromide (CTAB) has been carried out using AFM and still the mechanism for the long range hydrophobic attraction present in those systems is debated. In this work the interaction forces between a pair of silica particles (system I) and between a silica

67 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions particle and a silicon wafer (system II) in CTAB solutions were measured over a large range, from well below the CMC to concentrations above the CMC. A prediction of the aggregates structure on the silicon oxide surfaces in dependence on surfactant concentration is established based on the qualitative and quantitative analysis of the interaction curves.

4.2. Experimental Section

4.2.1. Materials

A suspension of silica particles of 4.63 µm in mean diameter (10% solid content) was purchased from Bangs Laboratories. Solutions from cetyl trimethyl ammonium bromide (CTAB, analytic grade, Aldrich) were prepared in a concentration range from 0.005 mM to 1.2 mM in pure water (milli-Q). CTAB was water soluble up to 1.2 mM at room temperature. Clear solutions were obtained at all concentrations, which shows that the experiments were performed above the Kraﬀt temperature.

4.2.2. Preparation and Methods

Preparation

The attachment of silica particles to the glass slide and to the cantilevers were performed as described in section 3.2.2. The silicon wafers (type-P Wacker Siltronic Burghausen) were cut and cleaned in piranha solution H2O2/H2SO4 50:50 for 30 minutes, thereafter washed with milli-Q water, and then immediately used for the experiments. This method allows the creation of an oxide layer at the silicon wafer surface and renders the surface highly hydrophilic.

Methods

The force measurements between the two silica particles [19] were performed using a MFP-3D Asylum Research atomic force microscope mounted on an inverted optical microscope (Olympus IX71) (see section 3.2.2). The measurements performed with one pair of spheres are reported. The force measurements between a silica particle and an oxidized silicon wafer were performed with the same apparatus (Olympus IX71), instead of a glass slide with attached silica particles, a clean silicon wafer is placed on the scanner. A new clean silicon wafer was used for each concentration. At least 50 repetitions were done. After each measurement an optical microscope was used to check, if the particle was still attached to the cantilever. For the particle-particle system this check was performed during the measurements using an inverse microscope placed in the AFM.

68 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

The zeta potentials were measured using a Malvern Zetasizer Nano ZS with a HeNe laser. Diﬀerent solutions of CTAB from 0.005 to 1.2 mM containing 0.1 % silica particles were prepared. These suspensions were placed in a clear disposable zeta cell (DTS1060C) and equilibrated for two minutes at 25oC in the equipment before starting measurements [20]. The average of 10 zeta potential measurements of the same sample was taken as the zeta potential value.

4.2.3. Simulations

The simulations are based on the DLVO theory and were performed as described in section 3.2.3.

4.3. Results

4.3.1. Interaction forces between two silica particles (system I)

Figure 4.1 shows the interaction forces between two silica particles at a surfactant concentration from 0 to 0.1 mM. The pH of the solutions was around 5.8 where silica is negatively charged. The diameter of the particles was 4.63 µm. The fitting of the experimental curves to the DLVO theory provides the potential, ϕ and ionic strength, I . Each experimental curve was fitted with both boundary conditions, constant charge and constant potential. For reasons of clarity, in figure 4.1 only the fit for the experimental curve at 0.1 mM ionic strength is shown. The constant charge boundary condition fits very well at larger distances but for distances smaller than 20 nm, the experimental curve lies between the two boundary conditions. Neither the charge nor the potential remains constant. The respective potentials, ϕ and decay length, k−1 are shown in table 4.1. The repulsion decreases with increasing surfactant concentration from 0 to 0.05 mM. The interaction in water was repulsive for all distances. Below 5 nm a small attraction was observed which can be interpreted as a van der Waals force. At 0.005 mM repulsion was observed for the whole range of studied forces. The diffuse layer potential ϕ was decreased from -30 mV (for water) to -14 mV and the Debye length k−1 of 42.87 nm was similar to the Debye length of water and corresponds to an ionic strength I of 0.005 mM. In force measurements, only the value of the potential can be determined. Based on zeta potential measurements under the same conditions, the sign of the potential is inferred. A further increase of the CTAB concentration gave a weaker repulsion at separation >10 nm, and for smaller distances attraction was observed. At 0.03 mM surfactant concentration only weak repulsion was detected, and at 0.05 mM CTAB concentration no repulsion was observed at all, only attractive interaction with a jump–in distance around 20 nm. When the concentration was increased further to 0.1 mM repulsion occurred again. This leads to the conclusion that the point of zero charge (pzc) is between 0.05 and 0.1 mM CTAB concentration.

69 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

0.25

water, pH=5.8, ö= - 30 mV, ë=42.18 nm 0.005 mM 0.20 0.008 mM 0.01 mM 0.03 mM 0.05 mM 0.15 0.1 mM DLVO_CC DLVO_CP

ðR [mNIm] 0.10 F/2

0.05

0.00

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Separation [nm]

Figure 4.1.: Forces between a pair of colloidal silica particles (system I) across CTAB surfactant solution, from 0 to 0.1 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J. DLVO_CC (constant charge) and DLVO_CP (constant potential) ﬁts are shown for 0.1 mM surfactant concentration

In figure 4.2 the respective interaction forces are shown for a concentration regime from 0.1 to 0.5 mM. For a better understanding only the best fits to the experimental curves are shown. The constant charge boundary condition is fitting the experimental curve really well at 0.2 and 0.3 mM for distances larger than 10 nm where repulsion was observed. At smaller distances short range attraction dominates the interactions. At 0.4 mM, the constant potential fits better almost until contact. Repulsive interactions are seen for larger distances and at smaller distances only a small attraction is observed. At 0.5 mM the experimental curve lies between the constant charge and the constant potential boundary conditions for distances smaller than 20 nm. No short range attraction is seen anymore, the interactions are monotonic repulsive at this concentration. The diffuse layer potential ϕ increased from 0.1 mM to 0.4 mM surfactant concentration (see table 4.1). The Debye length k−1 correlates well with the theoretical one and was decreasing with increasing surfactant concentration, as expected. In the concentration regime from 0.5 to 1.2 mM CTAB (see figure 4.3) repulsion domi-

70 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

CTAB Diﬀuse layer Diﬀuse layer Experim. Zeta Concentr. potential potential decay len. potential (mM) system I ϕ (mV) system II ϕ (mV) k−1(nm) ζ (mV) 0.00 -30.0±0.0 -45.0±0.0 42.2 -55.0 0.005 -14.0±1.6 0.0 42.9 0.0 0.008 -15.0±0.0 0.0 42.9 0.0 0.01 -15.0±0.0 0.0 42.9 0.0 0.03 0.0 0.0 n.d. +21.3±4.2 0.05 0.0 0.0 n.d. +26.9±5.3 0.10 +36.0±0.0 0.0 30.0 +20.5±2.0 0.20 +48.0±0.0 0.0 21.4 +35.0±6.8 0.30 +48.6±2.8 0.0 17.5 +62.7±3.6 0.40 +72.2±6.2 +46.0±2.6 15.1 +58.8±12.4 0.50 +57.2 |+75.0∗∗ ± 7.5 +75.0 |+50.0∗ ± 5.0 13.6 +95.8±1.3 0.80 +38.2 |+60.0∗∗ ± 0.6 +158.0 |+80.0∗ ± 2.7 10.7 +108.0±7.1 1.00 +30.0 |+45.0∗∗ ± 0.0 +44.0 |+85.0∗∗ ± 5.3 8.8 +127.0±2.7 1.20 +30.0 |+45.0∗∗ ± 0.0 not measured 8.78 +111.0±3.9

Table 4.1.: Results for simulations of direct force measurements by the Poisson-Boltz- mann theory. ∗ From fitting with the plane of origin of charge taken at 4 nm from each surface. ∗∗ Assuming that the surfaces were not in contact (micelles/bilayer adsorption). nated the interaction over the whole range. The fits shown for 0.5 mM (figure 4.2), 0.8 mM and 1.0 mM (figure 4.3) were calculated assuming that the surfaces were in contact. The constant charge and constant potential boundary conditions fit well at larger distances. For distances smaller than 20 nm, the experimental curve lies between the two boundary conditions. No adhesion in the retraction curves (data not shown) was seen for concentrations larger than 0.5 mM. An unexpected decrease in the repulsion with increasing surfactant concentration was observed in this concentration regime. Therefore the DLVO fits were also calculated assuming that the surfaces were not in contact taking into account the adsorption of aggregates that cannot be removed from the surface by the applied force. Under this assumption, the constant charge boundary condition fits the experimental curves well until contact (data not shown). The fitted diffuse layer potentials are reported in table 4.1).

4.3.2. Interaction forces between a silica particle and a silicon wafer (system II)

In order to get information about the adsorption of surfactant at the silicon wafer, force curves were recorded for interactions between a silica particle and a silicon wafer (system II) for the same CTAB concentration range (0.005 to 1 mM). The interaction curves were

71 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

0.25 water, pH=5.8, 0.1 mM 0.2 mM 0.3 mM 0.20 0.4 mM 0.5 mM DLVO_CC DLVO_CP

0.15 ðR [mNIm]

F/2 0.10

0.05

0.00

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Separation [nm]

Figure 4.2.: Forces between a pair of colloidal silica particles (system I) across CTAB surfactant solutions from 0.1 to 0.5 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J. DLVO_CC (constant charge) and DLVO_CP (constant potential) ﬁts are shown for 0.2 mM (DLVO_CC), 0.3 mM (DLVO_CC - overlaps ﬁt at 0.2 mM), 0.4 (DLVO_CP) and 0.5 mM (DLVO_CC and DLVO_CP) surfactant concentration

fitted with the DLVO theory. The Derjaguin approximation is still valid for this system and since it is difficult to simulate two different surfaces, a symmetric system (two planar surfaces) was assumed. The obtained diffuse layer potential serves to show the changes in the system. The fitted diffuse layer potentials are reported in table 4.1. The force curves for two concentrations below 0.3 mM are shown in figure 4.4. Only attraction was observed for these concentrations and the jump–in distance was around 20 nm for 0.005 mM and 50 nm for 0.05 mM. The interactions at 0.1 mM and 0.2 mM (data not shown) remain attractive with a jump–in distance around 50 nm. The interactions between two silica particles (system I) at 0.05 mM (jump–in distance around 15 nm) are also represented in this figure. It can be seen that the attraction is larger in magnitude and range for system II. Figure 4.5 shows the force curves for concentrations ranging from 0.3 mM to 0.8 mM. At 0.3 mM CTAB concentration the observed attraction has a jump–in distance of about 43 nm and no repulsion occurs at long range. From 0.4 mM up to 0.8 mM repulsion was

72 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

0.25

water, pH=5.8 0.5 mM 0.20 0.8 mM 1 mM 1.2 mM DLVO_CC DLVO_CP 0.15 ðR [mNIm] 0.10 F/2

0.05

0.00

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Separation [nm]

Figure 4.3.: Forces between a pair of colloidal silica particles (system I) across CTAB surfactant solutions from 0.5 to 1.2 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J. DLVO_CC (constant charge) and DLVO_CP (constant potential) fits are shown for 0.8 and 1 mM surfactant concentration. observed for distances larger than 20 nm. The better DLVO fits are shown for 0.4, 0.5 and 0.8 mM surfactant concentration. At 0.4 mM the constant charge as well as the constant potential boundary condition fits the experimental curve very well at distances larger than 20 nm. The interactions for distances smaller than 20 nm were attractive. At 0.5 mM the better fit is obtained by the constant charge boundary condition. A non-DLVO repulsion was observed from 20 nm down to 10 nm followed by a plateau down to contact. At 0.8 mM also the constant charge boundary condition fits better and a non-DLVO repulsion from 20 nm down to 13 nm was observed followed by a plateau down to contact. It is important to note that the plateau did not occur between silica particles (see figure 4.3). Since a plateau around 10 nm was observed at 0.5 and 0.8 mM corresponding to the adsorption of aggregates on the surface, the DLVO fits at these concentrations were calculated with the plane of charge displaced by 4 nm from each surface (see table 4.1). The obtained fitted diffuse layer potentials for the unshifted plane of charge are shown in table 4.1. At 1 mM the interaction was monotonically repulsive (figure 4.6). The experimental curve was shifted by 8 nm, considering the presence of aggregates on the surface, which could not be removed by the applied force at this concentration. The original and the shifted curves are both shown with the corresponding fits, the obtained

73 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions diﬀuse layer potentials are given in table 4.1.

0.10

0.005 mM (silica particle-silicon wafer) 0.08 0.05 mM (silica particle-silicon wafer) 0.05 mM (silica particle-silica particle)

0.06

0.04

0.02 ðR [mNIm] F/2 0.00

-0.02

-0.04

-0.06 0 10 20 30 40 50 60 70 Separation, D [nm]

Figure 4.4.: Forces between a colloidal silica particle and a silicon wafer (system II) across CTAB surfactant solutions at 0.005 and 0.05 mM surfactant concentration. Forces between two colloidal silica particles (system I) at 0.05 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J.

4.3.3. Point of zero charge

From figure 4.1 one can conclude that the point of zero charge (pzc) for silica particles (system I) occurred at about 0.05 mM surfactant concentration. Since a symmetric system is measured, the sign of the potential can also be inferred from the trend of the force curves. The interactions are changing with concentration, from repulsion at low surfactant concentration to attraction at 0.05 mM (pzc) to repulsion again at higher concentrations. Silica particles are negatively charged in water at normal pH [21]. Therefore, a negative sign is assumed for surfactant concentrations below 0.05 mM and a positive charge for concentrations above 0.05 mM. The dependence of the charge of the silica on surfactant concentration and the point of zero charge was confirmed by zeta potential measurements. The zeta potentials and the fitted diffuse layer potentials are shown in table 4.1. For surfactant concentrations below 0.03 mM the zeta potentials values were low and no stable values were obtained. From 0.03 mM onwards, positive values of zeta potentials were obtained, meaning that the pzc in zeta potential measurements

74 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

0.6

0.5 0.3 mM 0.4 mM 0.5 mM 0.8 mM DLVO_CC 0.4 DLVO_CP

0.3

ðR [mNIm] 0.2 F/2

0.1

0.0

-0.1

0 10 20 30 40 50 60 70 Separation, D [nm]

Figure 4.5.: Forces between a colloidal silica particle and a silicon wafer (system II) across CTAB surfactant solutions from 0.3 to 0.8 mM surfactant concentration. Hamaker constant A= 8.5 × 10−21 J. DLVO_CC (constant charge) and DLVO_CP (constant potential) fits are shown for 0.4 mM surfactant concentration. For the DLVO_CC fits shown at 0.5 mM and 0.8 mM the plane of charge was set 4 nm away from each surface. was around this concentration. The zeta potential increased with increasing surfactant concentration indicating an increase in adsorbed amount of surfactant. For system I the fitted diffuse layer and the zeta potentials show the same tendency at concentrations below 0.4 mM. Both are increasing with surfactant concentration. At concentrations above 0.4 mM the zeta potential continuously increased with concentration, as expected, whereas the fitted diffuse layer potential shows an "apparent" decrease. From figure 4.5 the point of zero charge for system II can be deduced. It occurred at about 0.3 mM surfactant concentration.

75 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

0.6

0.5 1 mM 1 mM (offset 8 nm) DLVO_CC1 0.4 DLVO_CP1 DLVO_CC2 DLVO_CP2 0.3

ðR [mNIm] 0.2 F/2

0.1

0.0

-0.1

0 20 40 60 Separation, D [nm]

Figure 4.6.: Forces between a colloidal silica particle and a silicon wafer (system II) across 1 mM CTAB surfactant solution. Hamaker constant A= 8.5 × 10−21 J. The experimental curve was oﬀset 8 nm under the assumption that micelles/patchy bilayers are adsorbed at the surface. DLVO_CC (constant charge) and DLVO_CP (constant potential) ﬁts are shown for the experimental curve (DLVO_CC1, DLVO_CP1) and the shifted curve (DLVO_CC2, DLVO_CP2).

4.4. Discussion

4.4.1. Interaction between two silica particles (system I)

The fitted potentials are called "effective potentials" because although the DLVO theory is applied for the analysis, there are other non DLVO forces which influence the interactions, like hydrophobic, hydration and steric forces. From zeta potential measurements it is known that the silica particles have a negative surface potential in water. The addition of 0.005 mM CTAB to water causes a decrease in the diffuse layer potential from ϕ=-30 mV to ϕ=-14 mV (see figure 4.1). The interaction curve is still repulsive. The potential is decreased due to the adsorption of positively charged surfactant monomers to the silica particles. The attraction observed at short range for the interaction in water was overcome at this surfactant concentration and short range repulsion was observed. A correlation between the measured interaction curves and the possible surfactant struc-

76 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions tures on the silica surface can be established. From the quantitative analysis of the interaction curve (the fitting to the DLVO theory) it is known that adsorption occurred at 0.005 mM because the diffuse layer potential is decreased with respect to the diffuse layer potential for the interactions in water. Based on the qualitative analysis of the force curve at this concentration a parallel arrangement is the most probable (see figure 4.7). If the surfactant molecules were arranged perpendicular to the surface, attraction would be observed at short range due to hydrophobic interactions. Since short range repulsion was observed, it is more probable that the surfactant monomers were arranged in parallel to the surface and that the observed short range repulsion was due to the dehydration of the ammonium head groups. At this concentration the hydration forces overcome the hydrophobic forces and/or van der Waals forces (see figure 4.7). Other authors [8] reported a perpendicular arrangement of the surfactant molecules with the hydrophobic tails facing towards water at low surfactant concentration. Increasing the concentration further from 0.008 mM to 0.05 mM (see figure 4.1) turned the interactions from repulsive to attractive. More and more surfactant molecules interact electrostatically with the silica surface. That is the most probable interaction because the silica particles are hydrophilic and negatively charged and the CTAB head group is positively charged. At 0.05 mM only attraction was observed with a jump–in distance at around 20 nm. The range is larger than for pure van der Waals forces, which indicates that additional interactions (like hydrophobic ones) come into play. The hydrophobic interactions observed in the force curves suggest that the most probable arrangement of surfactant aggregates is perpendicular to the surface (see figure 4.7). The point of zero charge (pzc) was obtained at this concentration. Parker et al. [12] also observed neutralization of the silica surface at this concentration. Although the range of the interaction was similar, the magnitude was larger in Parker’s work. Yaminski et al. [17] reported the point of zero charge for silica at the same concentration and argued that below this concentration the formation of patches on the surface is possible, since the energy gain is larger for patches than for isolated molecules. A further increase in the surfactant concentration to 0.1 mM resulted in a repulsive long range interaction meaning that a charge reversal from negative to positive occurred. The obtained potential was +36 mV. The sign of the potential can be inferred from the trend of the force curves and was also confirmed by zeta potential measurements. The interaction cannot be fitted by the constant charge/potential model for distances smaller than 20 nm indicating that at smaller distances there is a variation of both, the surface charge and the surface potential. At distances <10 nm attraction was observed, which may be due to the expulsion of aggregates present on both surfaces. Near to contact, for distances <2 nm, repulsion was observed again which may be steric in origin. The charge reversal of the surface is associated to the adsorption of surfactant molecules with the hydrophilic head facing out to the water. The aggregates can easily be removed during the force measurements (see figure 4.7). The repulsion continued increasing with increasing surfactant concentration (figure 4.2). The effective potentials are shown in table 4.1. At 0.2 and 0.3 mM the constant charge boundary condition matches the experimental curves really well at larger distances. For distances smaller than 10 nm no fit is seen at all because the interactions

77 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions are not DLVO like and correspond to the expulsion of aggregates from the surface. At 0.3 mM the attraction is weak and starts to be replaced by a soft short range repulsion. At 0.4 mM surfactant concentration, the experimental curve is best fitted with the constant potential boundary condition until smaller distances. A potential of 72 mV was obtained. The change in short range interaction from attraction at 0.2 mM to soft repulsion at 0.3 mM finishing with a strong repulsion at 0.4 mM can be correlated with the stiffness of the aggregates adsorbed to the surface. The aggregates are closer and stiffer with increasing concentration (see figure 4.7). A further increase in concentration to 0.5 mM led to a decrease in the diffuse layer potential to +57 mV (see table 4.1). The diffuse layer potential decreased until the critical micelle concentration was reached where it stayed constant (see figure 4.3). This decrease is unexpected. The experiments were repeated with another pair of silica particles show- ing the same tendency (data not shown). Interestingly, the interaction curves cannot be fitted by the constant charge/potential model at distances smaller than 20 nm. The fits were corrected for concentrations from 0.5 mM onwards, under the assumption that the surfaces were not in contact due to the adsorption of micelles/patchy bilayers on the surfaces. Still a decrease in diffuse layer potential is observed from 0.5 mM onwards (see table 4.1). That is explained by condensation of Br− counterions to the highly charged surfaces (surfaces with a denser outer layer of surfactant) during the force measurements. Hence lower diffuse layer potentials are obtained. In contrast the zeta potential increases with increasing surfactant concentration (see table 4.1), as expected. The interactions are monotonic repulsive in this concentration regime (from 0.5 mM to 1 mM) indicating that the aggregates are stiffer with increasing concentration.

4.4.2. Interaction between a silica particle and a silicon wafer (system II)

The interaction of a silica particle with a silicon wafer is represented in figures 4.4, 4.5 and 4.6. At concentrations as low as 0.005 mM a large attraction was observed. This could be caused by aggregate patches of CTAB with the hydrophobic tails facing out to the water present on the surface of the silicon wafer (see figure 4.7). At 0.05 mM the interaction between a silica particle and an oxidized silicon wafer still remained attractive. At this concentration the silica particle was neutralized, that means that the observed attraction has to be produced by hydrophobic interactions between the two surfaces. Therefore, it is probable that the surfactant was adsorbed at the silicon wafer surface with the hydrophobic tails facing out to the water, as shown in figure 4.7. The long range attraction occurred up to 0.3 mM (see figure 4.5). It is known from force and zeta potential measurements that at this concentration the silica particle was positively charged. Hence the observed attraction may be due to electrostatic interaction between the already positive silica particle (charge reversal at 0.05 mM) and the bare silicon wafer (the areas without surfactant adsorption). Therefore, it is proposed that the surfactant adsorbed patchwise to the silicon wafer surface (see figure 4.7). The validity of this statement is confirmed later in this paper (see figure 4.9).

78 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

Silica particle-silica particle interaction (system I)

0.005 mM 0.05 mM (pzc) 0.1 mM 0.4 mM 1 mM

Silica particle-silicon wafer interaction (system II)

0.005 mM 0.05 mM 0.3 mM (pzc) 0.4 mM 1 mM

Figure 4.7.: Possible surfactant morphologies depending on the concentration.

Figure 4.5 shows the interactions between a silica particle and a silicon wafer from 0.3 to 0.8 mM. The point of zero charge for this system was 0.3 mM. At 0.4 mM a long range double layer repulsion was present. The short range attraction started at around 17 nm and is non-DLVO. The surfactant starts to be adsorbed with the hydrophilic tails facing out to the water (see figure 4.7). At 0.5 mM the short range interaction occurred at around 8 nm which corresponds to the expulsion of micelles/patchy bilayers from the two opposing surfaces. At 1 mM (CMC) complete micelles/patchy bilayers were present on the silica/silicon wafer surfaces, and they could not be removed under the applied force. Two DLVO fits were done. The results are shown in table 4.1. The experimental curve can be fitted well with a diffuse layer potential of 44 mV, assuming that the surfaces (silica/silicon wafer) are in contact(D = 0) or with a diffuse layer potential of 85 mV assuming that the surfaces (silica/silicon wafer) are separated by 8 nm due to the adsorption of micelles/patchy bilayers. The non-DLVO repulsion observed in the shifted curve can be explained by steric forces between the micelles/patchy bilayers (see figure 4.6).

79 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

4.4.3. Comparison between the system silica particle–silica particle (I) and the system silica particle–silicon wafer (II)

A comparison of the interactions in the two systems gives the following results. At low CTAB concentration (e.g. 0.05 mM) the attraction is larger for system II. Since the attractions were due to hydrophobic interactions, the silicon wafer was more hydrophobic than the silica particles. Consequently the hydrophobic patches were larger at the silicon wafer (see ﬁgure 4.4). At 0.4 mM surfactant concentration (ﬁgure 4.8) the Debye

0.4

0.3 0.4 mM (silica particle-silicon wafer) , ö=+50mV, ë=15.2nm 0.4mM (silica particle-silica particle), ö=+75mV, ë=15.2nm

0.2 R [mNIm] ð F/2

0.1

0.0

0 10 20 30 40 50 60 70 Separation, D [nm]

Figure 4.8.: Interaction forces between two silica particles (system I) and between a particle and a silicon wafer (system II) at 0.4 mM surfactant concentration. The Debye length is 15.2 nm for both cases. length correlates well with the theoretical one for both systems but the interaction looks diﬀerent. A greater repulsion was obtained for system I compared to system II. That indicates that the outer layer of surfactant was more dense at the silica particle surface. The qualitative analysis of both systems from 0.4 mM onwards led to the conclusion that it was not possible to remove the aggregates from the silica particles under the applied force in system I whereas aggregate expulsion was seen for system II up to 0.8 mM. The aggregates at the silica particles must be stiﬀer and more closely packed than on the silicon wafer surface from 0.4 mM onwards. In general, the aggregate morphology at the

80 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions silica particles was different than at the silicon wafer, which was verified with the different force curves obtained for the same concentration. The silica particles and the silicon wafer have the same surface chemistry, but the surface treatment has a great influence on the surface charge and hence the type of aggregates on the surface. Other authors confirm that different preparation methods cause differences in surface charge [22]. The point of zero charge (pzc) for system I was obtained at 0.05 mM surfactant concentration and for system II at 0.3 mM. That indicates that the silicon wafer carries a larger surface charge.

4.4.4. Non DLVO forces

Non-DLVO forces were observed in system I and II at 0.05 mM. The jump–in distance for system II was about 43 nm which is larger than that for system I. The bridging mechanisms through nanobubbles is an explanation for the hydrophobic forces observed at this concentration. In previous experiments performed in our group nanobubbles were found on the silicon oxide surface at 0.05 mM surfactant concentration (see chapter5). When two surfaces with nanobubbles are approaching, a thin free–standing lamella is formed between the bubbles. When it breaks the bubbles will bridge and a jump-in contact will be observed in the force curves. At 0.3 mM a large attraction was observed for system II. It was proposed that the surfactant is adsorbed patchwise and that electrostatic interactions between the already positive silica particle and the still negative bare areas at the silicon wafer also play a role in the observed attraction. An AFM image of the silicon wafer at this concentration (see ﬁgure 4.9) shows that two diﬀerent morphologies were present on the surface, so-called "micropancake" (thin layers of air) with some nanobubbles on the top and some areas without surfactant adsorption. The observed morphology is similar to that in reference [24]. It indicates that the adsorption at the silicon wafer occurs patchwise at low surfactant concentration. Patchwise adsorption is also reported at mica surfaces in the presence of low concentration of C16TAB [25] and at silica surfaces in the presence of low concentration of C18TACl [26]. Micropancakes as well as nanobubbles may be covered with surfactant. Attractive electrostatic interaction between the silica particle (positive) and the silicon wafer (negative areas without adsorption) as well as the bridging through nanobubbles can explain the attractive interaction observed at this concentration. Non DLVO forces due to expulsion of aggregates or steric forces between micelles/patchy bilayers were also observed at higher concentrations.

81 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions

300 micropancake

250 4 200 3

mn 150 mn 2 100 1 50 0 0 300250200150100500 nm nanobubble

thgieH 1 0 -1 -2nm

0 50 100 150 200 250 300 nm

Figure 4.9.: AFM tapping mode of a silicon wafer at 0.3 mM surfactant concentration.

4.5. Conclusions

The interaction forces between two silica particles (system I) and between a silica particle and a silicon wafer (system II) in the presence of aqueous CTAB solutions with concentrations between 0.005 and 1 mM were measured using AFM. The force curves were correlated to the surfactant morphologies (figure 4.7). Both systems show a charge reversal from negative to positive caused by the adsorption of cationic surfactant (CTAB) at the former negatively charged silicon oxide surface. The interactions of the two systems were different for the same studied surfactant concentration. The point of zero charge was obtained at 0.05 mM for the silica particle–silica particle system (system I) as in Parker’s work [12] and at 0.3 mM for the silica particle–silicon wafer system (system II). This leads to different aggregate morphologies at the silica particle surface and at the surface of the silicon wafer. An explanation for the difference might be the surface treatment: the silica particles were plasma cleaned, whereas the silicon wafers were treated with a piranha solution. In another study the same surface treatment was applied to the silica particles and the silicon wafers and no differences in the interaction were observed [27]. At a low CTAB concentration (for example 0.05 mM) long-range attraction was observed. The attraction was larger in range and magnitude for the silica particle–silicon wafer system and starts at distances larger than 40 nm. They cannot be caused by van der Waals attraction, but they are explained by the presence of nanobubbles probing hy-

82 4. Interaction Forces between Silica Surfaces in Cationic Surfactant Solutions drophobic patches on the surfaces. The attraction occurs when the nanobubbles bridge. Obviously, on a silicon wafer surface larger hydrophobic patches are present than on the surface of the silica particles. Another explanation for long-range attraction is the electrostatic attraction between oppositely charged patches. At higher surfactant concentration (0.4 mM onwards), monotonic repulsion between the two silica particles was observed. In contrast to this, aggregate expulsion could be observed in the interaction curves for the silica particle–silicon wafer system up to a CTAB concentration of 0.8 mM. The difference is explained by the different stiffness of the surfactant aggregates at the two surfaces (silica particle and silicon wafer). For the same studied concentrations, the outer layer of surfactant was denser for the silica particles and the aggregates on the silica particles were stiffer and more closely packed. The stiffer aggregates are more difficult to remove.

83 Bibliography

[1] J. A. Zasadzinski et al. In: Biophys. J 89 (2005), pp. 1621–1629. [2] J. Stiernstedt et al. In: Langmuir 21 (2005), pp. 1875–1883. [3] D. F. Evans and B. W. Ninhman. In: J. Phys. Chem. 90 (1986), pp. 226–234. [4] V. Subramanian and W. A. Ducker. In: Langmuir 16 (2000), pp. 4447–4454. [5] S. Berr, R. R. M. Jones, and J. S. Johnson Jr. In: J. Phys. Chem. 96 (1992), pp. 5611–5614. [6] H. N. Patrick et al. In: Langmuir 15 (1999), pp. 1685–1692. [7] S. C. Biswas and D. K. Chattoraj. In: J. Colloid Interface Sci. 205 (1998), pp. 12– 20. [8] E. Tyrode, M. W. Rutland, and C. D. Bain. In: J. Am. Chem. Soc. 130 (2008), pp. 17434–17445. [9] S. B. Velegol et al. In: Langmuir 16 (2000), pp. 2548–2556. [10] R. Atkin, V. S. J. Craig, and S. Bigg. In: Langmuir 16 (2000), pp. 9374–9380. [11] M. W. Rutland and J. L. Parker. In: Langmuir 10 (1994), pp. 1110–1121. [12] J. L. Parker, V. V. Yaminski, and P. M. Claeson. In: J. Phys. Chem. 97 (), pp. 7706– 7710. [13] V. S. J. Craig, B. W. Ninham, and R. M. Pashley. In: Langmuir 14 (1998), pp. 3326– 3332. [14] A. Carambassis et al. In: Phys. Rev. Lett. 80 (1998), pp. 5367–5360. [15] V. S. J. Craig, B. W. Ninham, and R. M. Pashley. In: Langmuir 15 (1999), pp. 1562– 1569. [16] P. Kekicheﬀ and O. Spalla. In: Phys. Rev. Lett. 75 (1995), pp. 1851–1854. [17] V. V. Yaminski et al. In: Langmuir 12 (1996), pp. 1936–1943. [18] R. M. Pashley et al. In: Science 229 (1985), pp. 1088–1089. [19] G. Toikka, R. A. Hayes, and J. Ralston. In: Langmuir 12 (1996), pp. 3783–3788. [20] Zetasizer Nano Series User Manual. url: http://www.biophysics.bioc.cam. ac.uk/files/Zetasizer_Nano_user_manual_Man0317-1.1.pdf. [21] Y. Zeng. “Structuring of colloidal dispersion in slit-pore conﬁnement”. PhD thesis. 2011.

84 Bibliography

[22] J. Nawrocki. In: J. of Chromatography A 779 (1997), pp. 29 –71. [23] L. A. C. Lüderitz and R. von Klitzing. In: Langmuir 28 (2012), pp. 3360–3368. [24] X. H. Zhang et al. In: Langmuir 23 (2007), pp. 1778–1783. [25] B. G. Sharma, S. Basu, and M. M. Sharma. In: Langmuir 12 (1996), pp. 6506–6512. [26] J. Zhang et al. In: Langmuir 21 (2005), pp. 5831–5841. [27] M. Dishon, O. Zohar, and U. Sivan. In: Langmuir 25 (2009), pp. 2831–2836.

85 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

5.1. Introduction

Surfactants can be used as stabilizers/emulsifiers (above the CMC) in the cosmetic industry. Below the CMC, surfactants find applications in flotation processes since they can be adsorbed on a hydrophilic surface rendering the surface hydrophobic [1]. The surfactant morphologies found at the surface are a function of the surface charge, the type of head groups, and the hydrophobic part of the surfactant [2,3]. Many studies have been performed to clarify the structure of surfactants at the surface of mica or silica surfaces above the CMC. At the silica surface, micelles or flattened bilayers of CTAB close to the CMC have been reported [4]. Velegol et al. [5] described the CTAB adsorbed layer at the silica surface in the presence of 0.9×CMC and 10×CMC solutions. At 0.9×CMC surfactant concentration a coexistence of spheres and short rods was observed at the silica surface whereas wormlike micelles were observed at 10×CMC surfactant concentration. In some cases, a transition from the wormlike micelles to a laterally homogeneous structure (interpreted as a bilayer) similar to that observed on mica occurred. Ducker et al. [6] studied the adsorption of CTAB on mica at a concentration of 2×CMC. They obtained a flat sheet CTAB morphology in the absence of salt at the mica surface. Sharma et al. [7] reported that the adsorption of CTAB on mica at low concentration 10−5 M occurs patchwise. The distances between the patches was not constant, and a coexistence between patches of different heights was also observed, which was interpreted as surfactant molecules or aggregates. An increase of the concentration produced more closely packed surfactant patches. At 10−3 M surfactant concentration, a continuous wormlike admicellar structure with reduced separation compared to previous concentrations was observed at the mica surface. A further increase of the concentration to 10−2 M produced a continuous bilayer structure at the mica surface. They demonstrated that the variation of pH with the consequent variation in surface charge density influences the structure of the adsorbed micelles. The surface charge can also be varied by surface treatment [8]. Different substrates may also have different surface charges. The p.z.c for the adsorption of CTAB depends on the surface used: at the silica surface, 5 × 10−5 M was obtained, whereas the neutralization of the mica surface occurred at a lower concentration of 3.5 × 10−6 M[9]. Yaminski et al. [10] studied the adsorption of

86 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

CTAB to a silica surface in the presence of sodium acetate using a surface force apparatus. They reported a pronounced attraction between surfaces when the CTAB concentration is increased to 5 × 10−5 M. The attraction is explained by so-called hydrophobic interactions. Hence, one mechanism to explain the hydrophobic interactions is through bridging of nanobubbles, which are present at the opposing hydrophobic surfaces [11–15]. The presence of nanobubbles corresponds to a reduced density of water, which was detected at hydrophobic surfaces by neutron reflectometry [16]. It is known from the literature that the nanobubbles are stable for several days at hydrophobic interfaces and that they are present in solutions saturated with gas [17]. So far, nanobubbles have been studied on surfaces hydrophobized by chemical pretreatment (HOPG [18], OTS [18, 19]). The liquid phase was either water or surfactant solutions like SDS or CTAB [18]. Recently, nanobubbles were imaged on ultraflat gold covered with binary self-assembled monolayers (SAMs) with variable hydrophilic/hydrophobic balance [20]. Still, nanobubbles were found at the surface of SAMs with a macroscopic contact angle of 15◦, but they were very tiny. Ducker [17] explains the stability of nanobubbles at a hydrophobic surface by surface active contaminants. Under this assumption, the adsorption of surface active material to the nanobubble avoids the diffusion of gas out of the nanobubbles. The surface active material will stabilize the bubbles through creation of a diffusional barrier. The surface active contaminant may be adsorbed at the solid–liquid interface with the corresponding decrease of the solid–liquid interfacial tension. In that case, the liquid–vapor interfacial tension has to become extremely small to fit the observed low nanobubble contact angle (θ ≈ 16◦), which leads to a flattening of the nanobubbles [17]. Zhang et al. [18] studied the nanobubbles in the presence of two different surfactants, hexadecyl trimethyl ammonium bromide (CTAB) and (SDS) sodium dodecylsulfate at 0.5×CMC. Little or no variation in contact angle was observed for the nanobubbles present at HOPG or OTS surfaces when surfactant was added to the solution. This was later explained by the fact that the nanobubbles were already covered with some kind of surface active material (contaminants) so that the effect of surfactant on the nanobubbles was not seen [17]. A further proof for stabilization of nanobubbles by contaminants was their decreased stability in surfactant solutions well above the CMC, where the contamination is solubilized within the micelles. The bubbles on a graphite surface disappeared after 15 minutes exposure to 5×CMC SDS solution. A systematic investigation of the influence of the surfactant concentration on the stability of nanobubbles at a solid–liquid interface is still missing. Another open question is the hydrophilic/hydrophobic balance of silicon oxide surfaces at low CTAB concentration (below 0.5 mM). Therefore, in the present paper, hydrophilic silicon oxide surfaces are studied against aqueous solutions of a broad CTAB concentration regime (0.05–0.8 mM). In this regime, the silicon oxide surface is partially hydrophobic and nanobubbles may be present at the surface. Hence the surfactant fulfills two tasks: (1) modification of the silicon oxide surface via physisorption and (2) the stabilization of the nanobubbles. The nanobubbles are studied via Scanning Force Microscopy (SFM) and the results are correlated with the interactions between a silica microsphere and a planar silicon oxide surface against aqueous CTAB solutions in a Colloidal Probe AFM.

87 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

5.2. Experimental Section

5.2.1. Materials

Solutions from cetyl trimethyl ammonium bromide (CTAB, analytic grade, Aldrich, pu- rity > 99%) were prepared in a range of concentrations, 0.05 mM to 0.8 mM in pure water (milli-Q). CTAB was used as received. Surface tension measurements were made in the range of the studied concentration at 298 K with a tensiometer Krüss K11 using the ring method. The CMC of CTAB obtained was 1 mM, no minima was detected around the CMC, which indicates that no surface active contaminants were present in the CTAB used. The value of surface tension at 1 mM CMC of 36.2 mN/m correlates well with the literature value [21]. Nonporous silica particles, 4.63 µm in diameter, were used for the force measurements.

5.2.2. Preparation and Methods

Preparation

The silicon wafers (type-P Wacker Siltronic Burghausen) were prepared as described in section 4.2.2.

Methods

The scanning was performed using a MFP-3D Asylum Research atomic force microscope (AFM). The images of the spherical features in liquid were obtained in iDrive tapping mode using iDrive compatible cantilevers from Asylum Research. These were gold coated with a nominal spring constant of 0.09 N/m. iDrive is a patented technique that uses Lorentz force to magnetically actuate a cantilever with an oscillating current that flows through the legs [22]. This technique is recommended for imaging of extremely soft matter in liquid. The setpoint was adjusted to minimize the force on the sample while still tracking the surface. The tip curvature radius measured by SEM is Rtip = 15 nm ± 5 nm. A new experiment was performed for each concentration. Further, the images were flattened with a first or second order polynomial fit. The cross section of the observed spherical features was fitted to an arc of a circle, which allows the determination of the bubble parameters (see references [19, 23, 24]). At least 10 spherical features were analyzed (when possible) to obtain the parameters. The roughness (RMS) was obtained from a 300×300 nm image. The force measurements between a silica particle and an oxidized silicon wafer were also performed with a MFP-3D Asylum Research AFM mounted on an inverted optical microscope (Olympus IX71) (see section 4.2.2). The contact angles were measured with a Goniometer (OCA5) using the Sessile drop method. The clean silicon wafers were placed

88 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC on the sample stage. A syringe was used to place a drop of the surfactant solution on the silicon wafer surface. The software SCA 20 determined the contact angle using the Young–Laplace ﬁtting method [25].

5.2.3. Simulations

The DLVO theory is used to simulate the experimental curves (see section 3.2.3 for further details).

5.3. Results

Scanning of a Silicon Wafer from 0.05 to 1 mM CTAB

Figure 5.1 represents the surface of a silicon oxide surface in milli-Q water. A homogeneous surface is observed in water with a surface roughness of 0.43 nm. The contact angle of the cleaned silicon wafer is close to 0◦.

300

250 2.0 200 1.5 150 nm 1.0 nm 100 0.5 50 0.0 0 0 100 200 300 nm

Figure 5.1.: AFM tapping mode of a silicon oxide surface in water. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.265 V, setpoint ratio: 0.26, scan rate: 0.5 Hz, drive frequency: 6.91 KHz

89 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

When the concentration of surfactant is increased to 0.05 mM spherical features (figure 5.2), which resemble nanobubbles, are observed at the surface. At this concentration the features diameter is 50–80 nm. In the phase image, the spherical features are also recognized since a phase drop is observed along the features indicating the different nature of features and substrate. At 0.3 mM the features are still spherical, and the diameter is around 57 nm (figure 5.3). The spherical features dominate the surface topology. From this concentration onwards a slight phase shift between substrate and feature is observed in the phase image. At 0.4 mM surfactant concentration, a few spherical features could still be found at the surface (see figure 5.4). Since the observed features resemble a spherical cap, their parameters can be obtained by fitting a cross section to an arc of a circle (see table 5.1). Figures 5.5 and 5.6 represent the adsorbed surfactant layer at concentrations 0.5 and 0.8 mM. For 0.5 mM other types of features were found at the surface. They are flatter (height about 2 nm) than the features at lower CTAB concentration. At 0.8 mM, the roughness decreases to 0.27 ± 0.03 nm, which is lower than the original silicon wafer roughness (0.43 nm), and no features could be identified.

90 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

300

250

200 15

150 10 nm nm 100 5 0 50 -5 0 0 100 200 300 nm

10nm 5

Height 0 -5

0 50 100 150 200 250 300 nm

15° 10 5

Phase 0 -5

0 50 100 150 200 250 300 nm

Figure 5.2.: AFM tapping mode of a silicon oxide surface at 0.05 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.430 V, setpoint ratio: 0.43, scan rate: 0.5 Hz, drive frequency: 6.34 KHz

91 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

300

250 8 200 6 150 nm 4

100 2 nm 0 50 -2 0 0 100 200 300 nm

6nm 4 2 Height 0 -2

0 50 100 150 200 250 300 nm

2°

0 Phase -2

0 50 100 150 200 250 300 nm

Figure 5.3.: AFM tapping mode of a silicon oxide surface at 0.3 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.190 V, setpoint ratio: 0.19, scan rate: 0.5 Hz, drive frequency: 6.16 KHz

92 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

300

250

200 4

150 3 nm

2 nm 100 1 50 0 0 0 100 200 300 nm

6nm 4 2

Height 0 -2

0 50 100 150 200 250 300 nm

2 1 0 -1 Phase -2 -3°

0 50 100 150 200 250 300 nm

Figure 5.4.: AFM tapping mode of a silicon oxide surface at 0.4 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.215 V, setpoint ratio: 0.22, scan rate: 0.5 Hz, drive frequency: 6.03 KHz

93 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

300

250 400 200 200 150 nm 0 pm 100 -200

50 -400

0 0 100 200 300 nm

800pm 600 400 200 0 Height -200

0 50 100 150 200 250 300 nm

Figure 5.5.: AFM tapping mode of a silicon oxide surface at 0.5 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.234 V, setpoint ratio: 0.23, scan rate: 0.5 Hz, drive frequency: 7.1 KHz

94 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

300

250 600 200 400 200 150 nm 0 pm 100 -200 -400 50 -600 0 0 100 200 300 nm

200 0

Height -200pm

0 50 100 150 200 250 300 nm

Figure 5.6.: AFM tapping mode of a silicon oxide surface at 0.8 mM CTAB concentration. AFM images taken with a magnetic actuated cantilever; nominal spring constant: 0.09 N/m, amplitude setpoint: 0.210 V, setpoint ratio: 0.21, scan rate: 0.5 Hz, drive frequency: 7.1 KHz

5.4. Discussion

5.4.1. Nanobubbles

No features are observed at the surface of a cleaned hydrophilic silicon wafer. That correlates with the findings of reference [26], where no spontaneous formation of bubbles was observed on flat hydrophilic silicon wafers. The surface is smooth and hydrophilic with a roughness of 0.43 nm determined in water by iDrive tapping mode. The roughness correlates well with the values obtained in the literature [26]. In the presence of 0.05-0.4 mM CTAB, some features appear at the silicon wafer surface, which resemble nanobubbles. They are spherical, at 0.5 mM flatter (micropancakes), and they vanish close to CMC (0.8×CMC). Similar morphologies have been observed by other authors [19, 20, 27–31]. From the phase image one can conclude that the features are deformable at least at a concentration of 0.05 mM. The features are not present when a hydrophilic silicon wafer is imaged in water. The features with heights between 8 and 15 nm correlate well with that reported by other authors [32]. At low concentration of 0.05 mM, the feature height is around 15 nm. The observed features cannot be micelles because

95 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC micelles would have a diameter of 3.4 nm [5]. The properties of the spherical features suggest that they are nanobubbles on the surface of a partially hydrophobic silicon wafer (see figure 5.7). The presence of nanobubbles at the surface in figure 5.2 is an evidence that a partial hydrophobization of the silicon wafer surface occurs due to the adsorption of surfactant. The nucleation of the nanobubbles may be produced by air dissolved in water [17]. The nanobubbles have a regular shape and are stabilized by the surfactant in the solution. It cannot be excluded that surface active contaminants may also be present at the bubble interface [17, 33]. The contaminants in the reference [17] are due to the surface preparation (silanization process). Since in the present experiments the partial hydrophobization occurs in situ on a clean hydrophilic silicon wafer, only a tiny amount of contaminants should be present, if any. Another source of contaminant could be the tip itself [34]. Borkent et al. [34] proposed that contaminants coming from the gel package and deposited on the tip can precipitate on the organic surface once the tip is immersed in water for scanning. The polysiloxanes (from the gel package) could be adsorbed at the air–liquid interface (it is energetically unfavourable that polysiloxanes deposit on hydrophilic surfaces immersed in liquid [20]) and might be the reason for the discrepancy observed between the nanoscopic and macroscopic contact angles [34]. Song et al. [20] did not find experimental evidence that oligomeric siloxanes from the gel package influence the contact angle. Therefore, this source of contaminants can also be excluded from our experiments. The nanobubbles seem to be flattened at the surface (see table 5.1). We propose that the nanobubbles are seated at the hydrophobic tail of the CTAB (see figure 5.7) and that they label the hydrophobic domains. The size and distribution of the aggregates in the presence of nanobubbles may differ from the size and distribution of the aggregates in degassed solution since the bubbles could modify the distribution of surfactant. Water Nanobubble

Cationic surfactant molecule Silicon Wafer

Figure 5.7.: Schematic picture of nanobubbles (not to scale) seated on the hydrophobic tails of the surfactant molecules

The bubble parameters were obtained by ﬁtting a cross section of the nanobubble to an arc of a circle [19, 23, 24]. At 0.05 mM surfactant concentration, the height of the bubbles is 13 ± 2.2 nm and the radius is 37.7 ± 4.6 nm. With increasing surfactant concentration, the bubbles become smaller. At a surfactant concentration of 0.3 mM, the bubble height is 7.6 ± 1 nm and the radius 28 ± 4.3 nm (see ﬁgure 5.3). At 0.4 mM, the bubbles are smaller, which introduces more error in the analysis. At this concentration, nanobubbles with heights of 5–8 nm were found with a radius between 20 and 33 nm.

96 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

The nanoscopic contact angle remains constant between 140◦ and 150◦ irrespective of the CTAB concentration (see table 5.1) and is in good agreement with the literature [18, 34, 35].

Surfactant Avg. Avg. bubble Avg. radius of Contact Macroscopic Avg. nanoscopic

concentration height radius curvature angle contact angle contact angle

I (mM) h (nm) r (nm) Rc (nm) θair (degrees) θwater (degrees) θwater (degrees) 0.05 13.1 37.7 58.6 38.4 48.4 141.6 0.3 7.5 28.0 58.0 30.4 41.0 149.6 0.4 5.2 20.4 46.5 29.0 42.4 151.0

Table 5.1.: Parameters of the nanobubbles obtained by ﬁtting the cross section to an arc of a circle. The small size of the nanobubbles at 0.4 mM CTAB concentration introduces more errors in the ﬁtting and in the obtained parameters, but still the parameters of the nanobubbles are shown for comparison.

Surfactant Avg. Avg. bubble Avg. radius of Contact Avg. nanoscopic Nanoscopic

concentration height radius curvature angle contact angle ∆θwater (d)

I (mM) h (nm) rd (nm) Rcd (nm) θair (d) (degrees) θwater (d) (degrees) (degrees) 0.05 13.1 31.0 43.6 45.4 134.6 3.0 0.3 7.5 24.3 43.0 34.3 145.6 3.3 0.4 5.2 17.0 31.5 32.8 147.0 5.1

Table 5.2.: Parameters of the nanobubbles obtained after tip deconvolution, Rtip =15 nm±5 nm

Zhang et al. [19] also obtained no variation in contact angle in spite of a high polydisper- sity. This is in contrast to the work of Yang et al. [36] where a variation of the nanoscopic contact angle of the nanobubbles was observed after adding 2-butanol surfactant. They reported a slight decrease in the height of the nanobubbles and a more pronounced decrease in width after adding surfactant, which led to a decrease of the nanoscopic contact angle. The diameters of the nanobubbles in this work are smaller than those reported by other authors (see [18, 19]), because they are associated to the hydrophobic domains at the surface. They are stable at least for 30 minutes observation time. Yang et al. [36] reported small bubbles on a hydrophobic surface with a diameter of 100 nm whereas Kameda et al. [37] reported nanobubbles with a diameter between 10 and 100 nm on a Au(111) surface. Since the size of the bubbles is in the same order of magnitude as the size of the tip, it is necessary to perform tip deconvolution. The end of the tip is spherical with a curvature radius of Rtip = 15 nm ± 5 nm obtained by scanning electron microscopy. The deconvoluted radius of the curvature of the nanobubbles can be calculated as follows [20]:

97 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

Rcd = Rc − Rtip (5.1)

(Rcd : deconvoluted curvature radius of the nanobubble, Rc: convoluted curvature radius of the nanobubble, Rtip: tip curvature radius). The error propagation was stated (see table 5.2) taking into account the error in ﬁtting the nanobubble cross section to an arc of a circle and the error due to the uncertainty in the determination of the tip radius. Note that the nanobubble parameters vary slightly after tip deconvolution and that the nanoscopic contact angle still remains constant within the experimental errors (see table 5.2). The Laplace equation predicts, that small bubbles will have a large internal pressure. The pressure inside the bubbles is about 30 times the atmospheric pressure for the studied concentrations. Song et al [20] obtained similar Laplace pressures for the nanobubbles. In comparison, the internal pressure of the nanobubbles in water with a diameter of about 1 µm is in the range of 1 to 1.7 atm [19]. According to the Laplace equation

2σ ∆P = vl (5.2) Rc

(∆P :Laplace pressure, σvl:interfacial tension at liquid-vapor interface, Rc:curvature radius), the increase in surfactant concentration CTAB and the related decrease in surface tension leads to a decrease in bubble radius at constant ∆P (see figures 5.2, 5.3, 5.4). The nanoscopic contact angle througth water is much larger than the macroscopic one (see table 5.1). Song et al. [20] reported a correlation within the experimental errors of the macroscopic and nanoscopic contact angle of nanobubbles found on a hydrophilic sample ◦ (binary self-assembly monolayers; SAMs, θmacro = 37 ). In our experiments, only the local hydrophobicity of the (hydrophobic) domains/surfactant aggregates is taken into account for the determination of the nanoscopic contact angle. In contrast, different areas including hydrophobic domains and hydrophilic areas contribute to the macroscopic contact angle. A drop in phase angle is detected at the position of the nanobubbles at low surfactant concentration (see figure 5.2). At higher concentrations, the phase shift was small or not detected (see figures 5.3, 5.4, 5.5). Simonsen et al. [38] argued that the phase change is due to the deformation of the bubbles when the tip approaches the surface. The lack in phase shift for higher surfactant concentrations is due to the fact that the bubbles become stiffer and that their viscoelasticity approaches the one of the substrate.

5.4.2. Correlation with Force Curves

In order to correlate the appearance of nanobubbles with interactions between hydrophilic surfaces in the presence of CTAB, force curves between a silica particle and a silicon wafer

98 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC were recorded. As shown in ﬁgure 5.8, they cannot be ﬁtted with the DLVO theory at distances smaller than 20 nm.

0.6

ö=+50 mV, ë=15.2 nm 0.4mMC16TAB, 0.5mMC16TAB, ö=+70 mV, ë=13.6 nm 0.5 DLVO_CC DLVO_CP DLVO_CC (plane of charge taken 8 nm from the surface) DLVO_CP(plane of charge taken 8 nm from the surface)

0.4

0.3 ðR [mNIm] F/2

0.2

0.1

0.0

0 10 20 30 40 50 60 70 Separation, D [nm]

Figure 5.8.: Force curves between a silica particle and a silicon oxide surface in the presence of 0.4 and 0.5 mM CTAB concentration

The obtained Debye length coincides well with the theoretical one, which indicates that almost all surfactant molecules in solution are dissociated. From force measurements we obtained, that for concentrations below 0.4 mM only attraction is observed during the whole range (data not shown). At 0.4 mM surfactant concentration, a slight repulsion is observed at long range, which indicates a weak charge reversal (see figure 5.8). At this concentration, the nanobubbles have a height varying from 5 to 8 nm. The jump- in contact takes place at a distance of about 18 nm. Under the assumption that both opposing surfaces are decorated with nanobubbles, locally a foam lamella may be formed between 2 opposing nanobubbles (5 nm height), which ruptures at a distance of about 8 nm between the outer surface. This might lead to the observed attraction. Similar arguments are published in reference [39] for the rupture of aqueous wetting films on a hydrophobic surface [40]. When increasing the concentration to 0.5 mM, a strong repulsion is detected up to a distance of 10 nm. Then a jump-in contact is observed. The force curve can be fitted via Gouy–Chapman theory for distances larger than 20 nm. The strong repulsion between 10 and 20 nm cannot be fitted by DLVO theory, which might be an indication for steric forces

99 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC

0.6

ë= 9.6 nm 0.5 1mMC16TAB, DLVO_CC DLVO_CP

0.4

0.3 ðR [mNIm] F/2

0.2

0.1

0.0

0 20 40 60 Separation, D [nm]

Figure 5.9.: Force curves between a silica particle and a silicon oxide surface in the presence of 1mM CTAB concentration due to the formation of micelles or (patchy) bilayers at the silica/silicon oxide surface, respectively. Velegol et al. [5] found a jump-in contact at 5 nm for the interaction between a tip and a silica surface in the presence of CTAB solution at 0.9×CMC. In the present study, the jump-in contact occurs at 10 nm distance since the interactions are between a silica particle and a silicon oxide surface. The size of micelles in solution is about 5 nm and, at the surface, 3.4 nm [5]. In both studies, the jump-in contact is due to the expulsion of micelles or bilayer patches between the surfaces. In the height image, only micropancakes were detected at this concentration. Micropancakes (thin gas layers) and micelles or patchy bilayers may coexist at this concentration. The obtained Debye lengths at 0.4 mM and 0.5 mM are 15.2 nm and 13.6 nm respectively, which correlates well with the theoretical values. At 1 mM surfactant concentration, a monotonous repulsive interaction between a silica particle and a silicon wafer is obtained (see ﬁgure 5.9). According to the present results, the repulsion is assumed to be caused by the approach of two opposing micelles or bilayers adsorbed on the silicon oxide/silica surfaces. The micelles/patchy bilayers are then so closely packed and stiﬀ, that they cannot be pressed out from the surfaces. No jump-in contact is observed, and the 0 distance refers to the point of contact between opposing micelles/bilayers. No bubbles were detected by SFM at this concentration since the

100 5. Scanning of Silicon Wafers in Contact with Aqueous CTAB Solutions below the CMC silicon wafer surface is already hydrophilic. A decrease in height and radius of the bubble in the presence of surfactant with decreasing amplitude setpoint was reported in reference [18]. Interestingly, this eﬀect was not seen in the absence of surfactant. Therefore, in the present study, a similar amplitude setpoint was used to compare the variation of height of the nanobubbles with surfactant concentration. We concluded that the ﬂattening of the nanobubbles at 0.5 mM is due to the decrease in surface tension.

5.5. Conclusions

We reported the presence of small nanobubbles at the surface of a silicon oxide surface exposed to aqueous CTAB solutions. The eﬀect of the surfactant is twofold, it can partially hydrophobize the silicon wafer surface and stabilize the nanobubbles. The hydrophobic surfactant patches present at the silicon oxide surface at low concentration (below 0.5 mM) are labeled by nanobubbles, which are imaged. The diameter of the nanobubbles varies from 30 to 60 nm (after tip deconvolution). The nanoscopic contact angle through water remains constant between 140◦ and 150◦ and is independent of the CTAB concentration. This angle veriﬁes the hydrophobicity of the domains formed by the surfactant aggregates. It is much higher than the macroscopic contact angle of a CTAB solution droplet (about 40◦), which presents an average of hydrophilic and hydrophobic areas on a silicon wafer partially covered with CTAB. The Laplace pressure within the nanobubbles is about 30 atm. With increasing CTAB concentration, the nanobubbles become smaller and less prominent. This indicates that the silicon wafer surface becomes more hydrophilic. At low CTAB concentration (0.05–0.4 mM) the surface is partially covered with hydrophobic domains where the nanobubbles can be placed. Since nanobubbles were only observed at low surfactant concentration (below 0.5 mM), they may play a role in the hydrophobic interactions. A strong attraction (jump-in) is observed at short distances (≤ 20 nm) which is explained by the rupture of the lamella between opposing nanobubbles seated on the hydrophobic domains. With increasing concentration, more and more hydrophilic domains, i.e. micelles or patchy bilayers, are formed until there are no bubbles observed close to the CMC (1 mM). At 0.5 mM long range electrostatic forces as well as steric repulsion followed by micelle/bilayer expulsion are observed in the interaction curve between the two opposing silica/silicon oxide surfaces. In the height image, only micropancakes were detected. At 1 mM, the exerted force was not high enough to expulse the adsorbed aggregates from the surface.

101 Bibliography

[1] D. W. Fuerstenau and R. Herrera-Urbina. Cationic Surfactant: Physical Chemistry. Ed. by D. N. Rubingh and P. M. Holland. Vol. 37. Marcel Dekker: New York, 1991, p. 408. [2] V. Subramanian and W. A. Ducker. In: Langmuir 16 (2000), pp. 4447–4454. [3] S. C. Biswas and D. K. Chattoraj. In: J. Colloid Interface Sci. 205 (1998), pp. 12– 20. [4] M. W. Rutland and J. L. Parker. In: Langmuir 10 (1994), pp. 1110–1121. [5] S. B. Velegol et al. In: Langmuir 16 (2000), pp. 2548–2556. [6] W. A. Ducker and E. J. Wanless. In: Langmuir 15 (1999), pp. 160–168. [7] B. G. Sharma, S. Basu, and M. M. Sharma. In: Langmuir 12 (1996), pp. 6506–6512. [8] R. Atkin et al. In: Adv. Colloid Interface Sci. 103 (2003), pp. 219–304. [9] J. L. Parker, V. V. Yaminski, and P. M. Claeson. In: J. Phys. Chem. 97 (), pp. 7706– 7710. [10] V. Yaminski et al. In: Langmuir 12 (1996), pp. 3531–3535. [11] J. W. G. Tyrrell and P. Attard. In: Phys. Rev. Lett. 87 (2001), pp. 176104–1– 176104–4. [12] L. A. Palmer, D. Cookson, and R. N. Lamb. In: Langmuir 27 (2011), pp. 144–147. [13] N. Ishida et al. In: Langmuir 16 (2000), pp. 6377–6380. [14] Y. Takata et al. In: Langmuir 22 (2006), pp. 1715–1721. [15] J. W. G. Tyrrell and P. Attard. In: Langmuir 18 (2002), pp. 160–167. [16] D. A. Doshi et al. In: Proc. Natl. Acad. Sci. U.S.A. 102 (2005), pp. 9458–9462. [17] W. A. Ducker. In: Langmuir 25 (2009), pp. 8907–8910. [18] X. H. Zhang, N. Maeda, and V. S. J. Craig. In: Langmuir 22 (2006), pp. 5025–5035. [19] X. H. Zhang, A. Quinn, and W. A. Ducker. In: Langmuir 24 (2008), pp. 4756–4764. [20] B. Song, V. Walczyk, and H. Schönherr. In: Langmuir 27 (2011), pp. 8223–8232. [21] J. Mata, D. Varade, and P. Bahadur. In: Thermochim. Acta 428 (2005), pp. 147– 155. [22] S. Hohlbauch, H. Cavazos, and R. Proksch. iDrive: Theory, Applications, Installa- tion and Operation. Asylum Research. Santa Barbara, CA, 2008.

102 Bibliography

[23] Y. Wang, B. Bhushan, and X. Zhao. In: Nanotechnology 20 (2009), p. 045301. [24] L. Zhang et al. In: Soft Matter 6 (2010), pp. 4515–4519. [25] SCA 20, Version 2.04. DataPhysics Instruments GmBH. 2002. [26] A. Agrawal and G. H. McKinley. In: Mater. Res. Soc. Symp. Proc. 899E (2006), NO7–37.1. [27] B. M. Borkent et al. In: Phys. Rev. Lett. 98 (2007), p. 204502. [28] B. M. Borkent et al. In: Phys. Rev. E 80 (2009), p. 036315. [29] M. Holmberg et al. In: Langmuir 19 (2003), pp. 10510–10513. [30] X. H. Zhang, A. Khan, and W. A. Ducker. In: Phys. Rev. Lett. 98 (2007), p. 136101. [31] J. R. T. Seddon and D. Lohse. In: J. Phys. Condens. Matter. 23 (2011), p. 133001. [32] R. Steitz et al. In: Langmuir 19 (2003), pp. 2409–2418. [33] D. R. Evans, V. S. J. Craig, and T. J. Senden. In: Phys. A 339 (2004), pp. 101–105. [34] B. M. Borkent et al. In: Langmuir 26 (2010), pp. 260–268. [35] H. Elnaiem et al. In: Comput. Mat. Continua 14 (2009), pp. 23–34. [36] S. Yang et al. In: Langmuir 23 (2007), pp. 7072–7077. [37] N. Kameda and S. Nakabayashi. In: Chem. Phys. Lett. 461 (2008), pp. 122–126. [38] A. C. Simonsen, P. L. Hansen, and B. Klösgen. In: J. Colloid Interface Sci. 273 (2004), pp. 291–299. [39] K. W. Stöckkelhuber et al. In: Langmuir 20 (2004), pp. 164–168. [40] R. v. Klitzing. In: Adv. Colloid Interface Sci. 114-115 (2005), pp. 253–266.

103 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer

6.1. Introduction

Hydrophobic interactions are attractive strong interactions between hydrophobic objects or nonpolar molecules. These interactions play an important role in biology, since they determine the conformation of proteins and the structure of the biological membrane [1]. Several theories have been proposed to explain hydrophobic interactions. One theory explains that hydrophobic surfaces induce the formation of hydrogen bonds in the water molecules closer to the surface, this effect propagates at larger distances and the disrup- tion of the order produces a long range attractive force between the surfaces [2]. Another theory takes the electrostatic fluctuations at the hydrophobic surface as an explanation for the long range attraction observed between neutral bodies [3,4]. Yaminski et al. [5] argued that cavitation can explain the long range attractive forces between hydrophobic surfaces. Considine et al. [6] measured the interaction forces between two latex particles in an electrolyte solution. A big attraction was observed which varies from 20-400 nm, depending on the pair of spheres used. The proposed model to explain the attraction was based on the presence of gas bubbles attached to the particle surface. The bridging mechanism between bubbles of opposing surfaces results in an attractive force. They obtained a variation of the attractive force when the water was degassed. Carambassis et al. [7] also concluded that nanobubbles are responsible for the long range attractions observed between a silica and an oxidized silicon wafer hydrophobically modified with Tridecafluoro 1,1,2,2-tetrahydrooctyl-methyldichlorosilane. The range of the attraction from 50-100 nm was a measure of the size of the bubble. Craig et al. [8] studied the hydrophobic interactions between silica surfaces immersed in solutions of CTAB and CPC in the presence of 0.1 M NaCl and observed a decrease in the attractive force after removing the dissolved gas. Nonetheless the attraction was still present. They also obtained a larger attractive force for surfaces immersed in CPC solutions than for surfaces immersed in CTAB solutions. The magnitude and the range of the hydrophobic force depends on the material conforming the hydrophobic surface and the method of preparation [9]. Kokkoli et al. [9] cited that the range of the force between surfaces prepared from Langmuir-Blodgett deposition and silanization processes is about 250 nm, whereas hydrophobic surfaces formed from equilibrium adsorption have short range hydrophobic

104 6. Nanobubbles at the Surface of a Divinyl Disilazan modified Silicon Wafer interactions of about 10 nm. The hydrophobic interactions between polymer surfaces are measurable at distances of about 30 nm [9]. Divinyldisilazan is a silane with application in the rubber industry. This silane renders the surface hydrophobic with a contact angle close to 90◦. It is proven that nanobubbles are present at the interface of water and hydrophobic materials [10–12]. The contact angle of the nanobubbles varies depending on the composition of the solid surface [13]. This study was outlined to characterize the nanobubbles present at the interface of a divinyldisilazan modified silicon wafer immersed in water and surfactant solutions, and to study the influence of the surfactant on the nanoscopic contact angle of the observed nanobubbles. The interaction forces between two hydrophobically (divinyldisilazan) modified particles in the presence of surfactant solutions are also analyzed.

6.2. Experimental Section

6.2.1. Materials

A powder of silica particles 4.74 µm mean diameter was purchased from Bangs Labo- ratories and 1,1,3,3-tetramethyl-1,3-divinyldisilazan analytic grade was purchased from Aldrich. Solutions from cetyl-trimethyl-ammonium bromide (CTAB, analytic grade, Aldrich) were prepared in a range of concentration from 0.005 mM to 1.2 mM in pure water (milli-Q). The surface tension measurements were performed as described in section 5.2.1.

6.2.2. Preparation and Methods

Preparation

The colloidal silica particles were exposed to a vapor of 1,1,3,3-tetramethyl-1,3-divinyldisilazan under vacuum conditions. A small beaker containing 0.5 g of silica particles was placed in the reaction flask. The system was heated to 120◦C under vacuum (10−2T orr) for two hours to remove the adsorbed water. Thereafter the silane was added dropwise to the reaction flask. The system was left at room temperature and the particles were exposed to the silane vapor for 24 hours. The excess silane was removed at 120◦C under vacuum (10−2T orr). Finally hydrophobically modified silica particles were obtained [14]. The hydrophobic particle was glued to the end of a tipless AFM cantilever (CSC12, µ-mach, Lithonia) and another one to a glass slide (Menzel-Gläser, Germany) using an optical microscope and a micromanipulator. Colloidal probes and glass slides with attached particles were cleaned with ethanol and water and placed in an air plasma cleaner for 20 minutes (Diener electronic. Femto timer). The silicon wafers (type-P) Wacker Siltronic Burghausen were cut and cleaned in piranha solution for 30 minutes,

105 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer thereafter washed with milli-Q water and then modiﬁed by the same procedure as the silica particles.

Methods

The force measurements between the two silica particles [15] were performed using a MFP-3D Asylum Research atomic force microscope mounted in an inverted optical microscope (Olympus IX71) as described in section 3.2.2. The scanning was performed using the same apparatus (for further details see section 5.2.2). The contact angle of the modiﬁed silicon wafer was measured with a Goniometer (OCA5) using the Sessile drop method as described in section 5.2.2.

6.2.3. Simulations

The simulations are based on the DLVO theory. The description is found in section 3.2.3.

6.3. Results

Eﬀect of Surfactant

The silicon wafers were modified using the same method as for silica particle modification. The macroscopic contact angle of water on a modified silicon wafer is 89◦. The modified silicon wafer was exposed to the different surfactant solutions. Figure 6.1 shows the image of the silicon wafer in water. The surface is covered with nanobubbles of different sizes with an average height of 27 nm. The nanobubbles are stable and have an irregular form. The obtained roughness is about 9.492 nm. The fit of the cross section of the bubble to an arc of a circle [16–18] allows the determination of the bubble parameters. ◦ The radius of curvature Rc was 96 nm and the nanoscopic contact angle was about 136 on average. The morphology and distribution of the nanobubbles in the presence of 0.1 mM surfactant concentration are represented in figure 6.2. Different bubble sizes are present at the surface from 100 to 300 nm in diameter. It is not possible to obtain the nanobubble parameters with these experimental conditions, because the bubbles are flattened at the modified silicon wafer. The big nanobubbles are still recognized in the phase image. At 0.3 surfactant concentration, nanobubbles were still observed at the modified silicon wafer (see figure 6.3). The average height is between 4 and 8 nm and they are flattened at the surface. An increase of the concentration to 0.4 mM (see figure 6.4) does cause the nanobubbles at the surface to disappear. There are also other features which resemble micropancakes [19]. Nanobubbles will be present at the surface of the modified silicon

106 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer

1.0

0.8

0.6 5 µm 0 0.4 nm

-5 0.2

0.0 0.0 0.2 0.4 0.6 0.8 1.0 µm

20nm 10

Height 0 -10

0.0 0.2 0.4 0.6 0.8 1.0 µm

20° 10 0

Phase -10 -20

0.0 0.2 0.4 0.6 0.8 1.0 µm

Figure 6.1.: AFM tapping mode of a hydrophobically modified silicon wafer in water. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.292 V, setpoint ratio: 0.29, scan rate: 0.5 Hz, drive frequency: 6.27 KHz wafer until 1.2 mM (see figures 6.5, 6.6). It seems that the nanobubbles are deformed at 1.2 mM surfactant concentration. Interaction forces between two hydrophobically modified silica particles were also performed in order to get information about the range and magnitude of the hydrophobic forces. Figure 6.8 shows the interaction between two hydrophobic particles in water and at different surfactant concentrations. The diameter of the modified particles was obtained by SEM and is around 4.7 µm. The pH of the solutions is around 5.8. The experimental curves were fitted to the DLVO theory. It may be that nanobubbles are also present at the silica surfaces. Hence the fitted diffuse layer potential are not repre- sentative of the surface potential. The Debye length is a bulk parameter and can be used to obtain the experimental ionic strength. The interactions in the presence of water are attractive during the whole range with a jump into contact around 40 nm. An increase of the concentration to 0.03 mM gives rise to a slight long range repulsion with an attractive interaction at short range, which cannot be explained by van der Waals forces. The Debye length is 43 nm, which correlates well with the Debye length in water. A further increase in the concentration to 0.3 mM does not increase the repulsion, and attractive interactions are still observed at short range, but they are smaller than at lower surfactant concentration. The jump into contact is around 13 nm. The obtained Debye length at this concentration is 17.6 nm. At 0.4 mM still a small attraction is detectable at short range, but at long range strong repulsive forces dominate the interaction. For higher

107 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer

1.0

0.8

0.6 5 µm 0 0.4 nm

-5 0.2

0.0 0.0 0.2 0.4 0.6 0.8 1.0 µm

20nm 15 10 5 Height 0 -5

0.0 0.2 0.4 0.6 0.8 1.0 µm

100° 96

Phase 92

0.0 0.2 0.4 0.6 0.8 1.0 µm

Figure 6.2.: AFM tapping mode of a hydrophobically modiﬁed silicon wafer in 0.1 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 5.1 KHz

300

250

200 5 150 nm 0 nm 100 -5 50

0 0 100 200 300 nm

6nm 4 2

Height 0 -2

0 50 100 150 200 250 300 nm

8° 4 0 Phase -4

0 50 100 150 200 250 300 nm

Figure 6.3.: AFM tapping mode of a hydrophobically modiﬁed silicon wafer in 0.3 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 5.6 KHz

108 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer

300

250

200 2

1 150 nm 0 nm 100 -1

50 -2

0 0 100 200 300 nm

8nm 4

Height 0

0 50 100 150 200 250 300 nm

4 0

Phase -4 -8°

0 50 100 150 200 250 300 nm

Figure 6.4.: AFM tapping mode of a hydrophobically modiﬁed silicon wafer in 0.4 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 5.6 KHz

300

250 2 200 1 150 nm 0 nm 100 -1 50 -2 0 0 100 200 300 nm

6nm 4 2 0 Height -2 -4

0 50 100 150 200 250 300 nm

10° 5 0

Phase -5 -10

0 50 100 150 200 250 300 nm

Figure 6.5.: AFM tapping mode of a hydrophobically modiﬁed silicon wafer in 0.8 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 4.8 KHz

109 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer

300

250 1.5 200 1.0

150 0.5 nm 0.0 nm 100 -0.5

50 -1.0 -1.5 0 0 100 200 300 nm

4nm 2

Height 0 -2

0 50 100 150 200 250 300 nm

5 0

Phase -5 -10°

0 50 100 150 200 250 300 nm

Figure 6.6.: AFM tapping mode of a hydrophobically modiﬁed silicon wafer in 1.2 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 4.9 KHz

300

250 800 200 400 150 nm 0 100 pm -400 50 -800 0 0 100 200 300 nm

3nm 2 1 0 Height -1

0 50 100 150 200 250 300 nm

2° 0

Phase -2

0 50 100 150 200 250 300 nm

Figure 6.7.: AFM tapping mode of a hydrophobically modiﬁed silicon wafer in 1.2 mM CTAB solution. AFM images taken with a magnetic actuated cantilever, nominal spring constant 0.09 N/m, amplitude setpoint: 0.239 V, setpoint ratio: 0.24, scan rate: 0.5 Hz, drive frequency: 4.9 KHz

110 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer

0.4 water 0.03 mM

0.3

0.2

0.1 F/2

0.0

-0.1

0 20 40 60 80 100 Separation, D [nm]

Figure 6.8.: Forces between a pair of hydrophobically modiﬁed silica particles in water and from 0.03 to 1.2 mM surfactant concentration at pH=5.8, Hamaker constant A= 8.5 × 10−21 J. The continuous lines correspond to constant charge. DLVO_CC (constant charge) ﬁts are shown for 0.03 mM, 0.3 mM, 0.4 mM and 1 mM surfactant concentration concentrations repulsion is observed during the whole range and no adhesion is seen in the retraction curve. The experimental Debye lengths agree well with the theoretical ones.

6.4. Discussion

Eﬀect of the Surfactant Concentration on the Nanobubble Properties

The image of the hydrophobically modified silicon wafer in water is shown in figure 6.1. The surface is covered with nanobubbles of different sizes. The silanation was performed under vacuum (see section 6.2.2). Therefore, it is assumed that air was not trapped at

111 6. Nanobubbles at the Surface of a Divinyl Disilazan modified Silicon Wafer the surface before the AFM experiments and that nanobubbles formed spontaneously once the surface was exposed to water or aqueous solutions. It has been proven by Lou et al. [20] that nanobubbles can be induced on a hydrophobic surface after immersing the surface in water. Ducker et al. [21] argued that the stability of nanobubbles is due to the presence of contaminants at the vapor-liquid interface. That may be possible because an organic compound (divinyl disilazan) was used in the present study to modify the silica and the silicon wafer. However, an effect of the surfactant concentration on the bubble morphology is observed; the bubbles become flatter with increasing surfactant concentration, which means that surfactant is adsorbed at the vapor-liquid interface. The morphology of the bubbles at a divinyl disilazan interface exposed to water resembles a spherical cap. The parameters of the bubbles are obtained by fitting the cross section to an arc of a circle [16–18]. The average curvature radius Rc and the nanoscopic contact ◦ angle θwater are 96 nm and 136 respectively without tip deconvolution. Since the bubbles and the tip radius are in nanometer scale, tip deconvolution is needed. The end of the tip is spherical with a curvatures radius of Rtip = 15 nm ±5 nm obtained by scanning electron microscopy. The deconvoluted radius of curvature of the nanobubbles can be calculated as follows [22]:

Rcd = Rc − Rtip (6.1)

(Rcd : deconvoluted curvature radius of the nanobubble, Rc: convoluted curvature radius of the nanobubble, Rtip: tip curvature radius). The deconvoluted curvatures radius is 81 nm and from there the deconvoluted nanoscopic contact angle can be calculated (132◦ ± 13◦). The average nanoscopic contact angle of 132◦ is larger than the macroscopic one of 89◦. According to the Laplace equation

2σ ∆P = vl (6.2) Rc

(∆P : Laplace pressure, σvl: interfacial tension at liquid-vapor interface, Rc: curvature radius), taken σvl = 72 mN/m for water, the internal pressure of the nanobubbles in water is about 18 atm. Zhang et al. [16] reported an internal pressure of the nanobubbles in water with a diameter of about 1 µm in the range of 1 to 1.7 atm. At 0.01 mM surfactant concentration the surface is still covered with the same amount of nanobubbles and these nanobubbles still have a spherical cap morphology (data not shown). No interaction between tip and sample is seen at long range, but at short range the interactions are attractive (see ﬁgure 6.9) with a jump into contact around 10 nm. The origin of these attractive forces may be due to van der Waals forces or due to the penetration of the tip into the nanobubbles. The jump into contact correlates well with the observed height of the nanobubbles. The increase in the concentration to 0.3 mM does not reduce the amount of bubbles at the surface (see ﬁgure 6.3), but a decrease in height is detected in comparison with the

112 6. Nanobubbles at the Surface of a Divinyl Disilazan modiﬁed Silicon Wafer

3nm

0 Defl

-1

-2

-3

0 10 20 30 40nm Sep

Figure 6.9.: Interaction force between a tip and a bubble in 0.01 mM CTAB concentration bubbles on the modified silicon wafer without any surfactant. It seems that the surfactant is adsorbed at the gas–liquid interface decreasing the Laplace pressure, which leads to a flattening of the nanobubbles. The bubbles are not so well defined (more flattened), therefore it is difficult to determine the nanoscopic contact angle. Still nanobubbles are present at 0.4 mM surfactant concentration, but other kinds of features are also visualized. The features resemble flat air layers (micropancakes). A similar morphology has also been observed by other authors [19, 23]. At 0.8 mM nanobubbles were still found at the surface. For higher concentrations flattened circular domains are present at the surface with a typical hight of 3-4 nm. They may also be interpreted as aggregates at the modified silicon wafer surface (see figure 6.7). The 300 × 300 nm scale image shows flat domains at the surface with a typical height of 4 nm. Interestingly, at 1.2 mM surfactant concentration nanobubbles were still found at the surface, but they are flatter and more uniform than the bubbles present at the surface of a modified silicon wafer without any surfactant (see figure 6.7). Nanobubbles may also be present at the surface of the modified silica particles. The interaction of the hydrophobic silica particles in water are attractive over the whole range. It is proposed that the rupture of a water film between two opposing nanobubbles may cause this long range attraction (see figure 6.8). In that way the jump into contact will be related to the rupture thickness of the water film around 40 nm between two opposing bubbles. This value correlates well with the rupture of a foam film in water [24]. A similar behaviour was obtained by Ishida [25] for the interaction between two hydrophobic surfaces in water. At 0.03 mM a slight repulsion at large separation, due to the recharg- ing of the surface, is seen (see figure 6.8). The jump into contact is at around 17 nm. Long range repulsive interactions are observed at 0.3 mM surfactant concentration. For distances smaller than 10 nm a non-DLVO repulsion is observed until 8 nm, which may be attributed to the deformation of the bubble or to the steric repulsion of some aggre-

113 6. Nanobubbles at the Surface of a Divinyl Disilazan modified Silicon Wafer gates. The repulsion is followed by an attractive jump into contact. The obtained Debye length of 17.6 nm correlates well with the theoretical one. The "effective" diffuse layer potential was 42 mV. The interaction forces at 0.8 mM and 1 mM surfactant concentration are monotonic repulsive with a Debye length of 10.7 nm and 9.61 nm respectively. The corresponding effective diffuse layer potentials are +49 mV and +30 mV. Additional force curves, performed with another pair of modified silica particles at 0.8 mM and 1 mM (data not shown), yield an effective diffuse layer potential of +40 mV and +38 mV respectively. The obtained Debye lengths of 10.7 nm and 9.61 nm correlate well with the theoretical ones. The addition of surfactant increases the diffuse layer potential to more positive values. The apparent decrease in potential is due to the condensation of the Br− ions on the adsorbed layer during the force measurements, which leads to a screening of the surface charge. Also the uncertainty of the plane of charge due to aggregates at the surfaces or nanobubbles contribute to the low fitted potential obtained at higher surfactant concentrations. The lack of attraction at short range at higher surfactant concentration may be related to a closer packing of the aggregates at the surface or may be due to an increase in bubble elasticity with surfactant concentration (the nanobubbles will be covered with surfactant). The stiffness of a nanobubble will be greater than that of a microbubble 0.065N/m−1 [26]. That means that at large surfactant concentration, a weak cantilever, with a typical spring constant of 0.03N/m−1, is not able to penetrate the nanobubbles and come into contact with the hard surface because the nanobubble is more compliant than the cantilever itself. The same result was obtained by Zhang et al. [27]. The nanobubbles described in chapter5 have similar nanoscopic contact angles as the nanobubbles at the modified silicon wafer. For the same studied concentration (e.g. 0.3 mM), the nanobubbles at the CTAB hydrophobic patches (see chapter5) resemble a spherical cap, whereas the nanobubbles at the divinyldisilazan interface are more flattened at the surface. They are also bigger in size and their nanoscopic contact angle is closer to the macroscopical one. Different surfaces may induce different nanobubble morphologies.

6.5. Conclusions

Nanobubbles are formed at the interface of a hydrophobically modified silicon wafer (Divinyl-disilazan) exposed to water and surfactant solutions. In water the nanobubbles resemble a spherical cap with a height of around 27 nm. The nanoscopic contact angle through water of the modified silicon wafer after tip deconvolution is about 132◦, which is larger than the macroscopic contact angle of 89◦. The Laplace pressure inside the nanobubbles is 18 atm. A decrease in nanobubble height is observed with increasing surfactant concentration. At 0.4 mM micropancake like morphologies were imaged at the modified silicon wafer surface. At 1.2 mM surfactant concentration circular aggregates were visualized. Condensation of the Br− ions on the adsorbed surfactant layer, together with the presence of aggregates at the surface, may explain the low diffuse layer potential obtained at larger surfactant concentrations.

114 Bibliography

[1] J. Israelachvili and R. Pashley. In: Nature 300 (1982), pp. 341–342. [2] J. Ch. Erikkson, S. Ljunggren, and P.M. Claesson. In: J. Chem. Soc. Faraday Trans. 2 85 (1989), pp. 163–176. [3] R. Podgornik. In: Chem. Phys. Lett. 156 (1989), pp. 71.–75. [4] P. Attard. In: J. Phys. Chem. 93 (1989), pp. 6441–6444. [5] V. V. Yaminski and B. Ninham. In: Langmuir 9 (1993), pp. 3618–3624. [6] R. F. Considine, R. A. Hayes, and R. G. Horn. In: Langmuir 15 (1999), pp. 1657– 1659. [7] A. Carambassis et al. In: Phys. Rev. Lett. 80 (1998), pp. 5367–5360. [8] V. S. J. Craig, B. W. Ninham, and R. M. Pashley. In: Langmuir 15 (1999), pp. 1562– 1569. [9] E. Kokkoli and Ch. F Zukoski. In: Langmuir 14 (1998), pp. 1189–1195. [10] J. W. G. Tyrrell and P. Attard. In: Phys. Rev. Lett. 87 (2001), pp. 176104–1– 176104–4. [11] B. M. Borkent et al. In: Langmuir 26 (2010), pp. 260–268. [12] N. Ishida et al. In: Langmuir 16 (2000), pp. 6377–6380. [13] H. Elnaiem et al. In: Comput. Mat. Continua 14 (2009), pp. 23–34. [14] D. Akcakayiran. “Funktionalisierung der Porenwände und Einlagerung von Metall- Koordinations-Polyelektrolyten in SBA-15 Silika”. MA thesis. 2004. [15] G. Toikka, R. A. Hayes, and J. Ralston. In: Langmuir 12 (1996), pp. 3783–3788. [16] X. H. Zhang, A. Quinn, and W. A. Ducker. In: Langmuir 24 (2008), pp. 4756–4764. [17] Y. Wang, B. Bhushan, and X. Zhao. In: Nanotechnology 20 (2009), p. 045301. [18] L. Zhang et al. In: Soft Matter 6 (2010), pp. 4515–4519. [19] J. R. T. Seddon and D. Lohse. In: J. Phys. Condens. Matter. 23 (2011), p. 133001. [20] Shi-Tao. Lou et al. In: J. Vac. Sci. Technol. B 18 (2000), pp. 2573–2575. [21] W. A. Ducker. In: Langmuir 25 (2009), pp. 8907–8910. [22] B. Song, V. Walczyk, and H. Schönherr. In: Langmuir 27 (2011), pp. 8223–8232. [23] X. H. Zhang et al. In: Langmuir 23 (2007), pp. 1778–1783.

115 Bibliography

[24] D. Exerowa et al. In: Colloid Journal 63 (2001), pp. 50–56. [25] N. Ishida et al. In: Langmuir 16 (2000), pp. 5681–5687. [26] W. A. Ducker, Z. Xu, and J. N. Israeachvili. In: Langmuir 10 (1994), pp. 3279– 3289. [27] X. H. Zhang, N. Maeda, and V. S. J. Craig. In: Langmuir 22 (2006), pp. 5025–5035.

116 7. Conclusions and Future Work

7.1. General Conclusions

The discussed experiments show that ion specific effects are still present even at low ionic strengths. The long range interactions are similar for all the studied salts, but at 10−3 M ionic strength a slight tendency for Cs+ to be adsorbed at the surface is observed. The surface potentials are slightly lower for 10−3 M ionic strength indicating additional adsorption. The Debye length is different for 10−4 M and 10−3 M ionic strength and coincides with the ideal values. The same qualitative long range interaction behaviour is obtained for the studied ionic strengths. The short range interactions are different for both ionic strength studied. At 10−4 M ionic strength, the short range attraction decreases in the order Li+ >Na+ >K+ >Cs+ following the inverse Hofmeister series. The model of Pashley [1] and Higashitani [2] can explain the interaction between the silica particles at short range. It is possible that a gel layer is present at the silica surface which also influences the interactions between the silica spheres. A decrease in pH in the presence of electrolytes produces a synergetic effect between the protons and the cations, giving rise to hydration repulsion. An increase of the ionic strength to 10−3 M produces a short range repulsion due to hydration forces. It seems that the stability of silica under the studied conditions is defined by a balance of electrostatic forces at long range and a combination of hydration, depletion, and steric forces at short range. The interaction forces between two silica particles (system I) and between a silica particle and a silicon wafer (system II) in the presence of aqueous CTAB solutions with concentrations between 0.005 and 1 mM were measured using AFM. The force curves were correlated to the surfactant morphologies (figure 4.7). Both systems show a charge reversal from negative to positive caused by the adsorption of cationic surfactant (CTAB) at the former negatively charged silicon oxide surface. The interactions of the two systems were different for the same studied surfactant concentration. The point of zero charge was obtained at 0.05 mM for the silica particle–silica particle system (system I) as in Parker’s work [3] and at 0.3 mM for the silica particle–silicon wafer system (system II). This leads to different aggregate morphologies at the silica particle surface and at the surface of the silicon wafer. An explanation for the difference might be the surface treatment: the silica particles were plasma cleaned, whereas the silicon wafers were treated with a piranha solution. In another study the same surface treatment was applied to the silica particles and the silicon wafers and no differences in the interaction were observed [4]. At a low CTAB concentration (for example 0.05 mM) long-range attraction was observed. The attraction was larger in range and magnitude for the silica particle–silicon wafer system

117 7. Conclusions and Future Work and starts at distances larger than 40 nm. They cannot be caused by van der Waals attraction, but they are explained by the presence of nanobubbles probing hydrophobic patches on the surfaces. The attraction occurs when the nanobubbles bridge. Obviously, on a silicon wafer surface larger hydrophobic patches are present than on the surface of the silica particles. Another explanation for long-range attraction is the electrostatic attraction between oppositely charged patches. At higher surfactant concentration (0.4 mM onwards), monotonic repulsion between the two silica particles was observed. In contrast to this, aggregate expulsion could be observed in the interaction curves for the silica particle–silicon wafer system up to a CTAB concentration of 0.8 mM. The difference is explained by the different stiffness of the surfactant aggregates at the two surfaces (silica particle and silicon wafer). For the same studied concentrations, the outer layer of surfactant was denser for the silica particles and the aggregates on the silica particles were stiffer and more closely packed. The stiffer aggregates are more difficult to remove. A systematic study of the CTAB adsorption to a silicon wafer was performed by AFM. Small nanobubbles were present at the silicon oxide surface exposed to aqueous CTAB solutions. The effect of the surfactant is twofold, it can partially hydrophobize the silicon wafer surface and stabilize the nanobubbles. The hydrophobic surfactant patches present at the silicon oxide surface at low concentration (below 0.5 mM) are labeled by nanobubbles which are imaged. The diameter of the nanobubbles varies from 30 to 60 nm (after tip deconvolution). The nanoscopic contact angle through water remains constant between 140◦ and 150◦ and is independent of the CTAB concentration. This angle verifies the hydrophobicity of the domains formed by the surfactant aggregates. It is much higher than the macroscopic contact angle of a CTAB solution droplet (about 40◦), which presents an average of hydrophilic and hydrophobic areas on a silicon wafer partially covered with CTAB. The Laplace pressure within the nanobubbles is about 30 atm. With increasing CTAB concentration the nanobubbles become smaller and less prominent. This indicates that the silicon wafer surface becomes more hydrophilic. At low CTAB concentration (0.05-0.4 mM) the surface is partially covered with hydrophobic domains where the nanobubbles can be placed. Since nanobubbles were only observed at low surfactant concentration (below 0.5 mM) they may play a role in the hydrophobic interactions. A strong attraction (jump-in) is observed at short distances (≤ 20 nm), which is explained by the rupture of the lamella between opposing nanobubbles seated on the hydrophobic domains. With increasing concentration more and more hydrophilic domains, i.e. micelles or patchy bilayers, are formed, until there are no bubbles observed close to the CMC (1 mM). At 0.5 mM long range electrostatic forces as well as steric repulsion followed by micelle/bilayer expulsion are observed in the interaction curve between the two opposing silica/silicon oxide surfaces. In the height image only micropancakes were detected. At 1 mM the exerted force was not high enough to expulse the adsorbed aggregates from the surface. Nanobubbles are also formed at the interface of a hydrophobically modified silicon wafer (Divinyl-disilazan) exposed to water and surfactant solutions. In water the nanobubbles resemble a spherical cap with a height of around 27 nm. The nanoscopic contact angle through water of the modified silicon wafer after tip deconvolution is about 132◦, which is larger than the macroscopic contact angle of 89◦. The Laplace pressure inside the nanobubbles is 18 atm. A decrease in nanobubble

118 7. Conclusions and Future Work height is observed with increasing surfactant concentration. At 0.4 mM micropancake like morphologies were imaged at the modiﬁed silicon wafer surface. At 1.2 mM surfactant concentration circular aggregates were visualized. Condensation of Br− ions on the adsorbed surfactant layer together with the presence of aggregates at the surface may explain the low diﬀuse layer potential obtained at larger surfactant concentrations.

7.2. Future Work

Modiﬁcation of AFM Tips with Nanoparticles

Although the colloidal probe atomic force microscope allows the measurements of the interactions between the so called "colloidal particles", the particles are actually not real colloids. Their dimensions are in the µm range. Spalla et al. [5] postulated, that for the study of the hydrophobic interactions the use of macroscopic surfaces is a problem, since a huge area has to be hydrophobized and it is difficult to obtain a homogeneous hydrophobic layer. Rentsch et al. [6] studied the interaction forces between particles of different sizes and proved that the Derjaguin approximation holds true even at small distances. Will the Derjaguin approximation remain valid for the interaction between a nanoparticle of 500 nm diameter and a flat surface? The challenge is to modify an AFM tip with a nanoparticle of 500 nm or less in diameter in order to get nanoparticle terminated tips. Thereafter, interactions between the nanoparticle modified tip and other particles or planar surfaces have to be measured. Some previous experiments were performed using the following method. AFM cantilevers (PL2-CONT-10, NaNoAndMore) composed of silicon were cleaned with piranha solution. The cantilevers were then modified with APTS (aminopropyltriethoxy- silane) in the following way. First 0.1g of APTS was dispersed in toluene anhydrous. The cantilevers were placed in this solution. After 3 hours the cantilevers were removed and washed 3 times with toluene. The cantilevers were stored in a glass desiccator until further use. Modified silica particles around 500 nm in diameter with -COOH terminated groups were obtained from Bangs Laboratories. The idea is to perform an amidation reaction between the -NH2 terminated tips and the -COOH terminated silica particles. Surface functionalization with carboxylic acid functionalized silica was obtained by An et al. [7]. A suspension of the -COOH terminated silica particles was placed in DMF. The suspension was treated with a sonicator for 10 minutes. A small amount of DCC (N,N’- dicyclohexylcarbodimide) was dissolved in 10 ml of DMF (Dimethylformamide). The DCC solution was added to the particle suspension. Then the cantilevers were cleaned in a plasma cleaner (Diener electronic. Femto timer) and placed in the suspension for 24 hours. Thereafter, the expected particle terminated cantilevers were taken from the suspension and washed with DMF and water. Figures 7.1 and 7.2 show the resulting tip after the amidation reaction. Some particles are observed at the cantilever, but since the whole cantilever was modified, the particles can get attached to any part of the cantilever, not only to the end of the tip. Further work has to be done to find the right method for

119 7. Conclusions and Future Work the modiﬁcation of the tip with a single nanoparticle. The use of focus ion beam may facilitate the attachment of a nanoparticle to the cantilever.

Figure 7.1.: Scanning electron microscopy of a modiﬁed cantilever

Figure 7.2.: Scanning electron microscopy of a modiﬁed cantilever

120 Bibliography

[1] R. M. Pashley. In: Journal of Colloid and Interface Sci. 83 (1981), pp. 531–546. [2] I. U. Vakarelski, K. Ishimura, and K. Higashitani. In: J. Colloid Interface Sci. 227 (2000), pp. 111–118. [3] J. L. Parker, V. V. Yaminski, and P. M. Claeson. In: J. Phys. Chem. 97 (), pp. 7706– 7710. [4] M. Dishon, O. Zohar, and U. Sivan. In: Langmuir 25 (2009), pp. 2831–2836. [5] O. Spalla. In: Current Opinion in Colloid and Interface Sci. 5 (2000), pp. 5–12. [6] S. Rentsch et al. In: Phys. Chem. Chem. Phys. 8 (2006), pp. 2531–2538. [7] Yanqing An et al. In: Journal of Colloid and Interface 311 (2007), pp. 507–513.

121 A. Appendix

Figure A.1.: Scanning electron microscopy of silica particles

Figure A.2.: Scanning electron microscopy of a magnetic actuated cantilever

122 A. Appendix

Figure A.3.: Scanning electron microscopy of a magnetic actuated cantilever

Nanoscopic WATER contact angle

Height H AIR θnano

Width W Radius of curvature Rc

Figure A.4.: Schematic cross section of a nanobubble

123 A. Appendix

300

250 15 200 10 150 nm 5 100 nm 0 50 -5 0 0 100 200 300 nm

300

250 15 200 10 150 nm 5 100 nm 0 50 -5 0 0 100 200 300 nm

48nm 47 46 45

Amplitude 44 43

0 50 100 150 200 250 300 nm

Figure A.5.: Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.05 mM CTAB concentration (see ﬁgure 5.2)

124 A. Appendix

300

250 8 200 6 150 nm 4 nm 100 2

0 50 -2 0 0 100 200 300 nm

300

250 8 200 6 150 nm 4 nm 100 2

0 50 -2 0 0 100 200 300 nm

22.0nm 21.5 21.0

Amplitude 20.5 20.0

0 50 100 150 200 250 300 nm

Figure A.6.: Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.3 mM CTAB concentration (see ﬁgure 5.3)

125 A. Appendix

300

250

200 4

150 3 nm

2 nm 100 1 50 0 0 0 100 200 300 nm

300

250

200 26 25 150 nm 24

100 23 nm 22 50 21

0 0 100 200 300 nm

24.0nm

23.6

23.2 Amplitude

22.8

0 50 100 150 200 250 300 nm

Figure A.7.: Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.4 mM CTAB concentration (see ﬁgure 5.4)

126 A. Appendix

300

250

200 400

200 150 nm 0 pm 100 -200

50 -400

0 0 100 200 300 nm

300

250

200 400 200 150 nm 0 pm 100 -200

50 -400

0 0 100 200 300 nm

26.0nm

25.5

Amplitude 25.0

24.5

0 50 100 150 200 250 300 nm

Figure A.8.: Amplitude-distance data of nanobubbles on a silicon oxide surface at 0.5 mM CTAB concentration (see ﬁgure 5.5)

127 A. Appendix

1.0

0.8 15 0.6 10 5 µm 0

0.4 nm -5 0.2 -10 -15

0.0 0.0 0.2 0.4 0.6 0.8 1.0 µm

1.0

0.8 20

0.6 15 10 µm 0.4 5 nm 0 0.2 -5 -10 0.0 0.0 0.2 0.4 0.6 0.8 1.0 µm

11nm 10 9 8

Amplitude 7 6

0.0 0.2 0.4 0.6 0.8 1.0 µm

Figure A.9.: Amplitude-distance data of nanobubbles on a modiﬁed silicon wafer immersed in water (see ﬁgure 6.1)

128 A. Appendix

1.0

0.8 10

0.6 5 µm 0

0.4 nm

-5 0.2 -10 0.0 0.0 0.2 0.4 0.6 0.8 1.0 µm

1.0

0.8 30

0.6 28 µm 26 0.4 nm

24 0.2 22 0.0 0.0 0.2 0.4 0.6 0.8 1.0 µm

28nm 26 24 22 Amplitude 20

0.0 0.2 0.4 0.6 0.8 1.0 µm

Figure A.10.: Amplitude-distance data of nanobubbles on a modiﬁed silicon wafer at 0.1 mM CTAB concentration (see ﬁgure 6.2)

129 A. Appendix

300

250 3 200 2 150 1 nm 0 nm 100 -1

50 -2 -3 0 0 100 200 300 nm

300

250

200 4

150 2 nm 0 nm 100 -2

50 -4

0 0 100 200 300 nm

28nm 27 26 25

Amplitude 24 23

0 50 100 150 200 250 300 nm

Figure A.11.: Amplitude-distance data of nanobubbles on a modiﬁed silicon wafer at 0.3 mM CTAB concentration (see ﬁgure 6.3)

130 A. Appendix

300

250 4 200 2 150 nm 0 nm 100 -2 50 -4 0 0 100 200 300 nm

300

250

40 200 35 150 30 nm

25 nm 100 20 50 15

0 0 100 200 300 nm

28nm 27 26

Amplitude 25 24

0 50 100 150 200 250 300 nm

Figure A.12.: Amplitude-distance data of nanobubbles on a modiﬁed silicon wafer at 0.4 mM CTAB concentration (see ﬁgure 6.4)

131 A. Appendix

300

250

200 2 150 nm 0 100 nm -2 50

0 0 100 200 300 nm

300

250

200 30

150 28 nm 26 100 nm 24 50 22

0 0 100 200 300 nm

30nm 28 26 24 Amplitude 22

0 50 100 150 200 250 300 nm

Figure A.13.: Amplitude-distance data of nanobubbles on a modiﬁed silicon wafer at 0.8 mM CTAB concentration (see ﬁgure 6.5)

132 A. Appendix

300

250

200 2

150 1 nm 0 100 nm -1 50 -2

0 0 100 200 300 nm

300

250

200 30

150 28 nm 26 nm 100 24 50 22

0 0 100 200 300 nm

28nm

Amplitude 24 22

0 50 100 150 200 250 300 nm

Figure A.14.: Amplitude-distance data of nanobubbles on a modiﬁed silicon wafer at 1.2 mM CTAB concentration (see ﬁgure 6.6)

133