pubs.acs.org/Langmuir © 2009 American Chemical Society
Solvent Dependence of the Activation Energy of Attachment Determined by Single Molecule Observations of Surfactant Adsorption Andrei Honciuc, Denver Jn. Baptiste, Ian P. Campbell, and Daniel K. Schwartz* Department of Chemical and Biological Engineering, University of Colorado, Boulder, Colorado 80309 Received January 23, 2009. Revised Manuscript Received February 19, 2009 Single-molecule total internal reflection fluorescence microscopy was used to obtain real-time images of fluorescently labeled hexadecanoic (palmitic) acid molecules as they adsorbed at the interface between fused silica and three different solvents: hexadecane (HD), tetrahydrofuran (THF), and water. These solvents were chosen to explore the effect of solvent polarity on the activation energy associated with the attachment rate, i.e., the rate at which molecules were transferred to the surface from the near-surface layer. Direct counting of single-molecule events, made under steadystate conditions at extremely low coverage, provided direct, model-independent measurements of this attachment rate, in contrast with conventional ensemble-averaged methods, which are influenced by bulk transport and competing detachment processes. We found that the attachment rate increased with increasing temperature for all solvents. Arrhenius analyses gave activation energies of 5 ( 2 kJ/mol for adsorption from HD, 10 ( 2 kJ/mol for adsorption from THF, and 19 ( 2 kJ/mol for adsorption from water. These energies increased systematically with the solvent polarity and, therefore, with the expected strength of the solvent-substrate interaction. We hypothesize that the adsorption of amphiphilic solute molecules from solution can be regarded as a competitive exchange between solute molecules and surface-bound solvent. In this scenario, adsorption is an activated process, and the activation energy for attachment is associated with the solvent-substrate interaction energy.
Introduction The interfacial adsorption of amphiphilic molecules is one of the fundamental processes defining colloid science and underlies applications including detergency, colloidal stability, corrosion inhibition, and many others. The formation of self-assembled monolayers (SAMs)1 represents a special case of this process, where significant lateral associations between the adsorbed amphiphilic molecules result in a well-organized, stable, and robust surface coating. Several thorough reviews of surfactant adsorption and the structure of adsorbed layers have recently been published.2-6 There is an extensive literature dealing with the thermodynamics and kinetics of surfactant adsorption,5 which is typically a spontaneous process where the favorable enthalpy of adsorption compensates for the reduced entropy of the adsorbed solute. The classical mechanistic picture of surfactant adsorption7-9 divides the overall process into a number of conceptual steps including bulk transport to the near-surface layer, surface attachment from the near-surface layer, detachment from the surface, and interfacial mobility/organization. A number of experimental methods have been used to measure the overall net adsorption kinetics at the solid/solution interface, including solution depletion studies,10,11 *Corresponding author: e-mail
[email protected]; phone 303-735-0240; fax 303-492-4341.
(1) Schwartz, D. K. Annu. Rev. Phys. Chem. 2001, 52, 107-137. (2) Zhmud, B.; Tiberg, F. Adv. Colloid Interface Sci. 2005, 113 (1), 21-42. (3) Paria, S.; Khilar, K. C. Adv. Colloid Interface Sci. 2004, 110 (3), 75-95. (4) Zhang, R.; Somasundaran, P. Adv. Colloid Interface Sci. 2006, 123, 213-229. (5) Atkin, R.; Craig, V. S. J.; Wanless, E. J.; Biggs, S. Adv. Colloid Interface Sci. 2003, 103 (3), 219-304. (6) Parida, S. K.; Dash, S.; Patel, S.; Mishra, B. K. Adv. Colloid Interface Sci. 2006, 121 (1-3), 77-110. (7) Baret, J. F. J. Colloid Interface Sci. 1969, 30 (1), 1-12. (8) Koopal, L. K.; Avena, M. J. Colloids Surf., A 2001, 192 (1-3), 93-107. (9) Chang, C. H.; Franses, E. I. Colloids Surf., A 1995, 100, 1-45. (10) Biswas, S. C.; Chattoraj, D. K. J. Colloid Interface Sci. 1998, 205 (1), 12-20. (11) Torn, L. H.; Koopal, L. K.; de Keizer, A.; Lyklema, J. Langmuir 2005, 21 (17), 7768-7775.
Langmuir 2009, 25(13), 7389–7392
ellipsometry12-14 optical reflectometry,11,15-19 surface plasmon resonance,20-22 and spectroscopic methods.23 In all of these macroscopic, laterally averaging methods, the net amount of surface-bound adsorbate is determined as a function of exposure time. Although kinetic models have been developed to describe this combined process,7,9,24,25 in practice it remains difficult to tease apart the kinetic parameters associated with individual mechanistic steps. In particular, it is not always practicable to separate the net attachment rate (the attachment rate minus the detachment rate) from bulk transport, nor is it generally possible to separate the attachment and detachment rates with confidence. Thus, one cannot measure rates that are associated with specific molecular-level surface processes, like attachment and detachment, using macroscopic laterally averaging methods. We previously showed that total internal reflection fluorescence microscopy (TIRFM) could be used to explicitly count the number of molecules adsorbing with time at the fused silica/water interface and thereby directly calculate the rate of attachment from the near-surface layer without the need to compare with any particular model.26 The rates increased with temperature and (12) Tiberg, F.; Jonsson, B.; Lindman, B. Langmuir 1994, 10 (10), 3714-3722. (13) Tiberg, F. J. Chem. Soc., Faraday Trans. 1996, 92 (4), 531-538. (14) Samoshina, Y.; Nylander, T.; Claesson, P.; Schillen, K.; Iliopoulos, I.; Lindman, B. Langmuir 2005, 21 (7), 2855-2864. (15) Atkin, R.; Craig, V. S. J.; Wanless, E. J.; Biggs, S. J. Colloid Interface Sci. 2003, 266 (2), 236-244. (16) Atkin, R.; Craig, V. S. J.; Biggs, S. Langmuir 2000, 16 (24), 9374-9380. (17) Atkin, R.; Craig, V. S. J.; Biggs, S. Langmuir 2001, 17 (20), 6155-6163. (18) Geffroy, C.; Stuart, M. A. C.; Wong, K.; Cabane, B.; Bergeron, V. Langmuir 2000, 16 (16), 6422-6430. (19) Avena, M. J.; Koopal, L. K. Environ. Sci. Technol. 1999, 33 (16), 2739-2744. (20) Jung, L. S.; Campbell, C. T. J. Phys. Chem. B 2000, 104 (47), 11168-11178. (21) Damos, F. S.; Luz, R. C. S.; Kubota, L. T. Langmuir 2005, 21 (2), 602-609. (22) Sarkar, D.; Somasundaran, P. J. Colloid Interface Sci. 2003, 261 (1), 197-205. (23) Clark, S. C.; Ducker, W. A. J. Phys. Chem. B 2003, 107 (34), 9011-9021. (24) Dukhin, S. S.; Miller, R. Colloid Polym. Sci. 1991, 269 (9), 923-928. (25) Datwani, S. S.; Stebe, K. J. J. Colloid Interface Sci. 1999, 219 (2), 282-297. (26) Honciuc, A.; Howard, A. L.; Schwartz, D. K. J. Phys. Chem. C 2009, 113, 2078-2081.
Published on Web 03/24/2009
DOI: 10.1021/la900307f
7389
Article
Honciuc et al.
were consistent with an activation energy of ∼19 kJ/mol, suggesting for the first time that adsorption from solution is actually an activated process. Since attachment can be considered competitive exchange between an adsorbate molecule and a surfacebound solvent molecule, it seems reasonable to hypothesize that the activation barrier might be related to the strength of the solvent-substrate interaction. This hypothesis is directly tested in the current work by comparing adsorption rates in the presence of three solvents that span the entire range of polarity.
Experimental Methods Fused silica (FS) wafers were sonicated with cationic detergent (Micro 90, International Product Corp.) and thoroughly rinsed with 18 MΩ cm-1 water (Millipore Milli-Q UV+). After detergent cleaning the substrates were immersed in warm piranha solution for ∼1 h, followed by UV-ozone cleaning for 60 min. We have found that this cleaning procedure minimizes both total organic contamination and the amount of fluorescent impurities. Extremely dilute solutions of fluorescently labeled palmitic (aka hexadecanoic) acid (fl-PA: BODIPY FL C16, Invitrogen; see Figure 1) were used in these single-molecule experiments. For each solvent, an appropriate concentration of fl-PA was determined that resulted in a surface concentration that was convenient for single-molecule counting experiments; i.e., adsorption events were frequent enough for efficient counting, but adsorbed molecules were sufficiently separated that they could be easily resolved. The concentrations used were 3 10-9 mol/L in water, 6 10-10 mol/L in tetrahydrofuran (THF, 99.9% Optima grade, Thermo Fisher Scientific, Inc.), and 5 10-12 mol/L in hexadecane (HD, >99%, Sigma Aldrich, Inc.). These concentrations were well below the upper solubility limit reported for hexadecanoic acid in water, ∼2 10-5 mol/L27 (the system is below the Krafft point, so no micelles are expected to form). Upon introducing the solution of interest into the flow cell, the fluorophores were excited by 488 nm radiation from an Ar ion laser (model 543-A-AO3, Melles-Griot Inc.), and the exposure time was controlled with a Uniblitz shutter (model VMM-D3, Oz Optics Ltd.). A prism-based illumination system was used to attain the large angle of incidence required for the total internal reflection condition. The blue excitation light and the green emission light of the fluorophores entering the objective were separated with a dichroic mirror (cut-on wavelength ∼505 nm) and a green filter (band-pass ∼515-555 nm). An electron multiplied-CCD camera (model Cascade-II:512, Photometrics Inc.), cooled to -70 °C, was used as a photon detector. Metamorph 6.3 software (Molecular Imaging, Sunnyvale, CA) was used for the image and movie acquisition, data processing, and shutter controls. The images acquired were 512 512 pixels, and for a 60 magnification the corresponding pixel size was ∼0.07 μm2. Movies showing single-molecule events were acquired during continuous exposure to excitation light; frames were acquired at 2 s time intervals. In these movies, the individual molecules appeared as bright diffraction-limited spots. No fluorescent spots were observed in control experiments with pure solvent. When fl-PA was added, fluorescent spots appeared due to adsorption. Further details of the TIRF microscope, flow cell, general procedures, and data analysis methods used in the current work were presented previously.28
Results Movies showing adsorption events were analyzed frameby-frame to identify and count the appearance of new adsorbed fluorescent probes.26 With HD as a solvent, the adsorbate (27) Lide, D. R. CRC Handbook of Chemistry and Physics, 89th ed.; CRC Press: Boca Raton, FL, 2008. (28) Honciuc, A.; Harant, A. W.; Schwartz, D. K. Langmuir 2008, 24, 6562-6566.
7390 DOI: 10.1021/la900307f
Figure 1. Molecular structures of BODIPY FL C16.
molecules were observed to diffuse randomly on the surface following adsorption.28 With water or THF as solvents, the fluorescently labeled probes were not observed to move with time (within our ability to resolve them). In order to determine the raw adsorption rate, we counted the new spots that appeared in each frame. These data were averaged to determine the mean adsorption rate. Fluorescent spots were also observed to disappear due to desorption and/or photobleaching; however, these events were not considered relevant to the determination of the attachment rates and were not analyzed in detail. Some fluorescent emission from molecules moving in the near surface region reached the camera as background noise; however, these molecules in solution moved too quickly to be resolved under the conditions of these experiments. At least 15 independent movies were analyzed for each solvent. A total of 4202 adsorption events were counted for HD, 3297 for THF, and 3535 for water. The cumulative adsorption kinetics data were calculated as the sum of new molecules appearing as a function of time. Data from representative experiments are shown in Figure 2. In general, the cumulative adsorption was linear with time, and the slope increased with increasing temperature. The data are presented in units of molecules/site, where a site is taken as the approximate cross-sectional area of an adsorbed fl-PA molecule, 0.25 nm2; this approach is standard in the theoretical/computational literature. While in principle one could normalize the adsorption rate by any unit of surface area, it is sensible to normalize the adsorption rate by a physically meaningful surface area rather than by an arbitrary value such as the field of view. In previous work,26 we found that the adsorption rates from water were several orders of magnitude slower than would be expected for diffusion-limited adsorption. This was the case for adsorption from THF and HD as well. Also, for all solvents, the linear dependence of the measured data with time was consistent with an attachment-limited increase in surface coverage (this limit is sometimes referred to as reaction-limited adsorption). These observations suggest that equilibrium is rapidly reached between bulk solution and the near-surface layer and that the adsorption kinetics observed in the experiment is dominated by the rate of attachment from this near-surface region. Since the adsorption rates increased systematically with temperature, it is sensible to analyze these data using a classical Arrhenius approach. As described above, the mean attachment rates were determined by averaging the raw frame-by-frame data. Although the determination of the activation energy is not affected by constant multiplicative prefactors, it is conventional to analyze rate constants as opposed to raw rates. This has the effect of removing the explicit dependence on concentration. Therefore, we have defined an attachment rate constant k using the expression rate = kχ, where χ is the mole fraction of fl-PA. The rate constants calculated using this expression are given in Table 1. The attachment rate constants are presented graphically in Figure 3 in the form of Arrhenius plots; i.e., the natural logarithm of the rate constant is plotted versus the reciprocal of temperature. Langmuir 2009, 25(13), 7389–7392
Honciuc et al.
Article
Figure 2. Cumulative number of fluorescent molecules adsorbed per site (defined as described in the text) as a function of time for representative experiments performed at different temperatures as annotated using the following solvents: (a) hexadecane, (b) tetrahydrofuran, and (c) water. Table 1. Attachment Rate Constants and Activation Energies for Fluorescently Labeled Palmitic Acid on Fused Silica attachment rate constant, k (molecules site-1 s-1) solvent 2
hexadecane (/10 ) tetrahydrofuran water
19 °C
25 °C
30 °C
35 °C
40 °C
2.2 ( 0.2 2.4 ( 0.1 1.1 ( 0.1
2.4 ( 0.1 2.5 ( 0.1 1.3 ( 0.1
2.7 ( 0.1 2.8 ( 0.1 1.6 ( 0.1
2.7 ( 0.1 2.9 ( 0.1 1.8 ( 0.1
3.2 ( 0.1
45 °C
Ea (kJ/mol)
2.6 ( 0.1
5( 2 10 ( 2 19 ( 2
2.2 ( 0.1
However, this error due to photobleaching would be a temperature-independent scaling factor that would not affect the activation energy (Arrhenius) calculation to first order.
Discussion
Figure 3. Arrhenius plots for the attachment rate constants for flPA on FS.
Using classical transition state theory, the slope associated with these data is equal to -Ea/R, where Ea is the activation energy associated with the adsorption process and R is the gas constant. These analyses gave activation energies of 5 ( 2 kJ/mol for adsorption from HD, 10 ( 2 kJ/mol for adsorption from THF, and 19 ( 2 kJ/mol for adsorption from water. In these experiments, it is necessary to consider the possibility that molecules may bleach after such a short surface residence time that they cannot be distinguished from background fluorescence. While it is difficult to quantitatively separate photobleaching from desorption, as previously reported,26 we measured the mean surface residence time as a function of excitation intensity and determined that photobleaching was slow compared to the image acquisition time (and to desorption). Moreover, we note that even if the photobleaching rate were significant, the calculated activation barrier would not be affected. In such a situation, we would fail to count a certain number of molecules that bleached very quickly; this would result in a reduction of the measured absolute adsorption rate. Langmuir 2009, 25(13), 7389–7392
While the kinetics of surfactant adsorption from solution has been thoroughly studied for many years,10-23 until our recent work using single-molecule methods, it was not possible to unambiguously separate the rates of individual molecular processes in order to perform a rigorous mechanistic analysis. By directly counting individual molecular events under steady-state conditions at very low surface coverage (∼10-10), we were able to separate the attachment rate from all other kinetic factors. These measured rates can confidently be assigned as fundamental values associated with the transfer of fl-PA molecules from the nearsurface layer to the surface, i.e., the attachment rate. Using this method, we demonstrated that surfactant adsorption from aqueous solution was an activated process and hypothesized that the activation energy was related to the removal of a surface-bound water molecule prior to displacement by the fatty acid adsorbate. This was a reasonable hypothesis given the spectroscopic evidence for the preferential orientation of water molecules at silica surfaces (reviewed in refs 29 and 30) and the fact that the magnitude of the measured activation barrier (∼19 kJ/mol) was in reasonable agreement with that expected for breaking a OH—O hydrogen bond (∼21 kJ/mol31-35). An obvious test of this hypothesis involved comparing the activation barriers for adsorption of the same adsorbate molecule from solvents that interact with the fused silica surface in dramatically different ways. The fused silica surface is highly polar with a significant density of exposed silanol (29) Schrodle, S.; Richmond, G. L. J. Phys. D: Appl. Phys. 2008, 41 (3). (30) Shen, Y. R.; Ostroverkhov, V. Chem. Rev. 2006, 106 (4), 1140-1154. (31) Curtiss, L. A.; Frurip, D. J.; Blander, M. J. Chem. Phys. 1979, 71 (6), 2703-2711. (32) Feyereisen, M. W. J. Phys. Chem. 1996, 100 (8), 2993-2997. (33) Novoa, J. J.; Sosa, C. J. Phys. Chem. 1995, 99 (43), 15837-15845. (34) Feller, D. J. Chem. Phys. 1992, 96 (8), 6104-6114. (35) Reimers, J. R.; Watts, R. O.; Klein, M. L. Chem. Phys. 1982, 64 (1), 95-114.
DOI: 10.1021/la900307f
7391
Article
groups (∼5 1014 cm-2) that are capable of hydrogen-bonding interactions.36 Thus, the water-surface interaction is expected to be dominated by hydrogen-bonding between the protic solvent and surface hydroxyl (silanol) moieties. At the opposite extreme is the interaction between the nonpolar, hydrophobic hexadecane molecule with the polar, hydrophilic fused silica surface, which is due only to van der Waals interactions. THF represents an intermediate case, where the polar ether group can interact with the surface via dipole-dipole interactions. As one measure of polarity, the dipole moments of HD, THF, and water are ∼0, 1.75, and 1.85 D, respectively.27 However, the calculated dipole moment of individual molecules is not generally considered a good measure of overall solvent polarity because it neglects effects due to molecular correlations (i.e., solvent structure). According to most empirical scales of polarity, e.g., chromatography elution strength parameter37,38 or Snyder solvent polarity parameter,38 alkanes are at the nonpolar extreme, water is at the polar extreme, and THF appears roughly midway between them, typically slightly closer to the nonpolar end of the spectrum. A more quantitative measure of the interaction between the solvent and a silica surface is based on microcalorimetric measurements of the enthalpy of immersion or adhesion. Although these measurements vary due to a sensitivity to surface treatment, the enthalpies of immersion or adhesion of silica are typically 4-6 times more exothermic for water than for alkanes, with the values for aprotic polar solvents lying between the two.39-41 Garnier and co-workers40 measured enthalpies of immersion/adhesion of silica (and other minerals) in a variety of solvents, including heptane, (36) Fan, H. F.; Li, F. P.; Zare, R. N.; Lin, K. C. Anal. Chem. 2007, 79 (10), 3654-3661. (37) Wall, P. E. In Thin-layer Chromatography: A Modern Practical Approach; Royal Society of Chemistry: Cambridge, 2005; pp 93-96. (38) Menet, J.-M.; Thiebaut, D. Characterization of the Solvent Systems Used in Countercurrent Chromatography. In Countercurrent Chromatography; Menet, J.-M., Thiebaut, D., Eds.; Marcel Dekker: New York, 1999; pp 1-28. (39) Douillard, J. M.; Elwafir, M.; Partyka, S. J. Colloid Interface Sci. 1994, 164 (1), 238-244. (40) Garnier, J. M.; Martin, J. M.; Mouchel, J. M.; Chen, M. Colloids Surf., A 1995, 97 (3), 203-215. (41) Malandrini, H.; Sarraf, R.; Faucompre, B.; Partyka, S.; Douillard, J. M. Langmuir 1997, 13 (5), 1337-1341.
7392 DOI: 10.1021/la900307f
Honciuc et al.
diethyl ether, and water; the enthalpies of adhesion were in the approximate ratios 1:2:5, respectively. Since THF is slightly more polar than typical noncyclic ethers, these values again place THF roughly centered between water and alkanes in terms of the strength of interaction with silica. This is in good agreement with our measured values of the apparent activation energy for adsorption of fl-PA in these solvents, which are in the ratios 1:2:4 for HD, THF, and water, respectively. Interestingly, this picture suggests that these activation barriers depend only on solvent and substrate; i.e., they may be independent of the adsorbate molecule to first order. While this is likely to be an oversimplification, it will be interesting to test this hypothesis with another class of adsorbate molecules that interact with the silica surface in a qualitatively different way than do fatty acids.
Conclusions Single molecule TIRFM was used to count fluorescently labeled hexadecanoic (palmitic) acid molecules as they adsorbed at the interface between fused silica and three different solvents. The solvents;hexadecane, tetrahydrofuran, and water;were chosen to span a range of interaction types and strengths with the silica surface. Under steady-state conditions and at extremely low surface coverage, the rates that were determined provided unambiguous measures of the rate at which surfactant molecule were transferred from the subsurface layer to the surface (i.e., the attachment rate). These rates increased with temperature in all cases; Arrhenius analyses gave activation barriers of 5 ( 2, 10 ( 2, and 19 ( 2 kJ/mol for hexadecane, tetrahydrofuran, and water, respectively. The good agreement between these values and the expected energies of the surface-solvent interaction suggests that the activation energy for surface attachment is associated with a transition state involving the displacement of a surface-bound solvent. Acknowledgment. The authors acknowledge financial support from US National Science Foundation Awards CHE-0349547 (A.H. and D.K.S.) and EEC-0552903 (D.J.B.).
Langmuir 2009, 25(13), 7389–7392
