Nanoscale Chemical Composition Analysis Using Peptides Targeting

Feb 1, 2011 - Flucto-Order Functions Research Team, Advanced Science Institute, ... for performing chemical composition analyses of solid surfaces in ...
0 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/Langmuir

Nanoscale Chemical Composition Analysis Using Peptides Targeting Inorganic Materials Yuki Arai,† Ken-Ichiro Okabe,† Hiroshi Sekiguchi,‡ Tomohiro Hayashi,*,†,§ and Masahiko Hara†,§ †

Department of Electronic Chemistry, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, Kanagawa 226-8502, Japan ‡ Graduate School of Frontier Sciences, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa City, Chiba 277-8561, Japan § Flucto-Order Functions Research Team, Advanced Science Institute, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan

bS Supporting Information ABSTRACT: Chemical composition analysis by scanning probe microscopy (SPM) in water is a method whose introduction has been long-awaited. Here we propose a simple method for performing chemical composition analyses of solid surfaces in water using atomic force microscopes (AFMs) with probes functionalized with peptides targeting inorganic materials. In this work, bicompositional surfaces of gold and titanium oxide were scanned with AFM probes modified with the titanium-binding peptide (TBP). We found that surface chemical composition clearly appeared as contrast in the mapping images of adhesion forces with nanometer-scale resolution. In this Article, we further discuss appropriate designs of the AFM probes and appropriate imaging conditions for the chemical composition analysis based on the results of force measurements of the single TBP-titanium bond.

’ INTRODUCTION In the fields of nanoscience and nanotechnology, it has been considered to be important to develop techniques for observing surface morphology and the shapes of objects on solid surfaces with nanometer spatial resolutions while simultaneously acquiring their chemical information. Scanning electron microscopy combined with energy-dispersive X-ray analysis (SEM-EDX) and transmission electron microscopy integrated with electron energy loss spectroscopy (TEM-EELS) are powerful techniques for observing nanosized objects along with chemical (elemental) analysis. However, these techniques function under vacuum conditions and require vacuum-compatible samples. Therefore, these techniques cannot be used for in situ measurements in the fields of electrochemistry and colloid science. Since the invention of scanning tunneling microscopy (STM) in 1981,1 scanning probe microscopes (SPMs) have evolved as powerful imaging devices for obtaining nanometer-resolved topographic information on various objects in vacuum, air, and liquids. The objects investigated with the SPMs include solid materials, polymers, cells, proteins, biological tissues, viruses, and so on.2,3 A recent progress in SPMs is the introduction of methods to analyze local elemental properties with a variety of approaches having been reported in the past 5 years. These approaches include STM measurements assisted by a synchrotron light radiation,4 the discrimination of atoms based on the local interaction between the probe and substrate as measured by noncontact atomic force microscopy (NC-AFM),5,6 NC-AFM r 2011 American Chemical Society

assisted by X-ray irradiation.7 However, these techniques function only under ultrahigh vacuum, and the measurements cannot be performed with commercially available SPMs. Here we report an approach to perform chemical composition imaging with nanometer-scale resolution in water by utilizing peptide aptamers (binders) targeting inorganic materials. The applications of peptide aptamers have been expanding because of their specific affinities to their target materials, and they were already employed to build various interfaces between biomolecules and inorganic materials.8-10 In particular, the specificity of several peptide aptamers was examined by comparing the amounts of the peptides or peptide-modified biomolecules adsorbed on the surface, and the strong dependence of the amounts on substrates was confirmed. On the basis of the above findings, we employed chemical force microscopy (CFM), which maps interactions originating within probe-surface contact such as hydrogen bonding, host-guest interaction, capillary force due to water ad-layers, and antibody-antigen interaction.11-16 In this work, we investigated the chemical imaging in water by using AFM probes modified with the Ti-binding peptide (TBP)17,18 because our recent AFM measurements clearly revealed that the specificity and selectivity of the TBP appeared as a clear contrast in the adhesion force depending on the substrate.19,20 We mainly focus Received: October 17, 2010 Revised: December 12, 2010 Published: February 01, 2011 2478

dx.doi.org/10.1021/la104178h | Langmuir 2011, 27, 2478–2483

Langmuir

ARTICLE

Figure 2. Deflection of the cantilever as a function of time. Contact time (τc) was defined with a double-sided arrow. Figure 1. Primary species immobilized on three types of probes used in this work. Type I: TBP is immobilized directly on the Au-coated probe surface via gold-sulfur (cysteine residue) bond. Type II: A spacer of a poly(ethylene glycol) was introduced between the probe surface and TBP. Type III: The density of TBP on the probe surface was reduced by mixing a methoxy-terminated PEG chain.

on the following three issues: the design of AFM probes suitable for the chemical imaging, adequate conditions for the chemical imaging (especially duration of the probe-surface contact), and the character of the bond between the titanium oxide surface and TBP. We performed the dynamic force spectroscopy (DFS) measurements of the single TBP-Ti bond and discussed the adequacy of peptide aptamers for the chemical composition imaging.

’ EXPERIMENTAL SECTION Peptide Molecule. For the TBP, we purchased a commercially synthesized peptide (RKLPDAPGMHTWC, purity 95%, Biotech Lab, Japan). The cysteine residue was added to the C-terminal of the TBP-1 reported by Sano et al.18 to immobilize the peptide to an Au-coated probe via a gold-sulfur bond or to a spacer of poly(ethylene glycol) (PEG) by a coupling reaction between the maleimide and thiol. AFM and Probes. In this work, we employed a commercial AFM system equipped with a liquid cell (MFP-3D, Asylum Research, Santa Barbara, CA). We used three types of TBP-modified AFM probes, and their chemical structures are illustrated in Figure 1. Details of the preparation of the AFM probes are described in the Supporting Information, and we here describe the outline. For the Type I probe, the TBP was directly fixed to the Au-coated probe. For the Type II probe, a spacer moiety of the PEG chain (24 EG units) was introduced between the probe and TBP. As for the Type III probe, the density of TBP on the surface of the probe was reduced by mixing a methoxyterminated PEG spacer during the introduction of the PEG spacer (Supporting Information). The density of the peptide of Type III probe was estimated to be 10 to 15% of that of the Type II probe from the results of FT-IR reflection absorption spectroscopy measurements (from peak areas of amide I and II bands) of the Au substrates modified with the same procedure. For the preparation of the Type II and III probes, NHS- and maleimide-mediated conjugations were used. The Type I and II probes, whose spring constants were relatively high (60 and 120 pN/nm, NP-S, Veeco, Santa Barbara, CA), were used for the chemical imaging. As for the Type III probe, a soft cantilever (spring constant 6 pN/nm, Biolever, Olympus, Tokyo) was used to minimize an effect of thermal fluctuation of the cantilever on the observed rupture force of a single peptide molecule.

Pure Ti and Ti-Au Patterned Substrates. Ti substrates were prepared as follows. First, Si(100) wafers (1 mm thick, n-type, SUMCO, Tokyo, Japan) were cut into small pieces of about 10  10 mm, and a Ti layer of 30 nm was thermally evaporated at a growth rate of ∼0.1 nm/sec in high vacuum (base pressure, 1.0  10-4 Pa). For the patterned Ti-Au substrates, silica microspheres (diameter 4 μm, Polysciences, PA) dispersed in water were spread on the Ti substrates. After the substrate was dried in air, Au was evaporated onto it with a thickness of 50 or 100 nm at a rate of 0.1 nm/sec. Then, silica beads were removed by ultrasonication in pure water for a few seconds. The rms roughnesses of the substrate in Au and Ti areas were about 1.4 and 0.7 nm, respectively. General Force Measurements. All force measurements were carried out at room temperature in an aqueous solution containing Tween20 (0.05 wt %) and sodium chloride (10 mM) to minimize the nonspecific interactions between the probe and the substrate.21-23 The spring constants of the cantilevers were determined by monitoring the thermal noise of the cantilever.24 For the conversion of the deflection of the cantilever to the probe-surface separation, we simply defined the separation of zero as the point where linearity in the constant compliance region started in the force-displacement curve. Adhesion Force Mapping. Chemical imaging was carried out by force mapping with the Type I and II probes with bicompositional Ti-Au-patterned substrates. Force curves were taken at 64  64 points for each image, and topographic information and the adhesion force were recorded simultaneously. The relative trigger mode was used to keep the maximum loading force between the tip and the substrate ∼300 pN to avoid mechanical damage to the molecules immobilized on the probe. Analysis of the Kinetics of the Formation of the TBP-Ti Bond and Dynamic Force Spectroscopy. For the analysis of the kinetics of the formation of the TBP-Ti bond and DFS measurements of the single TBP-Ti bond, we used the Type III probe. For the analysis of the kinetics of the bond formation, we measured adhesion forces with different duration times for the probe in contact with the Ti surface (τc) from 0.03 to 10.03 s while keeping the loading force at 100 pN. The definition of τc is given in Figure 2. Approach and retraction velocities were fixed at 1000 nm/sec. We also investigated the dependence of the rupture force on the loading rate (so-called DFS) For this analysis, the retract velocity was varied from 30 to 10 000 nm/sec, whereas the velocity of approach was fixed at 1000 nm/sec. Other conditions were the same as those in the experiments on adhesion force mapping. Force curves were captured at a minimum of 256 different positions in the area of 20  20 μm. We extracted the rupture events and adhesion forces from the measured force-distance curves using a macro program written in Microsoft Excel software (Supporting Information). The histograms of the adhesion forces obtained at each loading rate (νf) were fitted by 2479

dx.doi.org/10.1021/la104178h |Langmuir 2011, 27, 2478–2483

Langmuir

Figure 3. (a) Topographic and (b) adhesion force images (10  10 μm) of the patterned Ti-Au surfaces obtained by the Type I probe.

Figure 4. Force-separation curves obtained with the Type I and II probes. multiple Gaussian functions (Supporting Information), and we assigned the mean force corresponding to the weakest interaction (first peak in the adhesion force histogram) as the most probable adhesion force of a single molecule (f*). The loading rate was calculated from the linear fitting of the slope of the force-time plot just before the rupture event (the length of the region was 3 nm).

’ RESULTS AND DISCUSSION Figure 3a shows a topographic image of the patterned Ti-Au substrate. The size of the circle and height of the bumps are exactly consistent with the diameter of the silica spheres (4 μm) and the thickness of the evaporated gold film (4 nm), respectively, indicating that Au and Ti areas were distinctly formed with a clear material contrast. Figure 3b shows the adhesion force image simultaneously recorded in the same region as Figure 3a. Despite the use of the probe modified with TBP (Type I), it was difficult to distinguish the Au and Ti areas from the adhesion force. We found that the probability of observing adhesion forces greater than 100 pN in the Ti region (denoted as p(fad > 100 pN) hereafter) is ∼10%. We next performed chemical imaging using the Type II probe, which possesses PEG spacer moieties between the probe and

ARTICLE

Figure 5. (a) Topographic and (b) adhesion force images (5  5 μm) of the patterned Ti-Au surfaces obtained by the Type II probe.

TBP. Representative force-separation curves obtained with the Type I and II probes observed in the Ti area are displayed in Figure 4. Comparing these two curves, the difference in the rupture lengths is obvious. The mean rupture length observed with the Type I probe was ∼2 nm, whereas that observed with the Type II probe was ∼9 nm (close to the theoretical length of the molecule immobilized on the probe in the all-trans configuration (9.4 nm)), ensuring that the observed adhesion force originated from the specific TBP-Ti interaction and not from other nonspecific interactions such as a hydrophobic interaction due to contaminants. As clearly seen in Figure 5, p(fad > 100 pN) increased up to 33% through the use of the Type II probe. We assume that the PEG spacer endowed a high degree of freedom in the structure and orientation of the TBP moiety, resulting in a high probability for the formation of the bond between TBP and Ti. We next discuss the duration time for the contact between the probe and substrate (τc) because we found that the observed adhesion force, which reflects the number of TBP-substrate bonds, critically depended on τc. Figure 6 shows the mean adhesion force (favr) and probability for observing adhesion [p(fad > 0 pN)] as a function of τc. In this experiment, the Type III probe was employed. As can be clearly seen, p(fad > 0 pN) dramatically increased up to ∼0.9 at τc = 1 and reached almost 1 at τc = 3. The time dependence of favr exhibited behavior similar to that of p(fad > 0 pN). As shown in Figure 7, the histograms for the observed adhesion forces at different τc values provide deeper insight into the kinetics of the formation of TBP-Ti bonds. At τc = 0.03, the rupture events of the single bond (∼60 pN) were dominant, and the ruptures of the double and triple bonds (about 120 and 200 pN, respectively) became major at τc = 1.03. These results suggest that higher contrasts in the adhesion force mapping images can be expected when we employ τc values longer than 1 s. Considering the time scales of molecular motion, the observed kinetics of the formation of the TBP-Ti bond seem to be too 2480

dx.doi.org/10.1021/la104178h |Langmuir 2011, 27, 2478–2483

Langmuir

ARTICLE

Figure 6. Probability of observing adhesion [p(fad > 0 pN)] and mean adhesion forces (favr) as a function of contact time (τc).

Figure 8. (a) Topographic and (b) adhesion force images (5  5 μm) of the patterned Ti-Au surfaces obtained by the Type I probe. The contact time (τc) was set at 3 s.

Figure 7. Histograms of the adhesion forces as a function of contact time (τc) (a) 0.03, (b) 0.09, and (c) 1 s.

slow because time scales for changes in the structural configurations of peptides are on the order of several to several tens of nanoseconds.25 We anticipate that the fixation of the TBP molecule to the probe and the confinement of the TBP moieties between the probe and substrate result in the low degree of freedom of the configuration of the TBP moieties because we

previously reported that the structure of the TBP molecule is a critical factor in its specific binding with Ti surfaces.20 The extremely slow kinetics of the bond formation were also observed for the system of selectin protein and its ligands.26 The results of the force mapping obtained with a longer τc value with the Type II probe are shown in Figure 8. Compared with Figure 5, the difference in the contrasts in the adhesion force images are obvious. In particular, in the case of τc equal to 3.06 s, p(fad > 100 pN) and p(fad > 0 pN) are 94 and 99%, respectively, indicating that the Au and Ti areas were perfectly distinguished by the force mapping using the probe modified with TBP. We here mention that the spatial resolution of this chemical imaging, which is considered to be governed mainly by the radius of the AFM probe, including the tip radius and the length of the PEG and TBP moieties, is below 100 nm. It should be noted here that no contrast in adhesion force images was observed with other probes, such as bare Si3N4, NH2-terminated, and PEG-terminated probes in our control experiments. We also performed the adhesion force mapping with the Type I probe at τc = 3.03. However the p(fad > 0 pN) was only 50%, indicating that the degree of freedom of the peptide is important for the formation of the TBP-Ti bond. We next consider the character of the single TBP-Ti bond to obtain prospects of the application of peptide aptamers for chemical composition imaging. The dependence of the most frequently observed rupture force of the single TBP-Ti bond (f*TBP) on the loading rate was investigated. In this experiment, the Type III probe was used. f*TBP was obtained by fitting histograms with Gaussian functions, and we regarded the first peak as the rupture force of the single bond. The obtained values of f*TBP were fitted to the Bell-Evans theoretical expression for rate dependence ! τð0Þ 3 xβ k T k T B B f ¼ lnðvf Þ þ ln ð1Þ xβ xβ kB T 2481

dx.doi.org/10.1021/la104178h |Langmuir 2011, 27, 2478–2483

Langmuir

ARTICLE

At least 40 min to 1 h is required to acquire one image of the chemical mapping when we set τc to 1 to 3 s. One alternative imaging approach is AC mode imaging, with which topographic and phase shift images can be captured simultaneously. With this method, however, the resonant curve for the cantilever is expected to have a low Q value in water. Therefore, we anticipate that the AC mode imaging with driving of the cantilever with magnetic force, which succeeded in the imaging of single molecular recognition events, may be adequate for the fast imaging.34,35

Figure 9. Most frequently observed rupture force of a single TBP-Ti bond (f*TBP) as a function of a loading rate (vf) and the fitting result to the Bell-Evans model (eq 1). The loading rate used for chemical composition mapping in this work is indicated by the arrow.

where f*, kBT, xβ, vf, and τ(0) are the most frequently observed adhesion force, the Boltzmann thermal energy, the distance to the transition state, the force loading rate, and the natural lifetime of the bond (the inverse of τ(0) corresponds to the kinetic offrate, koff), respectively. As can be clearly seen, f*TBP increased monotonically as a function of vf, suggesting that the unbinding is a dynamic process and that higher vf values result in a strong adhesion between TBP and Ti and are adequate for chemical imaging. Therefore, we employed the relatively higher vf values in this work (indicated with the arrow in Figure 9). In the general interpretation of the Bell-Evans theory, the number of slopes with different gradients corresponds to the number of potential barriers between the bound and unbound states along the reaction coordinate.27-32 As far as we investigated, only one slope was recognized for the TBP-Ti system. We previously reported that two electrostatic interactions between the charges of the TBP and Ti surface and one hydrogen bond are responsible for the specific TBP-Ti bond.19,20 Meanwhile, the results of computer simulations by Skelton et al. suggested that the charged residues are bound to the charged surface groups of the Ti substrate bridged by the water molecules between them.33 Our results suggest that these bonds cause the single potential barrier of the TBP-Ti bond and rupture simultaneously under the loading force. The fitting results showed that xβ and τ(0) are 0.27 nm and 0.99 s, respectively, indicating that the TBP-Ti bond is relatively weak compared with specific biomolecular interactions studied so far.31,32 On the basis of the above discussion, the applicability of other peptide aptamers to chemical imaging should be considered because peptide aptamers targeting various materials have been previously reported so far. Our results indicate that chemical composition analysis is possible if several noncovalent bonds are involved in the specific interaction between a peptide and its target. The amino acid sequences of the previously reported peptides targeting inorganic materials show that the peptides possess several polar and charged groups, indicating that several noncovalent bonds are responsible for their specific interactions with their targets. We therefore surmise that other peptide aptamers can be used for the chemical imaging. The applicability of the other peptide aptamers is still under investigation. Finally, regarding the imaging mode, we employed the method of force mapping in this study, which is a time-consuming approach.

’ SUMMARY In summary, we performed chemical composition imaging of the metal surfaces in water for the first time. The fixation of the TBP and the contact time are found to be critical factors in acquiring chemical composition images with a clear contrast. We believe that this approach can enable us to perform in situ observations of the surface chemical composition in water, especially in the field of electrochemistry and colloid science, where conventional methods of chemical composition analysis have not been successfully applied. ’ ASSOCIATED CONTENT

bS

Supporting Information. List of chemicals used in this work, preparation of cantilevers modified with TBP, filtering of the force curves, histogram of adhesion force obtained with the type II probe, and chemical composition images at a different location and histograms of the force measurements of the TBP-Ti bond at different loading rates. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT This work was partially supported by a Grant-in-Aid for Young Scientists (B) from MEXT. Dr. Ken-Ichi Sano and Dr. Yoshikazu Kumashiro are gratefully acknowledged for fruitful discussions regarding this research. ’ REFERENCES (1) Binning, G.; Rohrer, H.; Gerber, C.; Weibel, E. Phys. Rev. Lett. 1982, 49, 57–61. (2) Scanning Probe Microscopies Beyond Imaging: Manipulation of Molecules and Nanostructures; Samori, P., Ed.; Wiley-VCH: Weinheim, Germany, 2006. (3) Force Microscopy: Applications in Biology and Medicine; Jena, B. P., H€ orber, J. K. H., Eds.; Wiley: Hoboken, NJ, 2006. (4) Okuda, T.; Eguchi, T.; Akiyama, K.; Harasawa, A.; Kinoshita, T.; Hasegawa, Y.; Kawamori, M.; Haruyama, Y.; Matsui, S. Phys. Rev. Lett. 2009, 102, 105503. (5) Sugimoto, Y.; Pou, P.; Abe, M.; Jelinek, P.; Perez, R.; Morita, S.; Custance, O. Nature 2007, 446, 64–67. (6) Sugimoto, Y.; Namikawa, T.; Abe, M.; Morita, S. Appl. Phys. Lett. 2009, 94, 023108. (7) Suzuki, S.; Koike, Y.; Fujikawa, K.; Chun, W. J.; Nomura, M.; Asakura, K. Chem. Lett. 2004, 33, 636–637. (8) Sarikaya, M.; Tamerler, C.; Jen, A. K. Y.; Schulten, K.; Baneyx, F. Nat. Mater. 2003, 2, 577–585. (9) Tamerler, C.; Sarikaya, M. Acta Biomater. 2007, 3, 289–299. 2482

dx.doi.org/10.1021/la104178h |Langmuir 2011, 27, 2478–2483

Langmuir

ARTICLE

(10) Tamerler, C.; Khatayevich, D.; Gungormus, M.; Kacar, T.; Oren, E. E.; Hnilova, M.; Sarikaya, M. Biopolymers 2010, 94, 78–94. (11) Noy, A. Surf. Interface Anal. 2006, 38, 1429–1441. (12) Noy, A.; Vezenov, D. V.; Lieber, C. M. Annu. Rev. Mater. Sci. 1997, 27, 381–421. (13) Vezenov, D. V.; Noy, A.; Ashby, P. J. Adhes. Sci. Technol. 2005, 19, 313–364. (14) Vezenov, D. V.; Noy, A.; Rozsnyai, L. F.; Lieber, C. M. J. Am. Chem. Soc. 1997, 119, 2006–2015. (15) Kada, G.; Kienberger, F.; Hinterdorfer, P. Nano Today 2008, 3, 12–19. (16) Muller, D. J.; Krieg, M.; Alsteens, D.; Dufrene, Y. F. Curr. Opin. Biotechnol. 2009, 20, 4–13. (17) Sano, K.; Sasaki, H.; Shiba, K. Langmuir 2005, 21, 3090–3095. (18) Sano, K.; Shiba, K. J. Am. Chem. Soc. 2003, 125, 14234–14235. (19) Hayashi, T.; Sano, K.; Shiba, K.; Kumashiro, Y.; Iwahori, K.; Yamashita, I.; Hara, M. Nano Lett. 2006, 6, 515–519. (20) Hayashi, T.; Sano, K. I.; Shiba, K.; Iwahori, K.; Yamashita, I.; Hara, M. Langmuir 2009, 25, 10901–10906. (21) Brogan, K. L.; Shin, J. H.; Schoenfisch, M. H. Langmuir 2004, 20, 9729–9735. (22) Krautbauer, R.; Rief, M.; Gaub, H. E. Nano Lett. 2003, 3, 493– 496. (23) Zhang, Y. H.; Liu, C. J.; Shi, W. Q.; Wang, Z. Q.; Dai, L. M.; Zhang, X. Langmuir 2007, 23, 7911–7915. (24) Hutter, J. L.; Bechhoefer, J. Rev. Sci. Instrum. 1993, 64, 1868– 1873. (25) Heinz, H.; Farmer, B. L.; Pandey, R. B.; Slocik, J. M.; Patnaik, S. S.; Pachter, R.; Naik, R. R. J. Am. Chem. Soc. 2009, 131, 9704–9714. (26) Lu, S. Q.; Ye, Z. Y.; Zhu, C.; Long, M. Polymer 2006, 47, 2539– 2547. (27) Bell, G. I. Adv. Appl. Probab. 1980, 12, 566–567. (28) Evans, E. Faraday Discuss. 1998, 111, 1–16. (29) Evans, E. Annu. Rev. Biophys. Biomol. Struct. 2001, 30, 105–128. (30) Evans, E.; Ritchie, K. Biophys. J. 1997, 72, 1541–1555. (31) Lee, C. K.; Wang, Y. M.; Huang, L. S.; Lin, S. M. Micron 2007, 38, 446–461. (32) Bizzarri, A. R.; Cannistraro, S. Chem. Soc. Rev. 2010, 39, 734– 749. (33) Skelton, A. A.; Liang, T. N.; Walsh, T. R. ACS Appl. Mater. Interfaces 2009, 1, 1482–1491. (34) Han, W. H.; Lindsay, S. M.; Jing, T. W. Appl. Phys. Lett. 1996, 69, 4111–4113. (35) Stroh, C.; Wang, H.; Bash, R.; Ashcroft, B.; Nelson, J.; Gruber, H.; Lohr, D.; Lindsay, S. M.; Hinterdorfer, P. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 12503–12507.

2483

dx.doi.org/10.1021/la104178h |Langmuir 2011, 27, 2478–2483