Langmuir 2004, 20, 1647-1655
1647
Long-Range Order of Organized Oligonucleotide Monolayers on Au(111) Electrodes Hainer Wackerbarth, Mikala Grubb, Jingdong Zhang, Allan G. Hansen, and Jens Ulstrup* Department of Chemistry, Building 207, Technical University of Denmark, DK-2800 Lyngby, Denmark Received August 21, 2003. In Final Form: December 16, 2003 Oligonucleotides modified by a hexamethylene linker group adsorb on gold electrodes via Au-S bond formation. We have obtained novel data for adsorption of thiol-modified (HS) single-strand HS-10A and double-stranded HS-10AT oligonucleotides and for analogous thiol-free 10A (A ) adenine) and 10T (T ) thymine) nonspecifically adsorbed as reference molecules. Mercaptohexanol has served as a second reference molecule. The data are based on cyclic and differential pulse voltammetry, interfacial capacitance data, and in situ scanning tunneling microscopy (STM) directly in an aqueous buffer solution, with electrochemical potential control of both the sample electrode and the tip. All the data are based on single-crystal, atomically planar Au(111)-electrode surfaces. The high sensitivity of such surfaces provides accurate HS-10A and HS-10AT electrode coverages on the basis of the reductive desorption of the Au-S bond. The coverage is high and in keeping with dense monolayers of adsorbed HS-10A and HS-10AT in an upright or tilted orientation, with the oligonucleotide backbone repelled from the strongly negatively charged electrode surface. Adsorbed thiol-free 10A only gives a Au(111)-reconstruction peak, while 10T shows a subtle pattern involving pronounced voltammetric adsorption peaks indicative of both nonspecific adsorption via single thymine units and potential-dependent structural reorganization in the surface layer. In situ STM supports these findings at the molecular level. In situ STM of HS-10A discloses large, highly ordered domains at strongly negative sample potentials. Reversible domain formation and disordering could, moreover, be controlled by an electrochemical potential variation in the negative and positive directions, respectively. 10A and 10T did not form ordered adsorbate domains, substantiating that domain formation rests on adsorption of thiol-modified oligonucleotide adsorption in an upright or tilted orientation. The comprehensive, high-resolution information reported may hold prospects for single-molecule electronic conduction and molecular-scale mapping of oligonucleotide hybridization.
Introduction DNA-based molecules are uniquely organized molecular structures based on hydrogen bonding and base stacking of the component single-strand (SS) oligo- and polynucleotides into functional double-strand (DS) molecules.1 The past decade has witnessed wide-ranging attention to the multifarious physical and chemical properties of DNAbased molecules and the biotechnological development of the notion of DNA-based “chips” for gene-specific biological screening.2-5 Other high-technology perspectives are rooted in the single-molecule, highly base-pair-specific electronic conductivity of DNA-based oligonucleotides, and the vision of surface-immobilized DNA-based structures as electronic components of nanoscale functional devices.6-13 * Author to whom correspondence should be addressed. Fax no. +45 45883136, telephone no. +45 45252359, e-mail
[email protected]. (1) Sinden, R. R. DNA Structure and Function; Academic Press: New York, 1994. (2) Service, R. F. Science 1998, 282, 396-399. (3) Southern, E.; Mir, K.; Shchepinov, M. Nat. Genet. 1999, 21, 5-9. (4) Fritz, J.; Baller, M. K.; Lang, H. P.; Rothuizen, H.; Vettiger, P.; Meyer, E.; Gu¨ntherodt, H.-J.; Gerber, C.; Gimzewski, J. K. Science 2000, 288, 316-318. (5) Bashir, R. Superlattices Microstruct. 2001, 29, 1-16. (6) Kelley, S. O.; Barton, J. K.; Jackson, N. M.; McPherson, L. D.; Potter, A. B.; Spain, E. M.; Allen, M. J.; Hill, M. G. Langmuir 1998, 14, 6781-6784. (7) Kelley, S. O.; Jackson, N. M.; Hill, M. G.; Barton, J. K. Angew. Chem., Int. Ed. 1999, 38, 941-945. (8) Giese, B. Acc. Chem. Res. 2000, 33, 631-636. (9) Jortner, J.; Bixon, M.; Langenbacher, T.; Michel-Beyerle, M. E. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 12759-12765.
In DNA-based biotechnology toward the nanoscale and ultimately single-molecule levels, functional DNA-based molecules must be immobilized on well-characterized solid surfaces with hybridization, electron transfer (ET), and other properties retained. Chemical surface linking via attached, thiol-based linker groups has reached a high level.14-18 However, the vacuum or air environment in putative solid-state DNA-based electronics raises obvious problems caused by the conformationally flexible SS- and DS-structures, the role of counterions around DNA-based polyanions, and the electronic contacts to the solid surface. Biological or medical DNA-based “chip” technology is based on immobilized DNA fragments in aqueous buffer, which is the natural functional medium for DNA-based molecules. In contrast to notions of DNA-based molecules as (10) Bixon, M.; Jortner, J. J. Phys. Chem. B 2000, 104, 3906-3913. (11) Bixon, M.; Jortner, J. Chem. Phys. 2002, 281, 393-408. (12) Davis, W. B.; Hess, S.; Naydenova, I.; Haselsberger, R.; Ogrodnik, A.; Newton, M. D.; Michel-Beyerle, M. E. J. Am. Chem. Soc. 2002, 124, 2422-2423. (13) Berlin, Y. A.; Burin, A. L.; Ratner, M. A. Superlattices Microstruct. 2000, 28, 241-252. (14) Patel, R.; Lin, C.; Laney, M.; Kurn, N.; Rose, S.; Ullman, E. F. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 2969-2974. (15) Herne, T. M.; Tarlov, M. J. J. Am. Chem. Soc. 1997, 119, 89168920. (16) Petrovykh, S. Y.; Kimura-Suda, H.; Whitman, L. J.; Tarlov, M. J. J. Am. Chem. Soc. 2003, 125, 5219. (17) Demers, L. M.; Mirkin, C. A.; Mucic, R. C.; Reynolds, R. A.; Letsinger, R. L.; Elghanian, R.; Viswanadham, G. Anal. Chem. 2000, 72, 5535. (18) O’Brien, J. C.; Stickney, J. T.; Porter, M. D. J. Am. Chem. Soc. 2000, 122, 5004-5005.
10.1021/la035547g CCC: $27.50 © 2004 American Chemical Society Published on Web 01/28/2004
1648
Langmuir, Vol. 20, No. 5, 2004
potential electronic circuit elements, DNA-based molecular screening of biological liquids has, therefore, evolved into real devices.2-5 The spatial resolution of this approach to DNA-based chip technology is, however, at the micrometer level and remote from ultimate single-molecule levels. This is reflected in a second limitation, that is, the method of monitoring the functional response of the DNAbased chip such as fluorescence, radiolabeling, or microcantilever deflection,19 which are essentially statistically averaged, macroscopic responses. Electrochemical and in situ scanning probe technology, particularly scanning tunneling microscopy (STM) and atomic force microscopy, directly in aqueous buffer hold prospects for bringing high-resolution mapping of DNAbased molecules on metallic surfaces forward. Novel theoretical frames for the ET patterns of in situ STM have also been provided.20-22 Single-crystal, atomically planar electrode surfaces are crucial. This notion is established in physical electrochemistry. It is novel in interfacial bioelectrochemistry of large biological molecules but developing here as a powerful high-resolution tool in redox metalloprotein electrochemistry.23-25 A second crucial notion toward single-molecule in situ STM mapping of DNA-based molecules is that some level of supramolecular order must prevail. Otherwise imaged structures are too conformationally mobile for biomolecules in chemical action to be assigned. This needs subtle control of external parameters, particularly, the electrochemical potential. The molecular orientation and the accessibility of the oligonucleotides play key roles in DNA hybridization biosensors. Subtle control of the probe nucleotide immobilization is here required. Electric fields, caused by an electrochemical electrode, offer a unique tool. In this report, we have exploited the high sensitivity of singlecrystal electrochemistry and the high resolution of in situ STM. We have addressed oligonucleotides with 10 adenine (A) and 10 thymine (T) nucleobases. The SS oligonucleotides are denoted as 10A and 10T, respectively. Most of the focus is on 10A, to which a hexamethylene thiol linker is covalently attached to the 5′-end of the 10A strand, denoted as HS-10A. The single-stranded thiol-anchored HS-10A oligonucleotide is compared with thiol-linked HS10A hybridized with the complementary 10T oligonucleotide, denoted as HS-10AT. The thiol linker alone, 6-mercapto-1-hexanol (MCH), was finally used as a reference molecule. The oligonucleotide strands are short enough that detailed information about their interfacial structural and dynamic behavior can be obtained compared with the complex macromolecular structure of longer oligonucleotides. They are, however, long enough to offer insight into oligonucleotide collective properties and novel structural features compared with single nucleotide bases, (19) Raiteri, R.; Grattarola, M.; Butt, H.-J.; Skla´dal, P. Sens. Actuators, B 2001, 79, 115-126. (20) Friis, E. P.; Anderson, J. E. T.; Kharkats, Y. I.; Kuznetsov, A. M.; Nichols, R. J.; Zhang, J.; Ulstrup, J. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 1379-1384. (21) Kuznetsov, A. M.; Ulstrup, J. J. Phys. Chem. A 2000, 104, 1153111541. (22) Zhang, J.; Kuznetsov, A. M.; Ulstrup, J. J. Electroanal. Chem. 2003, 541, 133-146. (23) Zhang, J.; Chi, Q.; Kuznetsov, A. M.; Hansen, A. G.; Wackerbarth, H.; Christensen, H. E. M.; Andersen, J. E. T.; Ulstrup, J. J. Phys. Chem. B 2002, 106, 1131-1152. (24) Hansen, A. G.; Boisen, A.; Nielsen, J. U.; Wackerbarth, H.; Chorkendorff, I.; Andersen, J. E. T.; Zhang, J.; Ulstrup, J. Langmuir 2003, 19, 3419-3427. (25) Brask, J.; Wackerbarth, H.; Jensen, K. J.; Zhang, J.; Chorkendorff, I.; Ulstrup, J. J. Am. Chem. Soc. 2003, 125, 94-104.
Wackerbarth et al.
for which interfacial properties on Au(111) surfaces, particularly, of thymine, have been mapped in great detail.26-28 Material and Methods Reagents. Oligonucleotides were from TAG Copenhagen. MCH (97%) was from Aldrich Chemical Co. NaOH, K2HPO4, and KH2PO4 were of suprapure quality. Millipore water (Milli-Q Housing) was used throughout the experiments. Preparation of Single-Crystal Gold Electrodes. Singlecrystal gold electrodes for electrochemistry were prepared as bead electrodes by the method of Clavilier et al.29 and Hamelin30 or acquired from Surface Preparation Laboratory (The Netherlands). A Au(111) disk (Surface Preparation Laboratory, 10-mm diameter and 1-mm thick) was used as the substrate in STM. The bead electrodes and substrate, respectively, were electropolished in 0.1 M H2SO4 (+10 V, followed by soaking in 1 M HCl) and annealed in an oven at 850 °C for 6 h. Prior to use, the disk or bead electrodes were annealed for 2 min in a hydrogen flame. The disk substrates were cooled to room temperature in air. The bead electrodes were cooled above a Millipore water surface, followed by immersion into the water. The substrate or electrode was, finally, transferred to an aqueous solution of oligonucleotides or MCH. The quality of all the single-crystal gold electrodes was checked on a regular basis by recording cyclic voltammograms in 0.1 M H2SO4. Agreement with reported voltammograms at Au(111)electrode surfaces was taken as a satisfactory quality check.29 Sample Preparation. The disk or bead electrodes were immersed in a buffer, 1 M K2HPO4/KH2PO4, pH 6.9, containing 1 µM oligonucleotide, if not stated otherwise, and soaked for about 18 h at 5 °C. Only bead electrodes were used for interfacial capacitance, cyclic voltammetry (CV), and differential pulse voltammetry (DPV). The concentration of the HS-10AT for interfacial capacitance was 1 or 5 µM. For in situ STM and CV of HS-10A, the disk and bead electrodes were immersed in 0.01 M K2HPO4/KH2PO4, pH 6.9. MCH, 1 mM, was dissolved in H2O and adsorbed for several hours at room temperature. After completion of the surface immobilization, the samples were thoroughly rinsed with Millipore water. Glassware and other utensils were cleaned as previously described.31 Voltammetry and Capacitance Measurements. The hanging meniscus method was used in voltammetric and capacitance measurements. CV, DPV, and interfacial capacitances were recorded using an Autolab system (Eco Chemie, The Netherlands) controlled by the GPES software. The parameters in DPV and capacitance measurements were the same as previously described,31,32 that is, 10 mV s-1 scan rate and +4.05 mV step potential in DPV, and 100 Hz and a +5 mV modulation amplitude in the capacitance measurements. A coiled bright platinum wire and a saturated calomel electrode (SCE) were used as the counter and reference electrodes, respectively. All potentials refer to the SCE. 0.1 M K2HPO4/KH2PO4 buffer at pH 6.9 was the medium for the electrochemical measurements. The solutions were deoxygenated by bubbling purified argon (Chrompack, 5 N) through the solution prior to use and an argon atmosphere maintained above the solutions during experimental recordings. In Situ STM. A PicoSPM instrument (Molecular Imaging Co., U.S.A.) with a bipotentiostat for independent control of substrate and tip potential and in-house-built three-electrode KEL-F cells were used in the constant current mode. The substrate was a Au(111) disk, comparable with that previously described. The (26) Haiss, W.; Roelfs, B.; Port, S. N.; Bunge, E.; Baumga¨rtel, H.; Nichols, R. J. J. Electroanal. Chem. 1998, 454, 107-113. (27) Roelfs, B.; Port, S. N.; Bunge, E.; Schro¨ter, C.; Solomun, T.; Meyer, H.; Nichols, R. J.; Baumga¨rtel, H. J. Phy.s Chem. B 1997, 101, 754-765. (28) Tao, N. J.; DeRose, J. A.; Lindsay, S. M. J. Phys Chem. 1993, 97, 910-919. (29) Clavilier, J.; Faure, R.; Guinet, G.; Durand, R. J. Electroanal. Chem. 1980, 107, 205-209. (30) Hamelin, A. J. Electroanal. Chem. 1996, 411, 1-11. (31) Chi, Q.; Zhang, J.; Friis, E. P.; Andersen, J. E. T.; Ulstrup, J. Electrochem. Commun. 1999, 1, 91-96. (32) Chi, Q.; Zhang, J.; Nielsen, J. U.; Friis, E. P.; Chorkendorff, I.; Canters, G. W.; Andersen, J. E. T.; Ulstrup, J. J. Am. Chem. Soc. 2000, 122, 4047-4055.
Organized Oligonucleotide Monolayers
Figure 1. CV of SS oligonucleotides adsorbed on Au(111). 100 mM phosphate, pH 6.9. Scan rate 10 mV s-1. (A) Cyclic voltammogram of pure phoshate buffer (a) and 10A (b). (B) Two successive cyclic voltammograms of 10T: (a) first scan and (b) second scan. (C) DPV of 10T. Three consecutive scans: (a) first scan, (b) second scan, and (c) third scan. reference and counter electrodes were platinum wires. The supporting electrolyte was 0.01 M phosphate (pH ca. 7). Tungsten tips were prepared and coated as previously described.33
Results Voltammetry. Oligonucleotides were adsorbed at open circuit potential by immersing the electrodes into the appropriate solution. Figure 1A shows cyclic voltammograms of 10A and pure phosphate buffer on a Au(111) electrode. The anodic peaks in both cyclic voltammograms at ≈+0.16 V are caused by the lift of the Au(111) reconstruction. The CV of 10A also shows two cathodic peaks at +0.15 and -0.5 V. A cathodic peak around -0.53 V is also observed on glassy carbon34 and mercury electrodes.35 The adenine and cytosine bases are the only reduceable nucleic acid parts on a mercury electrode, (33) Hansen, A. G. Ph.D. Thesis, Technical University of Denmark, Lyngby, Denmark, 2002. (34) Zhu, Y.; Cheng, G.; Dong, S. Biophys. Chem. 2000, 87, 103-110. (35) Palecek, E. Bioelectrochem. Bioenerg. 1986, 15, 275-294.
Langmuir, Vol. 20, No. 5, 2004 1649
where reduction occurs below -1.1 V. Only adenine and guanine can be oxidized on a graphite electrode, at +1.2 and +0.9 V, respectively.35 These two cathodic peaks are, thus, likely to reflect non-Faradaic processes, related to the reorientation of the oligonucleotide on the surface. Figure 1B,C shows cyclic and differential pulse voltammograms of 10T adsorbed on Au(111). The broad cathodic peak at -0.01 V (Figure 1B) reflects a transition from a chemisorbed to a condensed physisorbed layer such as that found for a thymine film on Au(111) by Roelfs et al.27 In the physisorbed layer, thymine molecules lie flat on the surface in protonated and, hence, uncharged form. Formation of the chemisorbed layer is related to deprotonation and perpendicular orientation of thymine on the surface.27 A sharp cathodic peak appears at -0.38 V at the rising of a much broader cathodic peak with a maximum at -0.43 V. These peaks vanish in subsequent scans, suggesting that 10T is desorbed. The peak charges are in the range of several hundred nC cm-2, demonstrating the sensitivity of these non-Faradaic processes to the surface state of adsorbed 10T. These observations can be compared with the formation of multiple N(1s) peaks in X-ray photoelectron spectroscopy (XPS) shifts of 4-5 eV from the bulk thymine value,16 indicative of several adsorption modes. DPV provides enhanced sensitivity. Figure 1C shows two differential pulse voltammograms. Two strong peaks again appear around -0.4 V, that is, a sharp peak at -0.38 V in the first scan followed by a broader peak at -0.47 V. The peaks are completely separated but follow the CV pattern (Figure 1B). In the second scan, all the peaks have almost disappeared, and in the third scan, they have completely disappeared, again suggesting that the adsorbed molecules are liberated from the surface. No chargetransfer process is known for thymine at this potential.36 The 10T DPV and CV peaks must, therefore, be associated with non-Faradaic processes. The sharpness of the -0.38 V peaks is indicative of a phase transition of the surface monolayer, which could be associated with a disorderorder phase transition at -0.33 V such as that observed for a thymine film on Au(111).26 The broader CV and DPV peaks around -0.4 V can be interpreted as desorption, where the width of the peaks reflects the disorder of the 10T adlayer. Figure 2 shows cyclic voltammograms of MCH and the thiolated oligonucleotides. The sharp cathodic peak at -0.75 V in Figure 2A is caused by reductive desorption of MCH, that is, one-electron reduction of the Au-S bond and release of the adsorbed molecules.37-39 The origin of the anodic peak at -0.6 V is partial MCH readsorption. The dominating desorption process is reflected by the decrease of the cathodic peak in the second scan. The coverage of the organic thiols can be determined from the charge of the reductive desorption peak. Peak integration gives 66 ( 9 µC cm-2. Usually, reductive desorption is carried out in alkaline electrolyte to avoid masking the thiolate reduction peak by dihydrogen evolution. We have also recorded CV of MCH in alkaline media. In Table 1, we have summarized the key observables for the reductive desorption. The results show that the charge of the reductive desorption for MCH is independent of pH, which (36) Smyth, M. R.; Vos, J. G. Comprehensive Analytical Chemistry; Elsevier: Amsterdam, The Netherlands, 1992; Vol. 27. (37) Yang, D. F.; Wilde, C. P.; Morin, M. Langmuir 1996, 12, 65706577. (38) Yang, D. F.; Wilde, C. P.; Morin, M. Langmuir 1997, 13, 243249. (39) Esplandiu, M. J.; Hagenstro¨m, H.; Kolb, D. M. Langmuir 2001, 17, 828-838.
1650
Langmuir, Vol. 20, No. 5, 2004
Wackerbarth et al.
Figure 2. CV of MCH and thiolated oligonucleotides on Au(111). 100 mM phosphate buffer, pH 6.9. (A) Two cyclic voltammograms of MCH: (a) first scan and (b) second scan. Scan rate 5 mV s-1. (B) Voltammograms of (a) HS-10A and (b) HS-10AT. Scan rate 10 mV s-1. Table 1. Reductive Desorption Peak Measured in Aquous 0.1 M Phoshpate Buffer, pH 7, with Peak Potential versus SCE molecule
peak position [mV]
MCHa MCH HS-10A HS-10AT
-955 ( 14 -746 ( 10 -680 ( 26 -684 ( 14
a
charge [µC
cm-2]
72 ( 13 66 ( 9 27 ( 5 21 ( 6
Measured in 0.5 M NaOH.
is in keeping with observations by Morin et al.,38 demonstrating the reliability of reductive desorption also under neutral conditions. A charge around 70 µC cm-2 is typical for (functionalized) self-assembled monolayers with chain lengths between two and six carbon atoms on Au(111).39 Figure 2B displays the CV of HS-10A and HS-10AT. HS-10A shows a dominant peak at -0.671 V, which is absent in the 10A voltammogram (Figure 1A). A similar but smaller peak at -0.685 V is seen for HS-10AT. This peak has a shoulder on the negative side of the potential, which we have also observed several times for HS-10A. The shoulder reflects presumably the coexistence of different adsorption modes. These could be ordered adsorbate domains separated by regions of disordered adsorption, such as strongly suggested by in situ STM, compare with the following. We have observed a similar pattern for N-phenyl-mercaptoacetamide disulfide.25 The peaks close to -0.7 V in Figure 2B and the decrease of these peaks by consecutive scanning (not shown) point rather unambiguously to the reductive desorption of HS10A and HS-10AT as the origin of the peaks. The following observation illuminates further the CV patterns of HS10A and HS-10AT. By exposing the oligonucleotidecovered electrode to alkaline solution for several minutes prior to use, nonspecific adsorption via the backbone and
Figure 3. Interfacial capacitances of oligonucleotides adsorbed on Au(111)-electrode surfaces from 100 mM phosphate buffer, pH 6.9. (A) 10T (squares), HS-10A (circles), HS-10AT (diamonds), and 10A (triangles). (B) Desorption of HS-10AT in 100 mM phosphate solution, pH 6.9. Three successive scans: first scan (squares), second scan (circles), and third scan (triangles).
the nucleobases is avoided,40 presumably as a result of increased electrostatic repulsion from the negatively charged surface by deprotonation of the nucleobases. The CV of HS-10A was not affected by this procedure. Nonspecific interactions of HS-10A and HS-10AT can, therefore, be disregarded, substantiating that the cathodic peaks primarily represent reductive desorption. This is in keeping with the observation that nonchemisorbed bases dominate both the XPS and Fourier transform infrared (FTIR) patterns at high coverage.16 The reductive current peak of organic thiol layers has, however, two components. The dominant component stems from interfacial charge transfer from the electrode to chemisorbed thiolate, but there is also a capacitive component caused by the formation of the aqueous double layer on the uncoated gold surface after thiol desorption. This non-Faradaic contribution can be estimated to be 1.9 µC cm-2 for HS-10A and 2.3 µC cm-2 for HS-10AT by interfacial capacitance data, see Figure 3A. The resulting coverages are 260 ( 50 pmol cm-2 and 200 ( 60 pmol cm-2 for HS-10A and HS-10AT, respectively. In situ STM suggests that this high coverage is caused by the formation of a densely packed monolayer with an upright or a tilted molecular orientation. Figure 3 shows the interfacial capacitance of 10A, 10T, HS-10A, and HS-10AT. The capacitances of HS-10A and HS-10AT are almost constant, but strong peaks appear at -0.7 V. The apparent capacitive contributions to the reductive desorption charge are 1.9 µC cm-2,and 2.3 µC cm-2 for HS-10A and HS-10AT, respectively, compare with previous text. These values represent upper limits, (40) Yau, H. C. M.; Chan, H. L.; Sui, S.; Yang, M. Thin Solid Films 2002, 413, 218-223.
Organized Oligonucleotide Monolayers
inasmuch as the measured interfacial capacitance includes an ohmic resistance, which cannot be separated by capacitance data alone. However, the capacitive charges constitute about 10% of the total reductive desorption charge. In general, the capacitive component of voltammetric reductive desorption peaks of alkanethiolates is about 25% of the total charge,37,38 in keeping with our results. The interfacial capacitance of 10A displays no features in the potential window recorded, and the interfacial capacitance is significantly lower than that for HS-10A. This could be explained by the formation of a dense layer, where the adenines of 10A are closer to the surface than for the expectedly more permeable layer of upright-oriented HS-10A. The interfacial capacitance of HS-10A and HS-10AT shows no significant difference, indicating similar adsorption modes. Both display a peak around -0.7 V, representing the reductive desorption. The interfacial capacitance of 10T shows a strong and broad peak at +0.30 V and a second, smaller peak at -0.37 V. The capacitance on the positive side of the -0.37 V peak is only slightly lower for a clean Au(111) electrode in phosphate buffer, whereas the capacitances on the negative side of the peak are similar. This is again a strong indication that the 10T layer is desorbed. Comparable desorption peaks of organic molecules have frequently been observed on various electrode materials.41 The peak at +0.3 V is probably caused by charge transfer from the deprotonation of adsorbed thymine in the chemisorbed phase.26 The lift of the reconstruction of the Au(111) surface (cf. Figure 3B) underneath the 10T layer almost certainly contributes to this peak, which we have also observed by in situ STM. The thymine base, thus, appears to play a crucial role in the adsorption of 10T. Even subtle surface reorientation of the oligonucleotides, as well as rearrangements of thymine on the surface, was thus distinguishable by single-crystal electrochemistry as collective changes in the adsorption of the oligonucleotide. Figure 3B shows three consecutive interfacial capacitance scans of HS-10AT on a Au(111) surface. The interfacial capacitance of the first scan is constant with a peak at -0.7 V. The second scan shows a strong and broad peak around +0.3 V and a decrease of the -0.7 V peak. This trend is continued in the third scan. The height of the interfacial capacitance peak in the third scan approaches clearly that of Au(111) in pure phosphate buffer. The interfacial capacitance of the bare gold surface shows the hump at +0.3 V, with a tip around +0.16 V. 10A also shows a peak at +0.3 V in the second scan, but compared to the thiol oligonucleotide the height is significantly lower and has not reached the height of the peak on a clean Au(111) surface even after five or six excursions to negative potentials. The hump at +0.3 V can best be explained by changes in buffer anion adsorption patterns, as the potential crosses the potential of zero charge (pzc). The tip on the hump at +0.16 V reflects the lift of reconstruction of the Au(111) surface. The surface thus becomes accessible for anions because of HS-10AT desorption, and the hump appears. In contrast, 10A remains or is readsorbed on the surface even after several excursions to negative potentials, indicating that electrostatic interactions are not solely responsible for adsorption. In Situ STM of Adsorbed Oligonucleotides. The voltammetric and interfacial capacitance data show clearly that dense adlayers of the three SS oligonucleotides HS10A, 10A, and 10T are formed on the Au(111) surface. (41) Damaskin, B. B.; Petrii, O. A.; Batrakov, V. V. Adsorption of Organic Compounds on Electrodes; Plenum: New York, 1971.
Langmuir, Vol. 20, No. 5, 2004 1651
The in situ STM images directly in aqueous buffer solution overall support the voltammetric data, compare, however, with the discussion of 10T to follow. The in situ STM images add novel high resolution to the surface patterns of the thiolated compounds, and the image resolution of HS-10A approaches the single-molecule level. In situ STM images the electronic structure and electronic conductivity of the adsorbates. The electronic properties are determined by the intrinsic energetics and other properties of the highest occupied adsorbate molecular orbital (HOMO) and lowest unoccupied adsorbate molecular orbital (LUMO) relative to the substrate and tip Fermi levels.42,43 The HOMO and LUMO properties are also controlled by the molecular adsorbate conformation. Packing constraints in the adlayer can, for example, impose molecular conformations with poor electronic conductivity unfavorable for in situ STM imaging.42 Multilayer formation can also be a concern in the adsorption of molecules as large and conformationally mobile as the oligonucleotides. The strongly negatively charged oligonucleotide backbone and surface rinsing after oligonucleotide surface immobilization would, however, disfavor adsorption of more than a single monolayer. This is substantiated by both the voltammetric data and in situ STM. As still another general observation, reliable imaging approaching the single-molecule level requires that environmental conditions for the formation of monolayers of some lateral order are defined. Spatial constraint of the conformationally labile adsorbate molecules by collective lateral interactions is, thus, essential for robust highresolution imaging. Adsorbates brought to pack into dense monolayers have provided successful single-molecule in situ STM imaging of several redox metalloproteins.23-25 Formation of domains with long-range order has applied to a number of intermediate-size biomolecules such as single nucleobases,44,45 porphyrins,46 amino acids,47,48 and amino acid-related molecules.49,50 Two-dimensional domain formation also constitutes a basis for in situ STM imaging of HS-10A in which control of the electrochemical potential is a crucial factor. In situ STM of 10A, Figure 4A, shows that a dense monolayer on the Au(111) surface has formed but no longrange order was observed. This is in line with previous results.51 Au(111)-reconstruction lines could be detected underneath the oligonucleotide monolayer in the in situ STM images of both 10A (Figure 4A) and 10T (not shown). Single gold atoms on the top of terraces and sharp edges were, moreover, observed on the surface around the reconstruction potential in the presence of 10T. These are not visible on bare gold surfaces, presumably as a result of high surface mobility of the gold surface atoms. This could suggest that the 10T monolayer film attenuates the (42) Claypool, C. L.; Faglioni, F.; Goddard, W. A.; Gray, H. B.; Lewis, N. S.; Marcus, R. A. J. Phys. Chem. B 1997, 101, 5978-5995. (43) Sautet, P. Chem. Rev. 1997, 97, 1097-1116. (44) Dretschkow, T.; Dakkouri, A. S.; Wandlowski, T. Langmuir 1997, 13, 2843-2856. (45) Wandlowski, T.; Lampner, D.; Lindsay, S. M. J. Electroanal. Chem. 1996, 404, 215-226. (46) Tao, N. J. Phys. Rev. Lett. 1996, 76, 4066-4069. (47) Zhang, J.; Chi, Q.; Nielsen, J. U.; Friis, E. P.; Andersen, J. E. T.; Ulstrup, J. Langmuir 2000, 16, 7229-7237. (48) Dakkouri, A. S.; Kolb, D. M.; Edelstein-Shima, R.; Mandler, D.; Langmuir 1996, 12, 2849-2852. (49) Giz, M. J.; Duong, B.; Tao, N. J. J. Electroanal. Chem. 1999, 465, 72-79. (50) Sawaguchi, T.; Sato, Y.; Mizutani, F. J. Electroanal. Chem. 2001, 507, 256-262. (51) Tano, T.; Tomyo, M.; Tabata, H.; Kawai, T. Jpn. J. Appl. Phys. 1998, 37, 3838-3843.
1652
Langmuir, Vol. 20, No. 5, 2004
Wackerbarth et al.
Figure 4. In situ STM image of 10A and HS-10A. 10 mM phosphate buffer, pH 7.1. Scan rate 13 lines/s. Sample potential -0.21 V versus SCE. Constant current mode. (A) Tunneling current 0.3 nA. Bias voltage -0.10 V. (B) Tunneling current 0.5 nA. Bias voltage -0.15 V. The height profiles along a molecular row and across a set of rows following the lines a and b in Figure 4B, respectively, are shown in the figures that follow.
mobility of the gold surface atoms. Under the chosen conditions, direct mapping of adsorbed 10A and 10T could, however, not be achieved. This was unexpected because the electrochemical data suggest that a certain structural order on the surface prevails, particularly for 10T. It could also suggest that surface monolayers of large biomolecules are less robust in the confined in situ STM configuration than at the semi-infinite electrochemical surface, such as that noted for voltammetry and in situ STM of redox metalloproteins adsorbed on Au(111) surfaces.23-25 The adsorption pattern of thiolated HS-10A is quite different from those of thiol-free 10A and 10T. Figure 4B shows an in situ STM image of HS-10A adsorbed at an open circuit potential on a Au(111)-electrode surface in contact with aqueous buffer. The open circuit potential was measured to be +0.19 V. Figure 4B was recorded at a sample potential of -0.21 V after the potential had been set for 40 min to -0.61 V. Large areas of domains with long-range adsorbate order are visible. Pits typical for adsorbed organic thiolate layers on Au(111) surfaces were also present in the domain boundary regions. The domains are oriented with an angle of about 60° to each other. The sample potential is, thus, crucial in the adsorption
dynamics and the resulting supramolecular HS-10A adsorbate organization on the Au(111) surface. Figure 5A shows again an in situ STM image of HS-10A at a -0.21 V sample potential, after the potential had been set for 40 min at -0.61 V. On subsequent stepping to more positive potentials, that is, +0.09 V, the domains disappear slowly leaving the adlayer in an entirely disordered state (Figure 5B). Setting the potential back to -0.61 V for about 40 min and recording at -0.21 V, domains clearly appear again (Figure 5C). This shows that the domain formation process is reversible upon excursion to a more negative sample potential. However, there are some differences in the images compared to the first recording. The domains in Figure 5C are larger than those in Figure 5A but enclosed by areas with no ordered structure. This is not seen in Figure 5A. This could be because domain formation at -0.61 V is close to the reductive desorption potential, and some adsorbate molecules may leave the Au(111) surface upon successive potential scanning in this strongly negative range. The domains display rows of bright contrast with weaker contrasts in between. The distance between contrasts in a row is 5 Å, and the spacing between the spots in parallel rows 11 Å. The rows form a (x3 × 4)R30° surface lattice.
Organized Oligonucleotide Monolayers
Langmuir, Vol. 20, No. 5, 2004 1653
Figure 5. Successive in situ STM images of HS-10A. 10 mM phosphate buffer, pH 7.1. Constant current mode, tunneling current 0.5 nA. Scan rate 13 lines/s. Adsorption under open circuit potential. Prior to scanning, the potential was set for 40 min to -0.61 V to create an ordered adlayer. (A) Sample potential -0.21 V versus SCE; bias voltage -0.15 V. (B) Sample potential stepped to +0.19 V versus SCE; bias voltage -0.55 V. (C) Sample potential reversed to -0.61 V versus SCE for 45 min; then imaged at -0.21 V, bias voltage -0.15 V.
This is further illustrated by the height profiles in Figure 4. Each bright spot is, moreover, tentatively assigned as a bound thiol linker. This would give a coverage of 288 pmol cm-2, which is close to the value of 260 pmol cm-2 determined by CV. The difference is small and suggests that domain boundaries, pits, and disordered areas constitute a relatively small part of the total area. The in situ STM images and the coverage determined both from CV and in situ STM offer a view of HS-10A adsorption in which the adsorbate molecules are bound solely via the thiol linker and strong lateral interactions between adjacent oligonucleotides are present. It can be envisaged both that the lateral interaction must involve counterion screening and that this screening is strong enough to induce Au-S bond dissociation and reformation during the adsorption process. Discussion We have studied the adsorption of the SS oligonucleotides, 10A, 10T, and HS-10A, on single-crystal, atomically
planar Au(111) electrodes. A combination of cyclic and DPV, interfacial capacitance, and in situ STM was used. In addition, the hybridized HS-10AT was studied by CV and DPV. The oligonucleotides offer insight into collective adsorption properties, but their short length and uniform base sequences disclose details about interfacial structure and dynamics compared with the complex macromolecular structure of longer sequences. The capacitance data suggest that the pzc of the Au(111) electrode in 0.1 M phosphate buffer is ≈+0.3 V. Oligonucleotide adsorption, either at the open circuit potential (+0.19 V) or at controlled more negative potential, is, therefore, at a negatively charged surface. Three capacitance patterns emerge. 10A gives a small featureless capacitance in the whole potential range. 10T displays a more varied surface dynamics with peaks representing adsorption/desorption, proton transfer, and configurational surface reorganization. The HS-10A and the HS-10AT capacitance data are completely dominated by Au-S reductive desorption. The capacitance data
1654
Langmuir, Vol. 20, No. 5, 2004
overall suggest that thiol-free 10A and 10T are nonspecifically adsorbed by electrostatic image forces or in other ways, with large parts of the oligonucleotide in contact with the electrode. This mode is unimportant for HS-10A and HS-10AT as the strong Au-S bond links the adsorbate molecule to the Au(111) surface solely via this bond and the backbone part out of direct contact with the surface, giving a higher monolayer density than for the thiol-free oligonucleotides. CV and DPV support this view. In addition to the anodic Au(111)-reconstruction peak, 10A shows two weak nonFaradaic cathodic peaks, presently of elusive nature. 10T shows a strong reductive peak at -0.43 V, which matches the capacitive desorption peak. A satellite peak at -0.38 V is assigned to structural transition in the surface layer at the foot of the desorption peak. A broad peak at -0.01 V matches the capacitance peak but is shifted negatively by almost +0.3 V. Overall, the voltammetry and capacitance data are indicative of nonspecific adsorption of 10A and 10T in a flat-lying orientation. Adenine on Au(111) and mercury and adenine oligonucleotides of up to three adenine bases on mercury have been reported to adsorb in a flat-lying mode with the heterocyclic rings parallel to the surface.28,52 As for the capacitances, the voltammetry of HS-10A and HS-10AT is entirely dominated by Au-S reductive desorption. Clear stoichiometry is associated with this process and a precise value of 260 ( 50 pmol cm-2 of the SS coverage can be obtained. This value is high and suggests strongly that the adsorbate molecules are densely packed in an upright or tilted orientation. A flat-lying orientation would take up much more surface space and give lower monolayer coverage. An upright, or free-standing, configuration with bases out of direct contact with the surface was also concluded for HS-25T, at high coverage on the basis of XPS and FTIR spectroscopy.16 Voltammetry of the hybridized HS-10AT form is also dominated by Au-S reductive desorption in a similar pattern. The coverage is slightly lower, that is, 200 ( 60 pmol cm-2. A DS oligonucleotide should have roughly a four times lower coverage according to our model neglecting intermolecular interactions, where the oligonucleotides are adsorbed in an upright or tilted orientation and, hence, the diameter of the oligonucleotide primarily determines the coverage. In this respect, our SS coverage accords quite well with the DS coverage of 60-75 pmol cm-2 reported by Kelley et al.7 A decrease by only 25% in the coverage suggests a low hybridization level, which is understandable, inasmuch as the melting temperature of 10AT (30.4 °C) is only slightly above room temperature. The hybridization level can be increased by adding di- or highervalent cations. This may also affect aggregation of the monolayer of the SS oligonucleotide. However, this needs systematic studies, which are presently in progress. High-resolution in situ STM offers a molecular view of the different adsorption modes. In situ STM of 10A and 10T shows dense monolayers, but single-molecule mapping could not be achieved and the images are inconclusive as to the coverage and adsorbate orientation. In situ STM of thiol-modified HS-10A gives a quite different pattern. Imaging of HS-10A adsorbed at open circuit potential (+0.19 V) gives a dense, largely disordered layer. Stepping the potential to significantly negative values (-0.61 V), however, induces long-range ordered domain formation. Domain formation is reversible, and order/disorder can (52) Webb, J. W.; Janik, B.; Elving, P. J. J. Am. Chem. Soc. 1973, 95, 8495-8585.
Wackerbarth et al.
be induced sequentially by suitable potential steps. This supports the view that electrostatic repulsion of the phosphate backbone from the negatively charged electrode surface unfolds and loosens initial structurally disordered backbone parts from the surface. On the basis of XPS and FTIR spectroscopy, they may still be largely free-standing at high coverage and also in the disordered state.16 Stimulated by counterion screening, hydrogen bonding, phosphate solvation, and Au-S bond reshuffling, adsorbed HS-10A is rearranged into long-range ordered domains. The lines in the domains have different lengths. Together with the high coverage determined by CV (260 pmol cm-2), this is only compatible with HS-10A adsorption solely via the Au-S bond with the backbone oriented toward the solution. The coverages obtained both by CV and in situ STM are high compared to some reported values for 1215 base oligonucleotides,15,16 which range from 10 to 90 pmol cm-2. The coverage, however, depends on several factors such as adsorption time, ionic strength, and, not least, the electrode surface morphology. Amorphous Au films pretreated with “piranha” solution, used elsewhere, have, for example, surface morphologies quite different form the atomically planar Au(111) surfaces (see Supporting Information) used presently.15,16 Each bright spot in the in situ STM images can then be assigned to a single HS-10A molecule. This gives a domain-based coverage of 288 pmol cm-2, which is close to the electrochemically determined value considering that pits and disordered areas are also present. The image resolution is, thus, clearly at the single-molecule level. This level can only be achieved meaningfully for conformationally flexible molecules such as the SS oligonucleotides by prior immobilization into ordered domain structures, or possibly into ordered structures of other twodimensional matrices. In conclusion, this study has provided data for the mapping and control of the adsorption of short thiolmodified and thiol-free oligonucleotides of uniform composition on single-crystal Au(111) surfaces, based on CV and DPV, interfacial capacitance, and in situ STM. High sensitivity compared to polycrystalline electrodes has been achieved. Oligonucleotide behavior has, further, been directed toward reversible domain formation with high lateral order characterized to single-molecule resolution by in situ STM. The disorder-order interconversion can, tentatively, be framed by a qualitative mechanistic view. Thiol-modified oligonucleotide is first adsorbed via Au-S bond formation in a disordered conformation. Conversion to the ordered state requires Au-S bond reorganization by surface diffusional motion via the adsorbate linker part. The surface diffusion activation energy is compensated by attractive lateral forces. These are base stacking, adsorbate intermolecular hydrogen bonding, and cationinduced aggregation, of which the former two by far dominate and compare energetically with Au-S bond diffusion. The latter in principle also depends on the potential and is more labile at negative potentials. It can, moreover, be envisaged that the attractive lateral interactions are kinetically hindered in the disordered state. A major physical effect of the negative electrochemical potential excursion is, then, to drive the oligonucleotide backbone into an extended conformation in which these attractions come into full action and compensate the Au-S reorganization energy expenditure. Other areas of evolving high-resolution DNA science in which the reported approach would hold promise include, for example, SS and DS electronic conductivity mechanisms at the single-molecule level. Metal ion-induced two-
Organized Oligonucleotide Monolayers
Langmuir, Vol. 20, No. 5, 2004 1655
dimensional or columnar DNA-based aggregation is another area suggested by the single-molecule HS-10A arrays disclosed by in situ STM.53-55 Systematic extension to longer sequences of variable base composition, finally, offers a biotechnological nanoscale dimension in single-
molecule hybridization and other chemical processes of immobilized DNA-based molecules.
(53) Kornyshev, A. A.; Leikin, S. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 13579-13584. (54) Kornyshev, A. A.; Leikin, S. Phys. Rev. Lett. 1999, 82, 41384141. (55) Bloomfield, V. A. Curr. Opin. Struct. Biol. 1996, 6, 334-341.
Supporting Information Available: In situ STM images of Au(111) in 0.1 M phosphate (pH 7.0). This material is available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We acknowledge financial support from the Danish Technical Science Research Council (Contract No. 26-00-0034).
LA035547G