Nanoparticle Layers Assembled through DNA Hybridization

---

Langmuir 2002, 18, 1825-1830

1825

Nanoparticle Layers Assembled through DNA Hybridization: Characterization and Optimization Marc L. Sauthier, R. Lloyd Carroll, Christopher B. Gorman, and Stefan Franzen* Department of Chemistry, North Carolina State University, Box 8204, Raleigh, North Carolina 27695 Received August 7, 2001. In Final Form: November 5, 2001 The hybridization of nanoparticle-labeled DNA targets to surface-attached DNA probes has been investigated. Scanning tunneling microscopy (STM) and Raman and Fourier transform infrared (FTIR) spectroscopy were used to elucidate surface morphology, coverage, and the presence of aggregates. The factors that affect surface coverage, such as probe density, labeled target concentration, and particle size, were systematically investigated by STM in order to determine the best set of experimental conditions allowing the formation of dense monolayers with a minimal number of surface defects for both 5((1) nm and 10((2) nm gold nanoparticle labels on the target strand. Grazing-angle FTIR spectroscopy demonstrates that DNA is largely oriented once the labeled targets hybridized to the probes. Raman microscopy was used to probe the surface for the presence of large aggregates that would give rise to large scattering signals. Both STM and optical experiments provide evidence that dense surface layers can be formed without extensive aggregation. Nonselective binding was shown to be a function of the target concentration and nanoparticle size. Propensity for both aggregation and nonspecific binding is greater for 10((2) nm than for 5((1) nm gold nanoparticles.

Introduction The use of nanoparticles as a means to detect DNA hybridization has recently shown great promise. The optical properties of three-dimensional aggregates of gold nanoparticles have been used to detect hybridization of specific DNA sequences in solution1,2 and on surfaces as an alternative to fluorescent labeling of DNA.3-5 The efficient utilization of a detection strategy based on nanoparticles relies on the self-assembly properties of double-stranded DNA (dsDNA) labeled with nanoparticles. Although it has been shown that DNA hybridization can overcome electrostatic repulsion between gold nanoparticles in solution, the surface chemistry of dense dsDNAnanoparticle layers is an active area for research.5 These efforts will benefit from physical characterization of dsDNA-nanoparticle layers on surfaces. The systematic study of specific factors that affect surface coverage and surface defects as well as the determination of experimental conditions leading to the formation of dense nanoparticle layers are important for applications as well as a fundamental understanding of the surface chemistry of DNA-hybridized structures. One of the disadvantages of using nanoparticles as labels is the formation of aggregates. Nanoparticles are known for their tendency to aggregate at moderate or even low ionic strength. Previous studies using gold nanoparticles have shown the benefits of a relatively high sodium chloride concentration in the hybridization solution when (1) Storhoff, J. J.; Elghanian, R.; Mucic, R. C.; Mirkin, C. A.; Letsinger, R. L. J. Am. Chem. Soc. 1998, 120, 1959-1964. (2) Elghanian, R.; Storhoff, J. J.; Mucic, R. C.; Letsinger, R. L.; Mirkin, C. A. Science 1997, 277, 1078-1081. (3) He, L.; Musick, M. D.; Nicewarner, S. R.; Salinas, F. G.; Benkovic, S. J.; Natan, M. J.; Keating, C. D. J. Am. Chem. Soc. 2000, 122, 90719077. (4) Taton, T. A.; Mirkin, C. A.; Letsinger, R. L. Science 2000, 289, 1757-1760. (5) Taton, T. A.; Mucic, R. C.; Mirkin, C. A.; Letsinger, R. L. J. Am. Chem. Soc. 2000, 122, 6305-6306.

nanoparticles were used as DNA labels.1,2,6 Such a concentration leads inevitably to aggregation over time. One of the goals of the present work is to determine a set of conditions that will minimize aggregation. Reducing the number of aggregates is of the utmost importance as they tend to deposit nonselectively on the surface and are difficult to remove by simple washings or sonication. Aggregates may also be responsible for artifacts when optical techniques are used to characterize dsDNAnanoparticle layers. Experimental Section Materials. Gold nanoparticles (5 and 10 nm) were purchased from Ted Pella Inc., Redding, CA. Bis(p-sulfonatophenyl)phenylphosphine dihydrate (BSPP) was used as received from Strem Chemicals to stabilize gold nanoparticles.6 The hybridization buffer, saline sodium citrate (1xSSC), was prepared by diluting 20 times a 20X saline sodium citrate solution purchased from FisherBiotech. The 5′-thiol modified, single-stranded DNA (ssDNA) molecules used in the following studies were synthesized by the Nucleic Acid Facility at North Carolina State University. They each contained 25 bases, and their sequences were as follows: 5′-AACCAGGATATCCGCTCACAATTCC for the probe and 5′-GGAATTGTGAGCGGATATCCTGGTT for the target. The 5′-thiol modifier was purchased from Glen Research, Sterling, VA. ssDNA and nanoparticle concentrations were determined by measuring absorption at 260 and 520 nm, respectively. Goldcoated microscope slides were used as received. Instrumentation. A Hewlett-Packard HP8453 diode array absorption spectrometer was used to measure ssDNA and nanoparticle concentrations. The surface morphology and electrochemical properties of our samples were investigated with a Digital Imaging Nanoscope E. Infrared spectra of dsDNAnanoparticle monolayers were acquired on a Biorad FTS-6000 Fourier transform infrared (FTIR) spectrometer connected to a microscope fitted with a grazing-angle objective. The Raman microscope used to study optical surface properties was a SPEX Infinity with an excitation wavelength at 632.5 nm. (6) Loweth, C. J.; Caldwell, W. B.; Peng, X.; Alivisatos, A. P.; Schultz, P. G. Angew. Chem., Int. Ed. Engl. 1999, 38, 1808-1812.

10.1021/la0112763 CCC: $22.00 © 2002 American Chemical Society Published on Web 02/08/2002

1826

Langmuir, Vol. 18, No. 5, 2002

Sauthier et al.

Figure 1. Strategies for the synthesis of dsDNA-nanoparticle layers. (A) Sequential synthesis; first the probes are deposited on the surface, followed by MH deposition, target hybridization, and target labeling with gold nanoparticles. (B) Codeposition synthesis; probes and MH are deposited followed by hybridization of nanoparticle-labeled targets. Thiol-Modified ssDNA: Postsynthesis Workup. After synthesis, protected thiol-modified DNA strands were purified either by anion exchange HPLC or on poly-Pak cartridges to remove failure sequences that could interfere with subsequent experiments. Purified oligonucleotide solutions were then aliquoted, dried to a solid, and frozen for long-term storage. Target Labeling. The frozen aliquot of thiol-modifed target was diluted in 1 mL of a 100 mM solution of dithiothreitol (DTT) in water, pH 8.3-8.5. The mixture was then thoroughly mixed and allowed to react for 30 min at room temperature in order to deprotect the thiol group. The oligonucleotide solution was then syringed onto a Poly-Pak cartridge to remove the excess DTT and salt byproducts. Next, the oligo was eluted from the cartridge with 20% acetonitrile in water. The eluant was subsequently dried to a solid and resuspended in deionized water. This solution was finally added to a solution of BSPP-stabilized gold nanoparticles with a nanoparticle/ssDNA molar ratio of 10/9, that is, with a slight excess in gold particles. This specific stoichiometry produces nanoparticles substituted with only one strand of DNA.6 The resulting mixture was stirred vigorously over 36 h as the NaCl concentration in the mixture was gradually increased to 0.15 M using a 0.3 M solution of NaCl in water, pH 7.0. The conjugate solution not used immediately was stored at 4 °C. dsDNA-Nanoparticle Monolayers: Synthesis. Two different methods were used to create monolayers of dsDNAnanoparticle assemblies on gold surfaces (Figure 1). These methods showed similar results based on characterization by STM (vide infra) and grazing-angle FTIR microscopy. The first procedure consists of the sequential deposition of (1) a ssDNA oligonucleotide with a 5′-thiol (probe), (2) a 6-mercapto-1-hexanol (MH) passivating layer, and (3) the complementary ssDNA oligomer (target) with the 5′-end still trityl-protected from DNA synthesis.7 Following deprotection using silver nitrate to expose the free thiol group, the dsDNA-derivatized surface was exposed to a solution of BSPP-capped gold nanoparticles for 2-3 days. Alternatively, codeposition of (1) and (2) was followed by hybridization of nanoparticles, each derivatized with a single ssDNA target. Both methods are described more thoroughly below. Sequential Deposition Synthesis. As described above, this sequential synthetic strategy calls for initial deposition of probes onto gold substrates followed by MH deposition, target hybridization, and finally target labeling with gold nanoparticles. The thiol groups at the 5′-end of the ssDNA probes were deprotected by reaction with an excess silver nitrate immediately prior to deposition to prevent thiol oxidation to disulfide. The excess AgNO3 was removed by precipitation following the addition of DTT. The excess DTT was eliminated by passing the sample on a NAP-10 column using a 10 mM phosphate buffer, pH 7.0, as the mobile phase. Gold substrates (polycrystalline or [111]) were cleaned with warm piranha solution for 5 min (caution: piranha solution reacts violently with organic chemicals) and then rinsed

with deionized water before being immediately immersed in a 1.0 µM solution of 5′-thiol modified ssDNA in the presence of 1.0 M potassium phosphate buffer, pH 7.0. The thiol-modified ssDNA was allowed to react with the gold surface for 4 h. The substrate was then thoroughly rinsed with Nanopure water and subsequently dipped in a 1.0 mM solution of MH in water for 1 h. The MH addition reduces the number of noncovalent interactions between the ssDNA backbone and the gold.7 After rinsing with Nanopure water, the gold substrate carrying the mixed monolayer was immersed in a 1.0 µM solution of protected thiol-modified target in 1xSSC. The hybridization solution was brought to 85 °C for 1 h before being allowed to cool to room temperature over the course of several hours. Unlabeled DNA hybridization occurs within a 15 °C range around the theoretical melting point for a given DNA sequence, whereas it occurs only within a 5 °C range when nanoparticles are used as labels.5 The melting point for the oligonucleotides selected for our studies was estimated at around 62 °C when hybridization was carried out in 1xSSC. Once the hybridization process was completed, the sample was rinsed thoroughly with Nanopure water and the thiol group on the target was deprotected with an excess silver nitrate for 30 min. Immediately following deprotection, the sample was rinsed with Nanopure water and incubated for 2-3 days in a 5 nM solution of gold nanoparticles in water as the NaCl concentration was gradually increased to 150 mM. Finally, the adlayer was rinsed and dried with nitrogen gas. The sequential deposition procedure was employed for the Raman and infrared spectroscopic experiments mentioned herein. Codeposition Synthesis. This synthesis, as introduced earlier, consists of the codeposition of both probe and MH, followed by the hybridization of targets already labeled with gold nanoparticles. The thiol group at the 5′-end of the probe was deprotected using the same procedure as that described above. The gold substrate was immersed in warm piranha solution for 5 min and then rinsed with Nanopure water immediately prior to use. The 5′-thiol terminated probe was mixed with MH at a given molar percentage to create a 1.0 µM solution total in 1.0 M potassium phosphate buffer, pH 7.0. The clean gold substrate was immersed in this mixture overnight. The gold surface bearing the mixed monolayer was then rinsed with Nanopure water and immersed in the hybridization solution containing ssDNA target-nanoparticle conjugates in 1xSSC. The hybridization solution was then heated to 85 °C for 1 h and allowed to cool to room temperature over the course of 4 h. Finally, the sample was rinsed thoroughly with Nanopure water and dried under a flow of nitrogen. The advantages of this second synthetic route include faster experimental procedures and a better control on the amount of ssDNA probes on the surface. The codeposition procedure was used for all surface morphology and coverage studies described herein.

(7) Herne, T. M.; Tarlov, M. J. J. Am. Chem. Soc. 1997, 119, 89168920.

The primary goal of this study is to determine the conditions that lead to reproducibly dense, homogeneous

Results and Discussion

Nanoparticle Layers through DNA Hybridization

Figure 2. (A) Topography of a monolayer of dsDNA-nanoparticles on gold as analyzed by STM; 20% probe/80% MH in codeposition solution. (B) Control; 100% MH in codeposition solution. Experimental conditions for both images: 5 ( 1 nm particles; 5 nM labeled target solution; hybridization cooling rate, 5 h; dimensions of the probed area, 500 × 500 nm. Both images have been chosen as representative of a panel of four pictures taken in four distinct areas of the two samples.

layers of nanoparticles that are not bound as aggregates. As stated in the Introduction, aggregation is a common observation in nanoparticle suspensions. Therefore, detection of ordered layers or aggregates on the surface is desirable in order to establish the nature of surface interactions promoted by DNA hybridization. The characterization techniques in this study demonstrate the creation of ordered surface layers. Under appropriate conditions, aggregation can be minimized. Scanning probe (scanning tunneling microscopy, STM) and optical (Raman, FTIR) methods were used to elucidate monolayer morphology, surface coverage, and presence of surface defects such as aggregates and nonselective binding that could lead to the formation of multilayers. Surface coverage was studied by STM, and the factors that affect it, such as probe and labeled target concentrations, particle size, and hybridization conditions, were systematically investigated in order to determine the best set of experimental conditions allowing the formation of dense monolayers with a minimal number of surface defects for both 5((1) nm and 10((2) nm gold nanoparticles on the target strand. Grazing-angle FTIR spectroscopy demonstrates that DNA is largely oriented once nanoparticle hybrids are formed on the surface. Raman microscopy was used to probe the surface for the presence of large aggregates that would give rise to large scattering signals. Both STM and optical experiments provide evidence that dense surface layers can be formed without extensive aggregation. The current study has been limited to 5 nm particles as the smallest particle size and to BSPP capping. Probe codeposition solutions reported here had a molar percentage in ssDNA ranging from 5 to 20%, and labeled target solution concentrations ranged from 1 to 10 nM. The studies discussed below will lead to the conclusion that smaller particles and more robust capping are essential features of a successful DNA-hybridization array. A study of 5 nm nanoparticles demonstrates that dsDNA-nanoparticle layers can be formed on a gold substrate. Figure 2a shows the STM topographical picture of a relatively homogeneous layer of dsDNA-nanoparticles. To confirm that the particles were actually bound selectively to the surface through hybridization of complementary strands of ssDNA, a series of control experiments were conducted. An example of the control experiments is shown in Figure 2b where nonspecific binding of 5 nm particles is observed on a MH monolayer control. While good evidence for the formation of dsDNA will be presented thereafter, this contamination could not be eliminated even at target concentrations down to 500 pM (see

Langmuir, Vol. 18, No. 5, 2002 1827

Figure 3. Average single particle electrical behavior across a typical DNA-immobilized gold nanoparticle surface. The inner curve shows the average electrical behavior of a nanoparticle monolayer sample. The two outer curves represent the 95% confidence interval.

Supporting Information). Despite this contamination, optical and STM measurements show ordered surface structure for the majority of the dsDNA component. Although some of the nanoparticles bound to the gold surface appear to be in physical contact with one another in Figure 2a, they are indeed separated by several nanometers. Due to variations in tip geometry, STM imaging distorts an object’s lateral dimensions. In fact, the feedback loop of the STM is designed to give accurate height information while the dimensions of the tip must be convolved with the lateral dimensions. For this reason, particles shown in Figure 1a have an average apparent diameter of 14 ( 2 nm. Based on transmission electron microscopy analysis, the nanoparticles used in this study have a diameter of 5 ( 1 nm (Supporting Information). If this diameter is assumed, then the actual surface covered by gold nanoparticles is only 13% of the coverage apparent in Figure 2a. Similar conclusions are obtained for 10 ( 2 nm particles (data not shown). The observation of well-separated nanoparticles on a dsDNA hybridized surface is further confirmed by the results of the voltammetric analysis conducted on dsDNAnanoparticle monolayers using STM. Figure 3 shows that the electrical behavior of the tunneling junction composed of Au nanoparticle-dsDNA-Au substrate structures is dominated by a quantized electron-transfer mechanism characteristic of isolated nanoparticles. The I-V curve shown in Figure 3 is an example of a “Coulomb staircase”.8-11 The voltage steps are a result of the charging energy of the capacitor defined as the nanoparticle and surface V ) e2/2C. The capacitance of a spherical capacitor is 4πr, where  ) r0. For a 5 nm diameter nanoparticle, this corresponds to an energy of ≈10-19 J in a vacuum and ≈1.2 × 10-20 J in water where the solvent dielectric constant is r ≈ 78. This is in agreement with experimentally determined capacitance obtained from the I-V curve above C ) e/V ≈ 7 × 10-19 F for a step of V ≈ 0.25 V. Thus, the experimental charging energy is calculated to be ≈2 × 10-20 J in reasonable agreement with the value estimated above. Given that thermal energy is kT ≈ 4 × 10-20 J at 290 K and a partition function of (1 exp{-E/kT})-1 for energy levels evenly spaced with energy V, one would expect a mixture of roughly 2/3 ohmic and 1/3 quantized Coulomb staircase effects as observed in Figure 3. These observations are in agreement with literature precedents that suggest that the Coulomb staircase effect will be observed in nanoparticles of 10 nm diameter and (8) Andres, R. P.; Bein, T.; Dorogi, M.; Feng, S.; Henderson, J. I.; Kubiak, C. P.; Mahoney, W.; Osifchin, R. G.; Reifenberger, R. Science 1996, 272, 1323-1325. (9) Thomas, P. J.; Kulkarni, G. U.; Rao, C. N. R. Chem. Phys. Lett. 2000, 321, 163-168. (10) Feldheim, D. L.; Keating, C. D. Chem. Soc. Rev. 1998, 27, 1-12. (11) Brousseau, L. C.; Zhao, Q.; Shultz, D. A.; Feldheim, D. L. J. Am. Chem. Soc. 1998, 120, 7645-7646.

1828

Langmuir, Vol. 18, No. 5, 2002

Figure 4. FTIR grazing-angle microscopy analysis of dsDNAnanoparticle, dsDNA, and ssDNA monolayers. Monolayers were self-assembled on gold-coated microscope slides. Resolution: 4 cm-1. Inset: complete spectrum showing a broad band beyond 1200 cm-1 for dsDNA-nanoparticle monolayers. The data were obtained using a grazing-angle objective mounted on a UMA500 FTIR microscope (BioRad) attached to a FTS6000 FTIR spectrometer (BioRad).

smaller.12 The Coulomb staircase will not be observed for dimers, trimers, or multimer aggregates of 5 nm nanoparticles. The data shown in Figure 3 suggest that the 5 nm particles are acting as independent capacitors in an equivalent circuit for current flow from the tip to the gold surface. The I-V behavior shown in Figure 3 is indicative of the Coulomb staircase behavior even for densely packed dsDNA-nanoparticle monolayers. This interparticle distance obtained from the above analysis is consistent with the Debye-Hu¨ckel theory. The double layer due to the coating of BSPP molecules leads to mutual repulsion between the particles. Based on an analysis of the STM data, the calculated separation distance is 8 ( 3 nm. Consideration of the surface charge groups and double-layer effects provides a number consistent with this experimental value. Charged BSPP groups on the gold nanoparticle add roughly 1 nm to the radius. A Debye length of 1/κ ≈ 1 nm was estimated for the 150 mM monovalent salt concentration used to promote hybridization.13 The electrostatic potential due to the negative charge on these BSPP molecules will fall to 10% of its value at a distance of 2 nm from the surface. These factors suggest that 6 nm is a minimum separation distance between particles. The presence of oriented dsDNA-nanoparticle conjugates on the surface was further confirmed by grazingangle FTIR and Raman studies. FTIR grazing-angle microscopy was used to identify the presence of ssDNA, dsDNA, and dsDNA-nanoparticle adducts on the surface (Figure 4). ssDNA (Figure 4, dotted line) and dsDNA (Figure 4, dashed line) alone presented virtually featureless spectra while nanoparticle-labeled dsDNA (Figure 4, solid line) showed spectral features consistent with DNA and nanoparticles bound to the gold surface (Figure 4, solid): symmetric and antisymmetric stretching vibrations of the phosphodiester backbone (peaks at 1126 and 1222 cm-1), base-sugar vibrations (broad peak at 1410 cm-1), and weak stretching vibrations of the DNA bases (around 1650 cm-1).14 Carbon-hydrogen stretching (peaks between 2840 and 3000 cm-1) and oxygen-hydrogen stretching (broad band around 3500 cm-1) of the MH passivation layer were also observed. A broad band starting around (12) Matsumoto, K.; Ishii, M.; Segawa, K.; Oka, Y. Appl. Phys. Lett. 1995, 68, 34-36. (13) Sonntag, H.; Strenge, K. Coagulation Kinetics and Structure Formation; Plenum Press: New York, 1987. (14) Boncheva, M.; Scheibler, L.; Lincoln, P.; Vogel, H.; Akerman, B. Langmuir 1999, 15, 4317-4320.

Sauthier et al.

Figure 5. (A) Raman spectra for a monolayer of 5 nm BSPPcoated nanoparticles attached to a gold substrate via 25mer dsDNA. The presence of nanoparticles bound to the surface is confirmed by the detection of a BSPP Raman spectrum (∼1070 and 1580 cm-1, bottom curve); the presence of aggregates on the surface yields broad and intense spectra (top curve). (B) Raman spectra for a monolayer of 5 nm nanoparticles coated with a mixture of BSPP and MG and attached to a gold substrate via 25mer dsDNA. The presence of nanoparticles bound to the surface is characterized by a Raman spectrum presenting bands corresponding to both BSPP and MG (∼1080, 1170, 1295, 1375, and 1580 cm-1, bottom curve). Aggregates produce broad and intense Raman spectra (top curve).

1200 cm-1 and increasing in intensity at higher wavenumbers extended beyond the 7000 cm-1 cutoff of the mercury-cadmium telluride detector of the FTIR spectrometer. This band is attributed to a plasmon effect due to the presence of nanoparticles on the surface. The striking difference between the two spectra shown in Figure 4 arises from a change in orientation of the dsDNA when tethered to a nanoparticle. Surface selection rules indicate that the transition moments of groups perpendicular to the surface will be observed and those parallel to the surface will not be observed. The appearance of relatively strong phosphodiester asymmetric and symmetric modes relative to the base in-plane vibrations suggests that the dsDNA helix axis makes a significant projection along the surface normal. Thus, since the phosphodiesters are observed, we propose that the helix axis is likely not lying flat the surface. Raman microscopy is an ideal technique for optical detection of aggregates of nanoparticles on surfaces. Since nanoparticle aggregates are known for their strong surface-enhanced Raman scattering effect, it is reasonable to associate strong light scattering with the presence of aggregates on the surface.15,16 Scanned Raman microscope line images revealed two distinct classes of surface geometries (Figure 5). The first class of surface geometries was localized and presented strong Raman bands between 1100 and 1650 cm-1 (Figure 5A, top spectrum). These strong bands suggest surface enhancement and appear to arise from the BSPP coating of the nanoparticles rather than from the dsDNA. The position of these highly scattering areas coincided with the position of large surface aggregates visible with an optical microscope. The second class of surface geometries, representative of a majority of the surface, yielded weak Raman signals with bands around 1080, 1180, and 1570 cm-1 corresponding to the BSPP coating on the particles but not to the dsDNA (Figure 5A, bottom spectrum). These regions presenting weak Raman signals must correspond to dense dsDNA-nanoparticle monolayers. This general interpretation is substantiated by the STM images that we have obtained (15) Moskovits, M. Rev. Mod. Phys. 1985, 57, 783-826. (16) Freeman, R. G.; Bright, R. M.; Hommer, M. B.; Natan, M. J. J. Raman Spectrosc. 1999, 30, 733-738.

Nanoparticle Layers through DNA Hybridization

Figure 6. Noncooperative adsorption of ssDNA probes onto a gold substrate as analyzed by STM. Targets were labeled with 5 nm gold nanoparticles. Labeled target concentration: 5 nM. At probe percentages above 20%, a large number of aggregates were observed. Those aggregates made the precise determination of surface density impossible.

showing regions of relatively uniform coverage punctuated with a few large, isolated aggregates. Similar results were obtained when gold nanoparticles substituted with a mixture of BSPP and methyl green (MG), a dye absorbing in resonance with the laser excitation, were used as target labels (Figure 5B). While the MG spectrum can be seen as weak Raman bands in the lower spectrum of Figure 5B, surface aggregates seen in the upper spectrum of Figure 5B yielded intense Raman bands very similar to those observed in Figure 5A for only BSPP capping. dsDNA-nanoparticle monolayers prepared by sequential deposition or codeposition showed similar results. The ability to control surface coverage is crucial for the design of more efficient DNA hybridization detection strategies based on a DNA array format. Controlling the surface coverage of dsDNA-nanoparticle conjugates stabilized by DNA hybridization is a complex problem that involves an understanding of probe coverage in a mixed monolayer and of interparticle interactions at the solution-surface interface. It is well-known that nanoparticles tend to aggregate in solution by overcoming an electrostatic barrier.13 A similar process must occur at the surface, but with a lower dimensionality. Based on experiments described below, three factors have an effect on surface coverage: the density of probes on the surface, the concentration of nanoparticle-labeled targets in the hybridization solution, and the nanoparticle size. The density of probes on the surface can easily be modified by changing the percentage of ssDNA probes in the deposition solution used in the codeposition method. Figure 6 shows that for 5 nm particles, surface density increased when the molar percentage of probes was increased in the codeposition solution. However, a large number of aggregates were observed for surfaces made with >20% (mole percent) of probe molecules in the codeposition solution. Tarlov and co-workers observed reduced hybridization efficiency for dsDNA at similar high surface coverages.7,17 Although Figure 6 presents an indirect measurement of adsorbed probe concentration via final labeled target surface densities, it suggests that the coadsorption of ssDNA probes and MH onto gold substrates is noncooperative. It was found that 20% probes in the codeposition solution (1 µM total) was ideal for both (17) Steel, A. B.; Herne, T. M.; Tarlov, M. J. Anal. Chem. 1998, 70, 4670-4677.

Langmuir, Vol. 18, No. 5, 2002 1829

5 and 10 nm a nanoparticles in order to obtain dense monolayers and a minimal number of surface defects. Next, the influence of labeled target concentration on the surface density was studied. Figure 7 and Figure 8 show that increasing the concentration of DNA targetnanoparticle conjugates in the hybridization solution increases the surface coverage for both sizes of particles and for a given probe concentration. Low concentrations of labeled targets tend to give very sparse monolayers, whereas higher concentrations favor the formation of multilayers, which may be due to nonselective interactions between DNA targets in solution and neighboring nanoparticles already bound to the surface. In addition, Figure 8 illustrates that the adsorption of the labeled target is positively cooperative, that is, labeled target binding increases the affinity of neighboring sites. The corresponding Hill factor and the equilibrium constant for labeled target binding were determined to be 3.26 and 3.8 × 10-5, respectively, by using the model function θ ) K[A]n/ (1 + K[A]n) where [A] is the concentration of labeled targets, n is the Hill factor, and K is the equilibrium constant. Whether the cooperation occurs before or after the binding of the labeled targets to the surface remains unclear. Finally, it was shown that the size of the nanoparticle used as a label has a great influence on the final density of targets bound to the surface. Figure 9 shows that 10 nm particles tend to pack more densely than 5 nm particles for given probe and nanoparticle-labeled target concentrations. This is consistent with the fact that smaller particles have a greater charge-to-volume ratio. Therefore, Coulombic repulsion between two identically charged particles is stronger for smaller particles.13 The fact that 10 nm particles tend to form more densely packed monolayers means that they are also more prone to the formation of defects such as aggregation and formation of multilayers as illustrated in Figure 9. Therefore, 5 nm particles allow a better control of the monolayer homogeneity. Conclusion The formation of a monolayer of nanoparticles on a surface by DNA hybridization to surface-attached probe oliognucleotides provides the basis for a number of possible diagnostic methods. Electrochemical, surface plasmon resonance, optical, and scanometric methods all rely on the specificity of DNA hybridization on a surface. We have demonstrated that dense monolayers of nanoparticlelabeled dsDNA can be obtained with high homogeneity and a low occurrence of surface aggregates and multilayer defects. FTIR allowed identifying the presence of both dsDNA and nanoparticles on the surface, thus further confirming that the particles were bound to the surface by DNA hybridization. Raman microscopy proved to be a useful tool for the detection of large surface-bound aggregates and of dsDNA-nanoparticle monolayers by sensing the particle coating. STM characterization showed that nanoparticles that form a monolayer bound to the surface are well separated and electrically isolated from their neighbors, but STM also revealed that nonselective binding of labeled targets creates a background under the current conditions. The fact that 5 nm particles showed less aggregation than 10 nm particles suggests that nonselective binding can be greatly reduced by the use of still smaller particles. In addition, the capping ligand used will also strongly affect the results. For example, the data obtained here would not have been possible using citratecapped nanoparticles because their tendency to aggregate is much greater than that of the BSPP-capped particles.

1830

Langmuir, Vol. 18, No. 5, 2002

Sauthier et al.

Figure 7. Influence of target concentration on surface coverage. All samples were prepared from a codeposition solution containing 20% ssDNA probe; hybridization cooling rate, 5 h. Picture sizes are 500 × 500 nm. (A) 5 nm particles, 1 nM labeled target solution. (B) 5 nm particles, 5 nM labeled target solution. (C) 5 nm particles, 10 nM labeled target solution.

Figure 8. Positive cooperative behavior presented by the adsorption of ssDNA targets labeled with 5 nm particles. Surface densities were measured by STM. Probe concentration, 20%; hybridization cooling rate, 4 h. At target concentrations above 10 nM, the determination of surface coverage becomes difficult by STM due to a lack of resolution between adjacent particles and the presence of an increasing number of defects, nominally, aggregates and multilayers.

Figure 9. Influence of nanoparticle size on surface coverage. All samples were prepared from a codeposition solution containing 20% ssDNA probe; labeled target concentration, 5 nM; hybridization cooling rate, 5 h. Picture sizes are 500 × 500 nm. (A) 5 nm particles. (B) 10 nm particles.

Surface coverage optimization studies revealed that three factors affected surface coverage, that is, probe density, labeled target concentration, and particle size. Increasing the probe molar percentages in the codeposition solution resulted in higher labeled target surface densities for a given labeled target concentration. It was found that a mole fraction percent of 20% probe in the codeposition solution promotes the formation of dense monolayers of

labeled targets for 5 nm particles without forming large aggregates. Greater concentrations of ssDNA than 20% in the deposition solution led to saturation of the surface and promoted the formation of surface aggregates. A concentration of 10 nM labeled targets was found to be optimum for the formation of dense monolayers. At this concentration, the number of aggregates on the surface is low and surface coverage is nearing saturation. We have also shown that probe adsorption follows a noncooperative isotherm whereas labeled target adsorption shows positive cooperativity. Finally, it was found that 5 nm particles are preferable to larger nanoparticles because of their ability to form relatively homogeneous monolayers with fewer defects. In conclusion, we have demonstrated that a proper selection of probe density (20%), target concentration (10 nM), and nanoparticle size (5 nm) provides nearly complete surface coverage while minimizing the formation of surface-bound aggregates. Given that the selectivity of binding will be impeded by surface aggregates, these factors are important for future development of nanoparticle-based hybridization assays. Acknowledgment. We thank Dr. Dan Feldheim, his group, and the Nucleic Acid Facility at North Carolina State University for their input in this project. S.F. acknowledges North Carolina State University for financial support through start-up funds and Applied Biosystems for support through a sponsored research agreement. C.B.G. acknowledges support from the Air Force Office of Scientific Research MURI Program in Nanoscale Chemistry and by the National Science Foundation (CAREER Award, DMR-9600138). We thank Dr. Dwight Walker and GlaxoSmithKline for the use of their SPEX Infinity Raman microscope. Supporting Information Available: 1. The electrical characterization of the individual dsDNA-nanoparticle moiety compared to a control gold substrate; 2. STM characterization of dsDNA-nanoparticle monolayers for high probe densities; 3. concentration dependence of nonselective binding of labeled targets onto a MH-functionalized gold substrate; 4. determination of probe surface coverage by chronocoulometry of ssDNA; 5. additional topological data representative of the data given in the paper; 6. surface analysis before and after heating; 7. supplementary information on aggregates. This material is available free of charge via the Internet at http://pubs.acs.org. LA0112763