Abstract

We demonstrate that Au-cluster-decorated single-walled carbon nanotubes (SWNTs) may be used to discriminate single nucleotide polymorphism (SNP). Nanoscale Au clusters were formed on the side walls of carbon nanotubes in a transistor geometry using electrochemical deposition. The effect of Au cluster decoration appeared as hole doping when electrical transport characteristics were examined. Thiolated single-stranded probe peptide nucleic acid (PNA) was successfully immobilized on Au clusters decorating single-walled carbon nanotube field-effect transistors (SWNT-FETs), resulting in a conductance decrease that could be explained by a decrease in Au work function upon adsorption of thiolated PNA. Although a target single-stranded DNA (ssDNA) with a single mismatch did not cause any change in electrical conductance, a clear decrease in conductance was observed with matched ssDNA, thereby showing the possibility of SNP (single nucleotide polymorphism) detection using Au-cluster-decorated SWNT-FETs. However, a power to discriminate SNP target is lost in high ionic environment. We can conclude that observed SNP discrimination in low ionic environment is due to the hampered binding of SNP target on nanoscale surfaces in low ionic conditions.

1. Introduction

The ability to detect small sequence differences in DNA molecules is very important in molecular biology, especially in the context of personalized diagnostics and therapy [1]. PCR combined with fluorescence detection is the most widely used and commercialized technique to date; however, this method requires expensive, bulky instruments, which has hindered the popularization of DNA-based diagnostics. There is a great demand for cost-effective, miniaturized DNA diagnostic sensors. Electronic detection platforms such as electrochemical or field-effect transistor sensors may be ideal, because electronic sensors do not require expensive instrumentation, and large-scale production is easy using microfabrication technology. Extensive efforts have been made to develop such DNA sensors. Two distinct approaches have been taken. In one approach, oligomers are used to measure hybridization of target DNA, while in the other, mutations are probed using enzymes recognizing single nucleotide polymorphisms (SNPs), deletions, translocations, or inversions [2, 3]. The first method (hybridization detection) using hairpin probe DNAs with redox or enzyme tags has been quite effective in development of electrochemical sensors [46], and most reported DNA sensors of the FET type use this method. For example, Hahm and Lieber reported real-time detection of DNA using peptide nucleic acid (PNA-) modified Si nanowire devices [7], and DNA sensors based on single-walled carbon nanotube field effect transistors (SWNT-FETs) have also been reported [8]. Silicon nanowire devices have the advantage of well-established surface chemistry techniques borrowed from the silicon industry. However, immobilization of probe molecules onto carbon nanotubes remains challenging. If probe molecules are covalently bound to the nanotubes, the carbon bonds will be broken, thus destroying the excellent electrical transport characteristics of individual nanotubes. Thus, biosensors using SWNT-FETs bearing noncovalently attached linkers such as pyrene or CDI-Tween 20 have been developed [9, 10]. However, the successful use of noncovalent binding linkers for hydrophilic molecules such as DNA remains to be demonstrated.

Recently, Star et al. showed label-free detection of DNA hybridization using nanotube network transistors [8]. In their example, probe single-stranded DNAs (ssDNAs) were nonspecifically adsorbed onto carbon nanotubes by π-stacking between DNA bases and the sidewalls of SWNTs. Such nonspecifically adsorbed molecules cannot easily bind with complementary strands; it has been shown that ssDNA forms a complex with an SWNT by wrapping around the SWNT [11, 12]. Jeng et al. showed that although target DNA could hybridize with probe ssDNA on SWNTs, the process was relatively slow (13 hours at , compared to less than 10 minutes for free DNA hybridization) [13]. Therefore, an immobilization method better than that offered by simple nonspecific binding is required. To that end, we developed Au-nanoparticle-decorated SWNT-FETs as sensor platforms for label-free detection of DNA hybridization. Thiolated DNA capture probes bind to the Au nanoparticles on SWNTs, but the nanoparticles do not disturb the excellent electrical properties of SWNTs. Thus, we may covalently attach DNA capture probes and measure electrical responses upon hybridization. Metallic nanoparticles may offer a solid support for molecular recognition elements, and incorporating metal nanoparticles into SWNT sensors has been shown to effectively enhance the sensitivity [1416].

2. Results and Discussion

The fabrication of devices using patterned growth techniques and the fabrication details are described in the Method section. Figure 1(a) shows an atomic force micrograph of a device decorated with Au nanoparticles. Devices with only one or two nanotubes between electrodes were used for all experiments. Figure (b) (in Supplementary Material available online at http://dx.doi.org/10.1155/2011/105138) shows the change in characteristics upon Au decoration by electrochemical deposition. Several different approaches are available for Au nanoparticle decoration of SWNTs. For example, SWNTs may be decorated by chemical modification [17]. However, a “physical” decoration method is preferable. We have explored a simple evaporation method as well as spontaneous reduction [18]. With physical evaporation, nonspecific probe immobilization is problematic, and it is difficult to control Au nanoparticle size with spontaneous reduction. Therefore, we used electrochemical deposition [1921].

Unlike the electroless deposition method, electrodeposition of Au nanoparticles does not require oxidative consumption of nanotubes. Therefore, the observed hole-doping effect may be explained by the work function difference between the metal and the SWNT. When the high-work-function metal Au ( 5.2 eV) is brought into contact with an SWNT, electrons will be transferred from the SWNT to the Au by virtue of the work function difference [22]. One of the advantages with metal decoration on SWNTs is that the reactions can be observed with optical microscope. A single SWNT decorated with Au nanoparticles larger than 50 nm can be directly imaged with optical microscopes, and dye molecules immobilized on Au nanoparticles can be imaged with confocal laser microscope. As shown in Figure 1(b) and Figure in Supplementary Material, strong fluorescence signals observed upon Cy-3-labeled PNA binding. Usually, fluorescence signals from metal surfaces are low because of surface energy transfer [23]. The enhanced fluorescence that we observed presumably arose from coupling between Au nanoparticles [24].

As a probe, we used PNA rather than DNA. PNA is an artificially synthesized polymer similar to DNA and is used in both biological research and medical treatment [25]. Whereas DNA has a negatively charged phosphate backbone, the PNA backbone is composed of repeating N-(2-aminoethyl)-glycine units linked by peptide bonds. PNA has remarkable thermal, chemical, and enzymatic stability, and unique ionic strength effects relative to the natural analog, DNA. Hybridization of ssDNA with complementary target DNA can be performed only in an ionic environment that stabilizes the negatively charged backbones, because of Coulombic repulsion between negatively charged molecules. However, as PNA has a neutral peptide backbone, constraints on hybridization conditions are less severe than with DNA-DNA hybridization [26].

PNA probe sequences can be covalently immobilized to Au nanoparticles decorated on SWNT-FETs using Au-S binding. In our experiment, -HS-GAC  ATT  ACT  CAC  CCG-Cy3- , with the end terminated by a thiol, was used as a probe PNA, whereas -CGG  GTG  AGT  AAT  GTC- was the target. The DNA sequences were chosen from an Escherichia coli  library and synthesized by Bioneer Inc (Daejeon, Korea). Probe PNA solution was deposited on SWNT-FETs decorated with Au-nanoparticles and allowed to react for 6 hours in a humidified chamber at room temperature (Figure in Supplementary Material). To confirm successful immobilization, Cy-3-labeled probe PNA was employed. Figures 1(a) and 1(b) show an AFM image and a confocal microscope image of an Au-decorated SWNT-FET bearing Cy-3-labeled PNA.

To measure hybridization of target DNA electronically, we first immersed the PNA-immobilized, Au-decorated SWNT-FET in DI water with an Ag/AgCl reference electrode as a liquid gate and applied bias voltages (100–300 mV) to the device with measurement of current in real time. Positive bias voltages were applied to enhance the hybridization of target DNA [27]. When device signals became stabilized, we added 5  L of 100 pmol/ L target DNA solution in DI water. Figure 2(a) shows real-time measurements from the device. A decrease in conductance was observed as soon as complementary target DNA was added. After about 20 minutes, the conductance saturated at 40% of the original value. The inset of Figure 2(a) shows an AFM image of the Au-decorated device. Figure 2(b) shows the evolution of characteristics upon immobilization of PNA and hybridization of DNA.

The original device had ambipolar characteristics, and, as is clear from the graph, the gate threshold voltage showed a further positive shift after Au decoration, which may be explained by the hole-doping effect of Au nanoparticles. The decrease in the on-current observed after decoration is possibly attributable to enhanced scattering at the nanotube surface [15], but the change of electrical characteristics upon Au decoration is also dependent upon decoration condition (particle size, density, etc.). Immobilization of probe PNA results in a decrease in conductance, presumably because of a reduction in the Au work function upon Au-S binding [28, 29]. However, evolution of gate threshold is not clear in these 5  m-long channel Au-decorated SWNT-FETs after PNA immobilization. We can also consider the effect of capacitance change as a dominant mechanism behind the observed conductance change. It is known that capacitance mechanism plays a major role only in the case of nearly-full covered biomolecules [30], and we expect highly packed PNA molecules on Au nanoparticles with 6 hours incubation. Hybridization of target DNA results in a further conductance decrease, again probably arising from a drop in the Au work function upon hybridization or decreased capacitance. Recently, lowering of the Au work function after immobilization of probe DNA and hybridization with target DNA has been observed by Kelvin probe microscopy (KPFM) [31]. Also, DNA sensors based on nanotube Schottky transistors showed decreased conductance upon hybridization related to falls in Au work function [32, 33].

To confirm that the observed change of conductance indeed originated from hybridization of target DNA, we fabricated Au-decorated ultralong SWNT-FETs. Ultralong SWNTs were grown on SiO2/Si substrates [34], and electrical leads were patterned using photolithography and liftoff. Immobilization of PNA and hybridization of DNA were performed as in the case of 5  m-channel devices. Figure 3(a) shows a confocal microscope image of an Au-decorated ultralong SWNT-FET after hybridization of Cy-3 labeled target DNA. Inset of Figure 3(a) is an optical micrograph of the sample, and the white arrow indicates Au-decorated SWNT. The channel length of this device is 100  m, and the active channel length decreased to 80  m after electrode passivation. Note that SWNT is visible with normal optical microscope after Au decoration. As shown in the confocal micrograph of Figure 3(a), clear fluorescence signal observed from Au nanoparticles after hybridization of Cy-3-labeled DNA. We also confirmed that hybridization occurred mainly on Au nanoparticles by using an ultralong nanotube decorated with only a few Au nanoparticles (Figure in Supplementary Material). Figure 3(b) shows an evolution of gate-transfer characteristics from ultralong SWNT. Again, decoration of Au nanoparticles appears as hole doping (positive shift of gate threshold voltages), and immobilization of PNA appears as decrease of conductance as well as negative shift of gate threshold voltages. Hybridization of target DNA yields further decrease of conductance. Since the active channel length is much longer in ultralong SWNTs, we may expect that pronounced doping effect (gating effect) plays a role in them. Decrease of conductance upon hybridization of matching sequence was universal for both short-channel (5  m) and long-channel (80  m) devices. In Figure in Supplementary Material, we showed electronic transfer characteristics of 5 different devices [(a), (b), and (c) from short-channel devices, while (d) and (e) are from long-channel devices]. As is clear from the graphs, decoration of Au nanoparticles appears as hole doping, and immobilization of probe PNA causes decrease of conductance, as well as negative shift of gate threshold voltages. Hybridization of target DNA appears as a decrease of conductance and negative shift of gate threshold voltage in all five samples, though conductance change could vary 20 50% from different devices. To test the advantages of Au-decorated SWNT-FET, we performed the same experiment with nonspecifically bound probe PNA molecules. Bare SWNT-FET devices were incubated with PNA solution for 6 hours, and hybridization experiment was followed after thorough washing. However, devices with nonspecifically bound PNAs show little consistency. As shown in Figure of Supplementary Material, 60% of the devices showed decrease of conductance, 30% of the devices did not show any change of conductance, and the rest showed increase of conductance. Since the nonspecific binding of probe PNA (DNA) depends heavily on the chirality of individual nanotubes, it may be difficult to have consistency with individual SWNT devices that have random chirality.

To determine whether the Au-decorated SWNT-FET could detect an SNP, we used -CGG  GTG  AAT  AAT  GTC- as target ssDNA. Figure 4(a) shows real-time conductance values from this experiment. In contrast to the current change seen when a perfectly matched target was added, use of target ssDNA with a single mismatch did not alter the device current. After incubation for 20 minutes, the device was washed with an ample amount of DI water and blown dry with dry N2 gas. measurements taken after this did not show large changes either (Figure 4(b), black curve). Then, target ssDNA with perfect match was introduced to the same device baring 5  L DI water. As shown in Figure 4(a), clear decrease of conductance was observed upon introduction of matching DNA. Since we used the same concentration of target DNA in both SNP targets and matching targets (100 pmole/ L), we may assume that observed change of conductance is not due to the change of ionic conditions or change of pH. measurement taken after washing and drying shows a clear decrease of conductance as well (Figure 4(b), orange curve).

We next employed a hybrid buffer to improve hybridization between probe PNA and target DNA sequence with an SNP to determine whether a single mismatch could really alter electrical characteristics. The hybrid buffer was 3 M NaCl, 10 mM ethylenediaminetetraacetic acid (EDTA), and 380 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES). Target DNA oligomers in the hybridization buffer were allowed to react with PNAs immobilized on Au-decorated SWNT-FETs for 20 minutes. The chip was then washed with an ample amount of DI water to eliminate nonspecific binding and blown dry with N2 prior to further electrical characterization. We did not measure real-time changes in conductance as the high salt concentration of hybrid buffer made such assessment impossible. Figure 5(b) shows characteristics of the device before and after hybridization. In contrast to the absence of characteristic changes after hybridization in DI water, a clear decrease of conductance was observed after reaction in the hybrid buffer, as seen when sequences matched perfectly.

We thus conclude that the observed single-base mismatch discrimination shown by our device is likely not attributable to an SNP charge effect, as is the case in strain sensors [35, 36], but rather because hybridization of the SNP target is more difficult in DI water. To verify this, we labeled target SNP-bearing ssDNA with Cy-3 and hybridized this material to PNAs immobilized on Au-decorated SWNT-FETs in DI water for 20 minutes. No fluorescent signal was seen, in contrast to the result when perfectly matched target DNA was used for hybridization (data not shown). Park et al. have shown that ionic strength can affect PNA-DNA surface hybridization, although the influence of ionic strength is not as great as seen with DNA-DNA hybridization [37]. As we expect that probe PNAs are fully packed on Au nanoparticles, hybridizing sequences can communicate electrostatically with their neighbors, thereby hindering the hybridization of mismatched sequences.

In summary, we combined electrical transport in Au-decorated SWNT-FET with confocal microscopy to measure hybridization of target ssDNA with PNAs. Uniform-sized Au nanoparticles were formed on sidewalls of SWNTs by electrochemical deposition, and the effect of Au-decoration appears as hole doping in most cases. By addition of thiolated PNA probes to such a device, we achieved stable immobilization of molecular recognition elements with minimal effects on electrical properties. Hybridization of complementary DNA in DI water resulted in a decrease in real-time conductance, and we confirmed that hybridization has indeed occured on Au nanoparticles by measuring fluorescence signal from Cy-3-labeled target DNA. In contrast, introduction of SNP target yields negligible change of conductance is in DI water. However, clear decrease of conductance observed when hybridization with SNP target is performed in hybridization buffer; SNP target is indiscernible from perfect match in high ionic conditions. Therefore, we may conclude that hybridization of mismatched DNA with PNA immobilized on nanoscale surfaces is difficult in low ionic conditions, at least in the timescale of 20 minutes.

3. Experiments

3.1. Device Fabrication

SWNTs were grown on a Si/SiO2 substrate using a patterned chemical vapor deposition (CVD). Liquid catalyst consists of Fe(NO3)2·9H2O, and Mo(acac)2 in methanol was dispersed on substrates with patterned poly(methyl mettacrylate). After liftoff, SWNT growth was carried out in -heated furnace for 10 minutes with CH4 or ethanol as a carbon source. To grow ultralong SWNTs, 0.5 M FeCl3 in DI water was placed at the edge of the sample. Ethanol was used as a carbon source, and the growth was carried out at for 20 30 minutes. Patterns for electrical leads were formed by photolithography, followed by thermal evaporation of Ti and Au. For real-time characterization and metal decoration, all electrodes were insulated with Si or SU8-2000.5 negative photoresist.

3.2. Au Decoration

By cyclic voltammetry or chronoamperometry, it is possible to control both Au nanoparticle size and packing density. Scheme 1 shows a schematic diagram of our Au-decoration process. A droplet of 100  M HAuCl4 in 100 mM KCl (supporting electrolyte) was placed on the device, both source and drain electrodes were used as working electrodes, and a reduction potential was applied with respect to the Ag/AgCl reference electrode with current monitoring using a Pt counter electrode. We found that cyclic voltammetry afforded better decoration control than did chronoamperometry. Uniform 20- to 60-nm-sized Au nanoparticles were formed on SWNTs by sweeping the potential from 0.2 V to 0.6 V.

105138.sch.001

Supporting Information Available

Optical microscope images and AFM images of Au-decorated CNT and electrical transport that show the effect of Au decoration, confocal micrographs showing Cy-3-labeled probe PNA immobilization condition, both a confocal micrograph and a scanning electron micrograph of Au-decorated ultralong nanotubes after Cy3-labeled DNA hybridization, Electronic transfer characteristics of 5 different devices upon hybridization of DNA, and statistic of SWNT-FETs with nonspecifically bound PNA after the hybridization of matching DNA.

Acknowledgments

Financial support was obtained from the Korea Research Council for Industrial Science and Technology, and Ministry of Environment as “The Eco-technopia 21 project.” The authors thank Professor Hee Chul Choi and his group members for valuable discussion and assistance during this work and Professor Jing Kong’s group for their help in growing ultralong nanotubes.

Supplementary Materials

In Figure S1, we showed optical microscope images and AFM images of Au-decorated SWNT-FET. Figure S1(b) shows evolution of electronic transfer characteristics of SWNT-FET with Au decoration. Supplementary Figure S2 (a), (b), and (c) show confocal microscope images of Au-decorated ultra-long nanotubes with Cy-3 labeled PNA. It is possible to achieve a highly specific immobilization of probe PNA with 6 hrs incubation and thorough washing. Figure S3 also shows confocal microsope image and a SEM image of Au-decorated ultra-long nanotube. The figure displays that hybridization of target DNA occur mostly on Au nanoparticles. Figure S4 shows evolution of transfer characteristics from 5 different devices upon immobilization of Au, probe PNA, and hybridization of DNA. In Figure S5, we showed the results from control hybridization experiments with non-specifically adsorbed probe PNA. Change of conductance upon hybridization show little consistency in this case.

  1. Supplementary Material