DNA nanostructure-based “soft lithography” for the programmable

engineering of biosensing interfaces

Meihua Lin1, Jingjing Wang1,2, Guobao Zhou1,3, Jianxin Lu2, Jimin Gao2, Xiaoqing

Chen3, Jiye Shi1, Xiaolei Zuo1*, Chunhai Fan1*

1
Division of Physical Biology & Bioimaging Center, Shanghai Synchrotron Radiation Facility,

Shanghai Institute of Applied Physics, Chinese Academy of Sciences, Shanghai 201800, China

2
School of Medical Lab Science and Life Science, Wenzhou Medical University, Zhejiang 325035,

China

3
College of Chemistry and Chemical Engineering, Key Laboratory of Resources Chemistry of

Nonferrous Metals, Central South University, Hunan 410083, China

*Corresponding authors’ Emails: [email protected], [email protected]

The sensitivity of biomolecular detection is limited by the low mass transport rate of

the target molecules to macroscopic biosensors (millimetre to centimetre) and/or by

the low probability of probe-target collision for micro-/nano- biosensors1-11. A

promising solution to address these limitations is to introduce nanostructures on

macroscopic biosensing surfaces, which can increase both the mass transport rate of

the target molecules in solution and the absolute numbers of probe molecules on the

surface2,10,11. However, fabricating reproducible nanostructured surfaces via either

surface etching or the deposition of metal nanoparticles remains technically

challenging12-16. Here, we show that self-assembled DNA nanostructures with precise

sizes allow a programmable “soft lithography” approach to engineer the interface of

electrochemical DNA sensors. By using millimetre-sized gold electrodes modified

with several types of tetrahedral DNA nanostructures (TDNs) of different sizes, we

demonstrated that both the kinetics and thermodynamics of DNA hybridisation were

profoundly affected. Because each DNA probe is anchored on an individual TDN, its

lateral spacing and interactions are finely tuned by the TDN size. By simply varying

the size of the TDNs, we decreased the hybridisation time and increased the

hybridisation efficiency. More significantly, we were able to tune the detection limit

for DNA detection over three orders of magnitude with differentially nanostructured

electrodes, and achieved attomolar sensitivity with polymeric enzyme amplification.

Hence, our study provides a semi-quantitative approach to rationally engineer

biosensing interfaces and to modulate biomolecular interactions on heterogeneous

surfaces.
There have been increasing demands for the development of rapid and sensitive

biosensors in resource-limited settings, including in small clinics and for emerging

infectious diseases and biological warfare2,3,17-19. However, despite rapid progress in

biosensing studies, rationally designing biosensors with desirable speed and

sensitivity remains a hurdle2,10,11. In particular, it is critically important to engineer

biosensing interfaces to increase the kinetics and thermodynamics of biomolecular

recognition2,10. However, the currently available approaches are largely empirical and

usually rely on parameter optimisation5,7,20,21. To address these problems, we

developed a semi-quantitative approach for the rational interfacial engineering of

biosensors to improve biosensing performance.

The sensitivity of biomolecular detection is limited not only by the affinity of the

biomolecules but also by the interfacial properties of the biosensor2. At macroscopic

interfaces, the diffusion of biomolecules from the bulk solution to the surface is nearly

linear in nature, whereas mass transport is dominated by radial diffusion at micro- or

nanoscopic interfaces2,10. Thus, the size reduction of biosensors usually increases the

mass transport rate and improves sensitivity. Nevertheless, the limited space available

in nanosensors restricts the absolute number of immobilised probe molecules, which

reduces the probability of collision and binding of the probe and target molecules.

Indeed, theoretical studies have revealed that the detection limit of a nanosensor was

unlikely to exceed 1 femtomolar within a practical time scale2,10. In light of these

difficulties, trans-scale biosensors that incorporate nanoscale features into large-sized

macroscopic surfaces are expected to solve this size dilemma.

Intuitively, nanostructures can be incorporated into a macroscopic sensor surface by

either surface etching or the deposition of metal nanoparticles12-15. However,

reproducibly preparing such nanostructured surfaces with well-defined surface areas

remains technically difficult16. Photolithography potentially provides a route to the

high-resolution fabrication of nanostructures but is nevertheless hampered by its high

cost and the difficulty of making 3D structures22. Here, we developed a conceptually

new “soft lithographic” strategy to programmably and reproducibly engineer a

biosensing interface using well-defined 3D DNA nanostructures. Owing to their

highly specific Watson-Crick base pairing, DNA molecules can be self-assembled into

DNA nanostructures of various sizes, shapes and geometries with high predictability

and precision9,12,14,23-28. As an elegant example, tetrahedral DNA nanostructures

(TDNs) of sub-10-nm sizes have been shown to be a rigid scaffold for the oriented

anchoring of DNA probes. Here, we aimed to study the DNA surface hybridisation

regime (kinetics and thermodynamics) with nanometre precision and programmably

tune the detection limit of DNA sensors by using macroscopic gold electrodes

patterned with different sized TDNs (Figure 1).

As the test bed for our interrogation of DNA sensors, we designed three types of

TDNs with different sizes: TDN7, TDN13 and TDN17, in which each edge of the

TDN contains 7, 13 or 17 base pairs, respectively. All TDNs were rationally designed

such that no undesired secondary structures were present at the edges. Because each

base pair was separated by 0.34 nm in a double helix, we were able to precisely

calculate the edge length of these TDNs to be 5.8 nm, 4.4 nm and 2.4 nm, respectively.
Each TDN carries a pendant DNA probe at one vertex and three thiol groups at the

other three vertices, which can be firmly anchored at the Au surface (Figure 1 and

Figure S1) via well-established Au-S bonds. Native polyacrylamide gel

electrophoresis (PAGE) analysis showed that these TDNs could be self-assembled,

with high yields of 85-90% (Figure S2 and Figure S3). To further confirm the

formation of the TDNs, we modified two strands of the TDNs with a pair of

donor-acceptor dyes (Cy3 and Cy5) for fluorescence resonance energy transfer (FRET)

analysis. Cy3 and Cy5 were expected to be separated by 7, 13 and 17 base pairs for

TDN-7, TDN-13 and TDN-17, respectively. The presence of FRET confirmed the

successful assembly of the DNA strands (Figure S2). The FRET efficiency is

inversely proportional to the sixth power of the donor-acceptor distance. Consistent

with this, we observed a monotonic decrease of the FRET efficiency with the

increasing size of the TDNs (Figure S2). These findings indicated that the efficiency

of FRET can be used as an efficient nano-ruler to measure the dimensions of DNA

tetrahedra.

Because the lateral distance between the probes is dominated by the TDN scaffolds,

we were able to precisely tune the lateral distance at the nanoscale by varying the size

of the TDNs. To verify this tuning effect, we first quantified the surface density of the

TDN probes immobilised on a Au surface by using a cationic redox marker

(Hexaammineruthenium(III), Ruhex)29. Its saturated amount of charge compensation

is directly related to the number of anionic phosphate residues in the tetrahedron

monolayer. We observed that the surface density of the TDN probes was inversely
proportional to the size of the TDN (Figure 2a). The surface density increased from

(3.72±0.27)×1012 tetrahedra / cm2 to (7.14±0.21)×1012 tetrahedra / cm2 with the

decreasing size of the TDNs. We then estimated that the distances between the DNA

probes were ~5.2 nm, 4.5 nm and 3.7 nm for TDN-17, TDN-13 and TDN-7,

respectively, which coincided well with the theoretical values (5.8 nm, 4.4 nm and 2.4

nm for TDN-17, TDN-13 and TDN-7, respectively) (Figure 2b). In addition, a surface

density of (3.0±0.35)×1013 molecules/cm2, which corresponds to a distance of ~1.8

nm, was observed for single-stranded DNA (ssDNA). We concluded that the probe

distance could be precisely regulated by varying the size of the TDN.

With the ability to precisely regulate the distance between probes, we further

investigated the surface hybridisation capabilities, such as the hybridisation kinetics

and efficiency. The in situ hybridisation investigation revealed that the kinetics of

hybridisation were dependent on the distance between the probes (Figure 3a and 3b).

We observed a relatively slow hybridisation process (~90 min) with a distance of 1.8

nm. With an increase in distance, the hybridisation kinetics improved significantly.

Nearly saturated signals were observed within 20 min with a distance of ~5.2 nm. To

quantify the reaction rate, we calculated the pseudo-first-order reaction rate (k’) using

a well-established method. We discovered that the reaction rate was also

distance-dependent (Figure 3c). Without using the tetrahedron, a more densely packed

monolayer was formed, which reduced the reaction rate (0.01 min-1). With an increase

in the probe distance, the reaction rate increased monotonically. A 10-fold increase in

the reaction rate (0.10 min-1) was observed by using TDN-17 with a probe distance of
~5.2 nm.

The hybridisation efficiency also depends heavily on the distance between probes.

The hybridisation efficiency was as low as ~15.9% with a distance of ~1.8 nm. With

an increase in the distance, the hybridisation efficiency increased monotonically. With

a distance of 5.2 nm, a hybridisation efficiency as high as ~81.5% was achieved

(Figure 3c), representing a more than 5-fold increase in hybridisation efficiency.

Thermodynamic analysis revealed that the hybridisation became more favourable

when the distance between probes was increased (Figure 3d). Clearly, strong lateral

interactions hinder interfacial target hybridisation, whereas the reactivity and

accessibility of probes for hybridisation can be effectively improved by increasing the

probe distance.

After understanding this distance-dependent DNA hybridisation process, we further

explored the biosensing capability with TDN regulation. As the test bed, we integrated

the TDN probes with a well-established DNA sandwich assay. We observed that the

electrochemical signal increased with the increased lateral distance. When target DNA

was detected at 10 nM, an electrochemical current of ~7,800 nA was obtained using

TDN-17 (the lateral distance between probes was ~5.2 nm) (Figure S4, S5). In sharp

contrast, a current of only ~770 nA was obtained when the distance was decreased to

1.8 nm (Figure 4, S4, S5). We then programmed the detection limits of the biosensors

by regulating the distance between the probes via the use of different sized TDNs.

Interestingly, the detection limits shifted in intervals by one order of magnitude. With

increasing size of the TDNs, the detection limits shifted from 10 pM to 100 fM
(Figure 4, Figure S5). With this precise regulation, the detection sensitivity was

improved 100-fold. Without using the TDN scaffolds, the ssDNA-based biosensor

exhibited a detection limit of 10 pM by using appropriate diluent molecules (Figure

S6). By using an interfacial engineering strategy that employs mixed monolayers of

TDN-17 with and without the pedant DNA probes, we could further lower the surface

density. We demonstrated that a DNA sensor with a 1:10 mixed monolayer of TDN-17

and the use of a polymerized enzyme conjugate, poly-HRP40, led to a detection limit

of 100 aM (Figure S7). This attomolar sensitivity is comparable to those of previously

reported nanostructured biosensors 1,3,11,30.

We then investigated the specificity and selectivity of the biosensors. The TDN-based

sensors exhibited greater ability to discriminate single nucleotide mismatched targets

than ssDNA-based one (Figure 5). As demonstrated, the TDN-based sensors clearly

discriminated the signals from mismatched target (G) and mismatched target (A).

Furthermore, the TDN-based sensors were able to selectively recognise targets in

complicated matrices, such as foetal bovine serum (FBS), without an increase in

background signal or a loss in signal gain (Figure 5). The signal differences in FBS

and buffer solution are within ~10 %.

Molecules that function well in solution may not fully retain their properties when

immobilised on a surface2,8,31-33. The density, orientation and entanglement between

probes heavily affect their recognition process. A series of diluent molecules, such as

mercaptohexanol (MCH), oligo(ethylene glycol)-terminated thiols and ternary small

molecules, have been used as “backfillers” to help the DNA probes extend toward the
solution phase21,34. With these molecules, the non-specific adsorption of DNA probes

on the surface can be minimised, and the orientation of the DNA probes is improved.

However, finely tuning the density and minimizing lateral interactions remain

challenging for DNA immobilisation. DNA nanostructures provide a feasible solution

to this problem. Because the dimensions of DNA probes and the target recognition

process occur on the nanometre scale, the use of DNA nanostructures offer

unprecedented advantages for the bottom-up organisation of DNA recognition probes

with nanometre-scale precision.

To precisely control the probe density, we rationally designed TDNs with different

sizes. Through these well-defined TDNs, we were able to regulate the lateral distances

between the DNA probes at the nanometre scale, making it possible to investigate the

DNA hybridisation process with the same level of precision. Another advantage was

that we could precisely anchor a single DNA probe on a single DNA nanostructure,

which is very difficult to achieve with inorganic nanoparticles35,36. The third

advantage was that diluent molecules were not required for immobilisation, which

makes the assembly of monolayers simpler by eliminating the “backfill” step.

The performance of biosensors represents a trade-off: macroscopic biosensors have a

large surface area (high absolute number of probes), but their low mass transport rate

and surface heterogeneity are major limitations for sensitive and rapid sensing.

Nanometre scale sensors have fast binding kinetics; however, their physical

limitations in size will eventually dominate (owing to the very low probability of
target binding events on an individual sensor). Thus, impractically long times are

required for sensitive detection using nanoscale sensors. We present a programmable

“soft lithography” approach using DNA nanostructures for trans-scale surface

engineering. Through this method, we bridge the gap between nanometre scale

sensors and macroscopic biosensors by combining the inherent advantages of the two

approaches. Finally, this method allows the overall performance of biosensors to be

improved. For example, the detection speed can be decreased from hours to minutes,

and the recognition efficiency can be enhanced from ~15% to greater than 80%. The

detection limits can also be programmed and improved. Also importantly, the

biosensors based on our trans-scale surface engineering performed well in complex

matrices, revealing high applicability in real-world settings.

Methods

Equimolar quantities of four strands for the formation of the tetrahedrons were mixed

in buffer (20 mM Tris, 50 mM MgCl2, pH 8.0) at 95 °C and then cooled to 4 °C.

A 3 µL volume of 1 µM tetrahedron probes in I-buffer (containing 3 mM TCEP) was

added to the surface of the cleaned gold electrodes and allowed to immobilise

overnight at room temperature. The immobilisation for ssDNA was the same as that

for the tetrahedrons, and the concentration was also 1 µM. However, for the

immobilisation of ssDNA, the modified electrodes were then exposed to a 2 mM

MCH solution (in H-buffer) at room temperature for 1 hour to replace non-specific

interactions and form a self-assembled monolayer (SAM) that resists non-specific

adsorption of target DNA. Between each step, the electrode was rinsed with R-buffer
and dried lightly with N2 before hybridisation.

DNA detection was carried out in a sandwich assay format. Different concentrations

of target DNA were first mixed with the biotinylated reporter probe (100 nM) in 100

µL of H-buffer, heated to 80 °C for 5 min and then cooled under refrigeration for 5

min. ssDNA probe modified electrodes were exposed to a 2 mM MCH solution (in

H-buffer) at room temperature for 1 hour. Finally, the tetrahedron or ssDNA probe

modified electrode was incubated in the 100 µL solution for hybridisation at 37 °C for

2 hours. The electrodes were rinsed with 0.1 M PBS buffer and then incubated with 3

µL of avidin-HRP (0.5 U/mL) for 15 min at room temperature. The electrodes were

then extensively rinsed with R-buffer and subjected to electrochemical measurements.

Acknowledgements

This work was financially supported by the National Basic Research Program (973

Program 2012CB932600), the Shanghai Pujiang Project (13PJ1410700) and the

Chinese Academy of Sciences.

Author contributions

X. Z. and C. F. conceived and designed the experiments. M. L., J. W. and G. Z.

performed the experiments. J. L., J. G., X. C.and J. S. analysed the data. M. L., X. Z.,

and C.F. co-wrote the paper. All authors discussed the results and commented on the

manuscript.

Additional information

Supplementary information is available in the online version of the paper. Reprints

and permissions information is available online at www.nature.com/reprints.

Correspondence and requests for materials should be addressed to X. Z. and C. F.

Competing financial interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to: Xiaolei Zuo or Chunhai Fan

Figure legends

Figure 1. We developed a precisely programmable “soft lithography” method via 3D

DNA tetrahedral nanostructures (TDNs) for biosensing surface engineering. Three

types of TDNs with different sizes were designed. The lateral distance between the

DNA probes on the surface was precisely regulated through the different sized TDNs.

The surface hybridisation regimes (hybridisation kinetics and thermodynamics) were

investigated, and the detection limits were precisely programmed.

Figure 2. (a) The densities of the TDN probes decreased with increasing TDN size. (b)

The nanodistance between the probes was well controlled by tuning the size of the

TDN. The experimental data correlated highly with the theoretical data.

Figure 3. (a) Hybridisation kinetics are highly related to the nanodistance between

probes. (b) The pseudo-first-order reaction rates are shown. A ten-fold improvement

was obtained by tuning the nanodistance via the use of different sized TDNs. (c)
Hybridisation efficiency varies with the size of TDNs. The hybridisation efficiency

can be improved from 15.9% to 81.5% by tuning the nanodistance. (d) The free

energy of surface hybridisation decreased with increasing nanodistance between the

probes, indicating that the hybridisation process became more favourable.

Figure 4. Detection limits can be programmed by tuning the size of the TDNs. The

detection limits were 100 fM, 1 pM and 10 pM for TDN-17 (a), TDN-13 (b) and

TDN-7 (c), respectively.

Figure 5. (a)The mismatched targets exhibit relatively weak electrochemical signals,

indicating that our sensors can sensitively distinguish single nucleotide

polymorphisms. (b) The sensors functioned well when challenged with 50% FBS. The

background signals and the signals from the 10 nM targets were comparable to those

obtained in pure buffer solution.

References

1 Soleymani, L., Fang, Z. C., Sargent, E. H. & Kelley, S. O. Programming the

detection limits of biosensors through controlled nanostructuring. Nat.

Nanotechnol. 4, 844-848 (2009).

2 Squires, T. M., Messinger, R. J. & Manalis, S. R. Making it stick: convection,

reaction and diffusion in surface-based biosensors. Nat. Biotechnol. 26,

 417-426 (2008).

3 Lam, B. et al. Solution-based circuits enable rapid and multiplexed pathogen

detection. Nat. Commun. 4 (2013).

4 Bracha, D., Karzbrun, E., Shemer, G., Pincus, P. A. & Bar-Ziv, R. H.

Entropy-driven collective interactions in DNA brushes on a biochip. P. Natl.

Acad. Sci. U.S.A. 110, 4534-4538 (2013).

5 Gong, P. & Levicky, R. DNA surface hybridization regimes. P. Natl. Acad. Sci.

U.S.A. 105, 5301-5306 (2008).

6 Guan, J. J. & Lee, J. Generating highly ordered DNA nanostrand arrays. P.

Natl. Acad. Sci. U.S.A. 102, 18321-18325 (2005).

7 Opdahl, A., Petrovykh, D. Y., Kimura-Suda, H., Tarlov, M. J. & Whitman, L. J.

Independent control of grafting density and conformation of single-stranded

DNA brushes. P. Natl. Acad. Sci. U.S.A. 104, 9-14 (2007).

8 Pei, H. et al. Designed Diblock Oligonucleotide for the Synthesis of Spatially

Isolated and Highly Hybridizable Functionalization of DNA-Gold

Nanoparticle Nanoconjugates. J. Am. Chem. Soc. 134, 11876-11879 (2012).

9 Pei, H. et al. A DNA Nanostructure-based Biomolecular Probe Carrier

Platform for Electrochemical Biosensing. Adv. Mater. 22, 4754-+ (2010).

10 Sheehan, P. E. & Whitman, L. J. Detection limits for nanoscale biosensors.

Nano. Lett. 5, 803-807 (2005).

11 Soleymani, L. et al. Hierarchical Nanotextured Microelectrodes Overcome the

Molecular Transport Barrier To Achieve Rapid, Direct Bacterial Detection.

 Acs. Nano. 5, 3360-3366 (2011).

12 Pal, S., Deng, Z. T., Ding, B. Q., Yan, H. & Liu, Y. DNA-Origami-Directed

Self-Assembly of Discrete Silver-Nanoparticle Architectures. Angew. Chem.

Int. Edit. 49, 2700-2704 (2010).

13 Li, Z. W. et al. Hybrid Nanoimprint-Soft Lithography with Sub-15 nm

Resolution. Nano. Lett. 9, 2306-2310 (2009).

14 Jin, Z. et al. Metallized DNA nanolithography for encoding and transferring

spatial information for graphene patterning. Nat. Commun. 4 (2013).

15 Surwade, S. P., Zhao, S. C. & Liu, H. T. Molecular Lithography through

DNA-Mediated Etching and Masking of SiO2. J. Am. Chem. Soc. 133,

11868-11871 (2011).

16 Wirth, J. et al. Plasmonic Nanofabrication by Long-Range Excitation Transfer

via DNA Nanowire. Nano. Lett. 11, 1505-1511 (2011).

17 Chin, C. D. et al. Microfluidics-based diagnostics of infectious diseases in the

developing world. Nat. Med. 17, 1015-U1138 (2011).

18 Xiang, Y. & Lu, Y. Using personal glucose meters and functional DNA sensors

to quantify a variety of analytical targets. Nat. Chem. 3, 697-703 (2011).

19 Ferguson, B. S. et al. Genetic analysis of H1N1 influenza virus from throat

swab samples in a microfluidic system for point-of-care diagnostics. J. Am.

Chem. Soc. 133, 9129-9135, doi:10.1021/ja203981w (2011).

20 Vallee-Belisle, A., Ricci, F. & Plaxco, K. W. Thermodynamic basis for the

optimization of binding-induced biomolecular switches and

 structure-switching biosensors. P. Natl. Acad. Sci. U.S.A. 106, 13802-13807

(2009).

21 Wu, J., Campuzano, S., Halford, C., Haake, D. A. & Wang, J. Ternary surface

monolayers for ultrasensitive (zeptomole) amperometric detection of nucleic

acid hybridization without signal amplification. Anal. Chem. 82, 8830-8837,

doi:10.1021/ac101474k (2010).

22 Qin, D., Xia, Y. N. & Whitesides, G. M. Soft lithography for micro- and

nanoscale patterning. Nat. Protoc. 5, 491-502 (2010).

23 Goodman, R. P. et al. Rapid chiral assembly of rigid DNA building blocks for

molecular nanofabrication. Science 310, 1661-1665 (2005).

24 Surwade, S. P. et al. Nanoscale Growth and Patterning of Inorganic Oxides

Using DNA Nanostructure Templates. J. Am. Chem. Soc. 135, 6778-6781

(2013).

25 Fu, J. L., Liu, M. H., Liu, Y., Woodbury, N. W. & Yan, H. Interenzyme

Substrate Diffusion for an Enzyme Cascade Organized on Spatially

Addressable DNA Nanostructures. J. Am. Chem. Soc. 134, 5516-5519 (2012).

26 Zheng, J. P. et al. From molecular to macroscopic via the rational design of a

self-assembled 3D DNA crystal. Nature 461, 74-77 (2009).

27 Han, D. R. et al. DNA Origami with Complex Curvatures in

Three-Dimensional Space. Science 332, 342-346 (2011).

28 Wilner, O. I. et al. Enzyme cascades activated on topologically programmed

DNA scaffolds. Nat. Nanotechnol. 4, 249-254, doi:10.1038/nnano.2009.50

 (2009).

29 Steel, A. B., Herne, T. M. & Tarlov, M. J. Electrochemical quantitation of

DNA immobilized on gold. Anal. Chem. 70, 4670-4677 (1998).

30 Hahm, J. & Lieber, C. M. Direct ultrasensitive electrical detection of DNA and

DNA sequence variations using nanowire nanosensors. Nano. Lett. 4, 51-54

(2004).

31 Bin, X. M., Sargent, E. H. & Kelley, S. O. Nanostructuring of Sensors

Determines the Efficiency of Biomolecular Capture. Anal. Chem. 82,

5928-5931 (2010).

32 Pauloehrl, T. et al. Spatially Controlled Surface Immobilization of

Nonmodified Peptides. Angew. Chem. Int. Edit. 52, 9714-9718 (2013).

33 Abel, G. R., Josephs, E. A., Luong, N. & Ye, T. A Switchable Surface Enables

Visualization of Single DNA Hybridization Events with Atomic Force

Microscopy. J. Am. Chem. Soc. 135, 6399-6402 (2013).

34 Lao, R. J. et al. Electrochemical interrogation of DNA monolayers on gold

surfaces. Anal. Chem. 77, 6475-6480 (2005).

35 Zanchet, D., Micheel, C. M., Parak, W. J., Gerion, D. & Alivisatos, A. P.

Electrophoretic isolation of discrete Au nanocrystal/DNA conjugates. Nano.

Lett. 1, 32-35 (2001).

36 Jhaveri, S. D. et al. Isolation and characterization of

trioxyethylene-encapsulated gold nanoclusters functionalized with a single

DNA strand. Nano. Lett. 4, 737-740 (2004).