pubs.acs.org/Langmuir © 2009 American Chemical Society
Preparation, Characterization and Scanned Conductance Microscopy Studies of DNA-Templated One-Dimensional Copper Nanostructures Scott M. D. Watson,† Nicholas G. Wright,‡ Benjamin R. Horrocks,† and Andrew Houlton*,† †
Chemical Nanoscience Laboratories, School of Chemistry, Bedson Building, Newcastle University, Newcastle upon Tyne, NE1 7RU. U.K., and ‡School of Electrical, Electronic and Computing Engineering, Newcastle University, Newcastle Upon Tyne, NE1 7RU, U.K. Received July 15, 2009
The synthesis of one-dimensional metal nanostructures can be achieved through the use of DNA molecules as templates to control and direct metal deposition. Copper nanostructures have been fabricated using this strategy, through association of Cu2þ ions to DNA templates and reduced with ascorbic acid. Due to the possibility that the reduction of the Cu2þ can result in the preferential formation of Cu2O over metallic Cu0, X-ray photoelectron spectroscopy and X-ray diffraction have been carried out to establish the chemical identity of the nanostructures. Conclusive evidence is found that reduction of the Cu2þ ions does result in the formation of the desired metallic Cu0 structures. The morphology of the nanostructured Cu0 material has also been observed by atomic force microscopy, showing the structures to have a “beads-on-a-string” appearance and being 3.0-5.5 nm in height. The electrical properties of the structures have been investigated by scanned conductance microscopy, showing the Cu0 structures exhibit much larger electrical resistance than expected for a metallic nanowire. This is thought to be a consequence of their “beads-on-a-string” morphology and small lateral dimensions (sub-10 nm); both these factors would be expected to increase the electron scattering rate, and, further, there are likely to be significant tunneling barriers at the Cu0 particle-particle junctions.
Introduction Metallic interconnects in integrated circuits (IC) play several important roles in device operation including distributing clock signals, and providing power to the various circuits on a microprocessor. Current IC technology uses copper interconnects due to the materials low electrical resistivity (F = 1.7 10-8-1.9 10-8 Ωm) and resistance to electromigration.1,2 The fabrication of copper interconnects in microprocessor technology currently employs a three step approach known as the “Damascene” process, involving: (i) initial dry etching of narrow trenches in a silicon substrate, (ii) electroplating of the copper within the etched trenches, and (iii) chemical mechanical polishing to remove excess copper and planarize the substrate surface.3,4 Today, 45 nm node features is the current state-of-theart process technology used in high-volume microprocessor production.5,6 However, there remains a great motivation for further miniaturization of semiconductor technologies, driven by the continuing demands for improved device performance, e.g., processing speed and power consumption. The fabrication of *To whom correspondence should be addressed. E-mail: andrew.houlton@ ncl.ac.uk. Phone: þ44 (0)191 222 6262. Fax: þ44 (0)191 222 6929. (1) Lane, M.; Dauskardt, R. H.; Krishna, N.; Hashim, I. J. Mater. Res. 2000, 15, 203. (2) Strehle, S.; Bartha, J. W.; Wetzig, K. Thin Solid Films 2009, 517, 3320. (3) Edelstein, D.; Heidrenreich, J.; Goldblatt, R.; Cote, W.; Uzoh, C.; Lustig, N.; Roper, P.; McDevitt, T.; Motsiff, W.; Simon, A.; Dukovic, J.; Wachnik, R.; Rathore, H.; Schulz, R.; Su, L.; Luce, S.; Slattery, J. Full copper wiring in a sub250 μm CMOS ULSI Technology. In Technical Digest; IEEE International Electron Devices Meetings: New York, 1997. (4) Andricacos, P. C.; Uzoh, C.; Dukovic, J. O.; Horkans, J.; Deligianni, H. IBM J. Res. Dev. 1998, 42, 567. (5) Nisar, A.; Ekpanyapong, M.; Valles, A. C.; Sivakumar, K. Intel Technol. J. 2008, 12, 157. (6) Intel. Introducing the 45 nm next-generation Intel(R) Core(TM) Microarchitecture. In Intel White Paper, 2007; http://www.intel.com/technology/architecture-silicon/intel64/45 nm-core2_whitepaper.pdf (accessed May 2009).
2068 DOI: 10.1021/la902583j
interconnects beyond the current resolution limits using conventional “top-down” approaches has numerous drawbacks such as the high fabrication costs associated with the use of established methods for making sub-100 nm structures (e.g., electron beam lithography, extreme ultraviolet lithography),7,8 as well as the procedurally challenging task of depositing metal within the etched trenches in the silicon substrate, without voids resulting in the metal structure. In light of these problems, much attention is now being paid to alternative approaches to the construction of nanoscale semiconductor devices. “Bottom-up” strategies in particular, which focus upon the fabrication of individual nanostructured “building blocks” and their controlled assembly into organized structures, may offer a feasible route to the development of future generations of IC devices. Much work has already been carried out into the use of DNA as a template to direct the formation and placement of conducting wires upon a substrate surface.9-12 Although the syntheses of a range of metallic (7) Xia, Y.; Rogers, J. A.; Paul, K. E.; Whitesides, G. M. Chem. Rev. 1999, 99, 1823. (8) Ito, T.; Okazaki, S. Nature 2000, 406, 1027. (9) Dong, L.; Hollis, T.; Fishwick, S.; Connolly, B. A.; Wright, N. G.; Horrocks, B. J.; Houlton, A. Chem.;Eur. J. 2007, 13, 822. (10) Pruneanu, S.; Al-Said, S. A. F.; Dong, L.; Hollis, T. A.; Galindo, M. A.; Wright, N. G.; Houlton, A.; Horrocks, B. R. Adv. Funct. Mater. 2008, 18, 2444. (11) Al-Said, S. A. F.; Hassanien, R.; Hannant, J.; Galindo, M. A.; Pruneanu, S.; Pike, A. R.; Houlton, A.; Horrocks, B. R. Electrochem. Commun. 2009, 11, 550. (12) Houlton, A.; Pike, A. R.; Galindo, M. A.; Horrocks, B. R. Chem. Commun. 2009, 1797. (13) Park, S. H.; Prior, M. W.; LaBean, T. H.; Finkelstein, G. Appl. Phys. Lett. 2006, 89, 033901. (14) Keren, K.; Berman, R. S.; Braun, E. Nano Lett. 2004, 4, 323. (15) Braun, E.; Eichen, Y.; Sivan, U.; Ben-Yoseph, G. Nature 1998, 391, 775. (16) Keren, K.; Krueger, M.; Gilad, R.; Ben-Yoseph, G.; Sivan, U.; Braun, E. Science 2002, 297, 72. (17) Harnack, O.; Ford, W. E.; Yasuda, A.; Wessels, J. M. Nano Lett. 2002, 2, 919. (18) Mertig, M.; Ciacchi, L. C.; Seidel, R.; Pompe, W. Nano Lett. 2002, 2, 841.
Published on Web 09/16/2009
Langmuir 2010, 26(3), 2068–2075
Watson et al.
nanowires (e.g., Ag,13-15 Au,16,17 Pt,18,19 and Pd20,21) have previously been described, surprisingly few examples of DNAtemplated copper nanowires have been reported to date. Kudo et al.22 described the preparation of copper nanowires based upon traditional electroless copper deposition methods. This involved a two stage process in which DNA was initially “activated” through binding of Pd2þ to the DNA molecules, followed by reduction to Pd0. In the second stage, the Pd0 acts as a seed for electroless copper plating, by reduction of Cu2þ ions with formaldehyde. One example of DNA-templated copper nanowires prepared without the need for a second metal as a seeding surface has been reported by Woolley.23,24 Here, λ-DNA is treated in aqueous solutions with Cu(NO3)2 resulting in electrostatic association of the Cu2þ ions with the negatively charged phosphate backbone of the DNA molecules. Subsequent chemical reduction of the Cu2þ was used to generate DNA-templated metallic Cu0 structures, ca. 3 nm tall. It has since been reported that treatment of DNA with Cu(NO3)2 solutions prepared in dimethylsulfoxide (DMSO) can lead to more substantial Cu0 deposition upon the DNA molecules, attributed to the lower dielectric constant of the DMSO (ε = 46.7) in comparison to water (ε = 78.39), enhancing the Cu2þ/DNA interactions.25 To date, evidence for the formation of these one-dimensional copper structures has been largely limited to atomic force microscopy (AFM) studies which focus upon changes in the morphology of the DNA molecules following Cu2þ/ascorbic acid treatment, and provide no insight into their chemical composition. The need for such chemical characterization is significant due to the potential formation of Cu2O as an alternative product to metallic Cu0 upon the reduction of Cu2þ. In addition to this, investigating the electrical properties of these copper structures is also of fundamental importance toward their development and application as metal interconnects in future generations of nanoelectronic devices: the electrical resistance of the interconnects can be directly related to a parasitic contribution to the signal delay in integrated circuits. Chemical characterization reported to date has been limited to energy dispersive X-ray (EDX) spectroscopy. These studies, carried out upon two-dimensional DNA motifs having undergone similar Cu2þ treatments, provided elemental evidence of the presence of copper bound to the DNA structures.26 No studies to the best of our knowledge have so far has been reported describing the electrical behavior of such copper structures. In this paper, using X-ray photoelectron spectroscopy (XPS) and X-ray diffraction (XRD) studies, we provide significant evidence for the chemical reduction of Cu2þ ions upon DNA templates resulting in the formation of one-dimensional metallic copper nanostructures. We also show by scanned conductance microscopy (SCM) that despite their metallic composition, the copper structures exhibit high levels of electrical resistance, attributed to sub-10 nm line widths of the structures and their “beads-on-a-string” morphology resulting in an increase in the effective resistivity of the copper material. (19) Ford, W. E.; Harnack, O.; Yasuda, A.; Wessels, J. M. Adv. Mater. 2001, 13, 1793. (20) Richter, J.; Seidel, R.; Kirsch, R.; Mertig, M.; Pompe, W.; Plaschke, J.; Schackert, H. K. Adv. Mater. 2000, 12, 507. (21) Richter, J.; Mertig, M.; Pompe, W.; Monch, I.; Schackert, H. K. Appl. Phys. Lett. 2001, 78, 536. (22) Kudo, H.; Fujihira, M. IEEE Trans. Nanotechnol. 2006, 5, 90. (23) Monson, C. F.; Woolley, A. T. Nano Lett. 2003, 3, 359. (24) Becerril, H. A.; Stoltenberg, R. M.; Monson, C. F.; Woolley, A. T. J. Mater. Chem. 2004, 14, 611. (25) Stoltenberg, R. M.; Woolley, A. T. Biomed. Microdevices 2004, 6, 105. (26) Becerril, H. A.; Stoltenberg, R. M.; Wheeler, D. R.; Davis, R. C.; Harb, J. N.; Woolley, A. T. J. Am. Chem. Soc. 2005, 127, 2828.
Langmuir 2010, 26(3), 2068–2075
Article
Experimental Section Materials. All general chemical reagents were obtained from Sigma-Aldrich unless otherwise stated, and were of Analar grade or equivalent. Lambda (λ) DNA was from New England Biolabs, cat no. N3011S (New England Biolabs (UK) Ltd. Hitchin, Herts, United Kingdom). Calf thymus and herring testes DNA were purchased from Sigma. Silicon wafers, 3 in. diameter, 525 ( 50 μm thickness, 1-10 Ω cm resistance were purchased from Compart Technology Ltd. (Peterborough, Cambridgeshire, United Kingdom). Alignment of λ-DNA upon Substrates. Silicon wafers were cut into ca. 1 1 cm2 pieces with a diamond tip pen. The wafers were chemically oxidized through treatment in “piranha” solution (4:1 H2SO4/H2O2) for 45 min (Caution! Piranha solution should be handled with extreme care; it is a strong oxidant and reacts violently with many organic materials. It also presents an explosion danger), followed by rinsing in Nanopure water and dried in a clean oven. The oxidized Si wafer surface was modified through vapor deposition of a self-assembled monolayer of trimethylsilane (TMS): 100 μL of chlorotrimethylsilane (Me3SiCl) was placed in the bottom of a specimen bottle and the substrate placed on top of the specimen bottle (polished side facing up). The specimen bottle was placed in a larger, sealed specimen bottle and the substrate treated in the silane vapor for 5 min. Static contact angle measurements of the TMS-modified substrate were carried out using a CAM100 system (KSV Instruments Ltd., Helsinki, Finland), with Nanopure water as the probe liquid. Silicon nþþ wafers used for SCM experiments were degreased by boiling in a series of solvents (trichloroethylene, acetone, propanol) and dipped in 10% aqueous HF to remove the native oxide. Substrates were then placed in a furnace at 1000 °C with an oxygen flow of 80 mL min-1 for 5 h to produce an oxide layer of ca. 220 nm thickness as determined by spectrometric thin film analysis (Filmetrics F40). λ-DNA solution (3 μL; 500 μg mL-1, in 10 mM tris-HCl, pH 8, 1 mM EDTA) was diluted down to 300 μg mL-1 with Nanopure water immediately before use. Four 5 μL of the DNA solution was placed upon the TMS-modified substrate and left for 10 s before being dragged across the surface using the tip of the micropipet. “Molecular combing” of the DNA in this fashion was repeated several times to align the DNA across the substrate surface. Copper Metalization of Aligned λ-DNA. The surface aligned DNA was typically treated with 80-100 μL of 0.5 M Cu(NO3)2 (prepared in DMSO), for 20 min. After 20 min had elapsed, 80-100 μL of 1.0 M ascorbic acid (prepared in Nanopure water) was added to the Cu(NO3)2 solution and allowed to stand for a further 4 min, before the Cu(NO3)2/ascorbic acid solution was removed from the substrate surface and the substrate briefly washed in Nanopure water. Atomic Force and Electrostatic Force Microscopy. TappingMode AFM imaging of surface topography was performed in air on both a Multimode Nanoscope IIIa and Dimension Nanoscope V (Veeco Instruments Inc., Metrology Group, Santa Barbara, CA) using TESP7 probes (n-doped Si cantilevers, Veeco Instruments Inc., Metrology Group), with a resonant frequency of 234-287 kHz, and a spring constant of 20-80 Nm-1. Data acquisition was carried out using Nanoscope version 5.12b36 (Multimode IIIA) and Nanoscope version 7.00b19 (Dimension Nanoscope V) software (Veeco Instruments Inc., Digital Instruments) EFM measurements were carried out in air on a Dimension Nanoscope V system using MESP probes (n-doped Si cantilevers, with a metallic Co/Cr coating, Veeco Instruments Inc., Metrology Group), with a resonant frequency of ca. 70 kHz, a quality factor of 200-260, and a spring constant of 1-5 Nm-1. Data acquisition was carried out using Nanoscope version 7.00b19 software. For both AFM systems, vibrational noise was reduced with an isolation table/acoustic enclosure (Veeco Inc., Metrology Group). DOI: 10.1021/la902583j
2069
Article
Watson et al.
In EFM experiments, an electrostatic field was created between the tip and sample by applying an independently controlled bias to the sample, while the tip was grounded. EFM has also been referred to as scanned conductance microscopy (SCM), as the phase angle is related to the force gradient and is sensitive to the conductance of samples under scrutiny, as well as their polarizability. The reported EFM phase images show the phase of the tip oscillation at a set lift height above the sample surface (typically 40-50 nm). Samples used in EFM studies were prepared upon Si nþþ substrates with a thermally grown oxide layer, 220 nm thick on top, prepared as described earlier. Processing of data acquired from AFM and EFM experiments was carried out using both Nanoscope version 7.00b19 (Veeco Inc., Digital Instruments) and WSxM 4.0 Develop 12.6 (Nanotec Electronica S. L., Madrid, Spain) software.49 Fourier Transform Infra-Red Spectroscopy. FTIR spectra (in the range 600-4000 cm-1) were recorded in transmission mode with a Bio-Rad Excalibur FTS-40 spectrometer (Varian Inc., Palo Alto, CA) equipped with a liquid nitrogen cooled deuterated triglycine sulfate (DTGS) detector, and were collected at 128 scans with 4 cm-1 resolution. The DNA used was herring testes and all samples were prepared through drop casting of solutions of the DNA upon chemically oxidized Si substrates. Data acquisition and analysis were carried out using Digilab Merlin version 3.1 software (Varian Inc.). X-ray Photoelectron Spectroscopy. XPS was carried out upon samples of copper metalized λ-DNA supported upon chemically oxidized Si substrates. The λ-DNA was aligned upon the substrate through depositing 5 mL of the stock DNA solution upon the substrate and blowing it across the surface in a stream of N2. Copper metalization of the λ-DNA was achieved using solutions of 0.5 M Cu(NO3)2 (DMSO solvent) and 1.0 M ascorbic acid (aqueous solution) as described previously. XPS was carried out using a Kratos Axis Ultra 165 (Kratos Analytical Ltd., Manchester, UK) with an Al KR X-ray source (1486.6 eV) at a take off angle of 90°. The substrates were mounted on copper sample stubs with conductive carbon tape. A charge neutralizer gun was applied to reduce charging upon the substrate surface, and all peaks were referenced to the hydrocarbon C (1s) peak at 285.0 eV binding energy. Data acquisition was carried out using Kratos Vision software (Kratos Analytical Ltd.). Powder X-ray Diffraction. Powder samples were prepared through addition of aqueous solutions of 0.5 M Cu(NO3)2 and 1.0 M ascorbic acid to a 500 μg mL-1 solution of calf thymus DNA in a 1:1:1 (v/v) ratio, and left overnight. The resulting powder was washed three times with Nanopure water, followed by three ethanol washes, and allowed to dry in air at room temperature. Powder X-ray diffraction (XRD) data was obtained using a PANalytical X’Pert Pro diffractometer equipped with a Cu KR1 radiation source (λ = 1.540 A˚).
Results and Discussion Preparation of Copper Nanostructures. Copper metalization of the DNA templates was typically carried out following immobilization of the DNA upon a substrate support using wellestablished “molecular combing” techniques.9 The density of the DNA deposited upon the substrate surface during the combing procedure was controlled through tailoring the hydrophobic/ hydrophilic surface properties of the substrate. Combing of the DNA upon hydrophilic Si/SiO2 substrates (static contact angles = 19.3°, standard deviation = 3.3°) was employed to deposit dense networks of DNA for the fabrication of copper nanostructures with high surface coverage across the substrate (see Supporting Information (SI)). Spectroscopic characterization (FTIR, XPS) was carried out upon such networks of the copper nanostructures due to the sensitivity and resolution limits associated with these analysis techniques. 2070 DOI: 10.1021/la902583j
DNA immobilization was also carried out upon Si/SiO2 substrates modified with a trimethylsilane (TMS) self-assembled monolayer. The increased hydrophobic character of the TMSmodified substrate surface (static contact angles = 76.3°, standard deviation = 2.1°) favors the alignment of individual molecules of DNA during the molecular combing process, while limiting the occurrence of “ropes” or “networks” of DNA on the surface. The alignment of individual DNA molecules allows for the fabrication of individual copper nanostructures across the substrate surface. The ability to prepare the nanostructured material in this manner enables the copper structures to be studied independent of each other by scanning probe microscopy techniques, e.g., AFM, SCM. Metalization of the DNA was carried out using a two step procedure in which the immobilized DNA was first treated with a Cu(NO3)2 solution (0.5 M) prepared in DMSO to “dope” the DNA molecules with Cu2þ ions through metal/DNA complex formation. An equal volume of aqueous ascorbic acid solution (1.0 M) was then added to the Cu(NO3)2 solution on the substrate support. This acts to reduce the Cu2þ ions bound to the DNA molecules, forming one-dimensional metallic Cu0 architectures immobilized on the substrate. Cu0/DNA samples prepared for powder XRD studies were made up in aqueous solutions of calf thymus DNA (ca. 500 μg mL-1), through addition of equal volumes of Cu(NO3)2 (0.5 M) and ascorbic acid (1.0 M) aqueous solutions. Solution preparations of this nature were required because the amount of material obtained via metalization of DNA molecules immobilized upon substrate supports was insufficient to obtain useful XRD data. Cu2þ/DNA Complexation. The interaction of Cu2þ ions with DNA molecules has been extensively studied in the past and the nature of these interactions was found to be dependent upon the metal ion concentration.27-30 At very low metal concentrations, Cu2þ binding occurs through electrostatic interactions with the negatively charged phosphate backbone of the DNA molecules. At increased metal concentrations however, the Cu2þ ions favor direct association with the nucleobases of the DNA structure. Several models have been proposed to describe the exact nature of the Cu2þ/DNA binding, including coordination of the Cu2þ ions with the N-7 atom of guanine and a phosphate oxygen from the DNA backbone binding in a “sandwich complex”, between two adjacent guanines of the same DNA strand, or through chelation to both a guanine (N-7, O-6) and cytosine (N-3, O-2) of different strands.28 In the current studies, FTIR spectra recorded of DNA alone and DNA following exposure to a Cu(NO3)2 solution (but without the subsequent chemical reduction step) provide supporting evidence that Cu2þ coordination to both the phosphate backbone and nucleobases in the DNA structure takes place, (Figure 1 and SI). The spectrum of the DNA alone shows a broad, partially split band arising from PO2symmetric stretches (1096 cm-1), and P-O or C-O stretches (1074 cm-1) of the phosphate backbone. A shoulder on this band is also observed around 1024 cm-1 due to C-C stretches of the deoxyribose rings in the DNA structure. Upon Cu2þ/DNA complexation the intensity of the PO2- symmetric stretch is significantly reduced, whereas the P-O/C-O stretching in the phosphate backbone is shifted to lower frequency (1063 cm-1). Increased splitting of the P-O/C-O backbone and the deoxyribose C-C (27) Nucleic Acid - Metal Ion Interactions; Spiro, T., Ed.; John Wiley & Sons: New York, 1980; Vol. 1. (28) Andrushchenko, V.; Sande, J. H. V. d.; Wieser, H. Biopolymers 2003, 72, 374. (29) Tajmir-Riahi, H. A.; Naoui, M.; Ahmad, R. Biopolymers 1993, 33, 1819. (30) Zimmer, C.; Luck, G.; Fritzsche, H.; Triebel, H. Biopolymers 1971, 10, 441.
Langmuir 2010, 26(3), 2068–2075
Watson et al.
Article
Figure 1. FTIR spectra of DNA vs the Cu2þ/DNA complex in the 600 - 2000 cm-1 region.
stretches is also apparent. In addition to this the asymmetric PO2vibration (1248 cm-1) is also observed to shift to lower frequency (1242 cm-1). Direct interaction of the Cu2þ ions with the nucleobases is evident from several notable band shifts correlating to nucleobase vibrations: the in-plane cytosine/guanine vibration (1530 cm-1) is significantly reduced in intensity upon Cu2þ complexation and shifted to higher frequency (1535 cm-1), whereas the guanine ring vibration (1485 cm-1) is observed to shift to lower frequency (1477 cm-1). Changes in band positions and intensities are also apparent for the nucleobase vibrations in the 1570-1700 cm-1 regions of the spectra. X-ray Photoelectron Spectroscopy. Due to the possibility of Cu2O formation upon reduction of the DNA-templated Cu2þ ions (as an alternative product to metallic Cu0), it is important to establish the chemical composition of the nanostructures produced when using this DNA-templating method. XPS of the Cu2þ/ascorbic acid treated DNA detected the Cu 2p1/2 and 2p3/2 binding energies at 952.7 and 932.9 eV, respectively (Figure 2a); these fall within the expected range for metallic Cu0.31-33 Fitting of the Cu (2p) peaks to pseudo-Voigt functions shows that both peaks comprise of single components, with a separation of 19.8 eV. This is again in good agreement with the established values for the 2p1/2-2p3/2 splitting in metallic Cu0.31,34 The absence of Cu 2p3/2 satellite peaks around 942 eV confirms there are no Cu2þ species present in the nanostructured material.35,36 Inspection of the O (1s) spectrum shows a symmetrical band at 532.7 eV (fwhm = 1.68 eV), arising from the surface oxide layer of the Si/SiO2 substrate, Figure 2b.31,37,38 The absence of an oxygen (31) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Moulder, J. F. Handbook of X-Ray Photoelectron Spectroscopy; Perkin-Elmer: Ramsey, MN, 1979. (32) Chen, T.-Y.; Chen, S.-F.; Sheu, H.-S.; Yeh, C.-S. J. Phys. Chem. B. 2002, 106, 9717. (33) Capece, F. M.; Castro, V. D.; Furlani, C.; Mattogno, G.; Fragale, C.; Gargano, M.; Rossi, M. J. Electron. Spectrosc. Relat. Phenom. 1982, 27, 119. (34) Fuggle, J. C.; Kallne, E.; Watson, L. M.; Fabian, D. J. Phys. Rev. B. 1977, 16, 750. (35) Wang, L.; Wei, G.; Qi, B.; Zhou, H.; Liu, Z.; Song, Y.; Yang, X.; Li, Z. Appl. Surf. Sci. 2006, 252, 2711. (36) Partain, L. D.; Schneider, R. A.; Donaghey, L. F.; McLeod, P. S. J. Appl. Phys. 1985, 57, 5056. (37) Miller, M. L.; Linton, R. W. Anal. Chem., 57, 2314. (38) Netterfield, R. P.; Martin, P. J.; Pacey, C. G.; Sainty, W. G.; McKenzie, D. R.; Auchterionie, G. J. Appl. Phys. 1989, 66, 1805.
Langmuir 2010, 26(3), 2068–2075
Figure 2. (a) Copper (2p1/2 and 2p3/2) and (b) oxygen (1s) XPS spectra of λ-DNA following treatment in solutions of Cu(NO3)2 and ascorbic acid.
component around 530.1 eV indicates the nanostructures to be free of metal oxide O2- species,31,35 providing further support that the Cu (2p) signal arises from metallic Cu0 and not Cuþ species (which would be expected to be present if the Cu2þ reduction resulted in Cu2O formation). Both P (2p) and N (1s) signals are also observed around 134.1 and 400.1 eV, respectively, due to the DNA templates of the Cu0 nanostructures (see SI). X-ray Diffraction. In addition to our XPS findings, further evidence for the formation of metallic Cu0 is provided by XRD data. Figure 3a shows the XRD pattern of a powder sample prepared by the reduction of Cu2þ ions with ascorbic acid, in the presence of calf thymus DNA. The peak positions observed in the XRD pattern are consistent with those of metallic Cu0 (JPDS No. 00-004-0836) and can be indexed to the (111), (200), (220), (311), and (222) reflections. The small peak observed at 2θ = 36.4° is attributed to the Cu2O (111) reflection (JPDS No. 00-005-0667), indicating trace amounts of Cu2O in the powder sample, possibly as a result of a thin oxide coating forming upon the surface of the Cu0 crystallites in the powder. The average crystallite size of the Cu0 powder determined by Scherrers’ equation39 (using the fwhm (0.22°) of the Cu (111) peak following fitting to a pseudo-Voigt function, Figure 3b, was found to be 100 nm. Hence it should be noted that while we can use XRD (39) Nuffield, E. W. X-ray Diffraction Methods; Wiley: New York, 1966.
DOI: 10.1021/la902583j
2071
Article
Figure 3. (a) XRD pattern of Cu0 powder prepared in solutions containing calf thymus DNA, Cu(NO3)2 and ascorbic acid. Inset: enlarged region of XRD pattern for 2θ = 33-41°, showing trace amounts of Cu2O present in the powder. The peak at 38.9° arises from the radiation source (Cu Kβ). (b) Enlarged region of XRD pattern for 2θ = 41.5-45°, showing the peak arising from the Cu (111) reflection and the peak fitted to a pseudo-Voigt function.
methods to confirm that the chemical reduction of the Cu2þ results in the formation of Cu0, there are clearly marked differences between the crystallinity of the powder sample and the λ-DNA-templated Cu0 nanostructures which consist of Cu0 particles of significantly smaller dimensions, as determined by AFM, see the Atomic Force Microscopy section. Atomic Force Microscopy. Figure 4 shows AFM height images of DNA molecules (a) before and (b) after the Cu0 metalization process is carried out. A greater variation in structure height is apparent in the metalized DNA structure, indicating nonuniform deposition of the Cu0 along the length of the DNA molecule. Regions of the DNA which exhibit most extensive Cu0 deposition are up to ca. 7 nm in height, with the values as low as ca. 2.5 nm also observed where less significant levels of metalization have taken place. Closer inspection of the Cu0 structures reveals a distinctive “beads-on-a-string” appearance, with nanoparticles of Cu0 packed along the DNA template molecules, Figure 5. Variations in the packing density of the Cu0 nanoparticles along the DNA templates can also be seen, with some regions between adjacent Cu0 particles where less substantial Cu0 deposition has taken place (white arrows, Figure 5). The heights of these regions however (ca. 2.5-3.0 nm), suggest that Cu0 2072 DOI: 10.1021/la902583j
Watson et al.
deposition has still taken place upon the DNA template in these areas, but to a small extent (c.f. λ-DNA is well-known to appear in AFM to be 1.0-1.5 nm in height). The granular morphology associated with the nanostructures is indicative of Cu0 deposition along the DNA templates taking place via a “nucleation and growth” mechanism, similar to that described in the preparation of DNA-templated Pd0 nanowires.21 This is believed to take place through chemical reduction of Cu2þ ions which are bound to the DNA structure, yielding small metallic clusters of Cu0 along the template molecule. These Cu0 clusters act as seed particles with further reduction of Cu2þ taking place directly at their surface, resulting in Cu0 growth upon the DNA templates. This proposed mechanism is consistent with previous works which describe the formation of metal nanoparticles by radiolytic reduction of aqueous metal cation solutions, in which metal seeds were reported to play a fundamental role in the nanoparticle growth.40-42 Statistical analysis of the average structure height, recorded for >100 (bare and metalized) DNA molecules, provides a more comprehensive profile of the Cu0 coverage of the DNA templates, Figure 6. Analysis of the free DNA shows a positively skewed distribution with the peak DNA height observed to be 1.01.5 nm. Larger structure heights (>3.0 nm), observed in much lower frequency, arise from “bundling” of the DNA molecules when in solution, leading to a small number of DNA “ropes” aligned upon the substrate.10 The distribution of the measured heights of the metalized DNA does not show a well-defined modal value, but suggests that the Cu0 deposition results in a small spread of structure sizes within the range 3.0-5.5 nm. Moderate levels of nonspecific deposition of Cu0 particulates, typically ca. 3-6 nm in height, are also observed across the substrate surface (mean estimated to be ca. 19.0 particles/μm2, standard deviation = 3.4; corresponding to ca. 17% surface coverage). Larger particulates (up to ca. 13 nm in height) are also present upon the substrate, but occur in significantly lower densities (ca. 2.5 particles/μm2, standard deviation = 1.0; corresponding to ca. 2% surface coverage). Scanned Conductance Microscopy. Electrostatic force microscopy (EFM) provides a ‘contactless’ means of probing the electrical properties of individual nanowires on an insulating substrate background through phase image mapping of the force gradient experienced above the sample. It has also been referred to as “scanned conductance microscopy” (SCM) when used to detect conductive objects on a dielectric film through their effect on the capacitance between the tip and the substrate. It has been shown that the phase shift produced by conductive objects is distinct from that due to polarizable insulators. Both the technical details of EFM/SCM and the interpretation of EFM phase images for one-dimensional structures have previously been discussed in more detail elsewhere.10,43-45 Briefly, if a nonconductive, but polarizable nanowire is modeled as a thin dielectric strip lying directly under the tip (modeled as a disk of defined radius, Rtip), the phase shift can be estimated by10 " # 2πR2tip ε0 2πR2tip ε0 Q 2 tanðΔφÞ ¼ V tip 2k ðhþt=εOx Þ3 ðhþt=εOx þd=εÞ3
ð1Þ
(40) Henglein, A. J. Phys. Chem. B. 2000, 104, 1206. (41) Henglein, A. Langmuir 2001, 17, 2329. (42) Henglein, A.; Meisel, D. Langmuir 1998, 14, 7392. (43) Bockrath, M.; Markovic, N.; Shepard, A.; Tinkham, M.; Gurevich, L.; Kouwenhoven, L. P.; Wu, M. S. W.; Sohn, L. L. Nano Lett. 2002, 2, 187. (44) Staii, C.; Johnson, A. T.; Pinto, N. J. Nano Lett. 2004, 4, 859. (45) Zhou, Y.; Freitag, M.; Hone, J.; Staii, C.; Johnson, J., A. T.; Pinto, N. J.; MacDiarmid, A. G. Appl. Phys. Lett. 2003, 83, 3800.
Langmuir 2010, 26(3), 2068–2075
Watson et al.
Article
Figure 4. TappingMode AFM height images of (a) a λ-DNA molecule immobilized upon a Si/SiO2 substrate, modified with a TMS SAM, (b) a λ-DNA molecule following solution treatments with Cu(NO3)2 and ascorbic acid, with (c) and (d) showing the corresponding cross sections of (a) and (b), respectively (cross sections in (d), from left to right, correspond to the cross sections regions highlighted in (b) going from left to right). Scale bar = 800 nm, height scale = 5 nm.
Figure 5. TappingMode AFM height image of a section of a λ-DNA molecule following solution treatments with Cu(NO3)2 and ascorbic acid, highlighting the “beads-on-a-string” morphology associated with Cu0 material deposited along the DNA. The arrows highlight regions between Cu0 nanoparticles on the DNA template, where less substantial Cu0 deposition has taken place. Scale bar = 200 nm, height scale = 8 nm.
Figure 6. Height distribution of >100 λ-DNA molecules before
(Q is the quality factor, k = cantilever spring constant, t = oxide thickness, and d = nanowire diameter), where the first term derives from the tip-substrate capacitance and the second term from the tip-nanowire-substrate capacitance. Inspection of eq 1 shows that polarizability of an insulating one-dimensional structure alone, can only provide a positive phase shift. However, for a conducting nanowire which allows the charge stored on the
nanowire/Si capacitor to be spread along the length of the wire (L), the second term in eq 1, in effect, becomes significantly larger (as the capacitance is now determined by L rather than by Rtip), resulting in a negative phase shift. The Cu0 nanostructures were prepared for SCM experiments upon silicon substrates with a ca. 220 nm thick thermally grown SiO2 layer on top. EFM phase images were recorded with the tip
Langmuir 2010, 26(3), 2068–2075
and after treatment with solutions containing Cu(NO3)2 and ascorbic acid. The heights were determined from TappingMode AFM images.
DOI: 10.1021/la902583j
2073
Article
Watson et al.
Figure 7. EFM phase images of a DNA-templated Cu0 nanowire recorded at a range of different sample biases. Scale bar = 500 nm, and the color scale corresponds to a phase angle of 4°.
typically lifted 40-50 nm above the sample surface, whereas a dc voltage was applied to the sample Figure 7 shows a series of EFM phase images of a typical Cu0 nanostructure recorded at a range of bias potentials. The Cu0 nanostructure shows up in the images with a positive phase 2074 DOI: 10.1021/la902583j
(relative to the substrate background) indicating a lack of electrical conductivity along the length of the structure. The tangent of the phase shift also exhibits a parabolic dependence upon the sample bias as predicted by eq 1 for one-dimensional insulating structures, Figure 8. The phase shifts for several Cu0 Langmuir 2010, 26(3), 2068–2075
Watson et al.
nanostructures were recorded as a function of bias and fitted to a second order polynomial of the form y = ax2 þ bx þ c. The coefficient, a, provides a quantitative measure of the contribution to the electrostatic force experienced arising from tip-sample capacitance effects, whereas effects due to electrostatic forces from trapped charge determine b. From the EFM data of the Cu0 structures analyzed, an average value of the second order coefficient was determined to be a = 0.002, standard deviation = 7 10-5. The slight asymmetry associated with the parabola may be attributed to the presence of trapped charges on the metal nanostructure which exhibit a linear dependence on bias.46 The reason for the high resistivity associated with these metallic Cu0 architectures cannot be determined from SCM studies alone. At present however, two possible contributions to the high resistivity of the structures have been identified: (i) the presence of electron tunneling barriers along the metal nanostructures and (ii) electron scattering effects. The possible occurrence of these mechanisms can be rationalized when considering the structure morphology and reduced dimensions of the Cu0 architectures produced. The Cu0 nanostructures have been shown by AFM to have a “beads-on-a-string” appearance, proposed to be the result of a “nucleation and growth” mechanism. This irregular metal coverage of the DNA templates suggests that the presence of discontinuities or voids along the length of the nanostructure are highly likely, through metal deposition failing to take place in regions between some of the Cu0 nanoparticles. Such voids in the metal structure would introduce significant tunneling barriers at various points along the Cu0 architectures, preventing efficient charge transport along the entire length of the nanostructure. Even in regions of the metal structures where the constituent Cu0 nanoparticles appear more closely packed together, the structures may be subject to electron scattering at the interfaces between adjacent Cu0 particles. The occurrence of further electron scattering events should also be considered, arising as a consequence of the small line widths of the structures. It is well-known that the resistivity of metal structures may increase significantly as their lateral dimensions approach 100 nm and below. These “size effects” occur as a result of the contributions to the total electrical resistivity from grain-boundary and surface electron scattering increasing significantly as the structure dimensions become comparable to the mean free path of the electron (ca. 40 nm in the case of copper).47,48 The extent of the contribution to the total resistivity of the Cu0 nanostructures which arise from electron tunneling and grainboundary/surface electron scattering mechanisms is currently unclear. Further studies in which the resistivity of the Cu0 nanostructures is measured over a range of temperatures would be required in order to clarify which of these mechanisms is the dominant in causing the high electrical resistivity observed in such one-dimensional nanostructured Cu0 materials. (46) Jespersen, T. S.; Nygard, J. Nano Lett. 2005, 5, 1838. (47) Steinhogl, W.; Schindler, G.; Steinlesberger, G.; Engelhardt, M. Phys. Rev. B. 2002, 66, 075414. (48) Wu, W.; Brongersma, S. H.; Hove, M. V.; Maex, K. Appl. Phys. Lett. 2004, 84, 2838. (49) Horcas, I.; Fernandez, R.; Gomez-Rodriguez, J. M.; Colchero, J.; GomezHerrero, J.; Baro, A. M. Rev. Sci. Instrum. 2007, 78, 013705.
Langmuir 2010, 26(3), 2068–2075
Article
Figure 8. Tangent of phase shift vs sample bias for a single λ-DNA molecule following solution treatments with Cu(NO3)2 and ascorbic acid.
Conclusions In summary, XPS and XRD methods have been used to confirm that the preparation of one-dimensional nanostructures through the reduction of Cu2þ upon DNA templates with ascorbic acid results in the formation of metallic Cu0 structures. AFM shows the DNA-templated metal structures to be 3.0 5.5 nm in height, proving in good agreement with previous literature.23,25 However, despite their metallic composition, no evidence of electrical conductivity has been observed along these nanostructures during SCM studies. This is attributed to the “beads-on-a-string” morphology and small lateral dimensions of the metallic structures, resulting in electron scattering at the interfaces between the Cu0 nanoparticles as well as at the structure surfaces, inhibiting charge delocalization along the length of the whole structure. Further work is currently underway to refine the morphology of the Cu0 nanostructures, aimed toward forming more continuous one-dimensional metal architectures rather than the “beads-on-a-string” structuring that is reported here. Through reducing the number of particle-particle boundaries in the Cu0 structures, removal of tunneling barriers and/or attenuation of electron boundary scattering is expected to lead to a reduction in the resistivity of these structures. Acknowledgment. We thank One North East and Newcastle University for financial support of this work. Dr. Lidija Siller and Ross Little are acknowledged for XPS data. Supporting Information Available: Additional FTIR spectra (600-4000 cm-1) comparing DNA and Cu2þ/DNA complexes and tables stating band positions, N(1s) and P(2p) XPS spectra of DNA-templated Cu0 nanostructures, and additional AFM data of DNA-templated Cu0 nanostructures. This material is available free of charge via the Internet at http://pubs.acs.org.
DOI: 10.1021/la902583j
2075
