1. Introduction

Recently, surface plasmons on noble metal nanostructures have become one of the hottest topics in optical sciences [1]. The surface plasmon is a collective oscillation of quasi free electrons in metal nanostructures, and induces strong absorption and scattering at the resonance frequency [2]. In particular, the resonance characteristics strongly depend on the complex dielectric function of the material [2], the surrounding medium [3], and the shape of the nanostructure. Applications of surface plasmons on noble metal nanostructures include nano barcodes [4], metallic waveguides [5], surface plasmon assisted lithography [6], and nano biological labels [7]. In particular, nanowires are one of the best building blocks for bottom-up technology since they are small enough in diameter to retain nanoscopic nature, but large enough in length for manipulation such as delivery, arrangement, and elimination [8–11].

Among the noble metals, Cu nanostructures have attracted little attention due to their suppressed surface plasmon peak; few studies are available in the literature [12–13]. In this paper, however, we demonstrate that a strong, well-pronounced surface plasmon peak can develop in Cu nanowires in the deep red region.

2. Experimental methods

Copper nanowires used in our experiments were produced by a chemical solution method [14] using alkylamine as a surfactant. The method does not require a porous template or catalyst to induce wire growth. Rather, seeds are reduced from metal precursors and then grown into nanowires. The Ostwald ripening is widely accepted as the growth mechanism of nanowires [15]. Especially, long chain surfactant molecules can act like shape-controlling template during the ripening phase.

The detailed fabrication method is as follows: CuCl₂·2H₂O (2.5g, 0.014mol) and octadecylamine (ODA, 7.90g 0.029mol) were mixed and heated to 180°C under Argon atmosphere. Upon heating for 3 hours, the mixture formed a clear yellow solution that slowly turned to a red color, indicating the formation of copper nanoparticles in solution. The reaction was continued for 48 hours, after which it was cooled to room temperature. The clump of reaction product was washed with 30ml of hot dodecane three times to remove residual ODA. The residue was additionally washed with excess ethanol and deionized water. Isolated copper nanowires were filtered, collected, and dried under vacuum pressure. Transmission Electron Microscopy (TEM) and X Ray Diffraction (XRD) of our chemically produced nanowires clearly confirm the high aspect ratio cylindrical shape of the Face Centered Cubic (FCC) single crystal structure with 3.615Å lattice constant (Fig. 1). For optical measurements, the copper nanowires were dispersed in hexane.

 

Fig. 1. (a) SEM image of Cu nanowires. (b) TEM image of a Cu nanowire. Scale bar: 5nm. (c) High resolution TEM image of the same Cu nanowire. Box indicator shows 10 layers of (111) direction (2.08nm/10 lines). Growth direction is (011).

Download Full Size | PDF

We utilized total internal reflection (TIR) microscopy to investigate the plasmon resonance (Fig. 2) [9]. For light scattering measurements, one drop of Cu nanowire colloids was cast on a clean glass cover slip and dried under the ambient conditions. The glass cover slip was subsequently loaded on a right angle prism with index matching oil to maintain the total internal reflection condition. As shown in Fig. 2, white light of a 150W tungsten-halogen lamp impinged on one side of the prism, and reflected off the glass surface completely. A linear polarizer could be installed in front of the lamp whenever CCD images of Cu nanowires were acquired. By using total internal reflection, the background was sufficiently suppressed to obtain a high signal-to-noise scattering signal from individual single Cu nanowires.

 

Fig. 2. Experimental setup for total internal reflection microscopy. TH: tungsten-halogen lamp, P: prism, OL: objective lens, LP: linear polarizer, A: circular variable aperture, M: monochromator, I: ICCD.

Download Full Size | PDF

The scattered signal was collected with a 0.80 N.A. objective lens on an inverted optical microscope (Olympus, IX71), and then dispersed by a 32cm long monochromator. At the exit port of the monochromator, an intensified CCD (ICCD; Andor Technology, DH520) was mounted to measure the spectrum signal. Every spectrum was normalized by the input intensity of the light source. A linear polarizer was inserted between the microscope side port and the entrance slit of the monochromator to control the polarization state of the measured scattered fields. We could identify the spectra of single Cu nanowires because they were wellseparated from each other on the glass cover slip. To further exclude the signal from neighboring nanowires, we inserted a circular variable aperture at the image plane a few centimeters from the side port of the microscope, similar to a confocal microscope setup. In this configuration, when nanowires were moved out of the focal volume, we detected a very weak background signal that was used for background correction of all spectra presented in this paper.

3. Experimental results

Figure 3 depicts a typical image of Cu nanowires. Perpendicular polarization to the long axis of the Cu nanowire develops strong surface plasmon scattering as shown at Fig. 3(a). Due to the surface plasmon scattering, the Cu nanowire shines in a bright red color. As seen in Fig. 3, the TIR condition generated the high contrast image of individual nanowires. For polarization parallel to the long axis, the brightness of the nanowire scattering signal weakened significantly, but did not disappear completely. The origin of the parallel polarization light is not clear yet, but may involve bulk-like metal scattering, since the spectrum in Fig. 4(a) is almost flat below 1.8eV and quite similar to the reflectance spectrum of (bulk) Cu. One interesting feature in Fig. 3 is small protrusion in the upper part of the nanowire. The brightness level in this region is reversed for each given polarization state, suggesting that a small nanorod might have formed.

Using spectroscopic tools, the corresponding spectra for different polarization states of the nanowire were measured (Fig. 4(a)). Below 1.5eV, detection of the scattering signal was limited by the low light intensity of the Tungsten-Halogen lamp in the infrared and the relatively low quantum efficiency of the ICCD. However, above 1.5eV, the scattering signal was sufficient enough to identify the surface plasmon peak. Here, the spectrum shows a peak at around 720nm in the deep red visible region. According to Mie scattering theory, in general, scattering spectra of large diameter Cu nanowires have multi-polar plasmon resonances at shorter wavelength [2,11]. As shown in Fig. 4(b), the observed peak clearly exhibits a dipole pattern. Note that the dipole radiation in the sample plane was measured by using a collecting objective lens and a rotating linear polarizer. Therefore, the measured pattern represents the projection of the dipole radiation onto the sample plane, induced by the surface plasmon scattering of the Cu nanowire.

 

Fig. 3. CCD images of a Cu nanowire under white light illumination for polarization (a) perpendicular, or (b) parallel to the nanowire long axis. Arrows indicate incident E-field polarization. Scale bar: 5 µm. Inset: a typical SEM image of a Cu nanowire.

Download Full Size | PDF

 

Fig. 4. (a) Surface plasmon scattering spectra of a Cu nanowire with different polarization states. (b) Polar plot shows the angular variation of the peak.

Download Full Size | PDF

We measured scattering spectra for a set of individual nanowires with different diameters. Figure 5 displays scattering spectra that have been normalized for clarity. Most nanowires showed a strong scattering signal for perpendicular polarization to the long axis. In particular, the surface plasmon peak clearly shows a red-shift with increasing nanowire diameter. Small diameter nanowires show a rather weak and buried peak while, in general, nanowires as large as ~300nm show scattering peak heights 1–2 orders of magnitude higher than those of small nanowires under the same experimental conditions. After scattering measurement, the diameter of each Cu nanowire was determined by SEM imaging and plotted against its peak energy (inset of Fig. 5). At a diameter of 300 nm, the surface plasmon peak shifts down near to 1.75eV. Further increasing the diameter seems to saturate the red-shift effect. However, because of limitations on the size distribution for nanowires larger than 300nm in diameter, the saturation behavior could not be confirmed by our data. Nevertheless, with our current data, we demonstrated that Cu nanowires show a strong surface plasmon peak in deep red of the visible region, and believe that the most reasonable origin for the red-shift of comparably larger diameter nanowires is the phase retardation effect [2].

 

Fig. 5. Surface plasmon scattering spectra of a set of Cu nanowires having different diameters. Inset plots the peak energy position as a function of the diameter of Cu nanowires.

Download Full Size | PDF

4. Discussion

As is well known, the peculiar optical properties of noble metals like Ag, Au, and Cu are attributed to the contributions by bound electron transitions from d to sp conduction band [2]. Due to interband absorption, the actual dielectric function deviates from the simple free-electron-like Drude model. One of the most important results is the shift of bulk plasmon resonance.

For ellipsoids much smaller than the wavelength and with complex dielectric function ε in a surrounding medium of ε_m, Mie scattering can be approximated to the fundamental Fröhlich mode. The induced dipole moment along the axis i is therefore given by

where a, b, c are the dimensions of the ellipsoid, and L_i is the geometrical depolarization factor [9,16], which depends on the shape of the ellipsoid; one can find for a sphere L_a=L_b=L_c=1/3 and for a infinite cylinder L_a=L_b=1/2, L_c=0, taking c as the infinite cylinder axis. The resulting polarization perpendicular to the cylinder axis is inversely proportional to ε+ε_m, which shows resonance behavior when $\mid \epsilon +{\epsilon }_m\mid =\sqrt{{\left({\epsilon }_1+{\epsilon }_m\right)}^2+{{\epsilon }_2}^2}$ is minimized, where ε _1,2 are the real and imaginary part of the complex dielectric function respectively. Therefore, in alkali metals and Ag, the common relation ε ₁=-ε_m for the resonance can be true due to small ε ₂ [9]. However, Cu has large ε ₂ for the above condition [17], so the surface plasmon resonance of nanowires occurs where ε ₁~-6 in air, almost the same but slightly blue shifted from that of nanospheres [2].

When the diameter of nanowires is comparable to the wavelength, the electrostatic limit is not valid any longer, and the problem should be solved using full Mie theory. In this realm, cylinder problem can be solved analytically only for an infinite cylinder. For an infinite cylinder of radius a, the scattering efficiency per unit length for polarization perpendicular to the cylinder axis and normal incidence is given by [16],

with

where m=k ₁/k, x=ka.. Here, k ₁ and k are the wave vectors in the cylinder and medium, respectively, and J_n(x) and H ⁽¹⁾ _n(x) are the nth order Bessel functions of the first and third kinds (so called Hankel functions). The calculations for the infinite cylinder show the red-shift of the surface plasmon peak, though the peak is buried in the background (Fig. 6(a)). However, the surface plasmon peak could be resolved by decomposing each component of equation (3). Fig. 6(b) clearly shows the dipolar term (n=1) shifts to the red, in contrast to Fig. 6(a), in which higher multi-polar terms grow fast enough to bury the peak. Therefore, the limitation of our theoretical model, assuming infinite length, may overestimate the contribution of higher order terms (n=2,3,4,…). For comparison, we calculated the full Mie scattering efficiency of Cu nanospheres (Fig. 6(c)) and obtained a clear red-shift of the surface plasmon peaks with increasing diameter due to the phase retardation effect. Therefore, we come to the conclusion that our finite nanowires retain dipolar scattering like nanospheres, in a sense. Further theoretical investigation of the finite cylinder scattering problem is in progress.

 

Fig. 6. Calculated surface plasmon scattering spectra of Cu nanowires and nanospheres: nanowires by (a) full Mie scattering, (b) first order component only, and (c) nanospheres by full Mie scattering. For clarity, all the calculated spectra were normalized by their peak values.

Download Full Size | PDF

5. Conclusion

In conclusion, we have measured a set of scattering spectra of chemically produced Cu nanowires by using total internal reflection microscopy. High resolution spectroscopy made it possible to obtain the surface plasmon scattering spectra of individual Cu nanowires. In particular, we have observed that the surface plasmon peak red-shifts with increasing nanowire diameter. In addition, we have demonstrated that a strong, well-pronounced surface plasmon peak can develop in Cu, which normally has a suppressed surface plasmon peak, if the diameter is comparable to the wavelength. Strong red scattering from Cu nanowires may find applications in biological labeling, single molecule detection, nano barcodes, etc.