LETTER pubs.acs.org/NanoLett
Spatially Resolved Plasmonically Enhanced Photocurrent from Au Nanoparticles on a Si Nanowire Jerome K. Hyun and Lincoln J. Lauhon* Department of Materials Science and Engineering, Northwestern University, Evanston, Illinois 60208, United States
bS Supporting Information ABSTRACT: Semiconducting nanowires have been demonstrated as promising light-harvesting units with enhanced absorption compared to bulk films of equivalent volume. However, for small diameter nanowires, the ultrahigh aspect ratio constrains the absorption to be polarization selective by responding primarily to the transverse magnetic (TM) light. While this effect is useful for polarization-sensitive optoelectronic devices, practical light-harvesting applications demand efficient light absorption in both TM and transverse electric (TE) light. In this study, we engineer the polarization sensitivity and the charge carrier generation in a 50 nm Si nanowire by decorating the surface with plasmonic Au nanoparticles. Using scanning photocurrent microscopy (SPCM) with a tunable wavelength laser, we spatially and spectrally resolve the local enhancement in the TE photocurrent resulting from the plasmonic near-field response of individual nanoparticles and the broad-band enhancement due to surface-enhanced absorption. These results provide guidance to the development and the optimization of nanowire nanoparticle light-harvesting systems. KEYWORDS: Photocurrent, plasmonics, nanowire, scanning photocurrent microscopy
S
emiconducting nanowires have been demonstrated as efficient light-harvesting units because of their diameter-dependent spectral response to light.1 Additionally, by structuring the nanowire with radial junctions, long absorption lengths, and short minority carrier diffusion, lengths can be realized simultaneously, characteristics well-suited for solar energy applications.2,3 Superior absorption efficiencies compared to bulk films of equivalent volume have also been demonstrated due to the presence of multiple optical resonances in the nanowire.4 While this enhancement in absorption per equivalent volume generally increases for smaller diameters, the absorption in the limit of small diameters becomes polarization sensitive by responding primarily to light polarized parallel to the nanowire axis. This presents a challenge to practical light-harvesting applications where absorbing light of all polarizations is most useful. We propose a general solution to this problem by hybridizing metal nanoparticles with the nanowire, thereby providing additional degrees of freedom to engineer the polarization sensitivity and the conversion of light absorption into charge carrier generation through efficient surface plasmonic coupling. For the past two decades, surface plasmonic coupling between metal nanoparticles and semiconductor systems has been investigated extensively.5 8 Because the strong plasmonic nearfield from metal nanoparticles is driven parallel to the polarization and not in the direction of the light propagation, any metal nanoparticle resting on the entrance surface of a light absorbing material will provide poor plasmonic contribution to the absorption. Such was the case for bulk Si photodiodes supporting metal nanoparticles, where enhancements in the photocurrent were attributed to forward scattering from the nanoparticles rather than the plasmonic near-field.9 11 r 2011 American Chemical Society
On the other hand, semiconducting nanowires side-contacted by metal nanoparticles provide an opportunity to efficiently harness surface plasmons by satisfying the geometrical constraint of orthogonality between the plasmonic near-field and the direction of light propagation. Already, vertical arrays of Si nanowires have shown great promise as solar cells.2 With metal nanoparticles resting along the nanowire body, the strong nearfield intensity excited from light traveling along the direction of the nanowires is directed into the surface of the nanowire, generating a local density of photocarriers. Accordingly, measurements on arrays of vertical Si nanowires decorated with Pt nanoparticles have demonstrated considerable increases in the photocurrent.12 The efficient plasmonic near-field coupling can also be employed in geometries where the nanowire is oriented parallel to the substrate. This geometry has previously been demonstrated to be promising for broad-band absorption and wide-range angular responsivity.1,4 In this communication, we report the use of variable-wavelength scanning photocurrent microscopy11,13 16 (SPCM) to investigate a model device consisting of a Si nanowire device resting on the substrate and decorated with Au nanoparticles. We observed a 20% increase in the local TE photocurrent due to the plasmonic nearfield response of a single Au nanoparticle, which represents an increase in the effective light-harvesting cross-section of 200%. In addition we observe a nonplasmonic broadband enhancement in the TE photocurrent due to surface-enhanced absorption. The Received: March 27, 2011 Revised: May 13, 2011 Published: June 06, 2011 2731
dx.doi.org/10.1021/nl201021k | Nano Lett. 2011, 11, 2731–2734
Nano Letters combination of both effects results in larger than expected TE/TM photocurrent ratios across the visible range and the near-unity ratios at the plasmonic resonance. These findings provide guidance for optimal positioning schemes of the nanoparticles in hybrid nanostructured materials. The n-type Si nanowires were grown by the vapor liquid solid (VLS) method in a H2-rich environment as described in ref 18, which reported evidence for H2-assisted reduction of the active dopant concentration gradient. Nanowires with diameters of 50 nm were selected because Mie calculations showed a smooth monotonic decrease in absorption from λ = 475 to 700 nm (see Supporting Information, Figure S1). Ohmic contacts were defined by electronbeam lithography, metal deposition of 50 nm Ni and 50 nm Au, and liftoff. Commercial 50 nm-sized Au nanoparticles (BBInternational) immersed in water were then drop-cast onto the nanowire device. The 50 nm Au nanoparticles have a large Mie absorption crosssection with negligible scattering in the visible range (see Supporting Information, Figure S2). A scanning confocal microscope (Witec) coupled to a tunable laser source (NKT Photonics) was used to acquire the spatially and spectrally resolved photocurrent in the nanowire devices. A lock-in amplifier was used to record the photocurrent at the beam modulation frequency of 1837 Hz. The photocurrent was directly proportional to laser intensity and applied bias. Therefore we assume that the photocurrent is directly proportional to absorption. The laser power was maintained at 65 μW as the output wavelength was changed from 475 to 700 nm. Finite difference time domain (FDTD) simulations were conducted using Lumerical Solutions. Figure 1a shows the SEM image and the corresponding unpolarized reflection image at 532 nm of the hybrid device illustrated schematically in Figure 1b. Figure 1c displays the photocurrent maps at selected wavelengths between 475 and 700 nm for TM (E-field parallel to the nanowire) and TE (E-field is perpendicular to the nanowire) excitation. For all wavelengths the light was focused to a diffraction-limited spot using a 0.90NA objective lens. The bottom contact of the device was biased at 1 V and the top contacted grounded. To generate the photocurrent map, the piezostage supporting the device was scanned in 125 nm increments, and the photocurrent was measured at each point. Higher definition maps were obtained using 62.5 nm increments for scans showing localized resonances. Before identifying the plasmonic photocurrent induced by the particle, we describe the general characteristics of the SPCM images. The large positive (bright) and negative (dark) photocurrents at the top and bottom contacts, respectively, are due to contact-induced band bending as described previously.13,14 Within the channel, the TM photocurrent is generally larger than the TE photocurrent, and for increasing wavelengths, the overall intensity in the nanowire decreases monotonically. These observations are in qualitative agreement with Mie calculations of bare 50 nm diameter Si nanowires that show higher absorption for TM than for TE illumination and decreasing absorption with increasing wavelength (see Supporting Information). The gradual decay in intensity from top to bottom along the nanowire channel observed in both TM and TE scans is due to a gradient in the surface doping concentration as reported previously.17,18 As the photocurrent is inversely proportional to the majority carrier concentration, the lighter intensity near the bottom of the channel indicates an increase in dopant concentration.17 There are several regions of enhanced photocurrent in the 532 and 550 nm TE images (white arrows in Figure 1c) that are not seen in the TM images. Simulation of the polarization- and
LETTER
Figure 1. (a) (Left panel) SEM image of the Si nanowire device decorated by Au nanoparticles. Scale bar is 1 μm. (Right panel) The corresponding 532 nm reflection image of the same device. (b) Device schematic. (c) Photocurrent images for TM (top) and TE (bottom) at 1 V bias. Images are 4 16 μm2.
wavelength-dependent absorption using finite-difference timedomain (FDTD) methods (Figure 2) indicates that these features are due to plasmonically enhanced carrier generation in the nanowire. Figure 2a and b shows the spatial distribution of absorption per unit volume for TM and TE excitation. Both a bare nanowire control (Figure 2a) and a Au nanoparticlecoupled nanowire (Figure 2b) were simulated. In all cases, a Gaussian beam from a thin lens approximation (0.9NA) was used to model the diffraction limited laser spot centered on the nanowire. For the bare nanowire (Figure 2a), it is apparent that most of the TM absorption occurs near the center of the nanowire, whereas most of the TE absorption occurs near the surface. In each case, the absorption diminishes with increasing wavelength. By integrating the absorption within the nanowire at each wavelength (Figure 2c), one can see that the TE response is enhanced by the presence of the nanoparticle, whereas the TM response is minimally perturbed. Because the plasmon oscillation in the nanoparticle is driven parallel to the electric field, the nearfield intensity is peaked at the metal surface normal to the direction of polarization. The nanowire therefore experiences the largest absorption enhancement from the plasmonic nearfield when the polarization is parallel to the axis between the nanowire and the nanoparticle. The magnitude and the peak wavelength of the plasmonic enhancement are sensitive to the nanowire nanoparticle separation, as expected (Figure 2d). Because the near-field is evanescent, the response decays exponentially with separation. As the gap spacing between the nanoparticle and the nanowire decreases, the nanoparticle experiences a stronger dielectric presence introduced by the nanowire. This results in the red-shifting of the plasmonic response for smaller separation distances. The regions of enhanced photocurrent correspond to points where Au nanoparticles are attached to the nanowire (Figure 3). At the positions of the four ‘hot spots’ along the nanowire, scanning electron microscopy (SEM) images reveal at least one 48 ( 4 nm-sized Au nanoparticle in contact with the nanowire. The region labeled A shows an isolated nanoparticle with a gap distance of 10 nm, but there is no detectable plasmonic photocurrent, in agreement with simulations. One can also note the presence of two isolated nanoparticles in contact with the 2732
dx.doi.org/10.1021/nl201021k |Nano Lett. 2011, 11, 2731–2734
Nano Letters
LETTER
Figure 2. Absorption per unit volume in the center of an illuminated Si nanowire (a) and a Si nanowire in contact with a Au nanoparticle (b) for TM (top row) and TE (bottom row) excitation. (c) The integrated absorption in bare nanowire (blue lines) and in the nanowire-nanoparticle hybrid (red lines) for TM and TE excitation. (d) Simulated TE absorption in the nanowire nanoparticle hybrid for different gap sizes. The focal point is set at the particle in each case.
nanowire in region 2. The size of the smaller nanoparticle is 35 ( 4 nm, and it therefore contributes less to the photocurrent due to a smaller absorption cross-section. The SEM image also reveals regions along the nanowire in contact with clusters of Au nanoparticles. The detailed structure of these clusters cannot be easily determined with SEM scans. Therefore we restrict our analysis to single nanoparticles attached to the nanowire surface. To confirm the plasmonic origin of the photocurrent enhancement, Figure 3b compares the ratio of the experimental TE/TM photocurrent with the calculated TE/TM absorption for the four hot spots. The photocurrent at each wavelength was obtained by taking a horizontal line profile across the nanowire at the position of each hot spot and using the peak value of a fitted Gaussian function. A distinct plasmonic response is observed in the range of 525 575 nm in both the experiments and the simulations. In contrast, no such response is seen at position A (Figure 3c), confirming the mechanism of photocurrent enhancement. By comparing the plasmonic photocurrent in region 3 with a bare region located 0.75 μm away, the plasmonic enhancement is found to be 20% (Supporting Information, Figure S3). While an enhancement of 800% from the TE absorption is simulated for a particle in contact with the nanowire, the nanowire is coated by a 1 2 nm layer of native oxide. This will significantly reduce the enhancement in the plasmonic absorption (Figure 2d) to levels comparable to the experimentally determined value. Notably, the Au nanoparticle increases the geometrical cross-section of the illuminated region by only 10%, yet it increases the TE photocurrent by 20%. Therefore, the differential increase in effective light harvesting cross-section is 200%. Plasmonically induced heating of a nanoparticle, leading to heating of the nanowire, could in principle also affect the photocurrent signal. This effect would likely manifest as a decrease in photocurrent due to the decrease in mobility.19 In the case of single nanoparticles, the thermal response should be independent of polarization because thermal equilibration within the particle following plasmonic excitation should occur much faster than transport between the particle and the nanowire. We do not detect local photocurrent under TM excitation, but we do under TE excitation, indicating that the electromagnetic response dominates any thermally induced response.
Figure 3. (a) SEM and SPCM images of a nanowire device decorated with Au nanoparticles. The plasmonic response is analyzed in four regions, indicated by white boxes, in which Au nanoparticles contact the nanowire. Region A (yellow box) shows a Au nanoparticle with nanoparticle nanowire gap spacing of 10 nm. Scale bars are 1 μm and 300 nm for the low- and high-magnification images, respectively. (b) The experimental TE/TM photocurrent ratios (colored markers) compared to the calculated absorption ratio (black line). (c) The TE/ TM photocurrent ratio for region A compared to the calculated absorption ratio.
While the simulations provide clear confirmation of a plasmonic enhancement, the experimental TE/TM photocurrent ratio is consistently higher than the simulated absorption ratio. In fact, at the plasmonic resonance, the experimental TE photocurrent is as large as the TM photocurrent, and experimental TE/ TM ratio does not decay to the calculated ratio in the limit where the wavelength is much greater than the nanowire diameter (Figure 3c). These observations suggest that interface states, the response of which is not included in our model, may contribute to absorption. Because the TE absorption is concentrated near the surface (Figure 2a), interface-related absorption should disproportionately impact the TE photocurrent. Photocurrent measurements from surface states in optical waveguides generated by optical modes peaked at the surface have been reported previously.20 In further support of the influence of near-surface absorption, we note that there is a high concentration of active dopant species near the surface for nanowires grown under these 2733
dx.doi.org/10.1021/nl201021k |Nano Lett. 2011, 11, 2731–2734
Nano Letters conditions.18 Consequently, the recombination velocity at the surface is expected to be high, inhibiting the surface-sensitive TE photocurrent more than the bulk-sensitive TM photocurrent. The systematic decrease in experimental TE/TM photocurrent moving from the more lightly surface-doped region 1 to the more heavily surface-doped region 4 (Figure 3b) is consistent with this explanation. In conclusion, we have demonstrated a 20% increase in the plasmonically enhanced TE photocurrent for individual Au nanoparticles attached to a Si nanowire. As hybrid assemblies of nanowires and metal nanoparticles offer opportunities for lightharvesting applications, the geometrical considerations discussed here provide elementary guidance for further optimization. In addition, broadband enhancement of the TE photocurrent due to surface-enhanced absorption was shown, resulting in a higher TE/ TM photocurrent ratio. This result suggests the possibility of further enhancing and controlling photocurrent through surface modification.
LETTER
(13) Ahn, Y.; Dunning, J.; Park, J. Nano Lett. 2005, 5, 1367–1370. (14) Gu, Y.; Romankiewicz, J. P.; David, J. K.; Lensch, J. L.; Lauhon, L. J.; Kwak, E. S.; Odom, T. W. J. Vac. Sci. Technol., B 2006, 24, 2172–2177. (15) Freitag, M.; Tsang, J. C.; Bol, A.; Yuan, D. N.; Liu, J.; Avouris, P. Nano Lett. 2007, 7, 2037–2042. (16) Falk, A. L.; Koppens, F. H. L.; Yu, C. L.; Kang, K.; Snapp, N. D.; Akimov, A. V.; Jo, M. H.; Lukin, M. D.; Park, H. Nat. Phys. 2009, 5, 475–479. (17) Allen, J. E.; Perea, D. E.; Hemesath, E. R.; Lauhon, L. J. Adv. Mater. 2009, 21, 3067–3072. (18) Koren, E.; Hyun, J. K.; Givan, U.; Hemesath, E. R.; Lauhon, L. J.; Rosenwaks, Y. Nano Lett. 2011, 11, 183–187. (19) Klaassen, D. B. M. Solid-State Electron. 1992, 35, 961–967. (20) Baehr-Jones, T.; Hochberg, M.; Scherer, A. Opt. Express 2008, 16, 1659–1668.
’ ASSOCIATED CONTENT
bS
Supporting Information. Mie calculations for a 50 nm Si nanowire and 50 nm Au nanoparticle, FDTD simulation parameters, and calculations of the local TE photocurrent enhancement. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected].
’ ACKNOWLEDGMENT This work was supported by the Department of Energy, Basic Energy Sciences grant number DE-FG02-07ER46401. We thank Eric Hemesath for growth of the nanowires. ’ REFERENCES (1) Cao, L. Y.; White, J. S.; Park, J. S.; Schuller, J. A.; Clemens, B. M.; Brongersma, M. L. Nat. Mater. 2009, 8, 643–647. (2) Kelzenberg, M. D.; Boettcher, S. W.; Petykiewicz, J. A.; Turner-Evans, D. B.; Putnam, M. C.; Warren, E. L.; Spurgeon, J. M.; Briggs, R. M.; Lewis, N. S.; Atwater, H. A. Nat. Mater. 2010, 9, 239–244. (3) Tian, B. Z.; Zheng, X. L.; Kempa, T. J.; Fang, Y.; Yu, N. F.; Yu, G. H.; Huang, J. L.; Lieber, C. M. Nature 2007, 449, 885–U8. (4) Cao, L. Y.; Fan, P. Y.; Vasudev, A. P.; White, J. S.; Yu, Z. F.; Cai, W. S.; Schuller, J. A.; Fan, S. H.; Brongersma, M. L. Nano Lett. 2010, 10, 439–445. (5) Hagglund, C.; Zach, M.; Kasemo, B. Appl. Phys. Lett. 2008, 92, 053110. (6) Kirkengen, M.; Bergli, J.; Galperin, Y. M. J. Appl. Phys. 2007, 102, 093713. (7) Konda, R. B.; Mundle, R.; Mustafa, H.; Bamiduro, O.; Pradhana, A. K.; Roy, U. N.; Cui, Y.; Burger, A. Appl. Phys. Lett. 2007, 91, 191111. (8) Atwater, H. A.; Polman, A. Nat. Mater. 2010, 9, 205–213. (9) Lim, S. H.; Mar, W.; Matheu, P.; Derkacs, D.; Yu, E. T. J. Appl. Phys. 2007, 101, 104309. (10) Matheu, P.; Lim, S. H.; Derkacs, D.; McPheeters, C.; Yu, E. T. Appl. Phys. Lett. 2008, 93, 113108. (11) Sundararaian, S. P.; Grady, N. K.; Mirin, N.; Halas, N. J. Nano Lett. 2008, 8, 624–630. (12) Peng, K. Q.; Wang, X.; Wu, X. L.; Lee, S. T. Nano Lett. 2009, 9, 3704–3709. 2734
dx.doi.org/10.1021/nl201021k |Nano Lett. 2011, 11, 2731–2734
