báo plasmonic
Published: October 31, 2011 r 2011 American Chemical Society 5548 dx.doi.org/10.1021/nl203457v | Nano Lett. 2011, 11, 5548–5552 LETTER pubs.acs.org/NanoLett Plasmonic Photosensitization of a Wide Band Gap Semiconductor: Converting Plasmons to Charge Carriers Syed Mubeen, † Gerardo Hernandez-Sosa, ‡ Daniel Moses, ‡ Joun Lee, † and Martin Moskovits* ,† † Department of Chemistry and Biochemistry and ‡ Center for Polymers and Organic Solids, University of California, Santa Barbara, California 93106, United States b S Supporting Information P lasmonic concentration and propagation have recently been proposed as strategies for enhancing photovoltaic and photo- catalytic performance. 16À20 In this Letter we describe a device, fabricated entirely using foundry processes with which a wide band gap semiconductor is photosensitized by embedding plasmonic nanoparticles within it, thereby significantly broad- ening its photoconversion ability beyond the ultraviolet region. The active element of the device is a composite solid film consisting of multiple, dense two-dimensional planar arrays of gold nanoparticles with each layer well separated by TiO 2 . Ohmic Ti/Au metal interdigitating electrical contacts were fashioned on the upper surface of the film using photolithogra- phy (Figure 1a, and the methods section in the Supporting Information). The ultraviolet/visible absorption/extinction spectrum of gold nanoparticles produced on a quartz substrate shows a localized surface plasmon resonance (LSPR) maxima at 520 nm, indicative of well-separated gold nanoparticles (Figure 1b). When capped by a TiO 2 film of 200 nm (mass thickness), the LSPR red shifts by 100 nm and becomes more intense, primarily due to the increase in the dielectric constant of the surrounding medium (ranges from 55 to 130 for polycrystal- line anatase TiO 2 ) over that of air. 21À24 Surrounding the gold nanoparticles with titania also creates a Schottky junction at the metalÀsemiconductor interface, which (in this case) results in charge transfer from the TiO 2 to the gold nanoparticles (AuNPs), charging the gold negatively and the TiO 2 positively in the vicinity of the NPs, and producing a potential barrier ∼0.9 eV (Figure 1c). 25,26 TiO 2 (E g = 3.3 eV) was chosen on account of its excellent electron-accepting capability due to the high density of states in the conduction band and the absence of photoresponse in the visible, suggesting a low density of appro- priately located defect states. 27,28 The major goal of the present study is to determine to what extent the plasmonic energy resident in optically excited gold nanoparticles can be transferred to electrons that either over- come the metal semiconductor energy barrier or tunnel through it to become conduction electrons in the semiconductor, which can be probed as changes in the conductance of the TiO 2 /Au composite. Prima facie evidence that those TiO 2 conduction electrons began life as plasmonic excitation of the AuNPs would be the degree to which the device’s photoconductance tracks the plasmonic extinction spectrum of the composite material keep- ing in mind that the gold nanoparticles are spaced significantly below the conductance percolation threshold of the AuNPs. Received: October 3, 2011 Revised: October 28, 2011 ABSTRACT: A fruitful paradigm in the development of low- cost and efficient photovoltaics is to dope or otherwise photo- sensitize wide band gap semiconductors in order to improve their light harvesting ability for light with sub-band-gap photon energies. 1À8 Here, we report significant photosensitization of TiO 2 due to the direct injection by quantum tunneling of hot electrons produced in the decay of localized surface-plasmon polaritons excited in gold nanoparticles (AuNPs) embedded in the semiconductor (TiO 2 ). Surface plasmon decay produces electronÀhole pairs in the gold. 9À15 We propose that a significant fraction of these electrons tunnel into the semicon- ductor’s conduction band resulting in a significant electron current in the TiO 2 even when the device is illuminated with light with photon energies well below the semiconductor’s band gap. Devices fabricated with (nonpercolating) multilayers of AuNPs in a TiO 2 film produced over 1000-fold increase in photoconductance when illuminated at 600 nm over what TiO 2 films devoid of AuNPs produced. The overall current resulting from illumination with visible light is ∼50% of the device current measured with UV (pω > E g band gap) illumination. The above observations suggest that plasmonic nanostructures (which can be fabricated with absorption properties that cover the full solar spectrum) can function as a viable alternative to organic photosensitizers for photovoltaic and photodetection applications. KEYWORDS: Plasmonics, photoconductance, gold nanoparticles, titania, impedance spectroscopy 5549 dx.doi.org/10.1021/nl203457v |Nano Lett. 2011, 11, 5548–5552 Nano Letters LETTER Hence, the only material through which continuous conductance can occur is the TiO 2 . Very recently, evidence that such metal to semiconductor tunneling could occur was reported from gold nanoantennas fabricated on silicon to the silicon. 29 The wavelength-dependent photocurrent response measured between interdigitating electrodes patterned on AuNPÀTiO 2 samples with various levels of gold doping are shown in Figure 2, together with the corresponding absorbance/extinction spec- trum. The results show incontrovertibly that the visible-light response of the device is due to the AuNP loading and that the effect is significant, in aggregate on par with the current compo- nent resulting from direct band gap excitation of the TiO 2 . The photoinduced conductances and the quantum efficiency (see supplementary Figure S1 in the Supporting Information) faith- fully track the (plasmonic) absorbance/extinction spectra of the corresponding AuNPÀTiO 2 film, with the largest responsivities occurring for wavelengths in the 500À700 nm range and maximizing at ∼620 nm, implying that the plasmonic excitation is directly responsible for the increased numbers of carriers in the conduction band of the TiO 2 . The efficiency (electron per incident photon) at 620 nm is ∼0.2%. In the absence of gold, the device shows negligible photoconductance when illuminated with wavelengths longer than ∼370 nm. The large currents observed with wavelengths shorter than 370 nm are due to direct interband carrier photoexcitation in TiO 2 . The conductance changes were found to increase monotoni- cally with increasing numbers of nanoparticle layers in the AuNPÀTiO 2 composite layer (and with its corresponding optical absorbance). For the device with four layers of gold nanoparticles, over 30% of its integrated photoresponse is due to Figure 1. Plasmonic photosensitization of TiO 2 . (a) Cross section scanning electron micrograph of the device used in this work, consisting of multistack layers of AuNPs of diameter ∼14 nm embedded in a TiO 2 layer of total thickness ∼200 nm. The various components of the device were artificially colored for clarity. Scale bar, 100 nm. (b) UV/visible spectra of TiO 2 film, AuNPs and AuNPsÀTiO 2 composite on a quartz substrate at normal incidence. Absorption spectra of AuNPs were multiplied by a factor of 3 for clarity. The absorbance corresponding to the bare quartz substrate was subtracted from each absorption spectrum. (c) Schematic illustrating the band-bending effect of the Schottky junction between Au NP and the TiO 2 layer surrounding it. Band bending creates an energy barrier which allows an electron excited by an incoming photon of energy pω from a filled to an empty level of the metal’s conduction band to tunnel directly into the conduction band (E CB ) of the TiO 2 . Figure 2. Photoconductance of TiO 2 and AuNPÀTiO 2 films. Mea- sured UVÀvisible (solid lines) and photocurrent spectra (closed symbols) of TiO 2 and AuNPs in TiO 2 for varying nanoparticle densities. The photocurrent, normalized for the intensity of the monochromatized light source as a function of wavelength, is shown in units of micro- amperes per watt. Right side: the corresponding cross sectional SEM image of the devices measured (artificially colored for clarity). All scale bars correspond to 100 nm. Figure 3. Effect of photon fluence. Photocurrent responsivity of two AuNPs in TiO 2 devices with varying Au nanoparticle loadings (TiO 2 with four layers of gold nanoparticles (solid red squares) and TiO 2 with three layers of gold nanoparticles (solid blue circles)) as a function of the irradiating light intensity at 600 nm. The measurements were carried out with 500 W xenon lamp sources. Inset shows that at low intensities the photocurrent is linear with fluence. 5550 dx.doi.org/10.1021/nl203457v |Nano Lett. 2011, 11, 5548–5552 Nano Letters LETTER the contribution from the visible region of the spectrum. How- ever, gold loading reduced the device’s photoresponse in the UV region likely due to the Schottky depletion of TiO 2 carriers at the metal/semiconductor junction, and possibly also due to the increased extinction by the gold nanoparticles in the UV region, and the (somewhat) reduced volume fraction of TiO 2 . The large photoconductance observed for the AuNPÀTiO 2 device when illuminated with visible light might result from one or more of the following processes: (1) photoexcitation of filled midgap donor states associated with TiO 2 defects; (2) formation of excitons in the TiO 2 by a two-photon process excited by resonant energy transfer from the Au plasmon; and (3) the production of electronÀhole pairs in the gold as one of the decay channels of the surface plasmon excitation followed by electron transfer to the TiO 2 . The negligible currents pro- duced by visible light illumination of TiO 2 (devoid of Au) indicate that the numbers of filled midgap states are very few, so the first process cannot be a major contributor. Measure- ments of the dependence of the photocurrent intensity on the fluence of the incident illumination (Figure 3) indicate a largely linear dependence of the photocurrent on light in- tensity in the low intensity regime, ruling out a two photon absorption mechanism. Some saturation is found to occur at the higher photon flux values. However, the photon flux threshold at which photocurrent saturation occurs is found to be independent of AuNP loading (Figure 3). Referring to Figure 1c, the process begins with the production of electron hole pairs in the conduction band of the metal by the decaying SP, with the excited electrons occupying normally empty states in the metal’s conduction band. A (significant) fraction of these excited electrons is transferred (by quantum tunneling) to the TiO 2 . This process reduces the negative charge already present on the gold as a result of Schottky junction formation and restores to the TiO 2 some (or all) of the depleted negative charge. In contrast, direct band gap excitation of the TiO 2 or the AuNPÀTiO 2 composite medium produces electron hole pairs wholly resident on the TiO 2 , which would leave the overall charge states of the TiO 2 and the Au unchanged. Saturation may result from the fact that a sufficient number of carriers (hot electrons) are generated following surface plasmon decay to neutralize many of the surface states which initially caused the band bending. These differences in charge disposition resulting from alter- nately illuminating of the AuNPÀTiO 2 device by visible and UV light should have distinguishable effects on the low frequency (i.e., dc) dielectric constants of the materials when measured under red (600 nm) as opposed to UV (330 nm) illumination. This proposition was explored by performing impedance mea- surements on the devices in the dark and under UV and red light illumination in the range 400 Hz to 100 kHz. Representative results are shown in Figure 4 as plots of the imaginary versus the real part of the measured impedances. The overall trends in the behavior of measured impedance values as a function of fre- quency are not unlike what had been previously reported for polycrystalline TiO 2 . 30 Such films which can have complex micro- and nanostructures due to defects and various structural and compositional inhomogeneities are often modeled as con- stant phase elements. 31,32 The resistive and capacitive contribu- tions to a constant phase element can be evaluated using expressions such as those reported in refs 32 and 33. This is summarized briefly in the Supporting Information. Indeed, the equivalent circuit shown inFigure 4c was found to fitourmeasured impedance values very satisfactorily over the entire frequency range used. The electrical materials properties derived from the impe- dance measurements on the most highly AuNP loaded device, are summarized in Table 1 (see supplementary Table 1 in the Supporting Information for more details). It is noteworthy that despite the parametric freedom of the fit, the values of the isolation capacitance C SiO 2 returned are almost constant for all of the devices studied, and under all illumination conditions. More- over, the value of the dielectric constant calculated from this capacitance (averaging ∼3.9) corresponds closely to the re- ported dc dielectric constant of silica. 34 Gratifyingly, the contact resistance is found to be rather small (∼90 Ω) and also fairly constant for all measurements. By contrast the resistance values reflecting the conductance of the TiO 2 and AuNP-loaded TiO 2 depend markedly on the presence or absence of the gold nanoparticles and on the wavelength of illumination. For the pristine TiO 2 the dark resistance is measured to be ∼11.5 kΩ. This resistance value is almost unchanged (∼11.0 kΩ) upon 600 nm illumination, Figure 4. Impedance spectroscopy. Spectral dependence on the com- plex impedance plots (Nyquist plots) obtained for (a) TiO 2 and (b) AuNPs-TiO 2 films. Both plots: open symbols experimental data and solid lines simulated data using the equivalent circuit shown. Several frequencies are labeled in the Nyquist plot. (c) The equivalent circuit used to simulate the impedance measurements as a function of frequency. Briefly it consists of a single resistor that accounts for all contact resistances in series with a constant phase device in parallel with a shunt resistor which represents the AuNPÀTiO 2 composite material (with or without Au loading), and finally, a series capacitor represents the effect of the silica layer that insulates the device from the bottom ITO electrode. Table 1. Calculated Dielectric Constants of TiO 2 and AuNPsÀTiO 2 Thin Films Derived from the Impedance Measurements TiO 2 AuNPs + TiO 2 dark 330 nm 600 nm dark 330 nm 600 nm R Contact (Ω) 87.3 83.7 87.0 96.6 97.0 97.1 R 1 (Ω) 11500 5115 11047 18500 14000 17101 ε SiO 2 3.92 3.98 3.90 3.98 4.02 3.86 effective ε TiO 2 90.3 315 89.3 82.2 84.7 59 5551 dx.doi.org/10.1021/nl203457v |Nano Lett. 2011, 11, 5548–5552 Nano Letters LETTER indicating (not unexpectedly) that TiO 2 shows no photoconduc- tive response to visible light. Illuminating the device with 330 nm light, however, reduces the device’sresistanceto5.1kΩ— a consequence of the fact that UV light induces direct band gap transitions. The increased conductance reflects the increased density of the newly formed electrons and holes. The behavior of the AuNP-loaded TiO 2 is markedly different. First, the dark resistance of the device is some 60% larger (18.5 kΩ) than for the device fabricated with Au-free TiO 2 —due almost certainly to the charge depletion at the Au/TiO 2 Schottky junction. Second, the decrease in this resistance value induced by illumination with 330 nm light is only ∼24%—half what it was for the Au-free material. This also follows from the premise that Schottky depletion significantly reduces the density of accessible electrons in the valence band of the TiO 2 . With 600 nm illumination this device behaves radically differently from its Au-free counterpart. The device resistance drops by 7.5% to 17.1 kΩ, corresponding to ∼30% of the drop produced by UV (330 nm) illumination. More significantly, the photoconduc- tance change integrated between 500 nm and 750 nm is ∼50% of the integrated response in the UV. In other words, with AuNP loading the device is only ∼50% less conductive with red light illumination as it is in the UV. The capacitance results are even more distinctive. The di- electric constant for the (gold-free) TiO 2 calculated from capacitance measurements carried out in the dark ε TiO 2 ,dark ≈ 90, corresponds well with what is expected for polycrystalline anatase TiO 2 . The value measured with 600 nm illumination (ε TiO 2 ,600 ≈ 89) is almost identical to the dark value. However, with 330 nm illumination the value of the dielectric constant increases dramatically (ε TiO 2 ,330 ≈ 315) reflecting the significant increase in electron population in the material’s conduction band and hole population in the valence band, rendering it significantly more polarizable. 35 Once again the AuNP-loaded TiO 2 behaves dramatically differently. The dark value of the dielectric constant is reduced to 82. This reflects two opposing contributions. Schottky deple- tion: which reduces the value of the material’s dielectric constant, and metal loading: which increases its value due to the higher dc polarizability of the metal over that of the semiconductor. Illumination with 330 nm light increases the material’s dielectric constant only very slightly (ε AuNP‑TiO 2 ,330 ≈ 84). By contrast, exposure of the device to 600 nm light produces a dramatic effect. The calculated dielectric constant decreases to ∼72% of its dark value (from 82 to 59). This large decrease is consistent with the tentative proposition that at 600 nm, a significant fraction of electrons excited as electron hole pairs in the metal, cross from the metal into the TiO 2 thereby reducing the Au NP’s polariz- ability (and also the effective dc dielectric constant of the metal). As a result the decrease in the composite medium’s dielectric constant induced by Schottky depletion is no longer offset by the polarizability of the metal to the same extent as it was in the absence of illumination, and the composite’s dielectric constant should markedly decrease, as observed. CurrentÀvoltage measurements (see supplementary Figure S2 in the Supporting Information) are consistent with the above interpretation. The I(V) curves show almost Ohmic behavior when the device is in the dark or illuminated with 600 nm light. But when illuminated with UV (330 nm) it shows the I(V) characteristics of back-to-back diodes reflecting the changes in the effective location of the band edges with respect to the Fermi energies of the Ti/Au contact pads. In summary, devices fabricated by embedding Au nanoparti- cles in TiO 2 show significant additional photoconductances (∼30%) when illuminated by light with photon energies well below the band gap. The photoconductance is found to track the plasmonic absorption/extinction spectrum of the AuNPs faith- fully. This impressive change in photoconductance is ascribed both to quantum tunneling of hot electrons from the metal directly into the conduction band of the TiO 2 for those electrons with energies lower than the 0.9 eV needed to overcome the barrier and to energetic electrons going over the barrier trans- port. All of these electrons originate as electron hole pairs in the gold NPs produced by plasmonic excitation and decay. Device impedance measurements carried out in the dark and under illumination with UV and red wavelengths reinforce this me- chanism. Substantial improvement in performance can be ex- pected by further improving the light harvesting design. This could include (i) increased gold nanoparticle loading (it is particularly intriguing to consider what would happen if the nanoparticle density is high enough to produce electromagnetic hot spots); (ii) engineering a device architecture that increases the interfacial area between plasmonic nanostructure and a wide band gap semiconductor, for example, by using arrays of gold nanorods (absorption of which can be tuned to cover the entire solar spectrum); (iii) improving the crystallinity of the semicon- ductor; and (iv) selecting oxides with better electron mobility. ’ ASSOCIATED CONTENT b S Supporting Information. Detailed experimental meth- ods, calculation of quantum efficiency, capacitance and dielectric constants, detailed impedance measurements results, XRD dif- fraction patterns, schematic of the photocurrent measurement apparatus, and transient photocurrent measurement results. This material is available free of charge via the Internet at http://pubs. acs.org ’ AUTHOR INFORMATION Corresponding Author *E-mail: moskovits@chem.ucsb.edu. ’ ACKNOWLEDGMENT The authors thank Ashok Ramu for photolithography support, Namhoon Kim for technical support, and Peter Allen for graphic support. This work was supported by the Institute for Colla- borative Biotechnologies through Grant DAAD19-03-D-0004 from the U.S. Army Research Office and made extensive use of the MRL Central Facilities at UCSB supported by the National Science Foundation under award nos. DMR-0080034 and DMR- 0216466 for the HRTEM/STEM microscopy. We also gratefully acknowledge research support from the Institute for Energy Efficiency, an Energy Frontier Research Center funded by the U. S. Department of Energy, Office of Science, Office of Basic Energy Sciences, Award Number DE-SC0001009. ’ REFERENCES (1) O’Regan, B.; Gratzel, M. Nature 1991, 353, 737–740. (2) Bach, U.; Lupo., D.; Comte, P.; Moser, J. E.; Weissortel, F.; Salbeck, J.; Spreitzer, H.; Graetzel, M. Nature 1998, 395, 583–585. (3) Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. Science 2001, 293, 269–271. 5552 dx.doi.org/10.1021/nl203457v |Nano Lett. 2011, 11, 5548–5552 Nano Letters LETTER (4) Umebayashi, T.; Yamaki, T.; Itoh, H.; Asai, K. Appl. Phys. Lett. 2002, 81, 454–456. (5) Corma, A.; Atienzar, P.; Garcia, H.; Chane-Ching, J.-Y. Nat. Mater. 2004, 3, 394–397. (6) Hardin, B. E.; Hoke, E. T.; Armstrong, P. B.; Yum, J. H.; Comte, P.; Torres, T.; Frechet, J. M.; Nazeerudding, M. K.; Graetzel, M.; McGehee, M. D. Nat. Photonics 2009, 3, 406–411. (7) Chen, X.; Liu, L.; Yu, P. Y.; Mao, S. S. Science 2011, 331, 746– 750. (8) Woodhouse, M.; Parkinson, B. A. Chem. Soc. Rev. 2009, 38, 197– 210. (9) Tian, Y.; Tatsuma, T. J. Am. Chem. Soc. 2005, 127, 7632–7637. (10) Furube, A.; Du, L.; Hara, K.; Katoh, R.; Tachiya, M. J. Am. Chem. Soc. 2007, 129, 14852–14853. (11) Falk, A. L.; Koppens, F. H. L.; Yu, C. L.; Kang, K.; Snapp, N. L.; Akimov, A. V.; Jo, M.; Lukin, M.; Park, H. Nature Phys. 2009, 5, 475–479. (12) Yu, J.; Dai, G.; Huang, B. J. Phys. Chem. C 2009, 113, 16394– 16401. (13) Gather, M. C.; Meerholz, K.; Danz, N.; Leosson, K. Nat. Photonics 2010, 4, 457–461. (14) Nishijima, Y.; Ueno, K.; Yokota, Y.; Murakoshis, K.; Misawa, H. J. Phys. Chem. Lett. 2010, 1, 2031–2036. (15) Valverde-Aguilar, G.; Garcia-Macedo, J. A.; Renteria-Tapia, V.; Aguilar-FrancO, M. Appl. Phys. A: Mater. Sci. Process. 2011, 103, 659 – 663. (16) Brongersma, M. L. Nat. Photonics 2008, 2, 270–272. (17) Atwater, H. A.; Polman, A. Nat. Mater. 2010, 9, 205–213. (18) Konstantatos, G.; Sargent, E. H. Nat. Nanotechnol. 2010, 5, 391–400. (19) Schuller, J. A.; Barnard, E. S.; Cai, W.; Jun, Y. C.; White, J. S.; Brongersma, M. L. Nat. Mater. 2010, 9, 193–204. (20) Novotny, L.; van Hulst, N. Nat. Photonics 2011, 5,83–90. (21) Anker, J. N.; Hall, W. P.; Lyandres, O.; Shah, N. C.; Zhao, J.; Van Duyne, R. P. Nat. Mater. 2008, 7, 442–453. (22) Borensztein, Y.; Delannoy, L.; Djedidi, A.; Barrera, R. G.; Louis, C. J. Phys. Chem. C 2010, 114, 9008–9021. (23) O’Regan, B.; Moser, J.; Anderson, M.; Graetzel, M. J. Phys. Chem. 1990, 94, 8720–8726. (24) Boschloo, G. K.; Goossens, A.; Schoonman, J. J. Electrochem. Soc. 1997, 144, 1311–1317. (25) McFarland, E. W.; Tang, J. Nature 2003, 421, 616–618. (26) Park, J. Y.; Lee, H.; Renzas, J. R.; Zhang, Y.; Samorjai, G. A. Nano Lett. 2008, 8, 2388–2392. (27) Gong, X. Q.; Selloni, A.; Batzill, M.; Diebold, U. Nat. Mater. 2006 , 5, 665–670. (28) Thompson, T. L.; Yates, J. T. Chem. Rev. 2006, 106, 4428–4453. (29) Knight, M. W.; Sobhani, H.; Nordlander, P.; Halas, N. J. Science 2011, 332, 702–704. (30) Jung, H. S.; Lee, J. K.; Lee, J.; Kang, B. S.; Jia, Q.; Nastasi, M.; Noh, J. H.; Cho, C. M.; Yoon, S. H. Langmuir 2008, 24, 2695–2698. (31) Ito, S.; Zakeeruddin, S. M.; Comte, P.; Liska, P.; Kuang, D.; Graetzel, M. Nat. Photonics 2008, 2, 693–698. (32) Sanchez, M.; Rincon, M. E.; Guirado-Lopez, R. A. J. Phys. Chem. C 2009, 113, 21635–21641. (33) Brug, G. J.; van den Eeden, A. L. G.; Sluyters-Rehbach, M.; Sluyters, J. H. J. Electroanal. Chem. Interfacial Electrochem. 1984, 176, 275–295. (34) Kingon, A. I.; Maria, J.-P.; Streiffer, S. K. Nature 2000, 406, 1032–1038. (35) Spagnol, V.; Cachet, H.; Baroux, B.; Sutter, E. J. Phys. Chem. C 2009, 113, 3793–3799. . 97 .1 R 1 (Ω) 11 500 511 5 11 047 18 500 14 000 17 1 01 ε SiO 2 3.92 3.98 3.90 3.98 4.02 3.86 effective ε TiO 2 90.3 315 89.3 82.2 84.7 59 55 51 dx.doi.org /10 .10 21/ nl203457v. Published: October 31, 2 011 r 2 011 American Chemical Society 5548 dx.doi.org /10 .10 21/ nl203457v | Nano Lett. 2 011 , 11 , 5548–5552 LETTER pubs.acs.org/NanoLett