Letter pubs.acs.org/NanoLett
Ultrafast Modulation of the Plasma Frequency of Vertically Aligned Indium Tin Oxide Rods Daniel B. Tice,† Shi-Qiang Li,‡ Mario Tagliazucchi,† D. Bruce Buchholz,‡ Emily A. Weiss,† and Robert P. H. Chang*,‡ †
Department of Chemistry, Northwestern University, 2145 Sheridan Rd., Evanston, Illinois 60208-3113, United States Department of Materials Science and Engineering, Northwestern University, 2220 Campus Dr., Evanston, Illinois 60208-3108, United States
‡
S Supporting Information *
ABSTRACT: Light−matter interaction at the nanoscale is of particular interest for future photonic integrated circuits and devices with applications ranging from communication to sensing and imaging. In this Letter a combination of transient absorption (TA) and the use of third harmonic generation as a probe (THG-probe) has been adopted to investigate the response of the localized surface plasmon resonances (LSPRs) of vertically aligned indium tin oxide rods (ITORs) upon ultraviolet light (UV) excitation. TA experiments, which are sensitive to the extinction of the LSPR, show a fluence-dependent increase in the frequency and intensity of the LSPR. The THG-probe experiments show a fluence-dependent decrease of the LSPR-enhanced local electric field intensity within the rod, consistent with a shift of the LSPR to higher frequency. The kinetics from both TA and THG-probe experiments are found to be independent of the fluence of the pump. These results indicate that UV excitation modulates the plasma frequency of ITO on the ultrafast time scale by the injection of electrons into, and their subsequent decay from, the conduction band of the rods. Increases to the electron concentration in the conduction band of ∼13% were achieved in these experiments. Computer simulation and modeling have been used throughout the investigation to guide the design of the experiments and to map the electric field distribution around the rods for interpreting far-field measurement results. KEYWORDS: Localized surface plasmons, doped semiconductors, transient absorption, third harmonic generation, ultrafast optical modulation, carrier injection
T
components for telecommunications and molecular sensing. This Letter describes the study of ultrafast light modulation of vertically aligned indium tin oxide rods (ITORs) and addresses two important research questions: (i) What is the response of an ITOR when an ultrafast modulation occurs to its plasma frequency, ωp, by photoexcitation of electrons into (and subsequent relaxation of electrons from) its conduction band? and (ii) What is the spatial distribution of the electric field intensity within an ITOR as a result of the excitation? It is important to note that in this research the focus is on the response of individual ITOR to the radiation, and thus the
he interaction of light with nanostructured materials is of great interest due to the existence of rich phenomena that have very broad applications in energy,1 the environment,2 health,3 and security.4,5 In recent years, there have been extensive studies to probe the response of metal and semiconductor nanoparticles and their clusters to light.6 These studies have provided information on energy exchanges, excitations, radiation, and loss mechanisms of complex nanosystems. Fundamental understanding of the interaction between light and nanostructures will enable the fabrication of new devices with improved performances. The large oscillator strength and tunability of the surface-plasmon resonances of degenerately doped semiconductors, in particular, across the near- and mid-infrared (NIR and MIR) wavelengths7−11 make them potentially useful as active linear and nonlinear optical © 2014 American Chemical Society
Received: July 27, 2013 Revised: December 28, 2013 Published: February 14, 2014 1120
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
demonstration of an increase in ωp upon excitation of an interband transition in semiconductor-based plasmonic materials. Changes to ωp were not the dominant effect in previous TA experiments on the interband transition of copper chalcogenide nanocrystals.14,24 Experimental Procedure. ITOR Growth. While the concept of light interaction with a single ITOR is simple, the design for an experiment is more involved. To study how an individual ITOR will respond to light, we found after many experiments that the best approach is to grow periodic arrays of ITORs in order to produce near identical rods. We prepared the ITOR samples using a previously reported procedure.8 Briefly, we patterned square lattice arrays of gold nanoparticles on 30 nm thick ITO or In2O3 films grown epitaxially on yttriastabilized zirconia (YSZ) ⟨100⟩ substrates with pulsed laser deposition (PLD). We placed the substrates in a tube furnace at 800 °C, with the boats containing our metal precursors held at 900 °C upstream of the substrates. We used In mesh (100 mesh, 99.99% trace metal basis, Sigma-Aldrich) and SnO2 powder (99.99% trace metal basis, Sigma-Aldrich) in a 9:1 molar ratio and introduced a 25 sccm flow rate of 1% O2 in a N2 buffer gas during the growth process to provide the oxygen source. The gold nanoparticles serve as catalysts for vapor− liquid−solid (VLS) growth of vertically aligned ITORs from the substrate with a square cross section, consistent with the bixbyite crystal lattice of ITO [see Supporting Information (SI1) for processing details).25 We have grown a range of tapered vertical rods, both near periodic and periodic arrays (see Figure 1a,b).
experiments are designed accordingly with the aid of simulation. Unlike plasmons in metals, the energies of the resonances in doped semiconductors can be easily tuned through their elemental composition, which can be modified during synthesis7,9,10 or postsynthesis annealing.8,12−15 Additionally, the free carrier concentrations in doped semiconductors are much lower than in metals, so the plasmonic resonances of these materials are more sensitive than metals to absolute changes in the carrier concentration.16 The ITORs in this study have a carrier concentration on the order of 1020 e−/cm3, which is a factor of 100 lower than the carrier concentration of gold.17 This enhanced sensitivity to the carrier density makes doped semiconductors appealing as active optical components, which require reversible modulation of either the intensity or the frequency of the plasmon resonance and ultrafast response time for fast switching. While several groups have reported shifts in both the intensity and the frequency of the local surface plasmon resonance (LSPR), ωLSPR, from chemical doping,18 photocharging,19 or electrochemical doping of semiconductor nanocrystals,16,20,21 these methods are nonideal for active modulation of optical components because (i) it is uncertain whether the additional carriers lie in the core or surface states in some cases,20,21 (ii) shifts in ωLSPR through chemical treatments require oxidizing or reducing environments, and (iii) electrochemical methods, while reversible, are slow, are sensitive to nanostructure size, and require both electrolyte and electrodes, which limits potential applications. Recently, Krasvin and Zayats have introduced a novel concept of electro-optical fieldeffect nanoplasmonic modulator for nanoscale integration.22 In this work, we photoinject charge carriers into the conduction band of ITO upon direct photoexcitation at energies near its bandgap (∼3.6 eV); this excitation increases the carrier density in the conduction band and thereby increases ωp of the ITORs. We perform two different types of ultrafast transient pump−probe spectroscopy experiments on the ITORs to characterize the modulation of ωp upon carrier injection: 1. Using UV-pump and NIR-probe transient absorption (NIR-TA), we observe a shift in ωLSPR to higher energy and an increase in the intensity of the LSPR, consistent with an increase in ωp. The bandwidth at which we can modulate ωp is limited by the time scale of electronic relaxation of free carriers in the ITORs, which is on the order of one picosecond. 2. In a separate experiment, we pump the system at 325 nm and monitor the transient intensity of LSPR-mediated third-harmonic generation (THG) from the ITORs as a function of the time delay between the UV pump and the THG probe. We refer to this experiment as “THG-probe”. The NIR TA experiments are sensitive to changes in the extinction of the sample, which are not necessarily plasmonic in origin. The THG-probe is sensitive to the distribution of the local electric field within the ITORs,23 a distinctly plasmonic phenomenon that is dictated by detuning of the THG-probe fundamental relative to ωLSPR. At the chosen wavelength of our THG probe, we observe a decrease of the plasmon-induced enhancement of the electric field within the rod upon UV excitation. This effect is consistent with a shift of ωLSPR to higher energy which further verifies that the UV excitation of ITO enables dynamic modulation of ωp through carrier injection and decay. This work represents the first time-resolved study of plasmon dynamics in crystalline ITORs and the first study of LSPRenhanced THG in ITORs. Additionally, it is the first
Figure 1. Scanning electron micrographs of ITORs from the VLS growth process. (A) Near periodic distributed arrays with average spacing between rods of 0.6 μm and a height of 4 μm. (B) Ordered array of 10.8 μm tall tapered rods with spacing between rods of 1.6 μm.
Pitch and Rod Height Selection. Once periodicity is introduced into the array, it is then necessary to select a rodspacing (or pitch) to ensure that collective effects, such as mode coupling between LSPR resonances and photonic grating modes, are not present or minimized. To avoid the coupling of the LSPR fields between the rods, we utilized ITOR pitches >400 nm, which is much larger than the maximum decay distance of the surface fields. The effect of the distance between rods on their electrodynamics coupling is discussed in the Supporting Information section SI-2. In the design of our experiments we have chosen pitches and rod heights such that the phase space we work in has no observable mode couplings, and this is further confirmed by our simulation. We selected two different sets of samples for our experiments. For the LSPR modulation experiment (NIR-TA), we chose an ITOR array with 600 nm spacing to optimize detection with minimum broadening of the peak line width. On the other hand, for THG probe experiments, we have simulated an ITOR array with different spacings (800 nm, 1.2 μm, 1.6 μm, 2.4 μm), and found that the array with 1.6 μm spacing gave the maximum THG signal (see Figure SI-3.1.1 in Supporting Information SI-3.1). 1121
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
The simulation results compare well with our experimental measurements (see Figure SI-3.1.2). Rod Shape. The shape of the ITOR also contributes to our study especially for the THG-probe experiments (see Supporting Information SI-3.2). To observe THG in the rods, it is critical that the skin depth of the ITOR is comparable to the wavelength under study. Figure SI-3.2 in the Supporting Information compares the near-field distribution for rods of different shape and shows that, by having a tapered (along the length of the axis) rod, the condition of having a thickness much smaller than the wavelength of the incident light can be met easier in our experiments. Experimental Setup. To excite only one type of LSPR, our experiment is set up such that the incident Poynting vector is parallel to the long axis of the ITORs. With this geometry, only LSPR modes across the short axis of the rods (transverse modes) will be excited (i.e., longitudinal LSPR along the long axis of the rods will not be excited). Operating within this selected phase space per our design, we present here results from two samples in this study: for TA experiments we used a sample with a ITOR spacing of 600 nm, a height of 4 ± 0.2 μm, and sides of 265 ± 25 nm, while for the THG-probe experiments we used ITORs with a spacing of 1.6 μm and a height of 10.8 ± 0.1 μm. We provide additional experiments for ITOR arrays of different geometric parameters in the Supporting Information (SI) to demonstrate reproducibility. We prepared the sample for the TA experiments so the wavelength of the LSPR was within the spectral range accessible by our NIR-TA setup (800−1630 nm), while we prepared the sample for the THG-probe experiments so that the wavelength of the LSPR was resonant with the idler wavelength (2050 nm) of our optical parametric amplifier (OPA) when we used the signal wavelength (1300 nm) to generate the UV pump (see Figure 2 for the schematic diagrams of the instrument setup for each type of measurement). Ground State Extinction Spectroscopy of the LSPR of ITORs. To simulate the extinction spectrum of the ITOR array, we follow the same procedure used in earlier publications8,12,26 by calculating the frequency-dependent dielectric of the material with the Drude equation eq 1, ε(ω) = ε∞ −
Figure 2. (A) Optical setup for TA measurements and (B) optical setup for THG-probe measurements. The ITOR sample is first excited by a UV-pump pulse that promotes electrons from the valence band into the conduction band. In the NIR TA experiment (A), the changes in sample extinction induced by the pump beam in the LSPR region (1430−1630 nm) are probed with a NIR continuum using active background subtraction. In the THG-probe experiment (B), the probe beam is 2050 nm (the idler of our OPA), which generates an output pulse at 683 nm from the sample due to third harmonic generation. We also measure the THG-probe signal using active background subtraction.
ωp2 ω 2 + i Γω
(1)
where ε∞ is the high-frequency dielectric constant of the host medium (∼3.9 for ITO),8,9 ωp is the plasma frequency of the material, and Γ is its damping parameter. Using DDSCAT 7.3,27,28 a discrete dipole approximation (DDA) program (see SI-4), we varied ωp and Γ in eq 1 to fit the wavelength and line width of the LSPR, respectively. The best fit of the extinction spectrum corresponds to a value for ωp of 1.59 eV and Γ of 80 meV. We can relate ωp (using eq 2) to a carrier concentration of the ITO, n, of 7.35 × 1020 e−/cm3,
Figure 3. NIR extinction spectrum of the sample of ITO nanorods in Figure 1A (black) and the extinction spectrum calculated using DDA with the Drude equation (dashed teal trace) for a ωp = 1.59 eV and Γ = 80 meV. We have corrected the extinction spectrum to account for the underlying substrate and ITO thin film.
2
ωp =
ne ε0m*
(2) 29
which is in good agreement with values of n for ITO. In eq 2, e is the formal charge of an electron, ε0 is the permittivity of free space, and m* is the effective carrier mass in the medium of choice (0.4me for electrons in ITO, where me is the electron rest mass).8 Using these results, we simulated the extinction spectrum (see Figure 3) of a sample with vertically aligned ITORs as shown in Figure 1A. We see that the transverse-mode
LSPR of the ITORs comprises two convoluted peaksat 1770 and 2100 nm for this sampledue to the square cross-section of the ITORs, consistent with previous measurements.8 Transient Absorption Spectroscopy of the LSPR of ITORs upon UV Excitation. To study the effect of the photoinjection 1122
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
of carriers into the conduction band of the ITORs on ωp, we monitored the LSPR with a transient absorption system (see Figure 2A) that has been described in detail previously.30,31 Briefly, to generate the pump wavelengths, we used two barium borate crystals to upconvert the NIR signal wavelengths from our optical parametric amplifier (OPA) (TOPAS-C, Light Conversion) to UV wavelengths (300−350 nm). To generate a continuum probe for TA measurements, we focused a portion of the NIR signal output from the OPA to generate a NIR continuum in a 1 cm yttrium alumina garnate (YAG) crystal. We positioned the ITOR samples normal to the pump and probe beams, so that the Poynting vector of the light is along the long axes of the ITORs. This sample geometry results in selective excitation of the transverse mode LSPR of the ITORs.8 The polarization of the pump and the probe beams were parallel to each other for these experiments. Figure 4A shows the transient spectra of the ITORs with a spacing of 600 nm for probe wavelengths between 1430 and
1630 nm following photoexcitation of the ITORs at 300 nm which excites electrons in the valence band across the bandgapas a function of the fluence of the pump. For all fluences, we observe a photoinduced absorption that grows in intensity and shifts monotonically to higher energy with increasing pump fluence, with a peak clearly visible for higher fluences. We note that, since TA measurements report differential absorptions, this photoinduced feature corresponds to an increase in both the frequency and the oscillator strength of the LSPR, relative to LSPR of the ITORs in their ground state upon UV excitation. The measured absorbance of the sample at the excitation wavelength (300 nm) is 1.3 A.U.; from this value, we calculate a range of expectation values of 5.3 × 105 to 2.2 × 107 photons absorbed per rod on going from the lowest to highest fluences of the pump beam. If we assume each absorbed photon promotes a single carrier to the conduction band of the ITORs, then the 300 nm excitation increases the carrier concentration in the conduction band by 0.5% at the lowest excitation fluence, and by ∼13% at the higher excitation fluence, after accounting for the saturable absorption of the ITORs. The observation of a carrier concentration-dependent energy and intensity of ωLSPR agrees with eq 2 and is evidence that the photoinjected carriers are delocalized in the conduction band of the rod, as opposed to populating surface states.16 Figure 4B shows normalized kinetic traces, extracted from the TA spectrum in Figure 4A at 1570 nm as a function of increasing fluence of the 300 nm pump. The lifetime of the photoinjected carriers is insensitive to the fluence of the pump; the lack of change in the kinetics indicates that no new relaxation processes are introduced at higher pump fluences, which is consistent with our assignment of the photoinduced absorption to carrier injection into the conduction band of the ITO. The electrons promoted into the conduction band upon UV excitation are delocalized and behave collectively with the steady-state population of free electrons present in the conduction band from doping; we effectively “photodope” the ITO and achieve an increase in ωp of 100 meV with a single picosecond response time. We note that, as our excitation wavelength is similar to the bandgap of ITO, the excess energy of the injected carriers is minimized and we do not expect carrier-cooling dynamics to contribute to our measurements of the LSPR. As a control, we performed NIR-pump TA experiments to monitor hot electron-cooling lifetimes and observe nearly instrument response limited dynamics (180 fs for this NIR pump-NIR probe measurement (see SI-5). LSPR-Enhanced Third Harmonic Generation as a Probe of the Electric Field Distribution within ITORs. We have demonstrated that both ωLSPR and the oscillator strength of the LSPR increase upon carrier injection, but these TA experiments are sensitive only to the extinction of a transition, and are unable to probe the local electric field distribution of the rod, a primary characteristic of a plasmonic material.6,32 The probability of nonlinear optical processes, in contrast, depends not only on the intensity,6 but also the spatial distribution of the local electric field of a plasmonic excitationthat is, whether the induced field is inside versus outside of the nonlinear medium.33 Here, we use LSPR-enhanced THG from the ITORs to probe the local field distribution of the plasmon as a function of the excitation wavelength relative to ωLSPR. For plasmonic nanostructures of subwavelength dimensions, the efficiency of THG is dominated by the local electric field at a given excitation wavelength6,23 rather than the phase-matching
Figure 4. (A) Plot of the NIR transient spectra as a function of the fluence of a pump laser with a wavelength of 300 nm (light red to dark red traces with increasing fluence), for the ITORs from Figure 1A. All spectra are shown at time-zero of the temporal overlap of the pump and probe beams. The induced absorption increases in amplitude and shifts to higher energy with increasing fluence of the pump. Inset: Normalized transient spectra clearly show the spectral shift as a function of the fluence of the pump. The spectral cutoff at 1630 nm is due to a loss of sensitivity of our detector at wavelengths longer than 1630 nm. (B) Plot of the normalized kinetic traces extracted from the transient spectra in A at 1570 nm, for different fluences of the pump. The lifetime of the induced absorptions, which we fit with a single exponential time constant of 1.1 ps convoluted with our instrument response function of 200 fs (light blue dashed trace), does not depend on the fluence. 1123
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
Figure 5. (A) Plot of ηTHG for a given excitation wavelength (blue circles and lines) for the ITORs shown in Figure 1B, plotted with their ground state extinction spectrum (black trace), and their calculated extinction spectrum (dashed blue trace), which corresponds to ωp = 1.43 eV and Γ = 96 meV. We measured ηTHG in an integration sphere at a fluence of 790 μJ/cm2 for all wavelengths; the error bars come from three separate measurements at different locations on the sample. (B) Calculated local field intensity (|E|/|E0|) for the ITORs at the wavelengths (I−III) indicated in A. At I, where there is no significant extinction by the LSPR, there is minimal field enhancement in the volume of the rod. At II, extinction by the LSPR leads to increased efficiency of THG because the local field is enhanced inside the volume of the rod. At III, equal extinction by the LSPR does not enhance THG because the local field is outside the volume of the rod. For the near-field simulations, we propagated the field along the long axis of the rod (from the top to the bottom of the figure), with the polarization perpendicular to the long axis, and limited the scale of the intensity to enhance image contrast (the tips of the rod show higher field enhancements). Figure SI-3.2 in the Supporting Information shows |E|/|E0| profiles along the axial and transverse directions.
Figure 6. (A) Plot showing the wavelengths of the TA probe at 1630 nm (gray dashed line) and the fundamental of the THG-probe at 2050 nm (dashed teal line) plotted against the extinction of the LSPR (black trace) and ηTHG (blue circles and lines) from Figure 5A. (B) Normalized kinetics of the rod sample upon excitation with a 325 nm pump for the TA experiment, taken at 1630 nm (black trace), and the THG output for the fundamental probe centered at 2050 nm (teal trace), recorded at 695 nm. We plot the THG as −log(ΔηTHG) for ease of comparison to the TA experiment. We used a fluence of 12 mJ/cm2 for the 2050 nm fundamental of the THG-probe. The 325 nm pump had a fluence of 720 μJ/cm2 for the THG-probe experiment.
exactly track the extinction of the LSPR; we observe appreciable THG only at excitation wavelengths resonant with the high-energy side of the LSPR. Our DDA simulations of the ITORs explain this behavior: for wavelengths of light offresonance with the LSPR (Figure 5B, I), the rod behaves as a conventional dielectric, and, as expected, we do not measure appreciable THG. When the light is of a wavelength on resonance with the higher-energy side of the LSPR (Figure 5B, II), the near-field intensity of the light incident to the ITORs is enhanced not only on the surface of the rod, but also inside the volume of the rod, where ITO serves as the nonlinear medium, and we observe appreciable THG. For wavelengths resonant
considerations that dominate the efficiency for bulk, transparent media.6 The efficiency of the THG process, ηTHG, for the ITORs therefore increases at the excitation wavelengths where the LSPR-enhanced local electric field is confined within the volume of the rod, which increases the effective electric field the rod experiences.23 Figure 5A shows ηTHG for the ITORs shown in Figure 1B (spacing = 1.6 μm), for excitation at a series of wavelengths that span the LSPR spectrum of the ITORs. In this plot, ηTHG = powerTHG/powerfundamental (blue circles and lines). For reference, Figure 5A also shows the extinction spectrum of the LSPR (black trace). The THG efficiency spectrum does not 1124
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
lifetime of ΔηTHG matches that of the photoinduced absorption measured by NIR TA. This result confirms that the LSPRmediated local field distribution within the rod for a given excitation wavelength is modified upon UV excitation and that the decay of the photoinjected carriers from the conduction band of the ITO is responsible for the transient behavior of both the energy and the intensity of the plasmon resonance and the efficiency of the THG process. Ultrafast methods of altering the optical properties in materials are desirable for applications in active plasmonics and switching.6,32 Here, we have demonstrated (i) the first time-resolved studies of plasmon dynamics in ITO nanostructures, (ii) LSPR-enhanced THG from ITO rods, which we use as a probe of the local electric field, and (iii) the first demonstration of the modulation of the linear and nonlinear optical properties of the LSPR of a doped semiconductor through an increased ωp upon direct excitation of an interband transition.
with the lower-energy side of the LSPR, the extinction of the light by the rod is still strong, but the near-field intensity lies mainly outside of the rod. Due to the negligible nonlinear susceptibility of air, we observe no THG at these wavelengths. The behavior observed for light wavelengths resonant with the higher-energy side of the LSPR is a consequence of the NIR wavelengths of the LSPR; while resulting in a larger skin depth and enabling the field to penetrate into the rod, longer wavelengths also result in weaker surface fields compared to visible wavelength LSPRs in gold nanoparticles. The weaker surface field enhances the sensitivity of THG in ITORs to distribution of the field in the volume of the ITO. The Supporting Information (SI-6) contains THG experiments on another ITOR sample, as well as simulations using a finitedifference time domain method, (see SI-7). We performed a pump−probe experiment where we tune the pump to 325 nm to excite the electrons in ITORs across the bandgap, tune the probe to 2050 nm, and detect the THG wavelength at 695 nm. We detect at 695 nm instead of the “true” THG wavelength, 683 nm, due in part to the 200 cm−1 bandwidth of our laser and the Maker’s Fringing of the transmitted THG-probe, and to a probe-fluence dependent shift in the THG output wavelength at the probe fluences needed for the THG-probe experiment. We chose 2050 nm for the probe because it produces a near-maximum THG intensity, given by THG efficiency ηTHG, as shown in Figure 6A. Upon UV excitation, ηTHG decreases (Figure 6B, teal trace), which is what we predict for an increase of ωp because population of the conduction band with excess carriers shifts the LSPR and the field intensity spectrum to higher energy, and thereby lowers the efficiency of our THG-probe at 2050 nm. Note that 2050 nm is the wavelength that optimizes THG for the ground state, the wavelength of optimal THG efficiency will be shorter than 2050 nm for the excited state. In our experiments we fixed the THG probe wavelength to 2050 nm, and therefore, we observe a decrease in the THG efficiency upon UV photoexcitation of the ITORs. We additionally performed NIR TA with a 325 nm pump and a 1630 nm probe (Figure 6A, dashed gray vertical line), to measure the expected shift of the LSPR for this sample. We observe that upon UV excitation, the extinction at 1630 nm increases (Figure 6B black trace), which is consistent with our expected shift of the LSPR to higher energy. Figure 6B also shows that the kinetics of the TA and THG-probe experiments share the same lifetime, which indicates that the same mechanismthe modulation of ωp upon injection of electrons into and their subsequent decay from the conduction band of the ITORsis responsible for the processes we observe here. Conclusions. We have demonstrated the ultrafast dynamic modulation of ωp in ITORs upon photoinjection of electrons into the conduction band of ITO and mapped the field distribution of the plasmon resonance using pump−probe spectroscopies. In TA experiments with a UV pump and NIR probe, where we monitor the extinction of the LSPR, we observe a photoinduced absorption that increases in oscillator strength and energy with increasing pump fluence, which is consistent with the changes in LSPR expected for an increase of ωp. The lifetime of this process is 1.1 ps and is independent of the fluence of the excitation. We demonstrate LSPR-enhanced THG from the ITORs and show that the THG efficiency, ηTHG, at a given excitation wavelength of the LSPR corresponds to the local electric field distribution within the rods. Upon UV excitation of the rods, ηTHG at our probe wavelength decreases, which is consistent with an increase of ωLSPR, and that the
■
ASSOCIATED CONTENT
S Supporting Information *
Additional supporting data. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This research was funded under NSF-IGERT program (DGE0801685) and in part from NSF-DMR-1121262 and 0843962. The FDTD simulation was performed with QUEST computational resources located in Northwestern University under proposal p20447. Patterning of the gold nanoparticle array was performed with JEOL-9300 in the Center for Nanoscale Materials at Argonne National Laboratory (CNM 30831). Use of the Center for Nanoscale Materials was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.
■
REFERENCES
(1) Kamat, P. V. J. Phys. Chem. C 2007, 111 (7), 2834−2860. (2) Vaseashta, A.; Vaclavikova, M.; Vaseashta, S.; Gallios, G.; Roy, P.; Pummakarnchana, O. Sci. Tech. Adv. Mater. 2007, 8 (1), 47−59. (3) Huang, X.; Jain, P. K.; El-Sayed, I. H.; El-Sayed, M. A. Laser Med. Sci. 2008, 23 (3), 217−228. (4) Kirchain, R.; Kimerling, L. Nat. Photonics 2007, 1 (6), 303−305. (5) Lal, S.; Link, S.; Halas, N. J. Nat. Photonics 2007, 1 (11), 641− 648. (6) Kauranen, M.; Zayats, A. V. Nat. Photonics 2012, 6 (11), 737− 748. (7) Luther, J. M.; Jain, P. K.; Ewers, T.; Alivisatos, A. P. Nat. Mater. 2011, 10 (5), 361−366. (8) Li, S. Q.; Guo, P.; Zhang, L.; Zhou, W.; Odom, T. W.; Seideman, T.; Ketterson, J. B.; Chang, R. P. H. ACS Nano 2011, 5 (11), 9161− 9170. (9) Mendelsberg, R. J.; Garcia, G.; Li, H.; Manna, L.; Milliron, D. J. J. Phys. Chem. C 2012, 116 (22), 12226−12231. (10) Buonsanti, R.; Llordes, A.; Aloni, S.; Helms, B. A.; Milliron, D. J. Nano Lett. 2011, 11 (11), 4706−4710. (11) Rowe, D. J.; Jeong, J. S.; Mkhoyan, K. A.; Kortshagen, U. R. Nano Lett. 2013, 13 (3), 1317−1322. 1125
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
(12) Manthiram, K.; Alivisatos, A. P. J. Am. Chem. Soc. 2012, 134 (9), 3995−3998. (13) Scotognella, F.; Della Valle, G.; Srimath Kandada, A. R.; Dorfs, D.; Zavelani-Rossi, M.; Conforti, M.; Miszta, K.; Comin, A.; Korobchevskaya, K.; Lanzani, G. Nano Lett. 2011, 11 (11), 4711− 4717. (14) Kriegel, I.; Jiang, C.; Rodríguez-Fernández, J.; Schaller, R. D.; Talapin, D. V.; Da Como, E.; Feldmann, J. J. Am. Chem. Soc. 2012, 134 (3), 1583−1590. (15) Hsu, S.-W.; On, K.; Tao, A. R. J. Am. Chem. Soc. 2011, 133 (47), 19072−19075. (16) Garcia, G.; Buonsanti, R.; Runnerstrom, E. L.; Mendelsberg, R. J.; Llordes, A.; Anders, A.; Richardson, T. J.; Milliron, D. J. Nano Lett. 2011, 11 (10), 4415−4420. (17) Fann, W.; Storz, R.; Tom, H.; Bokor, J. Phys. Rev. B 1992, 46 (20), 13592. (18) Dorfs, D.; Härtling, T.; Miszta, K.; Bigall, N. C.; Kim, M. R.; Genovese, A.; Falqui, A.; Povia, M.; Manna, L. J. Am. Chem. Soc. 2011, 133 (29), 11175−11180. (19) Shim, M.; Guyot-Sionnest, P. Phys. Rev. B 2001, 64 (24), 245342. (20) Zum Felde, U.; Haase, M.; Weller, H. J. Phys. Chem. B 2000, 104 (40), 9388−9395. (21) Pflughoefft, M.; Weller, H. J. Phys. Chem. B 2002, 106 (41), 10530−10534. (22) Krasavin, A. V.; Zayats, A. V. Phys. Rev. Lett. 2012, 109 (5), 053901. (23) Zeng, Y.; Hoyer, W.; Liu, J.; Koch, S. W.; Moloney, J. V. Phys. Rev. B 2009, 79 (23), 235109. (24) Xie, Y.; Carbone, L.; Nobile, C.; Grillo, V.; D’Agostino, S.; Della Sala, F.; Giannini, C.; Altamura, D.; Oelsner, C.; Kryschi, C.; Cozzoli, P. D. ACS Nano 2013, 7 (8), 7352−7369. (25) Bourlange, A.; Payne, D.; Jacobs, R.; Egdell, R.; Foord, J.; Schertel, A.; Dobson, P.; Hutchison, J. Chem. Mater. 2008, 20 (14), 4551−4553. (26) Mendelsberg, R. J.; Garcia, G.; Milliron, D. J. J. Appl. Phys. 2012, 111 (6), 063515. (27) Draine, B. T.; Flatau, P. J. J. Opt. Soc. Am. A 1994, 11 (4), 1491−1499. (28) Flatau, P.; Draine, B. Opt. Express 2012, 20, 1247−1252. (29) Bel Hadj Tahar, R.; Ban, T.; Ohya, Y.; Takahashi, Y. J. Appl. Phys. 1998, 83 (5), 2631−2645. (30) McArthur, E. A.; Morris-Cohen, A. J.; Knowles, K. E.; Weiss, E. A. J. Phys. Chem. B 2010, 114 (45), 14514−14520. (31) Tice, D. B.; Frederick, M. T.; Chang, R. P.; Weiss, E. A. J. Phys. Chem. C 2011, 115 (9), 3654−3662. (32) Stockman, M. I. Opt. Express 2011, 19 (22), 22029−22106. (33) Utikal, T.; Zentgraf, T.; Paul, T.; Rockstuhl, C.; Lederer, F.; Lippitz, M.; Giessen, H. Phys. Rev. Lett. 2011, 106 (13), 133901.
1126
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Ultrafast Modulation of the Plasma Frequency of Vertically Aligned Indium Tin Oxide Rods Daniel B. Tice,† Shi-Qiang Li,‡ Mario Tagliazucchi,† D. Bruce Buchholz,‡ Emily A. Weiss,† and Robert P. H. Chang*,‡ †
Department of Chemistry, Northwestern University, 2145 Sheridan Rd., Evanston, Illinois 60208-3113, United States Department of Materials Science and Engineering, Northwestern University, 2220 Campus Dr., Evanston, Illinois 60208-3108, United States
‡
S Supporting Information *
ABSTRACT: Light−matter interaction at the nanoscale is of particular interest for future photonic integrated circuits and devices with applications ranging from communication to sensing and imaging. In this Letter a combination of transient absorption (TA) and the use of third harmonic generation as a probe (THG-probe) has been adopted to investigate the response of the localized surface plasmon resonances (LSPRs) of vertically aligned indium tin oxide rods (ITORs) upon ultraviolet light (UV) excitation. TA experiments, which are sensitive to the extinction of the LSPR, show a fluence-dependent increase in the frequency and intensity of the LSPR. The THG-probe experiments show a fluence-dependent decrease of the LSPR-enhanced local electric field intensity within the rod, consistent with a shift of the LSPR to higher frequency. The kinetics from both TA and THG-probe experiments are found to be independent of the fluence of the pump. These results indicate that UV excitation modulates the plasma frequency of ITO on the ultrafast time scale by the injection of electrons into, and their subsequent decay from, the conduction band of the rods. Increases to the electron concentration in the conduction band of ∼13% were achieved in these experiments. Computer simulation and modeling have been used throughout the investigation to guide the design of the experiments and to map the electric field distribution around the rods for interpreting far-field measurement results. KEYWORDS: Localized surface plasmons, doped semiconductors, transient absorption, third harmonic generation, ultrafast optical modulation, carrier injection
T
components for telecommunications and molecular sensing. This Letter describes the study of ultrafast light modulation of vertically aligned indium tin oxide rods (ITORs) and addresses two important research questions: (i) What is the response of an ITOR when an ultrafast modulation occurs to its plasma frequency, ωp, by photoexcitation of electrons into (and subsequent relaxation of electrons from) its conduction band? and (ii) What is the spatial distribution of the electric field intensity within an ITOR as a result of the excitation? It is important to note that in this research the focus is on the response of individual ITOR to the radiation, and thus the
he interaction of light with nanostructured materials is of great interest due to the existence of rich phenomena that have very broad applications in energy,1 the environment,2 health,3 and security.4,5 In recent years, there have been extensive studies to probe the response of metal and semiconductor nanoparticles and their clusters to light.6 These studies have provided information on energy exchanges, excitations, radiation, and loss mechanisms of complex nanosystems. Fundamental understanding of the interaction between light and nanostructures will enable the fabrication of new devices with improved performances. The large oscillator strength and tunability of the surface-plasmon resonances of degenerately doped semiconductors, in particular, across the near- and mid-infrared (NIR and MIR) wavelengths7−11 make them potentially useful as active linear and nonlinear optical © 2014 American Chemical Society
Received: July 27, 2013 Revised: December 28, 2013 Published: February 14, 2014 1120
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
demonstration of an increase in ωp upon excitation of an interband transition in semiconductor-based plasmonic materials. Changes to ωp were not the dominant effect in previous TA experiments on the interband transition of copper chalcogenide nanocrystals.14,24 Experimental Procedure. ITOR Growth. While the concept of light interaction with a single ITOR is simple, the design for an experiment is more involved. To study how an individual ITOR will respond to light, we found after many experiments that the best approach is to grow periodic arrays of ITORs in order to produce near identical rods. We prepared the ITOR samples using a previously reported procedure.8 Briefly, we patterned square lattice arrays of gold nanoparticles on 30 nm thick ITO or In2O3 films grown epitaxially on yttriastabilized zirconia (YSZ) ⟨100⟩ substrates with pulsed laser deposition (PLD). We placed the substrates in a tube furnace at 800 °C, with the boats containing our metal precursors held at 900 °C upstream of the substrates. We used In mesh (100 mesh, 99.99% trace metal basis, Sigma-Aldrich) and SnO2 powder (99.99% trace metal basis, Sigma-Aldrich) in a 9:1 molar ratio and introduced a 25 sccm flow rate of 1% O2 in a N2 buffer gas during the growth process to provide the oxygen source. The gold nanoparticles serve as catalysts for vapor− liquid−solid (VLS) growth of vertically aligned ITORs from the substrate with a square cross section, consistent with the bixbyite crystal lattice of ITO [see Supporting Information (SI1) for processing details).25 We have grown a range of tapered vertical rods, both near periodic and periodic arrays (see Figure 1a,b).
experiments are designed accordingly with the aid of simulation. Unlike plasmons in metals, the energies of the resonances in doped semiconductors can be easily tuned through their elemental composition, which can be modified during synthesis7,9,10 or postsynthesis annealing.8,12−15 Additionally, the free carrier concentrations in doped semiconductors are much lower than in metals, so the plasmonic resonances of these materials are more sensitive than metals to absolute changes in the carrier concentration.16 The ITORs in this study have a carrier concentration on the order of 1020 e−/cm3, which is a factor of 100 lower than the carrier concentration of gold.17 This enhanced sensitivity to the carrier density makes doped semiconductors appealing as active optical components, which require reversible modulation of either the intensity or the frequency of the plasmon resonance and ultrafast response time for fast switching. While several groups have reported shifts in both the intensity and the frequency of the local surface plasmon resonance (LSPR), ωLSPR, from chemical doping,18 photocharging,19 or electrochemical doping of semiconductor nanocrystals,16,20,21 these methods are nonideal for active modulation of optical components because (i) it is uncertain whether the additional carriers lie in the core or surface states in some cases,20,21 (ii) shifts in ωLSPR through chemical treatments require oxidizing or reducing environments, and (iii) electrochemical methods, while reversible, are slow, are sensitive to nanostructure size, and require both electrolyte and electrodes, which limits potential applications. Recently, Krasvin and Zayats have introduced a novel concept of electro-optical fieldeffect nanoplasmonic modulator for nanoscale integration.22 In this work, we photoinject charge carriers into the conduction band of ITO upon direct photoexcitation at energies near its bandgap (∼3.6 eV); this excitation increases the carrier density in the conduction band and thereby increases ωp of the ITORs. We perform two different types of ultrafast transient pump−probe spectroscopy experiments on the ITORs to characterize the modulation of ωp upon carrier injection: 1. Using UV-pump and NIR-probe transient absorption (NIR-TA), we observe a shift in ωLSPR to higher energy and an increase in the intensity of the LSPR, consistent with an increase in ωp. The bandwidth at which we can modulate ωp is limited by the time scale of electronic relaxation of free carriers in the ITORs, which is on the order of one picosecond. 2. In a separate experiment, we pump the system at 325 nm and monitor the transient intensity of LSPR-mediated third-harmonic generation (THG) from the ITORs as a function of the time delay between the UV pump and the THG probe. We refer to this experiment as “THG-probe”. The NIR TA experiments are sensitive to changes in the extinction of the sample, which are not necessarily plasmonic in origin. The THG-probe is sensitive to the distribution of the local electric field within the ITORs,23 a distinctly plasmonic phenomenon that is dictated by detuning of the THG-probe fundamental relative to ωLSPR. At the chosen wavelength of our THG probe, we observe a decrease of the plasmon-induced enhancement of the electric field within the rod upon UV excitation. This effect is consistent with a shift of ωLSPR to higher energy which further verifies that the UV excitation of ITO enables dynamic modulation of ωp through carrier injection and decay. This work represents the first time-resolved study of plasmon dynamics in crystalline ITORs and the first study of LSPRenhanced THG in ITORs. Additionally, it is the first
Figure 1. Scanning electron micrographs of ITORs from the VLS growth process. (A) Near periodic distributed arrays with average spacing between rods of 0.6 μm and a height of 4 μm. (B) Ordered array of 10.8 μm tall tapered rods with spacing between rods of 1.6 μm.
Pitch and Rod Height Selection. Once periodicity is introduced into the array, it is then necessary to select a rodspacing (or pitch) to ensure that collective effects, such as mode coupling between LSPR resonances and photonic grating modes, are not present or minimized. To avoid the coupling of the LSPR fields between the rods, we utilized ITOR pitches >400 nm, which is much larger than the maximum decay distance of the surface fields. The effect of the distance between rods on their electrodynamics coupling is discussed in the Supporting Information section SI-2. In the design of our experiments we have chosen pitches and rod heights such that the phase space we work in has no observable mode couplings, and this is further confirmed by our simulation. We selected two different sets of samples for our experiments. For the LSPR modulation experiment (NIR-TA), we chose an ITOR array with 600 nm spacing to optimize detection with minimum broadening of the peak line width. On the other hand, for THG probe experiments, we have simulated an ITOR array with different spacings (800 nm, 1.2 μm, 1.6 μm, 2.4 μm), and found that the array with 1.6 μm spacing gave the maximum THG signal (see Figure SI-3.1.1 in Supporting Information SI-3.1). 1121
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
The simulation results compare well with our experimental measurements (see Figure SI-3.1.2). Rod Shape. The shape of the ITOR also contributes to our study especially for the THG-probe experiments (see Supporting Information SI-3.2). To observe THG in the rods, it is critical that the skin depth of the ITOR is comparable to the wavelength under study. Figure SI-3.2 in the Supporting Information compares the near-field distribution for rods of different shape and shows that, by having a tapered (along the length of the axis) rod, the condition of having a thickness much smaller than the wavelength of the incident light can be met easier in our experiments. Experimental Setup. To excite only one type of LSPR, our experiment is set up such that the incident Poynting vector is parallel to the long axis of the ITORs. With this geometry, only LSPR modes across the short axis of the rods (transverse modes) will be excited (i.e., longitudinal LSPR along the long axis of the rods will not be excited). Operating within this selected phase space per our design, we present here results from two samples in this study: for TA experiments we used a sample with a ITOR spacing of 600 nm, a height of 4 ± 0.2 μm, and sides of 265 ± 25 nm, while for the THG-probe experiments we used ITORs with a spacing of 1.6 μm and a height of 10.8 ± 0.1 μm. We provide additional experiments for ITOR arrays of different geometric parameters in the Supporting Information (SI) to demonstrate reproducibility. We prepared the sample for the TA experiments so the wavelength of the LSPR was within the spectral range accessible by our NIR-TA setup (800−1630 nm), while we prepared the sample for the THG-probe experiments so that the wavelength of the LSPR was resonant with the idler wavelength (2050 nm) of our optical parametric amplifier (OPA) when we used the signal wavelength (1300 nm) to generate the UV pump (see Figure 2 for the schematic diagrams of the instrument setup for each type of measurement). Ground State Extinction Spectroscopy of the LSPR of ITORs. To simulate the extinction spectrum of the ITOR array, we follow the same procedure used in earlier publications8,12,26 by calculating the frequency-dependent dielectric of the material with the Drude equation eq 1, ε(ω) = ε∞ −
Figure 2. (A) Optical setup for TA measurements and (B) optical setup for THG-probe measurements. The ITOR sample is first excited by a UV-pump pulse that promotes electrons from the valence band into the conduction band. In the NIR TA experiment (A), the changes in sample extinction induced by the pump beam in the LSPR region (1430−1630 nm) are probed with a NIR continuum using active background subtraction. In the THG-probe experiment (B), the probe beam is 2050 nm (the idler of our OPA), which generates an output pulse at 683 nm from the sample due to third harmonic generation. We also measure the THG-probe signal using active background subtraction.
ωp2 ω 2 + i Γω
(1)
where ε∞ is the high-frequency dielectric constant of the host medium (∼3.9 for ITO),8,9 ωp is the plasma frequency of the material, and Γ is its damping parameter. Using DDSCAT 7.3,27,28 a discrete dipole approximation (DDA) program (see SI-4), we varied ωp and Γ in eq 1 to fit the wavelength and line width of the LSPR, respectively. The best fit of the extinction spectrum corresponds to a value for ωp of 1.59 eV and Γ of 80 meV. We can relate ωp (using eq 2) to a carrier concentration of the ITO, n, of 7.35 × 1020 e−/cm3,
Figure 3. NIR extinction spectrum of the sample of ITO nanorods in Figure 1A (black) and the extinction spectrum calculated using DDA with the Drude equation (dashed teal trace) for a ωp = 1.59 eV and Γ = 80 meV. We have corrected the extinction spectrum to account for the underlying substrate and ITO thin film.
2
ωp =
ne ε0m*
(2) 29
which is in good agreement with values of n for ITO. In eq 2, e is the formal charge of an electron, ε0 is the permittivity of free space, and m* is the effective carrier mass in the medium of choice (0.4me for electrons in ITO, where me is the electron rest mass).8 Using these results, we simulated the extinction spectrum (see Figure 3) of a sample with vertically aligned ITORs as shown in Figure 1A. We see that the transverse-mode
LSPR of the ITORs comprises two convoluted peaksat 1770 and 2100 nm for this sampledue to the square cross-section of the ITORs, consistent with previous measurements.8 Transient Absorption Spectroscopy of the LSPR of ITORs upon UV Excitation. To study the effect of the photoinjection 1122
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
of carriers into the conduction band of the ITORs on ωp, we monitored the LSPR with a transient absorption system (see Figure 2A) that has been described in detail previously.30,31 Briefly, to generate the pump wavelengths, we used two barium borate crystals to upconvert the NIR signal wavelengths from our optical parametric amplifier (OPA) (TOPAS-C, Light Conversion) to UV wavelengths (300−350 nm). To generate a continuum probe for TA measurements, we focused a portion of the NIR signal output from the OPA to generate a NIR continuum in a 1 cm yttrium alumina garnate (YAG) crystal. We positioned the ITOR samples normal to the pump and probe beams, so that the Poynting vector of the light is along the long axes of the ITORs. This sample geometry results in selective excitation of the transverse mode LSPR of the ITORs.8 The polarization of the pump and the probe beams were parallel to each other for these experiments. Figure 4A shows the transient spectra of the ITORs with a spacing of 600 nm for probe wavelengths between 1430 and
1630 nm following photoexcitation of the ITORs at 300 nm which excites electrons in the valence band across the bandgapas a function of the fluence of the pump. For all fluences, we observe a photoinduced absorption that grows in intensity and shifts monotonically to higher energy with increasing pump fluence, with a peak clearly visible for higher fluences. We note that, since TA measurements report differential absorptions, this photoinduced feature corresponds to an increase in both the frequency and the oscillator strength of the LSPR, relative to LSPR of the ITORs in their ground state upon UV excitation. The measured absorbance of the sample at the excitation wavelength (300 nm) is 1.3 A.U.; from this value, we calculate a range of expectation values of 5.3 × 105 to 2.2 × 107 photons absorbed per rod on going from the lowest to highest fluences of the pump beam. If we assume each absorbed photon promotes a single carrier to the conduction band of the ITORs, then the 300 nm excitation increases the carrier concentration in the conduction band by 0.5% at the lowest excitation fluence, and by ∼13% at the higher excitation fluence, after accounting for the saturable absorption of the ITORs. The observation of a carrier concentration-dependent energy and intensity of ωLSPR agrees with eq 2 and is evidence that the photoinjected carriers are delocalized in the conduction band of the rod, as opposed to populating surface states.16 Figure 4B shows normalized kinetic traces, extracted from the TA spectrum in Figure 4A at 1570 nm as a function of increasing fluence of the 300 nm pump. The lifetime of the photoinjected carriers is insensitive to the fluence of the pump; the lack of change in the kinetics indicates that no new relaxation processes are introduced at higher pump fluences, which is consistent with our assignment of the photoinduced absorption to carrier injection into the conduction band of the ITO. The electrons promoted into the conduction band upon UV excitation are delocalized and behave collectively with the steady-state population of free electrons present in the conduction band from doping; we effectively “photodope” the ITO and achieve an increase in ωp of 100 meV with a single picosecond response time. We note that, as our excitation wavelength is similar to the bandgap of ITO, the excess energy of the injected carriers is minimized and we do not expect carrier-cooling dynamics to contribute to our measurements of the LSPR. As a control, we performed NIR-pump TA experiments to monitor hot electron-cooling lifetimes and observe nearly instrument response limited dynamics (180 fs for this NIR pump-NIR probe measurement (see SI-5). LSPR-Enhanced Third Harmonic Generation as a Probe of the Electric Field Distribution within ITORs. We have demonstrated that both ωLSPR and the oscillator strength of the LSPR increase upon carrier injection, but these TA experiments are sensitive only to the extinction of a transition, and are unable to probe the local electric field distribution of the rod, a primary characteristic of a plasmonic material.6,32 The probability of nonlinear optical processes, in contrast, depends not only on the intensity,6 but also the spatial distribution of the local electric field of a plasmonic excitationthat is, whether the induced field is inside versus outside of the nonlinear medium.33 Here, we use LSPR-enhanced THG from the ITORs to probe the local field distribution of the plasmon as a function of the excitation wavelength relative to ωLSPR. For plasmonic nanostructures of subwavelength dimensions, the efficiency of THG is dominated by the local electric field at a given excitation wavelength6,23 rather than the phase-matching
Figure 4. (A) Plot of the NIR transient spectra as a function of the fluence of a pump laser with a wavelength of 300 nm (light red to dark red traces with increasing fluence), for the ITORs from Figure 1A. All spectra are shown at time-zero of the temporal overlap of the pump and probe beams. The induced absorption increases in amplitude and shifts to higher energy with increasing fluence of the pump. Inset: Normalized transient spectra clearly show the spectral shift as a function of the fluence of the pump. The spectral cutoff at 1630 nm is due to a loss of sensitivity of our detector at wavelengths longer than 1630 nm. (B) Plot of the normalized kinetic traces extracted from the transient spectra in A at 1570 nm, for different fluences of the pump. The lifetime of the induced absorptions, which we fit with a single exponential time constant of 1.1 ps convoluted with our instrument response function of 200 fs (light blue dashed trace), does not depend on the fluence. 1123
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
Figure 5. (A) Plot of ηTHG for a given excitation wavelength (blue circles and lines) for the ITORs shown in Figure 1B, plotted with their ground state extinction spectrum (black trace), and their calculated extinction spectrum (dashed blue trace), which corresponds to ωp = 1.43 eV and Γ = 96 meV. We measured ηTHG in an integration sphere at a fluence of 790 μJ/cm2 for all wavelengths; the error bars come from three separate measurements at different locations on the sample. (B) Calculated local field intensity (|E|/|E0|) for the ITORs at the wavelengths (I−III) indicated in A. At I, where there is no significant extinction by the LSPR, there is minimal field enhancement in the volume of the rod. At II, extinction by the LSPR leads to increased efficiency of THG because the local field is enhanced inside the volume of the rod. At III, equal extinction by the LSPR does not enhance THG because the local field is outside the volume of the rod. For the near-field simulations, we propagated the field along the long axis of the rod (from the top to the bottom of the figure), with the polarization perpendicular to the long axis, and limited the scale of the intensity to enhance image contrast (the tips of the rod show higher field enhancements). Figure SI-3.2 in the Supporting Information shows |E|/|E0| profiles along the axial and transverse directions.
Figure 6. (A) Plot showing the wavelengths of the TA probe at 1630 nm (gray dashed line) and the fundamental of the THG-probe at 2050 nm (dashed teal line) plotted against the extinction of the LSPR (black trace) and ηTHG (blue circles and lines) from Figure 5A. (B) Normalized kinetics of the rod sample upon excitation with a 325 nm pump for the TA experiment, taken at 1630 nm (black trace), and the THG output for the fundamental probe centered at 2050 nm (teal trace), recorded at 695 nm. We plot the THG as −log(ΔηTHG) for ease of comparison to the TA experiment. We used a fluence of 12 mJ/cm2 for the 2050 nm fundamental of the THG-probe. The 325 nm pump had a fluence of 720 μJ/cm2 for the THG-probe experiment.
exactly track the extinction of the LSPR; we observe appreciable THG only at excitation wavelengths resonant with the high-energy side of the LSPR. Our DDA simulations of the ITORs explain this behavior: for wavelengths of light offresonance with the LSPR (Figure 5B, I), the rod behaves as a conventional dielectric, and, as expected, we do not measure appreciable THG. When the light is of a wavelength on resonance with the higher-energy side of the LSPR (Figure 5B, II), the near-field intensity of the light incident to the ITORs is enhanced not only on the surface of the rod, but also inside the volume of the rod, where ITO serves as the nonlinear medium, and we observe appreciable THG. For wavelengths resonant
considerations that dominate the efficiency for bulk, transparent media.6 The efficiency of the THG process, ηTHG, for the ITORs therefore increases at the excitation wavelengths where the LSPR-enhanced local electric field is confined within the volume of the rod, which increases the effective electric field the rod experiences.23 Figure 5A shows ηTHG for the ITORs shown in Figure 1B (spacing = 1.6 μm), for excitation at a series of wavelengths that span the LSPR spectrum of the ITORs. In this plot, ηTHG = powerTHG/powerfundamental (blue circles and lines). For reference, Figure 5A also shows the extinction spectrum of the LSPR (black trace). The THG efficiency spectrum does not 1124
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
lifetime of ΔηTHG matches that of the photoinduced absorption measured by NIR TA. This result confirms that the LSPRmediated local field distribution within the rod for a given excitation wavelength is modified upon UV excitation and that the decay of the photoinjected carriers from the conduction band of the ITO is responsible for the transient behavior of both the energy and the intensity of the plasmon resonance and the efficiency of the THG process. Ultrafast methods of altering the optical properties in materials are desirable for applications in active plasmonics and switching.6,32 Here, we have demonstrated (i) the first time-resolved studies of plasmon dynamics in ITO nanostructures, (ii) LSPR-enhanced THG from ITO rods, which we use as a probe of the local electric field, and (iii) the first demonstration of the modulation of the linear and nonlinear optical properties of the LSPR of a doped semiconductor through an increased ωp upon direct excitation of an interband transition.
with the lower-energy side of the LSPR, the extinction of the light by the rod is still strong, but the near-field intensity lies mainly outside of the rod. Due to the negligible nonlinear susceptibility of air, we observe no THG at these wavelengths. The behavior observed for light wavelengths resonant with the higher-energy side of the LSPR is a consequence of the NIR wavelengths of the LSPR; while resulting in a larger skin depth and enabling the field to penetrate into the rod, longer wavelengths also result in weaker surface fields compared to visible wavelength LSPRs in gold nanoparticles. The weaker surface field enhances the sensitivity of THG in ITORs to distribution of the field in the volume of the ITO. The Supporting Information (SI-6) contains THG experiments on another ITOR sample, as well as simulations using a finitedifference time domain method, (see SI-7). We performed a pump−probe experiment where we tune the pump to 325 nm to excite the electrons in ITORs across the bandgap, tune the probe to 2050 nm, and detect the THG wavelength at 695 nm. We detect at 695 nm instead of the “true” THG wavelength, 683 nm, due in part to the 200 cm−1 bandwidth of our laser and the Maker’s Fringing of the transmitted THG-probe, and to a probe-fluence dependent shift in the THG output wavelength at the probe fluences needed for the THG-probe experiment. We chose 2050 nm for the probe because it produces a near-maximum THG intensity, given by THG efficiency ηTHG, as shown in Figure 6A. Upon UV excitation, ηTHG decreases (Figure 6B, teal trace), which is what we predict for an increase of ωp because population of the conduction band with excess carriers shifts the LSPR and the field intensity spectrum to higher energy, and thereby lowers the efficiency of our THG-probe at 2050 nm. Note that 2050 nm is the wavelength that optimizes THG for the ground state, the wavelength of optimal THG efficiency will be shorter than 2050 nm for the excited state. In our experiments we fixed the THG probe wavelength to 2050 nm, and therefore, we observe a decrease in the THG efficiency upon UV photoexcitation of the ITORs. We additionally performed NIR TA with a 325 nm pump and a 1630 nm probe (Figure 6A, dashed gray vertical line), to measure the expected shift of the LSPR for this sample. We observe that upon UV excitation, the extinction at 1630 nm increases (Figure 6B black trace), which is consistent with our expected shift of the LSPR to higher energy. Figure 6B also shows that the kinetics of the TA and THG-probe experiments share the same lifetime, which indicates that the same mechanismthe modulation of ωp upon injection of electrons into and their subsequent decay from the conduction band of the ITORsis responsible for the processes we observe here. Conclusions. We have demonstrated the ultrafast dynamic modulation of ωp in ITORs upon photoinjection of electrons into the conduction band of ITO and mapped the field distribution of the plasmon resonance using pump−probe spectroscopies. In TA experiments with a UV pump and NIR probe, where we monitor the extinction of the LSPR, we observe a photoinduced absorption that increases in oscillator strength and energy with increasing pump fluence, which is consistent with the changes in LSPR expected for an increase of ωp. The lifetime of this process is 1.1 ps and is independent of the fluence of the excitation. We demonstrate LSPR-enhanced THG from the ITORs and show that the THG efficiency, ηTHG, at a given excitation wavelength of the LSPR corresponds to the local electric field distribution within the rods. Upon UV excitation of the rods, ηTHG at our probe wavelength decreases, which is consistent with an increase of ωLSPR, and that the
■
ASSOCIATED CONTENT
S Supporting Information *
Additional supporting data. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This research was funded under NSF-IGERT program (DGE0801685) and in part from NSF-DMR-1121262 and 0843962. The FDTD simulation was performed with QUEST computational resources located in Northwestern University under proposal p20447. Patterning of the gold nanoparticle array was performed with JEOL-9300 in the Center for Nanoscale Materials at Argonne National Laboratory (CNM 30831). Use of the Center for Nanoscale Materials was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.
■
REFERENCES
(1) Kamat, P. V. J. Phys. Chem. C 2007, 111 (7), 2834−2860. (2) Vaseashta, A.; Vaclavikova, M.; Vaseashta, S.; Gallios, G.; Roy, P.; Pummakarnchana, O. Sci. Tech. Adv. Mater. 2007, 8 (1), 47−59. (3) Huang, X.; Jain, P. K.; El-Sayed, I. H.; El-Sayed, M. A. Laser Med. Sci. 2008, 23 (3), 217−228. (4) Kirchain, R.; Kimerling, L. Nat. Photonics 2007, 1 (6), 303−305. (5) Lal, S.; Link, S.; Halas, N. J. Nat. Photonics 2007, 1 (11), 641− 648. (6) Kauranen, M.; Zayats, A. V. Nat. Photonics 2012, 6 (11), 737− 748. (7) Luther, J. M.; Jain, P. K.; Ewers, T.; Alivisatos, A. P. Nat. Mater. 2011, 10 (5), 361−366. (8) Li, S. Q.; Guo, P.; Zhang, L.; Zhou, W.; Odom, T. W.; Seideman, T.; Ketterson, J. B.; Chang, R. P. H. ACS Nano 2011, 5 (11), 9161− 9170. (9) Mendelsberg, R. J.; Garcia, G.; Li, H.; Manna, L.; Milliron, D. J. J. Phys. Chem. C 2012, 116 (22), 12226−12231. (10) Buonsanti, R.; Llordes, A.; Aloni, S.; Helms, B. A.; Milliron, D. J. Nano Lett. 2011, 11 (11), 4706−4710. (11) Rowe, D. J.; Jeong, J. S.; Mkhoyan, K. A.; Kortshagen, U. R. Nano Lett. 2013, 13 (3), 1317−1322. 1125
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
Nano Letters
Letter
(12) Manthiram, K.; Alivisatos, A. P. J. Am. Chem. Soc. 2012, 134 (9), 3995−3998. (13) Scotognella, F.; Della Valle, G.; Srimath Kandada, A. R.; Dorfs, D.; Zavelani-Rossi, M.; Conforti, M.; Miszta, K.; Comin, A.; Korobchevskaya, K.; Lanzani, G. Nano Lett. 2011, 11 (11), 4711− 4717. (14) Kriegel, I.; Jiang, C.; Rodríguez-Fernández, J.; Schaller, R. D.; Talapin, D. V.; Da Como, E.; Feldmann, J. J. Am. Chem. Soc. 2012, 134 (3), 1583−1590. (15) Hsu, S.-W.; On, K.; Tao, A. R. J. Am. Chem. Soc. 2011, 133 (47), 19072−19075. (16) Garcia, G.; Buonsanti, R.; Runnerstrom, E. L.; Mendelsberg, R. J.; Llordes, A.; Anders, A.; Richardson, T. J.; Milliron, D. J. Nano Lett. 2011, 11 (10), 4415−4420. (17) Fann, W.; Storz, R.; Tom, H.; Bokor, J. Phys. Rev. B 1992, 46 (20), 13592. (18) Dorfs, D.; Härtling, T.; Miszta, K.; Bigall, N. C.; Kim, M. R.; Genovese, A.; Falqui, A.; Povia, M.; Manna, L. J. Am. Chem. Soc. 2011, 133 (29), 11175−11180. (19) Shim, M.; Guyot-Sionnest, P. Phys. Rev. B 2001, 64 (24), 245342. (20) Zum Felde, U.; Haase, M.; Weller, H. J. Phys. Chem. B 2000, 104 (40), 9388−9395. (21) Pflughoefft, M.; Weller, H. J. Phys. Chem. B 2002, 106 (41), 10530−10534. (22) Krasavin, A. V.; Zayats, A. V. Phys. Rev. Lett. 2012, 109 (5), 053901. (23) Zeng, Y.; Hoyer, W.; Liu, J.; Koch, S. W.; Moloney, J. V. Phys. Rev. B 2009, 79 (23), 235109. (24) Xie, Y.; Carbone, L.; Nobile, C.; Grillo, V.; D’Agostino, S.; Della Sala, F.; Giannini, C.; Altamura, D.; Oelsner, C.; Kryschi, C.; Cozzoli, P. D. ACS Nano 2013, 7 (8), 7352−7369. (25) Bourlange, A.; Payne, D.; Jacobs, R.; Egdell, R.; Foord, J.; Schertel, A.; Dobson, P.; Hutchison, J. Chem. Mater. 2008, 20 (14), 4551−4553. (26) Mendelsberg, R. J.; Garcia, G.; Milliron, D. J. J. Appl. Phys. 2012, 111 (6), 063515. (27) Draine, B. T.; Flatau, P. J. J. Opt. Soc. Am. A 1994, 11 (4), 1491−1499. (28) Flatau, P.; Draine, B. Opt. Express 2012, 20, 1247−1252. (29) Bel Hadj Tahar, R.; Ban, T.; Ohya, Y.; Takahashi, Y. J. Appl. Phys. 1998, 83 (5), 2631−2645. (30) McArthur, E. A.; Morris-Cohen, A. J.; Knowles, K. E.; Weiss, E. A. J. Phys. Chem. B 2010, 114 (45), 14514−14520. (31) Tice, D. B.; Frederick, M. T.; Chang, R. P.; Weiss, E. A. J. Phys. Chem. C 2011, 115 (9), 3654−3662. (32) Stockman, M. I. Opt. Express 2011, 19 (22), 22029−22106. (33) Utikal, T.; Zentgraf, T.; Paul, T.; Rockstuhl, C.; Lederer, F.; Lippitz, M.; Giessen, H. Phys. Rev. Lett. 2011, 106 (13), 133901.
1126
dx.doi.org/10.1021/nl4028044 | Nano Lett. 2014, 14, 1120−1126
