You are on page 1of 5

Letter

pubs.acs.org/NanoLett

Transient Reflection: A Versatile Technique for Ultrafast


Spectroscopy of a Single Quantum Dot in Complex Environments
Christian Wolpert,†,§ Christian Dicken,†,§ Paola Atkinson,‡ Lijuan Wang,‡ Armando Rastelli,‡
Oliver G. Schmidt,‡ Harald Giessen,† and Markus Lippitz*,†,§

4th Physics Institute and Research Center SCOPE, University of Stuttgart, Pfaffenwaldring 57, D-70550 Stuttgart, Germany

Institute for Integrative Nanosciences, IFW Dresden, Helmholtzstrasse 20, D-01069 Dresden, Germany
§
Max Planck Institute for Solid State Research, Heisenbergstrasse 1, D-70569 Stuttgart, Germany
*
S Supporting Information

ABSTRACT: Increasingly complex structures such as optical


antennas or cavities are coupled to self-assembled quantum
dots to harvest their quantum-optical properties. In many
cases, these structures pose a problem for common methods of
ultrafast spectroscopy used to write and read out the state of
the quantum dot. We present a pure far-field method that only
requires optical access to the quantum dot and does not
impose further restrictions on sample design. We demonstrate
Rabi oscillations and perturbed free induction decay of single
GaAs quantum dots that have a dipole moment as small as 18 D. Our method will greatly facilitate ultrafast spectroscopy of
complex quantum-optical circuits.
KEYWORDS: Coherent spectroscopy, quantum emitter, plasmonic nanocircuits

S emiconductor quantum dots (QDs) closely resemble


atomic dipole emitters in the solid state,1 which makes
them attractive building blocks for novel photonic devices. We
section. In order to overcome the weak light-matter interaction,
researchers resorted to nanoapertures to block unwanted
light,15,16 to electrodes that modulate the emission line by
can individually address, manipulate and read them out on the Stark shift,17 or to collecting the electrons tunneling out of
picosecond time scales using ultrafast spectroscopy.2 Individual the quantum dot.18 All these approaches hinder or even make
dots already find use as single photon sources for quantum impossible the fascinating plasmonic nanocircuit applications as
cryptography3 or as quantum bits for quantum computing.4 discussed above. In this Letter, we demonstrate that far-field
Hybrid quantum systems combining a quantum dot with a reflection spectroscopy is appropriate to gain ultrafast
microcavity5,6 or a particle plasmon7−10 can enhance light-
spectroscopic information of single quantum dots without
matter interaction. Even more fascinating applications require
the controlled interaction of many quantum systems in a further restrictions to the structure design. This allows for the
network. Recent publications envision the entanglement of first time experiments with quantum dots in complex systems.
quantum dots by means of plasmonic waveguides11 or the use We first introduce our experimental technique and then
of quantum dots as optical transistors.12 demonstrate its capabilities on two classical examples of
To realize these proposals, one needs to fulfill at least two ultrafast spectroscopy of quantum emitters, namely Rabi
requirements, namely the controlled positioning of the emitter oscillations19,20 and perturbed free induction decay.21−23
relative to the nanostructure and ultrafast optical spectroscopy Concept. The key idea (see Figure 1A) is that the
on the combined system. The first point can be realized by absorption by a dipole can be understood as the interference
using self-assembled quantum dots that grow at positions of the optical wave scattered at the dipole in forward direction
defined by electron-beam lithography.13,14 The second require- with that passing the dipole unperturbed, known as the optical
ment needs a noninvasive optical technique that does not theorem.24 Both absorption and reflection have their origin in
interfere with the structure around the quantum dot. Optical the scattered field ES of a dipole25
far-field spectroscopy is such a technique. It has been α0 −i γ
successfully applied to ultrafast coherent spectroscopy of ES = α̃ (ω)E0 with α̃ (ω) =
2 δ + iγ (1)
interface-fluctuation quantum dots (for a review see ref 2).
However, these stochastically formed dots are not compatible
with methods of deterministic positioning. Self-assembled Received: October 28, 2011
quantum dots, in contrast, can be grown in a site-controlled Revised: December 2, 2011
manner, but in turn have a much smaller absorption cross- Published: December 15, 2011

© 2011 American Chemical Society 453 dx.doi.org/10.1021/nl203804n | Nano Lett. 2012, 12, 453−457
Nano Letters Letter

Figure 1. (A) The field scattered at the quantum dot interferes with
the transmitted as well as with the reflected probe field. It gives rise to
transient absorption and transient reflection features. (B) The phase
lag ϕ changes the spectral line shape of the transient reflection signal
from absorptive to dispersive and back.
Figure 2. Spectrum of the reflected probe light R(ω,Δt) in the
presence of the pump (blue curve) and after subtraction of the
where E0 is the exciting probe field, α0 is the peak absorption of reference spectrum (no pump): ΔR = R(ω,Δt) − R0(ω) (red curve).
the dipole, δ = ω − ω0 is the probe detuning from the dipole’s The energy scale is relative to the exciton energy of the quantum dot
resonance, and γ ≪ ω0 is the resonance width. The total at 1.72 eV. The pump−probe delay was set to Δt = 150 ps. Pump-
transmitted and reflected fields are the respective coherent induced bleaching of the exciton transition in the QD is responsible
sums of the scattered field with the unperturbed transmitted or for the sharp feature at zero detuning. The noise is limited by photon
reflected probe fields, taking the Fresnel coefficients for shot noise to 2.4 × 105 photons, resulting in a signal-to-noise ratio of
about 17 with an integration time of 10 min. Inset: Energy level
reflection and transmission, r and t, of the sample surface
diagram of the fundamental transition of a quantum dot. The excitonic
into account.25 states |10⟩ and |01⟩ are separated by the fine structure splitting (FSS),
E T = tE0 + α̃ (ω)tE0 (2) while the biexciton state |11⟩ is reduced in energy due to Coulomb
interaction.
ER = rE0 + α̃ (ω)ei2ϕt 2E0 (3) equipped with a CCD camera (spectral resolution 100 μeV).
ϕ describes the phase lag that the field acquires by traveling The reflected pump light in turn is suppressed by crossed
from the sample surface to the dipole position. The phase lag ϕ polarizers and a double modulation scheme. The high quantum
influences the spectral shape of the signal, varying from an efficiency of the CCD camera allows us to acquire shot-noise
absorptive to a dispersive line shape, as shown in Figure 1B . limited reflection spectra with more than 1010 photons per pixel
This interference effect was used to detect single molecule in one minute integration time. In addition, fluorescence
absorption,26 steady-state absorption of single quantum dots,27 emission spectra and spatial emission maps can be taken.
and transient absorption of single interface-fluctuation quantum Details on experimental parameters and data processing can be
dots using a scanning near-field aperture23 or heterodyne found in the Supporting Information.
spectral interferometry.28 We will apply this technique to The transient reflection spectrum ΔR/R of a single quantum
ultrafast spectroscopy of a weakly absorbing self-assembled dot is acquired by taking the difference of the probe reflection
quantum dot. spectrum with and without preparing an excited state by the
Sample. Our quantum emitters are epitaxial GaAs quantum pump pulse. The pump and probe beams are switched on and
dots in AlGaAs barriers.29 The dots form in self-assembled off independently by acousto-optic modulators at a rate of
nanoholes that were prepared at low density in the lower 1 kHz. By subtracting consecutive spectra from each other,
AlGaAs layer. Detailed structural information is provided in the differential reflection spectra are obtained which are then
Supporting Information. The fundamental transition of the normalized to the spectrum of the probe pulse.
localized exciton is around 1.7 eV where the quantum efficiency Figure 2 shows an example of the probe spectrum R(ω,Δt)
of Si-based photodetectors is high. The asymmetry of the (blue curve) and the difference of subsequent spectra
quantum dot leads to linear polarization selection rules and a ΔR(ω,Δt) (red curve) as a function of the detuning ΔE from
fine structure splitting (FSS) in the order of 100 μV between the fundamental exciton transition in a single quantum dot.
the two fundamental bright exciton states, which we denote as One can discern two contributions to ΔR: a spectrally broad
|01⟩ and |10⟩ (see inset in Figure 2 and ref 30). The signal that reproduces the shape of the probe spectrum and a
polarization directions, Π90° and Π0°, are oriented along the sharp, dispersive feature at the exciton transition energy. The
crystal axes [11̅0] and [110]. The GaAs substrate is situated broad signal originates from a modified reflectivity due to the
only 20 nm below the QD layer. It absorbs at the exciton presence of photocarriers created by band absorption in the
resonance energy. This complex environment makes a GaAs substrate and cap.31,32 As the distance of the quantum dot
traditional transient transmission experiment impossible. to the sample surface corresponds to a phase lag ϕ = 0.2 π
Experiment. Orthogonally polarized pump and probe (Figure 1B), we find a dispersive spectral feature from the
pulses at the exciton resonance wavelength are derived from quantum dot at the exciton resonance. This sharp contribution
a mode-locked Ti:Sapphire laser in combination with two to the spectrum is due to bleaching of the |00⟩ → |01⟩
spectral pulse shapers. This results in a temporal resolution of transition (see inset in Figure 2). When the pump pulse
1 ps. The temporal delay Δt between both pulses is varied by a populates state |10⟩, the transition |00⟩ → |01⟩ cannot be
mechanical translation stage. The collinear beams are focused driven by the probe pulse. The amplitude of the dispersive
by a 0.7 NA objective onto the sample in a cryostat (T = 10 K). feature therefore monitors the population n10 of state |10⟩. We
The spatial resolution is in the order of 700 nm, which allows fit the line shape25,33 and obtain a zero-to-peak amplitude of
us to address single quantum dots. The reflected probe light is ΔR/R = 5.0 × 10−5. In comparison to steady-state frequency
collected by the same objective and analyzed by a spectrometer resolved experiments,25 in the present experiment the signal
454 dx.doi.org/10.1021/nl203804n | Nano Lett. 2012, 12, 453−457
Nano Letters Letter

amplitude is reduced by the ratio of the quantum dot line width the data by considering an effective pulse area θeff
(several μeV) to the resolution of our spectrometer (100 μeV).
⎛ P ⎞
In spite of this reduction, Figure 2 demonstrates that plain far- ⎜ ⎟
Psat
field reflection spectroscopy is able to measure transient signals θeff = θP⎜1 − f ⎟
P
of a single quantum dot. ⎜ 1+ ⎟
Rabi Oscillations. Let us turn now to coherent spectros- ⎝ Psat ⎠ (5)
copy of the quantum dot’s exciton transition. We manipulate where P and f are the pump power and a free fitting parameter,
the state of the quantum dot by a pump pulse which is resonant respectively. This relation between the bulk carrier concen-
with the |00⟩ → |10⟩ transition. The pulse can drive the tration and the pump pulse area reduction points to an effect
quantum dot to the |10⟩ state and, as it is shorter than the on a picosecond time scale, as the probe pulse coming at least
coherence time of that transition, also back again to the ground 50 ps later is unmodified (the height of the Rabi oscillations is
state, leading to Rabi oscillations in the population n10 constant). Simultaneously, no free carriers relax into the
proportional to sin2(θP/2). The pump pulse area θP is quantum dot, as the Rabi oscillation returns to zero. The
proportional to the square root of the pump intensity microscopic origin of this effect is under current investigation.
∞ EP⃗ (t ) ·μ⃗ Taking into account all experimental parameters that can be
θP = ∫−∞ ℏ
dt
(4)
determined independently (see Supporting Information) the
only remaining fitting parameters in Figure 3B are the
where E⃗ P(t) is the electric field of the pump pulse at the site of transition dipole moment |μ⃗|, the proportionality factor f, and
the quantum dot and μ⃗ is the transition dipole moment of the an overall amplitude scaling parameter. As the dipole moment
exciton. The system is completely inverted for a pulse with enters both the signal amplitude (via α0 in eq 1) as well as the
θP = π (π-pulse), whereas a 2π-pulse performs a full rotation pump pulse area (via eq 4) the amplitude scaling factor in the
and leaves the system in the ground state. order of one only accounts for experimental imperfections.
As described above, we monitor the population n10 by the To evaluate the stability and reproducibility of our
bleaching it imposes on the |00⟩ → |01⟩ transition. In the measurement scheme, we varied the delay time Δt from 50
spectral map (Figure 3A) every vertical line is a ΔR/R spectrum to 200 ps (details in the Supporting Information). The signal
shows no dependence on the pump−probe delay but a minor
reduction in amplitude due to the finite lifetime of the exciton.
Comparing several quantum dots we obtained transition dipole
moments |μ⃗| between 17 and 19 D. The saturation intensities
varied between 150 and 600 MW/cm2, which is in good
agreement with the literature values for defect-rich GaAs.36,37
The values of the fitting parameter f varied between 0.5 and 0.8.
The considerable range of these two parameters describing the
delocalized carriers suggests that the environment of the
different quantum dots is not very uniform. The transition
dipole moments are, however, almost unaffected by the
differing environments. This underlines the strength of optical
spectroscopy on the level of single nanoobjects.
Perturbed Free Induction Decay. While the Rabi
oscillations were measured as a function of pump power, we
can also follow the dynamics of the exciton in the time domain.
Figure 3. (A) Each vertical cut through the color map represents a A classical example is the perturbation of the free induction
ΔR/R spectrum at a given pump power and a fixed pump−probe delay decay by the pump pulse.21−23 In a density matrix picture the
Δt = 150 ps. The broad background has been subtracted. (B) By probe pulse creates coherence between the states |00⟩ and |01⟩.
fitting the dispersive line shape function to the spectra in (A), the The exponential decay of the coherence leads via Fourier
signal amplitude (blue dots) is extracted. The error bars are the transform to the Lorentzian spectral line of the scattered light.
standard deviation of four consecutive measurements. The scale of the
pump pulse area θp is calibrated to the first maximum. The data can be
When a pump pulse (coming after the probe pulse) perturbs
described by sin2(θ/2), considering an effective, reduced pulse area θeff the decay of the coherence in some way, the spectral shape
due to delocalized carriers around the quantum dot (red curve). changes, resulting in a differential reflection signal ΔR/R.
Figure 4 presents transient reflection spectra for pump−
at a certain pump power. In Figure 3B, the signal amplitudes probe delays varying from −30 ps to +30 ps. Positive delays
are extracted from the spectra by fitting a dispersive line shape describe a pump pulse that precedes the probe pulse, as was
function to the spectral data. Two complete cycles of the Rabi used to acquire Rabi oscillations where the pump pulse depletes
oscillation are visible. In contrast to the textbook example, they the ground state. We find a dispersive spectral feature that
decays with the exciton lifetime on a 200 ps time scale. As in
show a gradual stretching of the oscillation period. We find the
previous work,21−23 for negative delays the spectra show fringes
origin of this stretching in the carriers that the pump pulse that separate further with reducing pump−probe delay,
creates in the GaAs layers near the quantum dot. The density of resulting in hyperbolic features.
these carriers influences the overall reflectivity of the sample The whole data set can be simulated using a density matrix
and thus is responsible for the broad background signal34,35 model of only three levels (|00⟩, |01⟩, |10⟩) that are optically
(Figure 2). This signal follows a saturation law from which we connected in a V shape. We find perfect agreement in all details
can determine a saturation power Psat from it (see Supporting when taking into account the interference of the scattered and
Information). Using this saturation power, we can fully describe reflected probe field. The only free fitting parameters are the
455 dx.doi.org/10.1021/nl203804n | Nano Lett. 2012, 12, 453−457
Nano Letters Letter

and the effect of their coupling. Similar to the spectral signature


of the delocalized carriers presented above, the response of any
nanoenvironment can be taken into account when modeling
the coupled system. Transient reflection spectroscopy is not
limited to epitaxial samples. Colloidal nanocrystals or even
molecules could be investigated when prepared on a partially
reflecting substrate such as a cover slide. Our technique will
make coherent experiments feasible in a large group of
structures where other techniques are very difficult or
impossible to apply.


*
ASSOCIATED CONTENT
S Supporting Information
Growth of the GaAs/AlGaAs quantum dots. Details on the
experimental setup. Additional data sets for Rabi oscillations
and the background contribution. Density matrix model for a
V-system and calculation of differential reflectivity spectra.
Additional corresponding references. This material is available
free of charge via the Internet at http://pubs.acs.org.


Figure 4. (Top) Differential spectra as a function of pump−probe
delay in experiment and simulation. (Bottom) Temporal evolution of
the probed exciton polarization for the probe preceding the pump AUTHOR INFORMATION
pulse (negative delays, left figure) and the probe arriving after the
Corresponding Author
pump pulse (positive delays, right figure). In the first case, the pump
*E-mail: m.lippitz@physik.uni-stuttgart.de.


perturbs the polarization, leading to spectral fringes. In the second
case, the pump depletes the ground state, reducing the probed
polarization. ACKNOWLEDGMENTS
Our research was supported by the Deutsche Forschungsge-
coherence time T2 = 90 ps and the lifetime T1 = 230 ps of the meinschaft DFG (research unit FOR 730) and the Baden-
probed transition (for details see the Supporting Information). Württemberg Stiftung.


In the density matrix model, the scattered probe field equals
the imaginary part of the matrix element ρ|00⟩ |01⟩, describing the REFERENCES
probed exciton polarization. The lower row of Figure 4 depicts (1) Shields, A. J. Nat. Photonics 2007, 1, 215−223.
examples of the temporal evolution for negative and positive (2) Ramsay, A. Semicond. Sci. Technol. 2010, 25, 103001.
pump−probe delay. For positive delay the pump pulse bleaches (3) Michler, P.; Kiraz, A.; Becher, C.; Schoenfeld, W.; Petroff, P.;
the ground state |00⟩ so that less polarization is built up, Zhang, L.; Hu, E.; Imamoglu, A. Science 2000, 290, 2282.
resulting in a weaker scattered field and a dispersive feature in (4) Li, X. Q.; Wu, Y. W.; Steel, D.; Gammon, D.; Stievater, T. H.;
the transient reflection spectrum. For negative pump−probe Katzer, D. S.; Park, D.; Piermarocchi, C.; Sham, L. J. Science 2003, 301,
delays, the decay of the polarization is perturbed by the pump 809−811.
pulse as it bleaches the ground state. The resulting polarization (5) Badolato, A.; Hennessy, K.; Atatüre, M.; Dreiser, J.; Hu, E.;
trace can be seen as the sum of a reduced free induction decay Petroff, P. M.; Imamoglu, A. Science 2005, 308, 1158−1161.
(6) Kiraz, A.; Michler, P.; Becher, C.; Gayral, B.; Imamoglu, A.;
(dashed line in Figure 4C) and a rectangle with a width equal Zhang, L.; Hu, E.; Schoenfeld, W. V.; Petroff, P. M. Appl. Phys. Lett.
to the pump−probe delay (shaded area in Figure 4C). The 2001, 78, 3932−3934.
latter leads by Fourier transform to the fringes in the spectral (7) Akimov, A. V.; Mukherjee, A.; Yu, C. L.; Chang, D. E.; Zibrov, A.
domain. Like in the case of Rabi oscillations, this example S.; Hemmer, P. R.; Park, H.; Lukin, M. D. Nature 2007, 450, 402−406.
demonstrates again the advantage of a spectrally resolved (8) Fedutik, Y.; Temnov, V. V.; Schoeps, O.; Woggon, U.; Artemyev,
experiment. Only the rich information content of the M. V. Phys. Rev. Lett. 2007, 99.
differential spectra makes it possible to distinguish between (9) Pfeiffer, M.; Lindfors, K.; Wolpert, C.; Atkinson, P.; Benyoucef,
the quantum dot signal, perturbed free induction decay, and M.; Rastelli, A.; Schmidt, O. G.; Giessen, H.; Lippitz, M. Nano Lett.
background contributions. 2010, 10, 4555−4558.
(10) Curto, A. G.; Volpe, G.; Taminiau, T. H.; Kreuzer, M. P.;
Conclusion. We demonstrated that simple far-field Quidant, R.; van Hulst, N. F. Science 2010, 329, 930−933.
reflection spectroscopy is sufficient to determine the ultrafast (11) Gonzalez-Tudela, A.; Martin-Cano, D.; Moreno, E.; Martin-
dynamics of a single self-assembled quantum dot. As one Moreno, L.; Tejedor, C.; Garcia-Vidal, F. J. Phys. Rev. Lett. 2011, 106,
example, we determined the dipole moment of strain-free GaAs 020501.
quantum dots on an absorbing substrate by Rabi oscillations to (12) Chang, D.; Sorensen, A.; Demler, E.; Lukin, M. Nat. Physics
be 17−19 D. We also showed spectrally resolved perturbed free 2007, 03, 807−812.
induction decay of a localized exciton in full agreement with a (13) Lateral Alignment of Epitaxial Quantum Dots; Schmidt, O. G.,
density matrix model involving only a three level V-system. We Ed.; Springer: New York, 2007.
expect our method to find applications in ultrafast coherent (14) Atkinson, P.; Kiravittaya, S.; Benyoucef, M.; Rastelli, A.;
Schmidt, O. G. Appl. Phys. Lett. 2008, 93.
spectroscopy of complex quantum-optical structures, such as (15) Sotier, F.; Thomay, T.; Hanke, T.; Korger, J.; Mahapatra, S.;
quantum emitters in photonic or plasmonic circuits. Without Frey, A.; Brunner, K.; Bratschitsch, R.; Leitenstorfer, A. Nat. Phys.
any further requirements on the sample design, the transient 2009, 5, 352−356.
spectra contain enough information to separate the individual (16) Zecherle, M.; Ruppert, C.; Clark, E.; Abstreiter, G.; Finley, J.;
contributions of the quantum dot and the nanoenvironment Betz, M. Phys. Rev. B 2010, 82, 125314.

456 dx.doi.org/10.1021/nl203804n | Nano Lett. 2012, 12, 453−457


Nano Letters Letter

(17) Kim, E. D.; Truex, K.; Wu, Y.; Amo, A.; Xu, X.; Steel, D. G.;
Bracker, A. S.; Gammon, D.; Sham, L. J. Appl. Phys. Lett. 2010, 97,
113110.
(18) Stufler, S.; Ester, P.; Zrenner, A.; Bichler, M. Phys. Rev. Lett.
2006, 96, 037402.
(19) Stievater, T. H.; Li, X. Q.; Steel, D. G.; Gammon, D.; Katzer, D.
S.; Park, D.; Piermarocchi, C.; Sham, L. J. Phys. Rev. Lett. 2001, 8713,
133603.
(20) Kamada, H.; Gotoh, H.; Temmyo, J.; Takagahara, T.; Ando, H.
Phys. Rev. Lett. 2001, 8724, 246401.
(21) Fluegel, B.; Peyghambarian, N.; Olbright, G.; Lindberg, M.;
Koch, S. W.; Joffre, M.; Hulin, D.; Migus, A.; Antonetti, A. Phys. Rev.
Lett. 1987, 59, 2588−2591.
(22) Joffre, M.; Hulin, D.; Migus, A.; Antonetti, A.; Laguillaume, C.
B. A.; Peyghambarian, N.; Lindberg, M.; Koch, S. W. Opt. Lett. 1988,
13, 276−278.
(23) Guenther, T.; Lienau, C.; Elsaesser, T.; Glanemann, M.; Axt, V.;
Kuhn, T.; Eshlaghi, S.; Wieck, A. Phys. Rev. Lett. 2002, 89.
(24) Jackson, J. D. Classical Electrodynamics, 2nd ed.; Wiley: New
York, 1975.
(25) Karrai, K.; Warburton, R. J. Superlattices Microstruct. 2003, 33,
311−337.
(26) Plakhotnik, T.; Palm, V. Phys. Rev. Lett. 2001, 87.
(27) Alén, B.; Bickel, F.; Karrai, K.; Warburton, R.; Petroff, P. Appl.
Phys. Lett. 2003, 83, 2235−2237.
(28) Langbein, W.; Patton, B. Opt. Lett. 2006, 31, 1151−1153.
(29) Rastelli, A.; Stufler, S.; Schliwa, A.; Songmuang, R.; Manzano,
C.; Costantini, G.; Kern, K.; Zrenner, A.; Bimberg, D.; Schmidt, O. G.
Phys. Rev. Lett. 2004, 92, 166104.
(30) Plumhof, J. D.; Křaṕ ek, V.; Wang, L.; Schliwa, A.; Bimberg, D.;
Rastelli, A.; Schmidt, O. G. Phys. Rev. B 2010, 81, 121309.
(31) Oudar, J.; Hulin, D.; Migus, A.; Antonetti, A.; Alexandre, F.
Phys. Rev. Lett. 1985, 55, 2074−2077.
(32) Kim, S.; Oh, E.; Lee, J. U.; Kim, D. S. J. Korean Phys. Soc. 2006,
49, 1611−1614.
(33) Unold, T.; Mueller, K.; Lienau, C.; Elsaesser, T. Semicond. Sci.
Technol. 2004, 19, S260.
(34) Gupta, S.; Frankel, M.; Valdmanis, J.; Whittaker, J.; Mourou, G.;
Smith, F.; Calawa, A. Appl. Phys. Lett. 1991, 59, 3276−3278.
(35) Frankel, M.; Carruthers, T. Appl. Phys. Lett. 1994, 64, 1950−
1952.
(36) Loka, H. S.; Benjamin, S. D.; Smith, P. W. E. Opt. Commun.
1998, 155, 206−212.
(37) Minot, C.; Chavignon, J.; Leperson, H.; Oudar, J. Solid State
Commun. 1984, 49, 141−143.

457 dx.doi.org/10.1021/nl203804n | Nano Lett. 2012, 12, 453−457

You might also like