Letter pubs.acs.org/NanoLett
Full Spectral and Angular Characterization of Highly Directional Emission from Nanocrystal Quantum Dots Positioned on Circular Plasmonic Lenses Moshe G. Harats,*,† Nitzan Livneh,‡ Gary Zaiats,¶ Shira Yochelis,‡ Yossi Paltiel,‡ Efrat Lifshitz,¶ and Ronen Rapaport†,‡ †
Racah Institute of Physics and ‡The Department of Applied Physics, Selim and Rachel Benin School of Engineering and Computer Science, The Hebrew University of Jerusalem, Jerusalem 91904, Israel ¶ Schulich Faculty of Chemistry, Russell Berrie Nanotechnology Institute, Solid State Institute, Technion-Israel Institute of Technology, Haifa 32000, Israel S Supporting Information *
ABSTRACT: We design a circular plasmonic lens for collimation of light emission from nanocrystal quantum dots at room temperature in the near IR spectral range. We implement a two-dimensional k-space imaging technique to obtain the full spectral-angular response of the surface plasmon resonance modes of the bare plasmonic lens. This method is also used to map the full spectral-angular emission from nanocrystal quantum dots positioned at the center of the circular plasmonic lens. A narrow directional emitting beam with a divergence angle of only ∼4.5° full width at half-maximum is achieved with a spectrally broad bandwidth of 30 nm. The spectrally resolved k-space imaging method allows us to get a direct comparison between the spectral-angular response of the resonant surface plasmon modes of the lens and the emission pattern of the quantum dots. This comparison gives a clear and detailed picture of the direct role of these resonant surface waves in directing the emission. The directional emission effect agrees well with calculations based on the coupled mode method. These results are a step toward fabricating an efficient room-temperature single photon source based on nanocrystal quantum dots. KEYWORDS: Plasmonic lens, nanocrystal quantum dot, k-space spectroscopy, angular-resolved spectroscopy, directional emission, surface plasmons
N
conventional bulk optics is extremely challenging. Another limiting factor is the relatively long lifetime of the emission, which results in an uncertainty of the emission time of the photon. One possible solution for both challenges is the incorporation of light-emitting NQDs with plasmonic nanoantennas. Plasmonic nanoantennas, consisting of nanostructured metal− dielectric interfaces, have been shown to enhance local electromagnetic (EM) fields of light,6,7 as well as to control the direction of scattered light.7,8 These properties can be used to manipulate the lifetime and direction of the emission from nanoemitters in general and from NQDs in particular. Indeed, in recent years there have been several very promising demonstrations of the modification of the spontaneous emission rate and the emission pattern of NQDs positioned in the vicinity of plasmonic nanoantennas. These experiments
anocrystal quantum dots (NQDs) are promising as a source of deterministic single photons on-demand.1,2 Their discrete electronic energy states and the high quantum yield (QY) of the optical transitions between those states give rise to efficient emission of single photons.3 The material, size, and shape control during their growth give a unique design flexibility of the emission wavelength, lifetime, and polarization.4 One of the significant technological advantages of using NQDs as single photon emitters is their ability to efficiently emit single photons even at room temperature. The fact that NQDs are “substrate free” allows additional capabilities of integration on various platforms and material systems. A main obstacle for realizing a practical deterministic single photon source from NQDs is the ability to efficiently collect the emitted photons. A high collection efficiency with a low numerical aperture (NA) is critical as almost any envisioned application utilizing single photons requires a very high collection efficiency of the emitted photons.5 As the emission from NQDs is nondirectional, collecting it very efficiently with © XXXX American Chemical Society
Received: July 13, 2014 Revised: August 20, 2014
A
dx.doi.org/10.1021/nl502652k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
utilized different antenna geometries such as a patch antenna,9 a Yagi-Uda nanoantenna,10 and a split-ring resonator antenna.11 Because one common feature of all these antennas is that they are based on a dipole emission from scattered surface plasmon (SP) modes,9,10 or rely on asymmetric multipolar interference,11 they all yield either a relatively large divergence angle or a large anisotropy. Recently, we have demonstrated that the emission from NQDs can be directed if they are placed near a subwavelength linear metallic grating with a very small divergence angle of ∼3.4° along the grating dimension.12 This type of highly directional emission is due to efficient coupling to the resonantly confined modes of the grating (either surface plasmon or optical waveguide modes) combined with coherent interference of these modes when they are coupled to free propagating EM radiation. The combined effect induces highly directional emission. However, in the linear grating case the high directionality was only on a plane. A more symmetric design that has shown collimation of visible light is that of a circular plasmonic lens. In this case, the combined effect of coupling to SP mode and coherent interference of the scattered SP modes to propagating radiation13 can have a circular symmetry. Indeed, it was shown that a circular plasmonic lens can collimate transmitted incoherent light,8 the emission of the output of a laser,14 and the emission of molecules immersed in solution.15 A circular plasmonic lens was also used to show a proof-of-principle collimation of NQDs emission.16 However, no detailed characterization of the spectral-angular response of the NQDs emission was shown. Also, no comparison between the NQDs emission to the response of intrinsic resonant SP modes of the bare lens was done, so that a full quantitative understanding of the collimation effect mechanism, and its comparison to theory is still missing. We also note that in many previous works, the angular response of either the passive plasmonic elements and of the emission of emitters coupled to these elements were measured using a goniometric approach, which requires tilting the detectors and the sample. This approach is very sensitive to alignment errors, and because each angle is measured separately it requires very long measurement times especially for weak emission from few nanoemitters. The angular resolution is limited by the mechanical steps of the goniometric stage, and it also limits the ability to spectrally resolve the angular emission signals. A more suitable technique is that of spectrally resolved k-space imaging. k-space imaging technique was used for plasmonic structures in a few recent works10,11,16 where a specific spectral range was selected using bandpass filters. In order to characterize, understand, and optimize the collimation effect of the NQDs emission from the plasmonic lens, it is important to first fully characterize the passive response of the plasmonic lens in both the spectral and angular domains. This can be done by applying a spectrally resolved k-space imaging technique, which can fully map the SP resonances of the plasmonic lens. Then a similar spectral-angular characterization should be done to the emission of the NQDs, followed by a comparison between the two. This comparison should yield insights on which SP modes are responsible for the collimation of the emission. This can then be used to understand and optimize the performance of such devices. In this paper, we use a k-space imaging technique to show a full characterization of a device of NQDs embedded on a plasmonic lens. We demonstrate the characterization of the SP modes of the plasmonic lens, both in the spectral and angular domains, and a direct comparison to the directional emission
from the NQDs that are positioned at and in proximity to the center of a circular plasmonic lens. The comparison between the two k-space measurements allows us to directly show a very narrow directional emission with a divergence of only ∼4.5° full width at half-maximum (fwhm) and with a broad spectral bandwidth. We show that the emission properties of the NQDs are governed by the enhanced coupling to the SP fields propagating from the center of the lens and coherently scatter from the entire lens. We also perform spatially and spectrally resolved lifetime measurements on the NQD emission. This reveals a reduction of approximately 10% in the NQDs lifetime due to the moderate Purcell effect induced by the periodic structure of the lens. Our samples consist of an Au substrate, on which the plasmonic lens is fabricated by e-beam lithography followed by evaporation of Au (Figure 1a). Two samples are investigated. In
Figure 1. (a) An SEM image of the plasmonic lens. (b) A cross section of the plasmonic lens (along the dashed line in (a)) describing the geometrical dimensions of the lens. m = 412 nm, the width of the central cavity. a = 190 nm, d = 100 nm, p = 787 nm, width, depth, and period of the side grooves, respectively. f = 522 nm, distance between the main cavity to the first groove. The thin blue layer is a SiO2 layer. (c) The emission spectra of the CdSeTe/ZnS NQDs (blue line) and halogenated PbS/Cl NQDs (green line). (d) The k-space measurement setup (see Supporting Information for more details).
the first (sample A), the Au lenses are covered by a 10 nm thick SiO2 layer. NIR CdSeTe/ZnS NQDs (Life Technologies, Qdot 800 ITK amino PEG) with a peak photoluminescence (PL) at 785 nm are chemically attached (see Supporting Information) to a SiO2 layer (the blue line in Figure 1c shows their corresponding PL spectrum). Sample B is covered by a 50 nm thick SiO2 layer and halogenated PbS/Cl synthesized NQDs (see Supporting Information) with a peak PL at 880 nm are spin-coated on the sample (the green line in Figure 1c). The SiO2 layer buffers the NQDs from the Au to avoid quenching of the NQDs PL due to nonradiative energy transfer to the metal.9 In order to achieve the best collimation effect of the emission of NQDs from the center of the plasmonic lens, the different geometrical parameters of the elements of the lens denoted in Figure 1b are optimized by a set of simulations (see Supporting Information). For a resonant wavelength λ = 800 nm, we find that the optimum diameter of the main cavity (m = 412 nm) should be approximately equal to λ/2. The optimum grooves period is found to be p = 787 nm, which is slightly smaller than λ. The optimal depth was found to be d = 115 nm but a value of d = 100 nm (≪λ) was used as a result of limitations in the B
dx.doi.org/10.1021/nl502652k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
polar angles (θ,φ), as defined in the schematic drawing in Figure 2a,b. The response of three different wavelengths are presented as the following: Figure 2c shows the designed resonant wavelength (830 nm) and Figure 2d,e shows two nonresonant wavelengths (900 and 970 nm), respectively. We find that around 830 nm there is a maximum in the reflection at a small angular cone perpendicular to the lens surface, which is a result of a resonance that is excited and re-emitting with a narrow angular central cone, as is expected from the design of the plasmonic lens. As the wavelength is moved away from the resonance, the narrow angular cone of maximum reflection evolves into a circle with a growing radius, and a reflection minimum appears at the center. Next we excite the NQDs attached to sample A with a tightly focused HeNe laser at 632.8 nm and measure the emission of the NQDs with the same k-space measurement setup. In Figure 3a, we show a measured spectral-angular emission map showing
fabrication process. The distance from the main cavity to the first side groove, f = 522 nm, is optimized so that constructive interference of the re-emission from the main cavity and from the side grooves would occur perpendicularly to the lens surface.17 To verify that this optimization will indeed result in a directional emission from a nanoemitter embedded in the center of the lens, we calculate numerically the far-field angular distribution of a plane wave emerging from the main cavity of the lens as a function of its wavelength. The calculation is based on the coupled mode method (CMM).13,18 This calculation shows a clear resonant directional emission around 820 nm with a calculated angular fwhm of less than 4° and a spectral fwhm of 27 nm, as is shown in Figure 3b. In order to experimentally measure the spectral and angular response of the passive (without any NQDs) and active (with NQDs) plasmonic lens we use a spectral-angular k-space imaging technique (see detailed description in the Supporting Information). This allows a direct, fast, and high-resolution measurement of both the spectral and angular response of the lens without the need of rotating the samples or the detectors and without the need for a different measurement for each angle, as is done in the goniometric method (see, e.g., refs 8 and 12). We first perform a two-dimensional k-space reflection spectroscopy of the passive plasmonic lens antenna using white light. In this measurement, the angular reflection of white light focused on the plasmonic lens (without attached NQDs) is measured with unpolarized light and is normalized to the reflection from a flat Au surface. It is important to note that the illumination should be very precisely centered at the middle of the plasmonic lens19 and with the spot size smaller than the lateral diameter of the lens (∼10 μm). In Figure 2, we present the unpolarized, spectrally resolved angular reflection intensity from a plasmonic lens of sample A as a function of the two
Figure 3. (a) The normalized spectral-angular emission of NQDs from the center of the plasmonic lens at φ = 0. A clear resonant directional emission is observed around 820 nm. The spectral width of the directional emission is ∼20 nm. (b) The normalized angular emission pattern for λ = 820 nm and φ = 0° corresponding to the dashed black line in (a) (green line). The calculation for the same wavelength corresponding to the dashed white line in (d) (blue dashed curve). A clear directional emission with a fwhm of ∼4.5° is observed with good agreement with the calculation. (c) The FDTD calculation of a single dipole emitting from the center of the main cavity (green curve), and the contribution of an ensemble of dipoles (NQDs) on a circular area excited by the laser spot (blue curve). (d) The CMM calculation spectral-angular emission from the main cavity of the plasmonic lens. (e) Corresponding reflection k-space measurement for φ = 0. The dashed ellipses in (a,d,e) show the directional lobe. The doubleheaded arrows in (a,d,e) point at the converging emission modes.
the PL intensity as a function of λ and θ (for φ = 0°, along the entrance slit of the spectrometer) with linear polarization aligned to the entrance slit of the spectrometer (i.e., φ = 0°). This emission is normalized to the emission of NQDs from another area of the same sample where there is a flat Au surface covered with the same thickness of SiO2. A clear directional emission collimated perpendicular to the lens surface is observed. The collimation effect is visible only when the NQDs close to the center are optically excited by the focused laser, while excitation of NQDs outside the center of the lens does not show any collimation effect. We show a cross section of the angular emission pattern for λ = 820 nm in Figure 3b that shows a record low divergence of ∼4.5° fwhm. The experimental directional emission is compared to the calculated
Figure 2. (a) Definition of the polar angles θ, φ on the unit sphere. (b) The representation of the polar axis corresponding to the spectralangular images in (c−e) and in Figure 4. The whole 2π angle of the azimuthal angle φ is measured. θ is limited by the NA of the objective. In our case, θmax = θNA = 40°. (c−e) Normalized reflection measurements of sample A. (c) At a resonant wavelength (830 nm) a plasmon resonance with a cone around θ,φ = 0° is observed with reflectance of 98% compared to 70% in the surrounding angles. The dashed circles represent the lines of constant polar angle θ. (d,e) Nonresonant cases at 900 nm (c) and 970 nm (d). C
dx.doi.org/10.1021/nl502652k | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
spectral-angular emission from the plasmonic lens using the CMM described above, which is plotted in Figure 3c, and also to the corresponding φ = 0° spectral angular reflection measurement of the bare lens that is shown in Figure 3d. A good agreement between the emission, reflection, and calculation is observed. The emission pattern in Figure 3a shows two angular lobes at shorter wavelengths (marked by the double headed arrows) that converge into one directional lobe at θ = 0° around 820 nm. These converging lobes are also visible in the CMM calculation (Figure 3c) and in the reflection measurement (Figure 3d.) This comparison shows the underlying mechanism of the directional emission of the NQDs: the enhanced emission clearly follows the spectralangular response of the resonant SP modes of the plasmonic lens. Importantly, the effects of the losses of the propagating SP modes are included in the CMM calculation that takes into account the realistic metal parameters. Indeed, the main decay of the SP modes as they propagate along the lens surface is due to their rescattering to propagating EM modes via their interaction with the lenses’ grooves. The actual metal loss adds some additional decay that causes a slight angular broadening of the emission profile. This is shown explicitly in the Supporting Information. Note that in Figure 3b there is an appreciable emission background into a wide angular range. This is due to the spot size of the illuminating laser. As the main cavity diameter is m = 412 nm and the focused HeNe spot size is ∼1 μm, we can estimate that ∼90% of the excited NQDs are outside of the main cavity (see Supporting Information). These NQDs exhibit a lower directional emission and thus contribute to the experimental nondirectional background. We can at least partially verify this explanation using FDTD simulations20 with a near-field to far-field transformation21 of the dipolar emission. In Figure 3c, we show that the calculated emission pattern of a single dipole emitting from the center of the lens has low side lobes and additional nondirectional background (green curve). By taking into account the weighted incoherent contribution from many NQDs radially distributed around the center of the lens over a circular area of the laser excitation spot (blue curve), we find that the collimated cone does not change significantly while the side lobes are suppressed, and the background increases and flattens, as is observed in the experiment. We note that the simulations predicts a background that is half the measured one, so there might be other effects not taken into account in the simulation, such as vertical distribution of the NQD positions above the lens surface. The full angular emission pattern of the NQDs in proximity to the center of the plasmonic lens in sample B is presented in Figure 4a−c for three different emission wavelengths. A clear collimation of the NQD emission is observed at a resonant wavelength of λ = 835 nm with an almost perfect circular symmetry as is seen in Figure 4b. The divergence angle in this sample is slightly larger with a fwhm of 5°. The shift from the 820 nm expected resonance is ascribed to the thicker SiO2 layer that slightly changes the dielectric layer properties of the simulation. This is compared to nonresonant wavelengths shown in Figure 4a,c that exhibit an almost isotropic emission pattern. In order to quantify the efficiency of the collimation effect, we define the directivity (a dimensionless figure of merit) of the plasmonic lens emission. The directivity is defined in eq 1 as the ratio of the averaged intensity per unit solid angle measured in the solid angle cone Ωcone = 2π(1−cos θcone) where the
Figure 4. (a−c) Unpolarized and unnormalized two-dimensional emission of NQDs at different wavelengths. The collimation effect is evident at 835 nm. The colorbars represent the count per pixel. The dashed circles in (b) represent the polar angle θ. (d) The directivity (eq 1) as a function of the wavelength. At resonance, there is a higher probability to emit photons into the small angular cone.
directional lobe is evident (θcone is the fwhm extracted from Figure 3b), divided by the average emission intensity per unit solid angle of the total emission to an angle of 2π (as we have a mirror in our sample). This yields θ
2π
∫0 cone ∫0 I(θ , φ)dθ dφ
DΩcone =
2π (1 − cos θcone) π
2π
∫0 ∫0 I(θ , φ)dθ dφ 2π
(1)
The directivity is 1 when no collimation occurs, as there is no difference between the average emission into Ωcone compared to all other angles, and higher than 1 when there is a collimated beam. The directivity of the NQDs in the plasmonic lens as a function of the emission wavelength is plotted in Figure 4d. A peak is observed at the resonance of the plasmonic lens around 835 nm where there is 40% more probability to emit photons into the narrow collimated cone. The measured directivity at resonance is limited by the nondirectional background resulting from the appreciable excitation of NQDs outside the main cavity, as explained previously. Next we measure the temporal response of the emission of the NQDs as a function of their position on the lens. The NQDs on sample A were excited by a tight spot of a 100 fs laser pulses at 405 nm, with a spot size
