1. Introduction

Integration of electronic and optical devices in the 100nm scale poses a challenge for optics, requiring sub-wavelength confinement, guiding, manipulation, emission and detection of light. A possible way to face this challenge is by using metal-dielectric plasmonic structures [1–3], which offer sub-wavelength optical confinement, and are relatively easy to manufacture on glass or semiconductor in a fab-compatible environment. A nano-antenna [4,5] is designed to effectively couple the electromagnetic field generated by an emitter located in its near-field, to the far-field wave that can be collected and further used [6–8]. In addition, the high intensity of vacuum electric field confined at the vicinity of the nano-antenna modifies the density of states to which a close-by emitter can couple, therefore increasing the free space emission rate [9,10]. Properly designed metallic nano-particles and structures, incorporating either small size, sharp corner or narrow gap, have more significant local field enhancement, thus enhancing emission and absorption rates. On the other hand, larger size particles have greater scattering to extinction cross-section ratio and better coupling to the far-field. Nano-antenna design should maximize both characteristics.

Surface plasmon enhanced spontaneous emission was demonstrated in various systems: metal coated nanoshell [11], CdSe/ZnSe nanocrystals [12] and InGaN single quantum well (QW) [13,14] coupled to SPP waves propagating on a thin metal layer; Si-nanocrystals coupled to periodic metallic gratings [15]; Si Quantum Dots (QD) embedded in silica [16], and fluorescent dye molecules [17] coupled to localized plasmon fields of a single plasmon nano-antenna, or of an array of nano-antennas [18,19], and Er⁺³ emitters coupled to a quasi-periodic array [20]. While previous experiments were performed mainly on discrete emitters such as molecules or quantum dots, on low index substrate, emitting at visible or near-visible IR wavelengths of 400 to 900nm, this work focuses on a continuous semiconductor gain medium, InP-based multi quantum well (MQW) structure operating at room-temperature and emitting in the telecom wavelength regime of 1.5μm, coupled to an array of gold nano-antennas. Using a continuous MQW gain medium simplifies the sample manufacturing as no alignment of the emitter to the nano-antennas is required, moreover enhanced emission from a larger area can be accumulated.

We demonstrate enhancement of light emission from optically-pumped InP MQW using different arrays of gold single-strip and double-strip nano-antennas, selectively tuning the peak of enhanced emission according to nano-antenna resonance wavelength. Comparison to theoretical model shows that most of the enhancement is attributed to a reduction in the radiative recombination lifetime in the QW, which paves the way to the development of high-speed directly-modulated nano non-laser emitters.

2. Measurement results

Nano-antennas shape was chosen to be of short strip, and of two coupled short strips with a narrow gap in between [4,17–19,21,22], as elongated shape and high refractive index of our semiconductor substrate red-shifts the resonance for polarization parallel to the long axis [23,24]. The advantage of the double-strip structure is that the electric field is greatly enhanced in the narrow gap between the strips which enhances the emission rate, and the overall nano-antenna scattering cross-section is greater than a single-strip. An estimation of nano-antenna size resulted from a quasistatic analysis [24] of a gold ellipsoid embedded in InP, yielding a resonance at 1500nm for a particle length of 250nm. Following FDTD simulations (using the Lumerical software package) with the parameters of the manufactured antenna: width of 100nm and thickness of 60nm, yielded more accurate resonance values of 1400nm (Fig. 1(a) ) and 1640nm for 200nm and 250nm long single-strip antennas respectively. Simulations of the double-strip antennas exhibited enhanced field in the gap (Fig. 1(d)), and a spectrum showing a small 10nm red-shift for a gap size of 50nm (Fig. 1(b)).

 

Fig. 1 FDTD Simulated transmission and reflection spectrum of (a) 200nm single-strip, and (b) 2x 200nm double-strip with 50nm gap, and (c,d) Electric field amplitude respectively.

Download Full Size | PDF

The samples were grown by metal organic molecular beam epitaxy (MOMBE) on an InP substrate. The layer structure consisted of 4 InGaAs/InP 8nm/20nm QWs, with a 20nm intrinsic InP layer ending the QW stack (Fig. 2(a) ). Nano-antenna arrays were fabricated on top of the InP MQW stack by E-Beam Lithography, followed by evaporation of gold, and a subsequent lift-off process. Resulting nano-antennas dimensions were 60nm thick, ~100nm wide, lengths in the range 200-350nm, and gaps of 50 and 100nm for the double-strip antenna (Fig. 2(b)–2(d)). PL from the optically pumped sample by a 532nm laser showed strong emission with a peak at 1540nm.

 

Fig. 2 (a) Schematic of the Au double-strip nano-antenna on InP sample. SEM images of the nano-antennas on InP MQW, (b) 200nm single-strip, and (c,d) 2x200nm double-strip.

Download Full Size | PDF

Nano-antenna transmission spectrum measurements of the un-pumped nano-antennas, normalized to a reference having no nano-antennas, were performed using a white light source for both parallel and perpendicular polarizations relative to the long-axis of the antenna. Subsequently PL spectrum from the optically-pumped MQWs in the nano-antennas array region was measured, and normalized to the PL from a close region with no nano-antennas, resulting in the PL enhancement spectrum. Nano-antenna normalized transmission spectrum (Fig. 3(a) , 3(b), 4(a) , 4(b)), and respective PL and PL enhancement measurements (Fig. 3(c), 3(d), 4(c), 4(d)), are shown below for the single-strip and double-strip nano-antennas.

 

Fig. 3 (a,b) Normalized transmission spectrum of double-strip nano-antennas (a) 2x 200nm, (b) 2X250nm, with gap of 50nm (red line) and 100nm (blue line). (c,d) PL enhancement spectrum for respective nano-antennas (blue), emission from reference (dotted), and emission from nano-antenna region (full).

Download Full Size | PDF

 

Fig. 4 (a,b) Normalized transmission spectrum of short-stripe nano-antennas (a) 200nm, (b) 250nm long. (c,d) PL enhancement spectrum (blue), emission from reference (dotted), and emission from nano-antenna region (full).

Download Full Size | PDF

The nano-antennas measured resonance wavelengths were in good agreement with simulations results. PL enhancement of ~9 was measured for the 2x200nm double-strip antenna with 50nm gap (Fig. 3(c)), at a wavelength of 1340nm, corresponding to the resonance wavelength of the nano-antenna of 1370nm (Fig. 3(a)). The larger 100nm gap antenna gave a 'shallower' resonance and a smaller enhancement of ~5.2. For the 2x250nm double-strip antenna with 50nm gap enhancement of ~7.5 was measured (Fig. 3(d)), with shift in PL enhancement peak to ~1600nm relative to the antenna resonance wavelength of 1560nm (Fig. 3(b)). The single-strip nano-antenna gave maximum PL enhancement of ~5.2 (Fig. 4(c)), again with peak enhancement corresponding to the nano-antenna resonance wavelength.

3. Theoretical modeling and simulation

We used the theoretical model presented by Sun et al. [8] together with FDTD simulation results to interpret the experimental results and estimate the physical parameters involved. The plasmonic enhancement of the vacuum electric field by the nano-antenna modifies the density of states available for emission, thus increasing the radiative recombination time of the MQW by the Purcell-factor, τ^' _rad = τ_rad/F_P, and enhancing the measured PL. For InGaAs/InP QW's at room temperature non-radiative recombination dominates, with τ_nrad ~1ns and τ_rad/τ_nrad ~10 (values depending on carrier concentration, and material and sample details), leading to a spontaneous radiative emission efficiency ${\eta }_{rad}=1/(1+{\tau }_{rad}/{\tau }_{nrad})$ of ~9% [25]. The emitter is coupled to the plasmon modes of the nano-antenna, which are then coupled to radiation modes with an efficiency η_pr. The overall radiative recombination efficiency in the presence of plasmon nano-antennas is given by [8]: ${\eta }_{sp}={\eta }_{pr}\cdot \frac{F_P{\tau }_{rad}^{-1}}{F_P{\tau }_{rad}^{-1}+{\tau }_{nrad}^{-1}}$ and the enhancement in the PL is given by the ratio η_SP/η_rad .

Part of the enhancement measured in the experiment comes from the angular gain of the nano-antenna, namely the narrowing of the native PL angular distribution due to the interaction with the nano-antenna [26], causing increased light collection by the limited numerical-aperture (0.85) of our objective lens. Without nano-antennas 3.5% of the uniformly distributed emitted light (η_ext = (N.A./n_InP)²/2) is collected by the objective lens. FDTD simulations of the nano-antennas shows a narrower angular distribution, increasing the lens collection efficiency by a factor of F_lens ^(D) = 2.2 for the double-strip nano-antenna, and F_lens ^(S) = 1.5 for the single-strip. Correcting the maximum measured PL enhancement by this antenna-gain factor, results in radiative enhancement factor of 3.2 for the single-strip antenna, and of 4 for the double-strip antenna, which is 25% better than the single-strip.

PL can also be increased by resonant enhancement of the pump by the nano-antenna. However for these nano-antennas with resonances at 1360nm and 1540nm, no enhancement of the excitation at 532nm occurs.

The overall enhancement in measured PL, F_PL, is thus given by:

Apparently a significant reduction in radiative recombination time requires a large Purcell-factor, which together with a large plasmon-to-radiation coupling efficiency will lead to large enhancement in measured PL. It can be deduced from Eq. (1) that for this system of InAsAs/InP MQWs the maximum attainable radiative enhancement factor is of the order of the original τ_rad/τ_nrad ~10, and that Purcell factors higher than 100 are not required.

We used FDTD simulation results to evaluate the Purcell factor in our experiment using the expression [15]: $F_P=\frac{3}{4{\pi }^2}\cdot \frac{Q}{V_{eff}/{(\lambda /n)}^3}\cdot {(\frac{E}{E_{\max }})}^2$, where the quality factor Q is estimated from the PL spectrum to be Q ~10 for the short-stripe and ~12 for the double-strip nano-antennas, V_eff the effective mode volume calculated from electromagnetic energy distribution around the nano-antenna: $V_{eff}={\displaystyle \iiint [\frac{d(\omega \epsilon )}{d\omega }{\left|E\right|}^2+{\mu }_0{\left|H\right|}^2]dxdydz/\max [\frac{d(\omega \epsilon )}{d\omega }{\left|E\right|}^2+{\mu }_0{\left|H\right|}^2}]$, and the electric field is averaged over the simulation unit cell at the planes of the QWs. The resulted Purcell factor values are: F_{P, Double} ~25 and F_{P, Single} ~13, where the higher Purcell value of the double-strip nano-antenna comes from the higher field confinement in the gap and a slightly longer penetration into the substrate. Using the model of Eq. (1), with τ_rad/τ_nrad = 10, the calculated Purcell and lens collection factors, and with η_pr fitted to be ~0.5, gives calculated PL enhancement values of 4.8 for the single-strip and 8.6 for the double-strip nano-antenna, in good agreement with the measured maximum PL enhancement values of 5 and 9 respectively. This implies that with the 2x200nm double-strip nano-antennas the radiative recombination time is shortened by a factor of 1/F_P ~1/25 from ~10ns to ~0.4ns, and that the overall recombination time reduced by a factor of 1/3 from ~0.9ns to ~0.3ns. The plasmonic nano-antennas increased the radiative recombination rate such that the overall recombination rate is governed by radiative rather than non-radiative recombination.

4. Summary

Gold single-strip and double-strip nano-antennas were fabricated over InP substrate containing MQW. The photo-luminescence from the MQWs was measured and shown to enhance by a maximum factor of 9 and 5 for the double and single strip nano-antennas, where maximum enhancement wavelength tunes in resonance with the nano-antennas. The plasmonic nano-antennas increase lens collection efficiency, and shorten the radiative recombination time of the excitons in the QWs by Purcell factors - calculated from FDTD simulations of 25 and 13 for the double and single strip nano-antennas respectively, resulting in the enhancement values obtained for the measured PL. This implies that the coupling to the double-strip nano-antennas reduces radiative recombination time of carriers in the QWs, making it shorter than non-radiative recombination time, and reducing overall recombination time by a factor of ~3. Better designed nano-antennas, in closer proximity to the QWs, and gaps smaller than 20nm, will result in Purcell factors higher than 100, and in even shorter radiative recombination times. This can have greater impact than merely increasing PL efficiency – by substantiating a path for direct modulation of such LEDs in rates exceeding few Gb/sec, and thus enabling their use in optical short-reach data communications applications.