Plasmonic modes of electromagnetic fields in metal nanostructures can enable light confinement at subwavelength scales. This property has inspired researchers to explore the possibilities of nanotechnology in miniaturization of optics [1]. In particular, coupling of quantum emitters, such as quantum dots (QDs), to plasmonic metal structures has attracted significant attention. Enhancement of luminescence owing to interaction with plasmonic field has been demonstrated for both colloidal [2 –5] and epitaxial QDs [6 –9], and exploited in QD-based optoelectronic devices [10,11]. The benefit of using the epitaxial QDs is the possibility of monolithic integration with semiconductor optical and electronic device structures. To this end, site-controlled epitaxy provides means for precise positioning of high-quality QDs with respect to the strongly localized field of a metal nanostructure. In our previous work we have demonstrated site-controlled epitaxy of single InAs QDs [12 –14] and quantum dot chains (QDCs) [15 –17] by molecular beam epitaxy (MBE) on GaAs surfaces patterned by ultraviolet nanoimprint lithography (UV-NIL), which is able to produce sub-10-nm linewidths [18] with high throughput [19] and enables fast processing of large wafer areas [20]. The optical quality of the QDs fabricated by this method is manifested by exciton linewidth of only 41 μeV. Moreover, we have used site-controlled epitaxy to position QDs in microcavities and demonstrated coupling between single site-controlled InAs QDs and the intracavity optical field [21].

In this Letter, we demonstrate coupling of deterministically ordered chains of InAs QDs (site-controlled QDCs) with surface plasmon polaritons (SPPs) at GaAs–Ag interfaces. We investigate the plasmon–QD interaction by time-resolved and polarization-resolved photoluminescence (PL) experiments carried out in the cleaved-edge geometry. The site-controlled QDCs are promising candidates for providing gain or loss compensation in plasmonic waveguide networks [22,23] owing to the fact that ordered QDCs with different directions can be grown simultaneously depending on the design of the nanopattern [15,16]. Furthermore, hybrid materials consisting of ordered quantum emitters and plasmonic metal structures have great prospects in the fields of optical metamaterials [24] and quantum plasmonics [25].

Site-controlled InAs QDCs were fabricated by a combination of MBE and UV-NIL. First, a 500 nm GaAs layer was grown by MBE on a SI-GaAs(100) substrate. Then, UV-NIL was used to pattern the GaAs surface. The nanopattern consisted of [010]-oriented grooves with 1000 nm period. The depth and width of the grooves were 30 and 90 nm, respectively. Chemical cleaning and oxide removal treatments were performed after the patterning and the samples were loaded back into the MBE system. The UV-NIL process and chemical surface preparation are discussed in detail in [12]. In the second growth step, the pattern was first overgrown with a 30 nm GaAs buffer at 470°C. The QDCs were formed in the grooves by depositing 1.8 monolayers of InAs at 540°C. The formation of the QDCs is based on a variation of Stranski–Krastanov QD growth [26]. First a thin layer of InAs, the so-called wetting layer, is formed over the whole nanostructured surface [17]. Then, once a certain critical layer thickness is exceeded, the lattice mismatch between InAs and GaAs substrate causes formation of QDs on energetically favorable nucleation sites provided by the nanopattern. Finally, the QDCs were capped with a GaAs spacer, which separates them from the top surface. Three samples were prepared. The spacer thicknesses in these samples were 20, 40, and 80 nm. After MBE growth, each sample was cleaved into two pieces and one of the pieces was covered with a 200 nm Ag film grown by electron beam evaporation. The sample was tilted 45° with respect to the evaporation direction in order to shadow, and thus prevent Ag deposition on, the cleaved edge. Figure 1(a) shows an atomic force micrograph of the site-controlled QDCs before overgrowth. The detailed growth procedure and formation of QDCs with 1000 nm separation are presented in [27]. The full sample structure, including the Ag layer, is illustrated in Fig. 1(b). It should be noted that the QDCs are aligned at 45° with respect to the cleaved edge, as shown in Fig. 1(a).

 

Fig. 1. (a) AFM micrograph of [010]-oriented QDCs before overgrowth, and (b) illustration of the final structure. Optical measurements are performed on the edge of the sample along the [$01\overline{1}$]-direction. The vertical (V) and horizontal (H) polarizations of the optical field are indicated by the arrows.

Download Full Size | PDF

Microphotoluminescence ($\mu$-PL) was measured at 10 K temperature from the cleaved edge in the [$01\overline{1}$] direction, as sketched in Fig. 1(b). The sample was excited in resonance to the wetting layer with an 840 nm diode laser, which was focused on the sample with a microscope objective ($\mathrm{NA}=0.8$). The spatial resolution of the $\mu$-PL system is around 500 nm. The PL signal was collected with the same objective. In steady-state spectral measurements, the signal was dispersed with a 750 mm spectrometer and detected with a 2D CCD camera. An 875 nm longpass filter was used to filter the excitation laser. The $\mu$-PL intensity maps were recorded by using the same CCD camera for imaging. Large area excitation for imaging was obtained by using an additional lens to expand the excitation spot. The PL signal was filtered with a 925 nm longpass filter to exclude the photons corresponding to the excited states of the QDs. Vertical (V) and horizontal (H) polarizations, as defined in Fig. 1(b), were resolved with a half-wave plate and a linear polarizer.

The time-resolved PL experiments were performed by time-correlated single-photon counting (TCSPC) using pulsed excitation at 840 nm. The pulse width and the repetition rate were set to 50 ps and 80 MHz, respectively. The PL signal was again filtered with the 925 nm longpass filter and detected with a Si single-photon avalanche diode. The full width at half-maximum (FWHM) of the instrument response function of the TCSPC setup was 250 ps. The PL decay times and their uncertainties were extracted from the PL decay curves by iterative reconvolution. Good fits were obtained for all decay curves using a single-exponential decay function.

As shown in Fig. 2(a) for as-grown QDCs covered with 20 nm GaAs without Ag capping, the PL spectra of the QDCs are generally composed of narrow exciton lines at the spectral range from 900 to 940 nm. The decay rate of these excitonic states can be expressed in terms of the decay time:

 

Fig. 2. (a) $\mu$-PL spectrum for as-grown QDCs covered with a 20 nm GaAs layer. (b) PL decay rates $1/{\tau }_1$ and $1/{\tau }_2$ for as-grown QDCs (squares) and QDCs with Ag film (circles), respectively, obtained from the PL times for different spacer thicknesses. The inset in (b) shows the time-resolved PL signals for QDCs capped with a 20 nm GaAs spacer. The decay times are ${\tau }_{\mathbf{1}}=500\pm 15\mathrm{ns}$ for the as-grown QDCs and ${\tau }_{\mathbf{2}}=230\pm 19\mathrm{ns}$ for the QDCs with Ag film. The decay curves are normalized and vertically shifted for illustration purposes.

Download Full Size | PDF

The presence of the Ag film on top of the structure gives rise to SPPs propagating in the metal–semiconductor interface. The interaction between the QDs and free electrons in the metal film introduces additional decay channels. Now the decay rate can be expressed as

The plasmonic mode has electric field components parallel (horizontal) and perpendicular (vertical) to the metal–semiconductor interface. The emission of the QDCs couples to both of these field components, but the vertical electric field component is approximately 4.3 times stronger than the horizontal component for a GaAs–Ag interface [9]. Thus, the QD–plasmon interaction affects mainly the vertically polarized part of the QD emission. Figures 3(a) and 3(b) show polarization-resolved $\mu$-PL images for as-grown QDCs and QDCs with Ag film. The QDCs are represented in the $\mu$-PL images by a chain of spots. It should be noted that each QDC is shown as an individual spot due to the shallow depth of focus of the optical system, although they are aligned at a 45° angle with respect to the measurement direction. As shown in Fig. 3(a), the PL emission of the as-grown QDCs is predominantly horizontally polarized. This is typical for flat InAs QDs for which the polarization of PL emission is dictated by the compressive strain and strong vertical carrier confinement [31 –33]. However, as shown in Fig. 3(b), the ratio between the vertical and horizontal polarizations is changed when the Ag layer is present in the proximity of the QDCs. A more quantitative analysis was performed by calculating the degree of polarization $\mathrm{DOP}=(I_V-I_H)/(I_V+I_H)$ for each QDC. $I_V$ and $I_H$ are the intensities of the vertical and horizontal polarizations, respectively, which were determined by integrating the PL signal over a square region of interest that encloses the intensity spot of the QDC. The DOP was averaged over 10 QDCs per sample. The average DOP, along with its standard deviation, is shown in Fig. 4 as a function of spacer thickness for as-grown QDCs and QDCs with Ag film. According to Fig. 4, the DOP is around $-0.6$ for the as-grown QDCs, and it is independent of the spacer thickness. The Ag film has no significant effect for spacer thickness of 80 nm, but as the spacer thickness is reduced, the DOP increases and changes sign. The inversion of polarization anisotropy from horizontal to vertical polarization is a clear indication of plasmonic enhancement of the vertically polarized component emission of the QDCs. Furthermore, the enhancement of vertical polarization as a function of spacer thickness is consistent with the decay rate shown in Fig. 2(a). We can therefore conclude that they are both manifestations of QD–plasmon coupling. From the decay rates shown in Fig. 2(b), we obtain decay rate enhancement factors (${\tau }_1/{\tau }_2$) of 1.1, 1.3, and 2.2 for the 80, 40, and 20 nm spacer thickness, respectively. In comparison, the enhancement factor for the 20 nm spacer thickness is similar to what was observed for self-assembled InAs QDs placed 20 nm below a layer of Ag nanoislands [8].

 

Fig. 3. $\mu$-PL intensity maps measured for horizontal (H) and vertical (V) polarizations for QDCs with varying spacer thicknesses. (a) presents QDCs without and (b) QDCs with Ag layer. The color scale represents PL intensity, which is normalized individually for each spacer thickness.

Download Full Size | PDF

 

Fig. 4. $\mathrm{DOP}=(I_V-I_H)/(I_V+I_H)$ for as-grown QDCs and QDCs with Ag film.

Download Full Size | PDF

In conclusion, we have demonstrated QD–plasmon coupling in a structure consisting of a Ag film and site-controlled InAs QDCs embedded in a GaAs matrix. We observe plasmonic enhancement of both PL decay rate and vertical polarization of PL emission. These results pave the way for exploitation of high-quality site-controlled InAs QDs as active components in hybrid materials including plasmonic nanomaterials. Furthermore, we showed that the polarization-resolved edge-emitted PL can be used as an effective tool for investigating coupling between QDs and plasmons.