XXVI EPÍSTOLAS
EPÍSTOLAS PAULINAS
Site controlled QDs do not only show the weak mode localization in small enough
Lncavities, but also permit an alternative identification of the diffusive edge. On Fig.
3.14(a-c), the spectra obtained by exciting three isolated QDs placed at x=0, 4.5 or
-4.5μm with a high pump power. This distance is sufficient to ensure the selective
excitation of only one QD at a time. The mobility edge, which is typically 5meV above the lowest order mode, is shown at 1.268eV by a black arrow. As opposed to cavities with a uniform excitation, the FP modes show a modulation by an envelope function indicating variations of the QD and FP mode overlap. This assumption is validated by the 2D-FDM simulation of the normalized intensity of FP modes at the QD position in a disorderless L61cavity displayed as green dots.
When the central QD is pumped (Fig. 3.14(b)), in the blue part of the spectra only modes symmetric along the x direction are excited. The antisymmetric peaks are not even visible, which simultaneously validate the excellent alignment of the PhC pattern over the QD array and the absence of distortions of FP modes. Fig. 3.14(e) shows the simulated spectra obtained by exciting a disorderless L61cavity with a QD
displaced from the cavity center along x byΔx. The simulation was performed via 2D FDM. The spectra were simulated by adding lorentzian lines for each simulated mode. The intensity of each lorentzian line corresponds to the overlap between the QD and the photonic mode. The QD displacementΔx induces an increase of the antisymmetric modes intensity observed on Fig. 3.14(e). The relative intensity between antisymmetric and symmetric modes in the spectra for a displacement Δx = 20nm is already around 10 and would be visible in the experimental spectra of Fig. 3.14(b), which confirms that this structure exhibits an alignment between the PhC cavity and the QDs better than 20nm.
On the other hand, in the diffusive and localized regimes below 1.28eV, there are significant deviations between the calculated and measured mode intensities. This is also illustrated in Fig. 3.14(d), which shows the relative intensity of each FP mode when pumped by the QDs at x= +4.5μm (left or l) or x = −4.5μm (right or r) : (Il− Ir)/(Il+ Ir). Due to the modes symmetry, in the absence of disorder, this
quantity should be exactly zero, but with disorder it may fluctuate widely. A transition from a very ordered energy range (blue part) to a disordered one (red part) is indeed observed. From these results, the diffusive edge is identified near 1.276eV, in line with the ~10meV measured previously. These measurements demonstrate that pyramidal QDs are useful as sub-wavelength, site-controlled broadband sources, which makes them ideal probes of the local DOS.
Chapter summary
In this Chapter, ensembles of QDs embedded in Lncavities were used to study the
impact of disorder on light propagation in PhC waveguides.
First, propagation losses and reflection coefficients in long Lncavities simulating
PhCWs were measured. Increasing propagation losses at shorter wavelengths were attributed to GaAs band tail absorption. A sharper increase at higher wavelength was interpreted as a sign of scattering to the radiation modes enhanced by slow light near the band edge. These results lead to a deeper understanding of the main loss channels in Lncavities.
In a second part, the impact of disorder in PhCWs was measured via three com- plementary methods. Through spectrally resolved imaging of the modes, we clearly identified the mobility edge that separates the spectral zones of delocalized and lo-
0 103
x10
1.27 1.28 1.29 1.3 1.31 1.32 1.33
Photon energy [eV] -1 0 1 Relative Intensity (d) (c) x=4.5m 0 103 0 103 Intensity [ct/s] x10 x10 FDM simulation (a) (b) x=-4.5m x=0m Mobility edge Mobility edge Mobility edge Mobility edge
Diffusive and localised regime Dispersive regime
Diffusive and localised regime Dispersive regime
Intensity [ct/s]
Intensity [ct/s]
1.26 1.27 1.28 1.29 1.3 1.31 1.32 1.33 1.34
Photon energy [eV] 0 0.4 0.8 Intensity [a.u.] (e) уx=0nm уx=20nm уx=40nm
Figure 3.14 – (a-b-c) Selective excitation with single-QDs in the fast light regime in
L61cavities. The green circles represent the overlap between the QD and the in-plane
component of the mode intensity of each mode computed with 2D FDM. The sketches on the right show the position of excited QDs as a green triangle. (excitation power:
P = 500μW , Temperature: T = 10K , r = 61nm); (d) Measured relative intensity of
each modes when QDs on the left or on the right are pumped: (Il− Ir)/(Il+ Ir). The
black arrows indicate the limit of the mobility edge; (e) FDM simulations of spectra obtained by exciting a disorderless L61 cavity with a QD displaced from the cavity
calized modes. This technique also permitted the identification of scattering at the edges of L61 cavities indicating insufficiently optimized reflectors. In addition, we
used group index measurements and site-controlled QD excitation to identify the limit between diffusive and dispersive regimes of light propagation. The dispersive regime was characterized both by distortions of the envelop function and a highly irregular dispersion. Although photons in this spectral range can be transported across the waveguide channel, they suffer from scattering manifested by random phase variations in the wavefunction. On the contrary, selective excitation highlighted the weak disorder of modes in the dispersive regime and demonstrated the potential of site-controlled QDs for probing the local DOS in photonic structures.
More generally, these observations should be helpful in finding ways to reduce detrimental effects of photon localization and design optimal integrated photonic circuits.
IN recent years, significant efforts were realized towards the monolithic integration of QDs in waveguides, with the objective of obtaining high efficiency on chip single photon sources. One recurrent insight guiding these efforts was the objective of obtaining on chip reproducible and deterministic single photon emission. In recent publications, Stranski-Krastanov QDs were coupled with a broadband high coupling efficiency to PhC waveguides [107, 167, 106, 105, 168, 104]. Beyond single photon sources, the recent publication of directional coupling of SK QDs in chiral waveguides [103, 102] and nonlinearity at the single photon level [101] emphasizes the potential of QDs in PhCs to fabricate on chip quantum gates. However, the analysis of such structures is complicated by the lack of control over the position of QDs. Indeed, the measured properties may not give a faithful picture of the statistical behavior of QDs in such systems [104]. Besides, the coupling relies on a probabilistic approach: it is only a proof of concept for single QDs but it cannot be scaled to many QDs and will exhibit significant statistical variations for similar structures.
Scalability and reproducibility of photonic circuits both require a good spec- tral and spatial matching of the QDs and the photonic modes. In this context, site- controlled QDs in waveguide offer a remarkable potential. Indeed, the spatial match- ing is ensured by the accurate QD positioning and the spectral matching is made possible by the broadband photonic environment of a waveguide.
In this chapter, we demonstrate the first integration of five site-controlled QDs all coupled to a waveguide and the corresponding on chip single photon transfer over macroscopic distances. Using time resolved photoluminescence (TRPL) measure- ments, we infer the coupling efficiency of these QDs to waveguides and give optimal conditions to reach a reproducible, broadband coupling efficiency. We then show how a short slow light section could be harnessed to increase this coupling efficiency without any manifestation of Anderson localization.