Immunity of intersubband polaritons to inhomogeneous broadening
Extracted source text · arXiv 1711.09982
%% ****** Start of file template.aps ****** %
%%
%%
%% This file is part of the APS files in the REVTeX 4 distribution.
%% Version 4.0 of REVTeX, August 2001
%%
%% Copyright (c) 2001 The American Physical Society.
%%
%% See the REVTeX 4 README file for restrictions and more information.
%%
%
% This is a template for producing manuscripts for use with REVTEX 4.0
% Copy this file to another name and then work on that file.
% That way, you always have this original template file to use.
%
% Group addresses by affiliation; use superscriptaddress for long
% author lists, or if there are many overlapping affiliations.
% For Phys. Rev. appearance, change preprint to twocolumn.
% Choose pra, prb, prc, prd, pre, prl, prstab, or rmp for journal
% Add 'draft' option to mark overfull boxes with black boxes`
% Add 'showpacs' option to make PACS codes appear
\documentclass[aps,prb,twocolumn,showpacs]{revtex4} % for review and submission
%\documentclass[aps,preprint,showpacs,groupedaddress]{revtex4} % for double-spaced preprint
\usepackage{graphicx} % needed for figures
\usepackage{dcolumn} % needed for some tables
\usepackage{bm} % for math
%\usepackage{eqnarray,amssymb} % for math
\usepackage{subscript} % for math
\usepackage{xcolor}
\newcommand{\iac}[1]{{\color{black} #1}}
\newcommand{\jm}[1]{{\color{black} #1}}
\newcommand{\rc}[1]{{\color{black} #1}}
%\newcommand{\iac}[1]{{\color{black} #1}}
%\newcommand{\jm}[1]{{\color{black} #1}}
%\usepackage{bm}% bold math
%\usepackage{hyperref}% add hypertext capabilities
%\usepackage[mathlines]{lineno}% Enable numbering of text and display math
%\linenumbers\relax % Commence numbering lines
%\usepackage[showframe,%Uncomment any one of the following lines to test
%%scale=0.7, marginratio={1:1, 2:3}, ignoreall,% default settings
%%text={7in,10in},centering,
%%margin=1.5in,
%%total={6.5in,8.75in}, top=1.2in, left=0.9in, includefoot,
%%height=10in,a5paper,hmargin={3cm,0.8in},
%]{geometry}
\begin{document}
\title{Immunity of intersubband polaritons to inhomogeneous broadening}
\author{J-M. Manceau$^{1,\dagger}$, G. Biasiol$^{2}$, N.L. Tran$^{1}$, I. Carusotto$^3$, R. Colombelli$^{1,\ddagger}$}
\address{$^1$Centre de Nanosciences et de Nanotechnologies, CNRS \jm{UMR 9001}, Univ. Paris-Sud, Universit\'e Paris-Saclay, C2N - Orsay, 91405 Orsay cedex, France\\
$^2$Laboratorio TASC, CNR-IOM, Area Science Park, S.S. 14 km 163.5 Basovizza, I-34149 Trieste, Italy\\
$^3$INO-CNR BEC Center and Dipartimento di Fisica, Universit\'a di Trento, I-38123 Povo, Italy}
%\date{\today}
\begin{abstract}
We demonstrate that intersubband (ISB) polaritons are robust to inhomogeneous effects \jm{originating from the presence of multiple quantum wells (MQWs)}. In a series of samples that exhibit mid-infrared ISB absorption transitions with broadenings varying by a factor of 5 (from 4 meV to 20 meV), we have observed polariton linewidths always lying in the 4 - 7 meV range only. We have experimentally verified the dominantly inhomogeneous origin of the broadening of the ISB transition, and that the linewidth reduction effect of the polariton modes persists up to room-temperature. This immunity to inhomogeneous broadening is a direct consequence of the coupling of the large number of ISB oscillators to a single photonic mode. It is a precious tool to gauge \jm{the natural linewidth of the ISB plasmon} , that is otherwise masked \jm{in such MQWs system} , and is also \jm{beneficial} in view of perspective applications such as intersubband polariton lasers.
\end{abstract}
\pacs{}
\maketitle
%\section{\label{sec:level1}First-level heading}
% sections are not used for PRL papers
%
\section{Introduction}
\label{sec:intro}
The mechanisms responsible for broadening of optical transition lines are typically
classified in two classes: \textit{homogeneous} - related to the dynamics of each
emitter - and \textit{inhomogeneous} - inherent to the presence of multiple emitters.
In the textbook example of an atomic gas, the former class contains the radiative
broadening due to spontaneous emission. The latter class includes Doppler broadening due
to the wide thermal distribution of velocities, that randomly shifts each emitter resonance \cite{Siegman}.
%As a rule of thumb, inhomogeneous broadenings can be greatly suppressed by focusing on specific subgroups of emitters.
A similar physics is found in solid-state systems. For instance, optical transitions in assemblies of quantum dots suffer from a strong inhomogeneous broadening \cite{Gammon}. This comes on top of the radiative
linewidth and decoherence due to interaction with other degrees of freedom such
as phonons.
As a first step towards observing the natural linewidth {of quantum dots}, the inhomogeneous broadening due to the slightly different sizes and shapes of the {various dots} can be overcome by restricting to a single object~\cite{footnote}.
% If we are interested in the \textit{natural} linewidth,
%the inhomogeneous broadening due to the slightly different sizes and shapes of
%the emitters can be overcome by restricting to a single object.
%
In the last decades, ISB transitions in semiconductor quantum wells have been attracting a growing interest from the fundamental and applied physics community. This is due to their very strong coupling to the electromagnetic
field, and their easily tunable emission/absorption frequency across a wide spectral range (from terahertz up to mid-infrared frequencies), where few other compact emitters are available. As the frequency of the ISB transition is dependent on the QW thickness \cite{Liu}, and on the doping level \cite{Ando}, slight fabrication inhomogeneities are a major source of broadening of ISB plasmons in MQWs systems.
\iac{In addition to this, even in the single QW case the physics of ISB transitions is complicated by the homogeneous and inhomogeneous broadening mechanisms that affect the transition linewidth.} On one hand, the homogeneous linewidth is governed by the emission rate of photons and phonons, and \iac{by} the in-plane disorder which opens \iac{additional} decay channels for ISB plasmons. On the other-hand, the non-parabolicity of the electronic bands and the in-plane spatial variations of the QW doping level and thickness are responsible for inhomogeneous broadening \cite{Liu,Dupont,Campman,Ando,Warburton,khurgin}. \iac{Since the mid '90's this physics has been widely studied for interband excitons in the IR domain}, where several peculiar line-narrowing effects~\cite{whittaker,savona,Ell,Whittaker2} can be at play to \iac{reduce the transition linewidth} in strong light-matter coupling regimes (see also a review in Ref.~\onlinecite{Liti}).
%{In addition to this, also single QWs are typically affected by inhomogeneously broadened ISB transitions due to the non-parabolicity of the electronic bands and to in-plane spatial variations of their thickness and doping level~\cite{Liu,Dupont,Campman}. An analog of this latter effect has been widely studied for interband excitons in the IR domain, where several peculiar line-narrowing effects~\cite{whittaker,savona,Ell,Whittaker2} can be at play to suppress its importance in strong light-matter coupling regimes (see also a recent review in~\cite{Liti}). In the ISB case, however, the physics is made more complicate by the additional (homogeneous) decoherence and decay channels that in-plane disorder opens for ISB plasmons via, e.g., electron scattering processes off the in-plane disorder~\cite{xxx}.}
\iac{In the ISB context, all these} broadening mechanisms can have dramatic consequences for fundamental studies
of many body effects in 2D electronic plasmas \cite{Ando},
but also for optoelectronic devices. For instance, the threshold current density
of quantum cascade (QC) lasers is affected by the inhomogeneous broadening
of the material gain \cite{faist} and can only be partially
mitigated in the so called broadband QC lasers \cite{Gmachl,Rosch}.
%More generally, the relatively large linewidth of ISB transitions has been a major obstacle in the development of ISB quantum optics.
\iac{In the last decade, exciting new avenues are being opened by} devices operating in the strong light-matter coupling
regime, where ISB transitions coupled to a cavity photon mode form new bosonic
excitations of mixed nature called \textit{ISB polaritons}.
Such a regime is at the heart of intriguing proposed applications to low-threshold
lasers \cite{liberato,colombelli2} and quantum photon sources in the
far-infrared \cite{ciuti2}. Transferring into the realm of polaritons the design flexibility of ISB transitions - the key ingredient for QC lasers - has been one of the motivations behind the development
of ISB polaritonics \cite{Dini,Colombelli1}. Still, most of these applications require a narrow polariton
linewidth, which may seem incompatible with the wide absorption linewidth typically observed in devices containing a relatively large number of QWs.
In this {article}, we experimentally show how the strong coupling regime permits to
largely suppress the inhomogeneous broadening originating from the presence of a large number of QWs. The underlying mechanism is originally discussed in the atomic cavity-quantum electro-dynamics (QED) context \cite{Dalibard,ningyuan}, and it was theoretically translated to the solid-state context in Ref. \onlinecite{houdre}.
%Note also
%the existence of an interesting \texttt{"}classical\texttt{"} analogy in multi-layer
%optical metamaterials \cite{Ghasemi}.
The core idea is that by coupling the \jm {oscillators} to a single cavity mode, one effectively
singles out a single state out of the inhomogeneous broadened spectral distribution.
The resulting linewidth reduction turns out to be quantitatively important and
can \jm {be beneficial} in the development of ISB polaritonics.
% It is important to stress that this effect is different from the motional narrowing effect observed in excitonic polaritons\cite{whittaker, savona}.
% As motional narrowing follows from the delocalization of the polariton wavefunction along the cavity plane, it is a direct consequence of the small polariton mass and, thus, only indirectly of the strong coupling. On the other hand, the effect we report here crucially depends on the strong coupling of the many electronic oscillators with a single photonic mode.
%
%
\begin{figure}
\includegraphics[trim=0.5cm 0cm 0cm 0cm, scale=1.25]{fig1.pdf}
\caption{(a) Transmission measurements of the 4 studied samples performed at 78 K in a 45$^o$, multipass waveguide configuration. The data are offset for clarity. (b) Schematic of the polaritonic device and experimental probing conditions. (c) Reflectivity measurements performed on the 4 polaritonic devices at 15$^o$ incidence angle and at 78 K. The reflectivity dips correspond to the lower and upper ISB polaritons. }
%\vspace{-0.5truecm}
\end{figure}
%
%
\begin{figure}
\includegraphics[trim=0.75cm 0cm 0cm 0cm, scale=0.31]{fig2.pdf}
\caption{Summary of the experimental linewidth of ISB transitions (blue dots), empty grating resonator (red dots), upper polaritons (green up triangles) and lower polaritons (green down triangles). Panel (a) reports the results at 78 K. Panel (b) at 300 K. {The dashed
lines in the two panels represent the average of cavity and ISB transition linewidths, (${\Gamma}\textsubscript{ISB}+{\Gamma}\textsubscript{cav})/2$
that corresponds to the \textit{expected} polariton linewidth at zero detuning,
i.e. when the Hopfield coefficients are equal to 0.5.}}
%\vspace{-0.5truecm}
\end{figure}
%
%
In a series of samples with increasing absorption linewidth, we measure polariton linewidths consistently
much narrower than the average of cavity mode and ISB plasmon taken independently.
While hints of this phenomenon have been observed in Ref. \onlinecite{Murphy}, here we provide a full and quantitative characterization of it. We experimentally demonstrate the dominantly inhomogeneous origin of the ISB plasmon absorption linewidth \jm{in MQWs system}, and we quantitatively explain our observations using a theoretical model inspired from the seminal work Ref. \onlinecite{houdre}.
\section{The experiments}
\label{sec:expts}
The four samples investigated are GaAs/AlGaAs multiple quantum-wells exhibiting
ISB absorption resonances around 10 \ensuremath{\mu}m wavelength. They have been
grown by molecular beam epitaxy, and consist in 36-period repetitions of 8.3-nm-thick
GaAs QWs separated by 20-nm-thick Al$_{0.3}$Ga$_{0.7}$As barriers. The substrate is undoped GaAs, and \ensuremath{\delta}\ensuremath{-}doping
is introduced in the center of the barriers. The 4 samples differ in the nominal
modulation doping level \textit{n}\textsubscript{\textit{2D}}: 0.6$\times 10^{12}$ cm\textsuperscript{-2}
(HM3821), 1.7$\times 10^{12}$ cm\textsuperscript{-2} (HM3820), 4.4$\times 10^{12}$ cm\textsuperscript{-2}
(HM3872), and 7.6$\times 10^{12}$ cm\textsuperscript{-2} (HM3875).
All the absorption measurements are done using a Fourier transform infrared spectrometer (FTIR) equipped with a Globar thermal source and a deuterated triglycine sulfate (DTGS) detector operating at room temperature. The samples are shaped in multi-pass waveguide configuration and mounted on a continuous flow cryostat having anti-reflection coated Zinc-Selenide (ZnSe) windows implemented on the shroud. Resolution was set at 0.5 meV \iac{(4 cm\textsuperscript{-1})} for all the measurements and the polarization of the incoming radiation is selected with a wire grid holographic Thallium Bromoiodide (KRS-5) in order to get \iac{onto the sample the desired electric-field projection in the direction perpendicular to the growth plane.}
The low temperature (78 K) results are reported in Fig. 1a (results at 300 K are in Supplemental Information~\cite{SM}, Fig. S1). The peak absorption frequency blue-shifts with increasing doping because
of the depolarization shift\cite{Ando}. Most notably,
the ISB transitions dramatically broadens with increasing doping.
We have then inserted the four samples in dispersive, grating-based metal-dielectric-metal
resonators - as sketched in Fig. 1b - following the procedure described in Ref.\onlinecite{manceau}.
The tight electromagnetic confinement induced by the two metallic surfaces places
the system in the regime of strong light-matter coupling, and ISB polaritons can
be probed with surface-reflectivity measurements. \jm{The cryostat is placed within a fixed angle reflection unit (15 degree) inserted within the FTIR and the beam size is selected to cover the entire sample surface (2.5 mm x 2.5 mm). Care is taken to suppress any light that would impinge outside the sample surface.}
Figure~1c reports the reflectivity measurements on all the samples at 78 K and 15\textsuperscript{o}
incidence. For each sample, we have adjusted the grating period $\Lambda$ (filling factor
is kept constant at 80\%) in order to obtain the minimum polaritonic splitting
at \ensuremath{\theta}\ensuremath{\sim}35\textsuperscript{\ensuremath{o}}.
The measurements reveal the two polaritonic modes as reflectivity dips, whose energy
distance - the Rabi splitting - increases roughly as the square root of the doping
(Fig. S2 of Supplemental Information~\cite{SM}), i.e. proportionally to the electronic plasma frequency \label{HHFieldmarkHH367H1262489299}\label{HHFieldmarkHH366H1262489299} \cite{ciuti1,todorov}.
%
%
\begin{figure}
\includegraphics[trim=0.5cm 0cm 0cm 0cm, scale=0.3]{fig3.pdf}
\caption{(a) Transmission measurements at 78 K of sample HM3872 after removal of 0, 9, 15, and 26 wells respectively. The data are stacked for clarity and the number of remaining QW is reported for each measurement. \jm{The dotted line serves as guide to the eye to mark the central energy of the 36 QWs sample.}(b) Extracted linewidths, using a Voigt fitting procedure, as a function of the number of remaining QWs. }
%\vspace{-0.5truecm}
\end{figure}
%
%
The important observation concerns instead the polariton linewidths that appear
insensitive to the massive broadening of the bare ISB transition with doping. Fig.~2 summarizes the linewidths of ISB absorption transitions (\ensuremath{\Gamma}\textsubscript{ISB},
blue dots), upper and lower polaritons (\ensuremath{\Gamma}\textsubscript{UP} and
\ensuremath{\Gamma}\textsubscript{LP}, green triangles), and bare cavity resonators
(\ensuremath{\Gamma}\textsubscript{cav}, red dots) at 78 K and 300 K. The dashed
line represents the average of cavity and ISB transition linewidths, (\ensuremath{\Gamma}\textsubscript{ISB}+\ensuremath{\Gamma}\textsubscript{cav})/2,
that corresponds to the \textit{expected }polariton linewidth at zero detuning,
i.e. when the Hopfield coefficients are equal to 0.5 \label{HHFieldmarkHH412H1262489299} \cite{colombelli2}.
The experimental evidence is instead that \textit{both }\ensuremath{\Gamma}\textsubscript{UP}
and \ensuremath{\Gamma}\textsubscript{LP }are much smaller than the expected average
cavity and ISB transition linewidths. The phenomenon is particularly evident at
78 K and is not compatible with the expected polariton linewidth being the average of \ensuremath{\Gamma}\textsubscript{ISB}
and \ensuremath{\Gamma}\textsubscript{cav }\textit{weighted }by the Hopfield coefficients.
{As a final piece of experimental information,}
we demonstrated the dominantly inhomogeneous origin of the bare
ISB transition broadening observed in Fig.~1a in the absence of the cavity. The
presence of a low energy shoulder in the absorption spectra of the two highest
doped samples is already a strong indication. To gain further insight, we measured
the ISB absorption of three pieces of sample HM3872 (\textit{n}\textsubscript{\textit{2D}}=4.4$\times 10^{12}$ cm\textsuperscript{-2})
after removal by sulphuric-acid-based wet chemical etch of 9, 15, and 26 wells
respectively. The low-temperature (T=78 K) multipass waveguide absorption measurements
are presented in Fig. 3a: the peak absorption energy red-shifts and the low-energy
shoulder disappears, thus proving that the 36 QWs composing the sample are not
all identical. This leads to a linewidth reduction, at both 78 K and room temperature,
when a reduced number of QWs is probed (Fig. 3b). These measurements prove that
a large fraction of the ISB linewidth in these samples indeed stems from inhomogeneous
mechanisms {due to the different parameters of the various wells}.
\section{The theoretical model}
\label{sec:theory}
\iac{Based on this experimental input and taking inspiration from the seminal prediction made in Ref.\onlinecite{houdre},
we conjecture that ISB polaritons are not affected by inhomogeneous broadening
of the MQWs bare ISB transition, but only by the homogeneous one. To substantiate our claim, we generalize the temporal coupled-mode theory of Ref.\onlinecite{manceau} by including the spatial periodicity of the grating-based resonator and the inhomogeneous broadening of the multiple matter oscillators, and we show that this model is able to fully explain in a quantitative way the experimental findings.}
\subsection{Simplified model: planar cavity}
As a first step, we neglect the lateral patterning of the cavity and we build a simplified theory that does not include Bragg scattering processes. This simplification will be useful to fully appreciate the role of the inhomogeneous vs. homogeneous broadenings, before proceeding with the development of a complete theory in the next Subsection.
Under the approximation that the cavity is spatially homogeneous along the $xy$ plane, the in-plane wavevector $\textbf{k}$ is a good quantum number. The system's dynamics can then be written in terms of the following motion equations for the \iac{components at in-plane wavevector \textbf{k }
of the cavity field and of the ISB oscillator amplitude} in the \textit{j}-th quantum
well (with $j$=1,...,$N\textsubscript{QW}$) :
%
\begin{eqnarray}
i\frac{da_{\bm k}}{dt}&=&\omega^{cav}_{\bm k} a_{\bm k}-i\frac{(\gamma_{rad}+\gamma_{nr})}{2} a_{\bm k}+ \nonumber \\ &+&\Omega\sum_j b_{j,{\bm k}}+E_{inc}(t) \label{eq:a} \\
i\frac{db_{j,{\bm k}}}{dt}&=&\omega^{ISB}_{j} b_{j,{\bm k}}+\Omega a_{\bm k}-i\frac{\gamma_{hom}}{2} b_{j,{\bm k}}. \label{eq:b}
\end{eqnarray}
%
%\begin{eqnarray}
%i\frac{da_{\bm k}}{dt}&=&\omega^{cav}_{\bm k} a_{\bm k}-i\frac{(\gamma_{rad}+\gamma_{nr})}{2} a_{\bm k}+\Omega\sum_j b_{j,{\bm k}}+E_{inc}(t) \label{eq:a} \\
%i\frac{db_{j,{\bm k}}}{dt}&=&\omega^{ISB}_{j} b_{j,{\bm k}}+\Omega a_{\bm k}-i\frac{\gamma_{hom}}{2} b_{j,{\bm k}}. \label{eq:b}
%\end{eqnarray}
Here, $\omega^{cav}_{k}$ denotes the cavity mode dispersion,
and the frequencies $\omega^{ISB}_{j}$ of the ISB transition
in each well are assumed to be independent of the in-plane wavevector \textbf{k} and distributed
around their central frequency $\omega$\textsubscript{ISB} according to a Gaussian
distribution of standard deviation $\sigma$\textsubscript{inh}. \iac{Furthermore,} $\Omega$ is the
Rabi frequency of each ISB plasmon coupling to the cavity mode. $E_{inc}$
is the incident field. $\gamma_{rad}$ and $\gamma_{nr}$ are the radiative
and non-radiative linewidths of the cavity mode, and $\gamma_{hom}$ is the
homogeneous linewidth of the ISB plasmon, {resulting from all non-radiative decay and decoherence mechanisms due, e.g., to electron scattering on interface roughness~\cite{Liu,Dupont,Campman,Ando,Warburton,khurgin}.} The reflected field results then
from the interference of the directly reflected incident field and the cavity emission \cite{ciuti2},
{
\begin{equation}
E_{refl}=E_{inc} - i \gamma_{rad} a_{\bm k}.
\label{eq:refl}
\end{equation}
Reflection spectra are then straightforwardly obtained by inverting the linear set of equations describing the steady-state
of the motion equations (\ref{eq:a})-(\ref{eq:b}) and inserting the result into (\ref{eq:refl}). An intuitive physical understanding of this model can be summarized as follows along the lines of Ref.\onlinecite{houdre}.
In the absence of inhomogeneous broadening, all wells are identical $\omega^{ISB}_{j}= \omega^{ISB}$, so
the light-matter coupling singles out the fully symmetric combination of
ISB plasmons \textit {b}$_{B,\bm{k}}$= $\Sigma_{j}$\textit {b}$_{j,\bm{k}}/\sqrt{N_{QW}}$. This single \textit{bright} combination couples to the cavity mode with a collective Rabi frequency $\Omega_{R}=\sqrt{N_{QW}}\Omega$. Coherent mixing
of the bright ISB and the cavity modes} gives rise to the ISB polaritons, whose
linewidth results from a weighted average of the cavity and ISB homogeneous linewidths.
All other combinations of ISB's remain at $\omega$\textsuperscript{ISB }but are
dark and therefore do not appear in the optical spectra.
The situation is more interesting in the presence of some inhomogeneous broadening.
{As long as its standard deviation $\sigma$\textsubscript{inh} does not exceed the
collective $\Omega$\textsubscript{R}, the inhomogeneous broadening of the ISB plasmons is only responsible for a corresponding spectral broadening of the the dark states and a weak mixing of them with the bright states. As a consequence, dark states transform into a wide band of weakly optically active states located in between the polaritons, which however remain spectrally well separated and almost unaffected by the inhomogeneous broadening.}
Since large $\Omega$\textsubscript{R }values are a peculiar character of ISB polariton
systems, this mechanism explains why {ISB polaritons} are extremely robust against inhomogeneous
broadening.
The {behaviour} suddenly changes when the \iac{inhomogeneous broadening $\sigma$\textsubscript{inh} becomes comparable to} the collective $\Omega$\textsubscript{R}: in this case, \iac{the energy range over which dark and bright states are mixed by the inhomogeneous broadening} reaches the spectral position of polaritons. Seen in the polariton basis, the magnitude of the mixing terms can cross the polariton gap and \iac{effectively} contaminate the polaritons with the dark states. As a result, the two polaritons lose their character of spectrally isolated states and their
linewidth suddenly increases washing out the corresponding spectral features.
\subsection{Complete theory including Bragg processes}
While the simplified is able to account for the main features of the inhomogeneous broadening, it does not include the \iac{spatial periodicity of period $a$ of} the top mirror and the subsequent Bragg scattering processes that are responsible for the folding of the photonic and polaritonic bands and the consequent Bragg gaps that open between them.
In this Subsection we develop a complete theory that is able to quantitatively reproduce the experimental results. \iac{Restricting for simplicity our attention to the $k_y=0$ line that is addressed in the experiments, the photon and ISB bands can be labeled by the $k_x$ component along the spatial periodicity, denoted for brevity $k$ and belonging to the first Brillouin zone $k \in [-k_{Br}/2,k_{Br}/2]$ with $k_{Br}=2\pi/a$, and by the band index $n=1,2,3,\ldots$.}
The equation of motion for the amplitudes in the photonic and ISB modes read:
\begin{eqnarray}
i \frac{da_{n,{ k}}}{dt} &=& \omega^{cav}_{n,{ k}} a_{n,{ k}} +\Omega \sum_j b_{j,n,{ k}} + \Omega_{Br} [a_{n+1,{ k}} + a_{n-1,{ k}}] \nonumber \\ &-&i\frac{\gamma_{nr}}{2} a_{n,{ k}}-i\frac{\gamma_{rad}}{2} \sum_{n'} \sqrt{\eta_{n,{ k}}\eta_{n',{ k}}} a_{n',{ k}} + \nonumber \\
&+&\sqrt{\eta_{n,{ k}}}\,E_{inc}(t) \label{eq:a_br} \\
i \frac{db_{j,n,{ k}}}{dt}&= & \omega^{ISB}_j b_{j,n,{ k}} + \Omega a_{n,{ k}} - i \frac{\gamma_{hom}}{2} b_{j,n,{ k}} \label{eq:b_br}
\end{eqnarray}
The momentum-independence of the ISB transitions for wavevectors much smaller than the Fermi momentum of the electron gas reflects in the independence of $\omega^{ISB}_j$ from the in-plane wavevector ${ k}$ and the band index $n$. The inhomogeneous broadening of the ISB of the different wells (labeled by $j$) has the same shape as discussed in the previous section for the simplified model.
The periodic patterning of the mirror is described by letting the frequencies $\omega^{cav}_{n,{ k}}$ of the photonic bands to follow the folded dispersion
\begin{eqnarray}
\omega^{cav}_{n,{ k}} &=& \frac{c}{n_0} \left[k +(n-1) \frac{k_{Br}}{2}\right] \;\;\; \textrm{for}\;\;n\;\; \textrm{odd} \\
\omega^{cav}_{n,{ k}} &=& \frac{c}{n_0} \left[n\frac{k_{Br}}{2}-k\right] \;\;\; \textrm{for}\;\; n\;\; \textrm{even}
\end{eqnarray}
for $k>0$, and a symmetric one for $k<0$.
The coefficient $n_0$ is the effective refractive index of the cavity, determined by a combination of the bulk material index and the penetration into the mirrors. The amplitude of the Bragg processes opening gaps at the center and at the edges of the Brillouin zone is quantified by $\Omega_{Br}$.
The $\eta_{n,{ k}}$ coefficients account for the different radiative coupling of each band to the external radiative \iac{modes and are responsible for the peculiar radiative couplings~\cite{fano} that appear in the equations of motion (\ref{eq:a_br}) for the field amplitudes}. In our numerics, we have considered the simplest case where this coupling coefficient vanishes for the lowest \iac{$n=1$} band (which falls out of the light cone), is full and equal to 1 for the $n=2,3$ bands, and then drops to a small value of $0.1$ for the higher bands (to model the reduced large-angle scattering amplitude by the mirror patterning). While the details of the high-$n$ behavior are actually not important to match the experimental results, the light-cone condition on the lowest \iac{$n=1$} band and, even more, the simultaneous and comparable radiative coupling of the $n=2,3$ bands are essential to reproduce the \iac{experimentally observed dependence of the visibility of the different photonic bands on the band index $n$ and on the wavevector $k$.}
Finally, the reflection spectra can be obtained by numerically inverting the linear set of equations describing the steady-state
of the motion equations (\ref{eq:a_br})-(\ref{eq:b_br}) and inserting the result into the generalized reflection amplitude,
\begin{equation}
E_{refl}=E_{inc} - i \gamma_{rad} \sum_n \sqrt{\eta_{n,{ k}}}\,a_{n,{\bm k}}.
\label{eq:refl_br}
\end{equation}
\iac{This approach straightforwardly lead to reflection spectra such as the ones shown in Fig.4(b,c).}
\subsection{Comparison with experiments}
%
%
\begin{figure}
\includegraphics[trim=5.5cm 5cm 0cm 3cm, scale=0.65]{fig4.pdf}
\caption{(a) Experimental band-diagram of the polariton sample HM3872. The measurements are performed at 300 K. The dispersion is obtained by measuring the zero-order reflectivity of the sample between $\theta=13^o$ and $\theta=61^o$. The reflectivity minima of the polaritonic branches are marked with blue dots.
(b,c) Theoretically calculated reflectivity as a function of the in-plane wavevector and the frequency for no inhomogeneous broadening w$_{inh}$=0 (b) and for the experimentally realistic value $w_{inh}$=12 meV (c). All other parameters are chosen to match the experiment and are given in the text and in the SM. The black dashed lines indicate the edges of the experimentally accessible region. The dotted line indicates the $\theta=15^o$ line at which the spectra in Fig.5 are taken. }
%\vspace{-0.5truecm}
\end{figure}
%
%
{We are now going to show how this theoretical model} provides a quantitative
explanation of the experimental observations. Figure 4(a) reports the experimental
polaritonic band-structure of sample HM3872 (\textit{n}\textsubscript{\textit{2D}}=4.4$\times 10^{12}$ cm\textsuperscript{\textit{-2}})
experimentally acquired at room-temperature. It is obtained by measuring the zero-order
reflectivity of the sample between \ensuremath{\theta}\ensuremath{=}13\textsuperscript{\ensuremath{o}}
and \ensuremath{\theta}=61\textsuperscript{o}, and then applying the transformation
$k=\frac{\omega}{c}\sin\theta$ to obtain the in-plane wave-vector $k_{\parallel}$. \jm{For this measurement, the sample is placed in a motorized angular-reflection unit \iac{located} within the chamber of the FTIR. The entire surface of the sample is probed and care is taken to suppress any light impinging outside the sample surface, especially at high angles where the beam spread with a cos($\theta$) dependency.}
The corresponding numerically calculated polaritonic dispersions are shown in Figs.~4(b,c), respectively in the presence and in the absence of an inhomogeneous broadening
of magnitude $\sigma$\textsubscript{inh}=5.1 meV (that is FWHM $w_{inh}=12$~
meV) comparable to the experimental one around the central frequency $\omega^{ISB}=110$~meV. The homogeneous broadening is taken as $\gamma_{hom}$= 6 meV. {The cavity radiative and non-radiative linewidths are $\gamma_{rad}$=2.5
meV and $\gamma_{nr}$=3.5 meV, respectively, while the effective refractive index is $n_0=3.1$ and the Bragg coupling $\Omega_{Br}=5$~meV. The collective Rabi frequency is taken as $\Omega_R=11\,\textrm{meV}$.}
{The quantitative agreement between the theoretical model and the experimental data is very good: the two strongest lines corresponding
to the upper and lower polaritons are almost unaffected by the inhomogeneous broadening
and show linewidths of approximately $\gamma_{LP,UP }$=4 meV each. This
correctly reproduces the experimental observation of polaritonic linewidths that
are much smaller than the absorption linewidth of bare ISB transition shown in Fig.1(a).
The faint features visible in Fig.4(b) within the polariton gap between 105 and 120
meV are due to the folding of the polaritons by the Bragg periodicity. As it is shown in Fig.4(c), their structure is washed out by the inhomogeneous broadening and \iac{they} transform into an almost structureless band stemming from the weak mixing of the dark states with the bright ones. A weak trace of these features is anyway still visible in the experimental band-diagram of Fig.4(a) and in the reflection spectra of Fig.1(c). }
%
\begin{figure}
\includegraphics[trim=0cm 0cm 0cm 1cm, scale=0.3]{fig5.pdf}
\caption{Polaritons peaks positions and lower polariton linewidth as extracted from theoretically calculated reflectivity spectra at 15$^o$ for growing values of the total ISB broadening. Black {dots} correspond to the homogeneous case, while blue dots to the inhomogeneous one. \iac{In this latter case,} the homogeneous broadening is kept constant at an experimental value of $\gamma_{hom}$= 6 meV).
In (a), the polariton peaks positions are reported. A rapid reduction of the splitting is observed in the homogeneous case, while it is absent in the inhomogeneous one.
In (b), a similar effect is observed in the linewidth of the lower polariton state. }
%\vspace{-0.5truecm}
\end{figure}
%
%
To corroborate the findings, and to better illustrate the consequences of this
phenomenon, the different impact of inhomogeneous or homogeneous broadenings is theoretically illustrated
in Fig.5.
The two (a,b) panels respectively show the polariton peak positions and the linewidth of the LP
mode, extracted from the reflection spectra calculated at a given incidence
angle of 15\textsuperscript{o}, as a function of the total ISB linewidth $\gamma_{tot}$.
In the homogeneous case, $\gamma_{tot} = \gamma_{hom}$.
In the inhomogeneous case, %$\gamma_{tot} = \gamma_{hom}+\gamma_{inhom}$,
{ $\gamma_{tot} = \gamma_{hom}+\sigma_{inh}$,} and
the homogeneous contribution is kept constant at a value $\gamma_{hom}$= 6 meV.
Fig.~5(a) shows that the Rabi splitting immediately starts to quench with increasing homogenous broadening,
while it is much more stable against an equivalent increase of inhomogeneous one.
For instance, a purely homogeneous ISB linewdith of 22 meV significantly reduces the Rabi splitting to $2\Omega_{Rabi}=17$ meV, at the
onset of weak coupling.
On the other hand, {in the inhomogeneous case the Rabi splitting remains stable around a value $2\Omega_{Rabi}\approx 25$ meV well within the strong coupling regime and is even slightly reinforced by the spectral broadening as compared to the purely homogeneous case.}
Fig.~5(b) highlights the different effect of homogeneous and inhomogeneous broadening on the polariton linewdith.
In the homogeneous case, the LP FWHM increases approximately linearly with increasing ISB broadening, as expected.
In the inhomogeneous case, instead, the LP FWHM is essentially unaffected until $\gamma_{tot}$ is of the order of the
initial vacuum field Rabi splitting, $\approx$20 meV. At that point it starts increasing, but --importantly-- it is always smaller than in the homogeneous case.
These data highlight the qualitative difference between the two cases: the homogeneous broadening is responsible for a rapid reduction of the Rabi splitting with simultaneous
broadening of the polaritonic states, which gradually morph into a single bare cavity photon line as typical of weak coupling. On the other hand, the inhomogeneous broadening does not initially affect the Rabi splitting nor the polariton linewidths, which remain approximately
constant. Only when the inhomogeneously broadened ISB transition exceeds the polariton splitting, the polaritonic states suddenly collapse into a single broad photon line.
\section{Conclusions}
\label{sec:conclu}
In conclusion, we have experimentally demonstrated how the strong light-matter
coupling regime permits to strongly reduce the inhomogeneous broadening of ISB transitions in a multiple semiconductor QW system, at least its component stemming from the presence of a large number of slightly different QWs.
{The mechanism underlying the observed line narrowing effect is} a direct consequence of (i) the coupling between a large number of oscillators and a single photonic mode and (ii) the elevated coupling constants typical of ISB polariton systems.
In addition to offering a clear illustration of a general physical mechanism, the large line narrowing achievable makes this result an important step in the direction of extending solid-state quantum optical experiments to intermediate wavelengths between the visible/IR range of interband devices and the microwaves of circuit-QED devices. \rc{Furthermore, the elucidated mechanism could be useful to disentangle the homogeneous and inhomogeneous contributions of in-plane disorder to the ISB transition linewidth in single quantum wells.}
%
%
\\ \indent
We thank F. Julien, C. Ciuti, G. C. La Rocca, S. De Liberato for useful discussions. This work
was partly supported by the French RENATECH network. We acknowledge financial support
from European Union FET-Open grant MIR-BOSE 737017.
%
\vspace{-0.6truecm}
\begin{thebibliography}{30}
\expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
\expandafter\ifx\csname bibnamefont\endcsname\relax
\def\bibnamefont#1{#1}\fi
\expandafter\ifx\csname bibfnamefont\endcsname\relax
\def\bibfnamefont#1{#1}\fi
\expandafter\ifx\csname citenamefont\endcsname\relax
\def\citenamefont#1{#1}\fi
\expandafter\ifx\csname url\endcsname\relax
\def\url#1{\texttt{#1}}\fi
\expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
\providecommand{\bibinfo}[2]{#2}
\providecommand{\eprint}[2][]{\url{#2}}
\vspace{-0.5truecm}
\item[$\dagger$] E-mail: jean-michel.manceau@u-psud.fr
\item[$\ddagger$] E-mail: raffaele.colombelli@u-psud.fr
\vspace{0.15truecm}
\bibitem[{\citenamefont{Siegman}(1986)}]{Siegman}
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Siegman}},
\emph{\bibinfo{title}{Lasers}} (\bibinfo{publisher}{University Science
Books}, \bibinfo{year}{1986}).
\bibitem[{\citenamefont{Gammon et~al.}(1996)\citenamefont{Gammon, Snow,
Shanabrook, Katzer, and Park}}]{Gammon}
\bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Gammon}},
\bibinfo{author}{\bibfnamefont{E.~S.} \bibnamefont{Snow}},
\bibinfo{author}{\bibfnamefont{B.~V.} \bibnamefont{Shanabrook}},
\bibinfo{author}{\bibfnamefont{D.~S.} \bibnamefont{Katzer}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Park}},
\bibinfo{journal}{Science} \textbf{\bibinfo{volume}{273}},
\bibinfo{pages}{87} (\bibinfo{year}{1996}).
\bibitem{footnote} Alternative advanced spectroscopic techniques were adopted in P. Borri et al., Phys. Rev. Lett. {\bf 87}, 157401 (2001) to get rid of inhomogeneous effects but they are hardly applied in practical optoelectronic devices.
\bibitem[{\citenamefont{Liu and Capasso}(1999)}]{Liu}
\bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Liu}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Capasso}},
\emph{\bibinfo{title}{Intersubband transitions in quantum wells: physics and
device applications}}, vol.~\bibinfo{volume}{5} (\bibinfo{publisher}{Academic
press}, \bibinfo{year}{1999}).
\bibitem[{\citenamefont{Ando et~al.}(1982)\citenamefont{Ando, Fowler, and
Stern}}]{Ando}
\bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Ando}},
\bibinfo{author}{\bibfnamefont{A.~B.} \bibnamefont{Fowler}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Stern}},
\bibinfo{journal}{Reviews of Modern Physics} \textbf{\bibinfo{volume}{54}},
\bibinfo{pages}{437} (\bibinfo{year}{1982}).
\bibitem[{\citenamefont{Dupont et~al.}(1992)\citenamefont{Dupont, Delacourt,
Papillon, Schnell, and Papuchon}}]{Dupont}
\bibinfo{author}{\bibfnamefont{E.~B.} \bibnamefont{Dupont}},
\bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Delacourt}},
\bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Papillon}},
\bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Schnell}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Papuchon}},
\bibinfo{journal}{Applied Physics Letters} \textbf{\bibinfo{volume}{60}},
\bibinfo{pages}{2121} (\bibinfo{year}{1992}).
\bibitem[{\citenamefont{Campman et~al.}(1996)\citenamefont{Campman, Schmidt,
Imamoglu, and Gossard}}]{Campman}
\bibinfo{author}{\bibfnamefont{K.~L.} \bibnamefont{Campman}},
\bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Schmidt}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Imamoglu}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{A.~C.} \bibnamefont{Gossard}},
\bibinfo{journal}{Applied Physics Letters} \textbf{\bibinfo{volume}{69}},
\bibinfo{pages}{2554} (\bibinfo{year}{1996}).
\bibitem[{\citenamefont{Warburton et~al.}(1998)\citenamefont{Warburton,
Weilhammer, Kotthaus, Thomas, and Kroemer}}]{Warburton}
\bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Warburton}},
\bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Weilhammer}},
\bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Kotthaus}},
\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Thomas}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Kroemer}},
\bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{80}},
\bibinfo{pages}{2185} (\bibinfo{year}{1998}).
\bibitem[{\citenamefont{Khurgin}(2008)}]{khurgin}
\bibinfo{author}{\bibfnamefont{J.~B.} \bibnamefont{Khurgin}},
\bibinfo{journal}{Applied Physics Letters} \textbf{\bibinfo{volume}{93}},
\bibinfo{pages}{091104} (\bibinfo{year}{2008}).
\bibitem[{\citenamefont{Whittaker et~al.}(1996)\citenamefont{Whittaker,
Kinsler, Fisher, Skolnick, Armitage, Afshar, Sturge, and
Roberts}}]{whittaker}
\bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Whittaker}},
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Kinsler}},
\bibinfo{author}{\bibfnamefont{T.~A.} \bibnamefont{Fisher}},
\bibinfo{author}{\bibfnamefont{M.~S.} \bibnamefont{Skolnick}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Armitage}},
\bibinfo{author}{\bibfnamefont{A.~M.} \bibnamefont{Afshar}},
\bibinfo{author}{\bibfnamefont{M.~D.} \bibnamefont{Sturge}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~S.}
\bibnamefont{Roberts}}, \bibinfo{journal}{Physical Review Letters}
\textbf{\bibinfo{volume}{77}}, \bibinfo{pages}{4792} (\bibinfo{year}{1996}).
\bibitem[{\citenamefont{Savona et~al.}(1997)\citenamefont{Savona, Piermarocchi,
Quattropani, Tassone, and Schwendimann}}]{savona}
\bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Savona}},
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Piermarocchi}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Quattropani}},
\bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Tassone}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Schwendimann}},
\bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{78}},
\bibinfo{pages}{4470} (\bibinfo{year}{1997}).
\bibitem[{\citenamefont{Ell et~al.}(1998)\citenamefont{Ell, Prineas, Nelson,
Park, Gibbs, Khitrova, Koch, and Houdré}}]{Ell}
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Ell}},
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Prineas}},
\bibinfo{author}{\bibfnamefont{T.~R.} \bibnamefont{Nelson}},
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Park}},
\bibinfo{author}{\bibfnamefont{H.~M.} \bibnamefont{Gibbs}},
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Khitrova}},
\bibinfo{author}{\bibfnamefont{S.~W.} \bibnamefont{Koch}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Houdr\'e}},
\bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{80}},
\bibinfo{pages}{4795} (\bibinfo{year}{1998}).
\bibitem[{\citenamefont{Whittaker}(1998)}]{Whittaker2}
\bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Whittaker}},
\bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{80}},
\bibinfo{pages}{4791} (\bibinfo{year}{1998}).
\bibitem[{\citenamefont{Litinskaia et~al.}(2001)\citenamefont{Litinskaia,
La~Rocca, and Agranovich}}]{Liti}
\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Litinskaia}},
\bibinfo{author}{\bibfnamefont{G.~C.} \bibnamefont{La~Rocca}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{V.~M.}
\bibnamefont{Agranovich}}, \bibinfo{journal}{Physical Review B}
\textbf{\bibinfo{volume}{64}}, \bibinfo{pages}{165316}
(\bibinfo{year}{2001}).
\bibitem[{\citenamefont{Faist}(2013)}]{faist}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Faist}},
\emph{\bibinfo{title}{Quantum Cascade Lasers}} (\bibinfo{publisher}{OUP
Oxford}, \bibinfo{year}{2013}).
\bibitem[{\citenamefont{Gmachl et~al.}(2002)\citenamefont{Gmachl, Sivco,
Colombelli, Capasso, and Cho}}]{Gmachl}
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Gmachl}},
\bibinfo{author}{\bibfnamefont{D.~L.} \bibnamefont{Sivco}},
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Colombelli}},
\bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Capasso}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{A.~Y.} \bibnamefont{Cho}},
\bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{415}},
\bibinfo{pages}{883} (\bibinfo{year}{2002}).
\bibitem[{\citenamefont{R\"osch et~al.}(2015)\citenamefont{R\"osch, Scalari,
Beck, and Faist}}]{Rosch}
\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{R\"osch}},
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Scalari}},
\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Beck}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Faist}},
\bibinfo{journal}{Nat Photon} \textbf{\bibinfo{volume}{9}},
\bibinfo{pages}{42} (\bibinfo{year}{2015}).
\bibitem[{\citenamefont{De~Liberato and Ciuti}(2009)}]{liberato}
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{De~Liberato}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Ciuti}},
\bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{102}},
\bibinfo{pages}{136403} (\bibinfo{year}{2009}).
\bibitem[{\citenamefont{Colombelli and Manceau}(2015)}]{colombelli2}
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Colombelli}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{J.-M.} \bibnamefont{Manceau}},
\bibinfo{journal}{Physical Review X} \textbf{\bibinfo{volume}{5}},
\bibinfo{pages}{011031} (\bibinfo{year}{2015}).
\bibitem[{\citenamefont{Ciuti and Carusotto}(2006)}]{ciuti2}
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Ciuti}} \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Carusotto}},
\bibinfo{journal}{Physical Review A} \textbf{\bibinfo{volume}{74}},
\bibinfo{pages}{033811} (\bibinfo{year}{2006}).
\bibitem[{\citenamefont{Dini et~al.}(2003)\citenamefont{Dini, Kohler,
Tredicucci, Biasiol, and Sorba}}]{Dini}
\bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Dini}},
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Köhler}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Tredicucci}},
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Biasiol}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Sorba}},
\bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{90}},
\bibinfo{pages}{116401} (\bibinfo{year}{2003}).
\bibitem[{\citenamefont{Colombelli et~al.}(2005)\citenamefont{Colombelli,
Ciuti, Chassagneux, and Sirtori}}]{Colombelli1}
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Colombelli}},
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Ciuti}},
\bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Chassagneux}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Sirtori}},
\bibinfo{journal}{Semiconductor Science and Technology}
\textbf{\bibinfo{volume}{20}}, \bibinfo{pages}{985} (\bibinfo{year}{2005}).
\bibitem[{\citenamefont{Dalibard et~al.}(1992)\citenamefont{Dalibard, Raimond,
and Zinn-Justin}}]{Dalibard}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Dalibard}},
\bibinfo{author}{\bibfnamefont{J.-M.} \bibnamefont{Raimond}},
\bibnamefont{and}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Zinn-Justin}},
\emph{\bibinfo{title}{Systèmes fondamentaux en optique quantique}}
(\bibinfo{publisher}{North-Holland Amsterdam etc.}, \bibinfo{year}{1992}).
\bibitem[{\citenamefont{Ningyuan et~al.}(2016)\citenamefont{Ningyuan,
Georgakopoulos, Ryou, Schine, Sommer, and Simon}}]{ningyuan}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Ningyuan}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Georgakopoulos}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Ryou}},
\bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Schine}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Sommer}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Simon}},
\bibinfo{journal}{Physical Review A} \textbf{\bibinfo{volume}{93}},
\bibinfo{pages}{041802} (\bibinfo{year}{2016}).
\bibitem[{\citenamefont{Houdré et~al.}(1996)\citenamefont{Houdr\'e, Stanley,
and Ilegems}}]{houdre}
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Houdr\'e}},
\bibinfo{author}{\bibfnamefont{R.~P.} \bibnamefont{Stanley}},
\bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ilegems}},
\bibinfo{journal}{Physical Review A} \textbf{\bibinfo{volume}{53}},
\bibinfo{pages}{2711} (\bibinfo{year}{1996}).
\bibitem[{\citenamefont{Murphy et~al.}(2014)\citenamefont{Murphy, Bak,
Matthews, Dupont, Amrania, and Phillips}}]{Murphy}
\bibinfo{author}{\bibfnamefont{F.~J.} \bibnamefont{Murphy}},
\bibinfo{author}{\bibfnamefont{A.~O.} \bibnamefont{Bak}},
\bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Matthews}},
\bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Dupont}},
\bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Amrania}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{C.~C.} \bibnamefont{Phillips}},
\bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{89}},
\bibinfo{pages}{205319} (\bibinfo{year}{2014}).
\bibitem{SM} Supplemental Material can be found at...
\bibitem[{\citenamefont{Manceau et~al.}(2014)\citenamefont{Manceau, Zanotto,
Ongarello, Sorba, Tredicucci, Biasiol, and Colombelli}}]{manceau}
\bibinfo{author}{\bibfnamefont{J.-M.} \bibnamefont{Manceau}},
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Zanotto}},
\bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Ongarello}},
\bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Sorba}},
\bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Tredicucci}},
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Biasiol}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Colombelli}},
\bibinfo{journal}{Applied Physics Letters} \textbf{\bibinfo{volume}{105}},
\bibinfo{pages}{081105} (\bibinfo{year}{2014}).
\bibitem[{\citenamefont{Ciuti et~al.}(2005)\citenamefont{Ciuti, Bastard, and
Carusotto}}]{ciuti1}
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Ciuti}},
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Bastard}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Carusotto}},
\bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{72}},
\bibinfo{pages}{115303} (\bibinfo{year}{2005}).
\bibitem[{\citenamefont{Todorov et~al.}(2010)\citenamefont{Todorov, Andrews,
Colombelli, De~Liberato, Ciuti, Klang, Strasser, and Sirtori}}]{todorov}
\bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Todorov}},
\bibinfo{author}{\bibfnamefont{A.~M.} \bibnamefont{Andrews}},
\bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Colombelli}},
\bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{De~Liberato}},
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Ciuti}},
\bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Klang}},
\bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Strasser}}, \bibnamefont{and}
\bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Sirtori}},
\bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{105}},
\bibinfo{pages}{196402} (\bibinfo{year}{2010}).
%\bibitem[{\citenamefont{Wilson et~al.}(2011)\citenamefont{Wilson, Johansson,
% Pourkabirian, Simoen, Johansson, Duty, Nori, and Delsing}}]{wilson}
%\bibinfo{author}{\bibfnamefont{C.~M.} \bibnamefont{Wilson}},
% \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Johansson}},
% \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Pourkabirian}},
% \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Simoen}},
% \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Johansson}},
% \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Duty}},
% \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Nori}}, \bibnamefont{and}
% \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Delsing}},
% \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{479}},
% \bibinfo{pages}{376} (\bibinfo{year}{2011}).
\bibitem{fano} M. Ghulinyan, F. Ramiro Manzano, N. Prtljaga, M. Bernard, L. Pavesi, G. Pucker, I. Carusotto, %{\em Intermode reactive coupling induced by waveguide-resonator interaction},
Phys. Rev. A {\bf 90}, 053811 (2014).
\end{thebibliography}
\end{document}