Abstract
In a standard semiconductor laser, electrons and holes recombine via stimulated emission to emit coherent light, in a process that is far from thermal equilibrium. Excitonpolariton condensates–sharing the same basic device structure as a semiconductor laser, consisting of quantum wells coupled to a microcavity–have been investigated primarily at densities far below the Mott density for signatures of BoseEinstein condensation. At high densities approaching the Mott density, excitonpolariton condensates are generally thought to revert to a standard semiconductor laser, with the loss of strong coupling. Here, we report the observation of a photoluminescence sideband at high densities that cannot be accounted for by conventional semiconductor lasing. This also differs from an upperpolariton peak by the observation of the excitation power dependence in the peakenergy separation. Our interpretation as a persistent coherent electronholephoton coupling captures several features of this sideband, although a complete understanding of the experimental data is lacking. A full understanding of the observations should lead to a development in nonequilibrium manybody physics.
Introduction
Semiconductor lasers are one of the most fascinating systems for studying quantum manybody physics in a nonequilibrium regime arising from the interplay of an interacting electron–hole–photon (e–h–p) system. In the context of an ehp system in a semiconductor, localized excitons are typically treated as twolevel systems with no internal structure and coupled to a continuum of radiation modes confined in a twodimensional microcavity^{1}. On the other hand, at equilibrium without photons, there are several predicted phenomena^{2,3,4,5,6} when the electron–hole (e–h) internal structure is taken into account. One example is the e–h Bardeen–Cooper–Schrieffer (BCS) phase in the high e–h density regime^{2,7}, where the condensation of e–h Cooper pairs opens a gap around the Fermi energy in the electron and hole energy dispersions. Since such BCS physics is based on assumptions of equilibrium in e–h systems, with the complete lack of photons, it has been traditionally conceptually disconnected from semiconductor lasers (ehp system).
On the other hand, the phenomenon of excitonpolariton condensation has recently gained tremendous interest^{8,9,10,11,12}, while sharing the same basic structural elements with a semiconductor laser consisting of a number of quantum wells (QWs) coupled to a microcavity structure. The first characteristic difference between semiconductor lasers and excitonpolariton condensates is the lack, or presence respectively, of strong coupling between the bound eh pairs (excitons) and cavity photons. The strong coupling between the excitons and photons result in a new quasiparticle called the excitonpolariton, which condense into the zero momentum state via stimulated cooling, rather than stimulated emission as in a standard laser^{10}. The difference in the mechanism of coherence formation to conventional lasing has suggested that the excitonpolariton system be modelled as a nonequilibrium BoseEinstein condensate (BEC)^{13}, with a large number of experiments supporting this interpretation with expected properties such as superfluidity^{14}. However, such experiments typically take place at low densities, where the exciton density is several orders of magnitude below the Mott density. At higher densities, conventional wisdom has been that strong coupling is lost, and the system reverts to a standard photon laser^{15,16}. The mechanism of this loss of strong coupling is still not very well understood, with only a few theoretical^{17,18} and fewer experimental works^{19} analysing this regime in detail.
In this study, we directly probe the high density regime of excitonpolariton condensates towards the Mott density. We observe a photoluminescence (PL) sideband in the higherenergy side from the main peak. By the excitationpower dependence of the peakenergy separation, this is different from the upperpolariton peak and cannot be explained by the singleemitter model. The measured PL spectra shows that this is not taken as the conventional semiconductor lasing. Furthermore, we study the PL spectra based on the nonequilibrium ehp model including BCS physics, which allows for a treatment in whole regime from nonequilibrium lasing to equilibrium BECs, BCS states and inbetweens. While our measured PL spectra consist of the main peak and the highenergy one, this theory predicts the asymmetric triplet peaks. Our observation has a potential to demonstrate a strong coupling of an electron and a hole under a lasing phase and further leads to deepen nonequilibrium and dissipative manybody physics.
Results
Interacting electronholephoton (ehp) model
To discuss a high density regime of an excitonpolariton system in a semiconductor microcavity, the BCS physics^{20,21,22,23} and lasing phase, i.e., an e–h–p system in a nonequilibrium state^{23} should be simultaneously discussed. In this section, we qualitatively and quantitatively explain the interplay between the BCS physics and lasing phase from the viewpoint of PL spectra.
In the standard BCS theory without photons, it is wellknown that the spontaneous formation of coherence opens an energy gap in the excitation spectrum, which can be described by the BCS gap equation. In the e–h–p system, the corresponding energy gap in the spectrum should exist when the coherence is developed. Energy dispersion of the photondressed carriers in the e–h–p system are shown in Fig. 1B. The mechanism generating the energy gap can be understood in analogy with a twolevel atomic system (Fig. 1A). Dressed atom states are formed when the cavity photon field^{24} and the twolevel atom transition are in resonance and strongly coupled (say the coupling strength g^{JC} and the resonance energy μ). The resulting dressed state energy is split into two, where the energy difference between the two energies, 2g^{JC}(n)^{1/2} for the total excitation number n, corresponds to the rate of the coherent energy transfer between the twolevel system and the cavity photon field. As n increases, their coherent energy transfer occurs more rapidly; therefore, the energy splitting becomes large. In the limit of large n, the four possible radiative transitions from n + 1 to n sectors generate three emission peaks at μ and μ ± Ω_{R} (two of them degenerate at μ, and Ω_{R }≈ 2g^{JC}(n)^{1/2}). This is the cavity system version of the Mollow triplet in resonance fluorescence^{25,26}. We point out that the Mollow triplet is a direct signature of the coherent coupling between the matter and light fields, i.e. strong coupling. In the present e–h–p system, the corresponding energy gap is observed around the momentum where the valence band with n + 1 photons and the conduction band with n photons coincide (resonance condition), i.e. when there appears coherence (Fig. 1B). However, the gap formation is not only by the standard Rabi splitting due to the strong light field^{24} (large n), but further assisted by the BCS e–h Coulomb correlations^{21,22,27,28,29}. In the same way as the standard Mollow triplet with a single atom, the energy gap will show the Mollowtype triplet, which becomes a signature of coherence of the photons with the eh system and direct evidence of strong coupling.
In what follows, we have employed the recently developed formalism described in ref. 23 in order to investigate the equilibrium and nonequilibrium nature of the polariton condensate in the highdensity regime. In this formalism, electrons (the energy dispersion ), holes (the energy dispersion ), and photons (the energy dispersion ) are explicitly treated with the Coulomb interactions between carriers (U_{q}) and the lightmatter coupling (g) within the dipole approximation. One of the advantages of this approach is the applicability to highdensity regimes because electrons and holes are explicitly treated with their Coulomb interactions, in contrast to the other approaches assuming excitons as bosonic particles such as dissipative GrossPitaevskii equations. By taking into account the pumping baths and the vacuum photon bath, a closed set of equations for the cavity photon field a_{0}, the polarization function p_{k}, and the number of electrons n_{e, k} and holes n_{h, k} can be derived^{17,18} within the HartreeFock (HF) approximation for the steady state as
Here, and are the (Coulombrenormalized) energies of the particles measured on a rotating frame with an oscillation frequency μ, N_{k} ≡ n_{e, k} + n_{h, k} −1 is the population inversion, is the generalized Rabi frequency which represents coherence of the system, κ is the photon loss rate, and γ is the thermalization rate of the eh system. Here, we note that μ can be viewed as an additional unknown variable because the oscillations of a_{0} and p_{k} can be eliminated if μ is appropriately determined. Equations (1, 2, 3) are formally the same as the Maxwell–semiconductor–Bloch equations (MSBE) under the relaxation time approximation (RTA) but the major difference is that and are given by
through the retarded Green’s function,
the single particle spectral function,
and the Fermi distribution function in the electron (hole) pumping bath, , with the chemical potential and inverse temperatureβ (≡ k_{B}T). As a result, if ω_{e, k} = ω_{h, k} is assumed with , and γ and T are assumed to be small for simplicity (all assumed hereafter unless otherwise specified), the above equations result in the BCS gap equation when (I) min[2E_{k}] ≳ μ_{B}–μ (quasiequilibrium) with . In contrast, the MSBE under the RTA, which describes semiconductor lasers, can be recovered when (II) μ_{B} – μ ≳ min[2E_{k}] (lasing; nonequilibrium). This formalism is referred to as the BECBCSLASER crossover theory within the HF approximation.
Here we find from Eqs (6) and (7) the single particle spectral functions for electrons (holes) are given by
having peaks at ν = ± E_{k} given by
and weights, u_{k}^{2} and v_{k}^{2} respectively, given by the Bogoliubov coefficients,
where and . In Eqs (8)–(10), it is important to notice the remarkable similarities to the standard BCS theory describing superconductors. These single particle spectral functions describe the renormalized single particle energies, and it is wellknown that gaps are opened around the energy ±μ/2 with the magnitude of min[2E_{k}] (the minimum of 2E_{k} when the wavenumber k is scanned). The picture shown in Fig. 1(B) can thus be obtained, and one can now notice that the Mollow’s dressed picture in semiconductor is conceptually connected to the gaps in the BCS theory. However, we have to emphasize that the unknown variables in Eqs (8–10) are determined by the BECBCSLASER crossover theory rather than the BCS theory. The PL spectra shown later can be roughly discussed with the knowledge of the steady state populations in Eqs (1, 2, 3), and the singleparticle spectral functions (with the energy dispersion E = ±E_{k}) in Eq. (8), while they are calculated from the photon Green’s function in the practical numerical simulation (see Supplementary Information S3. 3 ).
Experiment
The time evolution of the PL spectra of the high density excitonpolariton condensates after a pulse excitation is studied by using a streak camera as shown in Fig. 2. Figure 2A is an example of the timeresolved PL spectra obtained at strong pump power (P/P_{th} = 75) as a function of time after a triggering pulse’s arrival at the streak camera, where the strong PL signal is observed at around an instant (Time~100 ps) with a strong main peak at 1.612 eV and an extra highenergy peak at 1.622 eV (the PL spectra at the instant is shown in Fig. 2B). The strong emission is followed by a relaxation decay of a hundred picosecond timescale. During the decay processes, the emission intensity decreases, while the main emission peak energy gradually decreases and is considered to approach finally to that of the lower polariton (LP) ground state (The decay process is focused in Fig. S9 in the Supplementary Information).
A remarkable finding, which is the main focus of this paper, is the emergence of an extra highenergy side peak, which is found only under strong pumping far above the condensation threshold, P/P_{th}≳ 20. This is clearly seen in Fig. 2C, a collection of such timeresolved PL spectra (at the instant of strong emission, similarly to Fig. 2B) taken at various pump power, 0.6 < P/P_{th} < 340. The blue shift is observed in the energies of the two peak emissions (main and side) as the pump power increases, and the main peak approaches the bare cavity photon energy (or slightly above it) at the highest pump power. The feature found in the main emission peak is consistent with past predictions of a high density polariton state^{1,20,21,22,30} and also with our simulation. As for this highenergy side peak, the emission energy increases with the pump power, and is clearly different from that of the upper polariton (UP) since it gradually evolves from the main peak energy (as predicted for cavity system in ref. 25). The pump power dependence of the energy separation from the main peak is shown in Fig. 2D. In contrast to the main peak, these observations as for the highenergy side peak require further explanation, since it is far beyond the prediction of literature; it is widely believed that the polariton condensate change its nature to conventional photon lasing at high density, and conventional photon lasing does not result in such a highenergy side peak. The deviation from the conventional photon lasing has also been supported by the temperature dependence of the PL in the same sample^{19}.
This side peak with the pumppower dependent peak separation reminds us of Mollow triplet spectra in coherently driven twolevel emitters discussed above (resonance fluorescence). Actually, the Mollowtriplet sidepeak separation is known to have a squareroot dependency on the pump power which is not too far from our observation (the square root dependency is shown by a black solid line in Fig. 2D). Of course, a considerable deviation from the square root dependency is not surprising as it is also seen in the theoretical results in the Supplementary Material. Before all, our emission sources are semiconductor carriers much more complicated than single twolevel atom systems due to e.g. the dispersion, dephasing, higherorder Coulomb effects including carrierinduced relaxation, and carrier heating etc. However, the most mysterious deviation from the conventional resonance fluorescence is that the lowenergy side peak theoretically predicted^{22,23,25} is missing at any excitation power in our experiments.
Comparison between experiment and theory
Now, let’s see what predictions on the PL spectra can be drawn from the simulation by our interacting ehp model, and then, compare them with our observation of the highenergy side peak emission.
Our theory predicts that the system is in quasiequilibrium in the lowexcitation regime when the quasiequilibrium condition (I) is satisfied. This condition, (I) the gap energy (=min[2E_{k}]) is larger than the difference μ_{B} −μ, means that the Fermi level of the pumping bath does not exceed the energy gap as shown in Fig. 3A. In this case, it is expected that the highenergy peak in the triplet does not become bright since the carriers cannot be supplied to the renormalizedband states above the gap (namely, the emission channel at E > μ +min[2 E_{k}] is closed). In contrast, in the highexcitation regime, the e–h–p system behaves differently from it would in the lowexcitation regime, and the system enters the nonequilibrium regime when another condition (II) is satisfied; the difference μ_{B} − μ becomes larger than the gap energy, min[2E_{k}]. In this case, the pumping baths can supply the carriers above the energy gap (Fig. 3B). Once this condition is satisfied, the emission channel at E > μ + min[2E_{k}] would be opened and the highenergy side peak emission begins to occur. These expectations are indeed confirmed by the simulated PL spectra shown in Fig. 4A,B, respectively, under the quasiequilibrium (I) and nonequilibrium (II) conditions (the small but nonzero highenergy side peak intensity in Fig. 4A is due to the nonzero thermal/quantum fluctuations). From this observation, a short conclusion from our ehp model is drawn; the bright highenergy side peak emission occurs only in the nonequilibrium condition.
We note, however, that manybody effects also play an important role in the simulated PL spectra. In semiconductor materials (with no cavity), it is well known that e–h Coulomb interactions can cause significant enhancement of the PL intensity around the Fermi edge at low temperature, even for a plasma state^{31,32}. As found in our previous studies on the carrier population and optical gain spectra^{17,18}, this effect was shown to survive also in our case (Fig. 4B). Furthermore, the Fermiedge enhancement becomes more pronounced when the difference of the Fermi energies (~μ_{B}) roughly coincides with the energy separation (μ + min[2E_{k}]) between the upper and lower edges of the gaps (see also Fig. 3)^{18}. As a result, the highenergy peak is further enhanced and exceeds the low energy peak when μ_{B} − μ ~ min[2E_{k}] in our calculations. It is also instructive to note that μ_{B} − μ ~ min[2E_{k}] is satisfied at the pump power where the system crosses over from the quasiequilibrium states into lasing states^{17}. Therefore, we can draw the second conclusion; the PL spectra exhibits asymmetric triplets with stronger highenergy side peak (than the lowenergy side peak) near the crossover regime into lasing (see also Fig. 9(d) of ref. 18). We note that such asymmetric PL spectra are never obtained for noninteracting twolevel atom models (see also the Supplementary Information and Fig. S6), showing an impact of the manybody effect present in highlyexcited ehp systems. By seeing the result, we have to recognize that our theoretical results show a large quantitative discrepancy from the measured PL spectra, since the latter exhibit only the main and highenergy side peak emissions with the missing lowenergy sidepeak emission. However, it is also interesting to notice a qualitative similarity between the theory and experiment, the nonmonotonic pumppower dependence of the highenergy peak separation (theory: blue points in Fig. 4C, and experiments: Fig. 2D); the energy separation (i.e. the gap, min[2E_{k}], in theory) increases with pump power, but decreases eventually at too strong pumping. If the gap is regarded as the binding energy of e–h pairs, the simulation result also implies that the eh pair binding survives at strong pumping above the crossover into lasing^{17}, and gradually quenches with increasing pump power where the e–h plasma is eventually formed.
Next, let us study whether the nonequilibrium condition, which is the requirement for the highenergy side peak to be found in our theory, is satisfied or not in the experiments, especially in the highexcitation regime. For this purpose, in Fig. 5A, we show the pumppower dependence of the PL intensity obtained in experiments. The second nonlinear increase in the PL intensity as reported in various studies^{9,15,16,33,34,35} is not seen in our experiments, even at our highest available pump power. This also implies that conventional photon lasing with e–h plasma gain does not occur here^{19}. A discrepancy exists between the experimental and theoretical PL intensities (Fig. 5A,B) regarding whether the second nonlinear increase occurs. We should discuss this point more carefully from the theoretical viewpoint; the simulation results depend on the parameters; roughly speaking, the second nonlinear increase (namely, the second threshold) is less visible for larger broadening factors by increasing T, γ, and κ, which is understood by the precise criteria (including these parameters) for the quasiequilibrium/nonequilibrium conditions; there is a crossover regime between them, i.e. min[2E_{k}] (2γ +2k_{B}T) ≲ μ_{B}−μ ≲ min[2E_{k}] + (2γ + 2k_{B}T), and therefore, the threshold feature becomes clear only if the crossover regime is negligibly small. We verified this understanding by simulating the temperature dependence of the PL intensity (not shown). The third conclusion drawn from this consideration is that the second nonlinear increase of the PL intensity cannot be a good clue indicating that the system transitions into the nonequilibrium regime. Here, we would like to add as well that the second nonlinear increase is not necessarily related to the e–h pair breaking, as already shown in ref. 17.
In order to clarify whether the ehp systems were in the quasiequilibrium or nonequilibrium conditions in our experiments, the dependence of the PL intensity on the PL energy of the main peak (μ in theory) is shown in Fig. 5C,D. As shown above, under the quasiequilibrium condition (μ_{B} −μ ≲ min[2E_{k}]), the ehp system is described within the thermal equilibrium theory, where the main peak energy μ is regarded as the chemical potential of the ehp systems. In this case, the cavity photon number should diverge when μ approaches the bare cavity energy^{21} (the right vertical dashed line) as plotted by a black line in Fig. 5D. This divergence originates from the basic physics of thermalequilibrium theory of bosons, which are photons in our case. This prediction leads to a criterion to distinguish between the quasiequilibrium and nonequilibrium states; if one finds nondivergent PL intensity with the main emission energy (=μ) reaching the bare cavity energy, it gives a proof to show the system is in the nonequilibrium. In the theoretical result (filled circles in Fig. 5D), a large discrepancy from the thermal equilibrium result (a black line) is found as μ approaches the bare cavity energy, where the system is found to be in the nonequilibrium. The beginning point of this discrepancy (μ − E_{cav} ~ −4 meV) indicates the crossover regime between the quasiequilibrium and nonequilibrium, which fully agrees with that obtained directly from the condition μ_{B} − μ ~ min[2E_{k}] (the two different regimes indicated by the red and blue plots in Fig. 5D). The experimental data in Fig. 5C clearly shows nondivergent PL intensity even when the main peak energy (μ) reaches the bare cavity energy. Therefore, we can safely state that the observed PL data taken at the regime (μ ~ E_{cav}) come from the nonequilibrium states of the ehp system. Furthermore, we show the region containing a highenergy side peak in the PL spectra by the blue shaded areas for both the experiments (Fig. 5C) and theory (Fig. 5D). Especially for the experimental results in Fig. 5C, the shaded area with the highenergy side peak emission largely overlaps with the theoretically supported nonequilibrium regime (μ ≥ E_{cav}). This consistency partly supports our theoretical interpretation of the highenergy side peak. However, by comparing Fig. 5C with Fig. 5D, we have to stress that the theoretical and experimental data do not fit quantitatively.
Discussion
In conclusion, we have performed a study of high density excitonpolariton condensates towards the Mott density and observed a highenergy sideband PL. We have compared this to a theory of nonequilibrium ehp system, generalizing polariton BCS theory to the nonequilibrium regime. After the comparison, several disagreements between our experiment and theory still exist. We point out that this theory is only one possible explanation of the physics observed in our experiment. Further work would be required to show that the observations are consistent with other possible explanations, some of which are discussed below.
Here, we add some discussions on the discrepancy between the theory and experiments, and mention some important factors and the effects which are neglected in our model. One is the observed relaxation dynamics after a pulse excitation (Fig. 2A), while the theory is dealing with the stationary state. Therefore, whether the stationary condition is fulfilled remains questionable. If the relaxation time to the (transient) stationary state (corresponding to Time ~100 ps in Fig. 2A) is taken into account, the highenergy peak emission could be enhanced, since the carriers initially supplied from the highenergy region. Dephasing omitted in our meanfield theory is known to enhance the emission from offresonant modes (the sidepeak emission in our case) in cavityQED research^{36,37,38,39}. It is also wellknown to destroy coherence and reduce the gap (the peak separation in our case) in the condensed matter research. Besides, the spontaneous emission from the quantum wells directly into free space was also neglected in our theory. Taking all these effects into account in the theory might reduce the large quantitative discrepancy from the experimental results, which is far beyond the scope of this paper.
One may seek for the origin of the highenergy side peak by other explanations different from our coherent ehp coupling scenario. In particular, the singleemitter Mollow triplet in the presence of detuning and dephasing has been shown to give an asymmetric Mollow spectrum^{40}. However, our experiments were performed at high densities towards the Mott density. Therefore, it is essential to take into account of the underlying Fermionic nature of the electrons and holes together with their Coulomb interaction. While a singleemitter Mollow triplet under suitable conditions may have superficial similarities, we believe that it is less plausible than the theoretical analysis we have presented in this work.
Another scenario would be for instance that the upper energy peak could be the band to band transition, with the lower energy peak being the bare cavity mode. We believe that it is unlikely that the bandedge emission account for the upper energy peak as the renormalized band edge is lower in energy than the cavity resonance (=QW exciton resonance) in this high density regime^{27}. In contrast, the Fermiedge emission, which was actually observed in Kim et al.^{41} in highlyexcited semiconductor QWs without coupling to a cavity by using a streak camera, can be a possible candidate for our upper energy peak. For this scenario, the Fermiedge should lower as time proceeds due to the radiation decay, hence, the emission energy should show a red shift, e.g., from 1.622 eV to 1.612 eV in Fig. 2A. In this case, there are two possibilities: (i) the red shift of the Fermiedge emission energy is very fast and occurs within the time resolution of our streak camera. This could be possible because the radiative decay rate should be higher than Kim et al. in our system with the microcavity and higher excitation density. (ii) the red shift of the emission energy occurs in a time scale longer than the timeresolution. In the former case (i), the observed PL spectra at the emission time should consist of a strong emission at cavity energy 1.612 eV plus a broad tail spread between 1.622 eV and 1.612 eV, which is quite different from the observed PL spectra with clear two peaks (e.g. Fig. 2B). Thus, this possibility can be safely excluded. In the latter case (ii), the red shift should have been observed by our streak camera. However, our experimental data does not exhibit such red shift (as in Fig. 2A), at any pump power with a welldefined side peak. Therefore, the latter possibility is also excluded. In either case, the Fermiedge emission does not account for the observed highenergy side peak.
On the other hand, this absence of the red shift of the highenergy peak (as in Fig. 2A) does not contradict with our coherent ehp coupling scenario. In our theory, the highenergy peak emission occurs only if the Fermiedge (=μ_{B} − μ, measured from the main emission) is close to the gap energy, whereas the gap energy robustly stays at the same energy position as long as the strong main peak emission exists (the intensity ~κ a_{0}^{2}). As time proceeds, the Fermiedge is quickly lowered and detuned from the gap energy. This violates the requirement for the emission to occur, resulting in a sudden quench of the upper energy emission as found in Fig. 2A.
Methods
We used an AlAs/AlGaAs distributed Bragg reflector microcavity sample in which 12 GaAs quantum wells (QWs) are embedded at central antinodes of the cavity photon field^{18,29}. The number of top (bottom) layers used to obtain the PL from the top surface is 16 (20). The 12 QWs are divided into 3 groups and positioned at the 3 highest mode intensity antinodes of the microcavity. The detuning between the microcavity photon energy and the QW exciton at around 1.612 eV is close to zero at the 8 K temperature of the present work; the normal mode splitting at zero inplane momentum is 14 meV. A modelocked Ti:Sapphire laser with a 76MHz repetition rate and an energy around 1.67 eV is used as the pump laser to utilize its 3ps pulse width for the aboveband excitation. The laser injection angle into the sample is around 50–60° from the normal which corresponds to k ~ 7 × 10^{4}/cm^{18}. The pump laser energy is set to maximize the injection rate into the sample reflection dip due to the cavity photon mode. The PL from the sample is focused onto the entrance slit of a spectrometer attached in front of a streak camera or a timeintegrated CCD camera by an objective lens and subsequent lenses, and then is reflected at the grating to extract PL energy. In this study, the central area (about 1.2 μm × 1 μm) of the pump laser spot with ~50 μm diameter is selected by the horizontal and vertical slits of the spectrometer and the streak camera. Hence, the outside of the central area with significantly different pumping intensities do not affect the observed signal.
Additional Information
How to cite this article: Horikiri, T. et al. Highenergy sidepeak emission of excitonpolariton condensates in high density regime. Sci. Rep. 6, 25655; doi: 10.1038/srep25655 (2016).
Change history
02 November 2016
A correction has been published and is appended to both the HTML and PDF versions of this paper. The error has not been fixed in the paper.
02 November 2016
References
Keeling, J. Eastham, P. R. Szymanska, M. H. & Littlewood, P. B. BCSBEC crossover in a system of microcavity polaritons. Phys. Rev. B 72, 115320 (2005).
Keldysh, L. V. & Kopaev, Y. V. Possible instability of the semimetallic state toward Coulomb interaction. Sov. Phys. Solid State 6, 2219 (1965).
Comte, C. & Nozières, P. Exciton Bose condensation: the ground state of an electronhole gas I. Mean field description of a simplified model. J. Physique 43, 1069 (1982).
Pieri, P. Neilson, D. & Strinati, G. C. Effects of density imbalance on the BCSBEC crossover in semiconductor electronhole bilayers. Phys. Rev. B 75, 113301 (2007).
Yamashita, K. Asano, K. & Ohashi, T. Quantum condensation in electron–hole bilayers with density imbalance. Journal of Physical Society of Japan 79, 033001 (2010).
Parish, M. M. Marchetti, F. M. & Littlewood, P. B. Supersolidity in electronhole bilayers with a large density imbalance. Euro. Phys. Lett. 95, 27007 (2011).
Versteegh, M. van Lange, A. J. Stoof, H. T. C. & Dijkhuis, J. I. Observation of preformed electronhole Cooper pairs in highly excited ZnO, Phys. Rev. B 85, 195206 (2012).
Kasprzak, J. et al. Bose–Einstein condensation of exciton polaritons. Nature 443, 409 (2006)
Deng, H. et al. Condensation of semiconductor microcavity exciton polaritons. Science 298, 199 (2002).
Deng, H., Haug, H. & Yamamoto, Y. Excitonpolariton BoseEinstein condensation. Rev. Mod. Phys. 82, 1489 (2010).
Carusotto, I. & Ciuti, C. Quantum fluids of light. Rev. Mod. Phys. 85, 299 (2013).
Byrnes, T., Kim, N. Y. & Yamamoto, Y. Excitonpolariton condensates. Nature Phys. 10, 803 (2014).
Wouters, M. & Carusotto, I. Excitations in a nonequilibrium BoseEinstein condensate of exciton polaritons. Phys. Rev. Lett. 99, 140402 (2007).
Amo, A. et al. Superfluidity of polaritons in semiconductor microcavities. Nat. Phys. 5, 805 (2009).
Dang, L. S. Heger, D. Andrè, A. Boeuf, F. & Romenstain, R. Stimulation of polariton photoluminescence in semiconductor microcavity. Phys. Rev. Lett. 81, 3920 (1998).
Tempel, J. S. et al. Characterization of twothreshold behavior of the emission from a GaAs microcavity. Phys. Rev. B 85, 075318 (2012).
Yamaguchi, M. et al. Second thresholds in BECBCSLaser crossover of excitonpolariton systems. Phys. Rev. Lett. 111, 026404 (2013).
Yamaguchi, M. et al. Generating functional approach for spontaneous coherence in semiconductor electronholephoton systems. Phys. Rev. B 91, 115129 (2015).
Horikiri, T. et al. Temperature Dependence of Highly Excited Exciton Polaritons in Semiconductor Microcavities. J. Phys. Soc. Jpn. 82, 084709 (2013).
Szymanska, M. H. Keeling, J. & Littlewood, P. B. Nonequilibrium quantum condensation in an incoherently pumped dissipative system Phys. Rev. Lett. 96, 230602 (2006).
Kamide, K. & Ogawa, T. What determines the wave function of electronhole pairs in polariton condensates? Phys. Rev. Lett. 105, 056401 (2010).
Byrnes, T. Horikiri, T. Ishida, N. & Yamamoto, Y. BCS wavefunction approach to the BECBCS crossover of excitonpolariton condensates. Phys. Rev. Lett. 105, 186402 (2010).
Yamaguchi, M. Kamide, K. Ogawa, T. & Yamamoto, Y. BEC–BCSlaser crossover in Coulombcorrelated electron–hole–photon systems. New. J. Phys. 14, 065001 (2012).
Mollow, B. R. Power spectrum of light scattered by twolevel systems. Phys. Rev. 188, 1969 (1969).
SanchezMondragon, J. J. Narozhny, N. B. & Eberly, J. H. Theory of spontaneousemission line shape in an ideal cavity, Phys. Rev. Lett. 51, 550 (1983).
Quochi, F. et al. Strongly driven semiconductor microcavities: From the polariton doublet to an ac Stark triplet. Phys. Rev. Lett. 80, 4733 (1998).
Haug, H. & Koch, S. W. Quantum Theory of the Optical and Electronic Properties of Semiconductors 5th ed. (Singapore: World Scientific, 2009).
SchmittRink, S. Chemla, D. S. & Haug, H. Nonequilibrium theory of the optical Stark effect and spectral hole burning in semiconductors. Phys. Rev. B 37, 941 (1988).
Henneberger, K. & Haug, H. Nonlinear optics and transport in laserexcited semiconductors. Phys. Rev. B 38, 9759 (1988).
Ishida, N. et al. Photoluminescence of highdensity excitonpolariton condensates. Phys. Rev. B 90, 241304(R) (2014).
SchmittRink, S. Ell, C. & Haug, H. Manybody effects in the absorption, gain, and luminescence spectra of semiconductor quantumwell structures. Phys. Rev. B 33, 1183 (1986).
Skolnick, M. S. et al. Observation of a manybody edge singularity in quantumwell luminescence spectra. Phys. Rev. Lett. 58, 2130 (1987).
Bajoni, D. et al. Polariton laser using single micropillar GaAsGaAlAs semiconductor cavities. Phys. Rev. Lett. 100, 047401 (2008).
Nelsen, B. Balili, R. Snoke, D. W. Pfeiffer, L. & West, K. Lasing and polariton condensation: Two distinct transitions in GaAs microcavities with stress traps. J. Appl. Phys. 105, 122414 (2009).
Tsotsis, P. et al. Lasing threshold doubling at the crossover from strong to weak coupling regime in GaAs microcavity. New J. Phys. 14, 023060 (2012).
Hennessy, K. et al. Quantum nature of s strongly coupled single quantum dotcavity system. Nature 445, 896 (2007).
Winger, M. et al. Explanation of Photon Correlations in the FarOffResonance Optical Emission from a QuantumDot–Cavity System. Phys. Rev. Lett. 103, 207403 (2009).
M. Settnes, P. Kaer, A. Moelbjerg & J. Mork Auger Processes Mediating the Nonresonant Optical Emission from a Semiconductor Quantum Dot Embedded Inside an Optical Cavity. Phys. Rev. Lett. 111, 067403 (2013).
Yamaguchi, M. Asano, T. & Noda, S. Third emission mechanism in solidstate nanocavity quantum electrodynamics. Rep. Prog. Phys. 75, 096401 (2012).
Valle, E. & Laussy, F. P. Regimes of strong lightmatter coupling under incoherent excitation. Phys. Rev. A 84, 043816 (2011).
Kim, J. H. et al. Fermiedge superfluorescence from a quatumdegenerate electronhole gas. Sci. Rep. 3, 3283 (2013).
Roumpos, G. et al. Single vortex–antivortex pair in an excitonpolariton condensate. Nature Phys. 7, 129 (2011).
Acknowledgements
The authors thank P. Littlewood, J. Keeling, I. Carusotto, M. KuwataGonokami, and M. Nomura for their helpful comments. This research was supported by JSPS through its FIRST Program and KAKENHI Grant Numbers 20104008, 24740277, 25800181, 26287087, 26790061 and 16K05492, Space and Naval Warfare Systems (SPAWAR) Grant N660010912024, the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), the Project for Developing Innovation Systems of MEXT, the National Institute of Information and Communications Technology (NICT), the joint study program at the Institute for Molecular Science, the Shanghai Research Challenge Fund, NSFC, and NTT Basic Laboratories. T.H. is supported by Toray Science Foundation, KDDI foundation, and REFEC. T.B. is supported by the Shanghai Research Challenge Fund, and National Natural Science Foundation of China grant 61571301, the Thousand Talents Program for Distinguished Young Scholars, and the New York University Global Seed Grants for Collaborative Research.
Author information
Authors and Affiliations
Contributions
T.H. performed optical experiments. M.Y. performed theoretical calculations. A.L., S.H. and A.F. grew and fabricated the sample. T.H. and Y.Y. designed the experiment. M.Y., K.K., T.B., N.I., Y.S. and T.O. prepared the theoretical work. T.H. and Y.M. analysed data. T.H., M.Y., K.K., Y.S. and T.B. prepared the manuscript. T.O., S.H. and Y.Y. led the discussion. Y.Y. organized this work.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Supplementary information
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Horikiri, T., Yamaguchi, M., Kamide, K. et al. Highenergy sidepeak emission of excitonpolariton condensates in high density regime. Sci Rep 6, 25655 (2016). https://doi.org/10.1038/srep25655
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/srep25655
Further reading

Quantum technology applications of excitonpolariton condensates
Emergent Materials (2021)

New optical solutions of complex Ginzburg–Landau equation arising in semiconductor lasers
Applied Physics B (2019)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.