Ultrafast exciton formation at the ZnO(10{\overline{\textbf{1}}}0) surface

Ultrafast exciton formation at the ZnO(100) surface

J.-C. Deinert deinert@fhi-berlin.mpg.de    D. Wegkamp    M. Meyer    C. Richter    M. Wolf    J. Stähler Fritz-Haber-Institut der Max-Planck-Gesellschaft, Abteilung Physikalische Chemie, Faradayweg 4-6, 14195 Berlin
July 18, 2019
Abstract

We study the ultrafast quasiparticle dynamics in and below the ZnO conduction band using femtosecond time-resolved two-photon photoelectron spectroscopy. Above band gap excitation causes hot electron relaxation by electron-phonon scattering down to the Fermi level followed by ultrafast (200 fs) formation of a surface exciton (SX). Transient screening of the Coulomb interaction reduces the SX formation probability at high excitation densities near the Mott limit. Located just below the surface, the SX are stable with regard to hydrogen-induced work function modifications and thus the ideal prerequisite for resonant energy transfer applications.

pacs:
79.60.Bm, 78.47.J-, 73.20.Mf, 73.20.At

with kind permission from the American Physical Society, published as Phys. Rev. Lett. 113, 057602 (2014)

The technological importance of zinc oxide (ZnO) originates from its large direct band gap () and high bulk exciton binding energy (60 meV) Wöll (2007), which make it a promising candidate for optoelectronic applications, using ZnO nanoparticles Richters et al. (2008); Slootsky et al. (2014); Klingshirn et al. (2010), epilayers Klingshirn et al. (2010), and hybrid organic-inorganic systems Della Sala et al. (2011); Xu et al. (2013); Slootsky et al. (2014). In this context, the optical and electronic properties of ZnO surfaces naturally play a significant role with regard to electronic and excitonic coupling with other materials, e.g., organic dye molecules. As dipole-dipole coupling is highly distance dependent Förster (1948), the rates for Förster resonance energy transfer (FRET) between different materials depend strongly on the presence of exciton dead layers at the surface Egelhaaf and Oelkrug (1996); Fonoberov and Balandin (2004a); Kiliani et al. (2011) or, on the contrary, the existence of surface excitons. The latter are a dominant species in photoluminescence (PL) studies of different types of ZnO nanostructures due to the comparably large surface-to-bulk ratio Fonoberov and Balandin (2004b); Richters et al. (2008); Kuehn et al. (2013); Slootsky et al. (2014). Despite the great scientific attention, detailed understanding of this technologically highly important species is still missing, as the systematic modification and characterization of nanoparticle surfaces is very challenging. Because of the lack of surface sensitivity of optical probes and PL, SX could only once be identified at a ZnO single crystal surface Travnikov et al. (1990). However, as the sample had been exposed to air, the surface condition was rather undefined. Neither observation nor systematic characterization of SX under ultrahigh vacuum (UHV) conditions, which provide reproducible surfaces at the nanoscale level Wöll (2007); Diebold et al. (2004), is known to date, and thus the origin and character of this very relevant species remains vague.

In this Letter, we use femtosecond (fs) time-resolved two-photon photoelectron (2PPE) spectroscopy to investigate the ultrafast carrier and exciton dynamics at the single crystal ZnO() surface under well-defined UHV conditions. Due to the intrinsic surface sensitivity of this technique we are able to observe the ultrafast relaxation of hot electrons in the conduction band and the subsequent formation an excitonic state near the surface on sub-picosecond timescales. High excitation densities near the Mott limit enhance the screening of the electron-hole (e-h) Coulomb interaction and thus reduce the formation probability of this state. Remarkably, even strong modification of the (static) surface charge density by hydrogen termination does not change the population dynamics of the SX, strongly suggesting that the exciton is localized in the subsurface region. This adjacence of the exciton to the surface (1–2 nm) combined with its stability with respect to surface modification demonstrates its direct relevance for applications of ZnO involving resonant energy transfer.

Figure 1: (color online). (a) 2PPE principle and relevant processes. Photoexcited () electrons relax by electron-phonon scattering (1) and exciton formation (2), which is probed by a time-delayed laser pulse () in photoemission. (b) Exemplary photoelectron spectrum of the hydrogen-terminated ZnO surface (black). The low-energy cut-off is a measure of the work function; the intensity below results from the CAL at the surface. Above band gap excitation leads to additional 2PPE intensity around (blue). Subtraction of the equilibrium background signal yields the pump-induced spectrum (purple). (c) Impact of hydrogen termination on sample work function and CAL intensity.

Figure 1(a) depicts the principle of time-resolved 2PPE spectroscopy. A first laser pulse initiates nonequilibrium dynamics by excitation of electrons from the valence band (VB) into normally unoccupied states in the conduction band (CB). A second laser pulse subsequently photoemits the excited electron population. The temporal evolution is monitored by variation of the time delay between pump and probe pulse 111Laser pulses were provided by a regeneratively amplified (RegA), tuneable fs-laser (200 kHz). The frequency-doubled output of an optical parametric amplifier provided and the third (4.65 eV) and fourth harmonic (6.2 eV) of the 800 nm RegA output were used for photoemission. The photoelectrons were detected using a hemispherical electron energy analyzer (PHOIBOS 100) held at a fixed bias voltage (1.5 V) with respect to the sample. Photoelectron (PE) spectra are referenced to the, in situ measured, tantalum sample holder Fermi energy , which was in electrical contact with the sample surface.

We investigated three different hydrothermally grown ZnO single crystals (Mateck) and found no quantitative differences. Their nonpolar, mixed-terminated () surfaces were prepared by sputtering and annealing cycles () under UHV conditions (). The VB maximum remains unchanged at for different types of preparations using oxygen background pressures up to (not shown). Neither charging, nor surface photovoltage effects have been observed 222As discussed further below, the observed spectral features exhibit a neither fluence nor temperature dependence.. Unless otherwise stated, experiments were performed at . Incident fluences between translate to maximum excitation densities of , which is below or close to the Mott limit 333Literature values for the Mott density vary between Versteegh et al. (2011); Hendry et al. (2007); Schleife et al. (2011). The calculated maximum excitation densities give an upper limit for the absorption directly at the surface.. Hydrogen termination of the surface was achieved by background dosing of 444The glowing filament of an ion gauge was set in line of sight with the sample surface leading to an increased dissociation of ..

Exposure of a semiconductor surface to an electron donor (like hydrogen) can lead to downward surface band bending. In the case of the intrinsically n-doped ZnO, this leads to the formation of a few thick charge accumulation layer (CAL) at the () surface Eger et al. (1976); Lüth (2010); Wang et al. (2005) with a maximum surface charge density of Ozawa and Mase (2011). We monitor the CAL formation by (1) the occurrence of photoelectron intensity directly below as shown in Fig. 1(b) and (2) by a reduction of the sample work function that is correlated with the surface band bending as sketched in Fig. 1(a). Figure 1(c) shows that increasing exposure leads, in agreement with literature Ozawa and Mase (2011), to an increasing CAL intensity (blue curve) 555The sticking coefficient is unknown and dissociation of occurs at filaments in the chamber plays a role, which inhibits quantitative comparison of the exposures to Ref. Ozawa and Mase (2011)., which is accompanied by a work function decrease that saturates at . This change of the surface dipole is a result of the electron donor character of the hydrogen, leaving the positively charged ion at the surface. Directly after preparation of a pristine ZnO surface, a slight but continuous increase of CAL intensity going along with a work function reduction of several 10 meV is always observed due to the inevitable background even in a UHV environment (). In order to stabilize the PE signal and to reduce the work function, we exposed the sample (unless otherwise stated) to of before experiments, which corresponds, based on Ref. Ozawa and Mase (2011), to a charge density in the CAL on the order of and a termination of surface oxygen of only a few percent.

Figure 2: (color online). (a) Ultrafast electron dynamics at the ZnO surface below the Mott limit as probed by 2PPE spectroscopy. False colors represent photoelectrons created/depleted by . Hot carriers in the CB relax on fs timescales by optical phonon emission. After a few 100 fs, additional electrons are verified below , indicating that the surface exciton has formed. The signal at negative delays and energies results from photoelectrons with reversed excitation scheme (pumped by 4.6 eV and probed by 4.2 eV) and are a replica of the above dynamics shifted to 0.4 eV lower final state energies and evolving towards negative delays. (b) Pump and probe XC traces at indicated energies above (hot electrons) and below (exciton) and the corresponding double exponential fits to the data (solid lines). The instrument response function is represented by the dashed curve. (c) Temperature dependence of the hot electron relaxation time and scheme of the different relaxation channels.

Figure 2(a) shows the photoinduced 2PPE intensity, i.e., after subtraction of the equilibrium background [cf. Fig. 1(b)], in false colors as a function of pump-probe delay and energy with respect to . Photoexcitation below the Mott limit (max. excitation density ) launches fast dynamics at high energies that significantly slow down for . Note that experiments at even lower excitation densities expose qualitatively similar dynamics (not shown) 666The observed dynamics thus cannot be attributed to collective phenomena of the excited electron-hole plasma like, e.g., Fermi edge singularities Kaindl et al. (1998); Perakis and Shahbazyan (2000); Huber et al. (2001).. Exemplary cross correlation (XC) traces are depicted in Fig. 2(b) (circles), which can be fitted (solid lines) using bi-exponential decays convolved with the laser pulses’ cross correlation (dashed). The fast time constant of this empirical fit can be related to the hot electron relaxation time at the respective energy. We performed similar experiments for temperatures between 50 and 300 K, the results of which are presented in Fig. 2(c). In agreement with previous 2PPE autocorrelation measurements Tisdale et al. (2008) and theoretical work Zhukov et al. (2010), we find extremely fast electron relaxation for carriers above the bulk conduction band minimum (CBM) [, gray shaded area in Fig. 2(c)]. These fast relaxation times compared to hot electrons in metals Lisowski et al. (2004) can be explained by the reduced screening in the semiconductor. They result from scattering with optical phonons and are therefore independent of temperature in the investigated region (). Theory Zhukov et al. (2010) predicts a slowing down of the electron relaxation in the bulk when approaching the CBM [dashed line in Fig. 2(c)], because fewer efficient scattering channels require more scattering events for the same energy loss. In the present experiment, at the ZnO() surface, such a “phonon bottleneck” does not occur, as the downward surface band bending provides density of states (DOS) down to [see inset in Fig. 2(c)].

The dynamics below differ significantly from the ones above. Integration of the 2PPE intensity yields the diamond-shaped markers in Fig. 2(b). They display an initial drop of intensity, which is due to bleaching of the CAL by , followed by a rise up to a constant positive value. Notably, this delayed increase of pump-induced 2PPE intensity occurs below , i.e., below the energy which defines the highest occupied electronic state in equilibrium. Consequently, must create additional states below leading to an increased 2PPE signal. Such photoinduced creation of states could be caused by (i) small polaron formation, (ii) photoinduced changes of the surface electronic structure like surface photovoltage shifts, band gap renormalization, or heating/bleaching of the CAL, or (iii) exciton formation.

It is well known that lattice distortions stabilize the ZnO exciton and lead to its large binding energy of 60 meV 777The Coulomb attraction between electron and hole and its screening in the dielectric material determines the binding energy and localization of the exciton. In particular in semiconductors with ionic character like and ZnO, the coupling of carriers and excitons to the lattice, even on ultrafast timescales, cannot be neglected Chiodo et al. (2010); Zhukov et al. (2010); Bothschafter et al. (2013). This leads to a polaronic character of the exciton (relaxation of ions localizes the e-h pair), leading to a change of exciton binding energy due to coupling with phonons Hsu and Hsieh (2004).. The formation of separate polarons of electron and hole is thus unlikely. Also, no experimental evidence of small polaron formation was found by recent time-resolved THz spectroscopy studies of ZnO Hendry et al. (2007), excluding scenario (i). We tested the validity of (ii) and (iii) by a variation of the excitation density. While, in the case of (ii), photoinduced changes of the surface band bending (SBB) and occupational changes due to laser heating should become stronger with increasing pump fluence, (iii) the probability of exciton formation should be independent of excitation density below the Mott limit, i.e., the density at which the screening of the Coulomb interaction sets in [see inset in Fig. 3(a)]. For stronger excitation the Coulomb interaction of the electron-hole pairs is reduced by the photoexcited e-h plasma and thus the exciton formation probability diminishes.

Figure 3: (color online). 2PPE spectra for (a) different excitation densities and (b) hydrogen terminations after background subtraction and normalized to the hot electron intensity at high energies () in order to compare changes relative to the single particle excitation density. (a) Stronger excitation reduces the SX formation probability due to screening of the Coulomb interaction as illustrated in the insets. (b) The SX intensity does not decrease with increasing termination showing localization in the subsurface region as illustrated by the inset.

Figure 3(a) compares 2PPE spectra of two different excitation densities near the Mott limit of ZnO at different time delays. While the overall dynamics do not change, the intensity of the spectral signature at is reduced by the increase of the excitation density. This observation unambiguously shows that (ii) photoinduced changes of SBB and CAL density cannot be the cause of the observed peak. On the contrary, the decreased intensity matches the expectations (iii) for an exciton whose formation probability is reduced by screening of the electron-hole interaction. As all of these processes must happen in the surface region 888The SX feature is caused by the downward SBB, which extends only 1 nm into the ZnO crystal Ozawa and Mase (2011). Also, PE is only sensitive to the first few nm due to the finite mean free path of electrons in a solid., we conclude that the photoinduced intensity below results from near surface-bound excitons (SX) with a binding energy with respect to the bulk conduction band minimum of 250 meV. It should be noted that, in the SBB picture, the local binding energy relative to the local CBM is probably smaller [cf. inset of Fig. 2(c), dotted curve]. The continuous evolution of the SX signature from the hot electron distribution is a direct consequence of the off-resonant excitation leading to carriers that relax to the band edges. The formation of SX occurs at least as fast as the observed rise time of PE intensity below of 200 fs, however, these excitons are in a highly excited ”hot” state that relaxes to the ground state on longer timescales 999The 1s ground state formation of the exciton usually occurs, depending on the material, on a ps to ns timescale. Chatterjee et al. (2004); Hendry et al. (2007); Kaindl et al. (2009). Due to the limited energy resolution (45 meV) and because PES averages over the energy distribution in the SBB region, the spectral signatures of electron and exciton blend smoothly into each other 101010This is different to the SX formation at the Si(100) surface, where the exciton results from electrons and holes in the electronic surface states and not the CB and VB continuum Weinelt et al. (2004) and also not comparable to the transient excitons at the Ag(111) surface, which decay on fs timescales Cui et al. (2014).. As the electron of the SX energetically lies below the Fermi level, nonradiative decay of this excitation is strongly suppressed: Dissociation of the exciton and separation of electron and hole in defect states within the band gap usually is a prominent nonradiative decay channel for excitons. It cannot occur here, as most possible final states for the electron lie below and are therefore occupied. The decay of the SX population can thus only happen through electron-hole pair recombination going along with luminescence or Auger-type processes, which results in the high stability of this excited state 111111The SX population decay is slower than the 400 ps accessible in our experiment. Comparison with the results achieved by Tisdale et al. Tisdale et al. (2008) using a 76 MHz oscillator suggests lifetimes exceeding 10 ns..

In order to test the stability of the SX, we modified the surface electronic properties by variation of the hydrogen termination. As mentioned above, this leads to a reduction of the sample work function by hundreds of meV and concurrent formation of a CAL at the surface. Such strong modification of the surface electronic structure should severely affect the formation probability of the SX if it was localized in the vacuum in front of the surface (through the modified ionization potential leading to spectral shifts) or in the first double layer (through the screening of the Coulomb interaction by the charge density in the CAL leading to a reduction of the SX intensity with increasing termination). As shown in Fig. 3(b), none of these effects occur when increasing the exposure from 3 to 44 L (). We thus conclude that the SX is localized in the subsurface region where additional static screening of the Coulomb interaction does not occur. On the contrary, increasing the hydrogen exposure by more than a factor of 10 leads to a comparably weak increase of the SX intensity and a weak shift to higher energies. This observation could be explained by SX localized close to hydrogen binding sites where the downward SBB is strongest and a denser CAL with increasing termination that could give the SX Mahan exciton character (see e.g. Schleife et al. (2011)), i.e. the energetic stabilization of the CAL due to a localized hole in the VB.

In summary, we present a systematic investigation of the ultrafast electron and exciton dynamics at the ZnO() surface, showing the formation of a subsurface-bound exciton on fs timescales that exhibits a very large binding energy with respect to the bulk conduction band, resulting in a remarkable stability of this feature. The study thus offers a complete and novel picture of the quasiparticle relaxation at this surface. The existence of a subsurface exciton which is stable with regard to surface modifications is of high relevance for applications of ZnO, e.g., involving FRET. The SX may dominate the dipole-dipole coupling across functional interfaces due to its adjacence to the surface, and it may even persist under non-UHV conditions.

Acknowledgements.
We are grateful for stimulating discussions with T. Kampfrath, P. Rinke and O. Hofmann at the FHI Berlin. This work was partially funded by the Deutsche Forschungsgemeinschaft through Sfb 951 and the European Union through Grant No. 280879-2 CRONOS CP-FP7.

References

  • Wöll (2007) C. Wöll, Progress in Surface Science 82, 55 (2007).
  • Richters et al. (2008) J.-P. Richters, T. Voss, D. S. Kim, R. Scholz,  and M. Zacharias, Nanotechnology 19, 305202 (2008).
  • Slootsky et al. (2014) M. Slootsky, X. Liu, V. M. Menon,  and S. R. Forrest, Physical Review Letters 112, 076401 (2014).
  • Klingshirn et al. (2010) C. Klingshirn, J. Fallert, H. Zhou, J. Sartor, C. Thiele, F. Maier-Flaig, D. Schneider,  and H. Kalt, Physica Status Solidi (B) 247, 1424 (2010).
  • Della Sala et al. (2011) F. Della Sala, S. Blumstengel,  and F. Henneberger, Phys. Rev. Lett. 107, 146401 (2011).
  • Xu et al. (2013) Y. Xu, O. T. Hofmann, R. Schlesinger, S. Winkler, J. Frisch, J. Niederhausen, A. Vollmer, S. Blumstengel, F. Henneberger, N. Koch, P. Rinke,  and M. Scheffler, Phys. Rev. Lett. 111, 226802 (2013).
  • Förster (1948) T. Förster, Annalen der Physik 437, 55 (1948).
  • Egelhaaf and Oelkrug (1996) H.-J. Egelhaaf and D. Oelkrug, Journal of Crystal Growth 161, 190 (1996).
  • Fonoberov and Balandin (2004a) V. A. Fonoberov and A. A. Balandin, Phys. Rev. B 70, 195410 (2004a).
  • Kiliani et al. (2011) G. Kiliani, R. Schneider, D. Litvinov, D. Gerthsen, M. Fonin, U. Rüdiger, A. Leitenstorfer,  and R. Bratschitsch, Optics Express 19, 1641 (2011).
  • Fonoberov and Balandin (2004b) V. A. Fonoberov and A. A. Balandin, Applied Physics Letters 85, 5971 (2004b).
  • Kuehn et al. (2013) S. Kuehn, S. Friede, S. Sadofev, S. Blumstengel, F. Henneberger,  and T. Elsaesser, Applied Physics Letters 103, 191909 (2013).
  • Travnikov et al. (1990) V. Travnikov, A. Freiberg,  and S. Savikhin, Journal of Luminescence 47, 107 (1990).
  • Diebold et al. (2004) U. Diebold, L. V. Koplitz,  and O. Dulub, Applied Surface Science 237, 336 (2004).
  • (15) Laser pulses were provided by a regeneratively amplified (RegA), tuneable fs-laser (200 kHz). The frequency-doubled output of an optical parametric amplifier provided and the third (4.65 eV) and fourth harmonic (6.2 eV) of the 800 nm RegA output were used for photoemission.
  • (16) As discussed further below, the observed spectral features exhibit a neither fluence nor temperature dependence.
  • (17) Literature values for the Mott density vary between Versteegh et al. (2011); Hendry et al. (2007); Schleife et al. (2011). The calculated maximum excitation densities give an upper limit for the absorption directly at the surface.
  • (18) The glowing filament of an ion gauge was set in line of sight with the sample surface leading to an increased dissociation of .
  • Eger et al. (1976) D. Eger, A. Many,  and Y. Goldstein, Surface Science 58, 18 (1976).
  • Lüth (2010) H. Lüth, Solid Surfaces, Interfaces and Thin Films, edited by W. T. Rhodes, H. E. Stanley,  and R. Needs (Springer Berlin Heidelberg, 2010).
  • Wang et al. (2005) Y. Wang, B. Meyer, X. Yin, M. Kunat, D. Langenberg, F. Traeger, A. Birkner,  and C. Wöll, Phys. Rev. Lett. 95, 266104 (2005).
  • Ozawa and Mase (2011) K. Ozawa and K. Mase, Phys. Rev. B 83, 125406 (2011).
  • (23) The sticking coefficient is unknown and dissociation of occurs at filaments in the chamber plays a role, which inhibits quantitative comparison of the exposures to Ref. Ozawa and Mase (2011).
  • (24) The observed dynamics thus cannot be attributed to collective phenomena of the excited electron-hole plasma like, e.g., Fermi edge singularities Kaindl et al. (1998); Perakis and Shahbazyan (2000); Huber et al. (2001).
  • Tisdale et al. (2008) W. A. Tisdale, M. Muntwiler, D. J. Norris, E. S. Aydil,  and X.-Y. Zhu, The Journal of Physical Chemistry C 112, 14682 (2008).
  • Zhukov et al. (2010) V. P. Zhukov, P. M. Echenique,  and E. V. Chulkov, Phys. Rev. B 82, 094302 (2010).
  • Lisowski et al. (2004) M. Lisowski, P. Loukakos, U. Bovensiepen, J. Stähler, C. Gahl,  and M. Wolf, Applied Physics A 78, 165 (2004).
  • (28) The Coulomb attraction between electron and hole and its screening in the dielectric material determines the binding energy and localization of the exciton. In particular in semiconductors with ionic character like and ZnO, the coupling of carriers and excitons to the lattice, even on ultrafast timescales, cannot be neglected Chiodo et al. (2010); Zhukov et al. (2010); Bothschafter et al. (2013). This leads to a polaronic character of the exciton (relaxation of ions localizes the e-h pair), leading to a change of exciton binding energy due to coupling with phonons Hsu and Hsieh (2004).
  • Hendry et al. (2007) E. Hendry, M. Koeberg,  and M. Bonn, Phys. Rev. B 76, 045214 (2007).
  • (30) The SX feature is caused by the downward SBB, which extends only 1 nm into the ZnO crystal Ozawa and Mase (2011). Also, PE is only sensitive to the first few nm due to the finite mean free path of electrons in a solid.
  • (31) The 1s ground state formation of the exciton usually occurs, depending on the material, on a ps to ns timescale. Chatterjee et al. (2004); Hendry et al. (2007); Kaindl et al. (2009).
  • (32) This is different to the SX formation at the Si(100) surface, where the exciton results from electrons and holes in the electronic surface states and not the CB and VB continuum Weinelt et al. (2004) and also not comparable to the transient excitons at the Ag(111) surface, which decay on fs timescales Cui et al. (2014).
  • (33) The SX population decay is slower than the 400 ps accessible in our experiment. Comparison with the results achieved by Tisdale et al. Tisdale et al. (2008) using a 76 MHz oscillator suggests lifetimes exceeding 10 ns.
  • Schleife et al. (2011) A. Schleife, C. Rödl, F. Fuchs, K. Hannewald,  and F. Bechstedt, Phys. Rev. Lett. 107, 236405 (2011).
  • Versteegh et al. (2011) M. A. M. Versteegh, T. Kuis, H. T. C. Stoof,  and J. I. Dijkhuis, Phys. Rev. B 84, 035207 (2011).
  • Kaindl et al. (1998) R. A. Kaindl, S. Lutgen, M. Woerner, T. Elsaesser, B. Nottelmann, V. M. Axt, T. Kuhn, A. Hase,  and H. Künzel, Phys. Rev. Lett. 80, 3575 (1998).
  • Perakis and Shahbazyan (2000) I. Perakis and T. Shahbazyan, Surface Science Reports 40, 1 (2000).
  • Huber et al. (2001) R. Huber, F. Tauser, A. Brodschelm, M. Bichler, G. Abstreiter,  and A. Leitenstorfer, Nature 414, 286 (2001).
  • Chiodo et al. (2010) L. Chiodo, J. M. Garcìa-Lastra, A. Iacomino, S. Ossicini, J. Zhao, H. Petek,  and A. Rubio, Phys. Rev. B 82, 045207 (2010).
  • Bothschafter et al. (2013) E. Bothschafter, A. Paarmann, E. Zijlstra, N. Karpowicz, M. Garcia, R. Kienberger,  and R. Ernstorfer, Physical Review Letters 110, 067402 (2013).
  • Hsu and Hsieh (2004) H.-C. Hsu and W.-F. Hsieh, Solid State Communications 131, 371 (2004).
  • Chatterjee et al. (2004) S. Chatterjee, C. Ell, S. Mosor, G. Khitrova, H. M. Gibbs, W. Hoyer, M. Kira, S. W. Koch, J. P. Prineas,  and H. Stolz, Phys. Rev. Lett. 92, 067402 (2004).
  • Kaindl et al. (2009) R. A. Kaindl, D. Hägele, M. A. Carnahan,  and D. S. Chemla, Phys. Rev. B 79, 045320 (2009).
  • Weinelt et al. (2004) M. Weinelt, M. Kutschera, T. Fauster,  and M. Rohlfing, Phys. Rev. Lett. 92, 126801 (2004).
  • Cui et al. (2014) X. Cui, C. Wang, A. Argondizzo, S. Garrett-Roe, B. Gumhalter,  and H. Petek, Nature Physics 10, 505 (2014).
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
""
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
   
Add comment
Cancel
Loading ...
184536
This is a comment super asjknd jkasnjk adsnkj
Upvote
Downvote
""
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters
Submit
Cancel

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test
Test description