Letter pubs.acs.org/NanoLett
Carrier Cooling in Colloidal Quantum Wells Matthew Pelton,*,† Sandrine Ithurria,‡,§ Richard D. Schaller,† Dmitriy S. Dolzhnikov,‡ and Dmitri V. Talapin†,‡ †
Center for Nanoscale Materials, Argonne National Laboratory, Argonne, Illinois 60439, United States Department of Chemistry and James Franck Institute, University of Chicago, Chicago, Illinois 60637, United States
‡
S Supporting Information *
ABSTRACT: It has recently become possible to chemically synthesize atomically flat semiconductor nanoplatelets with monolayer-precision control over the platelet thickness. It has been suggested that these platelets are quantum wells; that is, carriers in these platelets are confined in one dimension but are free to move in the other two dimensions. Here, we report time-resolved photoluminescence and transient-absorption measurements of carrier relaxation that confirm the quantum-well nature of these nanomaterials. Excitation of the nanoplatelets by an intense laser pulse results in the formation of a high-temperature carrier population that cools back down to ambient temperature on the time scale of several picoseconds. The rapid carrier cooling indicates that the platelets are well-suited for optoelectronic applications such as lasers and modulators. KEYWORDS: Quantum wells, semiconductor nanocrystals, carrier relaxation
Q
that are monodisperse on the atomic scale, for thicknesses from 4 to 7 monolayers (ML).6,9 Unlike epitaxial QWs, these colloidal systems appear to be free of monolayer thickness fluctuations. Moreover, the faces of these platelets are capped with organic ligands, and the platelets are typically surrounded by solvents or by air. This means that carrier confinement and exciton binding energies are much stronger in colloidal nanoplatelets than in epitaxial QWs, which are buried in a crystal of a second semiconductor material. The stronger confinement and binding energies, in turn, are likely to result in significantly different carrier dynamics in colloidal nanoplatelets as compared to epitaxial quantum wells. For example, experiments seem to indicate that carrier recombination rates are much slower in colloidal systems than in epitaxial systems.9,10 Another property that may be different in colloidal systems is the relaxation of high-energy carriers. This relaxation rate is of particular interest, since it plays a central role in potential optoelectronic applications of QWs: fast relaxation within the conduction and valence bands is generally desirable for lasers, including quantum-cascade lasers, and for modulators, whereas slow relaxation is helpful for quantum-well intersubband detectors and for hot-carrier photovoltaic devices. Relaxation processes are strongly dependent on the dimensionality of quantum-confined semiconductor structures. In colloidal CdSe QDs, hole states are closely spaced, and hole relaxation can
uantum wells (QWs) are thin semiconductor layers that confine electrons and holes in one dimension on a length scale comparable to their de Broglie wavelengths.1 This confinement results in tunable optical bandgaps and strong excitonic absorption and emission features even at room temperature. QWs have several advantages as gain media in semiconductor lasers, particularly, tunable emission wavelengths and low threshold currents.2 The strong, tunable optical absorption offered by QWs also makes them attractive for use in optical modulators, photodetectors, and solar cells. So far, though, QWs have been produced using expensive epitaxial crystal-growth techniques, primarily molecular beam epitaxy and metal−organic vapor-phase epitaxy. This has motivated research into the use of colloidal semiconductor nanocrystals in lasers3 and photovoltaics,4 among other applications, since these nanocrystals can be synthesized chemically in large volumes and at low cost. In these quantum-dot (QD) structures, however, carriers are confined in all three dimensions, and only a small number of exciton states exist at the optical bandgap energy. Since QDs cannot be packed closer together than their diameters, the maximum optical absorption or gain in even a dense layer of QDs is limited. Recently, colloidal synthesis methods have been developed that result in thin, flat semiconductor nanoparticles, with thicknesses of a few monolayers (ML) and lateral dimensions from 10 nm to several micrometers. Such structures have been referred to as nanoribbons,5 nanoplatelets,6 nanosheets,7 and quantum disks.8 All of these structures show narrow absorption and emission peaks characteristic of QW excitons. CdSe nanoplatelets, in particular, can be synthesized with thicknesses © XXXX American Chemical Society
Received: August 10, 2012 Revised: November 2, 2012
A
dx.doi.org/10.1021/nl302986y | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
occur efficiently through phonon emission. Electron states have energy separations that are too large to allow for efficient phonon-assisted relaxation, so electron relaxation occurs through Auger scattering with holes, through intermediate trap states, and by coupling to vibrational modes in ligands.11−14 If multiple excitons are excited simultaneously, all but one will rapidly recombine through a nonradiative Auger processes.15 In extended nanocrystals where carriers are free to move in one dimension, known as quantum rods (QRs), a continuum of energies is available. High-energy carriers in the QRs rapidly reach internal thermal equilibrium, with a welldefined temperature, and then cool by emitting phonons. Efficient Auger recombination still occurs, though, and can compete with phonon-assisted cooling.16 In epitaxial quantum wells, electron−phonon coupling is efficient, and relaxation rates for high-temperature carriers are similar to those in bulk semiconductors.17 In this Letter, we show that carrier relaxation in colloidal CdSe nanoplatelets is characteristic of a QW system, where carriers are confined in only one dimension, rather than a QD system, where carriers are confined in all three dimensions. As well as providing important information needed for applications, these results provide critical evidence that these nanoplatelets are indeed colloidal QWs. Other evidence has been provided for one-dimensional carrier confinement in these systems, particularly agreement between measured and calculated exciton energies and an increase in recombination rates, accompanied by an increase in emission intensity, as temperature is decreased.9 Different groups have obtained quantitatively different results in both cases, though;10 in addition, photoluminescence (PL) from single nanoplatelets shows blinking behavior resembling that of single nanocrystal QDs.18 Our results are thus an important, complementary demonstration of phenomena in colloidal nanoplatelets that are qualitatively different from those in colloidal QDs. The CdSe nanoplatelets are synthesized as previously reported;6 details are given in the Supporting Information. Briefly, two types of Cd carboxylate precursors, one with a long chain (for example, oleate or myristate) and one with a short chain (acetate), are heated in the presence of a selenium precursor (selenium powder or trioctylphosphine selenide). This results in flat platelets with zinc-blende crystal structure, lateral dimensions from 10 to a few hundred nanometers and precisely controlled thicknesses from 5 to 7 ML. Depending on the precursors and the temperature, the synthesis may also result in a byproduct of QDs, which can be separated from the nanoplatelets by selective precipitation. Figure 1 shows the linear absorption spectra of three samples with different thicknesses. Each spectrum is dominated by a pair of absorption peaks that occur at the energies expected for heavy-hole and light-hole excitons in quantum wells with infinite potential barriers.9,10 Absorption at higher energies can be attributed to the creation of unbound electron−hole pairs or to excitons in higher QW sub-bands. Luminescence from all of the samples occurs exclusively from the heavy-hole exciton state. For excitation at a photon energy of 3.0 eV, the PL quantum yields of the 5-ML, 6-ML, and 7-ML samples are 3%, 68%, and 15%, respectively. To monitor the relaxation of energetic carriers in these platelets, we measured the PL emitted when the platelets are excited with 35 fs laser pulses (which we call pump pulses) with a photon energy of 3.0 eV, focused within the sample to a spot with a diameter of approximately 225 μm. The pulses are
Figure 1. Absorption spectra of CdSe nanoplatelet samples with thicknesses from 5 monolayers (ML) to 7 ML. Background due to scattering has been subtracted from each of the spectra, and the spectra have been normalized for ease of comparison.
derived from an amplified Ti:Sapphire laser system (SpectraPhysics Tsunami and SpitFire Pro), operating at a repetition rate of 2 kHz, which pumps an optical parametric amplifier (Spectra-Physics TOPAS-C). The PL is resolved as a function of wavelength using a grating spectrometer (Acton SP2150) and as a function of time after excitation using a streak camera (Hamamatsu C5680) with photon-counting detection. All measurements are made at room temperature on solutions of nanoplatelets in hexane; the solutions are stirred during the measurements to avoid potential artifacts due to heating of the solution or charging of the platelets. Representative results are shown in Figure 2 for 6-ML-thick samples; similar results were obtained for 5-ML-thick and 7ML-thick samples. For low pump powers, emission occurs entirely from the heavy-hole exciton peak, and the PL spectrum does not change over the measured time range. This indicates that all of the carriers that are recombining and emitting light have already relaxed to their lowest-energy states within the approximately 5 ps time resolution of the measurement. By contrast, for high pump powers at short times after excitation, there is an exponential tail on the high-energy side of the PL spectrum. Such a tail has previously been observed in timeresolved PL measurements of epitaxial QWs17 and colloidal QRs16 and corresponds to recombination of higher-energy carriers. The photoexcited carriers reach internal thermal equilibrium much faster than the time resolution of the experiment, so that the exponential tail reflects a thermal distribution of carrier energies. As time passes, the carriers cool back down to ambient temperature, which is reflected in the increasing slope of the exponential tail. Time-resolved PL measurements on colloidal QDs give qualitatively different results.19 The short-time spectra show a clear shoulder on the high-energy side of the excitonic peak, rather than an exponential tail. The shoulder corresponds to emission from charged biexcitons, rather than a thermal population of carriers; indeed, it is impossible to obtain a thermal distribution of carrier energies in QDs, since only a discrete set of carrier energies are allowed. Similarly, the decay of the PL shoulder for QDs is due to Auger recombination of biexcitons, rather than carrier cooling. B
dx.doi.org/10.1021/nl302986y | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
The carrier temperature at a particular time after excitation can be estimated by fitting the high-energy luminescence tail. The luminescence intensity depends on the interband transition matrix element, the electron−hole joint density of states, and the energy distribution of the carriers. For carrier energies close to the band edges in a quantum well, the transition matrix elements and the joint density of states are approximately independent of energy.1 This means that the luminescence intensity is simply proportional to the carrier energy distribution, which can be approximated as the tail of a Boltzmann distribution:16 ⎛ ℏω ⎞ I(ℏω) ∝ exp⎜ − ⎟ ⎝ kBT ⎠
where ω is the photon frequency, T is the effective carrier temperature, and kB is Boltzmann’s constant. Representative time-dependent carrier temperatures are shown in Figure 3; similar results were obtained for all of the
Figure 2. Photoluminescence spectra for 6-monolayer-thick CdSe nanoplatelets at various times following excitation by an ultrafast laser pulse with a photon energy of 3.0 eV. Results are shown for pump pulse energies of (a) 5 nJ and (b) 1.7 μJ. These energies correspond, respectively, to the creation on average of 1.4 and 470 excitons in each nanoplatelet by each pump pulse, resulting in initial exciton densities of 5.6 × 1012 cm−2 and 1.9 × 1014 cm−2 (See the Supporting Information for an explanation of how these quantities were estimated.). Spectra are normalized by their maximum value, for ease of comparison, and are plotted on a semilogarithmic scale.
Figure 3. Carrier temperature in CdSe nanoplatelets, determined by fitting PL spectra, as a function of time following excitation by ultrafast laser pulses. Results are shown for platelet samples with thicknesses of 6 or 7 monolayers (ML) and for different pump-pulse energies. For the 6-ML sample, the pump pulse energies of 500 nJ and 1.6 μJ correspond, respectively, to the creation on average of 140 and 450 excitons in each nanoplatelet by each pump pulse, resulting in initial exciton densities of 5.5 × 1013 cm−2 and 1.8 × 1014 cm−2. For the 7ML sample, the pump pulse energy of 500 nJ corresponds to the creation of 620 excitons in each nanoplatelet by each pump pulse, resulting in an initial exciton density of 2.5 × 1014 cm−2.
We note that there is also emission on the low-energy side of the nanoplatelet PL spectrum. This emission is weak at low pump powers (visible as a small peak at approximately 2.25 eV in Figure 2a) but becomes more significant at high pump powers. Unlike the high-energy tail, the low-energy feature varies from sample to sample and does not decay on picosecond time scales. The low-energy absorption and emission may be due to nanocrystals that are produced as a byproduct of the platelet synthesis and that are not completely removed by size-selective precipitation.9 If these small nanocrystals have higher saturation intensities than the platelets, their emission will increase relative to the platelet emission as the excitation intensity is increased. The low-energy emission may be responsible for the apparent shift of the PL peak to lower energies under high excitation intensity, although other effects, such as bandgap renormalization, may also be playing a role.20
samples studied. The decay is nonexponential, reflecting complex relaxation dynamics. The carrier cooling occurs on time scales less than 100 ps, much faster than carrier recombination, which occurs on time scales of several nanoseconds at room temperature.9 Carrier cooling on similar time scales and with similar dynamics has been observed in bulk CdSe.21,22 In this case, it was proposed that high-energy carriers relax initially by emission of LO phonons; these phonons can be reabsorbed by the carriers, so that the LO-phonon population and the carrier population rapidly reach mutual thermal equilibrium. The two populations then cool together through interactions with acoustic phonons. This relaxation mechanism is also expected to apply for carriers in QWs and is C
dx.doi.org/10.1021/nl302986y | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
only weakly dependent on excitation intensity, consistent with our observations on nanoplatelets. This is in contrast to relaxation in QRs, for which even relatively modest excitation intensities produce high carrier densities, and multicarrier processes lead to modified relaxation rates.16 Our time-resolved PL measurements thus provide strong evidence that carriers in nanoplatelets behave like carriers in QWs and not like carriers in QRs or QDs. Further support comes from transient-absorption (TA) measurements. In these measurements, the samples are excited in the same way as in time-resolved PL measurements. Rather than measuring light emitted by the samples, though, we measure the absorption of a broadband probe pulse that passes through the sample after the pump pulse. The result of the measurement is the difference in the probe absorption in the presence and in the absence of the pump. The background signal due to scattering of pump light is subtracted from the data, and the time-dependent spectra are corrected to account for chirp in the probe pulse. Representative results are shown in Figure 4; similar results were obtained for all of the samples. At low pump powers,
negative transient-absorption peaks, or bleach signals, are seen at the heavy-hole and light-hole energies. In this case, the bleach features are due primarily to state filling (also known as phase-space filling): carriers occupy the states out of which the excitons are formed, preventing absorption of light at the exciton energies.20 An additional contribution to the transient signal comes from broadening of the excitonic transitions through collisions with the photoexcited carriers, or screening; this is reflected in the positive transient-absorption signal on the sides of the peaks. The transient spectra decay uniformly as carriers are lost through recombination or trapping. For high pump powers, the same features are present, but there is an additional, broad bleach at high energies that decays more quickly than the other bleach features. This high-energy bleach has the same origin as the high-energy tail on the time-resolved PL spectrum: occupation of higher-energy states within the bands by a thermal distribution of excited carriers.20 We note that there is also a low-energy tail for high pump powers, which likely has the same origin as the low-energy tail on the PL spectrum. The high-energy bleach due to high-temperature carriers overlaps the light-hole-exciton bleach. It is not straightforward to separate the two features and fit the TA spectrum to determine carrier temperature, as for the PL spectrum. On the other hand, we can isolate the effects of hot carriers by examining the TA signal at specific energies where the contribution from the excitonic peaks is minimal. For the data shown in Figure 4, this occurs around a probe-photon energy of 2.55 eV. The corresponding TA time trace is shown in Figure 5, superimposed with the carrier temperature
Figure 5. Transient-absorption kinetics (blue line) for 6-monolayerthick CdSe nanoplatelets, for a pump-photon energy of 3.0 eV and a probe-photon energy of 2.55 eV. The y axis is inverted so that the data can be superimposed with carrier temperatures (black points) obtained from photoluminescence data under the same excitation conditions (see Figure 3).
obtained from PL data under the same excitation conditions. The two time traces are nearly identical. This indicates that the time-dependent processes responsible for the two signals are the same; in particular, there is no indication in the transientabsorption signal of any additional processes, such as carrier trapping, occurring on time scales comparable to the carrier cooling. At lower probe-photon energies, in the near-infrared, the TA spectra show induced absorption (positive transient-absorption
Figure 4. Transient-absorption spectra from a sample of 6-monolayerthick CdSe nanoplatelets for various pump−probe delays. The pumpphoton energy was 3.0 eV. Results are shown for pump-pulse energies of (a) 20 nJ and (b) 200 nJ. These energies correspond, respectively, to the creation on average of 3.6 and 36 excitons in each nanoplatelet by each pump pulse, resulting in initial exciton densities of 1.6 × 1012 cm−2 and 1.6 × 1013 cm−2. D
dx.doi.org/10.1021/nl302986y | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
peaks are due to transitions between the heavy-hole exciton and a higher-lying state whose energy is nearly independent of nanoplatelet thickness. The decay of the induced-absorption peaks is thus primarily due to recombination of carriers, as for the bleach signals at the exciton absorption energies. In conclusion, the results of time-resolved PL and transientabsorption measurements on colloidal CdSe nanoplatelets are consistent with one-dimensional carrier confinement in the plateletsthat is, the nanoplatelets behave as colloidal quantum wells, not quantum dots or quantum rods. This provides an important confirmation of previous evidence of QW behavior in these materials.9,10 In addition, our measurements provide quantitative information about carrier relaxation in these materials, information that is critical for optoelectronic applications. The carrier relaxation appears to be similar to relaxation in bulk CdSe,23,24 indicating that carrier−phonon coupling is not strongly modified in these nanomaterials. Other measurements, however, have indicated reduced electron− phonon coupling in CdSe nanomaterials,10,25 and quantitative modeling of the time-resolved measurements will be required to reconcile the results. It will also be important to determine the universality of the current results by extending the timeresolved optical measurements to nanoplatelets made out of other materials and to nanoplatelet heterostructures.
peaks) rather than bleach, as illustrated in Figure 6. These peaks occur at energies well below the CdSe bandgap, where
■
ASSOCIATED CONTENT
* Supporting Information S
Details of nanoplatelet synthesis, determination of exciton density, and measurement of quantum yield. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]; Tel: 630-252-4598. Present Address §
Laboratoire de Physique et d’Etude des Materiaux, ESPCI, 10 rue Vauquelin 75231 Paris, France. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was performed, in part, at the Center for Nanoscale Materials, a U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences User Facility under Contract No. DE-AC02-06CH11357. R.D.S. and D.V.T. acknowledge support by the University of Chicago and the Department of Energy Contract No. DE.AC02-06CH11357 awarded to UChicago Argonne, LLC, operator of Argonne National Laboratory. D.V.T. also thanks the David and Lucile Packard Foundation and Keck Foundation for their generous support. This work used facilities supported by NSF MRSEC Program under Award Number DMR-0213745.
Figure 6. Transient-absorption spectra from a sample of 6-monolayerthick CdSe nanoplatelets for a pump-photon energy of 3.0 eV and a pump-pulse energy of 250 nJ. This energy corresponds to the creation on average of 45 excitons in each nanoplatelet by each pump pulse, resulting in an initial exciton density of 2.0 × 1013 cm−2. (a) Transient spectra for nanoplatelets with a thickness of 6 monolayers (ML), for various time delays. (b) Top panel: transient spectra for nanoplatelets with different thicknesses, averaged over all time delays from 1 to 2500 ps. Lower panel: linear absorption spectra for the same samples. The x-axis ranges are different for the two panels.
■
the samples show no linear absorption. The representative data in Figure 6a, for 6-ML-thick nanoplatelets, show that the induced-absorption peaks decay nearly uniformly. We can therefore average over all time delays to more clearly see the peak position and peak shape, as shown in Figure 6b. The linear absorption spectra are also reproduced in this figure, showing that the shift from sample to sample in the induced-absorption peak approximately tracks the shift in the heavy-hole exciton absorption peak. This suggests that the induced-absorption
REFERENCES
(1) Chuang, S. L. Physics of Optoelectronic Devices; John Wiley & Sons: New York, 1995. (2) Yariv, A. Quantum Electronics, 3rd ed.; John Wiley & Sons: New York, 1989. (3) Klimov, V. I.; Mikhailovsky, A. A.; Xu, S.; Malko, A.; Hollingsworth, J. A.; Leatherdale, C. A.; Eisler, H.-J.; Bawendi, M. G. Science 2000, 290, 314−317. (4) Sargent, E. H. Nat. Photon. 2012, 6, 133−135. E
dx.doi.org/10.1021/nl302986y | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
(5) Joo, J.; Son, J. S.; Kwon, S. G.; Yu, J. H.; Hyeon, T. J. Am. Chem. Soc. 2006, 128, 5632−5633. (6) Ithurria, S.; Dubertret, B. J. Am. Chem. Soc. 2008, 130, 16504− 16505. (7) Son, J. S.; Wen, X.-D.; Joo, J.; Chae, J.; Baek, S.-I.; Park, K.; Kim, J. H.; An, K.; Yu, J. H.; Kwon, S. G.; Choi, S.-H.; Wang, Z.; Kim, Y.W.; Kuk, Y.; Hoffman, R.; Hyeon, T. Angew. Chem., Int. Ed. 2009, 48, 6861−6864. (8) Li, Z.; Peng, X. J. Am. Chem. Soc. 2011, 133, 6578−6586. (9) Ithurria, S.; Tessier, M. D.; Mahler, B.; Lobo, R. P. S. M.; Dubertret, B.; Efros, Al. L. Nat. Mater. 2011, 10, 936−941. (10) Achtstein, A. W.; Schliwa, A.; Prudnikau, A.; Hardzei, M.; Artemyev, M. V.; Thomson, C.; Woggon, U. Nano Lett. 2012, 12, 3151−3157. (11) Efros, Al. L.; Kharchenko, V. A.; Rosen, M. Solid-State Commun. 1995, 93, 281−284. (12) Klimov, V. I.; McBranch, D. W. Phys. Rev. Lett. 1998, 80, 4028− 4031. (13) Guyot-Sionnest, P.; Wehrenberg, B.; Yu, D. J. Chem. Phys. 2005, 123, 074709. (14) Pandey, A.; Guyot-Sionnest, P. Science 2008, 322, 929−932. (15) Klimov, V. I.; Mikhailovsky, A. A.; McBranch, D. W.; Leatherdale, C. A.; Bawendi, M. G. Science 2000, 287, 1011−1013. (16) Achermann, M.; Bartko, A. P.; Hollingsworth, J. A.; Klimov, V. I. Nat. Phys. 2006, 2, 557−561. (17) Ryan, J. F.; Taylor, R. A.; Turberfield, A. J.; Maciel, A.; Worlock, J. M.; Gossard, A. C.; Wiegmann, W. Phys. Rev. Lett. 1984, 53, 1841− 1844. (18) Tessier, M. D.; Javaux, C.; Maksimovic, I.; Loriette, V.; Dubertret, B. ACS Nano 2012, 6, 6751−6758. (19) Achermann, M.; Hollingsworth, J. A.; Klimov, V. I. Phys. Rev. B 2003, 68, 245302. (20) Schmitt-Rink, S.; Chemla, D. S.; Miller, D. A. B. Adv. Phys. 1989, 38, 89−188. (21) Junnarkar, M. R.; Alfano, R. R. Phys. Rev. B 1986, 34, 7045− 7062. (22) Vengurlekar, A. S.; Prabhu, S. S.; Roy, S. K.; Shah, J. Phys. Rev. B 1994, 50, 15461−15464. (23) Junnarkar, M. R.; Alfano, R. R. Phys. Rev. B 1986, 34, 7045− 7062. (24) Vengurlekar, A. S.; Prabhu, S. S.; Roy, S. K.; Shah, J. Phys. Rev. B 1994, 50, 15461−15464. (25) Kelley, A. M. J. Phys. Chem. Lett. 2010, 1, 1296−1300.
F
dx.doi.org/10.1021/nl302986y | Nano Lett. XXXX, XXX, XXX−XXX
