Among the different photonic applications envisioned using semiconductor quantum dots (Qdots), Qdot lasers are one of the most investigated. In contrast to bulk materials or quantum wells (Qwells), the delta-like density of electronic states predicts a low, temperature-independent lasing threshold,1 providing enhanced device performance compared to other gain media, especially at elevated temperatures. Amplified spontaneous emission (ASE) and lasing have already been observed both in epitaxial2, 3 and colloidal4 Qdots. The molecular beam epitaxy (MBE)-grown Qdots, mostly based on III-V materials, have recently reached a level mature enough to move toward commercialization. Nevertheless, colloidal Qdots offer an interesting low cost alternative, being synthesized by wet chemistry approaches at low temperatures (25–350°C) and standard pressure. However, here progress toward efficient laser devices has been hampered by material quality and enhanced Auger recombination due to the small Qdot size (2–5 nm).
Only recently, the latter bottleneck was overcome by the use of quasi-type II CdSe/CdS giant-shell Qdots5 or anisotropic seeded quantum rods (Qdot-in-rods).6 Indeed, here the Auger recombination time could be extended from 10–50 ps to over 10 ns, resulting for instance in biexciton photoluminescence (PL) with a near-unity quantum efficiency (QE).7 Additionally, due to the reduced contribution of Auger recombination, a low ASE threshold5 in combination with a prolonged gain lifetime and a broad gain spectrum has been observed.6 However, it should be noted that these quasi-type II heterostructures may also intrinsically be of better quality compared to previously investigated colloidal Qdot-based gain media: as the lattice mismatch between the CdS shell and the CdSe core is reduced to 4% (much smaller than for instance the 12% mismatch in CdSe/ZnS Qdots or 8.5% in InP/ZnS Qdots), the formation of interface defects, which may reduce the QE and prevent ASE, will be suppressed. In this respect, we recently reported that the CdSe/CdS Qdot-in-rod PL is nearly mono-expontential from 80 K up to room temperature,8 which confirms that our samples do not suffer from significant carrier trapping once the exciton has reached the band-edge state. CdSe/CdS Qdots offer additional advantages unique to this materials system. Preferential localization of the hole in the CdSe core and electron delocalization over both core and shell introduces a strong Coulomb repulsion in the biexciton, blue shifting its absorption with respect to the single exciton.9 Consequently, gain may occur near the single-exciton regime, further suppressing Auger recombination.10
In this letter, we demonstrate an important prerequisite toward the realization of solution processed efficient laser devices: a nearly temperature-independent ASE in a close-packed thin film of colloidal Qdots. So far, only Kazes et al.11 have reported on such behavior in colloidal nanocrystals. Using CdSe/ZnS Qdot-in-rods, they observed that a temperature-independent threshold could only be maintained below 100 K, as the nanosecond pulses employed lead to strong, non-equilibrium Auger heating and concomitant suppression of the gain. Here, we observed efficient ASE even at room temperature, with thresholds Ith = 0.15–1.5 mJ·cm−2 and 0.4–2 mJ·cm−2 for core and shell emission, respectively, up to an order of magnitude lower than previously studied CdSe colloidal Qdots.12, 13 ASE threshold measurements as a function of pump wavelength highlight the importance of employing long CdS Qrod arms, as they enhance the absorption cross section and significantly reduce the gain threshold. Increasing the CdS Qrod length triggers ASE from shell states, which allows probing the carrier dynamics in view of the ASE core and shell recombination time. Finally, power- and temperature-dependent measurements yield the characteristic temperature coefficient for the ASE threshold. By determining a T0 of at least 350 K, we demonstrate that colloidal CdSe/CdS Qdot-in-rod gain materials could enable temperature-insensitive lasing, pivotal for miniaturized and integrated laser sources. This peculiar feature, not present in Qwells and bulk material, triggered wide-spread research on MBE-grown Qdots and has now been extended to colloidal systems.
Anisotropic colloidal CdSe/CdS hetero-nanocrystals offer size-tunable ASE over a wide range of wavelengths. The growth of the CdS shell provides surface passivation of the core and reduces the amount of surface defects, resulting in quasi type-II Qdots with high photoluminescence quantum efficiency.14 Additionally, it allows further extending the ASE spectral range as ASE in these systems can arise either from the core or the shell.15 Indeed, Table 1 and Figure 1(b) show that, by careful selection of the core and shell dimensions, the ASE can be tuned from a wavelength λ of 475 nm to 610 nm. Furthermore, by precise engineering of all dimensions, some samples display ASE from both core and shell states simultaneously, offering prospects for dual-emission applications.2, 16 In line with the PL data reported by Sitt et al.,9 the ASE from the CdSe core is blue shifted with respect to the PL peak, and this shift increases with decreasing size (Table 1). The blue-shift is expected to originate from Coulomb repulsion in the multi-exciton regime, which suggests that the ASE still occurs above the single exciton gain regime.
|Core diameter [nm]||Rod diameter [nm]||Rod length [nm]||PL core [nm]||ASE core [nm]||ASE shell [nm]|
To quantify the ASE threshold Ith, room-temperature PL/ASE spectra were fitted to a sum of Gaussian peaks, from which the PL and ASE amplitude Iout was extracted (Figure 1(c)). The data at high I0 were fitted by the following equation:
This yields the ASE threshold Ith, which typically varies from 0.15 to 1.5 mJ·cm−2 for the core ASE and from 0.4 to 2 mJ·cm−2 for the shell ASE. Values are comparable to organic17 and other efficient Qdot gain materials13, 15, 18, 19 (typically achieving 0.01–1 mJ·cm−2), demonstrating their potential for highly efficient, low-cost light-emitting devices with an enhanced photo-stability compared to organic materials, yet no need for high-vacuum fabrication techniques as used in the case of MBE-grown Qdot devices.
The key to obtain the low threshold here lies with the growth of a long CdS Qrod, which strongly enhances the short wavelength absorption cross section.19 Figure 1(d) shows the dependence of Ith as a function of the pump wavelength (sample S03). A strong reduction of Ith for off-resonant pumping is observed. In particular, we achieved an 8-fold decrease of Ith in sample S03, and a 3-fold decrease in sample S06 when pumping above the CdS absorption onset. The reduction, however, is not proportional to the increase in absorption. Comparing it to the ratio of the band-edge shell to band-edge core absorption (AR), we observe that the reduction is less pronounced for sample S03 (CdS Qrod length 52 nm, AR 65:1) than for sample S06 (CdS Qrod length 23 nm, AR 17:1). We can conclude that, by pumping well above the CdS band-edge, hot carrier trapping in the shell (at a rate τtr) may still limit the number of excitons reaching the lowest energy state in the CdSe core. Further support for this notion stems from PL excitation spectroscopy in CdSe/CdS Qdot-in-rods, showing a reduced PL QE when pumping the material above the CdS absorption edge,20 and from transient absorption measurements on CdSe Qdots, in which fast (ps-timescale) hot exciton trapping has also been observed.21 For such bare nanocrystals it lead to strongly suppressed gain when pumping them at 400 nm. Surface passivation using a ZnS shell,21 or increasing the nanocrystal volume by moving to single composition Qrods13 lead to reduced ASE thresholds. Here, using Qdot-in-rod heteronanocrystals we are able to further improve, decoupling the enhanced absorption cross section above the CdS band-gap from the emission tuning by the CdSe core size, thus adding a distinct advantage over single-composition Qrods. Further optimization of the Qdot-in-rods toward enhanced shell passivation may then lead to reduced shell trapping rates.
Table 1 also reveals a second limit for the Qrod length. Increasing it first leads to dual emission and eventually to shell-only ASE. This transition from core to shell ASE cannot merely be due to the increased carrier trapping rate in longer Qrods, as shell transitions should have an increased Ith as well. However, it can be rationalized by considering that, next to trapping, the shell-to-core exciton transfer time plays an important role in the ASE decay dynamics. In the regime of ASE, balancing the exciton relaxation rates is key to tuning the ASE between different emitting states,16 in contrast to spontaneous dual-color emission, where emission from excited states requires phase-space filling.22 To gain further insight, we extracted the PL (τPLc, τPLsh) and ASE (τASEc, τASEsh) lifetimes of our Qdot-in-rods using spectrally-resolved streak camera measurements (∼2 ps time resolution), and spectrally integrated fluorescence decay traces (time-correlated single photon counting). Figure 2 shows typical streak camera images (sample S03, dual emission), measured at pump intensities I0 just below and above the ASE threshold. In Figure 2(a) a weak PL signal from the shell can be discerned, with a τPLsh = 6 ps. The typical lifetime for radiative recombination in pure CdS Qrods is about 20 ns,23 hence the picosecond lifetime observed here already indicates that the decay is dominated by non-radiative relaxation into the CdSe core. When increasing I0 (Figure 2(b)), a sharp red-shifted ASE peak appears, with a detector response-time limited τASEsh = 2 ps. For the core emission, the single exponential PL decay at low intensities (with τPLc = 20 ns) becomes bi-exponential with the addition of a 72 ps component at higher I0 (Figure 2(c)). This component is blue shifted and has a linewidth similar to the main PL peak, hence we can exclude ASE and assign it to emission from a biexciton (or possibly multi-exciton) state. Being still below the ASE threshold, the image confirms our previous assumption that ASE occurs in the multi-exciton regime in CdSe/CdS Qdot-in-rods. Further increasing I0 allows to observe a sharp ASE peak (Figure 2(d)), here with τASEc = 3 ps. Streak camera images collected on sample S06 (core-only emission, τPLc = 12 ns) at low and high I0 revealed a similar behavior, where we determined τASEc = 4 ps. The relevant carrier dynamics are summarized in Figure 2(e). Our time-resolved PL data reveal that the lifetime of core and shell ASE are comparable to the picosecond shell-to-core exciton transfer time τtf determined using transient absorption (TA) measurements.24 Taking into account that TA data have already shown that longer CdS arms lead to a reduced transfer rate to the core,25 our measurements now correlate this with a suppression of the core ASE in combination with the enhancement of the shell ASE.
The wavelength-dependent and time-resolved measurements substantiate how careful engineering of the rod length allows to tailor enhanced absorption and increasing transfer time in order to optimize the ASE properties, for instance toward low threshold, core-emitting samples. Such a sample (S06), together with sample S03 which shows dual emission, was used to determine Ith as a function of temperature. Figure 3(a) shows Iout as a function of I0 for S06, demonstrating that efficient ASE is observed up to 325 K. Ith is again determined via a fit to Equation (1). From the resulting Figure 3(b) it is evident that, despite the large temperature range going from 5 K to 325 K, the threshold remains largely constant for both samples.
The change in optical gain is typically governed by the following exponential dependence of Ith on temperature:1
T0 is the characteristic temperature; a high T0 implies a nearly-temperature independent ASE threshold. Bulk or Qwell-based lasers have a typical T0 around 100 K,1 its value mostly being limited by the increased population of higher energy states at elevated temperature, with a concurring depopulation of the band-edge state. In contrast, fitting Equation (2) to our Qdot-in-rod data (Figure 3(b)) results in a T0 equal to 350 ± 75 K and 950 ± 565 K for samples S06 and S03, respectively. The larger error in the latter sample is due to its dual-emitting nature, which results to some extent in competition between core and shell ASE and consequently larger fluctuations in the ASE threshold. Interestingly, a nearly temperature-independent threshold is even obtained for the shell ASE of sample S03, which indicates that quasi-0D Qrods, with strong quantum confinement in the lateral dimensions and weak confinement along the 52 nm long Qrod axis, may also lead to enhanced temperature stability.
The temperature-independent ASE threshold is explained by considering the sparse density of electronic states in small colloidal Qdots. Indeed, the absorption spectra in Figure 1(b) reveal a typical energy separation of 190–250 meV and 210 meV between the first and second excited state of core and shell transitions, respectively. Hence, as these values are an order of magnitude larger than the thermal energy at room temperature, no significant thermal depopulation of the lowest emitting state will occur even up to 325 K. Support for this conclusion comes from the increased T0 for sample S03 (core diameter 2.2 nm) in comparison to S06 (core diameter 3.0 nm), as the reduced core dimensions in the former case result in a stronger excited-state splitting and thus a higher T0.
In summary, we have shown that CdSe/CdS Qdot-in-rods are excellent candidates for Qdot-based lasers, combining the advantages of wide spectral tuning of the ASE by varying the CdSe core size and a reduced gain threshold due to an enhanced absorption by the long CdS Qrod shell. Furthermore, they exhibit a high characteristic temperature T0 for the ASE threshold, enabled by the Qdot-in-rod's delta-like density of electronic states. Although future work on colloidal nanocrystals still needs to resolve the issue of efficient electrical injection (already available in epitaxial Qdot lasers), the first promising steps toward this goal have already appeared. Indeed, the fabrication of colloidal quantum dots with short inorganic ligands26 should allow overcoming the insulating barrier formed by the 1–2 nm thick organic ligand shell, paving the way toward a low cost, solution processable quantum dot laser.