Letter Cite This: Nano Lett. XXXX, XXX, XXX-XXX
pubs.acs.org/NanoLett
Probing Linewidths and Biexciton Quantum Yields of Single Cesium Lead Halide Nanocrystals in Solution Hendrik Utzat, Katherine E. Shulenberger, Odin B. Achorn, Michel Nasilowski, Timothy S. Sinclair, and Moungi G. Bawendi* Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, United States S Supporting Information *
ABSTRACT: Cesium lead halide (CsPbX3, X = Cl, Br, I) perovskite nanocrystals (PNCs) have recently become a promising material for optoelectronic applications due to their high emission quantum yields and facile band gap tunability via both halide composition and size. The spectroscopy of single PNCs enhances our understanding of the effect of confinement on excitations in PNCs in the absence of obfuscating ensemble averaging and can also inform synthetic efforts. However, single PNC studies have been hampered by poor PNC photostability under confocal excitation, precluding interrogation of all but the most stable PNCs, leading to a lack of understanding of PNCs in the regime of high confinement. Here, we report the first comprehensive spectroscopic investigation of single PNC properties using solution-phase photon-correlation methods, including both highly confined and blue-emitting PNCs, previously inaccessible to single NC techniques. With minimally perturbative solution-phase photon-correlation Fourier spectroscopy (s-PCFS), we establish that the ensemble emission linewidth of PNCs of all sizes and compositions is predominantly determined by the intrinsic single NC linewidth (homogeneous broadening). The single PNC linewidth, in turn, dramatically increases with increasing confinement, consistent with what has been found for II−VI semiconductor nanocrystals. With solution-phase photon antibunching measurements, we survey the biexciton-to-exciton quantum yield ratio (BX/X QY) in the absence of user-selection bias or photodegradation. Remarkably, the BX/X QY ratio depends both on the PNC size and halide composition, with values between ∼2% for highly confined bromide PNCs and ∼50% for intermediately confined iodide PNCs. Our results suggest a wide range of underlying Auger rates likely due to transitory charge carrier separation in PNCs with relaxed confinement. KEYWORDS: Cesium lead halide perovskites, perovskite nanocrystals, biexciton quantum yield, spectral lineshapes, single nanocrystal spectroscopy, photon correlation Fourier spectroscopy, semiconductor nanocrystals
T
down-shifting, and low-threshold lasing.12−15 The targeted advancement of these applications requires an in-depth understanding of the excited state dynamics in PNCs. For instance, identifying the origin of the emission linewidth may allow improvements in the color purity of PNC-based LEDs. Determining which physical parameters control the Auger lifetime would facilitate the realization of PNC-based single photon sources or lasers by either directed maximization or minimization of multiexciton quantum yields. Just as the large number of studies on the emission of CdSe NCs have guided the engineering of NCs with desirable minimal linewidth16,17 and unity biexciton emission quantum yield,18 similar studies of the emission of PNCs will undoubtedly facilitate their technological applications.
he extraordinary optical properties of bulk lead halide perovskites such as high absorption cross sections, high emission quantum yields, and defect tolerance have recently been translated into a variety of emissive nanomaterials.1−8 Notably, cesium lead halide (CsPbX3 X = Cl, Br, I) perovskite nanocrystals (PNCs) exhibit high emission quantum yields even in the absence of surface-passivating inorganic shells and can be prepared via remarkably robust colloidal syntheses.2,7,9 The emission energy of PNCs can be tuned very precisely with both halide composition and size-dependent quantum confinement. Owing to their labile anionic lattice, halide anion exchange can also be used to fine-tune the emission postsynthetically.10,11 The main drawback of PNCs is their limited chemical stabilityespecially toward polar media, and with increasing confinement. However, their compelling characteristics render PNCs a promising alternative for II−IV NC like CdSe. As such, they have been demonstrated in a plethora of applications like light-emitting devices (LEDs), light © XXXX American Chemical Society
Received: July 21, 2017 Revised: September 20, 2017
A
DOI: 10.1021/acs.nanolett.7b03120 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
reported values.2 TEM analysis allows us to conclude that our largest particles have a cubic shape. XRD analysis confirms a cubic crystal structure (TEM and XRD data in SI I,II).2 Figure 1 shows ensemble absorption and emission spectra. For the
Recently, single NC spectroscopy has been used to reveal some of the similarities and differences between the singleparticle emission of PNCs and that of established II−VI NCs. Similar to CdSe NCs, PNCs exhibit charge-induced (“A-type”) fluorescence intermittency (“blinking”) and a narrow ( Br/Cl). For both I- and Br-PNCs, the ESL increases significantly with increasing confinementan effect that is more pronounced for the I-PNCs. This size-dependent trend is consistent with results for core-only CdSe NCs, for which we have previously established that the single NC emission linewidth increases with increasing confinement.41 The broadening in CdSe may originate from highly polar surface phonons42 or surface related charge trapping, increasing coupling to optical phonon modes by inducing a permanent polarization in the NC.43,44 We also showed that, for II−VI NCs, surface passivation eliminates this effect of linewidth broadening with size, confirming that the surface plays a pivotal role in determining the NC linewidth. For PNCs, strong exciton−phonon coupling leads to a polaron effect in lead halide perovskites, the adaptation of the lattice to changes in the charge density due to excitation.45,39,46 This strong coupling leads to self-localization of excitonic carriers, or exciton self-trapping, which has already been observed for cesium lead halide nanowires.1,45 We suggest that, although in PNCs the surface does not significantly reduce the emission quantum yield through efficient nonradiative recombination, it may have a pronounced effect on the emission linewidth. This could possibly be due to enhanced coupling of the excited state to surface-phonon modes, or enhanced exciton self-trapping due to large reorganization energies of the labile PNC surface. Tailoring the surface structure synthetically, for instance by
Figure 4. Ensemble single excited state photoluminescence lifetimes and fits with the sum of three exponentials (dashed lines) for I-PNCs (a) and Br-PNCs and Br/Cl-PNCs (b). The long-time components of the decays become more significant with increasing size and from Br/ Cl to Br to I. We suggest that transitory carrier separation and shallow trapping in PNCs with relaxed confinement and PNCs with higher dielectric constants are responsible for the observed trend.
reported previously, PNCs exhibit multiexponential fluorescence decay behavior.48,2 We fit all traces with the sum of three exponentials, phenomenologically accounting for all confluent physical effects defining the ensemble PL lifetime (Figure 4, dashed black lines). The decay components range between τ = 2.2 ns and τ = 68 ns with bluer emitting PNC exhibiting shorter and more monoexponential PL decay behavior. Notably, the long time components become more dominant with increasing size and decreasing band gap of the underlying perovskite (I > Br > Br/Cl). Consistent with this trend, we also observe significant delayed emission following a power law decay up to D
DOI: 10.1021/acs.nanolett.7b03120 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters 1 μs after photoexcitation for large I1.84 eV-PNCs (see SI, VIII), as recently observed elsewhere.49 Similar delayed emission has also been observed in CdSe nanocrystals and nanoplatelets with relaxed confinement and has been attributed to transitory carrier separation.50,51 We thus suggest that, while confined PNCs exhibit mostly single-exponential decay dynamics indicating a predominantly excitonic recombination, excitations in large PNCs, in particular large I-PNCs, show the character of both excitonic and delayed emission due to temporarily separated carriers. Support for this hypothesis includes the fairly small exciton binding energies in lead halide perovskites compared to CdSe fueling the ongoing discussion in the literature about the primary nature of photoexcitations in perovskites (excitons vs carriers).52 Specifically, for cesium lead halide NCs, theoretical studies have yielded dielectric constants (ε = 6.32 for I-PNCs, 4.96 for Br-PNCs, and 4.07 for ClPNCs)14 giving rise to low exciton binding energies of 20, 40, and 75 meV, respectively.2 Even in highly confined 2D CsPbBr3 nanoplatelets, a relatively small exciton binding energy of 120 meV53 has been found that compares to several hundred of meV for CdSe based NC.54 These relatively small exciton binding energies may facilitate charge carrier separation in PNCs exhibiting low confinement energies. Delayed emission then results from a reduction of the electron−hole wave function overlap or thermal carrier detrapping from shallow defects prior to radiative recombination.50,51 This detrapping from shallow traps gives rise to power-law decay dynamics and has, for instance, been observed for bulk methylammonium lead halide perovskites.55 These results suggest that confining the excited state in PNCs can significantly suppress delayed emission, possibly reducing nonradiative losses from a longlived charge separated state. Taking the ensemble average single excited state quantum yield into account, we calculate an ensemble averaged radiative carrier lifetimea measure of the lifetime of the carrier recombination resulting in photon emission. With this definition, we account for the uncertainty in the nature of the excited state (bound exciton vs transitorily separated charges). In the limit of a purely excitonic recombination, the average radiative carrier lifetime equals the radiative lifetime of the excitonic transition. First, we define an average lifetime from the triexponential fits to the PL decay traces shown in Figure 4 as ⟨τ⟩ = ∑3i=1kiτi, where ki are the pre-exponential coefficients and τi are the lifetimes from our fits. We then estimate the average radiative carrier lifetimes of the single excited state manifold by taking the ensemble single exciton quantum yield into account: ⟨τr,X⟩ = ⟨τ⟩/QYX. The extracted radiative carrier lifetimes are ∼55 ± 8 ns, ∼17 ± 3 ns, and ∼22 ± 3 ns for I1.84 eV-PNCs, I1.92 eV-PNCs, and I2.02 eV-PNCs, respectively. The corresponding lifetimes for Br2.45 eV-PNCs and Br2.52 eV-PNCs are ∼9 ± 1 ns and ∼5 ± 1 ns. Our mixed Br/Cl2.75 eV-PNCs exhibit a radiative carrier lifetime of ∼11 ± 1 ns. For bromide and iodide particles similar to our Br2.45 eV-PNCs and I1.84 eVPNCs, radiative carrier lifetimes of 7.7 and 43 ns have been found using a similar methodology.48 We also obtain the ensemble averaged biexciton-to-exciton quantum yield ratio (BX/X QY ratio) using our solution-phase antibunching technique.24 For this, we detect the emission from freely diffusing PNCs in solution in a Hanbury−Brown−Twiss geometry utilizing four single-photon detectors and calculate the second order cross-correlation function of photons detected after excitation pulses separated by τ (see SI, IX for optical setup). In Figure 5a, we plot a typical peak integrated second
Figure 5. Second-order photon-correlation [g2(τ) − 1] as a function of the laser excitation pulse separation (τ) as measured for all PNCs studied (here shown for Br2.45 eV-PNCs). The correlation shows a clear decay on millisecond time scales due to particle diffusion and antibunching behavior for the center peak (τ = 0) (a). The evolution of the BX/X QY ratio for PNCs of different sizes and compositions as a function of the first absorption peak energy is shown in panel b. PNCs of different compositions, but a similar degree of confinement, exhibit largely different BX/X QY ratios indicating halide composition dependence for Auger recombination.
order correlation function corrected for uncorrelated emission [g(2)(τ) − 1], as obtained for all PNCs studied. The loss of correlation with characteristic time scales of milliseconds corresponds to the diffusion of single emitters out of our microscope’s confocal volume, akin to fluorescence correlation spectroscopy (FCS). On time scales much faster than the particle dwell time, the correlation function is not significantly influenced by particle diffusion. In contrast to a conventional single molecule antibunching experiment, at a given instant, more than one PNC can occupy the confocal volume, and twophoton detection events after a single excitation laser pulse can originate not only from biexciton emission but also from single exciton emission by two different PNCs. By correcting for the uncorrelated background emission from different PNCs [g(2)(τ ≫ τdwell) = 1], we recover the center- and side-peak correlation counts originating from single PNCs. As in a conventional antibunching experiment, the ratio of the correlation center peak [g(2)(0) − 1] to the correlation side peak [g(2)(τrep) − 1] corresponds to the BX/X QY ratio, however, due to particle exchange during the measurement, to the ensemble averaged BX/X QY ratio.24,56 The FCS traces of all PNCs measured showed clear decay behavior indicating the absence of particle aggregation in solution (see SI, X all traces). We stress that all measurements of the BX/X QY ratio have been conducted at very low excitation fluxes with average excitations created per pulse per PNC of
