Suppressed Blinking and Auger Recombination in Near-Infrared Type-II InP/CdS Nanocrystal Quantum Dots

Abstract

Non-blinking excitonic emission from near-infrared and type-II nanocrystal quantum dots (NQDs) is reported for the first time. To realize this unusual degree of stability at the single-dot level, novel InP/CdS core/shell NQDs were synthesized for a range of shell thicknesses (~1–11 monolayers of CdS). Ensemble spectroscopy measurements (photoluminescence peak position and radiative lifetimes) and electronic structure calculations established the transition from type-I to type-II band alignment in these heterostructured NQDs. More significantly, single-NQD studies revealed clear evidence for blinking suppression that was not strongly shell-thickness dependent, while photobleaching and biexciton lifetimes trended explicitly with extent of shelling. Specifically, very long biexciton lifetimes—up to >7 ns—were obtained for the thickest-shell structures, indicating dramatic suppression of non-radiative Auger recombination. This new system demonstrates that electronic structure and shell thickness can be employed together to effect control over key single-dot and ensemble NQD photophysical properties.

Fluorescence intermittency, or blinking, is exhibited by a number of fluorophores, including molecular dyes,¹ fluorescent proteins,² and nanocrystal quantum dots (NQDs).^{3, 4} Blinking behavior in NQDs has most commonly been attributed to the coupling of random charging/discharging events with activation of efficient non-radiative Auger recombination (AR) in charged or ionized NQDs.⁵ Attempts to suppress fluorescence intermittency in this class of emitter have often entailed addressing the charging portion of the blinking problem through supplementation with external reducing or electron-donating species (e.g., β-mercaptoethanol, propyl gallate) that electronically passivate surface trap states and suppress pathways for photo-induced charge formation,^6–8 or the use of charge-accepting species (e.g., TiO₂ nanoparticles) to provide enhanced routes for “de-trapping” and reneutralization.⁹ In contrast, modification of the NQD geometry or internal structure alters intrinsic AR, such as when AR is suppressed in elongated (larger volume) nanorods, where exciton-exciton coupling is reduced,¹⁰ or in type-II or quasi-type-II heterostructures (carriers spatially separated), where electron-hole overlap is reduced and even repulsive Coulombic interactions can develop.¹¹ Alloying of the core/shell interface can also suppress AR by inducing a smoothing of the confinement potential.^{5, 12}

To date, explicit experimental correlation of AR suppression with room-temperature blinking suppression has been limited to an alloyed CdZnSe/ZnSe NQD system¹² and a quasi-type-II, thick-shell CdSe/CdS system,^13–17 where the latter (so-called g-NQDs¹³) likely benefit from both suppressed charging (due to the thick, protective shell limiting ionization processes and/or possibly carrier trapping) and suppressed AR (due to a combination of size, carrier-separation and, possibly, interface effects¹⁷). The ability to simultaneously suppress blinking and non-radiative Auger recombination (AR) has important implications for a range of applications from single-particle tracking in the biosciences¹⁴ to low-threshold multi-color lasing¹⁷ and efficient multiexciton emission and extraction.¹⁸ Here, we move beyond the limited examples of CdSe-based core/shell engineering to show that judicious NQD heterostructuring can afford significantly suppressed blinking and AR in InP-based NQDs, demonstrating the approach as a general method for effecting control over key NQD light-emission properties. Specifically, we show that a new InP/CdS NQD system exhibits both suppressed blinking and AR, similarly to their thick-shell CdSe/CdS g-NQD counterparts. However, unlike the earlier system, blinking suppression is not strictly confined to thick shells. Furthermore, InP/CdS NQDs are distinguished by their fully type-II bandgap alignment, which affords complete spatial separation of excited-state carriers and transitions the room-temperature non-blinking excitonic emission for the first time into the near-infrared. Significantly, room-temperature blinking behavior for a fully type-II system has also not been previously demonstrated, as attempts have been thwarted by extreme sensitivity to photobleaching and low signal-to-noise at the single-dot level.¹⁹

In this work, we address the effects of both type-II electronic structure and shell thickness. To this end, we developed the novel InP/CdS core/shell NQD. Consideration of the alignment of electronic states for the bulk InP-CdS interface reveals that InP/CdS NQDs should afford a fully charge-separated (type-II) state, where electrons reside in the CdS shell and holes are confined to the InP core (Figure 1a). Furthermore, our core/shell synthesis approach allows us to fabricate CdS-overcoated InP NQDs with shell thicknesses ranging from 1 to >11 monolayers (MLs). Similar to our CdSe/CdS g-NQD synthesis, the CdS shell is deposited monolayer-by-monolayer onto the InP cores through a successive ionic layer adsorption and reaction (SILAR) method (see Supporting Information).^{13, 14, 20} However, compared to CdSe, InP cores are significantly more susceptible to surface oxidation and particle etching,²¹ the former requiring unusual care in applying air-free procedures prior to shell growth and the latter preventing use of high growth temperatures combined with acidic or basic ligands.²² It is likely for these reasons that relatively fewer successful high-temperature, SILAR-based preparations of core/shell NQDs with InP cores have been reported.²³ To render InP NQDs amenable to thick-shell CdS shell growth, we devised a method whereby the initial shell monolayer is applied using lower, non-ideal shell-growth temperatures and serves as a “protective” layer for subsequent shell additions that employ conditions optimized for higher quality shell growth.

Figure 1.

InP/CdS core/shell NQDs. a) Schematic of the InP/CdS core/shell structure and the relative bulk bandgap alignment of the core and shell materials. Localization of the excited electron in the CdS shell and the resulting hole in the InP core is indicative of type-II bandgap structure. b–e) TEM images of InP/CdS core/shell nanocrystals after 1, 4, 7, and 10 successive shell depositions, respectively. The scale bar for all 4 images is 5 nm.

Overall, InP cores (2.46 nm diameter) were exposed to 10 successive cation/anion addition cycles. Analysis of transmission electron microscopy (TEM) images (Figure 1b–e, Table 1) reveals increasing shell thickness with each shell addition cycle, with slightly more than one monolayer added per iteration. For example, one shell addition cycle (“InP/1CdS”) affords approximately 1.6 monolayers of CdS, while ten shell addition cycles (“InP/10CdS”) yields 11.4 monolayers of CdS shell. TEM analysis also demonstrates the reasonably uniform particle size and morphology achieved for this system, with deviations in diameter size of ~10% (Table 1). Small-angle x-ray diffraction (XRD) of thick-shell InP/CdS NQDs, which comprise predominantly CdS (>98% CdS by volume for InP/10CdS), reveals that CdS retains its thermodynamic hexagonal (wurtzite) crystal structure even though it has been applied to cubic (zinc-blende) InP NQDs (Figure S1, Supporting Information). Despite the inequivalencies in crystal structure and no compositional commonalities between the core and the shell materials, the InP/CdS NQDs benefit from a small lattice mismatch (<1%) and exhibit remarkably regular shapes, characterized by hexagonal faceting as shell thickness is increased (Figure 1b-e). The shell quality stands in contrast with some previously reported type-II systems,^{24, 25} but is similar to our CdSe/CdS g-NQD system.¹³ Based on these similarities in physical structure with the prototype g-NQD system, the InP/CdS NQDs allow direct assessments of blinking and AR suppression in terms of shell-thickness effects.

Table 1.

Summary of InP/CdS NQD properties.

Number of CdS shell-addition cycles	Diameter (nm)a	Shell thickness (nm)	No. of shell MLs from TEM analysis	PL peak position (nm/eV)	PL FWHM (nm/eV)	Average PL lifetime, τx (ns)	Biexciton lifetime, τxx (ps)
InP core	2.46	n/a	n/a	~600/2.1b	n/a	80–150c	n/a
InP/1CdS	3.52 ± 0.38 (11%)	0.53	1.6	696/1.78	124/0.32	190	58 ± 12
InP/2CdS	4.56 ± 0.49 (11%)	1.05	3.1	849/1.46	476/0.38	327	n/a
InP/3CdS	5.40 ± 0.52 (10%)	1.47	4.4	904/1.37	483/0.34	349	n/a
InP/4CdS	6.26 ± 0.65 (10%)	1.90	5.6	937/1.32	211/0.29	435	99 ± 5
InP/7CdS	7.97 ± 0.84 (10%)	2.76	8.2	982/1.26	220/0.28	575	612 ± 49
InP/10CdS	10.13 ± 0.90 (9%)	3.84	11.4	1000/1.24	229/0.28	702	7,212 ± 1,073

^aInP core diameter determined from 1S peak position.²⁶ InP/CdS core/shell diameter determined from TEM image analysis. Error represents standard deviation of 157–485 measured dots.

^bInP cores are non-emissive, so PL peak position derived from 1S peak position (544 nm/2.28 eV) plus an estimated Stokes shift.

^cLifetime values for InP from the literature.^{27, 28} See Supporting Information Table S1.

With respect to electronic structure, we find clear spectral and dynamic signatures of a type-II heterojunction. In type-II systems, the bandgap energy derives from a spatially indirect transition,^{25, 29} or charge-transfer state,³⁰ that involves electron-hole recombination across the core-shell interface, as the carriers are spatially confined to different regions of the heterostructure (Figure 1a). Thus, the bandgap can be smaller than that achievable by either the emitting core or the shell material alone, and exciton lifetimes are typically long.^{24, 31} For InP/CdS, we find that photoluminescence (PL) and bandgap energies indeed progress toward lower energies as CdS shells are added to InP (Table 1, Figure 2a,b), and PL lifetimes increase (Table 1, Figure 2c). Spectral red-shifting is clearly evident with the addition of the first several shell monolayers, after which changes in PL energy as a function of shell thickness are less substantial (Table 1, Figure 2b). Specifically, addition of only three CdS monolayers (InP/2CdS) results in a 1.46-eV core/shell emission energy that is >600 meV red-shifted from the ~2.1-eV starting InP core value (necessarily estimated here from the InP core absorption 1S peak, as core-only emission is negligible). The final emission energy reached by thicker shell InP/CdS NQDs (~1.25 eV) is less than the bulk bandgap of InP (1.34 eV). For all shell thicknesses, core/shell bandgap energies are smaller (“redder”) than the bandgap energies for equivalently sized InP NQDs (Figure 2b), consistent with the formation of a type-II electronic structure at very early stages of shell growth. This conclusion is supported by theoretical modeling of carrier localization behavior in InP/CdS NQDs using the effective mass approximation (Supporting Information), which affords a similar trend in bandgap changes as a function of shell thickness (Figure 2b, red trace). In contrast to energy shifting trends, PL lifetimes increase roughly linearly with shell thickness, reaching an ~7-fold enhancement compared to InP cores at the thickest shells (11.4 MLs) (Tables 1 and S1, Figure 2c).

Figure 2.

Ensemble spectroscopy. a) Absorption (dashed lines) and photoluminescence (PL) (solid lines) spectra of the InP core and InP/CdS core/shell structures (excitation at 405 nm). The dashed line at 800 nm indicates where PL detectors were switched. Shoulders in the PL peaks in this region are artifacts of using two different detectors. b) Bandgap energy vs. particle size. The InP core bandgap energy (data point without error bars) was determined by the 1S peak in the absorbance measurement, while the InP/CdS core/shell bandgap energies (remaining points) represent PL energies. One core size is depicted with multiple shell thicknesses. The blue line represents the bandgap energy expected for an InP NQD at various sizes. The red line depicts our predicted bandgap energy for a specific core/shell size combination (i.e., core size and shell thickness). Error bars in the x and y represent the standard deviation of the particle size and FWHM of the PL peak, respectively. c) PL lifetime measurements. Inset: plot of lifetime vs. shell thickness. d) Photoluminescence excitation spectra (PLE) of several InP/CdS structures. Emission set at 650, 875, and 900 nm for the InP/1CdS, InP/4CdS, and InP/7CdS samples, respectively.

Absorption spectra also reveal the transition to type-II behavior. Here, the well-defined 1S absorption peak of the starting InP cores is broadened and red-shifted by initial shell addition, consistent with previous accounts of type-II NQD absorption trends.²⁴ InP/CdS NQD absorption is subsequently dominated by the thick CdS shell at wavelengths <520 nm, resulting in blue-shifting of the absorption onset and a larger effective Stokes shift (separation between emission and absorption), like CdSe/CdS g-NQDs but unlike previously reported type-II NQDs. Significantly, we find that photoluminescence excitation (PLE) spectra follow similar trends as the absorption spectra (Figure 2d), suggesting that energy relaxation from a shell-derived excited state to the indirect or charge-transfer emitting state is efficient and dominates emission.

To determine how shell thickness and a type-II electronic structure influence charging and AR-mediated processes, we attempted to assess single-dot-level blinking behavior, as well as photostability and biexciton lifetimes. Notably, room-temperature blinking studies on fully type-II core/shell NQDs have not been reported to date. Though such studies are now well known in the case of quasi-type-II CdSe/CdS^13–15 and presumed quasi-type-II CdTe/CdTe_xSe_y/CdSe NQDs,¹⁹ photobleaching under single-dot excitation conditions has prevented such analyses on type-II systems.¹⁹ Perhaps more strikingly, single-NQD spectroscopy for any type of NQD emitting in the near-infrared is also almost unknown. Even nominally brighter type-I infrared-emitting NQDs have proven difficult to assess by this method, requiring use of highly specialized single-photon detectors to surmount materials issues of instability and relatively long emission lifetimes.³² This omission to date stands in stark contrast with self-assembled epitaxial quantum dot reports that have benefitted from the relative photostability and shorter radiative lifetimes of these non-colloidal systems.³²

In contrast to prior efforts, standard sample preparation and single-photon detection methods proved adequate, though not ideal (see below), for obtaining intensity-versus-time traces at the single-NQD level. Specifically, freshly prepared, ultra-dilute solutions were spread onto glass slides and the resulting dispersion of single InP/CdS NQDs were interrogated using a continuous wave (CW) laser at 405 nm. Photons were collected on an avalanche photodiode (APD) detector in single-photon-counting mode, and the PL intensity time trace was analyzed using a 20 ms bin time. We were able to obtain single-NQD data for the full range of shell thicknesses.

Interestingly, with respect to blinking behavior, a simple shell-thickness dependence was not observed (Figure 3), in contrast with the quasi-type-II CdSe/CdS system.¹⁴ For example, some relatively thin-shell InP/1CdS and InP/4CdS NQDs exhibited conventional blinking behavior, while others exhibited blinking suppression (Figure 3a), with ~15–25% of the dots demonstrating on-time fractions of >80% of the observation time (5 minutes). Notably, conventional blinking behavior (e.g., that typical of CdSe/ZnS NQDs) is characterized by 0% of a NQD population possessing on-time fractions >80%.¹⁴ Blinking suppression does improve as shell thickness is increased to 3.84 nm (InP/10CdS), though not as dramatically as observed for CdSe/CdS g-NQD shell-thickness trends. That said, >60% of the InP/10CdS NQDs are on >80% of the observation time, and several dots are observed to be essentially non-blinking, i.e., on-time fractions ~95% (Figure 3a, InP/10CdS “dot indices” 2, 9, 10, and 11). Given the difficulties of performing single-dot-level measurements in the near-infrared,³² our “blinking” data set is relatively limited (total dot numbers in the tens compared to hundreds typical for visible-emitting NQDs). Thus, while it is clear that blinking is suppressed in the InP/CdS system, the precise shell-thickness dependence and non-blinking dot fractions remain to be determined as measurement techniques improve.

Figure 3.

Blinking and photobleaching. a) Summary of blinking behavior for multiple individual NQDs for several shell thicknesses. Each dot index represents the blinking behavior of one dot. Black regions in each image represent times when the dot is in the “off” state. Colored regions correspond to “on” states. Green, orange, and red traces represent single dot data from InP/1CdS, InP/4CdS, and InP/10CdS, respectively. b) Representative intensity trajectories of single NQDs (both InP/10CdS). The top and bottom traces correspond to dot indices 4 and 10 in the top panel, respectively. c) Relative photobleaching from each sample over time. NQDs dropcast and imaged under widefield laser excitation (405 nm) using a CCD with 1 s integration time for three hours. The number of dots having a PL intensity above a threshold value was tabulated for each frame. The fraction of dots meeting this criteria (relative to the number of dots present at t = 0) is plotted as a function of time.

We also studied photobleaching behavior for thin (InP/1CdS and InP/4CdS) and thick-shell (InP/10CdS) samples over a 3 h period with continuous laser excitation. Here, shell-thickness dependency is more evident (Figure 3c). InP/1CdS NQDs essentially completely photobleach over the course of the experiment, while ~20% of InP/4CdS NQDs continue to emit for the duration of the experiment. In contrast, ~60% of InP/10CdS NQDs do not photobleach. In all cases, the observed declines in PL occur during the first ~30 minutes of the experiment, suggesting that NQDs that are inclined to photobleach do so in a manner that is consistent with that observed for conventional core/shell NQDs for which the photobleaching half-life (t_1/2) is ~15 min.^{13, 14} The remaining non-photobleaching dot population (~20% of InP/4CdS and ~60% of InP/10CdS) is photostable, exhibiting no further decline over the 3 h observation time. Thus, thicker CdS shells promote improved photostability, relevant for a range of potential applications from single-photon light sources to single-molecule tracking in biology.

In an attempt to understand the observed blinking suppression in terms of contributions from different aspects of the core/shell structure, especially shell thickness and electronic structure, we consider further the underlying processes that can lead to blinking in NQDs. Specifically, fluorescence intermittency in NQDs was recently shown to result from one of two mechanisms – (1) the “conventional” charging model that describes blinked-off events as resulting from efficient non-radiative AR in charged NQDs and (2) hot-electron capture to surface trap-states that short circuits radiative electron-hole recombination but does not invoke NQD-charging or AR processes.³³ Addition of a high-quality thick shell might be expected to suppress charging and hot-electron-capture pathways by minimizing carrier tunneling and core/surface-state interactions, respectively, thereby suppressing both blinking mechanisms. This seems a reasonable implication of applying thicker and thicker CdS shells to InP. However, our observations of this system show that blinking suppression is possible even for thin CdS shells, for which neither suppressed charging nor suppressed surface-state trapping should be expected. In this case, reduced AR efficiency is possibly a significant factor in the observed blinking-suppressed behavior. Indeed, AR efficiencies are in general reduced in type-II core/shell NQDs compared to type-I NQDs due to lessened electron-hole overlap and/or exciton-exciton repulsion.²⁴

To confirm AR suppression in the new type-II InP/CdS NQDs and to establish any shell-thickness dependencies, we obtained pump-intensity-dependent PL decay time traces for multiple shell thicknesses and extracted biexciton lifetimes (τ_xx). Longer biexciton lifetimes indicate reduced AR efficiencies.¹⁷ We find that biexciton lifetimes exhibit a dependence on shell thickness (Figure 4) and are substantially longer for the thickest-shell sample (by a factor of ~125 for 10CdS vs. 1CdS). Significantly, however, although this shell-dependent lengthening of the biexciton lifetime is dramatic, τ_xx remains short compared to τ_x (InP/10CdS: τ_xx = τ_x/97.5; Table 1). According to established theories of radiative decay in NQDs, radiative biexciton lifetimes should be reduced compared to radiative exciton lifetimes by only a factor of 2–4.²⁷ The observed deviation of InP/CdS τ_xx values from ideal radiative biexciton lifetimes [i.e., from ~τ_x /(2–4)] results from contributions from nonradiative processes, likely primarily AR.¹⁷ In this way, the ratio between single and biexciton lifetimes gives a measure of the relative strength of AR. It is then apparent that although nonradiative AR is greatly diminished in this system with increasing shell thickness (e.g., comparing InP/10CdS to InP/1CdS τ_x/τ_xx: 97.5 vs. 3276, respectively), AR is still not zero and is capable of contributing to less-than-optimal biexciton efficiencies. For this reason, we cannot conclusively state that the observed AR suppression is sufficient to explain the InP/CdS blinking suppression based on an AR model, and, thus, it remains to be determined what blinking pathways are important in this unusual system. Nevertheless, the remarkable enhancement in biexciton lifetime with increasing shell thickness has important implications for applications of this new g-NQD system in optical amplification and lasing, where stimulated emission most directly competes with Auger relaxation, and single-exciton recombination processes are less relevant.³⁴

Figure 4.

Biexciton lifetimes. (a) Spectrally integrated transient-PL measurements for a 5.2 nm diameter InP/CdS core/shell sample (InP/4CdS) in dilute toluene solution at different 3.1 eV pump fluences (1 to 170μJ/cm²). Traces normalized at 1.8 ns show that all multiexciton signals have decayed within less than 500 ps. The inset shows the extracted multiexciton decay component as well as a single exponential fit (red line) with a 99 ± 4 ps biexciton decay. (b) For an 8.3-nm diameter core/shell sample (InP/10CdS), pump-intensity-dependent transient-PL measurements normalized at 48 ns do not fully converge until ~30 ns, indicating long-lived multiexcitons. A single exponential fit (red line) to the extracted multiexciton component¹⁰ (shown in the inset) reveals long-lived multiexcitons. (c) Fitted biexciton lifetimes extracted from time-resolved PL data show a large lifetime increase as a function of shell thickness and total particle diameter.

From the perspective of intentional design of functional nanoscale structures and interfaces, we have shown that both the physical structure (thick shell) and the electronic structure (charge-separated excited state, or type-II alignment of energy bands) are important design parameters for realizing unusual and practical photophysical properties. This novel example of thick-shell heterostructuring demonstrates that photostability at both the single-NQD and ensemble levels, as well as intrinsic NQD properties such as AR, can be tuned by considering both physical and electronic-structure effects. These new InP/CdS near-infrared emitters—characterized by significant room-temperature photostability and blinking- and AR-suppressed behavior—open immediate opportunities from advanced bioimaging/single-particle tracking to low-threshold near-IR lasing. They have also permitted for the first time analysis by standard single-NQD spectroscopy methods of near-infrared emission at the single-particle level.