Graded Shells in Semiconductor Nanocrystals

1 Introduction

Semiconductor nanocrystals (quantum dots, QDs) have been known for more than 30 years for their superior photophysical properties. They have narrow emission lines, with energies determined by the size quantisation effect and ensemble peak widths that depend on inhomogeneous broadening from the size distribution in the particle ensemble. However, the high surface-to-volume ratio of quantum dots leads to a very strong contribution of surface effects on the photophysics of quantum dots. These can be observed by the appearance of a broad, red-shifted emission from surface trap states [1], decreased photoluminescence quantum yield (PL QY) and, on the level of individual QDs, as switching between intermittent luminescent and dark periods of photoluminescence, referred to as “blinking” or fluorescence intermittency, with power-law behaviour for the on and off time distributions [2], [3]. Two processes have been identified as the origin of these effects that are generally considered detrimental: transfer of charge carriers to localised surface states and non-radiative Auger processes, in which energy from exciton recombination or carrier cooling is used for intra-band excitation of a third charge carrier. Carrier trapping and Auger relaxation are connected in that charging of QDs by ejecting a single charge carrier is assumed to quench PL, because Auger relaxation of an exciton in a charged QD (a “trion”) or multiple excitons is 2–3 orders of magnitude faster than radiative recombination. Reversely, ejection of charge carriers from the nanocrystal can be mediated by Auger excitation. This manifests itself most prominently in the fluorescence intermittency or “blinking” of single quantum dots, which has been attributed to random charging and discharging of the nanocrystal under illumination with the charged state being dark due to the available Auger pathway [3].

In order to mitigate surface effects core/shell QDs have been synthesised in which a wider band gap material than that of the core spatially separates surface-related states from excitonic wave functions. These decay exponentially into the shell. This band alignment is called a type-I configuration. Examples include CdSe/ZnS, CdS/ZnS, InP/ZnS, and PbSe/PbS [4], [5], [6], [7], [8]. Alternative band alignments include the type-II configuration with a staggered band alignment that induces charge separation into core and shell (CdSe/CdTe, ZnSe/CdS) and the intermediate quasi-type-II configuration in which one carrier is confined to the core and the other is delocalised over core and shell (CdSe/CdS) [9], [10], [11], [12]. These materials replace the core surface with a sharp interface between core and shell, which brings a number of implications that arise from interfacing two different, crystalline materials: if the core has a larger lattice constant than the shell (as is generally the case for type-I structures) high tensile strain on the shell for thin shells and high compressive strain on the core for thicker shells is induced [13]. This is the case for one of the best studied systems, CdSe/ZnS, for which even moderately thick shells >2 monolayers (ML) leads to dislocations, additional defects and interface states, which directly counteracts the aim to spatially separate exciton and surface [14]. A move to smaller lattice mismatch, e.g. from CdSe/ZnS to CdSe/CdS, also reduces the band offset and thus exciton confinement. To further improve fluorescence properties an thin adapter layer has been introduced that reduces a large lattice mismatch to two smaller steps. CdSe/CdS/ZnS and CdSe/ZnSe/ZnS double shell QDs have been shown to significantly improve the properties over single shell particles with PL QY up to 70% [14], [15]. This approach has been followed further by adding further layers of mixed semiconductors, CdSe_xS_1−x and Cd_xZn_1−x S, [16] and finally by synthesising shells with a continuously graded shell.

Core/shell nanocrystals with sharp interafces have been reviewed in detail elsewhere [12], [17], [18]. In this review the recent developments and advances in the fabrication, crystallographic analysis, and photophysics of graded core/shell nanocrystals will be discussed. In Section 2 a brief overview of the current synthetic approaches to fabricate graded shell particles will be given. Section 3 addresses the impact of phase mixing and random distribution of ions in the graded shell on the photophysics of the nanocrystals. A detailed discussion of the effective mass approximation (EMA) model for charge carrier dynamics is given in Section 4, while the the implications of graded shells for applications involving multiple charge carriers such as optical gain and carrier multiplication will be discussed in Section 5.

2 Synthetic approaches and structural analysis

Core/shell nanocrystals have been made with monolayer compositional control using single ion layer adsorption and reaction (SILAR) method, in which solutions of cations and anions are added and reacted alternately to grow a shell monolayer by monolayer [19]. With this approach Xie et al. have synthesised CdSe/CdS/Cd_0.5Zn_0.5S/ZnS nanocrystals with a near-continuous potential gradient, limited by atomic resolution [16]. SILAR creates a near-perfect crystalline phase and miminises the occurrence of defect or trap states. However, the method is tedious and relies on both a precise knowledge of seed concentration and a complete turnover of reactants to be applicable for a large number of shells lest unreacted species from previous shells be accumulated in the reaction mixture. Therefore alternative synthetic routes have been developed. Graded shells are accessible by adding both Cd and Zn precursors simultaneously, either to the shell growth reaction or directly during the hot-injection step, if slightly different reactivities of the precursors lead to preferential reaction of one species over the other. This has been used to fabricate CdSe/Cd_xZn_1−xS/ZnS particles using metal acetates as precursors [20], and CdSe_xS_1−x particles with a Se-rich core and a S-rich shell using tri-n-butylphosphine (TBP) chalcogenides [21]. In the latter case it was shown that the degree of graduation depends crucially on reaction temperature: While at T=150 °C the TBP sulfide precursor was not activated, yielding pure CdSe particles, reaction at T=315 °C resulted in homogeneously alloyed particles. The temperature range from 220–260 °C led to graded shells, although a clear temperature-dependent trend of the gradient could not be established. McBride et al. argue that grading is a necessary prerequisite to achieve full surface passivation during shell growth, while hard interfaces (e.g. made by the SILAR method) favours growth on anion-terminated facets [22]. This and the mitigation of lattice strain at the interface allows to grow thick shells, leading to the class of “giant” shell CdSe/CdS nanocrystals, which show near-unity PL QY and no fluorescence blinking [23], [24], [25].

More recently we have employed a high temperature protocol developed by Chen et al. to create a graded shell in situ during shell growth by ion diffusion between new shell material and the surface of the seed particles [26], [27]. These particles exhibit PL QY between 75 and 95% before and after growth of the ZnS shell, but much weaker impact of electron and hole scavengers after ZnS growth. The reaction occurs at 270–310 °C under thermal decomposition of a thiol and metal oleates and employs slow addition of the precursor solutions to limit the reaction rate. A lower limit of 270 °C has been shown before for the occurrence of phase mixing to form Cd_xZn_1−xSe from CdSe/ZnSe core/shell nanocrystals. This temperature regime was observed as well for the combinations CdS/ZnS and ZnSe/CdS [26], [28], [29]. The degree of alloying strongly depends on temperature and could be observed as a blue-shift of the band-edge absorption and emission lines when going from CdSe/CdS to CdSe/CdS/ZnS or from ZnSe/CdS to ZnSe/CdS/ZnS nanocrystals. The shift is caused by mixing of CdS and ZnS phases, which increases the confinement potential of the conduction band of the inner shell (see Figure 1). Observation of a blue-shift is limited to a window in which alloying is incomplete and a core/shell/shell structure is preserved. For prolonged heating alloying smoothes out the confinement potential, carriers spill out into the shell, and the spectral shift of the PL is reversed.

Fig. 1:

(a) Absorption and PL spactra of CdSe (blue), CdSe/CdS (yellow) and CdSe/CdS/ZnS (green) showing the blue-shift caused by interface alloying. (b) Absorption and PL peak shifts during ZnS shell growth at 260 °C (■), 280 °C (▲), and 310 °C (*). (c) Schematic of the formation of graded shells in CdSe/CdS/ZnS particles and the corresponding confinement of wave functions during ZnS shell deposition at high temperatures. Reprinted with kind permission from Ref. [26] Copyright 2013 American Chemical Society.

Comparison studies between hard and soft interfaces have made it necessary to produce particles with sharp potential steps first and induce alloying in a subsequent annealing step. This has required shelling at low temperatures to preserve the size distribution of the seed particles and prevent premature alloying. Grumbach et al. have performed CdS growth on CdSe at temperatures ≤130 °C with annealing performed in 10 °C steps up to 160 °C. This protocol led to highly emissive CdSe/CdS particles with PL QY up to 92% [30]. Annealing has been employed intentionally or unintentionally in a number of publications to form graded interfaces and increase particle quality (judged by PL QY, mitigation of fast PL decay components, and long-term colloidal and fluorescent stability) [10], [27], [31], [32].

The formation of an alloyed layer can be observed using Raman spectroscopy, which detects the longitudinal (LO) and surface (SO) optical phonon modes of core and shell materials, CdSe at ~209 cm⁻¹ and CdS at ~290 cm⁻¹. Alloy formation causes an increase of the SO bands relative to those of the corresponding LO phonon, which can be attributed to signals from an alloyed CdSe_xS_1−x layer at the core/shell boundary [30], [33]. A complementary proof of alloying can be observed in powder XRD patterns, e.g. for CdSe/CdS particles treated at increasingly higher annealing temperatures, which show a systematic shift towards larger angles 2θ following Vegard’s law. The shifts occur in a range below what is expected for a homogeneously alloyed phase and confirm partial phase mixing [30]. Elemental mapping on the level of individual particles was performed by Keene et al. on CdSe_xS_1−x nanocrystals using energy-dispersive x-ray spectroscopy (EDX) in combination with high annular dark-field detected scanning electron microscopy (HAADF-STEM) [34]. Using this method they were able to prove that CdSe forms preferentially before CdS in a mixed precursor solution while preserving the chalcogenide ratio of the reaction mixture in the final product.

While this review focuses on the recent advances made for II–VI semiconductor nanocrystals, III–V (InP/ZnSe_xS_1−x) and IV–VI semiconductors (PbSe/PbSe_xS_1−x) have been synthesised as well with graded alloy composition [35], [36], [37], [38]. In all cases superior robustness was reported compared to similar particles with hard interfaces.

3 Impact of random phase mixing

Phase mixing at the interfaces leads to a random distribution of ions at the graded interface. This graininess of the real core/shell interface causes a deviation from spherical symmetry that cannot be modelled by EMA calculations based on either continuously or step-wise changing potential. For homogeneously alloyed, sulfur-rich CdSe_xS_1−x QDs Mourad, Hens, and co-workers have observed the appearance of additional transitions in the steady state absorption spectrum that are absent in both pure CdSe and CdS samples [39]. The particles were calculated using a multiband empirical tight-binding (ETBM) model and yielded electron states that could be classified as having s, p_x,y,z, or p± symmetries for pure CdSe (x=1) and CdS (x=0). Hole states showed some hole band mixing between s and p states. Alloying (e.g. for x=0.5) lead to visible distortions and loss of spatial symmetry of both electron and hole states (see Figure 2). The experimental observations can thus be rationalised by a symmetry breaking of electron and hole states that lift the optical selection rules for S/P excitons.

Fig. 2:

(a) Experimental UV/vis absorption spectra and (b) colour map representation of homogeneously alloyed CdSe_xS_1−x QDs showing an additional transition for sulfur rich particles. (c) Optical spectra obtained from stochastic ETBM. (d) Modulus squared wave functions for the first 4 electron and hole states for x=0, x=0.5, and x=1. Reprinted with kind permission from Ref. [29]. Copyright 2014 American Chemical Society.

We have recently observed evidence for a similar set of formally forbidden transitions in ZnSe/CdS and ZnSe/ZnS/CdS nanocrystals with graded shells. These structures have a type-II configuration with electrons localised in the CdS shell and holes confined to the ZnSe core. Without a ZnS barrier layer the electron ground state is partially delocalised into the core, approaching quasi-type-II behaviour, but overlap is mitigated when a ZnS barrier layer is added [29]. Transient absorption (TA) spectra reveal two bleaches, which correspond to the spatially indirect, band edge state 1S_3/2–1S_e (λ=465 nm) and the excited state 2S_3/2–1S_e (λ=530 nm), in which the hole extends to the CdS shell (see Figure 3). Both signals share the same electron state and thus decay with the rate of exciton recombination (τ≈30 ns). The two states are well separated and a broad, weaker bleach with faster decay kinetics (τ≈6 ns) can be observed. This bleach was attributed to mixed S/P states and is inhomogeneously broadened by the random distribution of ions in the alloyed shell across the particle ensemble. Alternatively, the broad bleaches could be signatures of carriers trapped in delocalised interfacial states, as some more recent reports suggest (vide infra).

Fig. 3:

TA spectra of ZnSe/CdS nanocrystals with graded interfaces up to 10 ps (a) and from 10 ps to 1 μs (b). State filling of the band edge electron determines the 1S_3/2–1S_e and 2S_3/2–1S_e decay, while the mixed states at 500 nm decays much faster (c). Reprinted with kind permission from Ref. [29]. Copyright 2015 American Chemical Society.

The oscillator strengths of the 1S_3/2–1S_e and 2S_3/2–1S_e excitons exhibit a strong dependence on the degree of grading, controlled by shell growth or annealing temperature. Between the lowest (260 °C) and highest (310 °C) shell growth temperature employed in the study a clear trend was observed for the ratio between the integrated bleaches corresponding to the two states. With increasing temperature the 1S_3/2–1S_e bleach decreased in favour of the excited 2S_3/2–1S_e state (see Figure 4). Control experiments performed at 260 °C with extended growth time could exclude a trivial effect of different shell thicknesses due to temperature-dependent growth kinetics [28]. Since both states are bleached due to state filling of the conduction band edge electron the integrals demonstrate that grading has significant effect on the radial distribution of holes in a core/shell particle. The exact distribution is dictated by two competing effects: the gradual diffusion of ions between core and shell, which leads to a stronger confinement and higher kinetic energy of carriers in regions with the lowest potential, and a gradual lowering and smoothing of potential barriers, which delocalises the wave function and averages the core and shell potentials. This is confirmed by the observation of a maximal red-shift of the band edge exciton and coinciding minimal non-resonant Stokes shift when the contribution of the confinement and potential energy term are similar. At higher temperatures the Stokes shift increases due to a high degree of alloy formation, as has been previously observed for homogeneously alloyed nanocrystals [40].

Fig. 4:

(a) TA spectra of ZnSe/CdS nanocrystals with shells grown at 270 °C (top) and 300 °C (bottom). (b) The ratio of the integrated bleaches plotted against reaction temperature. Reprinted with kind permission from Ref. [28]. Copyright 2014 American Chemical Society.

While phase mixing most often leads to an averaging of parameters such as effective mass and band edge energy it can also induce formation of additional potential barriers. In ZnSe/CdS interfacial alloying with a fast cation diffusion and a relatively slower anion diffusion leads to a region at the core with a composition close to Zn_xCd_1−xSe and an inner shell composed of Zn_1−xCd_xS. As the alloyed materials have band edges between those of the pure, binary compounds a potential trough at the valence band edge is formed in the sulfur-rich region that further confines the 1S hole to the core [28]. This effect can be enhanced by band bowing, which causes the band edges to change non-linear with composition x [41], [42], [43]. The effect is more pronounced when the gradient involves two anions.

The previous two Sections demonstrated the influences of graded shells on the steady state spectra of core/shell nanocrystals, most importantly engineering of exciton localisation and overlap and maximising PL QY. The latter has a dramatic effect on QD blinking, and nanocrystals that show no observable off periods have been reported [25], [26], [27]. The suppression of fluorescence intermittency is tied to the influence of the shape of the confinement potential on fast Auger-like processes.

4 Effects of graded confinement profiles on Auger relaxation

Auger processes play a limited role in bulk semiconductors due to a temperature threshold proportional to the band gap energy. However, they strongly reduce exciton lifetime in quantum confined systems with a rate constant that scales with volume. This can be understood based on confinement arguments: starting from a positive or negative trion state all carriers have cooled to the band gap, close to the centre of the first Brillouin zone with a small momentum component. Auger excitation causes a transition to a hot state with a large momentum far away from the zone centre. The transition is therefore momentum forbidden and slow in the bulk. The case is different for systems with spatially confined carriers. Here the wave functions require Fourier components with a higher momentum to allow for exponential drop off at the particle surface. These components relieve the restrictions on Auger transitions and make the process highly competitive for increasingly smaller nanocrystals [44], [45].

Cragg and Efros have postulated that Auger rates can be suppressed by changing the heterointerface in core/shell nanocrystals from an abrupt, hard to a graded, soft interface [46]. Auger efficiency is defined by the electronic transition matrix element M_if between the initial state Ψ_i and the final state Ψ_f.

Here we will discuss this for a positive trion state X⁺, although the same arguments apply for negative trions as well. Ψ_i is composed of two holes ψh0 with antiparallel spin occupying the valence band ground state. In the final state one hole is transferred to the conduction band by Coulomb interactions and recombines with the electron. It can be described by the complex conjugate of the electron wave function ψe∗. The second hole gets promoted to a higher valence band level ψhf.V(x₁−x₂) is the Coulomb repulsion between the two holes. Auger rates can be estimated following Fermi’s Golden Rule to enforce energy conservation:

δ(x) is the Dirac delta function, E_i and E_f the energy of the initial and final state, and ℛ_f the full set of variables defining the final state. The chief difference between the ground state and excited state holes ψh0 and ψhf is the large momentum ħk_f that the hole acquires from the Auger process. The magnitude of M_if is therefore proportional to the Fourier component of the ground state with the same momentum as the final state, ψ˜h0(kf). A soft interface drastically attenuates the high frequency components of ψh0. The calculations assume a potential

with 2a the effective width of the confinement potential and U₀ the potential height. ν is a parameter that controls the slope of the potential gradient with ν=∞ for a sharp step and ν=2 for a parabolic potential. It was shown that for 2a=5 nm the Auger rate changed by 3 orders of magnitude between the extremes for ν. Hence these calculations predict that the efficiency of Auger processes depends intrinsically on the shape of the potential well independently of the quantity of trap states at the interface.

Beane et al. argue that the the concentration of trapped holes does play an important role for Auger relaxation, because, based on the Uncertainty Principle, carriers that are strongly localised at trap states need to have a wider spectrum and larger contribution of Fourier components with large momentum. A trapped hole has therefore both a larger overlap with ψhf and a significant overlap with the valence band hole to be efficiently promoted to higher levels by the Auger process [45]. They have observed two characteristic biexciton lifetimes in strained CdSe/ZnSe particles with hard interfaces, τ_slow≈85 ps and τ_fast≈8 ps. After annealing the particles to form a graded shell both decay times were still observed, but the fast-to-slow amplitude ratio that was strongly reduced by a factor of 4–8, depending on particle size. This was attributed to strain-induced trap states, which get healed out during annealing, thereby mitigating the trap-mediated Auger pathway (see Figure 5).

Fig. 5:

Biexciton decay kinetics from CdSe/ZnSe nanocrystals with hard (a) and soft (b) interfaces. The red curves are a biexponential fit with 8 and 85 ps decay components. The slow component contributes with 11.5% and 51% for the hard and soft interfaces, respectively. (c) shows a schematic of the spatial (top) and momentum (bottom) distribution of core and trapped hole wave functions. R is the core radius, p* the momentum of the hot hole state after Auger excitation. Reprinted with kind permission from Ref. [45]. Copyright 2016 American Chemical Society.

García-Santamaria et al. and Bae et al. have observed two fast PL decay components in “giant” shell CdSe/CdS nanocrystals in the order of τ₁≈0.4 ns and τ₂≈5 ns, compared to the band edge decay with τ₁≈40 ns [47], [48]. The components were attributed to Auger relaxation of the biexciton XX (τ₁) and the trion X^* (τ₁). These values were similar for CdSe/CdSe_xS_1−x/CdS particles with the same core size and total diameter, but with an inner shell that was intentionally grown as an alloy, providing an intermediate potential step. These structures show strongly asymmetric behaviour in that the sharpness of the interface affects primarily the hole, while the electron is delocalised over the whole quasi-type-II structure. It was shown experimentally that alloying only affected the lifetime of biexcitons, which decay via Auger excitation of the second hole into the continuum and leave a negative trion, while single exciton and negative trion lifetime and QY was unaffected [48]. Keene et al. also confirm this model for graded CdSe_xS_1−x nanocrystals, which show a complete suppression of hole trapping (τ≈3 ps) for a sulfur content >34%, while electron trapping (τ≈20–30 ps) was unaffected [34]. They therefore validate the quasi-type-II alignment for CdSe/CdS interfaces in nanocrystals as well as the holes being the major contribution to Auger-mediated fluorescence quenching in II–VI semiconductor QDs.

These findings demonstrate that graded shells severely impact the dynamics of multiexcitons and charged QDs, while linear optics in the single exciton regime are only affected indirectly through charge trapping (which leads to trion states) and relief of lattice strain. In the following Section we will discuss the implications of graded shells on applications involving multiexcitons, such as quantum dot lasing and multi-exciton generation.

5 Implications on multi-exciton dynamics

Quantum dots constitute a 3-level system for stimulated emission in which the single exciton Stokes shift δ_X separates the level that is pumped (single exciton absorption) and the one from which stimulated emission occurs (single exciton emission). A long held assumption has been that single exciton gain is not possible in simple, type-I QDs, because reabsorption of emitted photons by already excited QDs causes formation of biexcitons. These relax rapidly and non-radiatively via Auger processes and compete with stimulated emission. This is made possible by the positive biexciton binding energy in type-I QDs that shifts the biexciton absorption below that of the single exciton [49]. The first successful demonstration of single exciton gain was demonstrated by Klimov’s group for type-II QDs, which have negative biexciton binding energy caused by the charge separation and repulsion of like carriers confined to the same volume. If the absorption of a second exciton is shifted above the level of the single exciton by more than the emission line width the gain threshold can be lowered a theoretical limit of 〈N〉=2/3 excitons per QD in the ensemble [50], [51].

Kambhampati and co-workers argue that QD lasing is limited by a high gain threshold, not gain lifetime, in CdSe nanocrystals. They have shown for CdSe/Cd_xZn_1−x S graded shell particles that biexciton binding energy was significantly lower compared to CdSe particles, 3 meV, while the single and biexciton Stokes shift remains similar in both systems. This excitonic structure causes no overlap between single exciton emission and biexciton absorption and hence allows for single exciton gain [52]. The CdSe/Cd_xZn_1−xS particles exhibit a strongly increased biexciton decay time of τ_XX=411 ps compared to τ_XX<60 ps for CdSe/ZnS or bare CdSe particles. This and the lack of red-shifted emission from surface or interface states at low temperatures points toward the lack of trap states in the presence of a soft interface. The small biexciton binding energy was therefore attributed to a less localised interface state and therefore reduced coupling with higher states, as discussed in Section 4. State-resolved nonlinear absorption spectra OD_NL=OD₀+ΔOD on pumping the band edge state reveal P symmetry for higher excitonic states, which are decoupled from the pumped levels, illustrated by the appearance of isosbestic points between the signals of excitons with different symmetries. When pumping high energy states the spectra show an extremely large gain bandwidth of 140 nm, compared to 40 nm for particles with hard interfaces and high exciton densities up to 〈N〉=6 due to the degeneracy of P states. These findings show that graded shell nanocrystals are strong contestants for breaking the limits imposed by Auger relaxation in strongly confined QDs.

A single exciton QD laser was recently demonstrated for the first time using CdSe/CdS seeded nanorods as gain medium deposited on silica spheres with 8–40 μm diameter as whispering gallery (WSG) mode resonators [53]. This kind of nanorod is fabricated at high temperatures of 350 °C so that interfacial alloying occurs, comparable to the “giant” QDs. The resonator was coupled to a tapered, single-mode optical fibre that was used for both pumping and outcoupling of the stimulated emission. The structure showed resonant WSG laser emission at a wavelength of 626 nm and a threshold of 〈N〉=0.675 excitons per particle on average, in good agreement with the theoretically predicted value. A second laser line emerged at 592.6 nm at higher pump fluence, corresponding to the formation of biexcitons. The two lines had pump power thresholds of 68.5 and 123.3 μW, and exhibited linear and quadratic dependence on pump power, respectively.

When exciting a QD with photons of at least twice their band gap the formation of multiexcitons through hot exciton fission can lead to quantum yields >100%. This carrier multiplication (CM) or multiexciton generation (MEG) is the reverse process to Auger excitation and is facilitated in confined semiconductors due to the volume scaling of Auger processes discussed above. However, the same Auger mechanism limits the lifetime of the multiexciton. Graded shell nanocrystals offer an elegant way to tune the electronic structure for an efficient extraction and use of MEG: they allow to effectively decouple the excitons from surface trap states and ligands, which dissipate the energy and allow non-radiative charge carrier cooling. At the same time interfacial states couple to hot excitons to facilitate exciton fission.

High MEG yields were observed for CdS tetrapods using InP/ZnS graded core/shell particles as seeds. InP and CdS have type-II band alignment with a small conduction band offset, while ZnS acts as a potential barrier for both electrons and holes [54]. As can be expected from the geometry the tetrapods exhibited absorption spectra dominated by CdS. A broad photoluminescence peak appeared at 700 nm, red-shifted by ~200 nm compared to the InP/ZnS seeds, with a quantum yield of ~20% and an average fluorescence lifetime of 545 ns. Dual emission from the indirect InP/CdS transition and from the CdS arms was observed at high pump fluences. CdS emission intensity exceeded that of the indirect transition at pump fluences above 100 μJ/cm² (〈N〉≈10 excitons per particle), a phenomenon that has not been observed with the better studied CdSe/CdS. Additionally a blue-shift of the indirect emission occurs with increasing pump fluence. The continuous shift cannot be explained by biexciton emission, which would appear as an additional, distinct peak. Rather, the shift was attributed to the Stark effect caused by the separated charges. These create an electric field along the arms of the tetrapod, which causes band bending at the interface. Ultrafast transient absorption spectra reveal that while at low pump fluences (〈N〉≈1) there is a fast decay component from the CdS arms to the type-II exciton no such decay can be observed at high pump intensities (〈N〉≈10). This as well is in contrast to CdSe/CdS tetrapods, which show a pronounced relaxation from CdS- to CdSe-based excitons at all excitation intensities. The low saturation threshold is achieved by the type-II structure, causing a steady state of exciton saturation in the InP/ZnS/CdS core region, combined with improved radiative recombination of excitons and mitigated Auger relaxation at the alloyed layer.

6 Conclusion

In conclusion it can be said that the recent years have seen a strong focus on the synthesis and photophysics of core/shell quantum dots with graded or alloyed shells. High temperature shell growth or annealing produces interfacial layers in which a random distribution of two cations or anions occurs over the distance larger than one unit cell. These particles have become known as the “next generation” QDs due to their superior fluorescence quantum yield, low lattice strain, and high performance in applications such as photostable fluorescent dyes, single exciton optical gain, and carrier multiplication. The underlaying mechanism is the suppression of fast Auger relaxation of multiexcitons and excitons in charged particles, caused by the delocalisation of interface states in the alloyed layer. Additionally, interdiffusion of core and shell can be used to engineer band structures that are difficult to obtain in classical core/shell particles, e.g. blue and green emitting nanocrystals with high PL QY and potential barriers with controlled height and width. More work is undoubtedly necessary to elucidate the atomistic processes that lead to the formation of a gradient, cation and anion diffusion, and to precisely control the slope of the material gradient.