Letter pubs.acs.org/NanoLett

Electrochemical Control of Two-Color Emission from Colloidal Dotin-Bulk Nanocrystals Sergio Brovelli,*,† Wan Ki Bae,‡ Francesco Meinardi,† Beatriz Santiago González,† Monica Lorenzon,† Christophe Galland,§ and Victor I. Klimov*,‡ †

Dipartimento di Scienza dei Materiali, Università degli Studi di Milano-Bicocca, via Cozzi 55, I-20125 Milano, Italy Chemistry Division and Center for Advanced Solar Photophysics, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, United States § ́ Ecole Polytechnique Fédérale de Lausanne (EPFL), 1015 CH Lausanne, Switzerland ‡

S Supporting Information *

ABSTRACT: Colloidal “dot-in-bulk” nanocrystals (DiB NCs) consist of a quantum confined core embedded into a bulklike shell of a larger energy gap. The first reported example of this class of nanostructures are CdSe/CdS DiB NCs that are capable of producing tunable two-color emission under both weak continuous-wave optical excitation and electrical charge injection. This property is a consequence of a Coulomb blockade mechanism, which slows down dramatically intraband relaxation of shell-localized holes when the core is already occupied by a hole. Here, we demonstrate electrochemical control of dual emission from DiB NCs. Spectroelectrochemical (SEC) experiments are used to tune and probe the photoluminescence (PL) intensity and branching between the core and the shell emission channels as a function of applied electrochemical potential (VEC). To interpret the SEC data we develop a model that describes the changes in the intensities of the shell and core PL bands by relating them to the occupancies of electron and hole traps. Specifically, application of negative electrochemical potentials under which the Fermi level is shifted upward in energy leads to passivation of electron traps at the surface of the CdS shell thereby increasing the total PL quantum yield by favoring the shell emission. Simultaneously, the emission color changes from red (VEC = 0) through yellow to green (VEC = −1). Time-resolved PL measurements indicate that as the Fermi level approaches the NC conduction band-edge electrons are injected into the NC quantized states, which leads to typical signatures of negative trions observed under optical excitation. Application of positive potentials leads to activation of electron traps, which quenches both core and shell PL and leads to the reduction of the overall PL quantum efficiency. A high sensitivity of emission intensity (especially pronounced for the shell band) and the apparent emission color of DiB NCs to local electrochemical environment can enable interesting applications of these novel nanostructures in areas of imaging and sensing including, for example, ratiometric probing of intracellular pH. KEYWORDS: Nanocrystal quantum dot, core/shell heterostructure, dual emission, spectro-electrochemistry, trapping, ratiometric sensing referred to as “giant” quantum dots or g-QDs) have received particular attention due to suppressed photoluminescence (PL) intermittency15−18 and reduced rates of Auger recombination,19−21 which is critical for practical realization of efficient NC lasers3 and LEDs.22 The newest member of this family is a dot-in-bulk (DiB) CdSe/CdS heterostructure in which a small quantum-confined CdSe core is embedded within a very large bulklike CdS shell with the overall size of ≥20 nm.23,24 A distinct feature of DiB NCs is that they can exhibit two-color emission due to core and shell states under weak continuous wave (cw) optical

C

olloidal semiconductor nanocrystals (NCs) are solutionprocessable functional materials that have been explored for a range of applications from light-emitting diodes (LEDs),1 photovoltaic cells,2 and lasers3 to biologic markers, 4,5 luminescent solar concentrators,6 and single photon sources.7 Electronic and optical properties of NCs can be tuned through particle-size control (quantum confinement effect) and/or socalled “wave function engineering” when the spatial distribution of electron and hole wave functions within a heterostructured NCs is controlled by introducing appropriate energy gradients.8,9 A combination of CdSe and CdS has been used to demonstrate several types of heterostructures including spherical core/shell10 and axial dot-in-rod11,12 NCs, tetrapods,13 and dot-in-plate14 nanostructures. NCs comprising a small CdSe core overcoated with a thick CdS shell (often © XXXX American Chemical Society

Received: March 19, 2014 Revised: June 2, 2014

A

dx.doi.org/10.1021/nl501026r | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

excitation24 and even electrical injection.23 This property has been attributed to the existence of a thin interfacial barrier, which impedes hole relaxation from the shell to the core states.23 DiB NCs have a polytypic CdS shell consisting of a thin zinc-blende (ZB) interfacial layer directly adjacent to the ZB CdSe core, which is followed by a thick wurtzite (WZ) layer.24 Because of the difference in the valence-band-edge energies between ZB and WZ CdS,25,26 the ZB layer introduces a potential barrier between the shell and the core regions that reduces the hole capture rate from the shell to the core24 and increases the lifetime of shell localized holes to ∼20 ps (Figure 1a).24 In addition, a dynamic Coulomb blockade mechanism, which “turns on” after the core traps the first hole, further reduces the hole capture rate, limiting the core occupancy to one hole. This results in suppression of Auger recombination

and the development of strong photon antibunching observed for core emission in single-NC measurements.24 The ability of DiB NCs to simultaneously emit red and green light can be utilized to tune the effective emission color by simply changing the branching ratio between the core and shell emission channels without modifying the nanostructure itself. As was demonstrated in refs 23 and 24, this branching ratio can be readily controlled by varying the intensity of optical excitation in the case of PL or driving current in the case of electroluminescence (EL). Further, the EL studies of ref 23 have suggested that the relative intensities of core- and shellemission bands are also influenced by modifications in the rates of relaxation associated with surface/interfacial defects sites. Here, we apply spectro-electrochemical (SEC) methods to DiB NCs in order to elucidate the effect of various trapping mechanisms on core- and shell-emission channels and further demonstrate the regime of color tunability through controlled activation/passivation of trap sites. By raising/lowering the Fermi energy we can modify occupancies of electron and hole traps (and hence their “trapping activity”)27 and thereby reversibly tune the relative intensities of core- and shell-related bands that results in effective emission color tunable from red to green. Using selective excitation of either core or shell states we observe that core emission is only weakly affected by defects at the surface of the CdS shell21 whereas shell emission is highly sensitive to changes in the activity of surface traps. The data further indicate the existence of defects at the core/shell interface that act as traps for core-localized holes without a significant effect on electrons. The CdSe/CdS DiB NCs used in the present study have been synthesized using a fast shell growth technique introduced in refs 23 and 28. In contrast to a traditional method based on successive ionic layer adsorption and reaction (SILAR), the new technique allows for a much faster deposition of a CdS shell, which has enabled exploration of the regime of extremely thick shells up to 9 nm.23,24 In these structures, the core excitations are still strongly quantized while the shell can be described in bulklike terms as its size is greater than the Bohr exciton diameter in CdS (∼6 nm). These properties of the novel nanostructures motivated us to term them as dot-in-bulk or DiB NCs.23,24 Some of the electronic behaviors of DiB NCs are similar to those of more traditional CdSe/CdS g-QDs20,29 or dot-in-rod NCs.11,30,31 Specifically, the CdSe/CdS interface is characterized by a large energy offset in the valence band while the energy offset in the conduction band is small. As a result, CdSe/CdS heterostructures are typically considered as quasitype II systems where the hole wave function is tightly confined within the CdSe core,11,32 while the electron wave function is delocalized over the entire NC volume. In both g-QDs and dotin-rod NCs, this band alignment leads to ultrafast relaxation31 (250 fs) of photogenerated holes from the CdS domain into the CdSe core. As a result, independent of whether a photon is absorbed by the CdS or CdSe part of the nanostructure, emission normally occurs from the CdSe core states. In order to observe shell emission from these NCs, one has to first saturate the core states, which however is not straightforward as when multiple holes are generated within the core, the carrier decay becomes dominated by nonradiative Auger recombination. Despite a large overall size of these nanostructures, multiexciton Auger decay is still very fast because close spatial separation between holes (they are confined within a small core) leads to large rates of a “positive-trion” Auger pathway

Figure 1. (a) CdSe/CdS DiB NCs feature an interfacial ZB CdS layer that separates a ZB CdSe core from a WZ CdS shell. (b) Schematics of a DiB NC that show a donor-like core exciton with radius Rd and binding energy εd and a bulklike shell-localized exciton. (c) PL (solid line) and absorption (dashed line) spectra of CdSe/CdS DiB NCs with R0 = 1.5 nm and H = 8.5 nm excited with a cw laser at 3.1 eV; excitation flux density 100 W/cm2. Core and shell PL bands are highlighted by red and green shading, respectively. (d) Left panel: Time transient for core emission (red line) (0.4 μJ/cm2, 3.1 eV excitation pulses; 800 ps times resolution). The decay time of core PL is about 240 ns. Right panel: Shell PL decay recorded with a streak camera using 3.1 eV excitation (green lines; 150 μJ/cm2; 10 ps time resolution). At a low fluence (0.7 μJ/cm2), the dynamics is dominated by a fast (19 ps) component followed by a slower (∼280 ps) decay. As the excitation fluence is increased up to 30 μJ/cm2, the relative contribution of the slow component increases and the measured decay (green dashed line) approaches that for core-only CdS NCs of 10 nm radius (black line). B

dx.doi.org/10.1021/nl501026r | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 2. (a) A series of PL spectra (0.5 s acquisition time per frame, 10 s steps) under excitation with photon energies hvexc = 3.1 eV (upper panel) or 2.35 eV (lower panel) of a submonolayer film of DiB NCs (R0 = 1.5 nm, H = 8.5 nm) deposited onto a ZnO-nanoparticle-covered layer of indium tin oxide (ITO) for a stepwise scan of the electrochemical potential to negative values. (b) Top: The total PL intensity (black circles; hvexc = 3.1 eV) as a function of VEC. Middle: The core PL intensity for hvexc = 3.1 eV (red circles; right axis) or 2.35 eV (pink triangles; right axis) as well as the shell PL intensity for hvexc = 3.1 eV (green circles; left axis) as a function of applied negative electrochemical potential; all intensities are normalized to their values at VEC = 0 V. Bottom: Photographs of the electrochemical cell excited at 3.1 eV at VEC = 0, −0.5, −1 V showing the change of emission color from red to yellow and finally green. (c) Electrochemical potential-dependent emission from the DiB NCs projected onto the Commission Internationale del’Éclairage (CIE) chromaticity diagram based on PL spectra shown in panel a for VEC ranging from 0 to −1 V. (d) Time transients of core PL excited at 3.1 eV at a zero electrochemical potential (black) and at VEC = −1 V (red).

which is similar to that of an exciton bound to a neutral donor observed in bulk CdS.37 Because of its relatively small binding energy (∼8.3 meV, ref 37), it is unstable at room temperature. Therefore, the second exciton in DiB structures is delocalized over the entire volume of the CdS shell (Figure 1b). We will refer to it as a “shell exciton.” Figure 1c reports the typical optical absorption (dashed line) and PL (solid line) spectra of DiB NCs with core radius R0 = 1.5 nm and shell thickness H = 8.5 nm under cw excitation at 3.1 eV with 100 W/cm2 flux density. The PL peak at 1.98 eV (red shading in Figure 1c) is due to recombination of core excitons occurring with decay rate Rco = 1/240 ns−1 (Figure 1d; left panel). The PL band at 2.4 eV (green shading in Figure 1c) is due to radiative decay of shell excitons.24 At low excitation fluence (solid green curve in Figure 1d; right panel), its decay is dominated by hole relaxation into the core states; the corresponding rate (rh) is 1/19 ps−1. As the excitation intensity is increased, the shell PL decay becomes slower due to progressive filling of core states (dashed green line in Figure 1d; right panel) and eventually becomes almost identical to that of a macroscopic bulk CdS crystal (black line in Figure 1d; right panel).24 On the basis of this high-pump-intensity trace, the recombination rate of shell excitons is Rsh = 1/280 ps−1. Because of the difference in their localization the core and shell excitons are expected to be affected in different ways by the defects at the core/shell interface and those at the outer surface of the CdS shell (below, referred to as “interfacial” and “surface” defects/traps, respectively). Because the core exciton occupies a limited volume around the CdSe domain of the NC,

whereby the electron−hole recombination energy is transferred to the second core-localized hole.33 Therefore, in order to achieve saturation of core valence-band states, required for activation of shell emission, one must use very high excitation rates (outpacing those of Auger decay)30,34 usually obtained using intense femtosecond pulses. In contrast to other CdSe/CdS systems, DiB NCs emit twocolor light even under very weak optical or electrical excitation.23,24 As described in refs 23 and 24, this behavior can be explained by a concerted effect of two factors: the existence of an energetic barrier at the core/shell interface impeding hole relaxation to the core (Figure 1a) and dynamic Coulomb blockade that develops after the core traps the first hole. In singly excited NCs, in the case when the electron−hole pair is generated in the shell, a hole relaxes into the core and the electron becomes trapped within the hole attractive potential in the near-core area of the NC (Figure 1b).35 This Coulombically bound exciton (below, referred to a “core exciton”) is similar to a standard donor state in bulk CdS where it is characterized by the binding energy of 32.7 meV36 and the effective radius of ∼2.6 nm. The latter value is in agreement with previous studies on giant NCs19,32 that showed saturation of the effective exciton radius at ∼2.9 nm occurring for shell thickness greater than 4 nm. The same donor-like exciton likely also exists in doubly excited DiB NCs. However, the electronic structure of the second electron−hole pair is different. Because of the dynamic Coulomb blockade,24 the second hole does not get captured by the core and together with the second electron it forms a state, C

dx.doi.org/10.1021/nl501026r | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 3. (a) A series of PL spectra (0.5 s acquisition per frame 10 s steps) recorded using excitation at 3.1 eV (top panel) and 2.35 eV (bottom panel) of DiB NCs (same sample as in Figure 2) during a stepwise scan of the electrochemical potential to positive values. (b) Top: The total PL intensity for excitation at 3.1 eV (black circles) as a function of VEC. Bottom: The core PL intensity for excitation with hvexc = 3.1 eV (red circles) or 2.35 eV (pink triangles) as well as the shell PL intensity (green circles; hvexc = 3.1 eV) as a function of applied positive electrochemical potential; all intensities are normalized to their values at VEC = 0 V. (c) Intensities of the core and shell PL bands (red and green, respectively) and the total PL intensity (black) for excitation at 3.1 eV during ON/OFF voltage cycles for the negative electrochemical potential (VEC = 0,−1 V). (d) Same for the positive electrochemical potential (VEC = 0,+1 V).

which is far away from the CdS surface, it recombines primarily via interfacial defects. On the other hand, the shell exciton is delocalized over the entire NC and, in principle, can recombine via both types of trap sites. However, because the surface area of the CdS shell is much larger than that of the CdSe/CdS interface, it is mostly affected by the surface defects. As was demonstrated previously,27 by using the means of SEC one can control the position of the Fermi level and thus the occupancy of trap sites, which in turn affects the PL by modifying the rate of nonradiative carrier losses. Here, we apply this method to study the influence of modulation of the occupancy of surface/interfacial traps on branching between radiative channels responsible for core and shell emission in DiB NCs. We start by monitoring the evolution of the PL in a cyclic stepwise scan resulting from changing the electrochemical potential (VEC) from 0 to −1 V (this corresponds to raising the Fermi level) and back to 0 V. Figure 2a displays a set of PL spectra for increasing negative potentials (0.5 s accumulation time, 0.25 V steps, 10 s per step) recorded using cw optical pumping at 3.1 eV (top panel) or 2.35 eV (lower panel). In Figure 2b, we plot the amplitudes of the shelland core-related PL bands together with the total PL signal (upper trace) derived from the recorded spectra. In the measurements with the 3.1 eV excitation, the flux density was 100 W/cm2 which corresponded to the average steady state NC occupancy (⟨N⟩) of ∼1 electron−hole pair per NC. Owing to a much larger absorption cross-section of the shell with respect to the core, 3.1 eV excitation generates carriers primarily in the shell region. In singly excited NCs, a

photogenerated hole quickly relaxes into the core, therefore, their PL is dominated by core excitons emitting in the red.24 For ⟨N⟩ of ∼ 1, a significant fraction of the NCs (∼25%) is excited with multiple electron−hole pairs. As discussed above, such NCs in addition to a core exciton contain also one or more shell excitons. As a result, the emission detected from the ensemble of DiB NCs exhibits both core- and shell-related PL bands. At VEC = 0 V, the ratio of the amplitudes of the core and shell emission bands is 3:1, which results in the effective yellow color. As the electrochemical potential is tuned to −1 V, we observe an initially weak but progressively greater brightening of the shell PL accompanied by the reduction of the core band amplitude (Figure 2b; middle panel). At VEC = −0.75 to −1 V, the shell PL is increased almost 4-fold that is accompanied by about 40% drop of the core PL. As a result of these changes, the total, spectrally integrated PL intensity grows by 30% (Figure 2b; top panel) and the overall emission color switches to green (Figure 2c). When returning back to a zero potential, we observe essentially hysteresis-free evolution of the PL peaks to their original intensities, indicating that potential sweeps to negative values do not cause any permanent degradation of the NCs. The behavior of the shell PL under increasing reductive potentials can be explained in terms of the deactivation of surface electron traps as a result of raising the Fermi level, which leads to progressive filling of intragap defect states. The role of changes in the hole trapping channel in this case is less significant, suggesting that a coefficient of capture at surface D

dx.doi.org/10.1021/nl501026r | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 4. (a) A diagram illustrating the model used to describe the effect of the electrochemical potential on core and shell emission bands via filling/emptying of trap bands at the surface of the CdS shell and the core/shell interface in response to changes in the Fermi level (FL; the dashed purple line for VEC = 0 and the solid purple line for VEC < 0). Filling defect bands with electrons (raising FL) suppresses electron trapping but enhances hole trapping; depleting defect bands (lowering FL) leads to the opposite effects. The band of trap states at the NC surface, which affects primarily shell excitons, is shown on the right of the NC and highlighted in green. The core PL is instead primarily affected by interfacial defects shown on the left of the NC and highlighted in red. The capture coefficients for core (shell) electrons and holes are denoted, respectively, as αC(S) and βC(S). (b) The calculated total PL intensity as a function of electrochemical potential using parameters listed in Table 1. (c) Calculated core (red) and shell (green) PL intensities normalized to their values at VEC = 0 for a varying negative electrochemical potentials. (d) Same for the positive electrochemical potentials.

transients measured for VEC = 0 and −1 V. We observe that the early time PL amplitude (time t = 0) at −1 V is higher with respect to that at the zero potential. This is consistent with the difference anticipated for charged versus neutral excitons as the former are characterized by a higher emission rate that ideally is twice that of a neutral exciton.38 We also observe that application of the −1 V potential leads to the reduction of the PL lifetime, again as expected for the enhanced radiative decay, and perhaps, increased contribution from hole trapping.19,38 For standard core-only or thin-shell NCs, the acceleration of the PL decay in charged NCs is primarily due to activation of nonradiative Auger decay;38 in DiB NCs, the rate of Auger recombination for negative trions is greatly reduced due to a large effective volume sampled by photoexcited electrons. The above observations confirm the presence of an extra electron in the NC under large reductive potentials and suggest that the unperturbed Fermi level in these structures is located approximately 1 eV below the conduction band edge, that is, close to the center of the bandgap, as expected for intrinsic (that is, undoped) NCs. Next, we analyze the effect of a positive electrochemical potential (Figure 3a,b). In this case, increasing VEC (that is, lowering the Fermi level) is supposed to activate electron traps but suppress hole trapping. Because shell PL is mostly affected

defects for holes is smaller than for electrons (see quantitative modeling in the next section). A much less dramatic effect of the electrochemical potential on core emission suggests the CdSe/CdS interface provides a much smaller contribution to nonradiative losses compared to the CdS shell. Some drop in the core emission intensity, which is still observed under negative potentials, is likely due to activation of hole trapping at both the shell surface (affects core PL by competing with hole capture by the core) and the CdSe/CdS interface (competes with radiative recombination of core excitons). The first of these processes can be eliminated by injecting holes directly into the core, which should reduce the detrimental effect of traps at the CdS shell surface. Indeed, in the case 2.35 eV excitation, which is below the energy gap of the CdS shell, the drop in the shell PL intensity at VEC = −1 V is decreased to 15% (middle panel of Figure 2b; triangles). As was mentioned earlier, at sufficiently high negative values of VEC when the Fermi level is raised above the conduction band edge of the CdS shell it should be possible to directly inject electrons into intrinsic quantized states of the NC. Photoexcitation of negatively charged NCs results in the formation of negative trions that are characterized by specific spectroscopic signatures.38,39 These signatures are indeed detected at VEC = −1 V. Figure 2d compares core PL time E

dx.doi.org/10.1021/nl501026r | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

weak effect of surface trapping on core PL associated with electron and/or hole losses in the early stage following their injection into the shell before they form a core exciton. We also neglect the effect of the electrochemical potential on branching of initial photoexcitations between the core and the shell emission channels and assume that these channels are independent. Below, we provide a description of the model for core emission; a similar description applies to shell PL. We assume that the defect states at the core/shell interface are distributed within continuous bands (defect bands), and the states above the Fermi level serve as electron traps while those below it as hole traps (Figure 4a). In our experiments, because the control of emission is achieved by applying the electrochemical potential, we introduce a proportionality constant k ([1/V]) which allows us to express the relative trap occupancies in the units of voltage. We thus denote the effective width of the defect band as Ñ 0 (Ñ 0 = N0/k; where N0 is the number of traps per individual NC) and assume that at a zero electrochemical potential, the Fermi level is located at the center of the band gap of the CdSe core; this defines the initial “dark” occupancy of the defect band (ñ0 = n0/k; where n0 is the number of occupied traps per individual NC) in the absence of photoexcitation. For simplicity, we consider the regime of continuous wave excitation. Steady-state per-NC occupancies of intrinsic core electron and hole states under photoexcitation are n and p, respectively, and both are assumed to be much smaller than N0. In addition to populating NC quantized states, photoexcitation also changes the occupancy of the defect band compared to its equilibrium value. We denote the effective filling of the defect band due to photoexcitation by ñd (ñd = nd/k where nd is the number of trapped electrons per individual NC following photoexcitation), which is related to the occupancy of quantized states by the condition of charge neutrality, n + kñd = p. To account for the effect of the electrochemical potential, we introduce an addition to the equilibrium (“dark”) occupancy of trap states expressed as −VEC, where the sign “minus” accounts for the fact that negative potentials correspond to injection of electrons into the defect band, while positive to their withdrawal. With this additional term, the total occupancy of the defect band in the presence of photoexcitation is given by ñt =ñ0 + ñd − VEC. Because of a finite width of the defect band, ñt varies between 0 and Ñ 0. Specifically, ñt = Ñ 0 if −VEC > Ñ 0 − (ñ0 + ñd) and ñt = 0 if VEC > ñ0 + ñd. The VEC-dependent effective trapping rates of band-edge carriers into the defect band (Re and Rh for electrons and holes, respectively) are determined by their total occupancies and can be presented as Re(VEC) = α·k·n[Ñ 0 − (ñ0 + ñd)+ VEC] and Rh(VEC) = β·k·n(ñ0 + ñd − VEC), where α and β are electron and hole capture coefficients. While recent studies of charge transfer from the NCs to molecular acceptors show that a fairly strong dependence of transfer rates on a driving force,40 here for simplicity we assume that the trapping coefficients are independent of the specific energy of the trap states, that is, the depth of the trap within the energy gap. As a result, in this model both electron and hole capture processes depend only on the total occupancy of the respective trap band. Because of fairly low PL quantum yields of our samples (

Electrochemical control of two-color emission from colloidal dot-in-bulk nanocrystals.

Colloidal "dot-in-bulk" nanocrystals (DiB NCs) consist of a quantum confined core embedded into a bulklike shell of a larger energy gap. The first rep...
3MB Sizes 0 Downloads 4 Views