Understanding Temporal Evolution of Electroluminescence Intensity in Lead Sulfide (PbS) Colloidal Quantum Dot Infrared Light-Emitting Diodes

Abstract

We, for the first time, report a temporal evolution of the electroluminescence (EL) intensity in lead sulfide (PbS) colloidal quantum dot (CQD) infrared light-emitting diodes. The EL intensity was varied during infrared light emission, and its origin is attributed to competition between the achievement of charge balance associated with interfacial charging at the PbS/ZnO CQD interface and the electric-field induced luminescence quenching. The effect of multi-carrier emission on the enhanced EL intensity is discussed relating to shifting in the wavelength at the peak EL intensity.

1. Introduction

Colloidal quantum dots (CQDs) have been assembled into a variety of devices, including light-emitting diodes (LEDs), solar cells, and biomedical sensors [1,2,3,4,5]. Excellent spectral tunability induced by quantum confinement effect associated with adjustable optical band gaps, simple solution process, and highly efficient photoluminescence (PL) quantum yields have motivated extensive research towards quantum dot light-emitting diodes (QLEDs), emerging as a next-generation display technology which can compete with organic light emitting diodes [5,6,7,8,9].

Understanding of the underlying mechanism for efficient emission has been one of the leading research themes, revealing that suppression of electric-field induced electrostatic interactions, including quantum dot (QD) charging and electric-field assisted exciton dissociation is crucial in reducing the external quantum efficiency (EQE) roll-off due to non-radiative loss, enabling integration of highly efficient QLEDs [10,11,12].

The main issue in structuring highly efficient QLEDs is to confine electrons and holes to the QD emission layer, suppressing the non-radiative recombination process [13,14,15]. Charge injection in the QD emission layer can lead to an unbalanced charge state that causes the non-radiative Auger process. QDs with a high surface to volume ratio are favorable for the formation of trap sites which facilitates the non-radiative process, quenching luminescence. To reduce the surface states, core QDs were shelled by inorganic or organic layers, and a variety of surface ligand chemistries were applied for surface passivation [16,17,18].

To improve the EQE, researchers have carried out structural engineering of hetero-structured QDs with adjustment of the shell thickness and the composition of the core/shell structure [19,20]. To prevent the non-radiative Auger recombination process due to QD charging, an interfacial alloy layer was introduced between the core and the shell QDs, mitigating the steepness of the confinement potential [19]. Bae et al. demonstrated an enhanced EQE by the insertion of an intermediate layer (CdSe_xS_1−x) between CdSe QD layers, suppressing the Auger decay process [12]. The conduction band edge was engineered for a higher energy level to reduce electron injection, improving charge balance within the QDs. Various strategies including QD core/shell alloying, the formation of giant QD, and the adjustment of QD stoichiometries have successfully led to significant improvement in the EQE [21,22,23].

Electric field-induced luminescence quenching arising from QD charging and spatial separation of electrons and holes has also been considered one of the primary sources of the low EQE in QD LEDs [10,11]. Therefore, understanding the mechanism of electric field-induced luminescence quenching is crucial in enhancing the EQE. Particularly, distinguishing between field-induced charging and field-induced luminescence quenching has remained elusive because the two effects occur simultaneously during the emission process, complicating the separation of the isolated effect on the EQE loss. Bozyigit et al. reported the fabrication of a test platform device in which multi-layers of QDs are sandwiched by SiO2 dielectric films [10]. This test structure enabled a fundamental study on the effect of the electric field on the EQE, excluding the effect of QD charging due to charge injection. They revealed that electric field-induced spatial separation of electrons and holes still occurs with the presence of the relaxed confinement potentials in the core/shell structures designed to extend either electrons or holes into the shell. Importantly, most of the charge carrier was localized over long time scales, revealing that the primary source of luminescence quenching is electric field-induced coupling to the surface states of QDs. Retaining the overlap of electron and hole wave functions at a reduced electric field is, therefore, a prerequisite in enhancing the EQE over structural engineering of the hetero-structured QDs.

As mentioned above, many efforts have been made to understand the electric field-induced luminescence quenching as well as QD charging. Surface states coupled luminescence quenching is expected to be more severe at the interface with adjacent functional layers such as carrier transport and blocking layers. Therefore, interfacial charging related to the surface states and/or the energy level as well as the structural engineering within the QD emission layer can be a key parameter in determining the EQE of the QD LEDs.

Compared with the huge success of QLEDs in the visible range, as discussed above, progress in the near-infrared (NIR) CQD LEDs is not so fast. NIR CQDs such as PbS and PbSe have attracted much attention due to their applications to biomedical imaging, telecommunications, and night vision [24,25]. Contrary to CQD LEDs in the visible range, luminescence efficiency in the NIR CQD LEDs lags in the presence of self-quenching in which efficient charge injection causes transport-assisted trapping of mobile carriers within QDs [25,26]. Various strategies to suppress self-quenching in CQDs, including organic ligand capping, insertion into the polymer or hybrid perovskite matrix, and shelling of the surface have led to enhanced EQE of up to ~7.9% [26,27].

Most of all, unlike CQD LEDs in the visible range, extensive studies of interfacial charging and electric field-induced luminescence quenching associated with PbS CQDs have not been conducted. Here, we report, for the first time, the temporal evolution of the electroluminescence (EL) intensity during light emission in a PbS CQD LED testbed device. We correlated the temporal evolution of the EL intensity to electric-field-induced interfacial charging between the emission and transport layers, and luminescence quenching. To articulate the temporal evolution, we investigated the spectral shift and the magnitude of the current simultaneously. Surprisingly, to the best of our knowledge, the effect of temporal charging over long time scales during light emission at the emission/transport layer interface on the EL intensity has not been reported in NIR CQD LEDs, despite the fact that the time scale associated with electric-field induced charging can extend up to tens of seconds depending on the charge injection into QDs and carrier trapping process on the QD surface.

2. Materials and Methods

PbS CQDs were synthesized by the hot injection method [28]; 1.45 mL of oleic acid (Sigma Aldrich, 90%) and 18.5 mL of 1-octadecene (ODE, Sigma Aldrich, 90%) were added to 0.45 g of lead (II) oxide (Sigma Aldrich, 99.999%) in a three-neck flask until the solute (PbO) is completely dissolved and becomes transparent. Temperature is increased to the target reaction temperature followed by injection of a hexamethyldisilathiane (TMS, Sigma Aldrich)-ODE solution containing 110 μL of TMS into 15 mL of ODE. The three-neck flask is immersed into ice water immediately after the reaction is completed, followed by a centrifugation process. For CQD washing, we centrifuged the synthesized solution, separating unreacted material. The supernatant was transferred to new tubes and centrifuged at 9000 rpm for 10 min. The precipitate is re-dispersed in hexane. We repeated the washing process several times and the sediments were re-dispersed in chlorobenzene (10 mL) to obtain a PbS CQD (~40 mg/mL).

To fabricate PbS CQD LEDs as structured in Figure 1a, PEDOT:PSS (70 nm, M124 HTL Solar from Ossila) was spin-coated onto the patterned indium tin oxide (ITO, 70 Ω□) substrate at 6000 rpm for 60 s. PbS CQD solution in chlorobenzene (40 mg/mL) was spin-coated onto the PEDOT:PSS film at 1000 rpm for 30 s. As an electron transport layer, ZnO CQDs were synthesized. Zinc acetate dehydrate (5 mmol) in Dimethyl sulfoxide (DMSO) was stirred for 10 min at room temperature. Simultaneously, 5 mmol of a tetramethylammonium hydroxide pentahydrate (TMAH) (0.968 g) in ethanol was prepared followed by being added to the zinc acetate solution. After the injection process, the mixed solution was cooled in air. For washing process, we added acetone (35 mL) into the solution (10 mL), followed by centrifugation at 9000 rpm for 10 min. After centrifugation, the solvent was removed and ZnO CQDs were dried for 1 min. We added ethanol (10 mL) into the conical tube containing ZnO CQDs for another centrifugation. After washing, ZnO solution was spin-coated at 1000 rpm for 60 s, producing a thickness of ~60 nm followed by thermal evaporation of aluminum (Al) electrodes [5]. Using photoelectron spectroscopy in air (PESA) and optical absorption spectroscopy, the lowest unoccupied molecular orbital (LUMO) and the highest occupied molecular orbital (HOMO) energy levels were calculated.

The EL intensity of a PbS NIR QLED device was monitored at a driving voltage applied to the ITO side. With a photo-spectrometer (Hamamatsu TG-cooled series) and a semiconductor parameter analyzer (Keithley 2400), the EL intensity and the electrical current were plotted as a function of time, respectively, during light emission at a particular voltage over the turn-on voltage. Interfacial charging/discharging experiments were carried out by applying positive and negative voltages to the ITO side.

3. Results and Discussion

The size of the synthesized PbS CQDs was estimated to be ~3 nm from the transmission electron microscopy (TEM) image, as seen in Figure 1b. An energy band gap of 0.9 eV and a highest occupied molecular orbital (HOMO) energy level of 5.1 eV were determined from the optical absorption and PESA characterizations, respectively, as seen in Figure 1c,d, which is consistent with the size determined from the TEM image. During PESA measurements to measure the HOMO level of the PbS CQDs, the photocurrent was recorded in the photon energy ranging between 4.2 and 6.2 eV. We plotted the emission yield as a function of the incident photon energy to estimate the HOMO level, which is determined from the incident photon energy at which the effective emission is negligible. To prevent charging, we deposited PbS CQDs films on ITO-coated glass substrates.

The temporal evolution of the EL intensity and current during light emission was simultaneously investigated through sampling measurements in which a constant voltage (1.5 V) over the turn-on voltage is applied. Figure 2a shows plots of the EL intensity as a function of the emission wavelength. Importantly, the EL intensity significantly increased during electrical biasing at the entire emission wavelength between 1100 and 1600 nm, while the current decreased, as seen in Figure 2b. We also observed that the abrupt increase in the EL intensity immediately after application of the electric field occurred with a significant decrease in the current in Figure 2b, which is addressed later.

Simultaneous change in the magnitude of the current and the EL intensity provides a hint of the origin of the rise in the EL intensity during emission. We hypothesize that charge balance in the emission layer can be attained by electrical bias-induced modification of the interfacial electrical contacts, altering the magnitude of the injection current across the interface between PbS and ZnO CQD layers. It is well-known that non-radiative Auger recombination can be mitigated by suppressing QD charging, achieving charge balance in the emission layer [10,19].

To address the origin of the current variation during light emission, we carried out experiments in which the ITO/PEDOT:PSS/PbS/ZnO/Al device is charged and discharged by applying positive and negative voltages to the ITO side, respectively, for tens of seconds, followed by measurement of the EL intensity at a driving voltage of 10 V. Prior to the charging/discharging experiments, we confirmed that, at a high voltage over 10 V, the EL intensity increased with voltage and time (data not shown), indicating that electrical biasing at such a high voltage for the charging/discharging experiments has not damaged the device. In Figure 3a, the PbS QLED device was charged by applying a high positive (10 V) or negative voltage (−10 V) for 2 min. Notably, the measured EL intensity under application of 1.5 V, after applying a negative voltage, was far lower than that after applying a positive voltage, indicating that the effect of electrical pre-biasing on emission is significant. It is noted that reproducible results were obtained after applying smaller voltages (5 V and −5 V) for charging and discharging, respectively.

A significant difference in the current over an order of magnitude was observed depending on the polarity of the electrical pre-biasing (10 V and −10 V for 2 min), as seen in Figure 3b. This suggests that the charge injection energy level between the transport (PEDOT:PSS or ZnO) and the emission (PbS) layer was altered, depending on the polarity of the electrical pre-biasing. Indeed, a subtle change at the interfacial energy level by electrical bias stress causes an exponential difference in the magnitude of the current, assuming the electric-field enhanced thermionic emission model [29,30]. An interfacial energy level-dependent diode current was confirmed in the I-V curve of the ITO/PEDOT:PSS/PbS/ZnO/Al CQD LED below the turn-on voltage in Figure 4 in which the forward and reverse bias current was clearly observed. The formation of the forward and reverse bias regions suggests that the test structure we used consists of energetically well-defined interfaces for asymmetric carrier transfer. In the test structure with distinct electron and hole energy barriers, under the application of a positive or negative voltage to the ITO side, for example, carriers are selectively accumulated and depleted at the PEDOT:PSS/PbS and/or PbS/ZnO interfaces, modifying the interfacial energy level.

During the charging process in Figure 3b at a high voltage (10 V), electrons are injected through Al and passivate the surface trap sites on the ZnO and PbS CQDs. The surface of ZnO CQDs are depleted of electrons due to adsorption of oxygen molecules as well as a high density of surface traps [31,32,33,34]. A high injection of electrons fills and passivates the surface traps on the ZnO CQDs close to the PbS/ZnO interface. According to Weaver et al., indeed, reductive passivation of surface traps through electron injection over the trap energy level of ZnSe CQDs led to large photoluminescence electro-brightening in the ZnSe CQDs, implicating interfacial carrier charging in determining luminescence intensity [35]. On the other hand, PbS CQDs are very well-known to form dynamic traps through which electrons are depleted, as will be discussed later. In contrast to the charging process, the discharging process under application of −10 V depletes the charged electrons at the PbS/ZnO interface. Therefore, the subsequent bias at 1.5 V, after the charging process, causes a far higher current in comparison with that after the discharging process, as shown in Figure 3b. It is important to note that the diode current increases with time at a higher voltage (10 V) while the current decreases at a lower voltage (5 V, Data not shown). At a high electric field, injected electrons fill the trap states fast and begin to passivate the interface while, at a low electric field, injected carriers are still in the process of being trapped. On the basis of the framework, the current decrease during the increase in the EL intensity at a low voltage (1.5 V) implicates electron trapping at the PbS/ZnO interface, achieving charge balance.

In our previous studies, we investigated interfacial charging of PbS CQDs, finding that the time scale of carrier trapping associated with surface traps corresponds to tens of seconds which is consistent with that of the current and EL intensity change in Figure 2b [31,32]. The long time scale over tens of seconds and the significant current change over an order of magnitude in Figure 3a in combination with the charging/discharging experiments in Figure 3b allow us to infer that the increase in the EL intensity during light emission is ascribed to the interfacial trapping of carriers at the PbS/ZnO interface.

Indeed, suppression of the injection current due to interfacial charge trapping should be considered in articulating the origin of the EL intensity increase during emission. The net injection current in the disordered material systems embedding CQDs is determined by the difference between the injected current across interface and the surface recombination current [36,37,38,39]. Many studies have reported that the electrical contact resistance at the interface involving CQDs is governed by the charge carrier mobility as well as the interfacial energy level [40,41]. PbS CQDs possess a high density of surface charge trapping states originating from unpassivated cationic lead sites and oxide species such as PbSO₃ and PbSO₄ (electron traps), and anionic sulfur sites (hole traps) [42,43,44,45,46].

It is important to emphasize that Auger recombination as well as CQD charging is mitigated due to the reduction in the net injection current. According to Bae et al., the onset of the EQE roll-off in the CQDs with reduced Auger recombination rate starts at higher currents with increased emission efficiency [12,19]. They demonstrated that the EQE roll-off in the CdSe/CdS structure is not affected by electric field-induced luminescence quenching due to spatial separation between electrons and hole. In other words, the optimum electric field at which the degree of charge balance is maximized exists, and a higher value of EQE is obtained, as the degree of CQD charging is reduced. Indeed, charge balance state was improved by increasing the interfacial energy barrier through the introduction of a thin outer shell (Zn_0.5Cd_0.5S) between the CdS (shell) and ZnO. Suppression of electron injection into the emission layer by the outer shell exhibited far improved roll-off behavior.

To elucidate the origin of the EL intensity increase, we traced the emission spectra in Figure 5a. The change in the EL intensity peak wavelength is plotted as a function of time under the application of a voltage in Figure 5b. The wavelength at the emission peak was, importantly, blue-shifted from 1450 to 1350 nm in 2 min, ruling out the possibility of electric-field-induced quantum-confined Stark effect [47]. The observed blue-shift could be a signature of multi-carrier emission, contributing to an increase in the EL intensity [12]. In a previous study, for example, the core/alloy layer/shell QD device showed a blue-shift of the EL spectra, which is translated into the contribution arising from multi-carrier emission increased [12,19]. In the CdSe/CdS QDs, a fraction of electrons and holes are separated in the core/shell structure. In case that multi-carrier emission takes place in the structure, spatial distribution of carriers causes repulsion, increasing the energy of the multi-carrier system. We eliminate the Joule-heating effect as a possible origin because the EL intensity increased despite a blue-shift [48,49,50].

The presence of competing mechanisms decreased the EL intensity in 2 min, as seen in Figure 6a. Suppression of the blue-shift occurred with a decrease in the EL intensity, as seen in Figure 6b, indicating that electric field-induced luminescence quenching mechanisms were activated, involving spatial separation of the electrons and holes, QD charging, and carrier coupling to surface states. In this regime, we speculate that trapping of the injected carriers into the localized surface states, i.e., carrier coupling to surface states, has a significant effect on the decrease in the EL intensity. Importantly, PbS CQDs have been reported to form dynamic traps under high electric field, originating from field-induced ionization of capping molecules and nano morphology change of the CQD surface [43]. In our previous study [31], indeed, we observed a time constant of tens of seconds in the time-domain measurements for pentacene/PbS CQD bilayer field-effect transistor devices, allowing us to connect the EL decay under high electric field to dynamic trap formation. From a far greater time constant than that with static electronic traps of fs to ms, we argued that dynamic traps originating from impurities associated with hydroxyl and hydrogen are present on the PbS CQD surface. It has been accepted that injection of charge carriers into PbS CQDs activates the reactions (O₂+2H₂O+4e⁻↔4OH^- and O₂+4H⁺+4e⁻↔2H₂O) associated with dynamic traps, causing a reduction of the mobile carrier density by depleting carriers through production of hydroxyl and hydrogen ions [51,52,53,54]. As mentioned before, indeed, the time scale of the current decay in the EL measurement is consistent with a long charge retention time on the order of minutes associated with dynamic trap formation, enabling us to infer that luminescence quenching is caused by the presence of the dynamic traps on the PbS CQD surface.

In addition to electric field-induced luminescence quenching due to dynamic trap formation, the electrical bias-stress effect is thought to decrease the EL intensity. Contribution of the bias stress effect to luminescence quenching is evidenced from the fact that the current decay in Figure 6a is well-matched with a stretched exponential function, I(t) = I_oexp(−(t/τ)^β), which has been extensively used to explain carrier trapping-induced bias stress effect as seen in the inset of Figure 6a. In the relation, I_o is the pre-exponential current, τ and β are the time constant and dispersion parameter, respectively. A relatively long time constant (τ) of 161 s is consistent with that of charge injection-induced dynamic trap formation. Importantly, the current decay during the increase in the EL intensity in Figure 2b is not well fitted to the stretched exponential function, clarifying that two different mechanisms compete to determine the EL intensity. It is noted that the contribution of the Joule-heating effect on the EL intensity is not substantial because blue-shift was suppressed.

In elaborating on the spectral shift during EL intensity variation, transition in the emission zone should be considered. Temporal blue-shift until 2 min can be interpreted that electron-hole recombination zone moved from the ZnO/PbS interface with an energy difference of 0.8 eV between the LUMO of the ZnO (−4.3 eV) and the HOMO of the PbS CQD (−5.1 eV) to the PbS CQD with a band gap of 0.9 eV. To elucidate the competing mechanisms, more controlled experiments in combination with transient absorption and luminescence measurements are required.

4. Conclusions

We report, for the first time, that the temporal evolution of the EL intensity in a PbS CQD NIR LED is a result of competition between change in the charge injection/extraction barrier due to interfacial charging that leads to charge balance and electric field-induced luminescence quenching. The increasing EL intensity is thought to be initiated by the achievement of charge carrier balance, deactivating non-radiative Auger recombination. The following EL intensity decrease is ascribed to electric field induced-QD charging coupled with surface states and bias stress effect based on the suppression of the spectral blue shift and the current decay due to carrier trapping. Our findings raise a critical issue of the electric field-induced dynamic trap formation associated with interfacial or QD surface charging, providing insights into designing functional interfaces to enhance device stability as well as the EQE in the NIR CQD QLEDs.