1. Introduction

Colloidal quantum dots (QDs) exhibit unique properties such as size-controllable color tunability by the quantum confinement effect, narrow emission linewidth and low-cost solution-based processing, which make QD a rising candidate material for optoelectronic device such as lasers, photodetectors, solar cells and light-emitting diodes (LEDs) [1–3]. Among these applications, colloidal QDs based LEDs (QD-LEDs) has been investigated popularly because it’s potential impacts to the display and lighting industry. Many research groups have attempted to enhance device performance using various QD compositions and device structures [4–6]. While the performance of QD-LEDs using II-VI semiconductor cadmium selenide (CdSe) colloidal QDs has been upgraded dramatically, a prominent future task is the substitution of Cd-containing QDs by less toxic materials. There have been only a few studies on QD-LEDs using Cd-free QDs [7–12]. Indium phosphide (InP) among III-V semiconductor nanocrystals is the most promising materials for Cd-free QD-LEDs owing to less ionic lattice, reduced toxicity and wide emission spectrum tunability covering the range of visible light [13–15]. The electroluminescent (EL) performance of QD-LEDs using InP QDs can be improved by the optimization of device structure and materials for the carrier balance in the multi-layered LED structure. However, the reported EL performance is insufficient and there are still many problems left such as EL quenching by Auger recombination, efficiency roll-off and device degradation, which limits the performance of QD-LEDs [16].

In this work, efficient Cd-free QD-LEDs were fabricated using InP/ZnSe/ZnS multishell QDs synthesized by the heating-up method [17]. The thicker ZnS shell of multishell QDs was adapted to improve the device efficiency, and the optoelectronic characteristics of QD-LEDs were investigated. Moreover, a serious QD charging problem was found in EL stability measurement, however it was also suppressed by the QD system of thicker ZnS shell to some extent.

2. Experimental

2.1 Synthesis of colloidal quantum dots

First, a mixture of indium(III) acetate (1 mmol) and zinc stearate (2 mmol) was heated until a homogeneous solution was obtained, then dodecanethiol (0.5 mmol) and tris(trimethylsilyl) phosphine (1 mmol) were added in the mixture subsequently and heated to 300 °C for 30 min. In order to grow ZnSe shell, trioctylphosphine selenide (0.5 mmol) was added first in the prepared InP core nanoparticles and heated to 280 °C for 10 min. Subsequently, InP/ZnSe nanoparticles were capped with ZnS outer shell in two ways by adding further zinc and sulfur precursor followed by the heating to 280 °C for 10 min. Between the ZnS outer shell growth steps, half the volume of the synthesis solution was subjected to work-up while the other half was further reacted with more ZnS precursor to get two samples with different thickness of ZnS outer shell. The raw solutions of multishell QDs were purified several times by acetone-induced precipitation and centrifugation, and the QD powders were redispersed in nonane for the device fabrication.

2.2 Fabrication of QD-LEDs

Using the colloidal multishell InP QDs, we fabricated the devices with patterned indium-tin oxide (ITO) on glass substrates. First, patterned ITO substrates were cleaned by sequential ultrasonication in acetone, methanol and isopropanol and dried in an oven and then treated with oxygen plasma. Then, PEDOT:PSS was spin-coated on ITO at a spin rate of 3000 rpm for 30 sec and dried on a hot plate at 180 °C for 30 min as hole injection layer (HIL). Then, poly-TPD dissolved in chlorobenzene (3.5 mg/ml) was also spin-coated at a spin rate of 4000 rpm for 60 sec on the PEDOT:PSS layer and dried on the hot plate at 130 °C for 30 sec as hole transport layer (HTL). In order to deposit emission layer, colloidal QDs with two different thicknesses of ZnS outer shell were spin-coated on the poly-TPD layer at a spin rate of 3000 rpm for 20 sec. After the solution processes, the deposition of both TPBi (75 nm) and LiF/Al (0.5 nm/150 nm) were carried out by thermal evaporation on top of the QD layer as electron transport layer (ETL) and metallic cathode respectively. Finally, the devices were encapsulated with a glass in a nitrogen glove box.

2.3 Characterizations

UV-visible spectra were recorded with a PerkinElmer Lambda 19 spectrometer. Photoluminescence (PL) spectra and quantum yield (QY) were measured by a Hamamatsu C9920-02. Energy levels were determined by UV photoelectron spectroscopy (Riken Keiki AC-2) and UV-visible measurement. Transmission electron microscopic (TEM) images were captured using a Phillips CM 200 instrument. Particle sizes were obtained using the ImageJ software by measuring at least 100 individual particles per image. A spectroradiometer (Minolta CS1000) was used to detect EL spectrum as well as reference luminance of device. Current density-voltage-luminance (J-V-L) characteristic measured with an experimental setup consisting of a Keithley 2400 source meter and a silicon photodiode at ambient condition. Luminance and current efficiency were calculated from the photocurrent of the photodiode and calibrated with the luminance detected from the spectroradiometer.

3. Results and discussion

Figure 1 shows the synthesis procedure for InP-based colloidal multishell QDs. InP/ZnSe/ZnS QDs which have different ZnS shell thickness were synthesized via the heating-up method. The heating-up synthesis is more simple and reproducible compared to the hot-injection method because the manual injection process can be avoided. Through the synthesis, PL QY of InP core QDs can be significantly improved compared to the conventional InP/ZnS single shell QDs due to the better match of lattice constants between InP and ZnSe [17]. Two different QD samples, which have identical InP core but different ZnS outer shell thickness, were used for the fabrication of QD-LEDs. They were obtained by several consecutive shell coating steps in the same InP core batch. Since PL characteristics of QDs are basically determined by core structure, both QDs have almost identical PL peak wavelength (539-540 nm) and full width at half maximum (FWHM, 52-54 nm) as shown in Fig. 2(a), as well as 40% of PL QY. There is only a slight change in PL wavelength with ZnS growth, which indicates that core properties are not affected by the shell growth. A substantial red shift from 523 to 535 nm is observed only after depositing the ZnSe shell. It has been reported in the literature that increasing ZnSe shell thickness results in increasing red shift, because the energy level confinement of the ZnSe shell is weaker than that of the ZnS shell [17]. Therefore, here we only use a thin ZnSe shell to obtain the desired smooth interface between core and shell, and then increase the outer ZnS shell thickness to achieve a strong exciton confinement. The thermogravimetric analysis (TGA) reveals the amount of inorganic content in the QDs (i.e. core and shells) by removing the ligands through heating to 600 °C. The TGA curves in Fig. 2(b) show the slightly increased residual mass with the thicker shell, which is an indication for the increase of particle size by the shell coating. Further, the high inorganic content proves the excellent purity of the samples regarding excess ligands or reaction side products, which is essential for efficient device operation. Moreover, inset of Fig. 2(a) which clearly shows that InP/ZnSe/ZnS-t2 has more absorption from shell than InP/ZnSe/ZnS-t1 under the range of 350 nm which matches the ZnS band gap of 3.54 eV. Therefore, this is an additional proof that the shell thickness was increased [18]. Based on the TEM images in Fig. 2(c) and 2(d), the particle size is estimated to 2.5 ± 0.3 nm and 2.9 ± 0.3 nm for the thinner shell and thicker shell QDs, respectively, which corresponds to a difference of approximately one monolayer of ZnS between the two types of QDs.

 

Fig. 1 A schematic of the one-pot synthesis of InP/ZnSe/ZnS QDs with different ZnS shell thickness.

Download Full Size | PDF

 

Fig. 2 (a) Normalized PL spectra and (b) TGA curves of synthesized colloidal InP/ZnSe/ZnS QDs with different core/shell structures. TEM images of (c) InP/ZnSe/ZnS-t1 and (d) InP/ZnSe/ZnS-t2. (inset of Fig. 2(a): Absorption spectra and difference of intensity caused by thicker ZnS shell)

Download Full Size | PDF

Figure 3 presents the schematic of the InP-based QD-LEDs and the corresponding energy diagram. The QD-LED structure was sophisticatedly designed to achieve efficient carrier injection and balance in the QD layer and to minimize parasitic luminescence of neighboring organic layers. The choice of organic charge transport layer (CTL) surrounding the QD emitting layer is critical for the efficient QD-LEDs considering energy band alignment and material properties. The device structure presented in this study consists of organic materials having high mobility, low potential barrier for carrier injection into the QDs, and spectral overlap between emission of organic materials and QD absorption [19]. PEDOT:PSS is used as the HIL on ITO anode mainly to increase the anode work function and to compensate the surface roughness of the anode in order to obtain a stable organic/inorganic interface. Poly-TPD is used as the HTL to reduce the energy barrier of 0.5 eV between the work function of PEDOT:PSS and valence-band edge (VBE) of the InP/ZnSe/ZnS QDs. The thin film of poly-TPD is chemically and physically stable to nonpolar alkane solvents such as hexane and nonane, so that the QDs can simply be spin-coated on top of the poly-TPD layer from these solvents. TPBi is adapted as the ETL because it exhibits a suitable LUMO energy and works as well as a good hole blocking layer. As shown in the energy band diagram of Fig. 3(b), electrons are easily transported from the cathode via TPBi to the emitting layer. The hole injection is facilitated by the low energy barrier of 0.3 eV between the highest occupied molecular orbital (HOMO) energy level of HTL and the VBE of QDs. The InP-based QD-LEDs generally have a great potential to compete with Cd-based devices because of a lower energy barrier between the HTL and the InP-based QDs.

 

Fig. 3 A cross-section schematic and band structure of InP/ZnSe/ZnS QD-LEDs and the corresponding energy levels.

Download Full Size | PDF

We investigated how the ZnS shell thickness of InP/ZnSe/ZnS QDs effects on the EL performance of QD-LEDs. Normalized EL spectra, current density-voltage (J-V) and current efficiency-current density characteristics of both devices are presented in Fig. 4. The peak wavelength of 555 nm and FWHM of 56 nm without any emission from organic layer are observed in the normalized EL spectra for both devices (Fig. 4(a)). However, there is asymmetric red shift in EL spectra around 15 nm compared to the peak of PL in QD solution (Fig. 2(a)). It may come from exciton migration to larger QDs in the QD ensemble facilitated by film formation, the quantum confined Stark effect known as the shift of exciton energy to lower values than band gap energy under electric field, or emission from defect states within the band gap [20,21]. While the maximum current efficiency and the luminance reached 3.32 cd/A and 1960 cd/m² for the InP QD-LEDs with thinner ZnS shell, the performance of the device with thicker ZnS shell was enhanced up to 4.65 cd/A and 2430 cd/m², respectively. The J-V characteristics of both device in Fig. 4(b) show ohmic conduction up to 2.4 V (J∝V), trap-limited conduction (J∝Vⁿ, n>2) up to 3.3 V and pseudo space-charge-limited conduction (SCLC) (J∝Vⁿ, n~4) behavior at higher voltage [6,22,23]. They show the same slope at the trap-limited conduction region, which refers to similar charge injection into the QD film above the threshold voltage (V_on) even though having different thickness of ZnS outer shell. This means the enhancement of current efficiency is not caused by better injection efficiency but by the suppression of quenching mechanism in QDs. It has been investigated by Char et al. that the electric field in the QD-LEDs localizes electrons to the shell phase or to the surface states because of weak confinement of electrons by lower energy offset between core and shell [8]. The localized electrons can cause the nonradiative exciton decay by Auger quenching or surface-state recombination. However, the enhancement of potential barrier by the thicker outer shell which provides strong confinement can reduce this luminescence quenching by reducing interaction of excitons with surface charges [8]. Moreover, Klimov et al. also have observed that the thicker shell reduced the rate of Auger decay and enhanced the device performance [24]. Consequently, the QDs with thicker ZnS outer shell remain efficiently emissive within our QD-LEDs and show better performance by the strong confinement and the suppression of luminescence quenching [25,26]. We have also tried to make a device with much more shell thickness ( + 1.1 nm, i.e. + 4 layers of ZnS), but EL spectrum shows serious emission from poly-TPD because QDs remained the aggregation or void in emission layer then it generated direct leakage path from TPBi to poly-TPD due to the low solubility of QDs (there are still chemistry issues remained and data are not shown here).

 

Fig. 4 (a) Normalized EL spectra, (b) current density-voltage (c) current efficiency-current density characteristics of InP/ZnSe/ZnS QD-LEDs.

Download Full Size | PDF

Figure 5(a) shows the luminance versus operation time graph for the InP-based QD-LEDs driven with a constant current of 0.84 mA/cm². The stability of QD-LEDs was investigated at low luminance in order to exclude field-induced luminance quenching [24,27]. The operating time to reach 90% of the initial luminance (LT90) was increased from 55 sec to 123 sec for the thicker shell QDs. This improvement is probably due to suppressed Auger recombination caused by charge carriers within QD core.

 

Fig. 5 (a) Stability data for InP-based QD-LEDs with different shell thickness, and (b) restored luminance during repeated device stability test after one day. The initial luminance was 10 cd/m² for thinner ZnS shell and 17 cd/m² for thicker ZnS shell under 0.84 mA/cm² of applied current.

Download Full Size | PDF

An interesting phenomenon was also observed in repeated lifetime measurements and is shown in the inset of Fig. 5(b). The device stability under operation was retested after one day of non-operation and the initial luminance of both devices was fully recovered. It indicates that the degradation of luminance under low electric field can be explained by QD charging. It is known as a critical problem of QD-LEDs caused by trapped carriers in QD itself or in the heterostructure of QD-LEDs [24]. All of these considerations suggest that thicker ZnS shell can suppress the Auger decay caused by the QD charging and increase the current efficiency as well as the device stability. Previous studies have reported that there is efficiency roll-off at high current density which is also observed in the Fig. 4(c) [27]. We suggest that the efficiency roll-off is not only a problem of field-induced quenching but also one of QD charging because the luminance quenching in our devices was observed even though under the condition of low electric field.

As a result, the suppression of QD charging can be controlled by thickness of ZnS shell in InP/ZnSe/ZnS multishell QD system, which also affects the device efficiency and stability. However, further fundamental investigation will be needed to understand more how the shell influences the charge behavior in the device. Moreover, there is still remained problem of efficiency roll-off influenced by both the strong electric field and the QD charging. However, this problem can be overcome by adapting other technology in the InP-based QD-LED system such as Auger assisted energy upconversion by oxide nanoparticle-based charge transport layer [28].

4. Conclusion

The influence of ZnS shell thickness of InP/ZnSe/ZnS QDs on the EL device performance was investigated in our study. We found that QD charging is responsible for luminance quenching in device operation, and it was also clearly observed by the refreshed luminance in repeated stability tests, which means that the charging is reversible. The EL device with thicker ZnS shell QDs shows better current efficiency and stability because the thicker outer shell suppressed the Auger quenching caused by the QD charging. Moreover, we suggest that efficiency roll-off is induced by both the strong electric field as well as the QD charging that results in serious luminance quenching even under the low electric field in our research. The performance of our InP-based QD-LEDs can be matched to that of traditional Cd-based QD-LEDs, which shows the great potential for the Cd-free future information displays. With further investigation, the performance of InP-based QD-LEDs would be more improved by optimizing the core/multishell structure of QDs and charge balance in multi-layered system.