1. Introduction

Quantum dot-light-emitting diode (QD-LED), where visible-fluorescent QDs as color converters are integrated with an LED chip as an excitation source, has been intensively investigated for last decade. Bi- or tricolored white light emission can be typically generated through the combination of the blue emission transmitted from a chip plus the QD emission(s) color-converted by the blue excitation. The primary QD compositional candidates for LED fabrication may be categorized into Cd-containing II−VI group [1–4] versus non-Cd III−V (e.g., InP [5–7]) and I−III−VI ones (e.g., ternary Cu−In−S (CIS) [8–11] and its quaternary derivatives such as Cu−In−Ga−S [12] and Zn−Cu−In−S [13,14]). Depending on the spectral position, coverage, and ratio of QD emissions in white QD-LEDs, the resulting white lights possess different figure-of-merits such as color rendering index (CRI) and color gamut, which are key metrics in general lighting and display backlighting sources, respectively. Using Cd-containing QDs high-quality white lights with a color gamut as wide as 104% relative to the National Television Systems Committee (NTSC) color space [1] as well as a CRI close to 90 [2] have been already demonstrated through wisely designing the white spectral distribution. On the other hand, the work on non-Cd QDs-integrated white LEDs has been exclusively targeted toward the fabrication of general lighting devices [5–14]. This is mainly because the emission bandwidths of non-Cd QDs available to date are quite broader, i.e., less color-pure, compared to Cd-containing rivals. Herein, highly fluorescent, color-pure green and red non-Cd QDs of InP/ZnS core/shell (C/S) are synthesized, embedded into a transparent polymer matrix to prepare free-standing QD-polymer composites, and then for the first time utilized as color converters for the fabrication of remote-type, high-color gamut white QD-LED.

2. Experimental details

InP/ZnS C/S QDs were synthesized with some modifications from our previous protocol [15], where tris(dimethylamino)phosphine (P(N(CH₃)₂)₃), P(DMA)₃), was used as phosphorus (P) precursor instead of a common, but extremely toxic substance of tris(trimethylsilyl)phosphine (P(Si(CH₃)₃)₃, P(TMS)₃). In a typical preparation of green InP/ZnS QDs, 2.0 mmol of InCl₃, 2.4 mmol of ZnO and 12 ml of oleylamine were loaded in a three-neck flask and evacuated/Ar-purged. The above mixture became optically clear upon overheating to 280−290°C, followed by lowering the temperature to 210°C. Then, 0.25 ml of P(DMA)₃ was injected and the InP core growth proceeded for 1 min at that temperature. For a consecutive ZnS shell overcoating, 3 ml of 1-dodecanethiol was slowly introduced to the above InP QD growth solution and the reaction was maintained at 200°C for 7−8 h. After that, the shelling was finalized by adding 6 mmol of Zn acetate dissolved in 6 ml of oleic acid and holding the reaction for 1 h. Synthesis of red InP/ZnS QDs was a bit modified from core growth details of green QDs, while the identical ZnS shelling was applied. Specifically, 1.4 mmol of ZnO was used, and the reaction temperature and time for core growth were set at 220°C and 4 min, respectively, to obtain larger-sized QDs. As-synthesized QDs were washed via the repeated precipitation/dispersion method with ethanol/hexane, and then dispersed in chloroform.

For the preparation of a 0.52 mm-thick free-standing polymeric composite plate embedded with green InP/ZnS QDs, 0.40 g of QDs was blended with 0.8 g of polyvinylpyrrolidone (PVP, M_w = 360000 g) in 12 ml of chloroform. 4 ml of this blend solution was poured into an aluminum dish (24 mm diameter × 8 mm depth) and slowly dried at 50°C for ~5 h. Another 0.25 mm-thick red QD composite plate was obtained similarly by blending 0.25 g of QDs with PVP/chloroform of the same quantities as in above and then taking 2 ml of this red QD-PVP solution. Dual-emitting bilayered plate was prepared simply by gluing these green and red QD plates with a mixture of epoxy resin/hardener (weight ratio of 1), with a thermal curing (60°C, 3 h) followed. For the fabrication of remote-type QD-LEDs, the resulting single- and dual-emitting QD plates, which were punched into 3.5 mm-diameter disk shape, were loaded onto an InGaN-based blue (450 nm) LED mold and then packaged with the above epoxy mixture under a thermal curing at 110°C for 1 h.

Absorption and photoluminescence (PL) spectra of QDs were collected by UV–visible absorption spectroscopy (Shimadzu, UV-2450) and a 500 W Xe lamp-equipped spectrophotometer (PSI Co. Ltd., Darsa Pro-5200), respectively. PL quantum yield (QY) was estimated by comparing the integrated emission of a dilute QD dispersion in chloroform with that of rhodamine 6G (QY of ~96%) ethanol solution. Transmission electron microscopy (TEM, FEI Tecnai G2 F20) work was performed on a QD composite plate that was sliced by a microtome. For PL decay data, QD plates were excited at 410 nm by 3 ps pulses from Ti:Sapphire laser operating at a repetition rate of 76 MHz and PL decay dynamics were resolved using a time-correlated single photon counting method. Electroluminescent (EL) performances of remote-type QD-LEDs were evaluated in an integrating sphere with a diode array rapid analyzer system (PSI Co. Ltd.).

3. Results and discussion

Figure 1 presents typical UV-visible absorption and normalized PL spectra of two green (516 nm) and red (614 nm) InP/ZnS QDs, where narrow PL bandwidths of 48 and 58 nm as well as well distinct excitonic absorption peaks at 470 and 582 nm, respectively, can be seen, indicating that both QDs possessed a good size monodispersity. The narrowest emission bandwidths achievable from the common, toxic P precursor (P(TMS)₃)-based InP/ZnS QDs in literature that emitted green (e.g., 515−525 nm) and red wavelengths (e.g., 610−620 nm) of interest in the display technology are in the ranges of 45−50 [16–18] and 78−100 nm [17,19], respectively. It is emphasized that our green and red QDs are comparable or even superior in color purity to the P(TMS)₃)-derived best InP/ZnS QDs available to date. Together with PL bandwidth, PL QY of InP/ZnS QDs reported previously is strongly dependent on emission wavelength, typically showing the high (e.g., 60−80%) and low values (e.g., 18−58%) in green and red wavelengths specified above, respectively [16,17,19]. Meanwhile, PL QYs of our green and red InP/ZnS QDs were measured to be 61 and 69%, respectively. From multiple standpoints of emission wavelength, bandwidth, and QY, therefore, the present InP/ZnS QDs may be regarded as promising Cd-free color converting materials to be applied for the fabrication of highly color-reproducible white QD-LED.

 

Fig. 1 Normalized absorption and PL spectra of green and red InP/ZnS QDs. The inset presents a fluorescent image of two QD dispersions under UV (365 nm) illumination.

Download Full Size | PDF

Analogous to phosphor-converted LEDs, most QD-LEDs have been fabricated by physically blending QDs with a resin mixture. However, this simple method generally accompanies the unwanted results such as substantial QD agglomeration and incomplete resin hardening, primarily originating from the processing incompatibility between strongly hydrophobic ligands of QD surface and resin encapsulants [5,11]. To address these issues, in this work, InP/ZnS QDs were embedded in a transparent polymer of PVP that is well miscible with hydrophobic QDs in chloroform, thus expecting the formation of QD agglomeration-free, homogeneous polymeric composite. The present processing for the fabrication of QD composite plate is not only simple, readily large-scalable, but also flexible in controlling QD load and thickness of plate. Figure 2(a) shows two bright polymeric composite plates embedded with red and green QDs under UV lamp illumination. A TEM image of green QD plate (Fig. 2(b)) also indicates that the QDs were uniformly distributed without appreciable agglomeration in the polymeric matrix. As shown in the schematic of device fabrication (Fig. 2(c)), the individual QD plates were integrated in a remote fashion with a blue LED. Figure 2(d) presents normalized EL spectra of green and red QD plate-loaded LEDs under a driving current of 20 mA. Green and red QD emissions in EL peaked at 532 and 639 nm, respectively, which were somewhat red-shifted compared to those (i.e., 516 for green and 614 nm for red QDs) in PL. These lower-energy emission shifts are ascribable to the combined factors of non-radiative Förster resonant energy transfer (FRET) as well as light reabsorption loss in the QD ensemble inside polymeric matrix [4,8]. The luminous efficacies of green and red QD plate-loaded LEDs at 20 mA were 30.5 and 10.8 lm/W, respectively. Taking into account that luminous flux (lm) is determined by human vision functions with a much higher sensitivity in green versus red regime, a higher luminous efficacy from the green device versus red one is understandable.

 

Fig. 2 (a) Photograph of red and green QD composite plates under UV illumination. (b) TEM image of a plate containing green QDs. (c) Schematic representation of remote-type QD-LED. (d) Normalized EL spectra of green and red QD plates-loaded LEDs at an operating current of 20 mA. The insets of Fig. 2(d) are the photographs of 20 mA-driven QD-LEDs.

Download Full Size | PDF

As a preliminary QD-LED fabrication test, a QD composite plate which comprises green and red QDs randomly distributed in a single matrix was prepared and then tested as a dual-emitting color converter. As noticed from EL spectrum of this blended QD plate-loaded LED in Fig. 3(a), red QD emission was overwhelmingly dominant, while green QD emission became nearly invisible. Such a dramatic quenching of green QD emission is a clear signature of the efficient FRET process from high-band gap (e.g., green) as donors to lower-band gap (e.g., red) QDs as acceptors. This energy transfer can be further elucidated by analyzing PL decay dynamics of donors and acceptors. Figure 3(b) shows a set of room temperature PL decay profiles of QD plates embedded with green, red, and green/red blended QDs. All decay curves were well fit with a triexponential function with the fitting parameters of amplitude (A) and decay time constant (τ), i.e., f(t) = A₁exp(−t/τ₁) + A₂exp(−t/τ₂) + A₃exp(−t/τ₃). Such a multiexponential behavior originates from the presence of various nonradiative channels that are involved in overall emission kinetics [20]. In such cases, the amplitude-weighted average lifetime (τ_ave) may be a representative measure of exciton lifetime. A τ_ave (53.6 ns) of green QD plate was reduced to 42.2 ns in the case of blended QD plate at the identical detection wavelength of 532 nm. In contrast, the red emission τ_ave (51.2 ns) of the blended QD plate became longer than that (45.8 ns) of red QD plate at the same detection wavelength of 638 nm. These lifetime variations, i.e., shortened green emission and lengthened red emission, of blended QD plate, are direct evidences of inter-dot FRET from green to neighboring red QDs. In the blended QD plate, FRET efficiency (η), according to the equation of η = 1 − τ_DA/τ_D (where, τ_DA and τ_D are the lifetimes of the donor with and without acceptor, respectively), was calculated to be ~21%. However, when considering a substantial quenching of green QD emission observed (Fig. 3(a)), this FRET efficiency is lower than expected, signifying that the high-energy photons emitted from green QDs would be inevitably reabsorbed by lower-band gap red ones to a considerable extent. For an effort to devise a dual-emitting color converter with a minimal FRET efficiency, the individual green and red QD plates in Fig. 2(a) were physically combined into a bilayered form (see Experimental details), as shown in the middle inset of Fig. 3(a). Note that green and red component τ_ave values of the bilayered plate were also collected (not shown here) and found to be identical to ones of monolayered green and red counterparts in Fig. 3(b). This bilayered design can afford the spatial separation of green QD donors from red QD acceptors, thereby naturally expecting the effective FRET suppression between two interacting QD dipoles. This bilayered QD plate was first applied in the same remote manner as in above with a red plate side facing a LED chip (i.e., green-on-red). Intriguingly, this green-on-red configuration led to a very weak green emission along with a predominant red one (Fig. 3(a)), similar to the EL result from the blended QD plate-loaded LED. This is attributable to a preferential absorption of blue excitation dose by red QD side, rendering the subsequent excitation of upper green QD side appreciably attenuated.

 

Fig. 3 (a) EL spectra of LEDs loaded with blended QD plate and bilayered QD plate with a green-on-red configuration at 20 mA. Photographs of UV-irradiating bilayered QD plate (middle inset) and the respective QD-LEDs (right insets) are also included. (b) Comparison of PL decay profiles of QD plates embedded with green, red, and green/red blended QDs.

Download Full Size | PDF

Upon loading the same bilayered QD plate with the reverse configuration of red-on-green, a spectrally well-separated, tricolored white EL spectrum, as intended, could be obtained as shown in Fig. 4(a). Analogous to the previous green-on-red configuration-based device, the QD-LED having this reverse configuration should exhibit a predominant blue light absorption by green QD side over red one. However, some portion of the resulting blue-to-green converted lights would sequentially excite the red QDs atop, producing down-converted red QD emission along with transmitted green one. White emission was spectrally resolved into its primary colors through applying commercial blue (B), green (G), and red (R) color filters (Fig. 4(a)). The white, B, G, and R emissions in Fig. 4(a) corresponded to (0.269, 0.239), (0.153, 0.056), (0.258, 0.652), and (0.656, 0.306), respectively, in CIE color coordinates, as denoted in Fig. 4(b). The color gamut of the resulting RGB triangle was calculated to be 87% relative to NTSC color space. Figure 4(c) shows the EL spectral evolution of QD-LEDs as a function of forward bias of 20−100 mA. With increasing bias the intensities of R, G, B emissions monotonically increased with the respective spectral positions unchanged. As a result, CIE color coordinates were varied only in (0.268−0.269, 0.235−0.239) within 20−100 mA. The luminous efficacy of the white QD-LED exhibited 16.7 lm/W at 20 mA, but steadily decreased to 8.9 lm/W at 100 mA (inset of Fig. 4(c)). This forward bias-dependent efficacy reduction is primarily associated with the limited performance of LED chip used, which exhibits a more efficiency drop at a higher drive current [8,12]. The values in color gamut and luminous efficacy obtained from the above InP QD-based white LED as a display backlight source are not maximal, and there is much room for further improvements of device performance mainly by developing higher-quality QDs with an even narrower emission and higher QY.

 

Fig. 4 (a) No, blue, green, and red color-filtered EL spectra and photographs of LEDs loaded with bilayered plate with a red-on-green configuration at 20 mA. (b) Color coordinates corresponding to color-filtered EL spectra of Fig. 4(a) (white stars) versus NTSC triangle (black line) along with no color-filtered EL (white circle).(c) EL spectral evolution and luminous efficacy change (inset) of the same device with increasing forward bias up to 100 mA.

Download Full Size | PDF

4. Conclusion

Highly efficient, color pure green (516 nm) and red (614 nm) InP/ZnS QDs having PL bandwidths of 48 and 58 nm and QYs of 61 and 69%, respectively, were synthesized and applied as color converters for remote-type LED fabrication. When a QD composite plate which contained green and red QDs randomly distributed in a single PVP matrix was applied, no white light was generated, showing an almost completely quenched green QD emission due to the efficient donor-to-acceptor FRET. Instead, two free-standing QD composite plates, where the respective green and red QDs were embedded into PVP matrix, were physically integrated to produce a dual-color emitting bilayered QD plate. Upon loading the bilayered QD plate with a red-on-green configuration, the resulting white QD-LED generated a well-balanced, tricolored EL spectrum. R, G, and B emissions transmitted through the respective color filters gave a color gamut as high as 87% relative to NTSC triangle standard.