Hyein
Yoo
ac,
Ho Seong
Jang
b,
Kwangyeol
Lee
c and
Kyoungja
Woo
*ad
aNanophotonics Research Center, Korea Institute of Science and Technology, P. O. Box 131, Cheongryang, Seoul 130-650, Korea. E-mail: kjwoo@kist.re.kr
bMaterials Architecturing Research Center, Korea Institute of Science and Technology, P. O. Box 131, Cheongryang, Seoul 130-650, Korea
cDepartment of Chemistry, Korea University, Seoul 136-701, Korea
dUniversity of Science and Technology, Daejeon 305-350, Korea
First published on 25th June 2015
Although the quantum efficiencies of quantum dots (QDs) are approaching unity through advances in the synthesis of QD materials, their luminescence efficiencies after mixing with resin and thermal curing for white light-emitting diodes (LEDs) are seriously lowered because of aggregation and oxidation of QDs and poor adhesion of QDs to the resin. To overcome these problems, QD-layer-encapsulated and phenyl-functionalized silica (SQSPh) spheres were synthesized and applied for white LEDs, whereby the QDs were homogeneously distributed at radial equidistance from the center and near the surface of approximately 100 nm-sized silica spheres and the surface was functionalized with phenylethyl groups. The inter-core distances of QDs were over ∼14 nm, which is over the limit (<10 nm) for Förster resonance energy transfer (FRET) that leads to photoluminescence (PL) reduction. This hierarchical nanostructure excludes a chance of FRET between QDs and provides the QDs a gradually refractive index matching environment, which yields ∼4-fold enhanced PL in SQSPh. More importantly, the SQSPh acquired a highly adhesive property to silicone resin due to their phenyl functional group matching, which resulted in remarkably improved light extraction in white LEDs. When incorporated along with a yellow-emitting Y3Al5O12:Ce3+ (YAG:Ce) phosphor and silicone resin on blue LED chips, the SQSPh spheres presented significantly improved performance [luminous efficiency (LE) = 58.2 lm W−1; colour rendering index Ra = 81.8; I/I0 = 0.98 after 60 h operation] than their original QDs (LE = 39.6 lm W−1; Ra = 78.1; I/I0 = 0.91 after 60 h operation) under a forward bias current of 60 mA.
Meanwhile, QD-layer-encapsulated silica (SQS) sub-microspheres were reported with highly enhanced photoluminescence (PL, 2- to 3-fold) and stability based on their ligand-exchanged QDs.12 The QDs were homogeneously distributed without aggregation at radial equidistance from the centre and near the surface. By reducing the aggregation between SQS spheres, roughly 100 nm-sized SQS spheres were improved to exhibit approximately 5-fold enhanced PL based on their ligand-exchanged QDs and 4-fold enhanced PL based on their original QDs.13 However, the mechanism for PL enhancement is not yet known.
Nevertheless, the SQS spheres can be a good starting material to overcome the aggregation and oxidation problems of QDs for white LEDs because the QDs are homogeneously distributed and the silica shell provides some stability to QDs.12 More importantly, for a good adhesion to silicone resin, the surface functional group of SQS spheres can be facilely matched with the key functional group (i.e., a phenyl group) of a silicone resin, which is expected to improve the LE of the resultant white LEDs. Also, determining the mechanism for PL enhancement is important not only for fundamental science but also for reliable application in various fields.
Today, the most prevalent white LED in solid-state lighting is composed of a blue LED chip and yellow-emitting Y3Al5O12:Ce3+ (YAG:Ce) phosphor and provides high LE with low cost.8,14,15 In this device, a part of the blue light emitted from InGaN chip is absorbed by YAG:Ce phosphor and converted to yellow light. Thus, the resultant white light consists of blue and yellow, showing a critical drawback: red-deficiency. There have been efforts to improve the colour rendering index (CRI) by incorporating red-emitting QDs as phosphors, which exhibited improved CRIs but seriously lowered LEs due to their intrinsic trade-off pairwise.8,15–18 Doped QDs with large Stoke-shift (CdSe:Cu/ZnS with PL QY up to 50%) were introduced as a promising phosphor, improving CRI with reduced sacrifice in LE because they exclude a chance of reabsorption originating from multiphase phosphors.19 However, commercial QDs with CdSe/ZnS core/multi-shell structure showing high PL QY (∼70%) and stability are predominant and conveniently available. Thus, in the current case, it is deserved to apply commercial QDs for the preparation of red-emitting SQS. More importantly, by surface-engineering the resultant SQS with phenyl, and by incorporating it into a commercial red-deficient white LED, it may be possible to achieve much higher LE and stability than the case incorporating the original commercial QDs, along with a high CRI.
In this report, we present QD-layer-encapsulated and phenyl-functionalized silica (SQSPh) along with a proposed mechanism for PL enhancement and its application for white LEDs yielding significantly improved performance (LE = 58.2 lm W−1; average CRI Ra = 81.8; I/I0 = 0.98 after 60 h operation) than the case incorporating the original commercial QDs (LE = 39.6 lm W−1; Ra = 78.1; I/I0 = 0.91 after 60 h operation) under a forward bias current of 60 mA.
UV-Vis absorption and PL spectra were obtained using a Perkin Elmer Lambda 25 and a Hitachi F-7000 spectrophotometer with a QD-based concentration of 5 × 10−8 M in all the traces using ethanol or chloroform as a solvent. The PL data were recorded with 485 nm or 550 nm excitation. PL QY data were recorded using a Hamamatsu Photonics C9920-02 absolute PL QY measurement system at Korea Institute of Machinery & Materials for the original QD, SQS, and SQSPh solutions only, because these solutions were stable for more than 1 year whereas Q and SQ solutions were not stable even after 1 day. FT-IR spectra were obtained using a Mattson IR 300 spectrometer with a KBr pellet including each sample. TEM and SEM images were obtained using a Philips CM30 with 200 kV and FEI Inspect F50 with 15 kV, respectively.
![]() | ||
Scheme 1 Synthesis of QD-layer-encapsulated and phenyl-functionalized silica (SQSPh) spheres and chemical structure of a hardening component (i.e., methylphenyl siloxane) of a silicone resin. |
A representative set of data for SQSPh emitting at 620 nm is displayed in Fig. 1 and S1 (ESI†), and a set of data for SQSPh emitting at 611 nm is included in Fig. S2.† The clockwise TEM images in Fig. 1a distinctly show each material corresponding to Scheme 1. That is, the darker QD particles are homogeneously distributed on a grey silica particle in SQ and the SQ was encapsulated with a thin silica shell around 10–15 nm to yield an SQS. In the SQS and SQSPh, the encapsulated QDs appear in the form of a dark ring and scattered dots within the ring because the TEM data exhibit projected images of the three-dimensional SQS and SQSPh on a two-dimensional plane. Thus, the ring pattern can be an indicator for homogeneous distribution of QDs as a layer within a large silica sphere. Within the ring pattern, the dark grey dots may reflect the QDs from the lower hemisphere and the distinctly darker dots may reflect the QDs from the upper hemisphere. Additional images to show a homogeneous distribution can be found in Fig. S2† and a previous report.12 The TEM image of SQSPh is not distinguishable from that of SQS because the organic monolayer consisting of phenylethyl is too thin to be observed. As can be noticed in the SEM images, intermittent sonication of the solution during silica shell formation greatly reduced the aggregation of roughly 100 nm-sized SQS particles (Fig. 1b, left) compared with the case without sonication (Fig. S1b†). Notably however, the SQSPh particles exhibited slightly higher aggregating tendencies than the SQS particles (Fig. 1b). The aggregation of SQSPh particles appears to be caused by the temporary stacking of peripheral phenyl groups between the neighbouring SQSPh particles. The phenyl group-stacking is an indicator of good mixing and adhesion of SQSPh to the silicone resin, as will be discussed later. In the photographs of Fig. 1c, the SQS particles produce an opaque solution in ethanol, whereas the SQSPh particles produce a transparent solution in chloroform because of their higher solubility after phenyl-functionalization. A better solvent was not found for the SQS particles. Nevertheless, both solutions exhibited highly bright red emissions.
![]() | ||
Fig. 1 (a) TEM images, (b) SEM images, and (c) photographs, and (d) FT-IR, (e) UV-Vis absorption, and (f) PL spectra of the materials (QD, Q, S, SQ, SQS, and SQSPh with λemit = 620 nm) corresponding to Scheme 1. |
The new FT-IR peaks appearing at approximately 3100 cm−1 (shoulder), slightly below 3000 cm−1, and 1453 cm−1 in SQSPh in Fig. 1d represent the vibrational stretching modes of aromatic and aliphatic C–H and the bending mode of CH2,20 respectively, and implicate surface-functionalization of SQS with the phenylethyl moiety. The absorption and PL spectra in Fig. 1e and f exhibit the same trends as those12,13 previously reported for SQS. Both the absorption and PL intensities of SQ and SQS increased moderately to dramatically compared with those of the original QDs in the order QD < SQ < SQS. Here, it is noteworthy that we assumed 100% yield to estimate the QD concentration in a consecutive synthetic process; however, there must be at least a minimal experimental loss that may contribute to decrease the absorption and PL intensities, which is in contrast to the current data. The increasing trend of the absorbance is partially attributable to the light scattering by aggregated and poorly soluble particles in the opaque solution. Thus, the PL intensity is expected to exhibit a further decreasing trend as the synthetic process proceeds. However, the SQS solutions encapsulating QDs with λemit = 620 nm and 611 nm, respectively, exhibited 3.9- and 3.7-fold enhanced PL compared with their original QD solutions. After phenyl-functionalization, the absorbance spectrum of SQSPh becomes much lower than that of SQS, whereas their PL intensities are comparable. Despite the temporary slight aggregation by phenyl group-stacking, the lowered absorbance spectrum and comparable PL intensity of SQSPh compared with those of SQS are attributable to the high solubility of SQSPh in chloroform. This behavior is clearly recognizable in the photographs of Fig. 1c.
The PL QYs of the original QD, SQS, and SQSPh solutions were 69.0%, 47.4%, and 34.7%, respectively. As described earlier, because the SQS solution is opaque, its PL QY may not be reliable. Anyway, it is certain that the PL QY of SQSPh is lower compared with that of the original QDs. However, the lower PL QY of the solution does not necessarily mean the lower LE of the resultant white LED as will be discussed later.
After careful study, we reasoned that the moderate PL enhancement (1.2-fold) achieved by forming SQ is attributable to the nanostructured rough surface of silica. It has been known that the light extraction efficiency from an emitter increases through surface roughening because the refractive index (RI) of a roughened surface appears as an intermediate between those of the surrounding medium (here, solvent) and the material without surface roughening and thus, the resultant gradual RI matching effect reduces internal light reflection and scatters the light outward.21–24 The local area of the silica sphere where a QD is located can be approximated as a roughened surface because the amorphous silica prepared by the Stöber process25 consists of a porous network structure with a porosity of 10–15%.26–28 In an on-going study, we confirmed that the SQ analogue consisted of a further roughened silica sphere and the original QDs showed 2.7-fold enhanced PL based on the original QDs. It supports our reasoning for PL-enhancement by surface roughening and the result will be published elsewhere after further study. The additional PL enhancement from SQ to SQS is also attributable to the gradual RI matching effect by the silica shell whose RI is intermediate between those of QD and the surrounding medium (solvent); however, the enhancement (3.9-fold) is superior to the case of SQ (1.2-fold) because the thin (10–15 nm) silica shell covers most part of the QD surface.
Furthermore, by virtue of structural characteristics, the SQS appears to exhibit the gradual RI matching effect efficiently. In the SQS, the QDs are homogeneously distributed due to electrostatic interaction at radial equidistance near the surface of the silica sphere. The average inter-QD-distance was over 20 nm in a previous report12 and is over 10 nm here because much larger QDs are used (i.e., 10.0 nm-sized QDs are used here compared with the previous case where 6.0 nm-sized QDs were used). Accordingly, considering the relationship between the diameter and band edge absorption wavelength of CdSe cores,29 the average inter-core-distance of QDs is estimated over 14 nm. This distance is over the limit for the Förster resonance energy transfer (FRET < 10 nm) which leads to serious reduction in PL intensity by self-quenching between QDs.3,10,28,30,31 It is in contrast with other studies that reported PL reduction by FRET in QD-embedded silica.9,10,32 Also, the homogeneous distribution of QDs at radial equidistance of the silica sphere is efficient for the light extraction compared with the accumulated one along the radial direction.12,32 All these structural characteristics of the SQS play an important role in efficient extraction of PL by gradual RI matching.
It is also noteworthy that light absorption in SQ and SQS can be enhanced by gradual RI matching, too.24 The absorbance values of QD, Q, SQ, SQS, and SQSPh at 485 nm in Fig. 1e were 0.027, 0.033, 0.177, 0.582, and 0.261, respectively. In the cases of opaque SQ and SQS solutions, the increased absorbance values appeared as a summation of the enhanced light absorption by gradual RI matching and the light scattering caused by their aggregation and poor solubility, which were not separable. On the other hand, in the case of transparent SQSPh, the enhanced light absorption was reflected as the major increment in absorbance value by approximately 9.7-fold based on the QDs, in Fig. 1e. (The absorbance value by silica only solution12 was ignorable compared with those by SQ, SQS, and SQSPh.) Thus, the lowered PL QY of SQSPh solution (34.7%) based on the original QDs (69.0%) is attributable to the relatively high enhancement in light absorption than in PL. Nevertheless, the remarkably enhanced PL with the proposed mechanism makes the SQS analogues as a promising candidate for phosphors and additionally, for other applications such as bio-imaging and bio-sensing materials.
Before white LED packaging, we simply investigated the miscibility and adhesive property of SQS analogues to silicone resin by mixing and curing them as a monolithic plate. The backbone of the hardening component (i.e., methylphenyl siloxane) of a commercial silicone resin (OE-6630 A/B KIT) has many dangling phenyl groups as illustrated in Scheme 1. In Fig. 2a, the SQS contains some aggregated dots that are detectable by the naked eye, as indicated by the arrows. However, SQSPh exhibits superior miscibility and adhesion to the resin than SQS because SQSPh and resins have the same phenyl groups, and the phenyl groups are alternately stackable.
Fig. 2b presents a representative image and diagram of the white LED package that was prepared by integrating SQS or SQSPh along with YAG:Ce phosphor and silicone resin on a blue LED chip. For comparison, nine different white LED samples were fabricated using different materials as the luminescence down-converters on blue LED chips. Their optical properties were investigated using a forward bias current of 60 mA, and the results are summarized in Table S1.† Here, the original QDs emitting at 611 nm and 620 nm are denoted as QD1 and QD2, respectively. The corresponding ligand-exchanged QDs and their derived products are denoted as Q1 and Q2, SQ1S and SQ2S, and SQ1SPh and SQ2SPh, respectively. When only the YAG:Ce phosphor was integrated for the white LED, the resultant CRI (Ra), which is the average value of R1–R8, and LE were 65.2 and 78.9 lm W−1, respectively. When the original QD was integrated along with the YAG:Ce phosphor, the respective Ra and LE were 78.1 and 39.6 lm W−1 for QD1 and 70.6 and 37.1 lm W−1 for QD2. As expected from a reference,8 a moderate increase in Ra and sharp decrease in LE were observed most likely because of the aggregation of QDs. The white LEDs with original QDs can be used as standards to compare the optical properties of the fabricated LEDs with SQS analogues. When SQSs were incorporated instead of QDs along with the YAG:Ce phosphor, the Ra values fluctuated slightly down and up, whereas the LEs were slightly improved in both LEDs (42.5 lm W−1 for SQ1S and 39.9 lm W−1 for SQ2S). The incorporation of a mixture of SQ1S and SQ2S instead of QDs resulted in better optical properties (Ra = 80.0 and LE = 44.1 lm W−1) than the cases incorporating a single product, SQ1S or SQ2S. When SQ1SPh and SQ2SPh were incorporated instead of each QD, the LEs of the corresponding LEDs were distinctly improved to 50.5 lm W−1 and 47.8 lm W−1, respectively, as we reasoned by functional group matching and their Ra values fluctuated slightly down and up similar to the cases of SQSs. Most importantly, when the mixture of SQ1SPh and SQ2SPh was incorporated instead of QDs, the LE of the corresponding LED was markedly improved to 58.2 lm W−1 with a moderate increase of Ra to 81.8. Therefore, replacing QDs with SQSPhs yielded a significant improvement in the LED performance, especially in the LE, and it was achieved through enhanced PL, better mixing with the resin, and better adhesion to the resin. This improvement was consistently observed throughout the repeated experiments.
The electroluminescence (EL) spectra showing the remarkable change are displayed in Fig. 2c. The EL spectrum of the white LED incorporating the mixture of SQ1SPh and SQ2SPh exhibits higher EL intensity than those of the other samples, whereas that of SQ2S exhibits lower EL intensity. The lower EL intensity of the white LED incorporating SQ2S is attributable to the aggregation of SQ2S particles in the resin due to bad miscibility and adhesion. In particular, the eye-visible aggregated SQ2S particles (indicated by arrows in Fig. 2a) can scatter the emitting light back and seriously lower the EL intensity at ∼550 nm as well as at ∼620 nm which correspond to the emission bands of YAG:Ce phosphors and SQ2S, respectively. In contrast, the higher EL intensity of the white LED incorporating the mixture of SQ1SPh and SQ2SPh is attributable to good miscibility and adhesion between SQSPh and resins by functional group matching as shown in Fig. 2a. In particular, the EL intensity of the white LED incorporating the mixture of SQ1SPh and SQ2SPh is higher than that incorporating the single product SQ1SPh or SQ2SPh. It occurred in the case of SQS too (Fig. S3†). Besides, when the relative ratio of the single product SQ2S was increased from 0.7 to 1, the EL intensity at ∼550 nm decreased but that at ∼620 nm increased as expected through the increased reabsorption of yellow light and emission of red light by the increased amount of SQ2S (Fig. S3†), and the resultant LE has decreased from 39.9 lm W−1 to 30.8 lm W−1. A plausible reasoning for the higher EL in the case of the mixture compared to the case of a single product is that the concentration of SQ1S (or SQ1SPh) is a half of SQ2S (or SQ2SPh) in the current white LEDs and the reabsorption of ∼611 nm light by SQ2S (or SQ2SPh) may not be efficient in such a diluted medium. Thus, the resultant EL spectrum can be higher and broader than those incorporating a single product at the fixed concentration of total SQS analogues (100:
10
:
0.70 or 0.75 for the weight ratio of resin
:
YAG:Ce
:
SQS analogues, 0.70 for the single product and 0.75 for the mixture). This reasoning was indirectly supported by a simplified test with SQS under excitation at 550 nm (Fig. S4†). The PL spectrum of the mixture solution of SQ1S and SQ2S [(2.5 + 5.0) × 108 M] was higher and broader than that of SQ2S only solution (7.0 × 108 M) and the integrated PL ratio was 1.4
:
1, which is far higher than the concentration ratio of 1.1
:
1. Thus, at the ratio 0.70 for the single SQS analogue, a concentration quenching may occur and the adjusted white light is emitted. On the other hand, at the combined ratio 0.75 for the mixture, the concentration quenching is minimized and the resultant white light is emitted. An image of the white light obtained from the LED with SQSPh mixture under a forward bias current of 60 mA is included as an inset.
The Commission Internationale de l'Eclairage (CIE) colour coordinates33 of LED VII (0.3516, 0.3702), VIII (0.3431, 0.3697), and IX (0.3360, 0.3355) incorporating SQ1SPh, SQ2SPh, and their mixture respectively are displayed in Fig. 2d. The white light of LED IX exhibits the highest Ra of 81.8 with a correlated colour temperature (CCT) of 5338 K. As displayed in Fig. S5,† the respective values of R1 and R8, which are related to red colour rendering, were improved from 64.4 and 55.9 for YAG:Ce phosphor-based white LED to 76.8 and 69.4 by the addition of QD2, and further improved to 88.1 and 77.6 by the addition of the mixture of SQ1SPh and SQ2SPh.
The PL bands of LED IX in Fig. 2e exhibit increasing intensities with increasing applied forward currents, implicating no saturation in their PL and thus a promising property for white LEDs. The operational EL stability of the white LED fabricated using each representative material was investigated 6 months after preparation. Notably, after 6 months, all the LED samples exhibited increased EL intensities compared with the corresponding as-prepared LED samples (Fig. S6†); the samples were then, slightly degraded for the initial several hours and then stabilized, though not perfectly as observed in the LEs and normalized EL intensities in Fig. 2f and S6.† The initial optical degradation is attributable to the oxygen and moisture contained in the integration sphere, where they can permeate into the QDs through the porous silica. Nevertheless, the stabilization of QDs by silica under LED operation was remarkable, which is consistent with other reports.9–11 The order of operational EL stability was SQSPh > SQS > QD. After 60 h operation, compared with I/I0 = 0.91 for QD2, the protection of QDs by SQS hybridization resulted in I/I0 = 0.95, which was further improved to 0.98 for SQSPh. The hydrophobic stackable phenyl groups dangling from the SQSPh appear to additionally protect the QDs from oxygen and moisture.
Footnote |
† Electronic supplementary information (ESI) available: TEM images of purchased and ligand-exchanged QDs, SEM image of SQS prepared without sonication, a set of data corresponding to Fig. 1 for SQS emitting at 611 nm, temporal normalized EL spectra, R1–R9 data, and summary of the performances of fabricated white LEDs. See DOI: 10.1039/c5nr02991k |
This journal is © The Royal Society of Chemistry 2015 |