1. Introduction

I–III–VI₂ metal chalcogenide semiconducting compositions of A–In–S (A = Cu⁺, Ag⁺) have been actively explored for synthesis of fluorescent colloidal quantum dots (QDs) during the past decade. This type of ternary QDs has been further extended readily to the quaternary composition of A–Zn–In–S through Zn²⁺-mediated alloying, which is facilitated mainly by high structural similarities between chalcopyrite A–In–S and zinc blende ZnS phases [1–6]. Since the radiative recombination of such QDs occurs through the intragap defect states associated with Cu or Ag vacancy (V_Cu or V_Ag as acceptor), they commonly exhibit a nonexcitonic photoluminescence (PL) with a large Stokes shift by a few hundreds of meV [7–9], by which the inter-QD light reabsorption and the resulting self-emission quenching can be effectively suppressed. Another intrinsic PL characteristic of the above I–III–VI₂ QDs is a highly broad emission with typical bandwidths of >300 meV [9]. Taking advantage of these unique PL properties, they have been combined as downconverters mainly with a blue LED pumping source for the fabrication of high-color rendering white solid-state lighting devices [10,11]. Despite a broad PL of those downconverters white QD-LEDs packaged with a single type of QD, producing a bicolored white light of blue LED plus QD emission components, possessed only moderate color rendering indices (CRIs) owing to the shortage of white spectrum mostly in cyan and/or red region. On that account, co-use of different-colored two I–III–VI₂ QDs was inevitable to attain a more balanced white spectral distribution towards a higher CRI. For example, Chen’s group reported the fabrication of white QD-LEDs with excellent CRIs of 90–95 by co-packaging green Cu–Zn–In–S and red Cu–In–S (CIS) QD emitters in a blue LED [1]. And even better CRIs of 94–97 was also demonstrated by choosing another QD combination of green Ag–Zn–In–S and red Cu–Zn–In–S [12]. Although the above dual QDs-based packaging is advantageous over single QD-based one in achieving a high CRI, it not only entails the reduction of luminous efficacy caused by the light reabsorption and Förster resonant energy transfer between different-colored QDs but requires more complicated, expensive fabrication processes and a greater difficulty in reproducibility of white color chromaticity [13]. In this context, the simple packaging with single-typed QDs would be the most viable approach from the viewpoints of device performance and processing, as long as one secures highly fluorescent QD downconverters with a much broader PL feature than I–III–VI₂ ones aforementioned.

Doping of QD hosts with impurity ions such as Mn²⁺ and Cu⁺ leads to the modulation of PL with their absorption unchanged. When these transition metal ions are successfully doped into the excitonic transition-based II–VI (e.g., ZnSe, ZnS) and III–V (e.g., InP) QDs, the photo-generated host energy is efficiently transferred to dopant states. As a result, an excitonic PL becomes entirely quenched and a dopant-related lower-energy PL appears instead [14–17]. Meanwhile, in the case of I–III–VI₂ QDs with A–In–S or its derivative compositions, Mn doping may not quench V_Cu- or V_Ag-associated host defect emission completely, with the characteristic Mn emission of ⁴T₁–⁶A₁ d–d transition superposed [18–20]. Such concurrent emissions are understandable, taking into account that both PLs originate in nature from the lattice defects, to which the host excitation energy is transferred competitively. Mn doping has been implemented on CIS QDs to produce a Mn emission-superposed broader PL to ultimately demonstrate the fabrication of a better-color rendering white QD-LED. However, the white lighting device possessed only a moderate CRI (~83) primarily due to the limited white spectral coverage resulting from a substantial overlap (or insufficient spectral separation) between two PL bands of CIS QDs and Mn dopant [18]. Very recently, we reported highly efficient green QDs from another I–III–VI₂ composition of Cu–Ga–S (CGS) [8], which is larger in band gap than the above CIS. In this work, we further explore Mn doping into green CGS QDs via surface adsorption and lattice diffusion strategy. As a result of successful doping, PL of Mn-doped, ZnS-shelled CGS (CGS:Mn/ZnS) QDs, consisting of spectrally well-separated two major bands of CGS host defect- and Mn-related emissions, is modulated by varying Mn concentration. A series of CGS:Mn/ZnS QDs with different doping concentrations are then individually packaged as single downconverters in combination with a blue LED. Among white QD-LEDs fabricated as such, the best-color rendering device integrated with optimally doped QDs produces CRIs of 84–95, depending on input current of 60–250 mA.

2. Experiment sections

2.1 Synthesis of CGS:Mn/ZnS QDs

Mn doping was implemented by means of surface adsorption and lattice diffusion, where Mn species is first adsorbed onto the pre-grown core QD surface and then diffused inwards in the course of the subsequent shelling process. For Mn doping into CGS QDs different nominal concentrations of Mn/Ga = 0, 0.5, 1, 1.25 and 1.5 were attempted. For a typical synthesis of CGS:Mn/ZnS core/shell QDs with a Mn/Ga = 1.25, 0.167 mmol of CuI, 0.5 mmol of GaI₃ and 1 mmol of sulfur were mixed with 1.5 ml of 1-dodecanethiol (DDT) and 5 ml of oleylamine (OLA) in 50 ml of three-neck round flask, followed by vacuum degassing and N₂ purging. This precursor mixture was heated to and then maintained at 240°C for 5 min for the growth of CGS core QDs. A Mn stock solution, prepared beforehand by dissolving 0.625 mmol of Mn acetate tetrahydrate in 0.6 ml of OLA and 0.6 ml of 1-octadecene (ODE) at 180°C, was injected into the above CGS QD growth solution and this reaction was maintained for 1 min. And then, consecutive ZnS shelling, consisting of multiple steps to secure high quantum yield (QY), was applied and individual shelling steps were experimentally optimized as follows; for the 1st shelling, a ZnS stock solution comprising 8 mmol of Zn acetate dihydrate dissolved in 8 ml of oleic acid (OA) and 4 ml of (ODE) was introduced into Mn ion-adsorbed CGS QD solution above and the shelling reaction proceeded at 240°C for 1 h. Then, the 2nd ZnS solution, prepared by dissolving 4 mmol of Zn acetate dihydrate, 4 ml of OA, 2 mL of DDT and 2 ml of ODE, and the 3rd ZnS solution, prepared by dissolving 4 mmol of Zn stearate, 2 ml of DDT and 4 ml of ODE, were sequentially injected. The 2nd and 3rd shelling were performed at 240°C for 30 min and 250°C for 30 min, respectively. As-synthesized QDs were precipitated by the addition of an excess of ethanol and repeatedly purified by centrifugation with a mixed solvent of hexane/ethanol, and then re-dispersed in chloroform.

2.2 Fabrication of white QD-LEDs

For the fabrication of singe QD downconverters-based white QD-LEDs, 3 ml of the dispersion of CGS:Mn/ZnS QDs in chloroform with an optical density of ~3.0 at 450 nm was mixed with 0.5 g of thermally curable epoxy resin (YD-128, Kukdo Chem., Korea). This pre-mixture of QD-resin was dried at 60°C for 30 min to remove chloroform and then 0.5 g of a hardener (KFH-271, Kukdo Chem., Korea) was added. The resulting QD-paste was dispensed into a surface-mounted typed, blue InGaN LED (λ = 455 nm) mold and then hardened by a sequential thermal curing process of 90°C for 1 h and 120°C for 30 min.

2.3 Characterization and analysis

Absorption and PL spectra of a series of CGS:Mn/ZnS QDs were recorded with a UV−visible spectroscopy (Shimadzu, UV-2450) and a 500 W xenon lamp-equipped spectrophotometer (PSI Inc., Darsa Pro-5200), respectively. Their PL QYs were measured with an absolute PL QY measurement system (C9920-02, Hamamatsu) in an integrating sphere. The particle image of QDs was collected with a transmission electron microscopy (TEM, JEM-F200, JEOL) operating at 200 kV. A powder X-ray diffraction (XRD, Rigaku, Ultima IV) with Cu Kα radiation was employed to obtain information on crystallographic structure of QDs and formation of secondary phase. Actual Mn concentrations in CGS:Mn/ZnS QDs were analyzed with an inductively coupled plasma optical emission spectrometry (ICP-OES, OPTIMA 8300, PerkinElmer). Various electroluminescent (EL) data of white QD-LEDs fabricated such as EL spectrum, CRI, luminous efficacy (LE), correlated color temperature (CCT), and the Commission Internationale de l’Eclairage (CIE) color coordinates of white QD-LEDs was acquired with a diode array rapid analyzer system (PSI Co. Ltd) in an integrating sphere.

3. Results and discussion

Tetragonal chalcopyrite, the most common crystal structure in metal chalcogenide I–III–VI₂ semiconductors, is a slightly modified form of cubic zinc blende. Such an affinity in structure plus proximity in lattice constants, as marked in Fig. 1(a), between chalcopyrite CuGaS₂ core and zinc blende ZnS shell would facilitate the formation of core/shell heterostructure in an epitaxial manner. Typical XRD patterns of three CGS:Mn/ZnS core/shell QDs synthesized with Mn/Ga = 0, 1, and 1.5 were compared in Fig. 1(a), basically showing no marked difference one another. Not only appearance of reflection peaks from single phase only but overall peak shift towards zinc blende ZnS phase point to the suitable formation of ZnS overlayer on CGS core surface [10,11]. Even in the case of the heaviest-doped QDs with a nominal Mn/Ga ratio of 1.5, MnS as a secondary phase was not observed, indicating that an excess Mn species which was not adsorbed onto CGS core surface would not participate in the formation of a separate MnS phase and thus be washed off after the whole synthetic reaction. Actual Mn concentrations of a series of CGS:Mn/ZnS QDs, assessed by an ICP analysis of Fig. 1(b), were found to be far below nominal ones. These tiny doping quantities are likely conceivable taking a very short duration (1 min) of the present doping reaction into account. The sizes of undoped and doped core/shell QDs, examined by a TEM, were observed not to be different one another. Figure 1(c) is a TEM image of the representative CGS:Mn/ZnS QDs with Mn/Ga = 1, showing an average size of 4.8 nm. High crystallinity of the present QDs can be also verified by clear lattice fringes of an individual particle from the inset of Fig. 1(c).

 

Fig. 1 (a) XRD patterns of CGS/ZnS QDs synthesized with Mn/Ga molar ratios of 0, 1, and 1.5. (b) ICP-based actual Mn/Ga molar ratios of CGS:Mn/ZnS QDs with different Mn doping concentrations. (c) TEM image of the representative CGS:Mn/ZnS QDs with Mn/Ga = 1. The scale bar is 20 nm. The inset of (c) presents a magnified high-resolution TEM image of a single QD showing clear lattice fringes.

Download Full Size | PDF

As seen from Fig. 2(a), undoped CGS/ZnS QDs exhibited a broad, asymmetric PL that by and large comprises a dominant green emission peaking at 525 nm and a long tail emission. As inferred from the following PL peak decomposition of doped QDs of Fig. 2(b), PL of undoped QDs is a combined result of largely two intragap-related emissions; one likely stems primarily from the radiative recombination of an electron in conduction band (CB) with a hole trapped in V_Cu acceptor (referred to as CB–V_Cu), while the other may be associated mostly with the recombination of an electron trapped at donor states presumably in the form of Ga_Cu and/or Zn_Cu with a hole at the same acceptor state [21,22], commonly called donor–acceptor pair (DAP) recombination. The most direct evidence for successful Mn doping can be obtained by PL analysis. Figure 2(a) presents Mn concentration-dependent PL spectra of CGS:Mn/ZnS QDs with an identical optical density of 0.1 at 370 nm, showing that with increasing Mn dopant concentration Mn-related red spectral contribution escalated accordingly at the expense of green CB–V_Cu and DAP emissions. This proportionality of the increment of Mn emission to its concentration clearly indicates that the present doping strategy of surface adsorption and lattice diffusion was indeed effective and further implies that a newly introduced Mn dopant-related recombination would compete in PL gain with originally present lattice defect states-related CB–V_Cu and DAP ones. As recognized from Fig. 2(b), PL of the representative QD doped with Mn/Ga = 1.25 was decomposed into three emission components, i.e., CB–V_Cu at 518 nm, DAP at 621 nm, and Mn²⁺ intrashell transition of ⁴T₁–⁶A₁ at 625 nm. The radiative recombination channels corresponding to the respective emission components are also schematically illustrated in Fig. 2(c). PL excitation (PLE) spectra of the same QD sample as in Fig. 2(b) were recorded at the detected emission wavelengths of 518 and 625 nm, yielding the identical spectral profiles, as shown in Fig. 2(d). This suggests that both emissions result commonly from the excitation of QD host and thus Mn ion is appropriately doped inside QD host lattice [20,23]. As noticed from Fig. 2(e), regardless of Mn doping and its concentration, all the QDs exhibited nearly the same absorption characteristics with the identical absorption onsets at ca. 500 nm, confirming that impurity doping does not affect the absorption feature of host QDs. Moreover, Mn doping did not lead to the deterioration of PL QY, which was also found to be insensitive to Mn concentration, showing similar values of 74–76% in Fig. 2(f). This may imply that Mn doping did not generate the additional lattice defects which serve as nonradiative channels and PL reductions in CB–V_Cu and DAP recombinations were fully compensated by PL of Mn transition.

 

Fig. 2 (a) PL spectra and UV-irradiated fluorescent image (inset) of a series of CGS:Mn/ZnS QDs with different Mn/Ga ratios. (b) PL spectral decomposition result of Mn/Ga = 1.25-based CGS:Mn/ZnS QDs. (c) Schematic diagram of radiative channels for CB–V_Cu, DAP and Mn²⁺ transitions. (d) PL excitation spectra of Mn/Ga = 1.25-based CGS:Mn/ZnS QDs recorded with detected emission wavelengths of 518 and 625 nm. (e) Absorption spectra and (f) PL QYs of CGS:Mn/ZnS QDs with different Mn/Ga ratios.

Download Full Size | PDF

Five CGS:Mn/ZnS QDs with different Mn/Ga ratios were individually combined as single downconverters with a blue LED chip for the fabrication of white QD-LEDs. Figure 3(a) shows a set of EL spectra, comprising transmitted blue (455 nm) emissions and blue-to-QD downconverted ones, of as-fabricated white lighting devices collected at a driving current of 60 mA. As Mn²⁺-associated spectral protrusion became more distinct with increasing Mn concentration of QDs, white lights became warmer, as sensed from EL images in the insets of Fig. 3(a) and their CIE color coordinates steadily shifted to the red territory specifically from (0.267, 0.352) for undoped QD-LED to (0.344, 0.304) for Mn/Ga = 1.5-doped one, as expressed in Fig. 3(b), corresponding to CCTs of 8864 and 4905K, respectively. Meanwhile, the overall spectral coverage and balance became markedly improved towards natural (or high-color rendering) white light with increasing Mn concentration up to Mn/Ga = 1.25, showing a progressive enhancement of CRI from 64 for undoped device up to 95 for Mn/Ga = 1.25-doped one, as plotted in Fig. 3(c). This CRI of 95 is by far superior to those (73–83) achieved from typical single QD-packaged white QD-LEDs [10,11,18,20] and is equal to the record value from the device packaged with single downconverters of Mn²⁺- and Cu⁺-codoped Zn–In–S QDs [13]. In the case of Mn/Ga = 1.5-doped device, however, excessive Mn²⁺-associated spectral contribution rather deteriorated white spectral balance, resulting in only a moderate CRI of 84. LE of white QD-LEDs was found to be inversely proportional to Mn concentration, showing a steady LE drop from 45.4 lm/W for undoped device down to 32.4 lm/W for Mn/Ga = 1.5-doped one at an operating current of 60 mA. It is well-known that in the white light spectrum the increment of red or particularly deep-red (wavelengths longer than ca. 690 nm) component is detrimental to photometric quantities such as lumen (lm) dueto the low or negligible sensitivity of those photons to the human eye [24,25]. On that account, the LE reduction observed above is attributable to increasingly strengthened red and deep-red spectral contribution in the overall white EL spectrum as a result of the enhancement of Mn emission by a heavier doping. PL versus EL spectra (normalized to green emission component) from Mn/Ga = 1.25-based CGS:Mn/ZnS QDs were compared in Fig. 3(d), showing a distinct spectral mismatch. First, compared to PL, green emission peak of EL was notably red-shifted by ca. 15 nm. Though a large Stokes shift of the present I–III–VI₂ QDs can suppress self-quenching substantially, inter-QD light reabsorption is still likely, as inferred from a nontrivial spectral overlap between absorption of Fig. 2(e) and green PL component of Fig. 2(a). Moreover, compared to a dilute QD dispersion for PL, the QDs packaged in LED are highly concentrated and thus the interacting QDs are nearby sufficiently to promote the light reabsorption event [25], leading to a self-quenching accompanying a red shift of green EL component to some extent. Meanwhile, a slight red shift by ca. 6 nm of the red emission component (consisting of the superposition of Mn²⁺ plus DAP emissions) of EL relative to that of PL was also observable. However, this seeming red shift would be a subordinate result primarily from the spectral shift of green EL component above, taking into account that two lower-energy Mn²⁺ and DAP emissions are scarcely overlapped with absorption of CGS:Mn/ZnS QDs in Fig. 2(b) and thus the light reabsorption is very unlikely.

 

Fig. 3 (a) EL spectra and variations of (b) CIE color coordinates and (c) CRI of 60 mA-driving white QD-LEDs packaged with single-phased CGS:Mn/ZnS QD downconverters with different Mn concentrations. (d) Spectral comparison of PL and EL from Mn/Ga = 1.25-based CGS:Mn/ZnS QDs.

Download Full Size | PDF

Another distinct spectral difference in Fig. 3(d) is a green-to-red (G/R) emission peak ratio, specifically showing the G/R peak ratios of 1 and 1.42 for PL and EL, respectively. This lowered red emission intensity in EL versus PL may be hinted from the following input current-dependent EL results. Figure 4(a) presents as-collected EL spectra of Mn/Ga = 1.25-based white QD-LED with increasing input current up to 250 mA. The disparity in increment of the respective spectral components with input current could be better expressed by normalizing them relative to blue component, as shown in Fig. 4(b). This clearly shows that

 

Fig. 4 Input current-dependent (a) as-collected, (b) normalized EL spectra (relative to blue component), and variations of (c) CIE color coordinates and (d) CRI–LE of Mn/Ga = 1.25-based white QD-LED.

Download Full Size | PDF

Mn²⁺-associated red emission became quenched noticeably with increasing input current, while green one was insensitive in quenching to input current. Such a Mn emission quenching under a higher input current or larger photon flux can be ascribed to its intrinsically long excited state lifetime (i.e., a few milliseconds) [13,20,26], which are long sufficiently to induce the ground-state depletion of Mn²⁺ ion. Under this circumstance an excited Mn²⁺ ion to ⁴T₁ level can be further promoted successively to a higher ⁴T₂ level via spin-allowed transition, followed by nonradiative relaxation back to ⁴T₁ level [27]. In this sense, one may expect Mn emission quenching to a certain degree even under a low driving current of 60 mA, being responsible for a higher G/R ratio of EL versus PL of Fig. 3(d). Being in line with the red spectral depression with increasing input current from Figs. 4(a) and 4(b), white lights became colder, specifically showing 6417K at 60 mA to 7884K at 250 mA (also corresponding to CIE color coordinates of (0.313, 0.344) and (0.286, 0.339), respectively), and simultaneously their color rendering property deteriorated from 95 at 60 mA to 84 at 250 mA in CRI, as shown in Figs. 4(c) and 4(d), respectively. LE of the white QD-LED was also strongly dependent on driving current, exhibiting a steady decrease from 34.5 lm/W at 60 mA down to 20.2 lm/W at 250 mA, as presented in Fig. 4(d). This LE reduction is universal in QD-LED and can be generally ascribed to two intrinsic events of efficiency drop of a blue LED chip and thermal quenching of QD emitters [10,25], both of which become intensified at a higher input current. Moreover, in the case of our Mn-doped QD-based LED, an increasing degree of Mn emission quenching at a larger photon flux is jointly responsible for LE reduction at a higher input current.

4. Conclusion

We presented an effective doping of Mn²⁺ ion into CGS host QDs by means of surface adsorption and lattice diffusion. This Mn doping yielded the red PL band associated with Mn^{2+ 4}T₁–⁶A₁ transition, which was concurrent and competitive with radiative intragap (i.e., CB–V_Cu and DAP) recombinations of undoped CGS/ZnS QDs. Mn²⁺-associated red spectral contribution in overall PL was readily controllable with and proportional to dopant concentration. Regardless of Mn doping and its concentration undoped and all doped QDs exhibited similar levels in PL QY of 74–76%. The respective CGS:Mn/ZnS QDs with different Mn concentrations were tested as single downconverters for the fabrication of solid-state lighting device. When Mn/Ga = 1.25-doped QDs were adopted, a high-color rendering white QD-LED with an exceptional CRI up to 95 was attainable, which is the highest quantity reported among white lighting devices with single-phased QD emitters. Then, Mn/Ga = 1.25-based white QD-LED was further operated under different driving currents, revealing the marked quenching of Mn emission compared to the green component (associated mostly with CB–V_Cu recombination) under a high input current or large photon flux due to its intrinsic long excited-state lifetime. This increasing spectral depression of Mn emission at a higher input current led to the deterioration of color rendering property and further partly contributed to the reduction of LE, showing CRIs and LEs of 95, 34.5 lm/W at 60 mA and 84, 20.2 lm/W at 250 mA, respectively.