1. Introduction

Since colloidal quantum dots (QDs) were investigated by L. E. Brus et al. [1], colloidal QDs have attracted keen attention as color saturated, highly luminescent and robust lumophors, due to their unique optical properties, such as facile bandgap tunability, wide absorption range, spectral purity and photo-/chemical stability. Among various colloidal QDs based on II-VI (i.e., CdS [2], CdSe [3–7], CdTe [8,9], ZnSe [10], or ZnO [11]), III-V (i.e., InAs [12] and InP [13]), IV-VI (i.e., PbSe [14,15] and PbS [16–20]), and IV(i.e., Si [21–24]) compounds, II-VI QDs comprising cadmium and zinc chalcogenides have shown considerable progress in terms of high color purity (full width at half maximum: 20 ~40 nm) and photoluminescence quantum efficiency (PL QE, above 70%) in the visible range. Such a success in materials science harnessed the development of colloidal QD-based light-emitting diodes (QD-LEDs) aiming toward economic, stable, and high performance next-generation devices, which will potentially replace conventional inorganic or organic based-LEDs.

The basic concept of colloidal QD-based LEDs is not much different from conventional LEDs: the radiative recombination of charge carriers supplied from two electrodes within the QD active layers. However, different material properties, in terms of size (below 10 nm), materials formulation (inorganic nanocrystals), and phase (solution phase), largely influence fabrication processes, device structures and their relevant physics. For example, colloidal QD films are typically fabricated by solution processes such as spin-coating, inkjet printing or dip coating because they are too heavy to be deposited with thermal evaporation method (i.e., molecular weight of 5 nm CdSe QD is about 18,300 g/mol). The limited process capability, in turns, affects to the choice of charge carrier transport layers: sub-layers (i.e., charge injection or transport layers formed prior to QD deposition) should have tolerance against a solvent containing QDs, and possess sufficient mechanical strength and adhesion property to substrates. Besides the processing, unique electronic properties of QDs originating from their structural and compositional nature are also critical factors to design device structure. Valence and conduction band of QDs are located relatively lower than the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) of organic semiconductors and, thus, conventional hole transport materials used in organic LEDs should be replaced with newly designed materials to facilitate the injection of carriers and the balanced exciton recombination within QDs.

As mentioned above, replacement of conventional lumophors in LEDs with QDs is not simple. The new materials (i.e., QDs), device structures, and fabrication processes are closely connected each other and only the optimal solution among various possibilities yields successful QD-based LEDs. Emphasizing these cross-related aspects of QD-LEDs, this review addresses recent advances in preparation of highly luminescent quantum dots, structural design for electroluminescence devices, and printing processes for QD-LEDs and display applications.

2. Fundamental aspects and nanostructure of quantum dots for efficient lumophors

2.1. Physical background and optical properties of quantum dots

QDs exhibit unique electronic properties depending on their size and shape, which have never been observed in bulk materials. In the case of bulk semiconductors, electrons are excited vertically to the “continuous” conduction band when semiconductors absorb photons with higher energy than the bandgap of materials (Fig. 1(a) , left), and, thus, the absorption spectrum of a bulk semiconductor shows monotonous and continuous increase above the optical bandgap energy of materials. However, if the spatial dimension of materials is reduced, the wavefunction of charge carriers (i.e., electron and hole) is spatially restricted (dimension = a), and the allowed wavenumbers for the optical transition, |k| = 2π/λ, gets quantized to discrete states which only satisfy the boundary condition (Fig. 1(a), right). This effect, the quantization of electronic structure in nanometer regime, is referred as the quantum confinement effect. At the strong confinement regime where the physical size of QDs are smaller than the bulk Bohr exciton size (a < a_B), the electron and hole wavefunctions of QDs are described by hydrogen atom-like symmetry, named as S, P, and D electronic states with 2-, 6-, and 10-fold degeneracy (Figs. 1(b) and 1(c)). Because of these similarities in electronic features with hydrogen atom, QDs are frequently described as artificial atoms.

 

Fig. 1 (a) The band structure for bulk semiconductor and nanocrystals with cubic lattice around k = 0. This illustration is reconstructed from reference [25]. (b) Atom-like S, P, and D orbitals of spherical semiconductor nanocrystals. Reprinted from reference [26] with permission. © 2009 American Chemical Society. (c) A schematic on building up of electrons in a strongly-confined QD (E_S: single-electron energy level; E_C: charging energy; E_e-e: Coulomb repulsion). This illustration is redesigned from reference [27].

Download Full Size | PDF

Colloidal QDs have been normally synthesized by thermal decomposition of organometallic precursors in hot reaction medium. In common synthetic methods, anionic precursors are rapidly injected into solution containing cationic precursors at elevated temperature, so-called as hot injection method (Fig. 2(a) , left). The hot injection method facilitates the instantaneous supersaturation of reactant over critical point for nucleation in very short period, which results in the monodisperse QDs during growth period (Fig. 2(a), right). Nucleation and growth kinetics depend on the reactivity of precursors, reaction variables (i.e., temperature and concentration), and surfactants (i.e., fatty acids, fatty amines, alkylphosphonic acids, or trialkylphosphine or trialkylphosphine oxides), and their mechanical aspects have been finely resolved by H. Weller, P. A. Alivisatos, or X. Peng [5,28–32].

 

Fig. 2 (a) (left) A schematic on conventional hot-injection method for group II-VI QDs. (right) The temporal change in degree of supersaturation (The LaMer plot). Rapid injection of precursors (region I) results in sudden supersaturation over critical point, resulting in burst nucleation of QDs (region II). In following period, QDs grow to finite size (region III), depending on the reaction parameters. ((b) Size-dependent optical spectra for monodisperse CdSe QDs. (c) An absorption spectrum of CdSe QDs and the manifolds of excitonic states.

Download Full Size | PDF

The quantized electronic states of monodisperse QDs are easily observed in optical transitions. The absorption edge and excitonic transition energy of QDs shift to a higher energy as the size of QD decreases (Fig. 2(b)). The size-dependent energy trace follows the quantum confinement effect of charge carriers and, thus, can be described as a function of 1/a². Another representative size-dependent feature of QDs is the wavy signature in the extinction coefficient (Fig. 2(c)), which reflects the density of state participating in the electronic transition from a ground state to excited states (i.e., 1S_3/2-1S_e at the first peak implies that the electron is excited from 1S_3/2 valence band to 1S_e conduction band) [33].

Although the particle-in-a-box (or sphere) framework provides an instinctive picture to understand the effect of quantum confinement on optical properties of QDs, such as bandgaps or quantization of electronic states of QDs, this approach does not fully resolve fine electronic structures and corresponding physical properties of QDs [34]. For better understanding A. L Efros [35,36], M. G. Bawendi and D. J. Norris [37,38] formulated multi-band effective mass approximation (EMA) for QDs. Based on several assumptions (i.e., strong-confinement regime, separable electron and hole wavefunctions, and location of those wavefunctions at the conduction band minima and valence band maxima), they were successful to unravel the fine electronic structure and corresponding optical properties of QDs, verified with theoretical calculation and experimental analysis.

Besides the size-dependent electronic features, QDs possess unique optical properties compared with other nanomaterials, for example broad absorption above the optical bandgap and narrow photoluminescence spectra. These distinguished features, which enables efficient light conversion from broad range of incident photons to photons with narrow spectral bandwidth (FWHM < 35 nm), harness their applications toward next-generation displays or lightings. The PL line has a symmetric Gaussian shape, which is distinguishable feature of QDs compared from conventional lumophors (e.g., organic molecules or inorganic phosphors) showing vibronic sub-peaks and long tail or phonon-related broad emission.

Until now, the optical properties of colloidal QDs are briefly introduced. Although major researches on physical properties of QDs are constrained on QDs with rather specific chemical compositions, such as cadmium chalcogenide or lead chalcogenide, the general understandings can be extended and applied to different types of QDs: the bandgap tunability based on quantum confinement effect, quantization of electronic states and corresponding absorption and emission properties. From the view point of QD-based LEDs, understanding the underlying physics in QDs and exploiting their unique optical properties are essential steps to realize high performance devices. In following sections, the state-of-the-art synthetic approaches to engineer the exciton dynamics for highly efficient and color-saturated QDs will be discussed.

2.2. Surface passivation of QDs through core/shell construction

Gaussian-shape PL is an expected photophysical result from the ensemble measurement of QDs at room temperature, as a result of 1) inhomogeneous broadening due to size and shape distribution within QDs and 2) homogeneous broadening due to the thermal energy (26 meV at room temperature). However, the different features in PL spectra of colloidal QDs are frequently observed. Figure 3(a) is a common absorption and emission spectrum of CdS QDs measured at room temperature. Although the absorption spectrum with well-resolved excitonic transition features signifies narrow size and shape distribution of QDs, the emission spectrum shows a sharp band-edge emission and unexpected broad emission at lower energy than the band-edge PL energy (denoted with a red arrow in Fig. 3(a)). This broad PL is known as surface trap emission or surface state emission. The QDs with surface state emission typically possess poor PL QE (below 5%) compared with the one with solely band-edge emission due to the non-radiative relaxation process through the surface states.

 

Fig. 3 (a) An absorption and photoluminescence spectrum of CdS QDs. The trap emission is denoted as a red arrow. (b) Schematics on the effect of surface states on the recombination process. The band structure as a function of k and the other relaxation processes are omitted for the simplicity. If the surface states are not passivated (left, bare QDs), an electron at the conduction band edge can be trapped at the surface states, resulting in a broad and weak surface state emission or nonradiative decay. On the other hand, if the surface states are passivated by organic or inorganic shells, then the trap emission is eliminated and only band-edge emission is occurred.

Download Full Size | PDF

As the size of QD decreases, the ratio of surface area to total volume is drastically increased. In nanoscopic view, the formation of solid surface results in the breakdown of periodic crystal lattice, that is, the chemical bonding between atoms are disconnected and remains as electron lone pair or vacant sites. For the colloidal QDs, dangling bonds on surface are bound with organic ligands with donating or accepting electron pair (i.e., amine, carboxylic acid or phosphonic acid, phosphine or phosphine oxide) from the synthesis. Due to the insufficient passivation with organic ligands, they are easy to be exposed to the environment and result in undesired non-radiative carrier relaxation processes or chemical reaction. These sites are referred as surface traps or surface states. In the energy band diagram, surface states broadly reside within the bandgap of QDs (as illustrated in Fig. 3(b), left), and thus the surface state emission has broad spectral features at lower photon energy regime than the band-edge emission [39,40]. It is obvious that the elimination of surface states (Fig. 3(b), right) or preventing non-radiative decay pathways via surface states improve the QE of QDs and enhance the color purity of QDs.

There are two general approaches to passivate the surface states of QDs for the enhanced QE and improved photochemical stability: 1) passivating surface dangling bonds with organic ligands or 2) passivating surface states with inorganic shells. Although the surface passivation with organic ligands is generally simple and straightforward (i.e., addition of ligands into QD solution), the organic capping is vulnerable to external stresses. For instance, primary amines, which efficiently passivate the hole trap states in CdSe QDs, are easily desorbed from the surface by adsorption-desorption equilibrium [41], and, thus, it does not guarantee reliable QE over time. Moreover, the addition of other organic ligands or solvents may change the density of original ligands on surface (i.e., thiols, frequently used functionality for surface modification), and easily affects the PL QE of QDs [42].

In order to enhance the reliability and stability of QE, the surface passivation with inorganic shells has been developed. This structural engineering of QDs in a core/shell formulation is regarded as a major advance in QD synthesis, because it provides robustness and tolerance to QDs against physical and chemical stresses enabling QDs applied to various research fields. Ever since A. R. Kortan et al. suggested the CdSe/ZnS core/shell QDs synthesized in inverse micelle media [45], numerous efforts have been followed to realize the core/shell structured QDs with high QE and stability.

The luminescent property of core/shell structured QDs can be engineered with relative position of core bandgap and shell bandgap. If core QD with a smaller bandgap is located within shell material with a larger one, the electron and hole wavefunctions are confined into core (type-I bandgap configuration, Fig. 4(a) ). In this case, the recombination probability of two wavefunctions increases while the non-radiative decay process via surface states decreases, and, thus, the PL QE of QDs increases. By contrast, if two bandgaps are staggered (type-II bandgap configuration, Fig. 4(b)), then two wavefunctions are spatially dissociated and the energy from radiative decay are determined by the difference between spatially-different energy levels, which is smaller than the bandgap of the core or shell. The type-II bandgap configuration give a huge opportunity to manipulate exciton recombination or dissociation process, so it is a strong candidate for photovoltaic applications [46]. From the QD-LED perspectives, type-I bandgap configuration is favored, and various types of semiconductor materials are realized in core-shell structures with type I configuration [47–50]. The information (i.e., bandgap and the electronic energy levels) for group II-VI, III-V and II-VI semiconductors are depicted in Fig. 4(c) [43,51].

 

Fig. 4 Schematics on (a) a type-I bandgap configuration and (b) type-II bandgap configurations of core/shell QDs. The type-I bandgap configuration confines electron and hole wavefunctions in a same space, improving the recombination probability and it produces the band-edge emission. While the type-II bandgap configuration divides electron wavefunctions spatially, as a result, the probability on the radiative recombination is reduced and photon energy is the difference between the conduction (valence) band of core and valence (conduction) band of shell. (c) Electronic energy levels of several group II-VI, III-V, IV-VI, VI semiconductor materials using the valence-band offsets from reference [43,44].

Download Full Size | PDF

Due to the ease of chemistry, semiconductor materials of group II-VI materials were firstly studied prior to group III-V and IV-VI. We will mostly review the advances in chemistry of group II-VI QDs with type I configurations to address general approaches for the synthesis of highly efficient and stable QDs.

Figure 5(a) represents that the evolution of optical spectra of the CdSe/CdS QDs during the shell formation. As the shell thickness increases, the QE of QDs rapidly increases from 1% to 84% and the overall optical spectra shifted a little bit to lower energy. The enhancement in QE and the red-shift in optical spectra are common phenomena during the core-shell formation. In the framework of “particle-in-a-box” model, the surroundings of bare QDs can be approximately regarded as infinite potential wall, implying the strict confinement of exciton wavefunctions inside the core (Fig. 5(b), top). By contrast, the shell is overgrown on the core, the height of potential wall is decreased and the exciton wavefunctions (In particular, electrons are delocalized widely due to smaller effective mass than holes) can be delocalized to the shell region by tunneling (Fig. 5(b), bottom). The extent of tunneling depends on the band offset between core and shell as well as the effective mass of charge carriers. As a result of sufficient protection of charge carriers by the shell with higher energetic barrier, the chance for charge carriers to leak out to the surface states decreases and, therefore, the photoluminescent efficiency as well as the photostability of QDs significantly increases, as illustrated in Fig. 5(c) [48].

 

Fig. 5 (a) Absorption (dashed line) and PL (solid line) spectra of CdSe/CdS QDs with 30 Å of CdSe core. The increase in QE (or Q.Y., quantum yield) and of coverage of CdS with each injection is also depicted. (b) Schematic illustration of potentials (solid lines) and electronic energy levels (dashed lines) of core (top) and core/shell (bottom) QDs. “x” represents the absorption onset for a CdSe core of 34 Å diameter, “y” for a core/shell with a same core diameter and a 9 Å thick shell. The conduction band offset is 0.27 eV, while the valence band offset is 0.51 eV. (c) Photostability comparison of core and core/shell QDs (The shell thickness is 7 Å). Absorption spectra of core (top) and core/shell (bottom) samples before (solid) and after (dashed) continuous wave irradiation at 514 nm with an average power of 50 mW for approximately 2 h. The solutions were saturated with oxygen and had identical optical densities at the excitation wavelength at the start of the experiment. Reprinted with permission from reference [48]. © 1997 American Chemical Society.

Download Full Size | PDF

Although CdSe nanocrystals with thin CdS shell show high QE and enhanced stability against physical and chemical stress, CdS is still vulnerable to photo-oxidation in presence of water or oxygen. Also the conduction band offset between CdSe and CdS is not sufficiently large to confine electron wavefunctions into core region, electron can be trapped to the surface states, leading to the PL QE lose and degradation of QDs. For the enhanced stability and reliability against various physical and chemical stresses, the formation of thick shell with larger bandgap and better chemical stability (e.g., ZnS: 3.61 eV) was needed. However, the direct introduction of ZnS on top of CdSe core did not give successful results. As shown in Fig. 6(a) , the PL QE of CdSe/ZnS QDs increases at the initial thin shell growth stage (below 2 atomic monolayer of ZnS), but dramatically decreases afterwards. The peak QE of CdSe/ZnS QDs is much lower (50%) than the case of CdSe/CdS [45,52–54]. The experimental results and theoretical calculation explained that the large lattice mismatch between core and shell materials gives rise to the interfacial defects and irregular shell growth (Fig. 6(b)) [53].

 

Fig. 6 (a) PL spectra of ZnS overcoated CdSe QDs with 42 ± 10% Å diameter. The spectra are for a) 0, b) 0.65, c) 1.3, d) 2.6 and e) 5.3 monolayers ZnS coverage. The spectra broaden with increasing ZnS coverage. The change in Q.Y. as a function of ZnS coverage is illustrated in an inset. Reprinted with permission from reference [53]. © 1997 American Chemical Society. (b) Z-STEM of CdSe/ZnS QDs with a QE of 34% (top). The signal from heavier core material and lighter shell material are colorized as yellow and red, respectively. The line profile shows the interface between core and shell clearly (bottom). A scale bar is 3 nm. Reprinted with permission from reference [55]. © 2006 American Chemical Society.

Download Full Size | PDF

For the uniform shell formation of higher bandgap with mitigated interfacial defects, the multi-shell structure (i.e., core-shell-shell) was proposed by D. T. Talapin et al (Fig. 7 ) [56]. The insertion of 2~3 atomic monolayers of CdS or ZnSe buffer layers between CdSe core and ZnS outer shell sufficiently mitigate the lattice strain, while retaining the type I bandgap configuration (Fig. 7(a)). The effect of the ZnSe or CdS intervening layer on optical properties of QDs is remarkable. While CdSe/ZnS QDs shows drastic decrease in QE after 2 monolayers of ZnS shell deposition, CdSe/ZnSe/ZnS QDs exhibit high QE above 70% even with thick ZnS shell formation above 5 monolayers (Fig. 7(b)). The multi-shell structured QDs with high PL QE and enhanced stability against physical and chemical stresses hastened the practicable applications of QDs toward LEDs or biomarkers.

 

Fig. 7 (a) i) schematics on the core-shell-shell QDs, ii) corresponding energy level diagram, and iii) relationship between bandgap energy and lattice parameter of bulk CdSe, ZnSe, CdS, and ZnS semiconductors with wurtzite phase. (b) PL QE of CdSe, CdSe/ZnSe, and CdSe/ZnSe/ZnS nanocrystals dissolved in chloroform at room temperature. For comparison, the dependence of PL QE on the shell thickness for various samples of CdSe/ZnS nanocrystals is shown. Reprinted with permission from reference [56]. © 2006 American Chemical Society.

Download Full Size | PDF

2.3. QDs with alloyed formulations

The emission wavelength of QDs, technically, can be modulated by changing the size of core and composition, but, in practice, it requires complicate and bothersome efforts for adjusting nucleation and growth kinetics; controlled reaction time, use of multiple ligands, or additional injection of precursors [12,57]. These difficulties have been critical to synthesize blue-emitting QDs based on CdSe, because CdSe QDs below 2 nm are hardly achievable with common reaction conditions. Limited by the lack of practical experiences in chemistry, orange or red-emitting core/shell QDs had been mainly synthesized in the early QD synthesis.

In order to overcome the practical limits in bandgap tunability of binary composition, alloyed QDs formulated with three or more elements has been developed [58]. The alloyed semiconductors generally exhibit mixed physical properties between the distinct original ones, in terms of the bandgap, exciton Bohr radius and lattice parameter. For example, incorporation of wide bandgap semiconductor (i.e., ZnSe or ZnS) into the narrow bandgap QDs (i.e., CdSe) enables the synthesis of QDs with larger bandgap, which is particularly useful to synthesize deep-blue emitting QDs. Besides the extended bandgap tunability, alloyed structures within QDs provide additional benevolent effects: 1) the alloyed interface composed of core and shell semiconductor can serve as a lattice adaptor, and 2) alloyed structured QDs possess less sensitive to inhomogeneous broadening and thus have narrow emission spectral bandwidth [59].

For the realization of homogeneous alloy QDs, several methods have been introduced, such as, thermal energy-assisted diffusion [60–62] and simultaneous reaction of three or more precursors with similar reactivity [63–71]. The thermal diffusion-assisted alloying method was firstly suggested by Z. Xinhua et al. [63] in a formulation of Zn_xCd_1-xSe QDs. Because of weaker bond strength of Zn-Se than Cd-Se, Zn can diffuse into CdSe phase at elevated temperature (over 270 °C), and vice versa [72]. By varying the Zn contents, wide range of emission window from near UV to red could be successfully realized in Zn_xCd_1-xSe QDs [60–62].

The simultaneous reaction of three or more precursors has been more widely used to synthesize alloyed QDs particularly comprising two or more of anions. Since anions possess larger ionic radius than cations within the crystal structure, the infusion-assisted alloying of anions hardly or very slowly occurs. Based on the subtle control of precursor reactivity, ternary or quaternary QDs have been successfully synthesized through the simultaneous reaction in various formulations of Zn_xCd_1-xSe [67], Zn_xCd_1-xS [63,68], CdSe_xSe_1-x [66], CdSe_xTe_1-x [73], ZnCdSSe [69] (group II-VI), InPZnS [71], InAs_xP_1-x [64] (group III-V), PbS_xSe_1-x, PbS_xTe_1-x or PbSe_xTe_1-x [70] (group IV-VI), encompassing the wide range of spectral window from near-UV, visible (Fig. 8(a) ), near-IR region (group III-V, Fig. 8(b)) to near-IR region (group IV-VI, Fig. 8(c)).

 

Fig. 8 (a) (left) STEM-EDS line scan along a single ~10 nm Zn_0.6Cd_0.4S_0.5Se_0.5 QDs, (middle) structural model of the QD lattice projected along the <001> orientation (cyan: Se, blue: S, red: Cd, green: Zn) and (right) composition-dependent photoluminescence spectra of ZnCdSSe QDs: 1 (0.90, 0.89), 2 (0.80, 0.71), 3 (0.69, 0.59), 4 (0.41,0.40), 5 (0.25, 0.24) and 6 (0.11, 0.10), where (x, y) is (Zn / (Cd + Zn), S / (S + Se) in QDs). Reprinted with permission from reference [69]. © 2009 American Chemical Society. (b) the bandgap of bulk materials (dashed line) and the emission peak of InAs_xP_1-x QDs as a function of arsenic content. Reprinted with permission from reference [64]. © 2005 American Chemical Society. (c) Calculated size- and composition-dependent bandgap of PbS_xS_1-x alloyed QDs. Bandgap in eV is notated on each contour line. Reprinted with permission from reference [70]. © 2010 American Chemical Society.

Download Full Size | PDF

2.4. Core/shell QDs with composition gradient interface

The stepwise change in lattice parameter improves the structural uniformity in shell growth and mitigates the interfacial defects originating from the abrupt structural difference. Although the insertion of buffer layer reduces lattice stress between core and the outermost shell materials (addressed in section 2.2), there still is lattice stress between core and buffer layer and buffer layer and the outermost shell, which possibly trigger the interfacial defect unless precise thickness of uniform shell is realized. In order to further reduce the lattice stress within QDs and exploit the best optical properties, QDs with chemical compositional gradient were suggested.

QDs with gradual changes in chemical composition from core to shell was firstly suggested by R. Xie et al. [74]. For the realization of such QDs with smooth interface, they have adopted successive ion layer adhesion and reaction (SILAR) method. The SILAR method is originally invented to the deposit thin solid films uniformly on solid substrates in solution bath [75]. By alternating deposition of a cation monolayer and an anion monolayer, R. Xie et al. introduced Zn_0.5Cd_0.5S buffer layer between CdS and ZnS shells. Due to the mitigated lattice stress between CdS and ZnS with intervening Zn_0.5Cd_0.5S alloyed layer, QDs showed very uniform shell morphology (Fig. 9(a) ) as well as high monodispersity in size and shape. The resulting QDs exhibited very high QE around 80% and also chemical/structural robustness against an intense purification or photo-oxidation (Figs. 9(b) and 9(c)).

 

Fig. 9 (a) A TEM image of CdSe QDs covered with 2 monolayers of CdS, 3.5 monolayers of Zn_0.5Cd_0.5S, and 2 monolayers of ZnS. (b) Reduction of the relative photoluminescence QE on repeated precipitation and redispersion of TOPO/ODA-covered CdSe cores and several ODA-covered core/shell particles in chloroform solution. (TOPO: trioctylphosphine oxide, ODA: octadecylamine) (c) Photochemical stability of QDs in oxygen saturated chloroform solutions under UV-irradiation. (top) Change in optical density of QD dispersion and (bottom) change in QE for CdSe core and different core/shell QDs. Reprinted with permission from reference [74]. © 2004 American Chemical Society.

Download Full Size | PDF

Although the SILAR is promising method to control the composition of shells, it requires long reaction time and multiple reaction steps. Instead of the discrete alloyed buffer layer by SILAR method, easier methods to achieve the composition gradient structure were developed based on 1) thermal energy-assisted infusion or 2) reactivity difference of precursors. For instance, W. K. Bae et al. made a smooth alloyed interface between Cd_xZn_1-xS core and ZnS shell by thermal annealing at elevated temperature (i.e., 310 °C) [68]. The interfacial alloying leads to the small extent of blue shift (less than 10 nm) in optical spectra (Fig. 10(a) ), but it causes a dramatic enhancement in QE: Cd_xZn_1-xS/ZnS nanocrystals with composition gradient structure exhibited QE above 80%, while Cd_xZn_1-xS/ZnS nanocrystals with abrupt interface only showed QE below 40% (Figs. 10(b) and 10(c)). This result is a good example of facile and reproducible synthetic method to realize the composition gradient structure within QDs by utilizing thermal energy-assisted atomic infusion.

 

Fig. 10 (a) PL emission wavelengths and (b) QE of Cd_xZn_1-xS/ZnS nanocrystals during heat treatment experiment. Blue arrows denote the increase in reaction temperature to 310 °C for the thermal treatment. (c) Schematics on the Cd_xZn_1-xS/ZnS nanocrystals with (top) alloyed interface and (bottom) discrete interface. Reprinted with permission from reference [68]. © 2008 American Chemical Society.

Download Full Size | PDF

Another synthetic method to formulate the compositional gradient structure is utilizing the reactivity difference between precursors. The reactivity of organometallic precursors is typically determined by differences in metal-ligand bond strength and a steric factor. In certain reaction conditions, such as extremely high temperature or the use of highly reactive precursors (i.e., organometallic compounds or elemental sulfur), the reactivity difference is negligible and resulting product has a homogeneous composition [63,69]. However, in general, precursors used in the synthesis of colloidal QDs shows a considerable difference in reactivity. For example, the bond strength of P-Te (280 ± 10.0 kJ/mol) is smaller than that of P-Se (364 ± 10.0 kJ/mol) [77], so the reaction involving coexistence of cadmium oleate, trioctylphosphine selenide and trioctylphosphine telluride precursors results in QDs with CdTe-rich core and CdSe-rich gradient shell.

By utilizing the reactivity difference, W. K. Bae et al. realized CdSe@ZnS QDs with composition gradient in a single-step hot-injection process (Fig. 11(a) ). In the early stage, cadmium oleate and TOP-Se pair leads the nucleation of a CdSe rich core and less reactive zinc oleate and trioctylphosphine sulfide participate in the formation of shell, resulting in the gradual chemical composition gradient along the radial direction (i.e., Cd and Se rich core phase to Zn and S rich shell phase) (Fig. 11(b)). Due to the mitigated lattice mismatch between CdSe core and ZnS outermost shell, QDs with composition gradient show high QEs above 80%, and also maintain their luminescent property even after surface treatment with various organic ligands. This method also enables bandgap tunability (PL peak wavelength from 500 nm to 610 nm) by simply changing the chemical contents introduced to the synthesis (Fig. 11(c)). Because of the simplicity in the synthesis (i.e., one-step reaction) as well as the superior optical properties of resulting structures, this method has been extended to the synthesis of QDs with different chemical composition (i.e., CdS_xTe_1-x [78], Zn_xCd_1-xS [79], Zn_xCd_1-xSe/ZnSe QDs [80], InP/ZnS [81] and InP/ZnSeS [82]).

 

Fig. 11 (a) (left) Schematic on the possible reaction mechanism for the single-step synthesis of QDs with chemical composition gradient, (middle) probable chemical composition and (right) electronic energy level of QDs. (b) Ratio of (left) Cd (blue) or Zn (yellow) to (Cd + Zn) and (right) that of Se (purple) or S (green) to (Se + S) for each shell from the center of the QDs. (c) Room temperature photoluminescence for different QDs prepared by the single-step synthesis. Reprinted with permission from reference [76]. © 2008 American Chemical Society.

Download Full Size | PDF

Ever since colloidal QDs were reported in early 1980s, ceaseless endeavors have been devoted to realize highly luminescent QDs for almost 30 years and recently PL QE up to the unity is reported in II-VI QDs [83]. Besides the high PL QE in solid state, the advances in synthesis enables to achieve QDs with exceptional color purity of QDs (PL FWHM < 40 nm), which has never been achieved by conventional organic/inorganic lumophors (Fig. 12 ). Due to these superb optical properties, QDs are regarded as the most promising next generation lumophors in display or lighting industries.

 

Fig. 12 The CIE (Commission Internationale de l’Eclairage) 1931 chromaticity diagram of hypothetical QDs emitting 420, 450, 470, 490, 510, 530, 550, 570, 590, 610, 630 and 650 nm (from the left) with 20 nm (red dot), 30 nm (green square) or 50 nm (blue triangle) of FWHM. NTSC 1987 (solid line) and 1953 (dashed line) color gamut are also illustrated. The emission spectra of QDs are assumed as Gaussian shape.

Download Full Size | PDF

Although the advances in the synthesis of group II-VI QDs provide a positive outlook, the hazardous chemical components (i.e., Cd or Pb) hamper the commercialization of QDs-based applications. In order to avoid such harmful substance, the research on group III-V QDs (i.e., InP-based QDs) is accelerated. Unfortunately, the current status of III-V QDs is still embryonic level and the technological advances for these materials are not matured. However, the physical background and the synthetic approaches to improve QE and stability are similar with these of group II-VI QDs. Recent progress on the development of InP/GaP/ZnS core/shell/shell QDs [84], composition gradient InP/ZnS QDs [81], or InPZnS alloyed QDs [71] supports such prospective. Novel synthetic strategy for size-controlled, monodisperse, and surface-passivated core/shell QDs will raise the possibility of non-toxic group III-V QDs. Based on this positive outlook, research on the QD-based electroluminescence device or down conversion light emitting diodes has been accelerated recently. In the next section, we will discuss about the major advance in QD-LEDs driven by electricity and look out the future of QD-LEDs.

3. Structural consideration and optimization of QD-LEDs

3.1. Multilayered structure of QD-LEDs for efficient electroluminescence

QD-LEDs can commonly be classified as electroluminescence (EL) devices or down-conversion devices. The down conversion LEDs utilizes QDs as phosphors that converts part of high energy incident photons originating from substrate LEDs into lower energy ones (Fig. 13(b) ). For commercial applications, green and red-emitting QDs are dispersed in optically clear and thermally stable polymeric resins and those composite resins are deposited on top of blue-emitting LEDs to produce white light consisting of three discrete colors with high color purity. Several companies such as Nexxus Lighting, LG, or Samsung are now trying hard to adopt those passive devices in conventional liquid crystal displays as back light units. In contrast to down-conversion LEDs with single-phase QD composites, QD EL devices require multilayered device structures stacked with other semiconducting materials (i.e., hole transporting layer (HTL) or electron transporting layer (ETL)) in between electrodes (Fig. 13(a)) for balanced and efficient carrier injection into QDs. Hereinafter, QD-LEDs refer QD-based EL devices, and recent progresses in the architecture of QD-LEDs are mainly addressed in this review.

 

Fig. 13 Schematics on simplified device structures of (a) an electroluminescence device and (b) a down-conversion device based on QDs.

Download Full Size | PDF

The fundamental principle of device operation for QD-LEDs is largely same as commercialized electrically-pumped LEDs (i.e., inorganic or organic LEDs): The light-emitting active materials are embedded between two electrodes in the form of thin layers. When electrons (or holes) are injected into the emission layer through a cathode with low workfunction (or an anode with high workfunction), these charged carriers are injected into the emission layer and radiate photons by radiative exciton recombination process. For the injection of electrons to conduction band/LUMO levels, low workfunction cathodes such as Mg, Ca, Mg/Ag, or LiF/Al are frequently used, whereas O₂-plasma treated indium tin oxide (ITO) has often been employed for high workfunction transparent anodes.

In order to prevent undesired exciton quenching within the active materials and also lead to the balanced carrier injection into the active materials, recent EL devices introduce one or mutiple components between the active layer and the electrode, as illustrated in Fig. 14 . For instance, a hole transporting layer (HTL) (in Fig. 14(a)) could prevent exciton quenching by the direct contact with the anode electrode and lower the hole injection barrier from an anode. An electron injection layer (ETL) also has the similar role, modulating electron concentration in the emission layer and preventing the exciton quenching by the direct contact with the cathode electrode. Therefore, the introduction of two or more (Figs. 14(b) and 14(c)) charge transport layers in EL devices conveniently enables the facilitated/balanced carrier transport and injection into the active layers, and also enhances radiative recombination within the active layer [85].

 

Fig. 14 Schematics on the energy diagrams of QD-LEDs reported. (a) A bilayer system consisting of a HTL and a QD layer. (b) The introduction of ETL between a QD layer and a cathode. (c) The insertion of a hole injection layer (HIL) in the middle of a transparent conducting oxide (TCO) and a HTL.

Download Full Size | PDF

The first QD-LEDs employed the multilayered structure of ITO as the high work-function anode, conjugated polymer (poly(p-phenylenevinylene) (PPV)) as a hole transporting layer (HTL), CdSe QDs as emitting materials, and Mg as a low work-function cathode (ITO/PPV/CdSe QDs/Mg) [86]. They displayed electroluminescence spectra clearly recognized as contributed by CdSe QDs, but device performance was not promising (i.e., EQE: 0.001 ~0.01% and luminance: ~100 cd/m²). Although material developments (i.e., CdSe/CdS [87] or CdSe/ZnS QDs [88]) improved both brightness by 10-fold of luminance and EQE up to 0.22% in the same device structure, the QD-LEDs still suffered from low PL efficiency and stability and the parasitic EL spectra contributed by PPV layers. This poor device performance is mainly attributed to the imbalanced carrier injection and recombination of excitons.

The success of QD-LEDs containing HTL and ETL was realized by S. Coe et al. [89]. They utilized N,N’-diphenyl-N,N’-bis(3-methylphenyl)-(1,1’-biphenyl)-4,4’-diamine (TPD) as HTL and (tris-(8-hydroxyquinoline)aluminium) (Alq₃) as ETL (ITO/TPD/QD/Alq₃/Mg) (Fig. 15(a) ). Their device structure (HTL/QD/EBL/ETL, EBL: electron blocking layer) demonstrated a significant improvement in QD-LEDs; they suggested the complete device structure to facilitate the carrier injection into QDs or the efficient Förster type energy transfer from neighboring organic layers to QDs and the recombination in the emission layer. In order to realize the multilayered device structure, they took advantage of the phase segregation of QDs from organic molecule layer (i.e., TPD) for the HTL/QD layers, followed by consecutive thermal evaporation of EBL/ETL/cathode layers. Because of low miscibility between TOP/TOPO-capped QDs and TPD layers, QDs are excluded from the TPD matrix layer and migrate to the top surface with closely-packed morphology (Fig. 15(b)). Although the QD monolayer realized in this report has several defects, causing parasitic emissions from the TPD layer, the introduction of a thin electron blocking layer (TAZ, 10 nm) considerably reduced the emission from organic layers (Fig. 15(c)). As a result, this device led to the significant improvement in terms of peak EQE (0.52% at 10 mA/cm²) and luminous efficiency (1.6 cd/A at 2000 cd/m²), which was the 25-fold improvement compared with previously reported QD-LEDs. Moreover, further optimization of QD monolayer [90], QD synthesis, and the introduction of new HTL materials in the HTL/QD/EBL/ETL device structure led to efficient green [91] and red QD-LEDs [92], representing enhanced EQE over 2% and saturated emission spectrum.

 

Fig. 15 (a) EL spectra and device structures of two kinds of QD-LEDs without TAZ layer (left) and with TAZ (right) as a hole-blocking layer. Dashed lines represent the deconvolution of the EL spectra into Alq₃ and QD components. The QDs used are illustrated in the inset that is composed of CdSe core (~38 Å in diameter) coated with 1.5 monolayers of ZnS. PL QE was 22 ± 2%. (b) An AFM phase image of a complete, hexagonally packed QD monolayer segregated from an underlying TPD layer. Grain boundaries between ordered domains of QDs are shown. (c) A proposed energy level diagram of an EL device shown on the left of (a). Reprinted with permission from reference [89]. © 1998 Nature Publishing Group.

Download Full Size | PDF

In contrast to the HTL/QD/EBL/ETL device structure shown above, another important progress in fabrication process and advanced device structure has been made since 2006. In order to facilitate the device fabrication, polymeric HTL materials such as poly(N,N’-bis(4-butylphenyl)-N,N’-bis(phenyl)benzidine) (polyTPD) [93] or thermally cross-linkable polyTPD-derivatives (i.e., PS-TPD-PFCB [94] or BiVB-MeTPD [95]) have been adopted as HTLs. These HTLs have good resistance to organic solvents in QD solutions, thus QD mono- or multilayers can be easily realized with a simple spin coating process. While the QD phase segregation technique requires subtle control on both solvent choice and concentration in the mixture to fabricate QD thin layers [90], this simple spin-coating process could yield homogeneous QD layers in large area without extensive void defects. In the aspect of the device structure, poly(3,4-ethylenedioxythiophene) doped with poly(styrenesulfonate) (PEDOT:PSS) was employed as the hole injection layer (HIL). The PEDOT:PSS has several roles in the device performance: the reduction in surface inhomogeneity of ITO substrates and the elevation of work-function (5.0 ~5.2 eV) [95]. The presence of HIL is critical to lower the turn-on voltage, owing to the reduced injection barrier from ITO/PEDOT:PSS to TPD below 0.3 eV [93] (Fig. 16(a) ).

 

Fig. 16 (a) Energy levels of materials involved in the QD-LEDs ITO/PEDOT:PSS/polyTPD/QD multilayer/Alq₃/Ca/Al. (b) EL spectra of red emitting QD-LEDs as a function of operation voltage with the device structure illustrated in (a). The inset shows the image of a large-area device under operation. (c) Luminous efficiency and power efficiency of the QD-LED as a function of luminance. Reprinted with permission from reference [93]. © 2007 Nature Publishing Group. (d) UV-vis and PL spectra of QDs, EL spectrum of a QD-LED (ITO/PEDOT:PSS/QD(green) multilayer/TPBi/LiF/Al) and a photograph of a QD-LED with a pixel size of 1.4 mm x 3.65 mm (turn-on voltage: 3.5 V, luminous efficiency: 5.2 Cd/A, maximum brightness: 10,000 cd/m²) Reprinted with permission from reference [96]. © 2009 Wiley-VCH. (e) Normalized PL spectra (dashed line) and EL spectra (solid line) of QD-LEDs (ITO/PEDOT:PSS/polyTPD:CBP/QD(blue) multilayer /TPBi/LiF/Al). The QD-LED pixel with 1.4 mm x 3.65 mm (left top) and CIE chromaticity color indices of QD-LEDs (right) are also shown in the insets. Reprinted with permission from reference [97]. © 2009 Institute of Physics. (f) (top) Composite photographs of 0.6 x 1.9 mm² QD-LED pixels for blue, skyblue, green, orange, and red. (bottom) PL spectra (dashed lines) of QD monolayers for red, orange, and green QDs and QDs in hexane solution for blue and skyblue QDs due to lack of absorption of blue and skyblue QD monolayers at wavelengths over 350 nm and EL spectra (solid lines) of QD-LEDs (device structure: ITO/PEDOT:PSS/spiro-TPD/QD monolayer/TPBi/Mg:Ag/Ag). Reprinted with permission from reference [98]. © 2009 American Chemical Society.

Download Full Size | PDF

Based on the improved device structure, Q. Sun et al. reported color-saturated red, orange, yellow, and green emitting QD-LEDs (Figs. 16(a)-16(c)) [102]. The important change in their system is the QD monolayer replaced with two or more QD layers. Thinner emissive layer has possibility to cause undesired emission from organic layers in the device due to current leakage through voids, grain boundaries, and interstitial space within QD monolayer. The optimal thickness of the QD emission layer varies with the size and shell structure of QDs, which is, in turn, attributed to different carrier injection barriers. That is to say that small QDs (i.e., green QDs) have larger difference in the HOMOs of HTL and QDs, requiring large amount of QDs to capture free carriers into QDs. The optimal thickness of QD emission layers were found to be 2, 2.5, 4, and 7 monolayers for red, orange, yellow, and green-emitting QLEDs, respectively. Through the improvement of the device structure and the optimization of HTL, ETL, and QD multilayer, they demonstrated significant improvement in device performance in terms of luminous efficiency, turn-on voltage, and lifetime. The optimized devices showed low turn-on voltage (3 ~4 V), improved luminous efficiency (1.1 ~2.8 Cd/A, Fig. 16(c)), saturated EL spectrum in the entire operation voltage in large area (1.5 cm x 2.5 cm, Fig. 16(b)), and long operation lifetime (half-life: 300 h at 13 V). Moreover, the maximum luminance was extended to 9064, 3200, 4470, 3700 cd/m² for red, orange, yellow and green QD-LEDs, respectively.

Based on such HIL/HTL/QD multilayer/ETL device structure, W. K. Bae et al. realized color saturated green [96] and blue [97] QD-LEDs in 2009. In the previous reports, QDs employed in device fabrication could not maintain PL QE after repeated purification steps [93], mainly due to the desorption of surface ligands followed by the formation of defect sites. In order to overcome this problem, Bae et al. utilized highly stable alloyed QDs [68,76] with composition gradient for the device fabrication. Due to the thick protecting shells, the prepared QDs showed a high PL QY even after repeated purification process. Because of improved quantum efficiency of QDs, the green emitting QD-LEDs exhibited significant enhancement in efficiency (luminous efficiency: 5.2 cd/A, EQE: 1.4%) as well as brightness (1,500 cd/m² at 7.7V, maximum brightness: 10,000 cd/m²) (Fig. 16(d)). In the case with blue-emitting QD-LEDs, they employed a composite HTL layer containing polyTPD and CBP in order to facilitate the hole injection into QDs. Because of the enhanced energy level matching between the composite HTL and QDs, the surface-state emission, originating from the exciton recombination at the surface states, could be effectively suppressed (< 1% of the total EL emission). The fabricated devices showed moderate turn-on voltage (6 V), EQE (0.1 ~0.3%), and color-saturated emission in the deep-blue region (Fig. 16(e)). Immediately after these reports, P. O. Anikeeva et al. reported full-color QD-LEDs based on the similar device structure incorporating a QD monolayer (ITO/PEDOT:PSS /spiroTPD/QD monolayer/TPBi/Mg:Ag/Ag) with more enhanced device performance: the peak EQEs were 1% for red, 2.7% for orange, 2.6% for green, 0.2% for cyan, and 0.4% for blue QLEDs (Fig. 16(f)) [98].

Although the operation of QD-LEDs has been demonstrated many times, their device performance is still far inferior to the cases with conventional organic or inorganic LEDs. Therefore, fundamental studies on the mechanism of EL involving QDs as emission layers are required to increase the IQE of QD-LEDs. There are generally two possible EL mechanisms for QD-LEDs: 1) direct carrier injection into QDs followed by the radiative recombination of excitons; 2) exciton formation in organics followed by the Förster energy transfer to QDs [99]. Besides, the Auger process related to QD charging [100] and the electric field-assisted exciton dissociation (Onsager exciton dissociation) [101] have been considered as major processes to limit the IQE of QD-LEDs.

These two EL mechanisms seem to coincide in QD-LEDs consisting of organic HTL/QD/organic ETL device structures. However, in some cases, the mechanism 2) has been considered as a major route for the emission in QD-LEDs [99]. It is believed that it originates from the large hole injection barrier from the HOMO level of HTL to the valence band of QDs. While the conduction bands of QDs used in the QD-LEDs are sufficiently low (below 4 eV) that electron injection from low work-fucntion electrodes or organic ETLs into QDs easily occurs, the valence bands of QDs are generally located lower (below 6 eV) than the HOMO energy level of common organic semiconductors (typically, 5 ~6 eV) and the hole injection is thus largely inhibited by this energetic barrier. As a result, excitons are not effectively formed within QDs, but instead HTL or ETL domains as a result of the charge carrier leakage through grain boundaries, interstitial sites, or vacancies in QD active layers. If excitons generated in HTL or ETL have larger energy than the energy of QDs, satisfying the energy transfer condition, the spectral overlap between the emission of a donor and the absorption of an accepter along with proper separation distance (i.e., Förster distance, ~11 nm for Alq₃ or TPD thin layer [99]), then the excitons firstly formed in organics are transferred to QDs through the non-radiative Förster energy transfer process [99]. In this operation mode, accurate placement of QD monolayer within the range of Förster energy transfer and balanced carrier injection are required to maximize the EQE of QD-LEDs.

Although the device operation based on the Förster energy transfer is a useful concept to overcome the hole injection barriers of current HTLs used in QD-LEDs, the internal quantum efficiency is inevitably limited by incomplete energy transfer. Therefore, the reduced contribution from the energy transfer and the increase in direct carrier injection into QDs guarantee high performance QD-LEDs. As we discussed above, the major hurdle for the direct injection into QDs has been regarded as the large hole injection barrier from HTL to QDs. The effect of lowered hole injection barrier has already been demonstrated by W. K. Bae et al. [97], thus, it is evident to develop new HTL materials with HOMO energy levels relevant to QDs for high-performance QD-LEDs. Moreover, detailed investigation on exciton quenching (i.e., Auger process) and deterioration mechanism (material stability and degradation process) should be conducted for practical applications. In the next section, we will discuss new approaches to overcome limitations imposed by materials or device structures of previous QD-LEDs.

3.2. Advances in carrier transport materials and device structures

Since QD-LEDs containing organic HTL and ETL have been reported, the basic operation mechanism and requirements for improving the device performance is mostly resolved. In particular, lowering the hole injection barrier from HTL to QDs is believed to be the critical factor to facilitate the direct carrier injection into QDs. Unfortunately, most hole transport materials used in organic LEDs or previous QD-LEDs have irrelevant HOMO position against the HOMO level of QDs, thus development of novel hole transport materials for QD-LEDs is highly desirable to realize high performance devices. In order to overcome the limited HOMO levels of typical organic semiconductors, inorganic carrier transport materials have been suggested since 2005. For example, A. H. Mueller et al. utilized p-doped and n-doped GaN injection layers for colloidal QDs fabricated by a novel technology, energetic neutral atom beam lithography/epitaxy (ENABLE). Although EQEs of their devices lie between 0.001 ~0.01%, their approach is meaningful in that colloidal QDs can be incorporated into all-inorganic carrier transport materials, similar to previous inorganic-LEDs.

Metal oxides, prepared by RF magnetron sputtering or atomic layer deposition [103], are much more robust than organic semiconductors against thermal stress and degradation by oxygen or moisture. Moreover, some of metal oxides have lower energy level compared with organic semiconductors [104], thus they are attractive candidates as new hole transport materials for QD-LEDs. J. M. Caruge et al. employed amorphous NiO and ZnO:SnO₂ as HTL and ETL for QD-LEDs based on RF magnetron sputtering (Figs. 17(a) and 17(b)) [102]. Although this all-inorganic device represented color-saturated EL spectra (Fig. 17(c)) and stable operation at high operation voltage (Fig. 17(d)), the overall performance in terms of turn-on voltage (3.8 V for red) and efficiency (maximum EQE: 0.09%, peak luminous efficiency: 0.064 cd/A) is much inferior to organic-based LEDs. The low performance of all-inorganic QD-LEDs is attributed as follows: exciton quenching of QDs in contact with the metal oxide layer due to high carrier concentration (~10¹⁴ /cm³) as well as the Auger nonradiative recombination caused by QD charging [105]. It seems that the favorable band alignment for electron injection and the large amount of charged carriers in operation facilitate the QD charging. Although all-inorganic colloidal QD-based LEDs still have poor device performance in spite of new hole transport materials with low VB (i.e., WO₃, 6.3 eV) and conductivity control (peak EQE: 0.2%, with ITO/NiO/QD multilayer/ZnO/ZTO/ZnS/ZTO/metal device structure) [105], stable operation over 15 V is highly relevant in realizing electrically-pumped lasing.

 

Fig. 17 (a) A schematic on the device structure and (b) a band diagram determined from UV photoemission spectroscopy and optical absorption measurements, denoting approximate electron affinities and ionization energies of QD-LED materials. (c) Electroluminescence spectra of the QD-LED of (a) at 6 V (0.46 A/cm²) and 9 V (1.14 A/cm²) applied bias. (d) EQE measured from the front face of the device as a function of current density, J. The maximum EQE of 0.09% and a luminance of 1,500 cd/m² were reached at 13.8 V and 2.33 A/cm². The inset shows a photograph of a bright and uniform pixel at 6 V applied bias. Reprinted with permission from reference [102]. © 2008 Nature Publishing Group.

Download Full Size | PDF

In contrast to the approaches taken toward all-inorganic colloidal QD-LEDs based on RF sputtering process, the use of solution-processed inorganic ETL and their combination with organic HTL have been suggested by several research groups [104,106,107]. K.-S. Cho et al. fabricated QD-LEDs using sol-gel based TiO₂ thin films as ETL (Fig. 18(a) ) [104]. Owing to the low band offset to aluminum electrode (0.4 eV) and higher electron mobility of the sol-gel TiO₂ film compared with Alq₃ layer (TiO₂: 1.7 x 10⁻⁴ cm²/V s, Alq₃: ~1.0 x 10⁻⁵ cm²/V s), TiO₂-based QD-LED represents an electron-dominant device and the highly enhanced electron injection (Fig. 18(b)). Moreover, it caused the decrease in turn-on voltage below the bandgap of QDs. They also adjusted other components in the QD-LEDs to realize the carrier balance and lower the hole injection barrier. In order to match the increased carrier density of the TiO₂-based QD-LED, the doping level of PEDOT:PSS HIL was adjusted by increasing the amount of PSS. The improved charge balance resulted in the enhancement of maximum luminance and luminous efficiency values of 12,380 cd/m² and 1.67 cd/A. Moreover, they also reduced the hole injection barrier from HTL to QDs by using primary amine-based crosslinkers. The crosslinking of QD multilayers followed by thermal annealing lowered the hole injection barrier around 0.6 eV and, at the same time, improved the contact between QDs and the hole transport layer. After extensive optimization of the components in the QD-LEDs, the TiO₂-based QD-LEDs exhibited high luminance, low turn-on voltage, and power efficiency (2.41 lm/W) (Fig. 18(c)). Significantly improved device performance led to fabricate a practical display with an active matrix drive backplane and a monochromatic QD-LED display was successfully demonstrated (Fig. 18(d)).

 

Fig. 18 (a) Energy level diagram of QD-LEDs based on sol-gel processed TiO₂ ETL. (b) Current density (J) versus voltage (V) characteristics of the QD-LEDs. A QD-LED with TiO₂ ETL depicted in (a) (black solid line), a QD-LED with Alq₃ as the ETL (blue broken line), and a reference, TiO₂-based QD-LED without QD layer (red broken line). Three different conduction regimes are apparent in the reference device (ohmic, trap-limited, and space-charge-limited conduction) whereas the other devices show partial (for the TiO2-based QD-LED) or no change (for the Alq₃-based Qd-LED) from trap- to space-charge-limited conduction due to large trap densities. The device turn-on voltage of the Alq₃-based QD-LED is larger (4.0 V) than the TiO₂-based QD-LED (1.9 V). (c) Luminous efficiency, EQE, and power efficiency as a function of luminance. (d) A display image of 4-inch crosslinked QD-LED using an amorphous-Si thin film transistor backplane with a 320 x 240 pixel array for the active matrix drive. The upper right inset is an image of light emission from all pixels under operation at 500 cd/m² and the lower right inset shows each pixel. Scale bar is 100 μm. Reprinted with permission from reference [104]. © 2011 Nature Publishing Group. (e) Energy level diagram for ITO/PEDOT:PSS/polyTPD/QD multilayer/ZnO NCs/Al. (f) EL spectra of blue, green, red-orange QD-LEDs with photographs (inset). Reprinted with permission from reference [106]. © 2011 Nature Publishing Group.

Download Full Size | PDF

Further progress in QD-LEDs based on organic HTL/QD/inorganic ETL device structure was recently reported by L. Qian et al. [106]. Although the TiO₂-based ETL mentioned above has the advantage of all-solution process capability as well as improved electron mobility, it requires a bothersome annealing process to hydrolyze TiO₂ precursors and its electron mobility is still poor owing to the amorphous phase of TiO₂ formed by the sol-gel process. L. Qian et al. solved this problem by using ZnO nanoparticles, which have higher electron mobility (2 x 10⁻³ cm²/V s) and better energy level alignment with Al electrode (Fig. 18(e)). Utilizing PEDOT:PSS and polyTPD as HIL and HTL, they achieved more improved device performance for blue, green and orange-red QD-LEDs in terms of maximum luminance (4,200 cd/m² for blue, 68,000 cd/m² for green, and 31,000 cd/m² for orange-red) and power efficiency (0.17 lm/W for blue, 8.2 lm/W for green, and 3.8 lm/W for orange-red) (Fig. 18(f)). Moreover, the device lifetime was extended over 250 hr with the initial brightness of 600 cd/m² under low vacuum, implying that the stability of carrier transport materials is one of the most important factors to extend the lifetime of QD-LEDs.

The use of ZnO ETL resulted in the huge improvement of device performance as well as lifetime. However, the hole injection barrier between HTL and QDs is still too large to induce the direct injection of carriers into QDs. Although the QD-LEDs fabricated by L. Qian et al. exhibited low turn-on voltage, it is mainly attributed to the Auger-assisted hole injection. Unfortunately, the Auger process consumes a large amount of carriers and also acts as the major non-radiative process, thus the device performance based on the Auger process is inevitably limited. Therefore, a proper choice of hole transport material is required to realize QD-LEDs operated by the direct-injection of carriers.

J. Kwak and W. K. Bae solved the limitation of HTLs by constructing an inverted device structure [108] (be published soon). In conventional device structure, an ITO electrode have solely been employed as an anode and the deposition of hole injection and transport materials on top of the anode should be able to realize the successive deposition of QD mono- or multilayers, which has largely hindered the choice of various hole transport materials. In contrast to such conventional ITO/HIL/HTL/QD layer structure, they placed a ZnO NC thin film on an ITO substrate as EIL/ETL followed by the deposition of a QD multilayer. Due to the inverted device structure, various kinds of hole transport materials of superior electrical properties with proved stability, typically in small molecules, can be easily deposited on top of the QD multilayer (Fig. 19(a) ). Through the inverted device structure as well as the proper choice of a hole transport material with a low HOMO level (CBP), the hole injection into QDs was significantly improved, as supported by the nearly flat EQE over the wide range of current density up to 1 mA/cm² (Fig. 19(c)). The improvement in the balanced carrier injection into QDs has direct consequences on the performance of QD-LEDs: peak EQE, peak luminous efficiency, and maximum luminance were 7.3%, 5.7 cd/A, and 23,040 cd/m² for red; 5.8%, 19.2 cd/A, and 218,800 cd/m² for green; 1.7%, 0.4 cd/A, and 2,250 cd/m² for blue (Fig. 19(b)). Moreover, their devices could be fabricated in large area (1.2 cm x 1.2 cm) with saturated red, green, and blue emissions (Fig. 19(e)), and the lifetime of the inverted QD-LEDs was shown to extend to hundreds of hours (half-life: 590 hr at 500 cd/m² of initial brightness), as demonstrated in Fig. 19(d).

 

Fig. 19 (a) Energy band diagram of inverted QD-LEDs, where electrons are injected from ITO and holes are injected from Al. (b) Maximum EQEs of red, green, and blue QD-LEDs using various HTLs with different HOMO energy levels. Higher EQEs were obtained as the HOMO energy level of HTL is close to the valence band of QDs. (c) EQE versus current density of red, green, and blue QD-LEDs with CBP as a HTL. (d) The lifetime characteristics of red QD-LEDs with standard (ITO/PEDOT:PSS/polyTPD/QD multilayer/TPBi/LiF:Al) and inverted device structure (ITO/ZnO/QD multilayer/CBP/MoO₃/Al). (e) Photographs of red, green, and blue QD-LEDs (with the emitting area of 1.2 cm x 1.2 cm) at applied voltages of 2.6 ~3.3 V displayed in the inset. Reprinted with permission from reference [108]. © 2012 American Chemical Society.

Download Full Size | PDF

The inverted device structure also has the great advantage in the fabrication of active matrix displays. In the realization of displays with layered current driven-LEDs (i.e., organic-LEDs), the fabrication of backplane TFTs with low-cost, high mobility, operation stability, and availability of both n- and p-type property is one of major technical issues. The commercialized organic-LEDs have employed low-temperature polysilicon (LTPS) as a bottom anode due to high carrier mobility as well as the availability of p-type TFT property. However, the LTPS-based TFTs requires complex fabrication process and they are limited by the size of available substrates [109], thus robust a-Si-based technology has been considered as an alternative because of low production cost with large-size substrates available. Because the inverted device structure uses the conventional ITO electrode as a cathode, n-type amorphous-Si or metal-oxide based TFTs, with low manufacturing cost and good uniformity, can be easily employed as backplane TFTs. Consequently, the transparent bottom cathode of the inverted QD-LEDs can be directly assembled with the drain line of n-type TFTs and such configuration could effectively lower the driving voltage [109].

Most reports on QD-LEDs have focused on the optimization of bandgap alignment between a QD layer and charge transport layers. However, those multilayered devices require multiple deposition steps based on either solution process or vacuum deposition, causing the increase in production and labor costs along with more defects likely formed during fabrication. In the following paragraphs, we will discuss novel approaches based on new materials and relevant device structures.

3.3. Novel materials and device structures

In many device fabrication processes, the QD and HTL layers are distinctively separated. In order to construct such a discrete interface, a crosslinkable HTL or an orthogonal solvent has been employed. As mentioned earlier, the phase separation of QDs out of hole conducting materials has once been used to have a distinctive interface between them but this process condition is not easy to generalize for other systems involving different materials and also very difficult to reproducibly produce large area displays adopting typical solution processing techniques such as drop-casting or roll-coating method. M. Zorn, W. K. Bae and associates came up with a novel material (QD hybrids) combining both high luminescence of QDs and extensive processability of polymers [110,111]. They designed and synthesized diblock copolymers consisting of hole conducting moieties (triphenylamine, TPA) and an anchoring block bound to the QD surfaces (cysteamine acrylamide, CAA) via the reversible addition-fragmentation chain transfer polymerization (RAFT) (Fig. 20(a) ). Those block copolymers are preferably associated with QDs through the short CAA blocks, yielding QD-polymer hybrids.

 

Fig. 20 (a) A schematic diagram of QD/poly(p-methyltriphenylamine-b-cysteamine acrylamide) (PTPA-b-CAA) hybrids. (b) The energy band diagram of QD-LEDs employing QD/PTPA-b-CAA hybrid emissive layers. (c) Normalized EL spectra of QD-LEDs including QD/PTPA-b-CAA hybrid emissive layers with different QD contents (0.5, 1.0, 1.5, 2.0, and 2.5 wt% for H1, H2, H3, H4, and H5). (inset) Magnified EL spectra from 440 nm to 480 nm. EL spectra were measured at a current density of 150 mA/cm². (d) (left) Plan-view and (right) cross-sectional TEM images of (top) QD/PTPA-b-CAA hybrid and (bottom) QD/PTPA-b-PFP blend films spun-cast with solutions containing 1 wt% polymers and 2.5 wt% QDs. (e) Cross-sectional TEM images of (top) a drop-cast QD/PTPA-b-CAA hybrid film and (bottom) a QD/PTPA-b-PFP blend film. (f) A fluorescence microscopy image of a QD/PTPA-b-CAA hybrid film with regular hole patterns (hole diameter: 1 μm, hole distance: 0.3 μm) prepared by the capillary force lithography. Reprinted with permission from reference [110]. © 2009 Wiley-VCH.

Download Full Size | PDF

The presence of short anchoring blocks in the block copolymer results in the drastic difference in the morphology of composite films. Due to the favorable interactions between QDs and the anchoring blocks, QDs were uniformly distributed in the hybrid polymer matrix (Fig. 20(d), top) while a blend film, QDs embedded in a polymer matrix without anchoring blocks, shows massive phase separation of QDs from hole conducting polymers (Fig. 20(d), bottom). This QD/conducting polymer hybrid material could eliminate the deposition of a separate HTL (Fig. 20(b)), apart from previous device structures. The device employing a QD/conducting polymer hybrid layer showed up to 2.2% of EQE in the range of practical brightness (100 ~1000 cd/m²) as well as the saturated color (Fig. 20(c)), outperforming the best green-emitting QD-LEDs based on similar device structure (ITO/PEDOT:PSS/QD layer/TPBi/Lif/Al) [96]. On top of improved device performance, the QD/conducting polymer hybrid also has a significant importance in terms of favorable processability for novel printing techniques. The QD/conducting polymer hybrid exhibited homogeneous morphologies even in thick films prepared by the simple drop-casting (Fig. 20(e)), which has a great advantage in inkjet printing requiring complex ink formulation and process optimization. Moreover, several unconventional lithographic techniques can be easily applied due to polymer-like properties of QD/conducting polymer hybrids. As an example, they demonstrated a hole-patterned hybrid film in the submicron range, as shown in Fig. 20(f).

W. K. Bae et al. also employed the layer-by-layer assembly with charged QDs for QD-LEDs [112]. In earlier studies on LEDs, the layer-by-layer assembly was used to realize PPV-based HTL layers [88] or QD emission layers [113,114], demonstrating the usefulness of thin film fabrication involving various nanomaterials based on the electrostatic interaction. However, the use of insulating polyelectrolytes between such functional materials has lowered the conductivity of assembled multilayers and, as a consequence, previous research showed poor EQE and low brightness compared with the common device structure mentioned in Section 3.2.

In order to eliminate insulating polymers in the layer-by-layer assembled films, W. K. Bae et al. modified the surface of QDs with positive and negative charges and fabricated all-QD multilayers. They deposited QD multilayers using the spin-assisted layer-by-layer assembly by the alternative deposition of positively- and negatively-charged QDs using a spin-coater, as illustrated in Fig. 21(a) . Since the multilayer deposition of QDs depends on the electrostatic attraction between opposite charges, the internal composition of QD films could be precisely controlled by changing the sequence of QD colors. Moreover, the fabricated thin film shows uniform morphology with a RMS roughness around 4 nm. Based on a simple device structure (Fig. 21(b)) without a separate HTL, they obtained reasonably good EQE (0.3% at 50 mA/cm²) with the maximum brightness (800 cd/m²) in the large area (3.4 x 3.8 cm²) when compared with previous QD-LEDs based on the layer-by-layer deposition [113,114], presumably due to better electronic structure of all-QD multilayers. Although the device performance is still lower than common devices with HTL, further optimization of the device structure involving the replacement with conducting brushes and suitable bandgaps is expected to increase the device performance.

 

Fig. 21 (a) A schematic on the preparation of all-QD multilayer films fabricated by spin-assisted layer-by-layer assembly. (b) Energy band diagram of QD-LEDs based on ITO/PAH (anode)/all-QD multilayer film/TPBi (ETL)/ LiF/Al (cathode). (c) (left) A schematic on the device structure and (right) an EL spectrum showing emitting QD layers (green QD layers) adjacent to the top TPBi layer. The inset shows an image of the QD-LED and corresponding CIE indices of the EL spectra. (d) (left) Fabrication scheme and (right) a photograph of a QD-LED exhibiting multiple colors (green, orange, and red) in a unit device (pixel size of 1.4 x 1.4 mm²). Reprinted with permission from reference [112]. © 2010 American Chemical Society.

Download Full Size | PDF

The easy patterning capability of all-QD multilayer-based LEDs is another merit in their research. With a series of all-QD multilayers assembled with green and red QD layers located at various positions across the film, they observed that the EL emission mainly originates from the QD monolayer adjacent to the TPBi layer and the energy transfer between two QDs with different bandgaps was not observed (Fig. 21(c)). This is very useful for solution-based printing processes such as inkjet printing or microcontact printing. Figure 21(d) demonstrates multicolored all-QD-based LEDs constructed on the same substrate with the inkjet printing. By arranging QDs with different emissions near the top of the multilayer, they were able to effectively integrate three different colors in the same device. This simple demonstration would lead to full-color QD displays based on all-QD multilayers and the inkjet process.

In the meantime, V. Wood et al. suggested a novel device structure mainly driven by electric field. Up until now, most of QD-LEDs have been driven by current, thus the alignment of bandgaps between carrier transport materials and QDs is crucial to achieve high efficiency. However, they instead suggested the n-i-n unipolar QD-LED structure [115], enclosing the QD multilayer with n-type ZTO layers by RF sputtering (ITO/ZTO (40 nm)/QD multilayer (50 nm)/ZTO (15 nm)/ZnS (30 nm)/ZTO (40 nm)/ITO, as illustrated in Fig. 22(c) ). Because of the large bandgap of metal oxide above 3 eV, the appearance of this device is transparent, as shown in Figs. 22(a) and 22(b). More interestingly, the dominant mechanism of electroluminescence is the field-driven ionization of QDs. As illustrated in Fig. 22(c), a large bias induces the accumulation of electrons at the interface between n-type layers (i.e., ZnS or ZTO), thus voltage drop is focused on the QD multilayer. If the electric field is sufficiently large, electrons and holes are then generated by the dissociation of excitons by the Onsager mechanism (namely, field-induced ionization of QD). An electron dissociated from one QD can recombine with a hole in another QD, resulting in the radiative recombination. By inserting an intervening ZnS buffer layer in the ZTO layer, they were able to prevent undesired QD charging, minimizing the decrease in current density and QD quenching. As a result, they demonstrated the EQE and the maximum brightness up to 0.15% and 1,040 cd/m², respectively. These field-driven QD-LEDs were also successfully demonstrated even with thin insulting Al₂O₃ or SiO₂ and such devices could be operated by AC modulation [116].

 

Fig. 22 (a) An absorption spectrum of unipolar device: ITO/ZTO (40 nm)/QD multilayer/ZTO (15 nm)/ZnS (30 nm)/ZTO (40 nm)/ITO, where ZTO is composed of ZnO:SnO₂. The inset photograph shows the device on top of text to demonstrate the transparency of wide bandgap ceramics and a thin QD layer. (b) A photograph of device operation at 18 V, demonstrating the uniformity of pixel illumination as well as device transparency. (c) A schematic band diagram of device structure in (a) under forward bias condition. (d) EQE plotted against absolute value of current density for two device structures, where the device ① is the same as (a) and the device ② is composed of ITO/ZTO (40 nm)/QD multilayer/ZTO (40 nm)/ITO. Reprinted with permission from reference [115]. © 2009 American Chemical Society.

Download Full Size | PDF

As discussed above, rapid improvement in device structure and efficiency has been achieved after around 2000. The QD-LEDs reported in the present review already satisfy the minimal requirements for full-color displays (i.e., 100 ~500 cd/m² for mobile displays, over 1000 cd/m² for TVs) and color purity (FWHM below 40 nm). Moreover, the recent progress in EQE is breath-taking: the rapid increase in EQE reported by industry (i.e., QD Vision Inc.) and academia, as illustrated in Fig. 23 . Particularly, EQEs of red QD-LEDs are rapidly approaching to the 20%, which is a theoretical limitation with the assumption of 20% of extraction efficiency. Although green and blue QD-LEDs are still far from the commercialization level, rapid increase in EQE represents bright future of QD-LEDs for practical devices. Because of the bright future of QD-LEDs, industries such QD Vision and Samsung Electronics have already paid keen attention to materials, device structures, and processing techniques involved in QD-LEDs to realize QD-based displays. Recently, they demonstrated mono- or full-color active matrix QD displays processed by the contact printing method.

 

Fig. 23 Progress in peak EQEs of QD-LEDs against time. The EQEs are classified into six colors in terms of red (770 nm ~620 nm), orange (620 nm ~580 nm), yellow (580 nm ~570 nm), green (570 nm ~510 nm), cyan (510 nm ~490 nm), and blue (490 nm ~430 nm) to give clear and fair comparison.

Download Full Size | PDF

From previous studies on QD-LEDs mentioned above, it is crucial that novel carrier transport materials (i.e., ZnO or TiO₂ as ETL) and device structure (i.e., introduction of HIL or inverted device structure) lead to considerable improvement in device performance. In the future, investigation on optimum carrier transport materials and degradation mechanism should be performed to improve both device performance and lifetime for practical applications. Moreover, highly cost-effective means for device fabrication and patterning should also be introduced for massive display production. In the following paragraphs, current status on printing technologies for patterned QD-LEDs for full-color displays is reviewed.

4. Printing technologies for QD displays

4.1. Microcontact printing

In previous studies, spin-coating of conducting materials or QD dispersions has extensively been used to fabricate thin films. Although spin-coating is superior in creating uniform and large-area thin films using solution-phase materials, there are some drawbacks for industrial applications. First, material loss during spin-casting is fairly large (around 94 ~97%) [117], increasing the production cost. Second, the solvent orthogonality has to be considered to stack two or more layers in one film. That is, a sublayer should have tolerance against the following solvent. For the fabrication of QD-LEDs, carrier transport materials with low or no solubility to common organic solvents (i.e., chloroform, toluene, hexane, and so forth) dissolving QDs are required and this requirement reduces the chances to find relevant HTL or ETL with suitable bandgap alignment. Third, the spin-coating is not good in realizing well-defined QD patterns. Precise control in thickness, uniformity, and lateral resolution of a patterned QD film is crucial to realize active-matrix QD displays. In order to form pixelated QD layers, several technologies have been suggested so far: microcontact printing, mist coating, and inkjet printing. In terms of uniformity and fidelity of patterns, the microcontact printing has frequently been adopted by QD Vision and Samsung Electronics to fabricate QD displays. However, the placement of QD layers on a stamp causes a large loss of QDs. Therefore, demands on the alternative processes for saving or recycling of QDs as well as improving the quality of patterns recently emerged. Moreover, the surface modification of QDs with functional ligands has also been suggested to improve the patterning process. In the sections below, we attempt to give a summarized overview of recent progress on printing technologies.

Microcontact printing (μCP) is one of soft lithographic techniques employing micro- or nano-sized polymeric stamps to form patterns of ink. Since this easy and straightforward technique was invented by A. Kumar et al. [119], it has frequently been used in various applications demanding micro-sized patterns consisting of self-assembled monolayers, nanoparticles, biomaterials, or polymeric materials in an easy and simple manner [120]. Since V. Santhanam et al. reported transferring the patterned gold nanoparticle arrays from a floating nanoparticle monolyaer on the surface of water using μCP, several approaches based on μCP have been introduced to make patterend QD active layers on various substrates for display applications.

L. Kim et al. simplified the QD transferring process by directly spin-coated a QD solution on a perylene-C coated poly(dimethylsiloxane) (PDMS) stamp and transferred the QD layer to an ITO/TPD substrate, as summarized in Fig. 24(a) [118]. The perylene-C layer on a PDMS stamp was adopted to modify the surface property of patterned PDMS substrate for multiple purpose: 1) to improve the wettability of QDs on patterned PDMS substrate, 2) to enhance the tolerance of PDMS substrate against swelling by the QD dispersion solvent (typically non-polar solvent), and 3) to modulate the surface energy of PDMS substrate for efficient detachment of QDs while transferring process. For the pattern transfer with μCP, the surface energy control of a stamp is crucial to deliver or transfer an ink to a substrate successfully: In order to prepare a homogeneous uniform ink layer on a stamp by the spin-coating of QD solution, the wettability between solvent and the surface of a stamp is critical to realize homogeneous and uniform QD layers, which is directly related with device performances. For the transfer process, interfacial energy between QD layer and a substrate should be lower than that between QD layer and the surface of a stamp. Due to the chemical nature of perylene, compatible with chloroform (typical solvent for QDs) and suboptimal for minimizing the interfacial energy, the preparation of QD monolayer and its transfer to a substrate were successfully demonstrated (Fig. 24(b)). As a result, they were able to integrate QD stripe patterns with 25 μm line features in QD-LEDs (Fig. 24(c)) as well as pixelated red and green QDs by crossing two strip patterns (in Fig. 24(d)). The printed QD-LEDs show high EQEs (i.e., 1.0%, 0.5%, and 0.2% for red, green, and blue QD-LEDs at 100 cd/m², respectively) and further improved [122]. In the meantime, A. Rizzo et al. also reported μCP of QD multilayers and the fabrication of QD-LEDs [123,124] through μCP double-transfer method (i.e., employing a second substrate to form a QD multilayer) or the use of a SU-8 buffer layer with the same purpose as the perylene-C.

 

Fig. 24 (a) A schematic on the four-step contact printing process. (b) AFM images of a QD film deposited on top of TPD layer using (left) a plain PDMS stamp and (right) a parylene-C coated PDMS stamp. A QD film on a plain PDMS exhibit spinoidal decomposition patterns with high surface roughness (RMS roughness = 23.0 nm) while a smooth hexagonally close-packed monolayer is formed on the parylene-C-coated PDMS stamp (RMS roughness = 0.5 nm). The chemical structure of parylene-C is shown in the inset. (c) (upper left) An EL of red and green pixels fabricated on the same substrate. A blue pixel is the result of TPD emission in the area where QDs were not deposited. (upper right) An EL of red QDs patterned with 25 μm wide stamp features. (bottom) Device structure of a QD-LED with an emissive layer consisting of 25 μm wide cross-stripes of green and red QD monolayers. (d) An EL of the device structure shown in (c) at 7 V of applied bias. Blue emission is due to emission from the TPD hole-transporting underlayer. The background TPD emission is not present in the image (c) due to lower applied bias. Reprinted with permission from reference [118]. © 2008 American Chemical Society.

Download Full Size | PDF

The feasibility of μCP for QD multilayers and integrated QD-LEDs with pixelation have been tested mostly in academa. However, from a practical point of view, just a small extent of contamination or misalignment during the μCP process leads to the failure in successful operation of active or passive matrix displays. The variation in QD layer thickness and uniformity is also an important factor for charge balance and corresponding quantum efficiency. In this context, Samsung Electronics recently reported a full-color QD display fabricated by the transfer printing (Fig. 25 ) [121]. In order to ensure the perfect transfer of QD patterns, the donor substrate was modified with octadecyltrichlorosilane (ODTS) which weakens the interaction between the donor substrate and a QD layer by decreasing the surface energy. Moreover, the kinetic control in the QD transfer/pick-up process was also engineered, where the applied pressure and peeling rate are critical to tune the adhesion force between the donor substrate/stamp as well as the stamp/receiving substrate interfaces. Due to this precisely-controlled transfer printing, the QD pick-up yield is almost 100% (Fig. 25(b)) with a micron-sized line-width and the fabrication of red, green, and blue stripes on the same substrate was also achieved with high reproducibility (Fig. 25(a), right bottom). Moreover, closely-packed QD layers by pressure and well-defined interfaces contributed to the enhancement in device performance, which increases the power efficiency by 71% (4.25 lm/W) with decreased leakage current.

 

Fig. 25 (a) A schematic illustration of solvent-free transfer printing. Briefly, (i) a donor substrate was modified with octadecyltriclorosilane (ODTS) to facilitate the delamination of QD multilayer. Then, (ii) red, green, and blue QD multilayers spin-coated were (iii) peeled off from the donor substrate using PDMS stamps and (iv-viii) they were transferred to conducting substrates with precise alignments. (right bottom) A photoluminescence image of the transfer-printed RGB QD stripes on a glass substrate excited at 365 nm UV irradiation. (b) (left) QD pick-up yield as a function of peeling velocity during lift-off of the stamp. The error bars are the standard deviation in the pick-up yields measured at various pressures. (inset) A fluorescence micrograph of QD stripes transferred onto a glass substrate, excited at 365 nm UV light. (right) A SEM image of nanopatterend QD stripes printed on a glass substrate by a microstructured stamp. (c) A 4-inch full-color QD display with a 320 x 240 pixel array. Reprinted with permission from reference [121]. © 2011 Nature Publishing Group.

Download Full Size | PDF

Based on such reliable printing process, they assembled a patterned QD layer with hafnium-indium-zinc oxide TFT arrays and finally realized a 4-inch full-color QD display driven by active matrix (Fig. 25(c)). The integration of finely-controlled QD patterns and novel metal oxide TFTs proves that next generation QD displays are possible based on the optimized μCP process. In addition to Samsung Electronics, QD Vision also demonstrated a 4-inch active matrix QD display [125].

4.2. Inkjet printing

Inkjet printing is a mask-free deposition technique based on liquid-phase materials such as molecules, polymeric materials, or nanoparticles. The core component of printing mechanics is the piezoelectric nozzle, ejecting small ink droplets of several pico-liters by electric signal. Once the voltage change is applied to the piezoelectricity, sudden reduction in the volume of an ink chamber occurs and a tiny liquid droplet is ejected from the nozzle. The liquid drop ejected from the nozzle falls on a substrate and spreads with a certain contact angle determined by surface tension. Solvent evaporation leaves behind an island of materials and the shape of deposited materials is determined by various factors: contact angle, the size and shape of a solute, and viscosity [126].

High accuracy and easy patterning without mask or mold are distinguishable features of the inkjet printing when compared with other patterning technologies. The accuracy of inkjet printing has been improved up to tens of micrometer range and the amount required for the inkjet printing is extremely small compared with the amount used in spin-coating. Moreover, the design of micro-patterns can be easily achieved by controlling the trajectory of inkjet nozzles using a computer. Because of these advantages, inkjet printing has received keen attention as the alternative printing process for organic electronics.

However, the removal of contaminants or aggregates in ink is required to prevent clogging and blocking of inkjet nozzles by large particles. In addition, delicate ink formulation is needed to make a uniform profile of deposited materials. In particular, defect or inhomogeneity in thin film electronics causes the leakage of current or non-uniform voltage profile, deteriorating device performance and lifetime. For instance, the coffee-ring effect [126], the accumulation of a large amount of materials at the evaporation edge resulting in a non-uniform profile, is a debating topic in printing technology. This effect arises from multiple variables in printing process, for instance, vapor pressure, solvent viscosity, transport of solutes via the motion of solvent, surface tension between ink and a substrate, and so on. The selection of a proper solvent with low vapor pressure, increasing contact angle via the chemical modification of a substrate, or the use of two or more solvents are common approaches to reduce the coffee-ring effect [126,127].

In 2009, H. M. Haverinen et al. first demonstrated the patterning of QD-LEDs using the inkjet printing process. Since QD-LEDs shows the best performance with a uniform and homogeneous monolayer of QD active layers, the morphological defects of QD layers, such as QD vacancies or multilayers as a result of coffee ring effect deteriorate device performance (i.e., efficiency and color purity) and reliability. In order to resolve the these issues, they carefully studied various solvents and selected chlorobenzene [128] or toluene/chlorobenzene mixture (1:1 volume ratio) [117] as QD dispersion solution in order to minimize the morphological defects (i.e., cracks or coffee-ring effect). Figure 26(a) shows the AFM height and phase images of inkjet-printed QD films on a polyTPD surface, using chlorobenzene used as a sole solvent. It indicates the presence of a mesa structure with a peak-to-valley value of 5.22 nm and a RMS roughness of 0.58 nm in 500 x 500 nm. Taking into account the size of QDs (~6 nm without a capping layer), this film represents reasonable uniformity. However, it seems that the overall quality of QD film is still inferior to the cases with spin-coated or μCP processed QD films, reporting peak-to-valley distances of 1.8 nm for μCP and 2.3 nm for spin-coating. Their initial QD-LEDs exhibited a wide-band emission centered around 450-750 nm and multiple depositions of QD ink to fill the voids cause the increase in turn-on voltage. A year later, they optimized the formulation of QD ink, a mixed solvent composed of toluene and chlorobenzene and reported full-color RGB devices (ITO/PEDOT:PSS/polyTPD/QD multilayer/TPBi/LiF/Al), shown in Fig. 26(b) [117]. Although the device performance is inferior to the unit device fabricated by the spin-coating process (100 cd/m² with a bias of 9.3 V; EQE of 0.24% for a RGB device), they were able to significantly lower the operation voltage (~5 V) and realize pixelated RGB QD-LEDs with a dimension of hundreds of micrometers (around 160 x 160 μm).

 

Fig. 26 (a) (left) An AFM height image (500 nm) of printed QD layer. The peak-to-valley height of 5.22 nm indicates no significant aggregation. (inset) A white line in the image shows the location of the section scan. (right) An AFM phase image of (a). Reprinted with permission from reference [128]. © 2009 American Institute of Physics. (b) A RGB spectrum measured at 10 V. The overall device area is 0.14 cm² with both ITO and metal as common electrodes to all array elements. (inset) A photograph showing strong emission from red pixels with overshadowed emissions from remaining G and B pixels. Reprinted with permission from reference [117]. © 2010 IEEE.

Download Full Size | PDF

Unfortunately, the printing technologies have not reached the quality of conventional micro-fabrication processes (i.e., photolithography or thermal evaporation with a mask) in terms of spatial resolution and process reliability. However, demands on economic fabrication processes for novel micro-electronics have steadily grown. In particular, many optotelectronic materials are now prepared in solution phase and numbers of such materials are not compatible with conventional processes. QDs are one of the examples of such limitations. Although basic properties of QDs as lumophors are expected to be better than properties with conventional organic or inorganic materials, the lack of relevant printing processes limits the rapid growth of QD-based devices in practical applications. In the future, we expect that the process optimization or development for efficient patterning of QDs would be the key to catch up with conventional display technologies such as liquid crystal-based displays or organic-based displays.

5. Concluding remark

In this review, we tried to provide a broad overview on the preparation and properties of QDs, the structural design for efficient QD-LEDs, and the printing processes for the realization of practical QD displays. Those three parts discussed are summarized as follows:

For realizing efficient and color saturated QD lumophors, the narrow size distribution of QDs and the introduction of uniform stable inorganic shells are critical. At the same time, alloying concepts open up the opportunity for controlling emission wavelength as well as achieving strain-reduced core/shell QDs with chemical composition gradient. Current achievement in the color purity of cadmium-based QDs is enough to render natural colors exceeding previous standards, but development of cadmium-free QDs and chemical processes for mass production remains to be solved. The performance of QD-based LEDs has rapidly been improved in the last decades. This success in device performance is closely connected to the integration of carrier transport layers, which balance the densities of holes and electrons within emissive QD layers. However, the large hole injection barrier between HTL and QDs remains as an obstacle for the direct injection of carriers followed by the radiative recombination in QDs. Therefore, new organic or inorganic hole transport materials are required to improve the device performance. Furthermore, the degradation mechanism of QD-LEDs and counterplans should be investigated in detail to extend the lifetime of the devices. Printing process with high precision and reliability is essential to fabricate practical QD displays in large area. Although current status on printing technologies is insufficient to compete with conventional lithography, a few companies recently demonstrated the possibility that μCP could be effectively utilized to pixelate QD-LEDs in large area. From a practical point of view, novel printing processes with high precision, reliability, and high throughput are thus highly desirable.

Although this review is arranged in the order of materials preparation, device structure, and printing processes, this sequence of the review never implies the order of priority in the development of efficient QD-LEDs. For example, the first QD-LED was reported in the middle of 1990s, at which research on QD synthesis emerged drastically. Moreover, sharp increase in the EQE of QD-LEDs coincides with the development of QDs with advanced core/shell nanostructure (in the middle of 2000s), for instance, the construction of core/multi-shell nanostructure as well as the introduction of alloyed cores or composition gradient shells. This close relationship between materials and devices will be extended to other aspects of QD-LEDs including fabrication processes. The QD/conducting polymer hybrid concept is one of good examples, taking into account the device performance or simplification as well as facilitated fabrication processes for complete QD displays. However, current status of QD-LEDs is still in their infancy compared with organic or inorganic LEDs in terms of efficiency and process developments. In order to compete with conventional LED technologies, further increase in device performance and lifetime as well as the development of economic printing processes for QDs should be addressed. We believe that cross-feedback among materials design, device optimization, and suitable process development will overcome current limitations.