Bandgap control and multishell passivation

The energy bandgap of QDs depends on their size variation in the range below ∼10 nm, due to quantum confinement effects. However, bandgap tuning based on size alone is not desirable because of several limitations. For example, it is difficult to synthesize uniform, small-sized QDs during the initial stage of the nanocrystal growth, and more defects can be generated on the surfaces as the QD size decreases. On the other hand, specific crystalline facets appear as a result of their different surface energies as the nanocrystals grow, resulting in asymmetric shapes. Therefore, it is more beneficial to prepare alloy structures that enable the bandgap to be tuned by changing the composition as well as the size.

Alloy QDs can be designed to have homogeneous or gradient compositions in their structures to cover a wide span of emission ranges, and various ternary and quaternary QDs have been reported, including CdSeTe (550–800 nm),^{Reference Bailey and Nie21} CdSeS (460–600 nm),^{Reference Jang, Jun and Pu22} ZnCdSe (370–470 nm),^{Reference Zheng, Yang and Ying23,Reference Zhong, Zhang, Liu, Han and Knoll24} ZnCdS (390–480 nm),^{Reference Zhong, Feng, Knoll and Han25} and ZnCdSeS (490–590 nm).^{Reference Bae, Char, Hur and Lee26} CdSeS QDs were prepared by performing a single injection of Se and S precursors into the reaction medium containing a Cd precursor.^{Reference Jang, Jun and Pu22} The resulting nanocrystals with a 5 nm diameter exhibit narrow photoluminescence (PL) at wavelengths between the energy levels of CdSe and CdS nanocrystals of the same size, and the PL wavelength can be controlled by changing the Se and S contents. Since the Se precursor reacts with the Cd precursor faster than S, the core of the CdSeS QD is composed of a CdSe-rich phase, and the exterior is composed of a CdS-rich phase. Therefore, the alloy has a gradient Type I structure, in which the shell has a larger bandgap than the core, resulting in a high QE of up to 85%. Annealing core–shell QDs at high temperatures is another way to make alloy QDs. In comparison, shell growth is commonly performed at a low temperature so as not to cause heterogeneous nucleation.^{Reference Dabbousi, Rodriguez-Viejo, Mikulec, Heine, Mattoussi, Ober, Jensen and Bawendi27} Coating of a ZnS shell on a CdSe core above 300°C induces the interfusion (diffusion between interlayers) of CdSe and ZnS to form alloy phases (In CdSe//ZnS, “//” indicates an interfused alloy structure).^{Reference Jun and Jang28} The CdSe//ZnS QDs exhibited considerably blue-shifted emission as the alloy phase expanded (Figure 1a), and the interfusion between the core and shell was promoted by higher reaction temperatures (Figure 1b). Highly luminescent blue QDs that showed peak PL emission at 470 nm with a QE of 98% could be prepared by using CdSe cores with diameters of 4 nm and peak PL emission at 520 nm. The CdSe/ZnS core–shell QDs prepared with the same CdSe core showed peak emission at 520 nm, with an improved QE around 50%. Since the interfused alloy QDs have reduced lattice strain because of the gradient interfaces, they show higher QEs than the core–shell QDs. It is further possible that in QDs with alloyed shells, QD blinking gets supressed, resulting in overall enhancement in QE, as reported by Wang et al.^{Reference Wang, Ren, Kahen, Hahn, Rajeswaran, Maccagnano-Zacher, Silcox, Cragg, Efros and Krauss29}

Figure 1.

(a) Conceptual drawing of core–shell, core, and interfused alloy structures. (b) Photoluminescence (PL) peak shifts of CdSe//ZnS depending on reaction temperature. (Inset: CdSe core and CdSe//ZnS solutions synthesized at 260, 300, and 320°C).^{Reference Jun and Jang28} (c) Conceptual drawing of core, core–shell, and core/multi-shell structures. (d) PL peak position and quantum yields (QYs) of the quantum dots (QDs) obtained by reacting the CdSe core with various amounts of Cd precursor and 0.1 mmol of Zn precursor. (Inset: QD solutions under 365 nm).^{Reference Jun, Jang and Lim38}

The bandgaps of QDs can also be controlled by manipulating the energy-level alignment of the cores and shells. For Type I QDs such as CdSe/ZnS^{Reference Hines and Guyot-Sionnest8} and InP/ZnS,^{Reference Li and Reiss30,Reference Xu, Ziegler and Nann31} the bandgaps are generally maintained after the shell coating. Type II QDs such as CdTe/CdSe (700–1000 nm), CdS/ZnTe,^{Reference Kim, Fisher, Eisler and Bawendi32} and ZnTe/ZnSe (500–600 nm)^{Reference Bang, Park, Lee, Won, Nam, Lim, Chang, Lee, Chon, Shin, Park, Choi, Cho, Park, Joo and Kim33} have staggered energy level alignments and thus show broad emission ranges. Reverse Type I QDs such as CdS/HgS (500–1000 nm),^{Reference Mews, Eychmuller, Giersig, Schooss and Weller34,Reference Braus, Burda and El-Sayed35} CdS/CdSe (500–650 nm),^{Reference Battaglia, Li, Wang and Peng36} and ZnSe/CdSe (400–680 nm),^{Reference Zhong, Xie, Basche, Zhang and Knoll37} in which the shell has a smaller bandgap than the core, exhibit red-shifted emissions. CdSe/CdS QDs are quasi-Type II, since the energy-level alignment changes from Type I to Type II as the CdS shell thickness increases. They showed red-shifted emission and a lower QE compared to CdSe/ZnS QDs.

CdSe/CdS/ZnS multishell passivated QDs with improved QEs could be prepared by performing consecutive injection of the precursors (Figure 1c).^{Reference Jun, Jang and Lim38} Their emission wavelengths were tuned over the entire visible range by controlling the CdS interlayer thickness with the same-sized CdSe core (Figure 1d). The CdS interlayer not only determines the emission wavelength but also reduces the lattice mismatch to improve the QE and stability. The gradient elemental composition of the multishell QDs could be determined for the crystalline structures using transmission electron microscopy (TEM) and local-probe energy-dispersive spectroscopy.

Based on the previously described methods of controlling the bandgaps of QDs, the optimal QD structures in terms of efficiency and stability for display applications could be prepared. To make relatively large QDs for green emission, structures with a blue CdSe//ZnS alloy core and a gradient CdS/ZnS multishell were combined. Figure 2a presents the concepts of nanocrystal growth of the CdSe core, the CdSe//ZnS alloy, and CdSe//ZnS/CdS/ZnS alloy/multishell QDs along with their TEM images and emission spectra. Green CdSe//ZnS/CdS/ZnS QDs with peak emission at 522 nm and a diameter of 7.4 nm were prepared from a blue CdSe//ZnS alloy core with a diameter of 4.5 nm. The growth of the CdS/ZnS multishell was carried out in a single batch by utilizing the difference between the reaction rates of Cd and Zn with S.^{Reference Jun and Jang39} The QE of the green CdSe//ZnS/CdS/ZnS QD was almost unity, and its FWHM was less than 30 nm. For the preparation of red-light-emitting QDs, CdS/ZnS multishells were grown two times on the CdSe core QDs to make CdSe/(CdS/ZnS)₂ QDs. The additional CdS/ZnS coating on the CdSe/CdS/ZnS QDs moved the emission maximum from 599 nm to 630 nm, while maintaining a FWHM of 35 nm; this improved the QE from 75% to 95%.^{Reference Jang, Jun, Jang, Lim, Kim and Kim40}

Figure 2.

(a) Schematic structures, transmission electron microscope images, and emission spectra of green and red quantum dots (QDs).^{Reference Jang, Jun, Jang, Lim, Kim and Kim40} (b) Photos taken under 365 nm UV light of green emitting CdSe//ZnS/CdSZnS (alloy core/multi-shell [AC/MS], “//” indicates interfused alloy structure) QDs, and CdSe/ZnS (core–shell [C/S]) QDs before and after the silicone curing process and UV exposure for 200 h.^{Reference Jun and Jang39}

TEM images clearly show that the particle size increased from d = 6.8 nm to d = 8.6 nm, while uniform size and shape distributions were maintained. The multishell-passivated QDs with relatively large sizes showed better efficiencies and stabilities than simple core–shell structures under intense UV radiation and high-temperature curing processes with silicone resin for further application to LEDs (Figure 2b). CdSe//ZnS/CdS/ZnS and CdSe/ZnS QDs that fluoresce at the same peak wavelength were drop-cast on the quartz substrate and exposed to 365 nm UV radiation at an intensity of 3 mW cm^–2. After 200 h, the multishell QDs retained 95% of their initial luminescence, whereas the core–shell QDs retained only 20%. When the QDs were mixed with silicone polymers and cured at 150°C for 2 h, the PL of the core–shell QDs decreased significantly to 10% of the initial luminescence, while the multishell QDs retained 71% of their initial luminescence.^{Reference Jang, Jun, Jang, Lim, Kim and Kim40}

Highly efficient color-converting QD-LEDs and backlight applications

Further encapsulation of QDs with a polymer or silica composite increases their photo- and thermal stability.^{Reference Lee, Sundar, Heine, Bawendi and Jensen5,Reference Alejandro-Arellano, Ung, Blanco, Mulvaney and Liz-Marzán41,Reference Selvan, Patra, Ang and Ying42} A stable, highly luminescent, homogeneously doped QD silica monolith (QD-SM) composite (comprising of QDs embedded within a silica composite) could be prepared by exchanging the original hydrophobic ligands on a QD surface with hydrophilic 6-mercaptohexanol and performing sol-gel condensation of silica precursors using propylamine as a base catalyst, as shown in Figure 3a.^{Reference Jun and Jang13} The QD-SM composite was stable on a high-power UV-LED operated at 1 W and maintained 90% of its initial intensity without any spectral change for 200 h, whereas the emission of the bare QD dropped to 65% of the initial intensity after 4 h. The external efficiencies of the LEDs with green and red QD-SM composites were 89% and 63%, respectively, while the efficiencies of the LEDs with bare green and red QDs were 70% and 50%, respectively.^{Reference Jun and Jang13} The white LEDs suitable for display backlights that were made with separately prepared green and red QD-SM composites showed the highest efficiency (47 lm/W) ever reported, and a white LED with a mixture of green and red QDs in the same silica composite showed an efficiency of only 33 lm/W because of re-absorption between the densely located QDs (Figure 3b). Since the average distance between the QDs is supposed to be larger than 10 nm, Förster-type energy transfer, in which non-radiative dipole-dipole coupling causes energy transfer between closely located QDs, rarely takes place. To obtain efficient QD-LEDs, it is important to protect the QD surfaces during the fabrication process and to control the QD concentration in limited spaces.

Figure 3.

(a) Quantum dot (QD)-silica composites prepared with red, green, and blue QDs, and the schematic light-emitting diode (LED) structure with QD-silica composites. (b) Spectra of white LEDs made with green and red QD silica composite prepared separately (solid blue) and mixed together (dotted purple). (Inset: photo of operating white QD-SM LEDs.)^{Reference Jun and Jang13}

A white QD-LED was fabricated as a backlight unit (BLU) for a 46” LCD TV panel (Samsung displays) and compared with a TV of the same size with conventional LED backlights (Figure 4a). Although the LED with inorganic phosphors emitted more radiation than the QD-LED over most wavelengths, the luminous efficacies of both LEDs were very similar. The reason for the disparity between the light intensity and the efficiency is based on the photopic sensitivity, which is relative sensitivity of the human eye to lights with different wavelengths: A considerable portion of the emission from the blue LED and the red phosphor does not contribute to the luminous efficiency (Figure 4b).^{Reference Jang, Jun, Jang, Lim, Kim and Kim40} The peak of the luminous flux from phosphors in the red region appeared at 594 nm because of the overlap of the broad emissions from green and red phosphors. However, the peak of the QD-LED’s luminous flux appears at its original peak wavelength of 625 nm. Therefore, the QD-LEDs produced more vivid color, and the area of their color triangle, which depicts color relationships, covered all of the NTSC color space (104.3%), while the conventional LEDs covered less than 80%.

Figure 4.

(a) 46-inch diagonal LCD TV panel (Samsung displays) adapting (quantum dot light-emitting diode backlight unit) QD-LED BLU (left) and phosphor LED BLU (right). (b) Light intensity spectra (solid line) and brightness (hatched area) of the QD-LED (blue) and the phosphor-LED (gray). Inset: Color triangles of the QD-LED (white) and the phosphor-LED (yellow) compared to the NTSC (National Television System Committee) standard (black line).^{Reference Jang, Jun, Jang, Lim, Kim and Kim40}

High-performance current-driven QD-LEDs

Future displays might need more efficient electrically operated light sources, and current-driven QD-LEDs show more potential than organic light-emitting diodes (OLEDs) because of their smaller Stokes shifts (difference between positions of band maxima of absorption and emission spectra), high stability against oxygen and moisture, and mechanical robustness. Colvin et al. first proposed the use of QDs as light-emitting layers in an electroluminescent (EL) device,^{Reference Colvin, Schlamp and Alivisatos43} and since then, various approaches to enhancing the performance of QD-LEDs have been investigated, including the design of QD structures tailored for EL,^{Reference Coe, Woo, Bawendi and Bulović6,Reference Schlamp, Peng and Alivisatos11,Reference Sun, Wang, Li, Wang, Zhu, Xu, Yang and Li12} the development of LED structures with appropriate charge transport layers and electrodes,^{Reference Cho, Lee, Joo, Jang, Kim, Lee, Kwon, Han, Kim, Choi and Kim14,Reference Colvin, Schlamp and Alivisatos43,Reference Qian, Zheng, Xue and Holloway44} and control over the composition and thickness of the QD layers.^{Reference Coe, Woo, Bawendi and Bulović6,Reference Sun, Wang, Li, Wang, Zhu, Xu, Yang and Li12,Reference Coe-Sullivan, Steckel, Woo, Bawendi and Bulović45} The performance of QD-LEDs was improved by balanced charge injection, which could be controlled by changing the QD layer’s thickness, energy level alignment, and nature of the interfaces. Compared to LEDs with thick QD-doped polymer active layers, the LEDs with ordered monolayer QD arrays had efficiencies an order of magnitude higher because the electrical resistance was minimized.^{Reference Coe, Woo, Bawendi and Bulović6} An inorganic sol-gel ETL (TiO₂ or ZnO) could induce efficient electron injection from the Al electrode to the QDs due to the small energy barrier between the layers. Inverted-type QD-LEDs with well-aligned energy levels showed a record efficiency due to their efficient charge injection and control of the recombination zone.^{Reference Anikeeva, Madigan, Coe-Sullivan, Steckel, Bawendi and Bulović46}

A monochromatic QD display has been produced by reducing the charge injection barrier using a cross-linking process.^{Reference Cho, Lee, Joo, Jang, Kim, Lee, Kwon, Han, Kim, Choi and Kim14} A schematic energy diagram of the QD-LED is shown in Figure 5a. The device consists of an ITO anode, poly(ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) and poly[(9,9-dioctylfluorenyl-2,7-diyl)-co-(4,40-(N-(4-sec-butylphenyl)) diphenylamine)] (TFB) as the hole injection and hole transport layer (HIL/HTL), QDs of the core/shell/shell (CdSe/CdS/ZnS) structure as the emissive layer, amorphous TiO₂ as an ETL, and an Al cathode. Electrons are easily transported to the QD layer because of the small energy barrier, but holes are mainly confined between the HTL and the QD layer because of the large band offset between the HOMO levels of the HTL and the QD layer (∼1.5 eV). Therefore, reducing this band offset is a key requirement for achieving efficiency of the QD-LEDs. Cross-linking was introduced not only to make QD films insoluble during the spin-coating of the sol-gel TiO₂ but also to modulate the energy band level of the QD layers. The process is carried out by dipping QD films into a 1,7-diaminoheptane/methanol solution, where the linker molecules are attached to the QD surfaces through exchange with pre-existing surfactants or by filling empty sites on the QD surface (Figure 5b).^{Reference Cho, Lee, Joo, Jang, Kim, Lee, Kwon, Han, Kim, Choi and Kim14,Reference Guyott-Sionnest and Wang47,Reference Yu, Wang and Guyott-Sionnest48} Figure 5c shows the EL characteristics of QD-LEDs fabricated under different cross-linking and thermal annealing conditions. Sample A was prepared by cross-linking the QD layer after spin-coating. For samples B and C, QD films were cross-linked and then annealed at 80°C and 180°C, respectively, for 30 min under nitrogen. In sample D, the QD layer was annealed at 180°C for 30 min without cross-linking. The cross-linking and/or subsequent thermal annealing remarkably improved the luminance and efficiency of the QD-LEDs.

Figure 5.

(a) Energy band diagram in quantum dot light-emitting diode (QD-LED). The device consists of a patterned ITO anode, PEDOT:PSS HIL (50 nm), TFB HTL (20 nm), QD layer as an emissive layer (30 nm), TiO₂ ETL (40 nm), and aluminum cathode. (b) Schematic illustration of the cross-linking process. As-coated films are dipped into the cross-linker (1,7- diaminoheptane) solution for 15 min. (c) Luminance–voltage (i) and luminous efficiency–voltage (ii) characteristics for the QD-LEDs with different QD manipulations. The QD layer cross-linked after spin-coating (sample A), cross-linked and annealed at 80°C and 180°C, respectively, for 30 min under nitrogen (samples B and C), and annealed at 180°C for 30 min without cross-linking (sample D). (d) He II UPS spectra of the valence band (right) and secondary cutoff region (left) for various QD films. Inset shows the QD band level shift (blue to red) after cross-link.^{Reference Cho, Lee, Joo, Jang, Kim, Lee, Kwon, Han, Kim, Choi and Kim14}

The effects of the cross-linker on the QD valence band were also studied using ultraviolet photoelectron spectroscopy (UPS) analysis.^{Reference Cho, Lee, Joo, Jang, Kim, Lee, Kwon, Han, Kim, Choi and Kim14} QD films were prepared on silicon substrates with different treatments. In particular, the films were as-coated without any post-treatment (A), cross-linked (B), cross-linked and then annealed at 80°C or 180°C for 30 min (C and D, respectively), or annealed at 180°C for 30 min without cross-linking (E). Figure 5d shows the UPS spectra of the valence and secondary cutoff regions of QD films. A remarkable shift of 0.6 eV in the QD valence band toward the vacuum level was observed for cross-linked films (red square). Thus, the energy band offset between the QDs and the HTL was reduced from 1.5 eV to 0.9 eV, which improved the efficiency of the QD-LEDs (C and D) through more efficient hole injection. Although the thermal treatment did not change the QD energy levels, it played an important role in increasing the brightness and efficiency of the devices (A–C), because it led to the removal of excess organics in the QD layer and improved the interfacial contact. Thermo gravimetric analysis indicated that 20 wt% of the QD layer was removed during thermal annealing at 180°C for 30 min. This removal of some portion of the electrically insulating organic residues surrounding the QDs decreases the resistance at the QD/HTL and QD/ETL interfaces, resulting in improved hole injection and charge balance.

Full-color active-matrix QD displays realized by dry pick-and-place QD transfer

The integration of QDs into full-color LEDs is still restricted because conventional wet-deposition methods do not provide a proper process for color-selective patterning of QDs into pixilated panels. Therefore, a new patterning method optimized for QD films is required for the realization of a full-color QD display. Mist deposition, inkjet printing, contact printing, and double transfer processes have been proposed as candidate QD patterning methods.^{Reference Zhu, Shanmugasundaram, Price, Ruzyllo, Zhang, Xu, Mohney, Zhang and Wang49–Reference Rizzo, Mazzeo, Palumbo, Lerario, D’Amone, Cingolani and Gigli53} However, these methods have problems with uniformity and controllability for large-scale applications. In order to suppress the swelling of the elastomer stamp and prevent contamination on the device stack, it is necessary to transfer fully dried QD films from one substrate to another instead of dropping QD solutions directly on the elastomeric stamp. Such highly robust solvent-free film transfer enables large-scale patterning for reliable device performance. Colloidal QDs can be integrated into various devices using transfer printing, which offers a convenient mechanism for lateral patterning of QDs in designed layouts.

Recently, transfer printing based on “pick-and-place” technology reported by Rogers and co-workers has become an essential, sophisticated approach for material assembly and device fabrication, and the variety of nanomaterials usable as ink and their applications have increased dramatically.^{Reference Meitl, Zhu, Kumar, Lee, Feng, Huang, Adesima, Nuzzo and Rogers16–Reference Kim, Wu, Carlson, Jin, Kovalsky, Glass, Liu, Ahmed, Elgan, Chen, Ferreira, Sitti, Huang and Rogers20} This technique involves selective subtracting of the necessary portion of the nanomaterials according to the stamp patterns, the so-called “inking,” and printing them at the required specific position through kinetic control without affecting the inherent properties of the nanomaterials. Therefore, this technique has advantages for large-area assembly of nanomaterials and will have applications in flexible electronics.

The first active-matrix full-color QD display fabricated by transfer printing was demonstrated in 2011.^{Reference Kim, Cho, Lee, Lee, Chae, Kim, Kim, Kwon, Amaratunga, Lee, Choi, Kuk, Kim and Kim15} It was achieved through the solvent-free transfer printing of QD films, in which the surface energy of the donor substrate and the speed of retrieving an elastomeric stamp were controlled to produce more compact QD networks. Unlike the additive transfer method, the subtractive transfer method needs perfect release at the interface between the QDs and the donor substrate because fully dried QD films on the donor substrates must be delaminated as desired. In order to achieve this, a self-assembled monolayer that reduces the surface energy of the donor dramatically while maintaining covalent bonding with the donor substrate was introduced.

Figure 6a presents a schematic illustration of the transfer printing process.^{Reference Kim, Cho, Lee, Lee, Chae, Kim, Kim, Kwon, Amaratunga, Lee, Choi, Kuk, Kim and Kim15} First, the surface of the donor substrate was modified with dense, hydrophobic alkyl self-assembled monolayers (SAMs). An octadecyltrichlorosilane (ODTS) monolayer was formed by dipping the Si substrates into the monomer solution and then heating them to form covalent bonds with the native oxide on Si substrates. These ODTS alkyl chains (C18) were physically ordered with a high packing density and tended to orient themselves perpendicular to the surface.^{Reference Kim, Lee, Yang, Yang and Cho54} The covalently bonded ODTS-SAM reduces the surface energy from 1140 mJ/m^{Reference Bruchez, Moronne, Gin, Weiss and Alivisatos2} to 21 mJ/m^{Reference Bruchez, Moronne, Gin, Weiss and Alivisatos2}.^{Reference Kim, Choi, Kim, Meitl, Menard, Jiang, Carlisle and Rogers17,Reference Kim, Lee, Yang, Yang and Cho54} The ODTS monolayer is molecularly flat over a large area and promotes spreading of a solution of aliphatic capped QDs, resulting in the production of uniform and smooth QD films over a wide area through spin-coating.^{Reference Meredig, Salleo and Gee55,Reference Lee, Yoon, Choi and Kim56} The QDs were placed over the ordered alkyl chains of the ODTS monolayer, and very weak interactions formed between the alkane group bonded on the donor substrate and the organic ligands that cap the QD surfaces, which permits the facile retrieval of QDs from the donor substrates.

Figure 6.

(a) Schematic illustration of the transfer printing process for the patterning of quantum dots (QDs). (b) QD pick-up yield as a function of the peeling velocity during lift-off of the stamp and optical image of the structured stamp with protruding stripes onto which QDs are transferred (right inset). (c) Scanning electron microscopy (SEM) image of nanopatterned QD stripes by the nanostructured stamp. (d) Fluorescence micrograph of the transfer printed RGB QD stripes excited by 365 nm UV irradiation. SEM images of (e) a spin-coated QD film after cross-linking and (f) a transfer printed QD film transferred using an applied pressure of 196 kPa, followed by cross-linking.^{Reference Kim, Cho, Lee, Lee, Chae, Kim, Kim, Kwon, Amaratunga, Lee, Choi, Kuk, Kim and Kim15} SAM, self-assembled monolayer.

QDs are retrieved from the donor substrate through the action of van der Waals forces^{Reference Hsia, Huang, Menard, Park, Zhou, Rogers and Fulton57} between the QDs and a structured PDMS stamp that featured a regular protruding pattern. However, in this case, the removal speed of the stamp has to be optimized to achieve a perfect peeling rate. The adhesion between the QD layer and the stamp is sensitive to the peeling rate because of the viscoelastic behavior of the elastomeric stamp.^{Reference Meitl, Zhu, Kumar, Lee, Feng, Huang, Adesima, Nuzzo and Rogers16,Reference Feng, Meitl, Bowen, Huang, Nuzzo and Rogers58} Pulling the stamp from the donor substrate with a sufficiently high peeling velocity leads to strong adhesion of the QD layer to the surface of the stamp. The pick-up and printing processes for the patterned QD films can be accomplished successfully through kinetic control, as reported by Meitl et al.,^{Reference Meitl, Zhu, Kumar, Lee, Feng, Huang, Adesima, Nuzzo and Rogers16} with assistance of the donor surface treatment. The interfacial contact area between the QD film and the stamp also influences the efficiency of the pick-up process.^{Reference Kim, Carlson, Ahn, Won, Wang, Huang and Rogers19} As the contact area between the QDs and the elastomeric stamp increases under the applied pressure, the QDs can be more easily retrieved from the donor substrate. Figure 6b shows the measured dependence of the pick-up yield on the peeling velocity. Perfect retrieval of QDs occurs at a high peeling velocity with a proper applied pressure. Below the threshold velocity, the efficiency drops substantially. If the pressure applied on the stamp is insufficient, perfect retrieval cannot be achieved at any peeling velocity because of the poor contact between the stamp and QD films.

The fully inked stamp with QDs only on the surfaces of the protruding stripes was used to print the QD pattern onto a receiving substrate with a prefabricated TFT array coated with an organic HTL and PEDOT:PSS on the ITO. Patterned QD films are printed on a device stack at a sufficiently low peeling rate of the stamp because of the low surface energy of the PDMS. Transfer of the QDs from the stamp to the polymer HTL is accomplished more easily than the picking-up process because the polymer film provides strong adhesion between QDs and the substrate.^{Reference Yan, Yao, Lu, Chen, Zhang and Yang59} QD stripes can be patterned and printed, even with a nanoscale line width, using the kinetically controlled process if the PDMS stamp has an appropriate nanostructured geometry (Figure 6c). The stamp picks up the QDs only from the Si substrate covalently bonded with the SAM, and no surface treatments are performed on the PDMS, so there is no transfer of contaminants that can act as luminescence quenching sites or insulators.^{Reference Yim, Zheng, Liang, Friend, Huck and Kim60} Green- and blue-emissive QDs can be repeatedly transferred by sequential printing.

Figure 6d shows a fluorescence microscope image of printed QD stripes with three individual colors under 365 nm UV excitation that were patterned using a sequential transfer-printing process. The pressure applied during the printing process also affects the morphology and interfacial contact of the QD film. The PDMS stamp distributed the applied pressure evenly onto the QD film, and the QDs were rearranged to form a densely packed QD layer. As the alkyl chains on QD surfaces are bendable and mobile, this in turn reduces inter-QD distances due to the pressure applied on the elastomer stamp. Scanning electron microscopy (SEM) images of a spin-coated QD film and a printed QD film with thicknesses of 30 nm after cross-linking on TFB-coated Si substrates are shown in Figure 6e–f, respectively. In a previous study, cross-linking of the QD film reduced the spacing between the QDs and introduced stress that caused fine cracks to form in the QD film, as shown in Figure 6e.^{Reference Cho, Lee, Joo, Jang, Kim, Lee, Kwon, Han, Kim, Choi and Kim14} Such cracks may cause current leakage in the device. In contrast, the printed QD films have fewer voids and cracks after cross-linking than the spin-coated films (Figure 6f) because the QDs are vertically close-packed because of the pressure applied to the stamp, and reordering proceeds through the filling of vacancies, which offers easy electron hopping in the QD layer and good mechanical properties.^{Reference Lee, Yoon, Choi and Kim61}

Printed RGB QD films were integrated into an LED structure with TFB and TiO₂ layers as the HTL and ETL, respectively. The QDs were cross-linked and thermally annealed after printing.^{Reference Cho, Lee, Joo, Jang, Kim, Lee, Kwon, Han, Kim, Choi and Kim14} The thermal annealing improved the adhesion between the organic layer (TFB) and the printed QDs, hence reducing the interfacial electrical resistivity.^{Reference Niu, Munro, Cheng, Tian, Liu, Zhao, Bardecker, Plante, Ginger and Jen62} The device characteristics of the spin-coated and printed QD devices were investigated and are compared in Figure 7a.^{Reference Kim, Cho, Lee, Lee, Chae, Kim, Kim, Kwon, Amaratunga, Lee, Choi, Kuk, Kim and Kim15} The wavelengths at the maximum EL peak for the printed red, green, and blue QD-LED are 615 nm, 530 nm, and 480 nm, respectively. The maximum power efficiency of the printed red QD-LED was improved by 71% up to 4.25 lm/W. The maximum brightness and the luminous efficiency of the printed QD-LED were improved by 25%–52% in all cases, owing to the morphological improvements of the printed film.

Figure 7.

(a) Luminous efficiency and brightness versus voltage characteristics of quantum dot light-emitting diodes (QD-LEDs) fabricated by transfer printing (-●-) and spin-coating (-■-) of red-emissive QDs. (b) Current density versus voltage characteristic of the QD-LEDs fabricated by transfer printing (blue solid line) and spin-coating (black dashed line). (c) A 4-inch full-color QD display image using an AM TFT backplane with a 320 × 240 pixel array. (d) Optical image of simultaneous RGB electroluminescent (EL) emission from all pixels under operation. (e) Flexible LED with RGB QDs patterned by transfer-printing. Inset, optical microscope image of EL emission of RGB patterned QDs from the flexible device. (f) Electrical and luminous properties for different bending radii.^{Reference Kim, Cho, Lee, Lee, Chae, Kim, Kim, Kwon, Amaratunga, Lee, Choi, Kuk, Kim and Kim15}

Figure 7b compares the J-V characteristics of the printed and spin-coated QD-LEDs.^{Reference Kim, Cho, Lee, Lee, Chae, Kim, Kim, Kwon, Amaratunga, Lee, Choi, Kuk, Kim and Kim15} The printed QD-LED exhibits Ohmic conduction up to 2.5 V, trap-limited conduction (J ∝ Vⁿ, n>2) up to 5 V, and then pseudo–space-charge-limited conduction (J ∝ V^{Reference Huynh, Dittmer and Alivisatos3}) at higher voltages. The current density of the printed QD-LED in the Ohmic conduction region is two orders of magnitude lower than that of the spin-coated QD-LED, implying that the leakage current of the printed QD-LED is considerably lower. In addition, the slope of the J-V curve for the printed QD-LED is steeper than that of the spin-coated QD-LED in the trap-limited conduction region, which means electron hopping occurs more easily in the printed QD-LED because of the smaller gap between QDs. Figure 7c shows a photograph of a 4” full-color active-matrix QD display with a resolution of 320 × 240 pixels formed by the sequential transfer printing of RGB QDs. The displayed image is nearly perfect, and white balance is achieved by balancing the current in the RGB pixels. The QD layer was transfer-printed onto a TFT backplane pixel array composed of amorphous Si or hafnium-indium-zinc oxide with charge transport layers. The emissive area of each pixel is 46 μm × 96 μm. Figure 7d shows an optical microscope image of the simultaneous RGB EL emission from the pixelated QD area during operation. RGB-patterned flexible QD devices were also demonstrated using the transfer printing procedure (Figure 7e). These flexible and colorful QD-LED devices were fabricated on flexible ITO/polyethylene naphthalate substrates using the RGB-patterned QDs by transfer-printing, and the LED structure was consistent with full-color QD displays. Uniform QD films can be printed onto curved or bumpy surfaces by the elastomeric stamp. The flexible RGB QD-LED fabricated by transfer printing had good electrical stability and mechanical properties, and the device experienced no performance degradation in a bending test, even after 50 bending cycles. The current density and luminous efficiency show little variation as a function of applied bias up to a bending radius of 3 cm (Figure 7f), which indicates that the printed QD films are also suitable for flexible displays and lighting.