Graphene quantum dot (GQD) light-emitting diodes (GQD-LEDs) are shown to have an electroluminescence exceeding 1000 cd m-2. These devices are possible due to a novel synthesis method to create GQDs with minimal oxidation, guaranteeing high quantum yields via the solvothermal formation of graphite intercalation compounds between graphite powder and sodium potassium tartrate. The GQDs are incorporated into polymeric host layers in a multilayer device and irradiate blue (~400 nm) emission. With strong optical absorptivity and a widely tunable bandgap, graphene is an attractive material for optical and optoelectronic devices.1, 2 Recently, emerging graphene quantum dots (GQDs), single- or few-layer graphene only several nanometers in size, have been implemented in bandgap engineering due to quantum confinement3, 4 and edge effects,5 which potentially can be exploited in new potential applications of graphene. Specifically, fluorescent GQDs with tunable emission are considered to be next-generation nanomaterials as a potentially inexpensive and safe alternative to semiconductor quantum dots (QDs).6, 7 Moreover, the chemical inertness and excellent biocompatibility of GQDs make them attractive for use in bio-sensor, bio-imaging, and photovoltaic applications.8-13 However, currently available GQDs have several drawbacks, including i) limited synthesis methods that provide desired optical properties (i.e., quantum yields); ii) the lack of a mass production method to create GQDs with uniform sizes and shapes without significant loss of their optical properties; iii) difficulty controlling the emission wavelength of the photoluminescence (PL); and iv) debate over the PL mechanism and contraints on the diversity of applications. Various methods and techniques to prepare GQDs, including the transformation of C60 molecules,14 electron beam lithography,15 organic synthesis,16 hydrothermal cutting,9, 17 and electrochemical processing,9 have been suggested to date. Despite some success in the synthesis of GQDs, these methods have limitations as well, such as the requirement of special equipment and extremely expensive raw materials, long fabrication times, low production yields, and the use of harmful chemical reagents. In this contxt, the method of oxidation and a further cutting reaction of graphite is appealing for the mass production of GQDs.9, 11-13, 17-20 However, numerous oxygenous functional groups and defects are inevitably generated both on the basal plain and at the edges of GQDs. This significantly influences the dispersion properties of GQDs and makes it difficult to control the emission wavelength of GQDs.21-24 This oxidation of GQDs thwarts both fundamental understanding of the origin of luminescence and attempts to realize practical applications. In particular, the use of GQDs as an alternative to organic active materials for light-emitting diodes (LEDs) remains challenging due to low production yields and low quantum yields. The development of novel strategies to synthesize high-quality GQDs with controlled oxidation on a large scale is thus crucial. Herein, we suggest an easy and mass-producible route to obtain high-quality GQDs with controlled oxidation by graphite intercalation compounds (GICs). The proposed method is cost-effective and eco-friendly, as potassium-sodium tartrate (KNaC4H4O6·4H2O) serves as both an intercalant into graphite and a solvent in the solvothermal reaction. After the formation of GICs, a dissolution process in water generates a large amount of GQDs (>60% yield) without requiring any surfactants or hazardous chemical solvents. More interestingly, in contrast to the visible emission of most GQDs synthesized from an oxidation method, our GQDs emit near the UV range (∼400 nm), which has been reported as the intrinsic emission wavelength of GQDs with low oxidation.21, 23, 25 This result extends and confirms our current understanding of the fluorescence mechanisms of GQDs with respect to quantum confinement effects,26-28 recombination of localized electron-hole pairs,8, 18 triplet states from zigzag sites,17 oxygenous emissive traps,11 and intrinsic/extrinsic states.25 We also found that the fluorescence of GQDs originated from two overlapped spectral bands with intrinsic and extrinsic states and first discovered the carrier transfer process from intrinsic to extrinsic states. Furthermore, the degree of oxygen content, which controls the fluorescence color, was controlled simply by changing the sizes of the GQDs. These results were realized by preparing high-quality GQDs with a narrowly distributed size via GICs. To confirm the superior performance of these GQDs and suggest the possibility of optoelectronic applications, we directly employed them as emitters in LEDs. In recent years, although GQDs have been made competitive in light-emitting performance, their full potential has not been demonstrated yet. From these LEDs fabricated with the GQDs, maximum luminance exceeding 1,000 cd m−2 was achieved. The performance levels of GQD-LEDs can be enhanced even further by using other polymers, improving the QYs of the GQDs by surface passivation, and optimizing the structural designs of the devices in which they are used. The GQDs fabricated by GICs will allow control of the fundamental properties of given materials through intrinsic/extrinsic effects. This will further allow new devices to be developed with extraordinary properties and functions for numerous applications ranging from bioimaging to optoelectronic devices. The overall process of the proposed synthetic route for GQDs is illustrated in Figure 1a. The initial step is the mixing and grinding of potassium sodium tartrate (200 mg) with graphite (20 mg, SP-1, Bay Carbon Inc.) at a mass fraction ratio of 10:1. Potassium sodium tartrate enables the synthesis of GICs at low temperatures with minimal damage or oxidation in the basal plane of GQDs. In the second step, the ground homogeneous mixtue is reacted in an autoclave vessel at 250 °C for 24 h. The GICs prepared in this way are stable in air, in contrast to conventional KC8 GICs.29-31 The abrupt reduction between the GICs and water not only exfoliates the graphene layers but also breaks the graphene flakes into GQDs. This aqueous solution of GQDs is then centrifuged in filters (8000 and 10 000 NMWL, Amicon Ultra-15) to separate the GQDs by size (below 5 nm, from 5 nm to 10 nm and over 10 nm) at 10 000 rpm for 30 min. The proposed method simply allows scalable synthesis of high quality GQDs by increasing the amount of reactants. Further details of the experimental method are described in the Supporting Information. The phase transition by intercalation is proven by X-ray diffraction (XRD), as shown in Figure 1b. The main peaks related to the interlayer spacing of the GICs (∼6.27 Å) and graphite (∼3.37 Å) are 14.1°, and 26.4o, respectively. The slightly larger interlayer spacing of the GICs in this experiment compared to KC8 GICs suggests that highly ordered GICs can be fabricated by inserting not only potassium metal but also a tartrate chain. This is also supported by a comparison of mass losses from the TGA data (Figure S2.1.1). The thermogravimetric analysis (TGA) result shows that the weight of the potassium-sodium tartrate steeply decreases in the range of ∼250 °C, which implies that the potassium sodium tartrate begins to dissociate into two metal ions (Na+ and K+) and tartrate linker. The graphite is expanded by reacting with metal ions salt in hydrothermal condition, and the expanded graphite is co-intercalated by tartrate. The co-intercalation of tartrate chains with alkali metal may be associated with both the unusual stability of the GICs in air and minimal oxidation of the GQDs, even after the GICs undergo a reduction process in water. High-resolution transmission electron microscope (HRTEM) images and the corresponding Fourier transform (FFT) pattern of the GQDs presented in Figure 1c and Figure S2.1.2, respectively, reveal a highly crystalline structure with a lattice spacing of 2.11 Å, consistent with previous reports.21, 22 The atomic force microscopy (AFM) images shown in Figure 1d and Figure S2.1.3 present the topologic morphology of the GQDs. Their diameters are mainly distributed in a range of 1∼5 nm, and their topological heights fall in a range of 0.5∼1.5 nm (Figure 1e). To verify the lower oxidation and defect formation of GQDs fabricated through GICs, we fabricated GOQDs using the modified Hummers method.22 The GQDs and the GOQDs are similar in terms of their size (below 5 nm) and thickness (below 2 nm) (Figure S.2.2.1). The X-ray photoelectron spectroscopy (XPS) results, as shown in Figure 2a and Figure S2.2.2, indicate that the GQDs have a dominant sp2 carbon peak at 284.5 eV with negligible oxygenous peaks caused by their tartrate functional groups. In contrast to the GQDs, the XPS spectrum of the GOQDs indicates the presence of numerous oxygen functional groups of carbon atoms for the C-C (284.5 eV), C-O (286.6 eV), C=O (288.2 eV) and OH-C=O (289.1 eV) bonds. From fourier transform infrared spectroscopy (FT-IR) analysis of the GQDs and the GOQDs (Figure 2b), characteristic peaks related to the C-H band (2923 cm−1), C=O/COOH band (1720 cm−1), and the -OH band (1379 cm−1) are observed in both samples, while the epoxide band at 1072 cm−1 is completely absent in the GQDs. Note that epoxy groups can serve as chemically reactive sites for the rupture of the underlying C-C bonds and these groups usually induce non-radiative recombinations of localized e-h pairs.11, 20 The Raman spectrum shown in Figure 2c also provides convincing evidence of the formation of high-quality GQDs. The GQDs show a disorder (D) band at 1356 cm−1 and a sp2-bond C-atom (G) band at 1594 cm−1, as well as a D-to-G ratio (ID/IG) of 0.873, which is smaller than that of the GOQDs (1.105). These results indicate that the synthesized GQDs have a pure sp2 carbon crystalline structure with fewer oxygen defects. These are extremely important characteristics when attempting to understand the optical PL origin of GQDs and GOQDs. In Figure 2d, which presents the UV/Vis spectra of the GQDs and GOQDs, a typical absorption peak at 260 nm is observed for the GQDs. This peak is assigned to the π–π* transition of the aromatic sp2 domains. For the GOQDs, however, a new absorption band caused by the n-π* transition of C=O at 320 nm is also observed.10, 20 In the PL under excitation at 310 nm, as shown in Figure 2e and Figure S2.2.3, the GQDs show a strong peak at ∼400 nm (blue emission), whereas the GOQDs have a maximum peak at 520 nm (green emission). This result strongly suggests that the GQDs fabricated by GICs have a high crystal quality. It also indicates their luminescence mainly originated from the transition of aromatic sp2 nano-domains, as the luminescence derived from the oxygen functional groups with regard to GOQDs is known to have a green spectral output, whereas that derived from aromatic sp2 domains in high-quality GQDs generally shows a blue spectral output (Supporting Video 1).25 In order to clarify the carrier dynamics in the GQDs, we carried out time-resolved PL (TRPL) using a Ti:Sapphire laser and a streak camera detector at room temperature. By changing the excitation wavelengths (λex) in the GQDs, the optical transient states for the (intrinsic) π–π* transition of the aromatic sp2 nano-domains and the (extrinsic) n-π* transition of the oxygen functional group of the basal plane and/or the edge are selectively excited. Figures 3a and c present streak images of the PL from the GQDs with different values of λex ranging from 266 to 450 nm. For a λex value of 266 nm, both the intrinsic and extrinsic states can become excited, whereas when the λex value is 450 nm, only the extrinsic states associated with the oxygen functional groups can become excited. The insets show the temporal profile of the time-integral PL (TIPL) spectra from the streak images. As time passes, the temporal PL spectra at a λex value of 266 nm reveal a red shift of the PL peak position from 422 nm to 448 nm (inset of Figure 3a), whereas the spectra at a λex value of 450 nm did not change significantly (inset Figure 3c). This indicates that the optically excited carriers in GQDs with a λex value of 266 nm transfer from intrinsic to extrinsic states, as depicted in Figure 3b. On the other hand, this type of carrier transfer does not occur at a λex value of 450 nm, as only the extrinsic states are excited, as illustrated in Figure 3d. The PL decay time at short excitation wavelengths (i.e., 266 nm and 300 nm) shows much slower recombination times than that at long excitation wavelengths (i.e., 350 nm, 400 nm, and 450 nm) due to the effect of the carrier transfer (Figure S2.3.1). Because the carrier transfer increases the relaxation time,32 we analyzed the variation of the rise time from TRPL spectra at different λex values. The rise times become faster with an increase in λex from 266 nm to 450 nm due to the disappearance of the carrier transfer process (Figure S2.3.2). We also prepared four types of GQDs with different average diameters (d): GQDs-A (d < 5 nm), GQDs-B (5 < d < 10 nm), GQDs-C (10 < d < 20 nm), and GQDs-D (d > 20 nm) through filtration, as described in Figure S2.4.1. When d increases, the emission peaks are red-shifted from 436 nm to 487 nm under excitation by a He-Cd CW laser (325 nm). Several recent studies have reported that the size-dependent PL behavior originates from the quantum confinement effect10 or from shapes33 created by zigzag and armchair edges. In contrast, in other studies, size-dependent PL behavior was not observed due to the self-passivated layer of the GQDs.34 Further study is therefore required to solve the controversial mechanisms of this type of size-dependent PL behavior. In addition, the intensity of the UV/Vis absorbance (Figure S2.4.2) increases from the GQDs-A to the GQDs-C samples, whereas that of the PL spectra show the opposite tendency (Figure 4a). Hence, GQDs with a small diameter (GQDs-A) show the strongest optical efficiency among the GQDs with various diameters. To determine the cause of the strong luminescence from the GQDs with a small diameter (GQDs-A) as discussed above, the characteristics of the PL mechanisms of diameter-dependent GQDs were investigated through a comparison of the GQDs-A samples and the GQDs-C samples. From the XPS spectra, we found that the GQDs-C samples have more oxygen functional groups than the GQDs-A samples (Figure S2.4.3). In addition, the PLE spectra were measured at the maximum PL peak position of both GQDs-A and GQDs-C samples (the blue spectra in Figures 4b and c). For the GQDs-A samples, the PLE spectra show a sharp peak at ∼250 nm which originates from the π–π* transition in GQDs' sp2 nano-domains as well as a broad shoulder near 300 nm related to the n-π* transition for oxygen defects.10, 13, 33 For the GQDs-C samples, however, a sharp PLE peak originating from the π–π* transition was not observed, although a PLE shoulder peak near 350 nm was noted, indicating that the origin of PL from the GQDs-C samples is dominated by the extrinsic states. We propose that the diameter-dependent PL in the GQDs can be attributed to the different numbers of oxygen functional groups. To confirm the proposed PL mechanism, we performed PL and PLE by changing the pH of the solvent. The oxygen functional groups should be very sensitive to the pH, while the pure sp2-bonding structure should not respond.35 At a pH of 2 (red diagrams in Figures 4b and c), the PL peaks were blue-shifted by ∼22 nm for the GQDs-A samples and by ∼59 nm for the GQDs-C samples. In addition, both the GQDs-A and the GQDs-C samples exhibit distinct PLE peaks at 250 nm. On the other hand, the results did not significantly change at a pH of 12 (green spectra of Figures 4b and c). Because the oxygen functional groups of the GQDs are protonated under acidic conditions, the luminescence from the extrinsic states becomes inactive in the PL spectrum (the upper box in Figure 4f),18, 22, 36 resulting in a blue shift of the PL peak position and the appearance of a distinct PLE peak at 250 nm at a pH of 2. On the other hand, the oxygen functional groups are deprotonated under alkaline conditions, and the luminescence from extrinsic states can be maintained (the lower box in Figure 4f). These mechanisms are further confirmed by TRPL experiments on the GQDs-A samples under different pH conditions (Figures 4d and 4e). We found that the carrier transfer from the intrinsic to the extrinsic state disappears at a pH of 2 (Figure 4d), whereas it is maintained when the pH is 12 (Figure 4e). Based on the pH-dependent PL under acidic conditions, the blue shift for GQDs-C is larger than that for GQDs-A due to the existence of more oxygen functional groups in the GQDs-C samples. From the results of the XPS analysis and the optical characterization assessments, we found that GQDs with small diameters (GQDs-A) have fewer oxygen functional groups, resulting in the strongest luminescence among the GQDs with different diameters. This suggests that the origin of the diameter-dependent PL is associated with the degree of oxygenation of the GQDs resulting mainly from the functionalized tartrate on the basal plain and edges of the GQDs. Finally, to investigate the optical efficiency of GQDs (GQDs-A) with respect to those of amino-GQDs, rGOQDs, and GOQDs, we measured the QYs for each type using an absolute PL QYs measurement system under the same conditions, as shown in Figure S3.1. We found that the value of the QYs depends on the concentration of the GQDs and that it becomes saturated at about 4%, whereas the other GQD derivatives with the amino-GQDs,37 rGOQDs,21 and GOQD functional groups have QYs of only about 2%, as shown in Figure 5a and Figure S3.1. Given that our GQDs have relatively high QYs, we assessed GQD-based LED devices (Supporting Video 2). Figure 5b shows the GQD-LED structure, consisting of ITO/PEDOT:PSS/PVK:GQDs/TPBi/LiF/Al, as well as the corresponding energy diagram in which PEDOT:PSS, PVK, and TPBi refer to poly(3,4- ethylenedioxythiophene):poly(styrene sulfonate), poly(N-vinylcarbazole) (PVK), and 2,2′,2″-(1,3,5-benzinetriyl)-tris(1-phenyl-1-H-benzimidazole), respectively. Figure 5c and Figure S3.2 show the current-density voltage (J-V) and the luminance-voltage (L-V) characteristic curves of the fabricated GQD-LEDs. While the turn-on voltage (Vt) is as large as 13 V for the reference device containing PVK without GQDs, it decreases to ∼8 V for the devices with PVK:GQDs (3 wt%). The GQDs dispersed in PVK appear to provide additional carrier transport/injection pathways, resulting in enhancement of the overall current density. This in turn increases the chance of radiative recombination from GQDs; luminance in excess of 1000 cd m−2 was measured for a device with 3.0 wt% GQDs at 16 V, which is, to the best of our knowledge, the highest reported value from GQDs. Figure 5d compares the luminous efficiencies and emission spectra of the devices. Luminous efficiency of 0.65 cd A−1 is obtained for the 3.0 wt% GQDs device at a current density of 1 mA cm−2, while reference devices without GQDs show an efficiency level of only 0.49 cd A−1. This increase in the luminous efficiency is attributed to the fluorescent emission of the GQDs dispersed in the devices, which can be identified in the emission spectrum of the device incorporating GQDs, in sharp contrast to that of the pure PVK sample. In this study, we demonstrated a novel GIC-based route for synthesizing GQDs with high PL tunability and efficiency. The proposed method is cost-effective, eco-friendly, and can easily be scaled up, as it allows the direct fabrication of GQDs using water without a surfactant or chemical solvent. We carried out in-depth investigations of the carrier dynamics in the GQDs and the origin of their PL via an ultrafast time-resolved PL technique. The obtained results show that the fluorescence from the GQDs realized by the proposed method consists of two spectrally overlapped bands that correspond to the transitions associated with the intrinsic bands from sp2 nano-domains with a highly crystalline structure and with extrinsic bands correlated with the surface states. Furthermore, the PL QYs of the GQDs were found to be significantly greater than those of similar variants such as amino-GQDs, rGOQDs, and GOQDs. Owing to these superior optical properties, LED devices containing the GQDs fabricated in this work as an emissive material exhibited far better electroluminescence (EL) than that of the devices without GQDs. Luminance exceeding 1000 cd m−2 was measured for a device with 3.0 wt% GQDs at 16 V; this is, to the best of our knowledge, the highest reported value from EL devices based on GQDs.38, 39 Furthermore, with the achievement of uniformity in size, shape and edges, multicolor PL, and a high QY, the GQDs synthesized using GICs are shown to be highly promising for use in diverse applications ranging from bioimaging to optoelectronic devices. Preparation of GQDs: Potassium sodium tartrate was selected in order to synthesize GICs at low temperature to minimal damage. The initial steps start from the mixing and grinding of potassium sodium tartrate (200 mg) with graphite (20 mg, SP-1, Bay Carbon Inc.) by the mass fraction ratio. Then, the grinded homogeneous mixtures are reacted to the autoclave vessel at 250 °C for 24 h. The manufactured GICs are exfoliated in water and then well-defined GQDs are produced. Then, the sizes of the GQDs were controlled by filtration (10,000 and 8,000 NMWL, Amicon Ultra-15) methods, and the dialyzed in dialysis tubing to remove the remaining salts. Finally, the GQDs were obtained after dried under vacuum for several days. Characterization: Morphology of GQDs was analyzed using an atomic force microscope (AFM, SPA400, SII, Japan) in tapping mode under ambient conditions. UV/Vis spectra (UV-3101PC spectrometer), fluorescence spectra (Perkin-Elmer LS 55 luminescence spectrometer), X-ray photoelectron spectroscopy (XPS, Sigma Probe, AlKα), transmission electron microscopy (TEM, Tecnai G2 F30) analyzes were condcuted. TEM samples were prepared by drying a droplet of the GQDs suspensions on a carbon grid. Raman spectra were obtained from 1200 to 3000 cm−1 using a Raman spectrometer (LabRAM HR UV/Vis/NIR, excitation at 514nm). The FT-IR spectrum was measured using a FT-IR-4100 type-A FT-IR spectrometer with pure KBr as the background from 1000 and 3000 cm−1. All the luminescent data were obtained by handmade setup using the precision cells made of quartz suprasil. The photoluminescence (PL) measurements, such as pH-dependent, size-dependnet, and excitation wavelengths-dependent PL behaviors, were carried out using a 325 nm He-Cd continues-wave (CW) laser, a monochromatic light from a 300 W-xenon lamp, and UV spectrometers (Maya2000, Ocean Optics, USA) as a PL detector at room temperature. The PL excitations were measured by monochromatic light from a 300 W Xenon lamp and a high-sensitive photomultiplier tube as a PL detector. In order to elucidate the recombination dynamics, we carried out time-resolved PL experiments. A mode-locked femto-second pulsed Ti: sapphire laser (Coherent, Chameleon Ultra II) system was used as an excitation source, and the five wavelengths of the pulsed Ti:sapphire laser (266 nm, 300 nm, 350 nm, 400 nm, and 450 nm) were employed. A streak camera (Hamamatsu, C7700–01) was utilized to measure the decay profile of the PL spectra at room temperature. Device Fabrication: GQD-LEDs having the structure of ITO(150 nm)/poly(3,4- ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) (30 nm)/3 wt% GQDs doped poly(N-vinylcarbazole) (PVK)/2,2',2"-(1,3,5-benzinetriyl)-tris(1-phenyl-1-H-benzimidazole) (TPBi) (60 nm)/lithium fluoride (LiF) (1 nm)/Al (100 nm) was fabricated. Glass substrates pre-coated with 150nm thick ITO were cleaned in an ultrasonic bath using soapy water, DI water, acetone and isopropyl alcohol and treated by air plasma using a plasma cleaner (PDC-32G, Harrick Plasma). S. H. Song and M.-H. Jang contributed equally to this work. This research was supported by the Converging Research Center Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2013K000175) and through a grant (2011–0031630) from the Center for Advanced Soft Electronics under the Global Frontier Research Program of the Ministry of Education, Science and Technology, Korea. It was also partially supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (CAFDC/No. 2007–0056090) and by the GRC project of KI for the NanoCentury. The authors acknowledge valuable discussions with J. Park, S. Cho, and G. Chea in LG Display R & D Center. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
