Quantum-Dot Light-Emitting Diodes with Nitrogen-Doped Carbon Nanodot Hole Transport and Electronic Energy Transfer Layer

Electroluminescence efficiency is crucial for the application of quantum-dot light-emitting diodes (QD-LEDs) in practical devices. We demonstrate that nitrogen-doped carbon nanodot (N-CD) interlayer improves electrical and luminescent properties of QD-LEDs. The N-CDs were prepared by solution-based bottom up synthesis and were inserted as a hole transport layer (HTL) between other multilayer HTL heterojunction and the red-QD layer. The QD-LEDs with N-CD interlayer represented superior electrical rectification and electroluminescent efficiency than those without the N-CD interlayer. The insertion of N-CD layer was found to provoke the Förster resonance energy transfer (FRET) from N-CD to QD layer, as confirmed by time-integrated and -resolved photoluminescence spectroscopy. Moreover, hole-only devices (HODs) with N-CD interlayer presented high hole transport capability, and ultraviolet photoelectron spectroscopy also revealed that the N-CD interlayer reduced the highest hole barrier height. Thus, more balanced carrier injection with sufficient hole carrier transport feasibly lead to the superior electrical and electroluminescent properties of the QD-LEDs with N-CD interlayer. We further studied effect of N-CD interlayer thickness on electrical and luminescent performances for high-brightness QD-LEDs. The ability of the N-CD interlayer to improve both the electrical and luminescent characteristics of the QD-LEDs would be readily exploited as an emerging photoactive material for high-efficiency optoelectronic devices.


S1. SEM inspection of multilayer HTLs
The N atoms can be doped into the basal (or surface) plane of a graphene sheet in several different forms. The schematic illustration of Figure S2 shows atomic configurations of doped N in graphene (or CD layer), which are pyridinic, pyrrolic, and quaternary N. The characteristic XPS peaks for pyridinic, pyrrolic, and quaternary N are located at about 398.3, 400.3, and 401.4 eV, respectively, as shown in Figure 1e. It is known that the electron density in pyridinic and pyrrolic N-doped graphene is much less than undoped graphene 2 . Additionally, the N-related functional group acts as a p-type acceptor impurity.
Especially, in Figure 1f, the integrated peak intensities of pyridinic, pyrrolic, and functional group N are higher than that of quaternary N, thus our N-CD with electron deficiency is thought to behave ptype conductor, similar to the findings in N-doped carbon nanotubes 3 . Figure S3. Peak-fitted (a) C 1s and (b) O 1s XPS spectra of N-CD layer. The spectra were de-convoluted using Doniach-Šunjić functions with Lorentzian line shapes. Figure S3a shows the C 1s spectrum of the N-CD exhibiting strong multiple peaks at above 285 eV, which is noticeably distinguished from nearly symmetric C 1s peak of high-quality undoped graphene 4,5 .

S3. C 1s and O 1s XPS results of the N-atom-doped CD layer
The C 1s band was de-convoluted into the following characteristic peaks: carbon in graphite (sp 2  peak, the CONH groups were formed on the N-CD to some extent using our N-CD synthesis method. The FT-IR data (Figure 1d in main text) also support the argument.  Figure S4. UV-Vis absorption spectrum of the N-CD layer and PL spectrum of the QD layer. Figure S4 shows the UV-Vis absorption spectrum of the N-CD layer and the PL spectrum of the QD layer. The QD yielded a sharp recombination spectrum with a peak centered at 622 nm, while the spectrum of N-CD contained a broad absorption tail originated from N-or O-related functional groups at the surface of N-CD. Since the spectral overlap region was not observed in Figure S4, the luminescence of QD was not surmised to be absorbed to N-CD in the QD/N-CD bilayer. Figure S5. Peak-fitted photoluminescence spectrum of the N-CD layer. The spectra were deconvoluted with Gaussian line shapes.

S5. Peak-fitted PL spectrum of the N-atom-doped CDs
To date, the PL emission of N-CD has been intensively studied, which referred the quantum size effect 7-9 , zigzag sites 10 , different degree of π-conjugation 11,12 , edge states 13 , surface defects 14,15 , surface groups 16 , N-doping 1, 9,17 , and surface passivation 18 . Figure S5 shows that PL band for the N-CD layer,     In order to measure the thickness of each layer, we complimentarily employed scanning electron microscopy and transmission electron microscopy. For the electron microscopic inspections, the samples were cross-sectioned by ion milling machine. Roughly, the thickness of each layer was measured: the thickness of PEDOT:PSS, PVK, QD, and ZnO was ca. 30, 22, 15, and 30 nm, respectively ( Figure S9). From combined TEM and EDX measurements ( Figure S10), the thickness of ZnO/QD was measured at 45 nm by mapping Zn element, and the thickness of Cd-content layer (QD layer) was 15 nm. Thus, the ZnO layer is estimated to be 30 nm. From high-magnification TEM inspection, the average thickness of N-CD layer was estimated to be 4-5 nm, as shown in Figures S11b and 11c. Figure S12. Time-resolved PL spectroscopic analysis. TR-PL spectra of N-CD and QD/N-CD layers measured at 590 nm.

S9. Time-Resolved Photoluminescent Characteristics of QD/N-CD Layer
For the TR-PL analysis, the PL decays were recorded at emission wavelengths of 590 nm ( Figure   S12), which are one of the characteristic PL emission peaks of the N-CD layers (shown in Figure 3a) in the spectral overlapping region between QD absorption and N-CD PL emission. The PL decay curves were numerically analyzed using a bi-exponential model fitting and the fitting results are summarized in Table S1. Figure S12 shows the PL decay of the N-CD layer and QD/N-CD bilayer measured at 590 nm, exhibiting that the PL emission of the bilayer diminished faster than that of the N-CD layer because the carriers in the energy-donor N-CD layer quenched as a result of excitation energy transfer to the acceptor QD layer in the QD/N-CD bilayer. The FRET efficiency (ξ) was determined to be 0.19 and the FRET rate (k FRET ) was calculated to be 4.5 ×10 7 s -1 . It is imperative to optimize the annealing temperature for high-efficiency QD-LEDs because too high temperature degrades (or oxidizes) as-coated layers while too low temperature does not sufficiently de-gas the organic solvent and residual moisture. It is well known that 2-methoxyethanol used as solvent of N-CD is evaporated over 125 °C. If the annealing is carried out above 250 °C, PSS in PEDOT:PSS can be certainly degraded, as reported elsewhere 28,29 . Also, PVK is known to be highly degraded over 300 °C 30 . Considering the literatures, the annealing temperature of 125-250 °C may be desirable.

S10. Effect of annealing temperature on electrical and electroluminescent performances of QD-LEDs
For optimizing the annealing temperature, we carried out the annealing at 60, 150, and 200 °C after spin coating of each layer, and compared the device performances. In Figure S13, the QD-LEDs annealed at 150 °C showed the highest current density, luminance, and current efficiency, representing superior QD-LED performances. We believe that 60 °C was insufficient to vaporize the solvent and moisture that can degrade device performances. The annealing temperature of 200 °C was found to be better than 60 °C, but was inferior to 150 °C in terms of luminance and current efficiency. Since we performed all the spin-coating and annealing procedures under normal air ambient condition, too high temperature (200 °C) might possibly oxidize the surface of as-coated layer. Hence, we believe that the annealing at 150 °C was most effectual for removing the moisture and organic solvent with little degradation among a series of annealing experiments shown in Figure S13. Table S1. Photoluminescence decay parameters of the N-CD and QD/N-CD layers derived from the equation (1) in the main text. The emission wavelength was adjusted to the donor and acceptor emission, respectively. The sum of the individual amplitude W i is normalized to unity and <τ> = ∑ w i • τ i is the amplitude-weighted average decay time.