SUPPLEMENTARY INFORMATION DOI: 10.1038/NNANO.2012.63 Bright infrared quantum-dot light-emitting diodes through inter-dot spacing control Liangfeng Sun, Joshua J. Choi, David Stachnik, Adam C. Bartnik, Byung-Ryool Hyun, George G. Malliaras, Tobias Hanrath, Frank W. Wise NATURE NANOTECHNOLOGY www.nature.com/naturenanotechnology 1
A. Estimation of radiance The radiance of the LEDs was not presented in Reference 3 cited in our report. However, it can be estimated according to the current density and external efficiency data plotted in Fig. 2B in the reference. For instance, the maximum radiance of the LED in Reference 3 can be estimated, as shown below. According to the definition of external quantum efficiency ( external efficiency used in Reference 3 has the same definition as external quantum efficiency, confirmed by the corresponding author of Reference 3), EQE = (number of photons exit the device structure per second) / (number of charges injected to the device per second). The number of photons exit the device structure per second is equal to the light power divided by the photon energy (P light / hν, where h is Plank constant, ν is frequency), the number of charges injected to the device per second is equal to the current divided by the elemental charge (I / 1.6X10-19 ), then we have EQE = (P light / hν) / (I / 1.6X10-19 ) P light is in unit of Watt (W), while hν is usually expressed in unit of ev (electron volt). Then a conversion factor (1.6X10-19 J/eV) is need for calculation, then EQE = (P light / [1.6X10-19 J/eV (hν in ev)] / (I / 1.6X10-19 ) = (P light /I ) * [1/(hν in ev)] There is a relationship between the photon energy (in ev) and the photon wavelength (in nm), photon wavelength (in nm) = 1240 / photon energy (in ev). Then EQE can be written as, EQE = (P light /I ) *[ photon wavelength (in nm) / 1240]. According the spectrum of the electroluminescence plotted in Fig 3B of Reference 3, we chose the photon wavelength to be 1240 nm (for simplification purpose) which is close enough to the central wavelength of the electroluminescence, and then we calculated the maximum electroluminescence power, P light = EQE * I / (1240/1240) = 0.005 * 0.5 ma = 2.5 µw, where, I = J A = (0.5 ma/mm 2 ) * (1 mm 2 ) = 0.5 ma Then, Radiance = W/(pixel area)/(solid angle)=2.5 µw/1mm 2 /π = 0.8 W sr -1 m -2. B. Experimental error estimation The error of the current measurement (made with a Keithley 236) is < 0.1% in the current range we have measured, and this contributes negligible error to the EQE. The uncertainty from the light detector is < 0.2 nw (±0.1 nw). This contributes greater relative errors in radiance and EQE at low current density than at high current density. For instance, in the
measurement of the highest EQE (2%), we measured an average of 0.925 nw light power with a statistical error ±0.05 nw (±5.4% relative error), where the uncertainty (±0.1 nw) of the detector contributed ±11% relative error. There is a time-dependent shift (in a few minutes) of the detected light power, which could contribute to the systematic error. We suppressed this by zeroing the detector before each measurement. The uncertainty of the sample position relative to the detector is the other error source in the light power measurement. This error is estimated by intentionally shifting the sample around the nominal position while recording the change of the light power. This error is estimated to be < 20% (±10%). The overall relative error in light power measurements is ±16% (square root of the sum of 5.4% square, 11% square and 10% square) at the lowest current density shown in Fig. 2C inset. The relative error of EQE is the same since the current-density error is negligible. The error will be less at higher current density, since the uncertainty of the detector contributes less relative error in the measured light power. The highest EQE can be written as (2.0±0.3) %. C. Current-voltage characteristic of a QD-LED Figure S1. Current density - applied voltage characteristic of a typical QD-LED. The current density increase sharply when the applied voltage is larger than 1 V, but is negligible at inverse bias. D. QD size-dependent EQE of the QD-LEDs
Figure S2. EQE of the QD-LEDs reaches the maximum when the diameter of the QD is 3.5 nm. All the QDs are capped by MOA linkers. The error bars indicate the EQE fluctuation in a broad range of bias (0 ~ 11 V), and from diode to diode. E. Inter-NC distance measurements with GISAXS Figure S3. D-spacing data from GISAXS measurements show a trend of longer inter-qd spacing with longer ligands between PbS QDs with absorption peak at 670 nm.
Figure S4. D-spacing data from GISAXS measurements show a trend of longer inter-qd spacing with longer ligands between PbS QDs with absorption peak at 905 nm. Figure S5. D-spacing data from GISAXS measurements show a trend of longer inter-qd spacing with longer ligands between PbS QDs with absorption peak at 1140 nm.
Figure S6. D-spacing data from GISAXS measurements show a trend of longer inter-qd spacing with longer ligands between PbS QDs with absorption peak at 1425 nm. F. EQE dependence on the current density for the devices treated with different linker molecules Figure S7. The EQE for each device increases slightly, and then decreases with current density. It reaches to a maximum at certain current density which depends on the linker molecules (MPA, MHA, MOA and MUA).
G. Metal-oxide nanoparticle synthesis The ZnO was synthesized in ambient air using an adaptation of a previously published method 1. 1.76 g of zinc acetate dihydrate was dissolved in 150 ml of ethanol with vigorous stirring and heated to 60 o C for 1 hour. Parafilm was used to block solvent evaporation during the heating and the entire duration of the reaction. In a separate container, 6.4 ml of a tetramethyl ammonium hydroxide (TMAH) solution (28% in MeOH) was added to 50 ml of ethanol. Over 10 minutes, the TMAH solution was slowly added to the zinc acetate solution at regular intervals. During this time the temperature of the zinc acetate solution was maintained at 60 o C with continuous stirring. Subsequently, the solution was heated at 60 o C for 30 minutes after which the heater and stirrer were removed. After cooling to room temperature, the solution was kept refrigerated at ~ 5 o C. This mother-liquor solution was stable for over ~5 months. ZnO nanoparticles were collected from the refrigerated solution immediately before device fabrication. To prepare devices, 5 ml of the solution was typically mixed with 20 ml of hexane to precipitate the ZnO particles. After centrifugation, the supernatant was removed and the white precipitate was dissolved in a 1:2 by volume chlorobenzene:methanol mixture. H. Conductivity of ZnO film before and after photodoping
Figure S8. The current-voltage curve shows that illumination with AM1.5 light increases the conductivity of ZnO film (about 100 nm) sandwiched in between ITO and aluminum electrode. This effect was previously attributed to increased concentration of mobile electrons due to UV photodoping 2. I. Comparison QD-LED performance in terms of both EQE and radiance
Figure S9. Relation between EQE and radiance of our QD-LED device (black solid squares) and the device reported in Ref.3 (red dots). The data indicated by black solid squares are extracted from the plots of Fig. 2C in our manuscript. The data indicated by red dots are extracted from the plots of Fig. 2B in Reference 3. J. Time-resolved PL from QD films Figure S10. Time-resolved photoluminescence from PbS QDs films on top of PEDOT:PSS films coated on a silicon wafer. The QDs are treated with MPA, MHA, MOA and MUA
ligands respectively. The PL lifetimes (after deconvolution from the system response) defined as the times to decay to 10% of the initial are 20 ns (MPA), 168 ns (MHA), 324 ns (MOA) and 468 ns (MUA), respectively. The ratios of integrated PL (proportional to PL QE) values for the different films are MPA : MHA : MOA : MUA = 0.06 : 0.50: 0.70 : 1. K. PL quenching and red-shifting with shorter spacing ligands Figure S11. PL spectra from PbS QDs (diameter 4.5 nm) capped by OA (cyan), MPA (black), MHA (red), MOA (green) and MUA (blue) ligands. The QDs were spin-coated on top of PEDOT:PSS/ITO/glass. Ligands exchange was followed by treating four samples with MPA, MHA, MOA and MUA solutions respectively. L. Dependence of EQE on QD size and ligand length
Figure S12. Dependence of EQE from the QD-LEDs made of 2.7 nm (squares), 3.5 nm (circles), 4.5 nm (up triangles), 5.6 nm (down triangles) and 6.5 nm (diamonds) diameter QDs on the number of carbon atoms in the the ligands (MPA[3], MHA[6], MOA[8] and MUA[11]). M. Size-dependent energy levels of the QDs and turn-on voltages of the QD-LEDs The size tunable energy levels in QDs provide several advantageous experimental degrees of freedom. Quantum confinement can be exploited to systematically adjust the energy level offsets as well as to tune the emission wavelength. As mentioned above, the built-in potential between the ZnO and PEDOT:PSS prevents direct charge injection into the QDs 3. In addition, there are potential differences between the QD layer and the carrier-transport layers, and these differences depend on the size of the QDs. The variation of the QD energy levels is shown in Figure S13A. Based on this energy level alignment, we expect the LED will not turn on until the applied bias is enough to overcome all potential barriers. The expected trend is validated in Figure S13B, which
shows the relationship between QD size and the turn-on voltage (arbitrarily defined as the voltage that produces electroluminescent power greater than 100 nw). We observe two distinct performance domains: for QDs with diameter above 5.5 nm, the barrier for electron (hole) injection at the interface between the ZnO and the QDs (PEDOT:PSS and QDs) is negligible, and the turn-on voltage is mainly determined by the built-in potential (Figure S13B). For diameters below 5.5 nm, the turn-on voltage increases as the size of the QDs decreases, since the barriers for electron and hole injection increase as QD size decreases. An analogous trend between open circuit voltage and QD diameter was observed in PbSe QD solar cells 4. Figure S13. (A) Calculated ionization potential (dash line) and electron affinity (solid line) levels of PbS QDs. The fuzzy bars represent the electron affinity of ZnO nanoparticles and the ionization potential of PEDOT:PSS film. (B) Dependence of the turn-on voltage on the quantum-dot size. The QD energy gap is calculated by subtracting the ionization potential from the electron affinity for each QD diameter. The horizontal error bars indicate the size
distributions of the QDs, the vertical error bars indicate the fluctuation of the turn-on voltage over the different linker molecules and from diode to diode. References 1. Wood, A., et al. Size Effects in ZnO: The Cluster to Quantum Dot Transition. Aust.J.Chem. 56, 1051-1057 (2003). 2. Lakhwani, G., et al. Probing Charge Carrier Density in a Layer of Photodoped ZnO Nanoparticles by Spectroscopic Ellipsometry. The Journal of Physical Chemistry C 114, 14804-14810 (2010). 3. Malliaras, G. G., Salem, J. R., Brock, P. J. & Scott, C. Electrical characteristics and efficiency of single-layer organic light-emitting diodes. Phys. Rev. B 58, R13411-R13414 (1998). 4. Choi, J. J., et al. PbSe nanocrystal excitonic solar cells. Nano Lett. 9, 3749-3755 (2009).