5002

OPTICS LETTERS / Vol. 39, No. 17 / September 1, 2014

All-solution-processed, multilayered CuInS2/ZnS colloidal quantum-dot-based electroluminescent device Jong-Hoon Kim and Heesun Yang* Department of Materials Science and Engineering, Hongik University, Seoul 121-791, South Korea *Corresponding author: [email protected] Received March 10, 2014; revised July 22, 2014; accepted July 22, 2014; posted July 24, 2014 (Doc. ID 207948); published August 18, 2014 While significant progress of electroluminescent (EL) quantum dot light-emitting diodes (QD-LEDs) that rely exclusively on Cd-containing II–VI quantum dots (QDs) has been reported over the past two decades with respect to device processing and performance, devices based on non-Cd QDs as an active emissive layer (EML) remain at the early stage of development. In this work, utilizing highly luminescent colloidal CuInS2 (CIS)/ZnS QDs, allsolution-processed multilayered QD-LEDs are fabricated by sequentially spin depositing a hole transport layer of poly(9-vinlycarbazole), an EML of CIS/ZnS QDs, and an electron transport layer of ZnO nanoparticles. Our focus in device fabrication is to vary the thickness of the QD EML, which is one of the primary determinants in EL performance but has not been addressed in earlier reports. The device with an optimal EML thickness exhibits a peak luminance of 1564 cd∕m2 and current efficiency of 2.52 cd∕A. This record value in efficiency is higher by 3–4 times that of CIS QD-LEDs reported previously. © 2014 Optical Society of America OCIS codes: (160.4236) Nanomaterials; (230.3670) Light-emitting diodes; (230.5590) Quantum-well, -wire and -dot devices. http://dx.doi.org/10.1364/OL.39.005002

I–III–VI2 -based metal chalcogenide semiconductor nanocrystals [quantum dots (QDs)] have proven to be promising active materials as light absorbers and emitters for application to photovoltaic and light-emitting devices [1–3]. Among the various QD composition candidates for realizing an efficient visible-luminescence, ternary Cu–In–S (CIS) [4–9], Ag–In–S [10,11], and quaternary Zn–Cu–In–S (ZCIS) [12–18], Cu–In–Ga–S QDs [19] have been the most intensively investigated for the past five years. Their visible emission wavelength could be widely tuned by engineering the bandgap via precise control of size and/or composition. Based on the efficient photon absorption in the blue region and the resulting high photoluminescent (PL) quantum yield (QY), those multinary QD materials have been successfully utilized as downconverters that are optically excited by a blue lightemitting diode (LED) pumping source, for the fabrication of typically bicolored white QD-LEDs [6,7,16,19]. Along with the advancement of the above downconversion device, an electrically driven luminescent [electroluminescent (EL)] QD-LED has been highlighted as the next-generation planar light-emitting device that may compete with state-of-the-art organic LEDs (OLEDs). In the most commonly adopted device architecture of an EL-type QD-LED, the emissive layer (EML) of close-packed QDs is directly sandwiched by two auxiliary layers, an electron transport layer (ETL) and a hole transport layer (HTL), that enable an efficient charge injection from cathode and anode, respectively, and an improved balance of injected charge carriers. Stimulated by the demonstration of high-performance QD-LEDs comprising high-fluorescence quality Cd-based II–VI QDs, non-Cd III–V (e.g., InP) [20,21] and I–III–VI2 (e.g., CIS or ZCIS), QDs [8,17,18] have also been utilized for device fabrication. In 2011, Zhang et al. [17] reported the early version of an all-solution-processed ZCIS QD-based EL device with the following multilayered 0146-9592/14/175002-04$15.00/0

structure: indium-tin oxide (ITO) // poly(ethylenedioxythiophene):polystyrene sulphonate (PEDOT:PSS) // poly(N, N 0 -bis(4-butylphenyl)-N, N 0 -bis(phenyl)benzidine) (poly-TPD, HTL) // red ZCIS/ZnS core/shell (C/S) QDs //Al. Since an ETL was absent in this device, however, it exhibited not only dual EL components of red and blue-green from QDs and poly-TPD, respectively, but relatively poor EL performance. Specifically, it showed a peak luminance of 450 cd∕m2 at a high voltage of 16 V and a peak external quantum efficiency (EQE) of 0.033%. Shortly afterward, it was shown that improved EL characteristics could be obtained by incorporating a thermally evaporating ETL of tris-(8-hydroxyquinoline) aluminum (Alq3) into the above device architecture, resulting in peak luminances of 1200–1600 cd∕m2 at still high voltages of 12–20 V and peak current efficiencies of 0.49–0.62 cd∕A at a low luminance level of 10 cd∕m2 [18]. More recently, by using yellow and red C/S structured CIS QDs and adopting the same multilayered structure as in [17], Chen et al. published somewhat better EL values of peak luminances of 1700–2100 cd∕m2 at 14–15 V and peak current efficiencies of 0.88–0.92 cd∕A at 10 cd∕m2 [8]. Unlike the above QD-LEDs consisting of poly-TPD HTL and either no ETL or thermally evaporated organic ETL (e.g., Alq3), here an all-solution-processed CIS/ZnS QDbased EL device is constructed by sequentially depositing a HTL of poly(9-vinlycarbazole) (PVK), QDs, and an inorganic ETL of ZnO nanoparticles (NPs). The device fabrication is optimized mainly by varying the thickness of the QD EML, which was not investigated earlier, but critically influences EL figures of merit (i.e., luminance and efficiency). As a result, a bright, efficient QD EL with peak values of 1564 cd∕m2 in luminance and 2.52 cd∕A in current efficiency is obtained from the device with an optimal QD EML thickness. © 2014 Optical Society of America

September 1, 2014 / Vol. 39, No. 17 / OPTICS LETTERS

-1.0

(a) Al

600

-5.0

ITO PEDOT: PSS

-3.4 eV

PVK

-4.0

QDs

(b)

-5.0 eV -5.4 eV

-6.0

-5.7 eV

-4.3 eV

Al

-7.0 -8.0

(b)

(c)

CuInS2 (JCPDS 85-1575)

30

40

50

60

QDs PEDOT:PSS // PVK

70

2 theta (degree) ITO

(c) 100 nm

500

-1.9 eV

-3.0

(a) 400

(d)

-2.0

ZnO NPs

Intensity (a.u.)

PL intensity (a.u.)

Absorbance (a.u.)

ZnS (JCPDS 77-2100)

close proximity to those of the cubic ZnS phase, which is consistent with an appropriate formation of a ZnS overlayer on the CIS core surface [6,7]. The sizes of core and core/shell QDs were found by high-resolution transmission electron microscopy (TEM) to be distributed in the range of 1.7–2.0 nm (not shown here) and 2.7– 3.2 nm [Fig. 1(c)], respectively, leading to an approximate nominal shell thickness of 0.5–0.6 nm. All-solution-processed, multilayered CIS/ZnS QDbased LEDs were fabricated with a conventional device scheme [Fig. 2(a)] as follows: onto a patterned ITO (anode) glass substrate, which was ultrasonically cleaned sequentially with acetone and isopropanol for 15 min and then treated with UV ozone for 20 min, an ∼20 nm thick PEDOT:PSS (AI4083) hole injection layer (HIL) was spin deposited (3000 rpm, 60 s) and baked at 150°C for 30 min in a N 2 -flowing glovebox. PVK (average M W
25; 000–50; 000) in chlorobenzene (10 mg∕ml) was spin-coated and baked at the same conditions as in PEDOT:PSS, forming an ∼30 nm thick HTL on top of the HIL. For multilayer processing compatibility, hexane was chosen as a QD dispersion medium. That is, it disperses the QDs completely, but does not dissolve PVK, thus anticipating no damage of the underlying PVK layer after QD EML deposition. The thickness of the QD EML, determined by a cross-sectional scanning electron microscopic (SEM) measurement, was varied by spin-casting four QD–hexane solutions with different QD concentrations, i.e., 7.1, 9.6, 12.4, and 15.5 mg∕ml, with the coating condition (2000 rpm, 20 s) fixed, resulting in a thickness range of 12–40 nm. The respective spin-deposited EMLs were dried at room temperature. For ∼35 nm thick ZnO ETL deposition, colloidal ZnO NPs were synthesized with the recipe reported in [21]. A scatter/sediment-free ZnO NP–ethanol dispersion with a concentration of 25–30 mg∕ml was spin-coated (1500 rpm, 60 s) on top of the EML, followed by 60°C baking for 30 min. The multilayered device fabrication was then finalized with a thermal evaporation for a 100 nm thick Al cathode. The spin-cast QD EML possessed a uniform surface morphology without noticeable cracks and voids, as shown in a surface SEM image [Fig. 2(b)]. Figure 2(c) presents a representative cross-sectional SEM image of

Energy (eV)

Highly luminescent CIS/ZnS QDs have been prepared by substantially modifying the colloidal chemical synthesis reported in the literature [5,7]. One of the major synthetic differences is to replace a commonly used 1-dodecanethiol (DDT) by 1-octanethiol (OTT), which serves as an organic surface ligand as well as an Sfeeding precursor. Our preliminary test on QD-LED fabrication revealed that, when a QD was capped with DDT with a longer (more insulating) alkyl chain versus OTT, the charge injection and transport across the QD EML were rather limited, leading to low levels in current flow and luminance. In a typical synthesis of CIS/ZnS QDs, a mixture of 0.125 mmol of Cu (I) iodide, 0.5 mmol of In acetate, and 4 ml of 1-octadecene (ODE) were placed in a 50 ml three-neck flask, degassed/Ar-purged during heating to 100°C, and then kept at that temperature for 30 min. The mixture was further heated to 230°C with a rate of 25°C/min, staying turbid. Then, upon swiftly injecting 1 ml of OTT, the growth solution become clear and the reaction for CIS core growth was held for 5 min. For C/S structured QDs with ZnS shells, the first Zn stock mixture, comprising 4 mmol of Zn acetate, 4 ml of oleic acid (OA), and 2 ml of ODE, was added into the CIS core growth solution at 240°C, followed by a 1 h long shelling reaction. Then, the second Zn stock solution with the same formulation as above but a different Zn precursor, Zn stearate instead of Zn acetate, was sequentially introduced, and this shelling proceeded for 2 h at 240°C. As-synthesized CIS/ZnS QDs were repeatedly washed by centrifugation with a solvent combination of hexane/ethanol and finally dispersed in hexane for optical characterization and device fabrication. As shown by the absorption and PL spectra of second-shelled CIS/ZnS QDs [Fig. 1(a)], large Stokes-shifted emission (peaking at 580 nm) plus broadband emission (bandwidth of 114 nm) indicate that the radiative recombination is involved with the intrinsic defect energetic levels inside the bandgap [5–9]. PL QY of CIS/ZnS QDs increased from 59% for first shelling to 65% for second shelling. As noticed from a PL blueshift from first to second shelling, the alloying of a CIS core with ZnS to a certain degree is likely to occur throughout the shelling procedure [19]. As seen from an x-ray diffraction (XRD, Rigaku, Ultima IV) pattern [Fig. 1(b)], the reflection peaks of CIS/ZnS QDs became a bit separated from those of the tetragonal chalcopyrite CuInS2 phase, being in

5003

100 nm

700

Wavelength (nm)

Fig. 1. (a) Absorption and PL spectra (inset: fluorescent image under UV illumination), (b) XRD pattern, and (c) TEM image of second-shelled CIS/ZnS QDs. PL spectra of first- versus secondshelled CIS/ZnS QDs are also compared.

Fig. 2. (a) Device schematic of all-solution-processed, multilayered QD-LED comprising ITO // PEDOT:PSS // PVK // QDs // ZnO NPs // Al. (b) Surface and (c) cross-sectional SEM images of ITO // PEDOT:PSS // PVK // QDs with an ∼20 nm thick QD EML. (d) Proposed energy levels of the multilayered device.

10

10

10

(a)

0

4

10

2

10

0

10

-2

10

-4

-2

10

-4

10

-6

10

-8

device A device B device C device D

3

4

5

6

7

8

9

1.5

(b)

device A device B device C device D

1.0

0.5

0.0

Voltage ( V)

0

100 200 300 400 500 600 700 2

Current density (mA/cm )

Fig. 4. Variations of (a) current efficiency and power efficiency as a function of drive voltage and (b) EQE as a function of current density for devices with different QD EML thicknesses.

These peak efficiencies are higher by at least 3–4 times in current efficiency and by 2 times in power efficiency as compared to the CIS QD-LEDs reported earlier [8,18]. It is also worth of mention that the peak efficiency values in the literature were obtained at a negligibly low luminance level of 10 cd∕m2 , while our results were collected at a much higher luminance of 791 cd∕m2 at 6 V. Figure 4(b) presents the variations of EQE as a function of current density. The highest performance device B exhibited a peak EQE of 1.1% at a current density of 31.2 mA∕cm2 (at 6 V). All four devices showed a similar roll-off behavior, i.e., the reduction of EQE with increasing current density. This efficiency roll-off characteristic at higher drive voltage or current regimes, which has been intensively investigated in Cd-containing II–VI QD-LEDs in recent years, is presumably attributable to a more spatial separation of electron/hole wave functions at a higher drive voltage [25] and/or a higher degree of QD charging at a higher current [26]. Figure 5 shows the voltage-dependent evolution of EL spectra and images (insets) of the optimized device, i.e., device B, with regard to luminance and efficiency. EL originated entirely from CIS/ZnS QDs without observing any parasitic PVK emissive component, indicative of the occurrence of the radiative combination of injected carriers in the EML region only. The luminances corresponding to drive voltages of 5, 6, and 7 V were 16, 791, and 1564 cd∕m2 , respectively. Compared with the solution PL in Fig. 1(a), EL at 5 V was markedly redshifted, peaking at 603 nm. Such a spectral diffusion to a lower energy side is a common observation in II–VI QD-LEDs,

device A device B device C device D

2

400

200

(b)

device A device B device C device D

1600

Luminance (cd/m )

600

(a)

1200

800

EL Intensity (a.u.)

7V

800

2

Current Density (mA/cm )

ITO // PEDOT:PSS // PVK // QD EML, where an ∼20 nm thick EML was deposited with a coating solution with 9:6 mg∕ml QD concentration. Figure 2(d) shows the proposed energy levels of the multilayered device. The designated energy levels of the HTL and the ETL were the values reported from literature [22]. The valence band level of QDs was tentatively approximated based on the values obtained by cyclic voltammetry measurements of the earlier investigation [3,23], and their conduction band level was determined by applying the optical bandgap (2.3 eV) estimated from the absorption spectrum of Fig. 1(a). Using QD–hexane solutions with the different QD concentrations specified above, four QD-LEDs with respective EML thicknesses of 12, 20, 32, and 40 nm were fabricated (termed devices A, B, C, and D, respectively) and characterized with a Konica-Minolta CS-2000 spectroradiometer coupled with a Keithley 2400 V and current source under ambient conditions. The EML thicknessdependent variations of current density-voltage and luminance-voltage are shown in Figs. 3(a) and 3(b), respectively. As expected, the current density decreased throughout the whole bias range with increasing QD EML thickness as a result of a smaller charge injection from a lower effective electric field. Compared to the previous CIS QD-LEDs with an organic ETL of Alq3 [8,18], our devices with an inorganic ETL of ZnO NPs exhibited substantially higher current flows at the same drive voltage, primarily ascribable to a much higher electron mobility of ZnO (∼2 × 10−3 cm2 ∕Vs) versus Alq3 [24]. Consistent with the EML thickness-dependent variations of current density, the luminance tended to be higher from the device with a thinner QD EML, even though device B exhibited a slightly brighter EL than the thinnest QD EML-based device. The peak luminance of device B was 1564 cd∕m2 at a drive voltage of 7 V. As recognized from the relations of current efficiency and power efficiency versus drive voltage in Fig. 4(a), the current densities of devices C and D with thick QD EMLs were not sufficiently low to compensate for low luminances observed, leading to the low device efficiencies. The peak current and power efficiencies in devices C and D were 1.66 and 1.26 cd∕A and 0.87 and 0.56 lm∕W, respectively. Meanwhile, in the case of two thin QD EML-based devices, device B with a lower current density and higher luminance compared to device A possessed the best EL performance with a peak current efficiency of 2.52 cd∕A and power efficiency of 1.32 lm∕W.

Power Efficiency (lm/W) Exteranl Quantum Efficiency (%)

OPTICS LETTERS / Vol. 39, No. 17 / September 1, 2014

Current Efficiency (cd/A)

5004

7V

6V 6V

5V × 20

400

5V

0

500

0 0

1

2

3

4

5

6

Voltage (V)

7

8

9

0

1

2

3

4

5

6

7

8

9

Voltage (V)

Fig. 3. QD EML thickness-dependent variations of (a) currentdensity-voltage and (b) luminance-voltage.

550

600

650

700

750

800

850

Wavelength (nm)

Fig. 5. Voltage-dependent evolution of EL spectra and corresponding EL images (insets) of the optimized device (i.e., device B). The size of a square-shaped pixel was 3 mm × 3 mm.

September 1, 2014 / Vol. 39, No. 17 / OPTICS LETTERS

primarily originating from the inter-QD energy transfer in a closely packed QD EML [27,28]. It should be noted that, despite a substantially Stokes-shifted emission of the present QDs, a certain degree of Förster resonant energy transfer (FRET) in the QD ensemble, as sensed from a nontrivial spectral overlap between their absorption and emission [Fig. 1(a)], would lead to redshift as well as quenching of EL. Also, conventional II–VI QD-LEDs have exhibited a voltage-dependent gradual EL redshift, which is possibly caused by either a larger local Joule heating from a larger current flux [26] or larger exciton polarization under a higher electric field [28]. Intriguingly, the opposite trend of a systematic blueshift in EL with increasing voltage, i.e., 603, 597, and 595 nm at 5, 6, and 7 V, respectively, was observed in our device. This unusual blueshifted EL cannot be clearly elucidated for the present, but may be correlated to a signature of the increasing contribution from multicarrier states involved emission, as proposed in a recent work on QD-LEDs containing CdSe∕CdSe0.5 S0.5 ∕CdS core/alloy layer/shell QD of a size of 14 nm [26], or the variation in population density of intragap states as a function of applied voltage. In summary, 580 nm emitting CIS/ZnS QDs with a PL QY of 65% were first synthesized and then applied for ELtype QD-LED fabrication. Multilayer-structured QD-LEDs comprising hybrid charge transport layers of ∼30 nm thick organic PVK HTL and an ∼35 nm thick inorganic ZnO NP ETL were constructed by all-solution processing. In an effort to optimize device performance in regard to luminance and efficiency, the thickness of the QD EML was varied in the range of 12–40 nm. The device with an ∼20 nm thick EML achieved the best results, with a peak luminance of 1564 cd∕m2 , current efficiency of 2.52 cd∕A, power efficiency of 1.32 lm∕W, and EQE of 1.1%. These peak efficiencies far surpassed the highest values from the earlier CIS QD-LEDs. In addition, an unusual blueshift in EL with a higher drive voltage, which is in contrast to the trend observed in the conventional Cd-containing II–VI QD-LEDs, was for the first time reported, to the best of our knowledge. This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (No. 2013R1A2A2A01068158). References 1. D. Aldakov, A. Lefrancois, and P. Reiss, J. Mater. Chem. C 1, 3756 (2013). 2. H. Z. Zhong, Z. L. Bai, and B. S. Zou, J. Phys. Chem. Lett. 3, 3167 (2012). 3. X. Yuan, J. L. Zhao, P. T. Jing, W. J. Zhang, H. B. Li, L. G. Zhang, X. H. Zhong, and Y. Masumoto, J. Phys. Chem. C 116, 11973 (2012).

5005

4. L. Li, T. J. Daou, I. Texier, T. K. C. Tran, Q. L. Nguyen, and P. Reiss, Chem. Mater. 21, 2422 (2009). 5. L. Li, A. Pandey, D. J. Werder, B. P. Khanal, J. M. Pietryga, and V. I. Klimov, J. Am. Chem. Soc. 133, 1176 (2011). 6. W. S. Song and H. Yang, Appl. Phys. Lett. 100, 183104 (2012). 7. W. S. Song and H. Yang, Chem. Mater. 24, 1961 (2012). 8. B. K. Chen, H. Z. Zhong, W. Q. Zhang, Z. A. Tan, Y. F. Li, C. R. Yu, T. Y. Zhai, Y. S. Bando, S. Y. Yang, and B. S. Zou, Adv. Funct. Mater. 22, 2081 (2012). 9. J. Kolny-Olesiak and H. Weller, ACS Appl. Mater. Interfaces 5, 12221 (2013). 10. Y. Hamanaka, T. Ogawa, M. Tsuzuki, and T. Kuzuya, J. Phys. Chem. C 115, 1786 (2011). 11. T. Uematsu, T. Doi, T. Torimoto, and S. Kuwabata, J. Phys. Chem. Lett. 1, 3283 (2010). 12. X. Tang, W. Cheng, E. S. Choo, and J. Xue, Chem. Commun. 47, 5217 (2011). 13. L. De Trizio, M. Prato, A. Genovese, A. Casu, M. Povia, R. Simonutti, M. J. P. Alcocer, C. D’Andrea, F. Tassone, and L. Manna, Chem. Mater. 24, 2400 (2012). 14. J. Zhang, R. G. Xie, and W. S. Yang, Chem. Mater. 23, 3357 (2011). 15. W. Zhang and X. Zhong, Inorg. Chem. 50, 4065 (2011). 16. B. Chen, H. Zhong, M. Wang, R. Liu, and B. Zou, Nanoscale 5, 3514 (2013). 17. Y. Zhang, C. Xie, H. Su, J. Liu, S. Pickering, Y. Wang, W. W. Yu, J. Wang, Y. Wang, J. I. Hahm, N. Dellas, S. E. Mohney, and J. Xu, Nano Lett. 11, 329 (2011). 18. Z. Tan, Y. Zhang, C. Xie, H. Su, J. Liu, C. Zhang, N. Dellas, S. E. Mohney, Y. Wang, J. Wang, and J. Xu, Adv. Mater. 23, 3553 (2011). 19. W. S. Song, J. H. Kim, J. H. Lee, H. S. Lee, Y. R. Do, and H. Yang, J. Mater. Chem. 22, 21901 (2012). 20. X. Y. Yang, Y. Divayana, D. W. Zhao, K. S. Leck, F. Lu, S. T. Tan, A. P. Abiyasa, Y. B. Zhao, H. V. Demir, and X. W. Sun, Appl. Phys. Lett. 101, 233110 (2012). 21. J. Lim, M. Park, W. K. Bae, D. Lee, S. Lee, C. Lee, and K. Char, ACS Nano 7, 9019 (2013). 22. K. H. Lee, J. H. Lee, W. S. Song, H. Ko, C. Lee, J. H. Lee, and H. Yang, ACS Nano 7, 7295 (2013). 23. E. Arici, N. S. Sariciftci, and D. Meissner, Adv. Funct. Mater. 13, 165 (2003). 24. L. Qian, Y. Zheng, J. G. Xue, and P. H. Holloway, Nat. Photonics 5, 543 (2011). 25. D. Bozyigit, O. Yarema, and V. Wood, Adv. Funct. Mater. 23, 3024 (2013). 26. W. K. Bae, Y. S. Park, J. Lim, D. Lee, L. A. Padilha, H. McDaniel, I. Robel, C. Lee, J. M. Pietryga, and V. I. Klimov, Nat. Commun. 4, 2661 (2013). 27. B. S. Mashford, M. Stevenson, Z. Popovic, C. Hamilton, Z. Q. Zhou, C. Breen, J. Steckel, V. Bulovic, M. Bawendi, S. Coe-Sullivan, and P. T. Kazlas, Nat. Photonics 7, 407 (2013). 28. V. Wood, M. J. Panzer, J. M. Caruge, J. E. Halpert, M. G. Bawendi, and V. Bulovic, Nano Lett. 10, 24 (2010).