Air-Stable Operation of Transparent, Colloidal Quantum Dot Based

Dec 22, 2009 - ... voltages (Vrms) of 10−15 V at frequencies up to 50 kHz. Figure 3a presents the typical current density versus voltage (J−V) cha...
0 downloads 0 Views 3MB Size
Download PDF
pubs.acs.org/NanoLett

Air-Stable Operation of Transparent, Colloidal Quantum Dot Based LEDs with a Unipolar Device Architecture Vanessa Wood,† Matthew J. Panzer,† Jean-Michel Caruge,‡ Jonathan E. Halpert,‡ Moungi G. Bawendi,‡ and Vladimir Bulovic´*,† †

Department of Electrical Engineering and Computer Science and ‡ Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 ABSTRACT We report a novel unipolar light-emitting device architecture that operates using direct-current, field-driven electroluminescence of colloidally synthesized quantum dots (QDs). This device architecture, which is based only on transparent ceramics and QDs, enables emission from different color QDs and, for the first time, constant QD electroluminescence during extended operation in air, unpackaged. KEYWORDS Quantum dots, electroluminescence, metal oxides, ionization

C

olloidally synthesized quantum dots (QDs) have shown promise as the active material in light-emitting devices (LEDs) because of their tunable, narrow band emission and high photoluminescent efficiencies throughout the visible wavelength region. Efficient electroluminescence (EL) from QDs with different emission wavelengths and chemical compositions has been demonstrated and explained in LEDs with small molecule organic charge transport layers.1-6 More recently, QD-LEDs incorporating p- and n-type inorganic charge transport layers demonstrated electrical excitation of QDs in air-stable structures that operate via direct charge injection.7-9 In these devices, efficient EL is dependent on favorable energy band alignment between the charge transport layers and the film of QD emitters.9 To obviate the restriction of precise band alignment, in this study we introduce a new unipolar device architecture in which a QD multilayer is embedded in a transparent, n-type ceramic matrix and which is designed to enable only one type of charge carrier (in this case, electrons) to be injected into the device. We present results indicative of a novel, field-driven mechanism for QD EL and demonstrate both green and red QD emission, with a peak luminance of 1000 Cd/m2 and a luminous efficiency of 1 Cd/A from one face of our most efficient transparent devices. These LEDs exhibit long shelf lives and enable constant luminance over extended operating times in air, unpackaged. Previously, QDs have been incorporated into thin film LEDs that use some combination of semiconducting polymers, small organic molecules, III-V materials, and ceramics as charge transport layers.1-13 All of these devices rely on a p-i-n type architecture, which requires the simulta-

neous transport of electrons and holes toward the QD emissive layer. For example, the most efficient visibleemitting QD-LEDs to date utilize a monolayer of closely packed QDs sandwiched between hole and electron transporting organic layers.4 The use of molecular organic materials in these QD-LED structures introduces the fabrication challenges similar to those facing organic LEDs (OLEDs), namely, the need for packaging in order to prevent degradation due to atmospheric oxygen and water vapor exposure. We demonstrated previously that radio frequency (rf) sputter-deposited metal oxides, which are chemically and morphologically stable in air, can be used as charge transport layers to achieve robust QD-LEDs that do not require packaging.8 These first all-inorganic devices exhibited lower efficiency than state of the art organic-based QD-LEDs due in part to the difficulty of balancing hole and electron injection into the luminescent QD film.9 Furthermore, since the most efficient red, green, and blue QDs have different chemical compositions,4 a series of carefully selected transport layers is required to achieve multicolor EL with comparable efficiency and luminosity.9 In response to these concerns, in this work we demonstrate QD-LEDs that utilize a unipolar (electron-transporting) ceramic matrix. We form an n-i-n QD-LED structure by surrounding the QD film with n-type, alloyed ZnO and SnO2 (ZTO). ZTO was used previously as the electron transport layer in our initial QD-LEDs with metal oxide charge transport layers, resulting in stable device operation.8 Two types of device structures (structure 1 and structures 2A-2C), differentiated by the presence of at least one ZnS layer in structures 2A-2C, are depicted schematically in Figure 1a. As discussed below, ZnS layers are included in order to gain a deeper understanding of operating mechanism and improve device stability. A

* Corresponding author, [email protected]. Received for review: 07/28/2009 Published on Web: 12/22/2009 © 2010 American Chemical Society

24

DOI: 10.1021/nl902425g | Nano Lett. 2010, 10, 24-29

FIGURE 1. (a) Cross-sectional schematics of the device architectures described in this Letter. Structure 1 consists of colloidal QDs sandwiched between two layers of ZnO:SnO2 (ZTO). Structures 2A, 2B, and 2C incorporate at least one layer of ZnS within the ZTO layers, located 15 nm away from the QD film. The ZnS can be positioned in the ZTO above the QDs (structure 2A), below the QDs (structure 2B), or on either side of the QDs (structure 2C). (b) Plot of structure 2A device absorption versus wavelength. The inset photograph shows the device on top of text to demonstrate the transparency of the wide band gap ceramics and thin QD layer. (c) Photograph of device structure 2A operating at 18 V, demonstrating the uniformity of pixel illumination as well as device transparency.

detailed description of device fabrication and material properties is given in the Supporting Information. We demonstrate that these novel, n-i-n architectures are optically transparent and enable electroluminescence (EL) from multicolored QDs. For structure 2A, we measure an absorption of less than 4% at wavelengths above λ ) 500 nm (see Figure 1b), with most of the absorption due to the 50 nm thick QD layer. There is negligible absorption from the ceramic thin films, all of which have band gaps greater than 3 eV. The inset photograph of the completed device provides a visual indication of device transparency. A photograph of a structure 2A device biased at 18 V highlights the uniform EL over the entire pixel area (Figure 1c). Figure 2 shows the energy band diagrams of structure 1 and structure 2A under bias to elucidate the device operating mechanism. The equilibrium energy bands for each material in our device are determined using ultraviolet photoelectron spectroscopy and optical absorption measurements (see Supporting Information). The location of the voltage drop in the device structures is determined through conductivity and capacitance measurements of the metal oxide and QD layers and through atomic force microscopy (AFM) measurements of the QD layer thickness. The band diagrams show that the electron affinity of ZnS is 0.6 ( 0.2 eV lower than that of ZTO, indicating that the ZnS shell on the QDs and the ZnS layers in structures 2A, 2B, and 2C serve as electron blocking © 2010 American Chemical Society

FIGURE 2. Schematic band diagrams of device structures 1 (a) and 2A (b) under forward bias conditions. A larger applied bias is needed to generate the same voltage drop across the QD film in structure 2A as in structure 1.

layers. When a voltage is applied across the device structure, electrons are injected into the n-type ZTO and accumulate 25

DOI: 10.1021/nl902425g | Nano Lett. 2010, 10, 24-29

at the first ZnS barrier they encounter, which can be either a ZnS shell of a QD or a sputtered ZnS layer, depending on the device structure and bias direction (1). This causes most of the voltage applied to the device to be dropped across the QD film (in the case of structure 1) or across both the ZnS layer and the QD film (in the case of structures 2A, 2B, and 2C). If the electric field generated across the QD film is sufficiently large, an electron can be extracted from the valence band of the QD, leaving behind a hole (2), in a process that we refer to as field-induced QD ionization. We demonstrate below that the voltage drop required across each QD to achieve EL is related to the band gap energy of the QD, which is consistent with the QD ionization mechanism. An electron extracted from a neighboring QD (or injected from ZTO) can then couple with this hole to form an exciton on the QD and lead to radiative recombination (3). To achieve steady-state electroluminescence, a sufficiently large electric field must be maintained across the QD film by way of charge accumulation in the device. In the case where the first energy barrier electrons encounter is the QD layer (for example in structure 1, in structure 2A under reverse bias, or in structure 2B under forward bias), the electron accumulation occurs in the QD film as well as in the ZTO layer. Electrons can be injected from the ZTO into QDs causing buildup of space charge in the first layer of QDs that will limit further injection of electrons into the QD film and thereby enhance the accumulation of electrons in the ZTO layer adjacent to the QDs (4). In this case, QD charging aids in the accumulation of electrons that in turn enables a sufficient voltage drop across the QD layer to achieve electroluminescence via QD ionization; however, QD charging will quench luminescence and lead to a decrease in the device current14,15 that is ultimately detrimental to efficient device operation. Figure 2b shows that a ZnS layer between the electron injecting electrode and the QD layer displaces the electron accumulation away from the QD film (4), limiting the amount of QD charging. In Figures 5 and 6, we provide experimental confirmation of this reduced QD charging for devices in which the electrons encounter a sputtered ZnS layer prior to the QD film. As the proposed device operating mechanism suggests, these unipolar devices display low-voltage, forward and reverse bias direct current (dc) as well as alternating current (ac) operation. In ac operation, luminescence is typically observed for applied square wave pulses with root-meansquare voltages (Vrms) of 10-15 V at frequencies up to 50 kHz. Figure 3a presents the typical current density versus voltage (J-V) characteristics. The inset device schematic indicates our biasing convention where “reverse bias” refers to the bottom ITO layer as the source of electrons (i.e., the cathode). Device structures 2A, 2B, and 2C require higher voltages than structure 1. While the sputtered ZnS layer is better at blocking electrons than the QD film, the additional voltage dropped across the insulating ZnS layer necessitates a larger applied bias in the case of structures 2A, 2B, and © 2010 American Chemical Society

FIGURE 3. (a) Current density versus voltage characteristics for device structures 1 (black line), 2A (red line), 2B (purple line), and 2C (blue line) under forward and reverse dc bias conditions. The inset schematic indicates the bias convention. (b) External quantum efficiency (EQE) versus the absolute value of current density for device structures 1 and 2A. The dotted red lines in panels a and b show the current density and EQE measurements for the optimized structure 2A device. (c) Luminance (solid symbols) and luminous efficiency (open symbols) for device structure 1 (black) and structure 2A (red). Data for the optimized structure 2A are shown in red with circle symbols.

2C to achieve QD electroluminescence. As shown in Figure 3b, a peak external quantum efficiency (EQE) of 0.15% is recorded from the glass side of a structure 2A device with additional ZnO buffer layers, as described in the Supporting Information. With a luminance of 1040 Cd/m2, which is 10 times video brightness, recorded at a current density of 92 mA/cm2, the luminous efficiency of these transparent struc26

DOI: 10.1021/nl902425g | Nano Lett. 2010, 10, 24-29

FIGURE 4. (a) Spectra corresponding to photoluminescence (PL) of red and green QDs in a dilute solution (dashed black lines), PL of QDs in the device structure (dashed red and green lines), and electroluminescence (EL) from red and green QDs in device structure 1 at 14 V bias (solid red and green lines). The PL of the concentrated solution of red QDs is shown by a solid black line. (b) Plot of current density (solid line) and EQE (dashed line) as a function of voltage drop per QD. We observe a 0.3 V difference between the voltage drop per QD required to observe EL in a device containing red QDs and a device containing green QDs. Part a shows that the difference in band gap energy between these red and green QDs is 0.3 eV, suggesting that QD ionization is the mechanism governing device operation.

tures is 1.1 Cd/A. (Note that a similar amount of light is emitted through the top ITO surface, corresponding to an overall external luminous efficiency of over 2 Cd/A.) The EL spectra for green and red QDs in structure 1 at 14 V demonstrate that a unipolar, n-i-n structure can be used to obtain multiple color emission from the same device structure (Figure 4a) where the EL is due entirely to QD emission. The red-shifting and broadening of the EL spectra in comparison to the dilute QD solution PL spectra (dashed black lines) can be explained by well-characterized interactions in QD ensembles.16-18 We note that the PL spectrum of the concentrated solution of red QDs used for the QD layer deposition (solid black line) matches the device PL spectrum (dashed red line). This suggests that QD proximity, through effects such as energy transfer16 and dielectric dispersion,17 can explain the red shift and some of the broadening. At 14 V applied bias, an increase in the full width half-maximum (fwhm) between the device PL and EL spectra of 7 nm for the red device and 8 nm for the green device is observed. The red shift in peak position of the PL spectra as a function of applied voltage across the device displays a quadratic dependence for all measured structures (1, 2A, and 2C). This red shift in peak position as a function of applied voltage is a reflection of the quantum confined Stark effect in QD ensembles and the spectral broadening as a function of applied voltage is consistent with the increased LO (longitudinal optical)-phonon coupling expected with larger exciton polarization in the presence of an electric field.18 The hypothesis of field-induced QD ionization is also supported by our observation of red and green QD luminescence from the same device structure. We expect the QD ionization process to be dependent on the band gap energy © 2010 American Chemical Society

of the QD. Figure 4b relates the electroluminescent (EL) turn on to the voltage drop across each layer of QDs. We find that EL turn on in a device containing green QDs requires 0.3 ( 0.1 V more per QD than in a device containing red QDs. Figure 4a shows that these green-emitting QDs have a band

FIGURE 5. Plot of normalized photoluminescence (PL) as a function of applied voltage in forward bias for device structures 1 (squares), 2A (upward pointing triangles), 2B (downward pointing triangles), and 2C (diamonds). In the case of device structures 2A and 2C in forward bias, electrons injected into the ZTO encounter a ZnS layer prior to the QD film. In contrast, electrons in the ZTO in structures 1 and 2B do not encounter a ZnS layer and accumulate at the QD film. Consequently device structures 2A and 2C show less PL quenching with increasing voltage than structures 1 and 2B, indicating that accumulation of electrons at the ZnS layer instead of at the QD film reduces charging of the QD layer. The trend in PL quenching of device structures 1 and 2B are the same but offset in voltage because of the presence of the insulating ZnS layer in structure 2B. This plot illustrates that the extent of electron accumulation at the QD film, not the electric field across the QDs, determines the amount of QD charging and luminescence quenching. 27

DOI: 10.1021/nl902425g | Nano Lett. 2010, 10, 24-29

FIGURE 6. (a) Plot of luminosity (top, left axis) and voltage (bottom, right axis) as a function of time for device structure 2A operated in forward bias at a constant current of 30 mA/cm2. The device was unpackaged and operated in air. Voltage and luminescence data were recorded every 5 s for 20 consecutive hours and then binned and averaged into 1 min intervals. (b) Scatter plot of luminosity versus voltage using all of the data in panel a shows that despite fluctuations in the voltage and luminosity over time, the luminosity at a given voltage is constant, indicating that these fluctuations do not correspond to changes in the device performance with time. For example, the voltage and luminosity at 1 h 50 min (R) and 10 h (β) are comparable.

gap of approximately 0.3 eV more than the red-emitting QDs. Comparing Figures 4a and 4b shows that the magnitudes of the electroluminescent turn-on voltages for the red and green QD devices are comparable to the bandgap energies of the red and green QDs. These two observations suggest that the band gap energy of the QDs determines the voltage drop across each QD needed extract an electron (i.e., ionize the QD), generate free charge, and thereby achieve EL. The addition of at least one ZnS layer in the device structure improves device performance by limiting QD charging. To demonstrate this, we measure the photoluminescence (PL) quenching in the QD layer due to an applied voltage across the device structures. This measurement provides an indication of the number of Auger nonradiative recombination and exciton dissociation events due to QD charging or due to the presence of an electric field across the QD layer. The data presented in Figure 3 indicate that a certain minimum electric field across the QDs is required for luminescence. The addition of the 30 nm of ZnS in device structures 2A, 2B, and 2C shifts the voltage at which this electric field is established by 4 V. However, Figure 3b shows that once this field is obtained, a device with or without ZnS exhibits similar external quantum efficiencies. If we assume that above the turn on, device structures 1, 2A, 2B, and 2C all exhibit similar degrees of exciton dissociation, then the differences observed in PL quenching between the structures can be attributed to charging of the QD layer. Figure 5 shows that device structures 2A and 2B, which are identical except for the placement of the ZnS either above or below the QD layer, exhibit different PL quenching trends at the same voltage. When structure 2A is under forward bias, the electrons are injected from the top contact and are blocked by a ZnS layer barrier prior to reaching the QD film. Consequently, the electric field needed for electroluminescence is achieved with relatively little charging of the QDs and the PL from the QD film shows little quenching as a © 2010 American Chemical Society

function of voltage. In the case of structure 2B in forward bias, the electrons arrive first at the QD film. Accordingly, structure 2B exhibits the same charging properties and trend in PL quenching as structure 1 but offset in voltage (by 4 V) due to the presence of the insulating ZnS. The slightly larger amount of PL quenching in structure 2C compared to structure 2A can be explained by the fact that, in forward bias, electrons encounter a 15 nm thick ZnS layer following injection in structure 2C but a 30 nm thick layer in structure 2A. The thinner ZnS layer in structure 2C is a weaker electron blocking layer that allows for more QD charging. By using ZnS layers, we are thus able to validate our understanding of the device operating mechanism and controllably reduce QD charging to improve device performance. Finally, we demonstrate stable device operation in air with unpackaged devices, enabled by the environmental stability of the ZTO layers19 and the device operating mechanism based on QD ionization. All device measurements presented here were performed in ambient laboratory conditions. Repeated J-V and EQE measurements recorded every week over the course of 40 days on devices stored in air showed stable luminescent response. This long shelf life suggests the possibility for extended operation in ambient conditions. An understanding of the device operating mechanism enables us to choose favorable bias conditions for extended testing. A device held under constant dc bias exhibits gradual charging of the QD layer, which is manifested by a corresponding decrease in device current over time.14,15 Device structure 1 will, in fact, no longer exhibit EL after approximately 5 min at constant applied bias due to a lack of sufficient current through the device. Structure 2A, in which the QD charging occurs at a slower rate due to the presence of the ZnS layer, will exhibit EL for approximately 45 min when forward biased in a constant voltage mode. In order to circumvent eventual device turnoff with a constant applied voltage, we therefore perform extended lifetime testing on structure 2A run in a constant 28

DOI: 10.1021/nl902425g | Nano Lett. 2010, 10, 24-29

Supporting Information Available. Descriptions of device fabrication, optimized device structure, the band structure, and device characterization and a figure of AFM images and spectra showin QD photoluminescence. This material is available free of charge via the Internet at http://pubs.acs. org.

current mode, with voltage and luminescence measured in 5 s intervals. We choose to test a device with an average EQE to ensure that our results are representative. Figure 6a shows luminescence as a function of time during 20 h of continuous device operation. The overlap onto a single curve of the 14400 luminosity and voltage data points from panel a (Figure 6b) shows that for a given voltage and current, at any point during the 20 h of testing, the device exhibits the same luminosity. This indicates that an unpackaged device does not degrade over 20 h of continuous operation in air. In summary, we developed a unipolar, n-i-n lightemitting device architecture that operates via a new method of QD electrical excitation: field-induced QD ionization. The device structure facilitates the use of stable transport materials and reduces the need for complicated band alignment engineering. We demonstrated that introduction of an appropriate thin film of ceramic material (in this case, ZnS) into the charge transport layer to create a potential barrier for majority carriers limits charging of the QD layer. This enables the first QD-LED with constant luminance over extended operation. These unique unipolar devices can yield multicolor QD emission in transparent device structures that exhibit stable operation in air, without packaging.

REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

Acknowledgment. The authors acknowledge Lisa Marshall and Dr. Polina Anikeeva for their assistance and Professor John Wager for helpful discussions. We thank QD Vision, Inc., for supplying some of the QDs used in this work. This work is supported by the Institute for Soldier Nanotechnologies (DAAD-19-02-0002), a Presidential Early Career Award for Scientists and Engineers (V. Bulovic´), and a National Defense Science and Engineering Graduate Fellowship (V. Wood). This work made use of MRSEC Shared Experimental Facilities at MIT, supported by the National Science Foundation under award number DMR-02-13282.

© 2010 American Chemical Society

(15) (16) (17) (18) (19) (20) (21) (22) (23)

29

Coe, S.; Woo, W. K.; Bawendi, M. G.; Bulovic´, V. Nature 2002, 420, 800. Steckel, J. S.; et al. Angew. Chem., Int. Ed. 2006, 45, 5796. Kim, L.; Anikeeva, P. O.; Coe-Sullivan, S. A.; Steckel, J. S.; Bawendi, M. G.; Bulovic´, V. Nano Lett. 2008, 8, 4513. Anikeeva, P. O.; Halpert, J. E.; Bawendi, M. G.; Bulovic´, V. Nano Lett. 2009, 9. Anikeeva, P. O.; Madigan, C. F.; Halpert, J. E.; Bawendi, M. G.; Bulovic´, V. Phys. Rev. B 2008, 78, No. 085434. Anikeeva, P. O.; Halpert, J. E.; Bawendi, M. G.; Bulovic´, V. Nano Lett. 2007, 7, 2196. Mueller, A. H.; et al. Nano Lett. 2005, 5, 1039. Caruge, J.-M.; Halpert, J. E.; Wood, V.; Bawendi, M. G.; Bulovic´, V. Nat. Photonics 2008, 2, 247. Wood, V.; Panzer, M. J.; Halpert, J. E.; Caruge, J.-M.; Bawendi, M. G.; Bulovic´, V. ACS Nano 2009, 3, 3581. Colvin, V. L.; Schlamp, M. C.; Alivisatos, A. P. Nature 1994, 370, 354. Zhao, J.; et al. Nano Lett. 2006, 6, 463. Stouwdam, J. W.; Janssen, R. A. J. J. Mater. Chem. 2008, 18, 1889. Caruge, J.-M.; Halpert, J. E.; Bulovic´, V.; Bawendi, M. G. Nano Lett. 2006, 6, 2991. Drndic´, M.; Jarosz, M. V.; Morgan, N. Y.; Kastner, M. A.; Bawendi, M. G. J. Appl. Phys. 2002, 92, 7498. Morgan, N. Y.; Leatherdale, C. A.; Drndic´, M.; Vitasovic, M.; Kastner, M. A.; Bawendi, M. G. Phys. Rev. B 2002, 66, No. 075339. Kagan, C. R.; Murray, C. B.; Bawendi, M. G. Phys. Rev. B 1996, 54, 8633. Leatherdale, C. A.; Bawendi, M. G. Phys. Rev. B 2001, 63, 165315. Empedocles, S. A.; Bawendi, M. G. Science 1997, 278, 2114. Go¨rrn, P.; et al. Appl. Phys. Lett. 2007, 9, No. 063502. Dabbousi, B. O.; et al. J. Phys. Chem. B 1997, 101, 9463. Ivanov, S. A.; et al. J. Phys. Chem. 2004, 108, 10625. Colvin, V. L.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem. Soc. 1992, 114, 5221. Blackstock, J. J.; Li, Z.; Freeman, M. R.; Stewart, D. R. Surf. Sci. 2003, 546, 87.

DOI: 10.1021/nl902425g | Nano Lett. 2010, 10, 24-29