Efficiency in QD-LEDs

Efficient electron and hole injection, balance of charge carriers arriving at the QD active layer, and minimization of the electric field across the QDs are all important design criteria for ensuring high-performance QD-LEDs.^{Reference Mashford, Stevenson, Popovic, Hamilton, Zhou, Breen, Steckel, Bulovic, Bawendi, Coe-Sullivan and Kazlas2,Reference Anikeeva, Madigan, Halpert, Bawendi and Bulović5,Reference Wood and Bulović6} However, these design guidelines are highly device specific and difficult to achieve in the same device for different color emitters with various chemistries and sizes.^{Reference Anikeeva, Halpert, Bawendi and Bulović3,Reference Wood and Bulović6}

Given the extensive discussion of optimization in device structure in prior literature, in this article, we do not consider the challenges of bringing charge carriers to the QD layer and forming excitons on the QDs; rather, we examine efficiency in the last step of the light generation process in a QD-LED. Namely, when an exciton is present on the QD, what is the probability that it will recombine to emit a photon, which can be quantified by the luminescent quantum yield (QY). The device EQE can thus be assumed to be proportional to the QY of the emitters in the device structure, which depends on the exciton nonradiative (k _nr) and radiative (k _r) recombination rates:

(1)$${\rm{EQE}}\,\alpha \,{\rm{QY}} = k_r /\left( {k_{nr} + k_r } \right).$$

As illustrated schematically in Figure 1, k _nr and k _r are determined by the QDs themselves and the interaction of the QDs with the electric field (F) and charge (Q) resulting from the voltage and current needed to operate the LED.

Figure 1.

Schematic of a standard quantum dot light-emitting diode (QD-LED) architecture featuring QDs sandwiched between an electron transport layer (ETL) and a hole transport layer (HTL). In addition to the presence of defect or surface states on QDs, which can result in an increase in the trap-assisted recombination rate (k _t), the voltage and current required for LED operation results in an electric field on the QDs and charge, that, in turn, affects the radiative (k _r) and non-radiative (k _nr) recombination rates for excitons generated in the QD-LED.

In a QD-LED, the two major contributors to the non-radiative rate are electronic trap states and free-charge carriers. For example, if the QD emitter has surface state defects, trap-assisted recombination can occur, whereby the electron or hole in an exciton relaxes to a trap state, and the two carriers subsequently recombine without emission of a photon.^{Reference Kuno, Lee, Dabbousi, Mikulec and Bawendi7} This results in an increase in k _nr, which, as seen in Equation 1, decreases the QY. Alternatively, the presence of charge on a QD, resulting from an improper balance of electron and hole injection, for example, can increase the Auger non-radiative recombination, where energy is dissipated as kinetic energy to a charge carrier instead of as a photon.^{Reference Klimov8,Reference Galland, Ghosh, Steinbrück, Sykora, Hollingsworth, Klimov and Htoon9} Again, this increases k _nr and decreases the QY.

In most QD-LEDs, the QD active layer also experiences an electric field on the order of 1 MV/cm. It has recently been shown, using a combination of experiment and theory, that in a QD, while k _nr is not affected by the electric field in the device, the electric field can induce a spatial separation of the electron and hole wave functions that is sufficient to significantly reduce the radiative rate of the exciton.^{Reference Bozyigit, Yarema and Wood10} In the limit where k _r is smaller or on the order of k _nr (k _r ≤ k _nr), Equation 1 shows that a reduction of the radiative rate can cause a significant reduction in the EQE of the QD-LED.

Designing QDs for LED applications

Over the past 20 years, excellent progress has been made in the development of QD chemistries to improve the QY of the QD emitters. These chemistry advances can be broadly categorized into two trends: overcoating of the QD core with a shell material and grading the QD core composition, which is referred to as alloying.^{Reference Talapin, Lee, Kovalenko and Shevchenko11} As shown schematically in Figure 2, these modifications to the QD can result in a change in the degree of confinement of an exciton, in the extent of surface passivation, or a combination of these two effects. We discuss how these changes in QD chemistry have led to improvement of the QYs through a reduction of the non-radiative processes.

Figure 2.

Schematic summarizing the positive (green check) or negative (red cross) impact a specific quantum dot (QD) structure has on the trap-assisted (k _t) or Auger (k _a) non-radiative recombination rate. QD structures are organized according to their extent of surface passivation and electronic confinement with respect to a CdSe core located in the upper left. Each QD structure is illustrated along with its energy band diagram showing the wave functions of the electron (blue) and hole (red).

Starting with a CdSe core (upper left of Figure 2), the addition of a shell passivates the surface of a QD core and offers physical separation of the exciton from defect states on the surface of the QD (right side of Figure 2). This results in a decrease in the non-radiative trap-assisted recombination rate (k _t), thereby improving the luminescence QY of the QD.^{Reference Hines and Guyot-Sionnest12–Reference Chen, Zhao, Chauhan, Cui, Wong, Harris, Wei, Han, Fukumura, Jain and Bawendi14} This is highlighted by the recently developed “giant” shell CdSe/CdS QDs, where CdSe cores are overcoated with CdS such that the thickness of the shell is more than double the core diameter.^{Reference Mahler, Spinicelli, Buil, Quelin, Hermier and Dubertret15,Reference Chen, Vela, Htoon, Casson, Werder, Bussian, Klimov and Hollingsworth16} These QDs are reported to sustain a high degree of thermal stress and maintain their luminescence even when the passivating ligands are removed.

While the addition of a shell tends to consistently improve passivation of the QD, depending on the energy levels of the shell material relative to those of the core, the shell can either increase or decrease confinement of the exciton. For example, in a CdSe/ZnS QD, where a ZnS shell is added to a CdSe core as shown in the upper right of Figure 2, the wave functions of the electron (blue shading) and hole (red shading) that make up the exciton are more strongly spatially confined in the CdSe core due to the large energy offsets between the valence and conduction bands of the CdSe and ZnS. This increase in exciton confinement can be easily observed as shift of the emission in the QD to higher energies, as expected from a quantum mechanical particle-in-a-box picture. Specifically, a red-emitting CdSe QD that is overcoated with ZnS shell will emit in the green. This spatial confinement promotes strong carrier-carrier interactions, which results in the QDs retaining the high Auger non-radiative recombination rate (k _a) of the core.^{Reference Klimov8}

In contrast, when a CdSe QD is overcoated with CdS (lower right of Figure 2), the hole wave function remains confined on the CdSe core, while the electron wave function extends into the shell. This decrease in the electron confinement, which reduces carrier-carrier interactions, has been shown experimentally to decrease the Auger non-radiative recombination rate (k _a).^{Reference García-Santamaría, Chen, Vela, Schaller, Hollingsworth and Klimov17,Reference García-Santamaría, Brovelli, Viswanatha, Hollingsworth, Htoon, Crooker and Klimov18}

QDs with alloyed composition, such as ZnCdS or ZnCdSe, are also understood to have low k _a.^{Reference Wang, Ren, Kahen, Hahn, Rajeswaran, Maccagnano-Zacher, Silcox, Cragg, Efros and Krauss19} This has been explained theoretically by a smoothing of the shape of the confinement potential (see lower left of Figure 2) that is believed to occur in these alloyed QDs.^{Reference Cragg and Efros20} Alloying is also thought to be present at the core–shell interface in the “giant” shell QDs, which, in addition to the reduced electron confinement, can explain the observation of extremely low k _a in these QDs.^{Reference Cragg and Efros20}

Based on these considerations, it would seem that “giant” shell QDs, which offer the optimal passivation and reduction of Auger non-radiative recombination, would be the ideal choice for an emitter in a QD-LED. However, record efficiencies in QD-LEDs have not been reported with “giant” shell CdSe/CdS QD materials.^{Reference Pal, Ghosh, Brovelli, Laocharoensuk, Klimov, Hollingsworth and Htoon21} Instead, QD-LEDs with multilayered alloyed structures such as ZnCdS/ZnS, CdZnSe/CdZnS, or ZnCdSe exhibit the highest EQEs.^{Reference Anikeeva, Halpert, Bawendi and Bulović3,Reference Steckel, Snee, Coe-Sullivan, Zimmer, Halpert, Anikeeva, Kim, Bulovi and Bawendi22,Reference Caruge, Halpert, Wood, Bulović and Bawendi23} This can be understood by considering the impact of the electric field on the QY on the different types of QDs discussed previously.

As shown in Figure 3a, increasing the electric field across CdSe/CdS QDs decreases the luminescence QY.^{Reference Bozyigit, Yarema and Wood10} The extent of the decrease, which can exceed a factor of 10 for electric fields encountered in a QD-LED, is dictated by the thickness of the CdS shell. Significantly less quenching is observed for a ZnS shell; however, it is still present and important for QD-LED efficiency.^{Reference Bozyigit, Wood, Shirasaki and Bulovi24} Indeed, a recent study has shown that this field-induced luminescence quenching is responsible for the efficiency droop seen in EQE versus current density plots for many QD-LEDs.^{Reference Shirasaki, Supran, Tisdale and Bulović25}

Figure 3.

(a) Experimental measurement of the photoluminescent quantum yield (QY) and (b) tight binding calculations of the optical matrix element as a function of the electric field experienced by the quantum dot (QD). (c) Schematics of electron and hole wave functions in CdSe/ZnS and CdSe/CdS QDs with and without the influence of an electric field showing that in the presence of an electric field, relaxed confinement is not desirable, as it causes a reduction in the radiative rate due to exciton polarization. Adapted with permission from Reference 10.

While electric field-induced luminescence quenching has been observed for QDs and other nanocrystal shapes^{Reference Kraus, Lagoudakis, Rogach, Talapin, Weller, Lupton and Feldmann26} at both the single QD^27,28 and ensemble levels,^{Reference Jarosz, Porter, Fisher, Kastner and Bawendi29} until recently,^{Reference Bozyigit, Yarema and Wood10,Reference Shirasaki, Supran, Tisdale and Bulović25} no clear consensus was reached on its origins. As with Auger non-radiative recombination, the origins of field-driven luminescence quenching relate to QD band structure. Time-dependent photoluminescence measurements of the QDs exposed to varying electric fields reveal the exciton lifetime to be invariant with the field. This implies that the electric field does not change the non-radiative rate. Calculations of the optical matrix element (Figure 3b), which is proportional to the radiative rate (k _r) and reflects the spatial overlap in the electron and hole wave functions, confirm that the decrease in luminescence QY can be assigned to a decrease in the radiative rate.^{Reference Bozyigit, Yarema and Wood10} As illustrated schematically in Figure 3c, selecting a QD chemistry with decreased electronic confinement (e.g., a CdS shell instead of a ZnS shell for a CdSe core QD) allows for spatial separation of the electron and hole wave functions (blue and red shading) when the QD is exposed to an electric field. This effect can be referred to as exciton polarization and reduces the radiative recombination rate (k_r).

The previous discussion highlights the challenge in developing a QD for use as an emitter in a QD-LED, where the QDs are subject to both charge carriers and electric fields. The trade-off associated with choosing the optimal degree of exciton confinement that emerges from this discussion is depicted in Figure 4. A QD with minimal electronic confinement (left side of Figure 4) will likely not suffer from reduced QY even as charge accumulates on it during LED operation. However, this QD will experience increased electric field-induced luminescence quenching. Conversely, while a QD with a large confinement potential (right side of Figure 4) will not experience a significant loss in QY when exposed to an electric field, charging of the QD will decrease the QY and pose a significant challenge for high-performance QD-LED operation.

Figure 4.

Schematic illustrates the challenge in selecting a quantum dot (QD) emitter for a light-emitting diode (LED) application. QDs with reduced electronic confinement have low Auger non-radiative recombination, but also exhibit reduced radiative recombination in the presence of an electric field. Alternatively, a QD with strong electronic confinement shows high Auger non-radiative recombination and low field-induced luminescence quenching.

QD luminescence from localized states

One strategy to improve upon QD emitters is to minimize field-induced luminescence quenching in QDs already exhibiting low Auger non-radiative recombination rates and high photoluminescent QYs. One scheme for this could be the rapid transfer of an exciton formed in an alloyed or thick-shelled QD to a localized state such that exciton polarization in the presence of an electric field is reduced. Such a scenario could be achieved through the introduction of a luminescent impurity such as Mn⁺ into an alloyed QD (See Figure 5a) or through selection of a QD, such as CuInS₂ or CuInSe₂, where luminescence occurs through a donor-acceptor process.^{Reference Thakar, Chen and Snee30,Reference Aldakov, Lefrançois and Reiss31}

Figure 5.

Two possible strategies to avoid luminescence quenching due to an electric field are shown. (a) A dopant can offer a luminescent state that may be less sensitive to high electric fields. One type of common dopant in quantum dots (QDs) is a Mn⁺ impurity that emits in the red (solid line) even if the host QD has a large bandgap, as evidenced here by the absorption onset at 450 nm (dashed line).^{Reference Thakar, Chen and Snee30} (b) Embedding QDs in a host matrix with high dielectric constant (ε_host) can reduce the effective field experienced by the exciton (F_ex). For example, an emissive layer consisting of a TiO₂ host with a 20% QD volume fraction reduces the effective field experienced by the QDs by over a factor of 10, as compared to a neat QD film.

QD active layer placement and engineering

Another approach to circumvent the limitations to EQE due to electric field-induced luminescence quenching is to minimize exposure of the QDs to high electric fields. As a first step, this implies that QDs should be placed away from material interfaces, where electric fields are typically very high (∼1 MV/cm). It has been shown that this strategy can significantly improve QD-LED EQE efficiencies.^{Reference Anikeeva, Madigan, Halpert, Bawendi and Bulović5}

A different way to protect QDs from high electric fields is through their encapsulation in a high-k host material, which could be carried out, for example, using chemical bath deposition^{Reference Kinder, Moroz, Diederich, Johnson, Kirsanova, Nemchinov, O’Connor, Roth and Zamkov32} or atomic layer deposition in-filling of the QD active layer.^{Reference Liu, Gibbs, Perkins, Tolentino, Zarghami, Bustamante and Law33,Reference Pourret, Guyot-Sionnest and Elam34} Using Maxwell–Garnett and Clausius–Garnett theory, we estimate how placement of the QDs in a host material with given dielectric constant, ε_host, reduces the field (F_ex) experienced by excitons on the QDs. To compare QD active layers with different ε_host, we assume that the electric field in the QD-LED transport layers adjacent to the QD-host remains constant. This is equivalent to assuming a fixed displacement charge, which for normal QD-LED operation is roughly on the order of 0.2 μC/cm², assuming 10 V bias over a 100-nm-thick device with a relative dielectric constant of 2.2. As shown in Figure 5b, we find that for low ε_host, the field experienced by the exciton (F_ex) is hardly influenced by the exact host dielectric constant or the volume fraction of the QDs. However, a high dielectric constant material such as TiO₂ (ε ≥ 86) can reduce the effective field by a factor of 10 for a QD volume fraction of 20%, which demonstrates the possibility for a significant alleviation of field-induced luminescence quenching.

Novel device architectures and mechanisms for QD electroluminescence

Another option to reduce the challenges associated with designing a QD that exhibits high QY in the presence of both charge and electric field is to consider device architectures that are not pn-junctions and that achieve QD-electroluminescence in new ways. An example of such a device architecture, which highlights the possibility for a paradigm shift away from charge injection into QDs as a means for electroluminescence is an electric field-driven QD-LED.^{Reference Wood and Bulović35}

Figure 6 shows a typical field-driven device structure where the emissive QD-containing layer is sandwiched between two insulating, wide bandgap oxide layers such as Al₂O₃ or SiO₂.^{Reference Kobayashi, Tani and Kawazoe36,Reference Wood, Panzer, Bozyigit, Shirasaki, Rousseau, Geyer, Bawendi and Bulović37} The device structure is completed with two electrodes. When a voltage is applied across the device, no charge is injected from the contacts into the QD layer; however, electroluminescence is still observed. As depicted in Figure 6, when the voltage applied across each QD exceeds its bandgap energy, an electron can transfer from the valence band of one QD to the conduction band of a neighboring QD (i.e., QD ionization), creating a spatially separated electron and hole in the QD film that can subsequently radiatively recombine.^{Reference Wood, Panzer, Bozyigit, Shirasaki, Rousseau, Geyer, Bawendi and Bulović37} Application of a pulsed applied electric field enables sustained electroluminescence in such a device. In addition to high-voltage pulsed operation of these capacitive devices, it is also possible to achieve field-driven electroluminescence in relatively low-voltage, constant current-driven devices by incorporation of thin (∼15 nm thick) electron blocking layers into the device, which permit sufficient buildup of an electric field to allow for the QD ionization process described previously.^{Reference Wood, Panzer, Caruge, Halpert, Bawendi and Bulović38}

Figure 6.

Schematics showing field-driven quantum dot light-emitting diode (QD-LED) device architecture and operating mechanism. Photographs of devices during operation and electroluminescence (EL) spectra (solid lines) show that different QDs can be excited within the same device structure. Comparison of the EL spectra to the photoluminescence spectra (dashed lines) of the QDs when the device is not in operation highlights that EL occurs at low electric field conditions. Adapted with permission from Reference 38. © 2011 American Chemical Society.

To study the mechanism of field-driven electroluminescence in detail, time-resolved luminescence studies were carried out while charge and electric field in the QD film were simultaneously measured.^{Reference Bozyigit, Wood, Shirasaki and Bulovi24,Reference Wood, Panzer, Bozyigit, Shirasaki, Rousseau, Geyer, Bawendi and Bulović37} It was found that following the ionization step, the applied electric field transports electrons and holes away from each other toward opposite QD/oxide interfaces. This redistribution of charges creates an internal electric field that screens the external applied field. When the external applied electric field is removed, the internal field, which is present due to the spatial separation of electron and hole populations, causes these electrons and holes to drift toward each other and recombine.

Therefore, despite the high electric fields on the order of 5 MV/cm needed to generate the free charge, luminescence in field-driven QD-LEDs occurs under lower electric field conditions (∼1 MV/cm). This can be readily observed in Figure 6 from the lack of a pronounced red shift between the QD photoluminescence and electroluminescence spectra, which one would expect for QDs in high electric fields due to the Stark effect.^{Reference Empedocles27} Furthermore, because electric field-driven luminescence is inherently a local process, the emissive layer need not be a continuous QD film, but could consist of clusters of QDs embedded within an insulating matrix. Indeed, field-driven electroluminescence has been demonstrated for high QY, QD-insulating polymer blends, which had previously been restricted to applications involving optical excitation of colloidal QDs.^{Reference Wood, Panzer, Bozyigit, Shirasaki, Rousseau, Geyer, Bawendi and Bulović37}

The field-driven QD-LED alleviates the band alignment considerations that typically dictate which emissive materials can be electrically excited using particular charge transport layers. As highlighted by the photographs and spectra in Figure 6, luminescent materials that have different chemistries and absolute energy level positions, and whose peak emission wavelengths span the visible to near-infrared regions, can all be excited within the same device structure. The first report of electroluminescence from a Mn⁺-doped QD material was also achieved in a field-driven QD-LED.^{Reference Wood, Halpert, Panzer, Bawendi and Bulović39}

While field-driven devices remove some of the design constraints associated with traditional QD-LEDs and emphasize the opportunity for new types of QD-LEDs, significant optimization of these devices is still needed. Engineering of the QD band structure and the emissive layer to minimize the electric fields needed for ionization and control of the charge distributions within the active layer following ionization will be critical in achieving high EQEs in field-driven QD-LEDs.