1.1.

Thermally Activated Delayed Fluorescence as an Emerging Trend in Optoelectronics

The transformation of electrical energy into visible light has remained an important issue for scientists and engineers, even more than a century after the development of the incandescent lightbulb by Swan, Edison, and others.^1,2 One of the new emerging topics in this niche of material sciences is emitting materials that feature thermally activated delayed fluorescence (TADF). Although TADF has been known since 1960,³ the potential use of TADF emitters in optoelectronic devices has been proposed independently by several groups led by Adachi et al.^4–6 (organic and tin emitters), Deaton et al.,⁷ Yersin et al.,⁸ and others.^9–12 The rise of the TADF-topic is reflected by an exploding number of new peer-reviewed publications dealing with it in general (Fig. 1), peaking close to 200 in 2015. Note that Refs. 4–8, which propose and show the use of such materials in organic light-emitting diodes (OLEDs) for the first time, have been published between 2009 and 2012, suggesting that the rise of this concept was directly related to a potential (industrial) application in OLEDs. Occasionally, synonyms such as “singlet harvesting”^13–16 or “E-type fluorescence”^7,17,18 are also being used in the literature.

Fig. 1

TADF-related publications per year according to Scifinder as of February 2016.

The aim of this review article is to briefly introduce this concept and outline its impact on photonics, especially with respect to its potential concerning the realization of energy-efficient solid-state lighting and display devices. The scope of this article does not include any in-depth discussion of the molecular structure of TADF molecules or device architectures, which can be found elsewhere.^19–22

1.2.

What Is Thermally Activated Delayed Fluorescence?

Upon recombination of charge carriers in any electroluminescent device, two different types of excitons are formed: singlet excitons having a spin $S=0$ and triplet excitons with $S=1$.²³ Due to quantum statistics, the relative probability for the formation of singlet and triplet excitons is 25% to 75%.^13,24,25 In order to realize efficient OLEDs, it is crucial to harvest both singlet and triplet excitons for the generation of photons. Currently, there are three main emissive mechanisms available to harvest excitons: fluorescence, phosphorescence as well as TADF. A schematic overview of these mechanisms is shown in Fig. 2.

Fig. 2

Schematic comparison between fluorescence, phosphorescence, and TADF. Straight arrows represent emissive transitions, while dotted arrows indicate up- or down-conversion processes (ISC and rISC).

On a side note, there are also more complex schemes, which usually require at least a bimolecular, often trimolecular emitting mechanism (e.g., multiple species between which energy can be transferred) such as $P$-type delayed fluorescence or triplet–triplet annihilation,^26–29 hyperfluorescence,³⁰ exciplex emission,⁶ and exciplex-assisted emission.^31–34 These mechanisms will not be covered in this article.

Fluorescent molecules have a relatively large energetic difference between the $S_1$ and $T_1$ state. This difference is also known as the singlet–triplet splitting or $\mathrm{\Delta }E$ ($S–T$).¹³ In addition, they usually show very weak intersystem crossing (ISC) and reverse intersystem crossing (rISC), meaning that the transformation of singlet ($S_1$) into triplet excitons ($T_1$) and vice versa is quantum-mechanically forbidden, therefore both slow and unlikely.²³ The same is true for any transitions between $T_1$ and the singlet ground state $S_0$.^35,36 Because of these limitations, triplet excitons cannot be harvested with fluorescence, limiting the overall efficiency of such OLEDs to 25%.¹³ An archetypical fluorescent OLED emitter is aluminum-tris-(8-hydroxyquinoline) (${\mathrm{Alq}}_3$; Fig. 3).^37–40

Fig. 3

Molecular structure of ${\mathrm{Alq}}_3$, ${\mathrm{Irppy}}_3$, and 4CzIPN.

Phosphorescent molecules⁴¹ share a rather larger $\mathrm{\Delta }E$ ($S–T$) value with their fluorescent counterparts. On the other hand, ISC is very strong, leading to a prompt transformation of any singlet into triplet excitons. Because of the ISC, which is the result of the presence of heavy transition metals such as iridium, platinum, and osmium, the transition between $T_1$ and $S_0$ is possible.^35,36 Connected to this is the term triplet harvesting, because all excitons are harvested via the triplet state $T_1$.¹³ With this, it was first possible to harvest all generated excitons, which resulted in highly efficient OLEDs.^42,43 An archetypical phosphorescent OLED emitter is iridium-tris-(2-phenylpyridine) (${\mathrm{Irppy}}_3$; Fig. 3).^44–45

TADF molecules have a low $\mathrm{\Delta }E$ ($S–T$), typically in the order of several hundreds of meV, and mediocre ISC and rISC.²¹ Although direct transitions between $T_1$ and $S_0$ are still forbidden, it is possible to go from $T_1$ back to $S_1$ by means of rISC, if the thermal energy $k_{\mathrm{b}}T$ is sufficient. Compared with normal fluorescence, where the excited-state lifetime is in the order of ns, the excited-state lifetime is longer, often in the order of $\mu \mathrm{s}$ or more—the term TADF accounts for these observations. Occasionally, the term singlet harvesting^14–16,46 is used, because all excitons are harvested via the singlet state $S_1$.¹³ A typical TADF emitter is tetrakis-$N$-carbazoyl-isophthalonitrile (4CzIPN; Fig. 3).^5,47–49

In summary, TADF is an alternate concept, which can help to realize efficient OLEDs with maximum efficiency. It competes with phosphorescence, the commercial state-of-the-art solution for efficient OLEDs.

2.1.

Molecular Design Principles

Without going too much into detail, all TADF emitters consist of donor and acceptor moieties. In a way, they are similar to so-called ambipolar host molecules and often even contain similar building blocks, which are either arranged in a different way or used in different combinations in order to realize the large bandgap, which is needed to accommodate dopant molecules.^50–55 In both hosts and TADF emitters, donor moieties are electron-rich functional units that have a rather deep HOMO, which allow them to be oxidized easily and partake in the transport of holes. Contrarily, acceptor moieties are electron-deficient functional units with a deep LUMO. They can be reduced and thus partake in the transport of electrons. Figure 4 contains the molecular structure of various donor units, while acceptor units are shown in Fig. 5.

Fig. 4

Schematic structures of various donor units.

Fig. 5

Schematic structures of various acceptor units.

It is apparent that—as of today—there is much more variation found on the acceptor part than on the donor part of TADF emitters.¹⁹ As indicated in Fig. 4, the vast majority of all published organic TADF emitters contain substituted carbazole-^56–59 or arylamine-type donors.^60–62 In most cases, the connection to the acceptors is realized via the aromatic nitrogen. On the acceptor-side (Fig. 5), one can find a plethora of nitrogen-containing heterocycles,^63,64 as well as various benzonitriles.^{47–49,56,57,59,65,66} A third acceptor-class is sulfones,^{31,54,58,60,67,68} which seems to be particularly useful to realize deep-blue TADF emission.

The color of the TADF emitter depends on the choice of the donor and acceptor moieties. Approximately, the choice of the donor dictates the HOMO, while the acceptor dictates the LUMO. The energetic difference between HOMO and LUMO is the bandgap, which is closely connected to the emission color of the emitter. Roughly, a bandgap of ca. 2.8 eV (450 nm) relates to blue emission, while 1.9 eV (650 nm) gives red emission.²³ In Sec. 1, $\mathrm{\Delta }E$ ($S–T$) was introduced as the most crucial parameter for TADF. This parameter is highly dependent of the localization of the Frontier orbitals HOMO and LUMO.^69,70 The better the separation (or the smaller the overlap between HOMO and LUMO), the smaller $\mathrm{\Delta }E$ ($S–T$), which facilitates efficient ISC and rISC. From a design point of view, the overlap of HOMO and LUMO can be manipulated by the way in which the donor and acceptor moieties are connected. This is emphasized by analyzing the properties of different combinations of triphenyltriazine-type acceptors (Trz) with carbazole-type donors (Cz and mCz or ${\mathrm{Me}}_2\mathrm{Cz}$), which have been published by Lee et al.^71,72 This work lead to the TADF emitters DCzTrz, TCzTrz, DCzmCzTrz, and TmCzTrz, which are shown in Fig. 6. Addition of an additional carbazole or dimethylcarbazole to DCzTrz changes the triplet energy $\mathrm{\Delta }E$ ($S–T$) from 0.23 to 0.16 and 0.07 eV, respectively, in TCtTrz and DCzmCzTrz.

Fig. 6

Impact of the connectivity of donors and acceptors on photophysical properties.^71,72

The photoluminescence quantum yield and excited-state lifetime are—in this case—connected to $\mathrm{\Delta }E$ ($S–T$): it changes from 43% at $31\mu \mathrm{s}$ (DCzTrz) to 100% at $14\mu \mathrm{s}$ (TCzTrz) and 99% at $14\mu \mathrm{s}$ (DCzmCzTrz). Note that the color of the four emitters shown in Fig. 6 depends on the choice of the donor unit. The carbazole-emitters DCzTrz and TCzTrz show peak electroluminescence around 480 nm, while the emitters with dimethylcarbazole, ${\mathrm{Me}}_2\mathrm{Cz}$, emit around 500 nm. This can be attributed to a variation of the HOMO energy of carbazole, which is slightly raised when electron-rich methyl groups are added.

When comparing ambipolar host molecules with TADF dopands, the differences are often subtle. Examples are pCzOXD,⁷³ a fluorescent, ambipolar host molecule and 2PXZ-OXD,⁷⁴ a TADF emitter, which are shown in Fig. 7. They both feature diphenyloxadiazole (OXD) as the acceptor unit (not shown in Fig. 5) as well two carbazole- (Cz) or phenoxazine-donors (PXZ) per molecule. Apart from the different HOMO energy, which leads to a smaller bandgap for 2PXZ-OXD due to the shallower HOMO of phenoxazine, the twist angle between the donor and the phenyl-substituent is different with a reported value based on DFT calculation of 52 deg for pCzOXD⁷³ and 75 deg for 2PXZ-OXD.⁷⁴ The larger twist breaks the conjugation between the donor and acceptor, which leads to a well-separated HOMO and LUMO and a small $\mathrm{\Delta }E$ ($S–T$) for 2PXZ-OXD,⁷⁴ while the HOMO is totally delocalized for 2PXZ-OXD, as suggested by DFT calculations.⁷³

Fig. 7

Differences between ambipolar host molecules⁷³ and TADF emitters⁷⁴ are often subtle.

2.2.

Organic Versus Copper Thermally Activated Delayed Fluorescence Organic Light-Emitting Diodes

In Sec. 2.1, we highlighted organic, small-molecule TADF emitters. For metal–organic TADF emitters, the same design principles are also essentially true (Fig. 8). In this case, the HOMO is often located on metal (or metal-halide) centered orbitals, while N-heterocyclic ligands (which are also being used in organic TADF emitters, refer to Fig. 5) house the LUMO. The same is also true for polymeric TADF emitters, where the HOMO and LUMO are located on different monomers of the polymer strand.⁷⁵

Fig. 8

Examples for metal–organic TADF emitters.^4,76,77

Both organic^66,78,79 and metal–organic TADF emitters⁷⁷ have been shown to reach the maximum theoretical OLED efficiency, which is in the order of 20% to 25% external quantum efficiency in cases where no light-extraction technology is applied.¹³ For clarity, the molecular structures of metal–organic TADF emitters are not discussed here. This has already been covered elsewhere.^80–82