A Simple Molecular Design Strategy for Delayed Fluorescence toward 1000 nm.

Harnessing the near-infrared (NIR) region of the electromagnetic spectrum is exceedingly important for photovoltaics, telecommunications, and the biomedical sciences. While thermally activated delayed fluorescent (TADF) materials have attracted much interest due to their intense luminescence and narrow exchange energies (ΔEST), they are still greatly inferior to conventional fluorescent dyes in the NIR, which precludes their application. This is because securing a sufficiently strong donor-acceptor (D-A) interaction for NIR emission alongside the narrow ΔEST required for TADF is highly challenging. Here, we demonstrate that by abandoning the common polydonor model in favor of a D-A dyad structure, a sufficiently strong D-A interaction can be obtained to realize a TADF emitter capable of photoluminescence (PL) close to 1000 nm. Electroluminescence (EL) at a peak wavelength of 904 nm is also reported. This strategy is both conceptually and synthetically simple and offers a new approach to the development of future NIR TADF materials.

All-organic materials exhibiting thermally activated delayed fluorescence (TADF) have emerged as next generation dopants in organic light-emitting devices (OLEDs).[1,2] Efficient emission has been reported throughout the visible spectrum, facilitated by modular synthesis and rational design.[3] Despite this, it has not yet been possible to realize efficient emission in the near-infrared (NIR) region (ca. 700–1000 nm).[4−7] The longest wavelength TADF emitter reported to date exhibits a maximum peak emission wavelength (λmax) of ca. 800 nm,[8] marginally outside the visible spectrum, and considerably below the values of ≥1000 nm that are well established for first generation fluorescent emitters.[5,7,9,10] Consequently, shifting the emission from TADF materials to comparably low energies remains an immediate challenge if potential applications in telecommunications, night vision, organic photovoltaic devices (OPVs), organic lasers, photodetectors, and NIR bioimaging are to be realized.[5,6,11]

All-organic TADF emitters are typically assembled from electron-rich donor (D) and electron-poor acceptor (A) heterocycles, with twisted linkages to ensure a small exchange energy (ΔEST). Emission occurs from an intramolecular charge transfer (ICT) state. Since the seminal structures reported by Adachi et al. (e.g., 4CzIPN),[12,13] there has been an overwhelming trend toward acceptor cores functionalized with multiple peripheral donors, i.e., polydonor D–A structures where n > 1.[2] This also applies to the vast majority of deep-red and NIR TADF candidates published to date.[6,14−21] For example, Jiang, Liao and co-workers reported a D2–A acenaphthene-based emitter (APDC-DTPA) (Figure ) which exhibits emission at λmax = 687 nm and a photoluminescence quantum yield (PLQY) of 63% when doped into TBPi at 10 wt % (λmax = 756 nm, PLQY = 17% in neat film).[15] Adachi et al. also studied two NIR TADF dyes based on unconventional boron curcuminoid acceptors.[8,16] The redder D4–A2 analogue (Figure ) emits at λmax = 801 nm with a PLQY of 4% at 40 wt % in CBP. These examples constitute the current state-of-the-art NIR TADF emitters.

Figure 1

Molecular structures of CAT-1 and state-of-the-art NIR TADF materials and their lowest-energy reported photoluminescence λmax values.

An exceptionally stabilized ICT state is required for NIR TADF. This is intuitively achieved through ensuring the D–A interaction is as strong as possible, which is compromised in common polydonor designs. This is because the multiple peripheral donors that are typically employed in TADF materials are not directly electronically coupled, reducing their additive effect.[12,22,23] An increase in the number of donors, while enhancing the net donor strength, also suppresses the strength of the acceptor through functionalizing it with an increasing number of discrete electron-donating groups, compromising any red shift. For many red TADF materials,[19,22−25] any bathochromic shift in emission afforded by additional donors is incremental at best compared to single-donor D–A dyads, which now comprise a noteworthy portion of the most efficient red examples.[26,27] It was also recently demonstrated that converting D2–A APDC-DTPA into its D–A analogue TPAAP (Figure ) red-shifts the PL by 21 to 777 nm in a neat film (PLQY = 20%).[28] Herein, we demonstrate a simple concept to afford unprecedented progress into the NIR through adopting an uncomplicated D–A dyad structure.

The structure of the new NIR TADF emitter, CAT-1, is shown in Figures and 2. It can be easily synthesized (Scheme S1) on a multigram scale for thermal evaporation (Td = 409 °C, Tg = 128 °C) (Figure S3) with minimal chromatography. The removal of an electron donor in APDC-DTPA to afford TPAAP facilitates an incremental 21 nm red shift in neat film emission.[28] In CAT-1 we take greater advantage of the D–A dyad structure by replacing the second electron donor with a more beneficial electron-withdrawing CN group.[29−33] The additional CN group is both proximal to the single electron donor and directly conjugated to it via the pyrazine moiety, which may contribute to the greatly increased acceptor strength (Figure S1). This simple and rational addition substantially stabilizes ICT to obtain unprecedentedly red-shifted PL from a NIR TADF material.

Figure 2

Numbering scheme, parameters calculated by TD-B3LYP/6-31G* and HOMO and LUMO distributions for CAT-1.

CAT-1 was designed using DFT/TD-DFT (B3LYP/6-31G*) (Figure , Figure S4, Tables S2–S8). In the optimized geometry the phenyl spacer-acceptor dihedral angle of 61°, ensures a narrow calculated ΔEST of 0.11 eV. Nevertheless, an oscillator strength (f) of 0.1052 for the S0 → S1 excitation is maintained (Table S3). This compares favorably with other NIR emitters that have narrow calculated ΔEST values[14,19,30] (e.g., APDC-DTPA (ΔEST = 0.10 eV, f = 0.0266; Table S4)).[15] We note that TD-PBE0/def2-SVP, TDA-PBE0/def2-SVP, and TDA-PBE0/def2-TZVP all also predict narrow ΔEST values ≤0.15 eV (Tables S6–S8).[34,35] Cyclic voltammetry (Figure S2, Table S1) indicates that the additional CN affords a deep[14,15,28,29] LUMO energy of −4.11 eV (HOMO = −5.64 eV) and a very narrow electrochemical band gap (1.53 eV), highlighting the high strength of the new electron acceptor unit.

The photophysical properties of CAT-1 were studied in dilute solution, doped evaporated films (with CBP as host), and neat films prepared by both thermal evaporation and drop-casting. λmax and PLQY data are summarized in Table (further data are recorded in Table S10).

Table 1

Summary of the PL Data for CAT-1

sample preparation	λmax PL/nm	PLQYa /%
toluene solution	770	3.9 ± 0.4
10 wt % in CBP	763	8.8 ± 0.2
40 wt % in CBP	820	1.98 ± 0.04
neat evaporated	887	0.18 ± 0.04
neat drop-cast	950	≤0.18

Absolute PLQY measured using an integrating sphere.

The absorption spectrum in toluene consists of three major bands (Figure ). The high energy region (≤380 nm) is ascribed to local π–π* transitions, the broad low energy band centered at 560 nm is assigned to D–A ICT, and the shoulder around 450 nm is mixed character. This assignment is supported by TD-DFT (Figure S7). The photoluminescence (PL) is broad and unstructured (toluene λmax = 770 nm) and exhibits positive solvatochromism (Figures S10 and S11, Table S9), indicating emission from an ICT state. A large red shift in solution PL compared to APDC-DTPA(15) (130 nm) highlights the intrinsic efficacy of the design strategy in the absence of aggregation.

Figure 3

(a) Normalized absorption and PL spectra for CAT-1 in toluene. λex = 520 nm. (b) Normalized Steady state PL spectra for CAT-1 in doped and neat films. λex = 520 nm. Spectra are reported to 1000 nm due to the low detector sensitivity beyond 1000 nm. (c) Time-resolved PL intensity of CAT-1 doped into CBP at different weight ratios recorded in 10–3 mbar vacuum. Instrument response function (IRF) plotted in blue. λex = 400 nm.

When doped into CBP at 10 wt %, CAT-1 exhibits low-energy emission with λmax = 763 nm and a PLQY of 8.8 ± 0.2% (Figure ). As well as prompt fluorescence (τ < 5 ns), a delayed component (τ = 80 μs, 13.5% contribution) to the PL is also observed (Figure , Figure S13). The delayed PL is spectrally similar to the prompt (Figure S14) and is thus assigned to delayed fluorescence from the same state. The intensity of the delayed fluorescence increases with temperature between 10 and 292 K, confirming a TADF mechanism (Figure S17). It is also significantly quenched in the presence of oxygen (Figure S13), indicating a contribution from triplet states, as expected. Furthermore, a narrow ΔEST of ca. 0.04 eV was estimated experimentally from the onsets of the low temperature fluorescence and phosphorescence spectra of CAT-1 in CBP at 10 wt % (Figure S19).

Increasing the doping ratio of CAT-1 in the evaporated films leads to significant bathochromic shifts in emission (Figure , Table , Table S10). This is common for highly polar TADF materials and is attributed to the effects of solvatochromism and aggregation.[7,28,36] This is also accompanied by a sequential decrease in PLQY,[16,28] indicating a trade-off between PL wavelength and quantum efficiency when exploiting aggregation to red-shift emission. When CAT-1 is doped into CBP at 40 wt %, emission is recorded at λmax = 820 nm with a TADF component still present (τ = 8 μs, 5.1% contribution) (Figures S13 and S18). This is lower in energy than for the D4–A2 curcuminoid derivative reported by Adachi et al. (λmax = 801 nm) (Figure ),[8] although our PLQY is lower (2% vs 4%). Unlike the curcuminoid the molecular weight of CAT-1 is also sufficiently low for desirable thermal evaporation.[14,15,29] To the best of our knowledge this is the lowest-energy TADF confirmed by variable temperature time-resolved measurements to date (Table S11).

In evaporated films of CAT-1 PL is recorded with λmax = 887 nm (Figure ). While the signal-to-noise ratio is poor due to the low PLQY, the longer lifetime component of the PL decay also appears to be quenched in the presence of oxygen (Figures S15 and S16). Emission from films drop-cast from chlorobenzene solution is further reproducibly red-shifted to λmax = 950 nm with a sharpening of the onset (Figure ) and an associated red shift of the absorption spectrum (Figure S9). A similar PL red shift upon solution processing was recently reported for a red TADF emitter.[21,37] Analysis of the drop-cast film by thin film X-ray diffraction indicates a degree of solid state ordering consistent with enhanced aggregation (Figures S26 and S27). The large PL red shift in the drop-cast films is attributed to this.[28] No such crystallinity could be observed for the evaporated film, although this may be related to the low film thickness (Figure S28). While reabsorption in the comparatively thick drop-cast films may also influence this red shift in PL, such a dependence of the emission color on the processing method (e.g., thermal evaporation or solution processing) is well recognized for TADF materials.[38−41]

In summary, strategic replacement of the second electron donor of APDC-DTPA with a CN group in CAT-1 affords large bathochromic shifts in PL: 130 nm in toluene, 76 nm in 10 wt % doped film, and 131 nm in evaporated film. This is considerably greater than the evaporated film red shift of 21 nm obtained upon sole removal of an electron donor (TPAAP) (Figure S20).[15,28] Notably, when doped into CBP at 10 wt % CAT-1 also emits with a longer wavelength PL λmax (763 nm) than APDC-DTPA in a neat evaporated film (756 nm). While the low PLQY precludes unambiguous confirmation of TADF in the drop-cast film, the PL λmax of 950 nm is red-shifted by 149 nm compared to the longest wavelength PL λmax reported thus far for a TADF-capable emitter (curcuminoid, Figure , Table S11).[8] Furthermore, CAT-1 can display an emission spectrum for which the entirety of the PL falls >700 nm, which is considered “real” NIR.[7,11] In fact, the drop-cast film has an impressive PL onset of ca. 800 nm, which is exceptional for a material capable of TADF.[11]

The electroluminescence (EL) performance of CAT-1 was also evaluated in preliminary thermally evaporated OLED devices (Figures S21–S25). Crucially, we note that undoped devices exhibit complete NIR (>700 nm) emission with an EL λmax of 904 nm (Figure ). This red shift in the undoped EL compared to the neat evaporated film PL (904 nm vs 887 nm) may be because excitons are formed via a diffusional process in the device rather than direct photoexcitation. Optical interference is another possible explanation. While the maximum external quantum efficiency (ca. 0.019%) and radiance (ca. 1 × 103 mW sr–1 m–2) are low, the peak EL λmax is much lower in energy than for any previously reported TADF material (>100 nm), while being competitive with conventional fluorescent devices and red-shifted compared to almost all phosphorescent examples.[11] These data importantly confirm that the favorable long-wavelength PL of a neat film of CAT-1 can be retained in an EL device.

Figure 4

EL spectrum for an undoped CAT-1 OLED at 5 V. λmax = 904 nm.

In conclusion, this work provides a great advance into the NIR from a delayed fluorescent material. At the high donor and acceptor strengths required for NIR TADF, the incorporation of multiple donors is essentially ineffectual at red-shifting emission.[19,22−25,28] In contrast we have shown that adopting an intrinsically simple D–A system structurally liberates molecules for rational functionalization to greatly stabilize the ICT state. We have succeeded in obtaining a dramatic red shift in PL λmax of over 100 nm from a material capable of TADF compared to previous work (Figure ). This facilitated fabrication of OLEDs displaying impressive EL λmax values of >900 nm. The next step is clearly to improve the PLQY in this newly accessed low-energy region. This may be possible through incorporation of a more rigid electron donor capable of suppressing aggregation-induced quenching. Nevertheless, with respect to emission wavelength, CAT-1 affords rapid and important progress. These unprecedented results crucially stem from a design strategy that is conceptually simple and easy to implement synthetically. It will be broadly applicable to other systems and heavily influence future NIR TADF design.