TADF Parameters in the Solid State: An Easy Way to Draw Wrong Conclusions.

The successful development of thermally activated delayed fluorescence (TADF) OLEDs relies on advances in molecular design. To guide the molecular design toward compounds with preferable properties, special care should be taken while estimating the parameters of prompt and delayed fluorescence. Mistakes made in the initial steps of analysis may lead to completely misleading conclusions. Here we show that inaccuracies usually are introduced in the very first steps while estimating the solid-state prompt and delayed fluorescence quantum yields, resulting in an overestimation of prompt fluorescence (PF) parameters and a subsequent underestimation of the delayed emission (DF) yield and rates. As a solution to the problem, a working example of a more sophisticated analysis is provided, stressing the importance of in-depth research of emission properties in both oxygen-saturated and oxygen-free surroundings.

Introduction

According to spin statistics, only 25% of excitons in the typical OLED device are of a singlet nature. To enhance the internal quantum efficiency of a device with singlet emitters, nonemissive triplet excitons should be employed. As a solution, the thermal activation of triplet excitons and the subsequent reverse intersystem crossing (rISC) in TADF compounds allow us to utilize nearly all of the excited states and attain efficient emission.[1−3] To enable triplet recycling, the lowest-energy singlet and triplet states should be nearly isoenergetic.[4] Furthermore, to ensure high TADF efficiency, a prompt fluorescence radiative decay rate (kr) should be greater than the nonradiative decay, and the rISC rate (krISC) should exceed that of nonradiative triplet decay.[5] Moreover, TADF OLED stability[6] and a low external quantum yield (EQE) roll-off[7] also rely on maximizing the kr and krISC values. To optimize the material properties and later relate to device performance, fluorescence and electroluminescence yields should be estimated thoroughly. The most important parameters of prompt and delayed fluorescence (e.g., the rates of intersystem crossing (ISC) and reverse intersystem crossing), radiative and nonradiative fluorescence rates are calculated starting from the simplest ones–prompt and delayed fluorescence quantum yields (ΦPF and ΦDF, respectively) and the corresponding fluorescence decay rates (kPF and kDF, respectively).[5,8,9] Prompt and delayed fluorescence quantum yields usually are estimated either by simply measuring the efficiencies under oxygen-saturated (ΦPL+O) and oxygen-free (ΦPL–O) ambient conditions[10,11]or by deconstructing the fluorescence decay transient into prompt and delayed parts by fitting the fluorescence decay with a biexponential model and later estimating emission yields as[12,13]where A1 and A2 are the fractional intensities of prompt and delayed fluorescence and τPF and τDF are the prompt and delayed fluorescence lifetimes. The first method relies on the assumption that TADF is quenched by molecular oxygen under oxygen-saturated conditions and only prompt fluorescence is observed. This is typically observed in dilute solutions, but when TADF emitters are dispersed in solid films, this is rarely the case. Dense solid surrounding efficiently prevents oxygen diffusion inside the film, when the emitter molecules close to the surface are susceptible.[14−16] Typically, an evident part of only weakly quenched TADF still exists under +O2 conditions, making the direct application of eqs and 2 inaccurate.[17] Also, the unquenched part of TADF is larger for compounds with larger rISC rates since the rapid upconversion of triplet states reduces the chance of nonradiative collision with molecular oxygen,[17] especially complicating the analysis of novel TADF materials with rapid rISC. The second method relies on the assumption that all of the delayed fluorescence is collected during the measurement. However, the TADF lifetime usually is evidently prolonged in solid films, as compared to that in solutions, due to the presence of conformational disorder.[17−20] For compounds with less rigid molecular structure, weak delayed emission (e.g., 107 times weaker than the initial intensity) can be observed even after 0.1 s,[17] making the measurements of the TADF transient rather complicated.

In this article, we showcase an easy risk to estimate the prompt and delayed fluorescence parameters with large variation, which might eventually lead to inaccurate conclusions. We show that fluorescence decay rates may be estimated within the 1 order of magnitude error, depending on the accuracy of the initial emission parameters. Such variation of TADF rates significantly complicates the analysis and comparison of material parameters and the prediction of OLED performance. On the other hand, we show that reliable emission parameters can be obtained after the thorough analysis.

Methods

TADF compounds were analyzed in 1 wt % PMMA (PXZPM, 4CzPN), 7 wt % mCP (tCz-ND), and 3 wt % TSPO1 (ARCPyr) films. A larger doping concentration in mCP/TSPO1 films was used to ensure the full energy transfer from host to emitter and simultaneously prevent concentration quenching. Films were prepared by dissolving each material and host in appropriate ratios in toluene solutions and then wet-casting the solutions on quartz substrates. Time-integrated fluorescence spectra and fluorescence decay transients were measured using nanosecond YAG:Nd3+ laser NT 242 (Ekspla, τ = 7 ns, pulse energy 200 μJ, repetition rate 1 kHz) and time-gated iCCD camera New iStar DH340T (Andor). Fluorescence transients were obtained by exponentially increasing the delay and integration times.[21] Fluorescence quantum yields (±5% error) were estimated using the integrated sphere method[22] by integrating the sphere (Sphere Optics) connected to CCD spectrometer PMA-12 (Hamamatsu) via optical fiber. Solid-state samples were mounted in a closed-cycle He cryostat (Cryo Industries 204N) for all fluorescence measurements (for oxygen-saturated and oxygen-free conditions).

Results and Discussion

Four TADF compounds were analyzed (Figure ). Extensively analyzed compound PXZPM(23−26) was selected as a model compound to showcase the peculiarities of ΦPF and ΦDF. However, PXZPM has a rather flexible molecular core and shows evident conformational disorder.[26] Compounds 4CzPN,[1,17]tCz-ND,[27] and ACRPyr,[28] however, were selected due to the rigid molecular core and minor conformational disorder, enabling the comprehensive analysis of solid-state emission properties.

Figure 1

Molecular structures of compounds PXZPM, 4CzPN, tCz-ND, and ACRPyr.

Initially, phenoxazine-pyrimidine compound PXZPM was analyzed. Compound PXZPM was shown to be an efficient green TADF emitter with a fluorescence quantum yield of 1 in the mCPCN host and prompt and delayed fluorescence lifetimes of 20.2 ns and 2.56 μs, respectively, with similar parameters in the PMMA polymer host.[25] Prompt fluorescence was shown to dominate the emission with ΦPF = 0.65 and kr = 3.22 × 107 s–1. Fluorescence decay transients of PXZPM dispersed in the PMMA host at a 1 wt % doping level are shown in Figure a. The intensity and temporal ranges were selected to be identical to those reported in ref (25). Namely, the fluorescence intensity scale ranged from 1 to 10–4, while the timescale ranged from 0 to 20 μs. As we can see, the temporal profile of PXZPM decay is very similar to the one reported in ref (25), where the intense initial PF was observed, followed by the long-lived DF. The initial TADF decay followed a nearly single-exponential decay profile, similar to that in ref (25) with a comparable decay constant (τTADF = 5 μs). However, the situation in Figure a is only a small part of the big picture. Actually, the weak delayed emission of PXZPM is observed even up to about 20 ms, as evident from the TADF transient over a wide intensity and time range (Figure b). Indeed, the fractional intensity of the delayed fluorescence, according to the analysis by eqs –5, is clearly larger, amounting to about 73% of the total emission, which is more than twice the value stated in ref (25). Similar fractions of prompt and delayed fluorescence were also estimated by measuring the fluorescence intensity enhancement under −O2 conditions (Figure c). However, the direct use of eqs and 2 would also lead to wrong conclusions. As we can see, the fluorescence intensity under ambient −O2 is 1.44 times larger than that under oxygen-saturated conditions. From this ratio, the DF fraction would be 59%, nearly 26% lower than the actual value. As seen in Figure b, a considerable part of TADF still exists under oxygen-saturated conditions, amounting to about 52% of the total emission under +O2 conditions. Therefore, PXZPM actually yields ΦPF of 0.25 and ΦDF of 0.67 in the PMMA film, together with a radiative decay rate of 1.29 × 107 s–1 (kr = ΦPF × kPF), almost the same as in toluene.[26] However, as the delayed emission, shown in Figure b, was clearly multiexponential due to the evident conformational disorder, it was impossible to estimate the exact TADF lifetime and compare the solid-state TADF parameters.[26] For this purpose, three TADF emitters with rigid molecular structure and nearly single-exponential TADF decay, namely, 4CzPN, tCz-ND, and ACRPyr, were analyzed (Figures and 3).

Figure 2

(a) Fluorescence decay transient of a 1 wt % PMMA film of PXZPM in a narrow intensity and temporal range under −O2 conditions. (b) Fluorescence decay transient of a 1 wt % PMMA film of PXZPM over a broad intensity and temporal range under +O2/–O2 conditions. The shaded area represents the range used in Figure a. (c) Fluorescence spectra of 1 wt % PMMA films of PXZPM under +O2/–O2 conditions. The black line represents the emission spectrum of solely prompt fluorescence, excluding the existing DF part.

Figure 3

(a–c) Solid-state fluorescence decay transients of compounds 4CzPN, tCN-ND, and ACRPyr under +O2 and −O2 conditions. The PF share in the total decay under +O2 conditions is shown for every compound. (d–f) Solid-state fluorescence spectra of compounds 4CzPN, tCN-ND, and ACRPyr under +O2, −O2, and +O2 conditions without the DF part (+O2*).

All three compounds showed intense and rather short-lived delayed fluorescence (τTADF ranged from 1.76 to 46 μs). This rapid delayed fluorescence was weakly quenched under +O2 conditions (Figure a–c), leading to a minor PL intensity difference under oxygen-saturated and oxygen-deficient conditions (Figure d–f). The direct use of eqs and 2 would give ΦPF for all three TADF compounds in the range from 0.44 to 0.7 (0.44, 0.6, and 0.7 for ACRPyr, 4CzPN, and tCN-ND, respectively). However, as we can see from Figure a–c, the PF share (ηPF = ∫IPF/∫IPL) in the total emission under +O2 conditions was only 0.14–0.30, leading to a remarkably lower real ΦPF of 0.07–0.23 (0.07, 0.09, and 0.23 for ACRPyr, 4CzPN, and tCN-ND, respectively, equation ΦPF = ΦPL+2ηPF) and a remarkably larger real ΦDF of 0.53–0.63 (0.59, 0.64, and 0.53 for ACRPyr, 4CzPN, and tCN-ND, respectively; eq ). Such variation in ΦPF and ΦDF values leads to very large discrepancies between accurate and inaccurate TADF parameters. This is shown in Figure , where the radiative fluorescence decay and rISC rates are compared. Both kr and krISC were showcased as both strongly depending on the emission yield, and both rates are used for the estimation of other major fluorescence parameters.[5,8]

Figure 4

(a) Fluorescence radiative decay and (b) reverse intersystem crossing rates for compounds 4CzPN, tCN-ND, and ACRPyr. Accurate values are shown as black bars, and inaccurate values are shown as blue bars. The numbers close to the arrows denote the ratio between both values. krISC was calculated according to the models used in the initial reports. krISC of 4CzPN was calculated according to Kreiza et al.[8]

As shown, accurate kr ranged from 3.9 × 106 s–1 (ACRPyr) to 44.2 × 106 s–1 (tCz-ND). Such a high kr for tCN-ND was in line with the rapid PF decay and high oscillator strength of the S0 → S1 transition.[27] However, the rapid kr with values exceeding 107, 1 order of magnitude larger than the accurate ones, could be derived if the overestimated ΦPF was used. In this case, kr ranged from 22.7 (ACRPyr) to the remarkable 135 × 106 s–1 for tCz-ND. kr values of >108 s–1 are typical for organic lasing materials with strong LE emission[29] and are hardly likely for CT-based TADF compounds.[30] On the contrary, the rISC rate was underestimated even more. The actual krISC ranged from 0.2 × 106 s–1 for 4CzPN to 5.7 × 106 s–1 for ACRPyr. When the enlarged ΦPF was used, krISC decreased down to 0.011 × 106 s–1 (4CzPN) to 0.9 × 106 s–1 (ACRPyr). Clearly, such a deviation in the fluorescence parameters by up to 1 order of magnitude complicates the material optimization and may provide wrong guidelines, as the impact of delayed fluorescence is evidently underestimated. Somewhat similar results should be obtained if only the initial and intense delayed fluorescence is accounted for in the fluorescence transients, as shown in Figure d–f. Therefore, to avoid such tremendous errors in solid-state TADF parameters, great care should be taken. For instance, the existing DF part should be eliminated from ΦPF under +O2 conditions. Concomitantly, the TADF transients should be measured over large intensity and temporal ranges, including the weak DF at the largest delays.[21,31]

Conclusions

We have shown that solid-state TADF parameters can be estimated with high inaccuracy. The specific solid-state surrounding prevents the full delayed fluorescence quenching in ambient air; therefore, it is critically important to exclude the remaining DF part in order to get the correct prompt and delayed fluorescence quantum yields according to eqs and 2. On the other hand, the conformational disorder existing in the solid state usually remarkably extends the delayed fluorescence lifetime, when the latest weak delayed fluorescence is difficult but critical to assess. Failing to do that, prompt and delayed fluorescence parameters, according to eqs –5, can be estimated within 1 order of magnitude error, which is highly unfavorable for material and device optimization.