The role of molecular oxygen (O2) and UV light in the anion radical formation and stability of TCNQ and its fluorinated derivatives

We report the electronic absorption spectroscopy of 7,7,8,8-tetracyanoquinodimethane (TCNQ) and its fluorinated derivatives (F2TCNQ and F4TCNQ), well-known electron-accepting molecules in common organic solvents (toluene, chlorobenzene, acetonitrile, and ethanol) under controlled exposure to air (O2) and UV light. All compounds (FxTCNQ (x = 0, 2, 4)) were stable in a neutral state (FxTCNQ0) in toluene and chlorobenzene, even under both O2 and UV light. On the other hand, in EtOH, the formation of FxTCNQ·− was monitored upon controlled exposure to O2 or UV light. Especially in air-equilibrated ethanol upon the UV-illumination, efficient α,α-dicyano-p-toluoylcyanide anion (DCTC−) and its fluorinated derivatives were generated evinced by the absorption peak near 480 nm, whereas the reaction was shut off by removing O2 or blocking UV light, thereby keeping FxTCNQ·− stable. However, even in deaerated ethanol, upon the UV-illumination, the anion formation of TCNQ and its fluorinated derivatives (FxTCNQ·−, x = 0, 2, 4) was inevitable, showing the stability of FxTCNQ0 depends on the choice of solvent.

For further energy level engineering, several fluorinated derivatives, tetrafluorinated/difluorinated TCNQ (F4TCNQ and F2TCNQ), were developed and successfully have shown the tunability of the first reduction potential with respect to TCNQ: E 1/2 (TCNQ ·− /TCNQ) = − 54 mV; E 1/2 (F2TCNQ ·− /F2TCNQ) = + 165 mV; E 1/2 (F4TCNQ ·− / F4TCNQ) = + 365 mV versus Ag/Ag + in CH 3 CN (Le et al. 2011;Miyasaka et al. 2010). Furthermore, these fluorinated TCNQ derivatives have been shown to form radical anions and dianions in polar solvents such as acetonitrile and ethanol (Ma et al. 2014;Vo et al. 2018;Chae et al. 2014), while external conditions such as O 2 or UV light for anion species formation have not been well characterized. Therefore, to utilize the electron-accepting property of TCNQ and its fluorinated derivatives in many devices, it is crucial to understand the conditions to keep TCNQ derivatives neutral (TCNQ 0 ) and to prevent from producing by-products such as anions or DCTC − . In this regard, the role of O 2 and UV light on the spectroscopic properties of TCNQ derivatives is worth investigating. Here, we report the electronic absorption spectroscopy of 7,7,8,8-tetracyanoquinodimethane (TCNQ) and its fluorinated derivatives (F2TCNQ and F4TCNQ) in various common organic solvents (toluene-Tol, chlorobenzene-CB, acetonitrile-ACN, and ethanol-EtOH) either as air-equilibrated or as N 2 -purged with controlled UV-illumination. TCNQ showed the production of DCTC − upon the reaction of TCNQ ·− with O 2 in EtOH under UV-illumination, whereas in N 2 -purged EtOH, no or minimal DCTC − was observed. On the other hand, stable neutral form TCNQ 0 was confirmed in Tol, CB, or ACN. Similarly, F2TCNQ and F4TCNQ in air-equilibrated EtOH exhibited the effective production of fluorinated DCTC − derivatives with an absorption peak near 480 nm under UV, which was suppressed by removing O 2 or blocking UV light.

Instrumentation
Electronic absorption spectra were acquired using a Hitachi U-3900 UV/visible/NIR spectrophotometry system in a quartz optical cell. Baseline corrections for the transmittance of an optical cell with solvent were made prior to each measurement. In addition, the evolution of an absorption spectrum upon the UV-illumination (365 nm, 4 W, CW UV lamp) was monitored either with molecular oxygen in an air-equilibrated solution or without molecular oxygen in an N 2 (99.999% purity)-purged solution to examine the absorption behavior of TCNQ derivatives, dependent on the presence of O 2 .

Results and discussion
To fabricate OPV devices or to prepare solution samples for photocatalyst characterization, electrical or photophysical studies involving TCNQ, solution processes are commonly used. In this regard, electronic absorption spectra of TCNQ were explored in various common organic solvents: toluene (Tol), chlorobenzene (CB), acetonitrile (ACN), and ethanol (EtOH). Additional file 1: Figure S1 displays the electronic absorption spectra of TCNQ in air-equilibrated solvents, and the spectroscopic results are tabulated in Table 1. In all solvents, the absorption spectra of TCNQ revealed the characteristic S 0 → S 1 transition in the range of 394-403 nm, consistent with the previous literature results (Suchanski and Duyne 1976). However, in Additional file 1: Figure S1d (black line), TCNQ in EtOH displays additional dual peaks in 700-900 nm, which are typical peaks for TCNQ anion radical (TCNQ ·− ) (Melby et al. 1962).
The spectroscopic behavior of TCNQ was further tracked upon UV light illumination every 3-min to examine its photostability for a total of 15 min, as shown in Fig. 2a, Additional file 1: Figures S1 and S3. The spectroscopic results of TCNQ, either in air-equilibrated Tol, CB, or ACN, showed a virtually consistent absorption spectrum. In contrast, TCNQ in air-equilibrated EtOH (EtOH-air) exhibited a substantial drop of the intensity for the absorption band at 396 nm even in 3 min in Fig. 2a. With the decrease of 396 nm band, multiple absorption bands at 421, 743, and 841 nm, corresponding to TCNQ ·− increased, and a new band at 474 nm emerged. At 6 min (green in Fig. 2a), even the absorption bands of TCNQ ·− at 421, 743, and 841 nm began to decrease, and the 474 nm band continuously gained its intensity up to 15 min (magenta line in Fig. 2a), when the original absorption peak at 396 nm was wholly disappeared. The spectral change for TCNQ in EtOH-air in Fig. 2a suggests that TCNQ reacted to generate TCNQ ·− , in contrast to stable TCNQ 0 in Tol, CB, or ACN. Furthermore, the spectral evolution upon the UV-illumination from 3 to 15 m, featuring the decrease of 743 and 841 nm band, implies that TCNQ ·− was reacted to produce the species that shows an absorption peak at 474 nm. The new species that shows the absorption band at 474 nm has been ascribed to α,α-dicyano-p-toluoylcyanide anion (DCTC − , Additional file 1: Figure S2) previously (Suchanski and Duyne 1976;Hertler et al. 1962;Mizoguchi et al. 1978;Kryszewski et al. 1981;Grossel et al. 2000;Xiulan et al. 2012), and the details of DCTC − generation are discussed below. No sign of TCNQ 2− with a peak at ~ 330 nm was observed during this spectroscopic evolution, contrasting to the previous studies in ACN (Suchanski and Duyne 1976;Chae et al. 2014). Due to the potential role of molecular oxygen (O 2 ) in air-equilibrated EtOH in DCTC − generation, the spectroscopic behavior of TCNQ was further examined in degassed EtOH (EtOH-N 2 ) by pre-purging EtOH solvent and purging TCNQ solution in EtOH with high purity (99.999%) N 2 gas. Figure 2b displays the spectral evolution upon the same UV-illumination. Like the absorption spectrum of TCNQ in EtOH-air at 0 m in Fig. 2a (black line), TCNQ in EtOH-N 2 also exhibited the characteristic 396 nm band for TCNQ 0 in addition to 743 and 841 nm bands corresponding to TCNQ ·− in Fig. 2b (black line). At 3 m (red line), TCNQ ·− peaks increased with the decrease of TCNQ peak, and the additional band at 421 nm is also the spectroscopic fingerprint of TCNQ ·− , which was buried with TCNQ 0 band at 0 m due to the spectral proximity. From 6 min (green line), the spectral change in EtOH-N 2 contrasts with that in EtOH-air, lacking the growth of the DCTC − absorption band at 474 nm. Also, the absorption peak intensity of TCNQ ·− peaks in EtOH-N 2 was kept constant, suggesting stable TCNQ ·− and no further reactions consuming TCNQ ·− occurred. The formation of DCTC − in EtOH-N 2 could be blocked because of the absence of TCNQ ·− even without O 2 . Additional file 1: Figure S4 presents the absorption spectra of TCNQ as a function of time passed since blocking UV light in the middle of UV-illumination (exposed to UV light for 9 min), clearly showing the presence of TCNQ ·− . Over the 10 min since blocking UV light, no DCTC − generation was noticed, highlighting the role of both O 2 and TCNQ ·− .
Since the initial report of TCNQ, several DCTC − generation mechanisms from TCNQ have been proposed, whether the reaction starts either from TCNQ 0 , TCNQ ·− , or TCNQ 2− Mizoguchi et al. 1978;Kryszewski et al. 1981;Grossel et al. 2000;Xiulan et al. 2012). To identify the origin of DCTC − formation, we  plotted the absorbance (Abs 474 ) of DCTC − product, monitored at 474 nm against the absorbance (Abs 841 ) of TCNQ ·− at 841 nm using the data in Fig. 2a. If DCTC − is produced from TCNQ ·− , as the stoichiometry of TCNQ ·− and DCTC − is 1:1 based upon the potential mechanism shown in Additional file 1: Figure S2, the concentration of the produced DCTC − ([DCTC − ] p,t ) at a given t should be equivalent to that of the consumed TCNQ ·− ([TCNQ ·− ] c,t ) at the same t.
From the Beer's law: where ε x (A) is the molar absorption coefficient at x nm for A species. Therefore, from the equivalence in Eq.
(1) as well as Eqs. (2) and (3) Grossel et al. 2000;Melby et al. 1962), ε 474 (DCTC − )/ε 841 (TCNQ ·− ) can be calculated to be 0.892, which matches reasonably well despite the solvent difference. In addition, Abs 841 (TCNQ ·− ) is plotted against the UV-illumination time in Fig. 3b, showing the pseudofirst-order reaction for TCNQ ·− behavior is observed, again supporting that the reaction of TCNQ ·− is not mediated by the self-collisions, consistent with the proposed mechanism by Hipps et al. (Qi et al. 2012) and the resulting fit determined the rate constant to be 0.13 m −1 .
Structurally related fluorinated TCNQ molecules (F2TCNQ and F4TCNQ: FxTCNQ, x = 2 and 4) have been widely utilized as comparative and tunable functional molecules to TCNQ, as their spectroscopic signatures are similar in spite of the modified reduction energy levels. The absorption spectra of FxTCNQ, either EtOH-air or EtOH-N 2 , in Fig. 4a, b, d, e showed substantial change over time, like TCNQ in EtOH. On the other hand, as presented in Additional file 1: Figure S5a, b, d, Abs 474 = − ε 474 DCTC − /ε 841 TCNQ ·− Abs 841 e, no appreciable spectral change was probed in Tol or CB even under UV-illumination, while in ACN, F4TCNQ showed a noticeable decrease in 387 nm band, contrasting to F2TCNQ that displayed no spectral change ( Figure  S5c, f ). This is likely due to the higher electron affinity of F4TCNQ, resulting in the reduction of F4TCNQ. On the other hand, in EtOH-air, both F2TCNQ and F4TCNQ in Fig. 4a, b, d, e displayed a dramatic change as in TCNQ. As the formation of an oxidized product, DCTC − from TCNQ ·− was evident from a new absorption band at ~ 480 nm in Fig. 2a, F2TCNQ and F4TCNQ in EtOH-air similarly show the rise of ~ 480 nm band under UV light. Congruent to TCNQ, in EtOH-N 2 , the effective growth of the 480 nm band was not observed, again suggesting that fluorinated DCTC − derivatives were not produced without O 2 . Additionally, even in EtOH-air, without UV light, FxTCNQ ·− (x = 0, 2, 4) was not effectively generated in Fig. 4c, f, evincing the stable neutral TCNQ 0 and F2TCNQ 0 , contrary to unstable F4TCNQ that underwent F4TCNQ 2− formation (Melby et al. 1962).

Conclusions
In summary, the electronic absorption spectroscopy of FxTCNQ (x = 0, 2, 4) probed the stability of FxTCNQ (x = 0, 2, 4) in common organic solvents upon controlled exposure to O 2 or UV light. In general, all FxTCNQ compounds were stable in Tol and CB even under both O 2 and UV light, whereas in EtOH, the FxTCNQ ·− formation was monitored. Furthermore, in air-equilibrated EtOH upon UV-illumination, efficient α,α-dicyano-ptoluoylcyanide anion (DCTC − ) and its fluorinated derivatives were generated, whereas the reaction was shut off by removing O 2 or blocking UV light. With the significance of TCNQ and its derivatives (F2TCNQ and F4TCNQ) not only in molecular electronics but also in electrochemistry, this work will provide an understanding of