Photophysical, electrochemical, and spectroelectrochemical investigation of electronic push–pull benzothiadiazole fluorophores

Introduction

Benzothiadiazole (BZT) has been widely used in an array of applications [1–3]. This is in part due to its intrinsic electron accepting character that gives rise to interesting opto-electronic properties. These properties are enhanced when BZT is introduced between electron rich neighbors. The resulting π-conjugated donor–acceptor–donor arrangement results in BZT derivatives having combined optical and electrochemical properties that are well-suited for their use as active functional materials in plastic electronics [4, 5], including electrochromic devices [6], organic field effect transistors [7–9], and solar cells [10–14]. BZT derivatives are also known as active components that signal pathogen attack [15] and stimulate defense reactions of plants [16].

Among the many useful properties of BZT, is its intense fluorescence [17]. Its inherent high fluorescence yield has proven beneficial for emitting applications such as organic light emitting diodes [18–20]. The intense fluorescence of BZT derivatives has also found uses in bioimaging [21] and for nucleic base molecular recognition in DNA sensing [22]. The versatility of BZT as a fluorophore is in part owing to its emission color that can be adjusted contingent on the aromatics substituted in the 4,7′-positions [23]. Emission tailoring is also possible by conjugating BZT with various electron donating and withdrawing groups. These groups perturb the excited state, resulting in intramolecular charge transfer [2] and fluorescence solvatochromism [24].

N,N-Dimethylamino and nitro groups are typically used as complementary electron donating and withdrawing functionalities, respectively, for solvatochromic push–pull fluorophores. However, the nitro group can be problematic. It can significantly quench the fluorescence by many deactivation modes, including photoinduced electron transfer, aryl–NO₂ bond rotation, and solvent–solute interactions [25–27]. It would be therefore advantageous to design a fluorophore whose fluorescence yield would be consistent in a wide range of solvents, yet whose emission color would dependent on its environment. These are ideal features of a universal fluorescence probe that could potentially span a wide range of applications from biological sensors to polymer deformation probes.

Given their similar electron affinity, the nitrile group is a potential viable replacement for the nitro group. We were therefore motivated to investigate the electronic push–pull analogues 1 and 2 (Chart 1) to better understand the effect of the similar electron withdrawing groups on the photophysical properties. Although BZT derivatives have been extensively examined, to the best of our knowledge, 1 and 2 have not been previously investigated and their collective spectroscopic and electrochemical properties remain unexplored. Bearing this mind, the spectroscopic, spectroelectrochemical, and electrochemical properties of the similar BZT fluorophores are herein presented to evaluate these similar electronic push–pull derivatives as universal fluorescence probes.

Chart 1:

Fluorophores examined.

Experimental section

Results and discussion

To probe the solvent effect on the spectroscopic properties, the electronic absorption and steady-state fluorescence of 1 and 2 (Scheme 1) were measured in aprotic solvents of different polarity. As seen in Table 1, the absorption of 1 is shifted by 10 nm (475 cm⁻¹) as a function of solvent, whereas 2 is shifted by 15 nm (685 cm⁻¹). The larger spectral shift observed for 2 is from its ground state being more perturbed by solvent effects than 1. This implies that 2 has a greater ground state dipole than 1, courtesy of a stronger electronic push–pull effect between the –NO₂ and –NMe₂ termini.

Scheme 1:

Synthetic scheme for the preparation of 1 and 2: a) Cs₂CO₃, TBAB, Pd(PPh₃)₄, toluene/H₂O (3/1), 90 °C.

The effect of solvent polarity on the fluorescence wavelength was also examined. The fluorescence of 1 and 2 was significantly shifted in different solvents. This is evident in Fig. 1, where the emission maximum of 1 is shifted upwards of 175 nm (3710 cm⁻¹), while 2 is shifted by 160 nm (3455 cm⁻¹) in similar solvents. To some extent, the observed shifts are not surprising given the excited state is known to be more perturbed by solvent effects than the ground state. This is a result of the excited state being more polar than the ground state for the π to π* transition. The π to π* transition was confirmed by DFT calculations (vide infra) with the corresponding HOMO and LUMO calculated orbitals seen in Fig. 3. The conjugated electron donating and withdrawing groups can further give rise to an intramolecular charge transfer that is favored in polar solvents [40]. Although the exact nature of the excited state cannot be unequivocally identified, the excited state is significantly perturbed by different solvent polarities. This is further supported by the large Stokes shift. The effect of solvent on the emission color is qualitatively seen in the photographs of Fig. 1c and d. It is evident that the fluorescence of 1 shifts from yellow to green with increasing solvent polarity. Similarly, the emission color of 2 shifts from yellow to blue. It should be noted that the fluorescence of both 1 and 2 was poorly correlated with the Lippert–Mataga relationship (Fig. S3). This implies specific solute–solvent interactions take place in the excited state. These interactions are more pronounced with 2, whose linear regression for the Stokes shift as function of E_T(30) was r² = 0.89. To some extent, this is not surprising, owing to the nitro group that is known is sustain specific NO₂–solvent interactions that are not taking into account in the general Lippert–Mataga equation. An improved linear regression was found for 1 (r² = 0.93), confirming reduced specific solvent–solute interactions involving the CN group in the excited state compared to 2. Although ethanol would be an ideal solvent for examining the hydrogen bonding effect on the fluorescence behavior, both 1 and 2 were insufficiently soluble in alcohols for quantitative measurements.

Fig. 1:

Normalized emission spectra of 1 (a) and 2 (b) measured in dichloromethane (■), DMSO (□), acetonitrile (

), hexane (

), acetone (

), ethyl acetate (

), diethyl ether (

), THF (

), toluene (

) by exciting at the respective maximum absorption. Photographs showing the fluorescence of 1 (c) and 2 (d) in hexane (1), toluene (2), ether (3), dichloromethane (4), ethyl acetate (5), THF (6), acetone (7), acetonitrile (8), and DMSO (9), when excited with a UV lamp at 360 nm.

Of particular interest is the fluorescence quantum yield (Φ_fl) dependence on solvent polarity. Rather than using relative actinometry to determine the Φ_fl, the values were calculated using an integrating sphere. The benefit of this method is that fluorescence references absorbing and emitting in the same region and having roughly the same Φ_fl as the compound of study are not required. This is particularly advantageous for studying the BZT fluorophores whose emission wavelength and intensity varied considerably with solvent (vide supra). The Φ_fl values derived from the integrating sphere method are additionally absolute values. To validate the results, the sphere was calibrated against anthracene in degassed anhydrous ethanol (Φ_fl = 28 %) [28]. As seen in Table 1 and Fig. 1b and c, both 1 and 2 fluoresce appreciably in apolar solvents. In fact, the Φ_fl of both 1 and 2 are similar within experimental error in hexane, toluene, ether, and dichloromethane. The emission of 2 decreased significantly in acetone, acetonitrile, and DMSO, such that it could not be quantified with the instrument. While the same trend was observed for 1, its emission could nonetheless be quantified using the integrating sphere method. Although 1 and 2 are electronically similar, the emission yield of the fluorophore with the nitrile group is less perturbed by solvent effects than with the nitro group.

The change in Φ_fl with solvent polarity (Table 1) is not surprising given the narrowing of the ¹S-¹T energy gap that can occur in polar solvents. This can potentially increase intersystem crossing to the triplet manifold. Transient absorption spectroscopy was therefore used to determine whether a triplet state was formed when exciting 1 and 2. The transient absorption spectra of 1 and 2 were examined in toluene, dichloromethane, and acetonitrile. These solvents were chosen because of the contrasting measured Φ_fl values. A transient was observed only in the case of 1 in acetonitrile having an absorption at 360 and 540 nm. The absorption in the visible region was assigned to the triplet owing to its unimolecular decay (τ = 2.6 μs) and its quenching with oxygen. The absence of detectable signal on the μs time scale suggests a triplet is not formed for 2. Its quenched fluorescence therefore arises from solute-solvent interactions. These occur to a lesser extent for 1, whose fluorescence deactivation occurs by intersystem crossing in acetonitrile.

Two short-lived transients within a 100 ns time window were observed for 1 and 2 by transient absorption spectroscopy. While these were ascribed to the singlet excited state, the interesting feature is their interdependence. For example, the transient at 530 nm was converted into the transient at 655 nm (3600 cm⁻¹ shift) for 1 in dichloromethane. The red-shifted transient observed for 1 in acetonitrile (715 nm; 4880 cm⁻¹), toluene (620 nm; 2740 cm⁻¹), and dichloromethane (675 nm; 4055 cm⁻¹) was consistent with the emission measured by steady-state fluorescence. The two states were further probed by time-resolved fluorescence. The lifetime of the red-emission of 1 in dichloromethane was 7.4 ns, whereas the blue emission was 5.4 ns. The fluorescence lifetime of 1 was consistent in the solvents examined. In contrast, the lifetime of 2 became shorter when progressing from toluene, dichloromethane, THF, acetonitrile, and DMSO (Table S1). This trend was consistent with the decrease in Φ_fl of 2 in the different solvents, confirming the importance of non-radiative deactivation modes involving the nitro group with increasing solvent polarity.

The emission scans of 1 and 2 confirmed the dual-emission of the two fluorophores. This is evident in Fig. 2 where an emission at 480 and 675 nm can be seen for 1 in dichloromethane. The emission intensity was contingent on the excitation wavelength. The isosbestic point at 580 nm confirms the independence of the dual-emission. 2 similarly had a dual-emission with the red-shifted emission being less intense. In contrast, the blue-emission of 1 was essentially quenched in toluene. The origin of the dual-emission cannot be from aggregates, excimers, and bimolecular processes given the ratio of the blue/red emission of 1 and 2 did not change with concentration. In fact, no change was observed when diluting the samples up to 370 fold (Figs. S13 and S16). This provides strong evidence for intramolecular effects. Given the shift was contingent on solvent polarity, the red emission is the intramolecular charge transfer band between the donating amine and accepting nitrile and nitro, respectively, whereas the blue-emission is the global emission.

Fig. 2:

Fluorescence spectra of 1 in dichloromethane by exciting between 380 and 450 nm.

To gain a better understanding of the spectroscopic properties, DFT calculations were done to investigate both the ground and excited states. This level of theory was chosen because it is well-known to accurately afford both the optimized geometry and corresponding properties of π-conjugated molecules [41]. Moreover, the intrinsic errors of the theoretically calculated values are consistent such that the relative values can be accurately calculated. The X-ray data of a similar compound was used as the starting geometry of 1 and 2 to accelerate the geometry optimization calculations [36]. The resulting calculated optimized geometries were in good agreement with the XRD data. Figure 3 shows the Frontier orbitals and the HOMO/LUMO energy levels calculated for 1 in different solvents (Fig. S35 for 2). Little difference in the HOMO-LUMO energy levels was calculated, suggesting these orbitals are only weakly perturbed by the solvent. The calculations provide evidence for the HOMO-1 and LUMO+1 orbitals contributing to the change in spectral properties, albeit only weakly. Time-dependent density functional theory (TD-DFT) was used to calculate the excited energies. The calculated oscillator strengths and excitation energies (Table S2) were used to calculate the major electronic transition in the theoretical absorption spectrum, according to Stephens methods [42]. As seen in Table 1, the calculated absorptions are consistent with the experimental data with the same trend in absorption shift with solvent. The Frontier orbitals provide further insight into the origin of the observed spectral changes. For example, the charge density of the HOMO is concentrated on the donor portion of the fluorophore. Similarly, the charge density of the LUMO is uniquely localized on the acceptor portion of the fluorophore. The electronic transition therefore involves intramolecular charge transfer between the HOMO to the LUMO for both fluorophores. This is consistent with the experimental results.

Fig. 3:

Top: calculated HOMO (

) and LUMO (

) energy levels and Frontier orbitals (right) of 1 in different solvents. Bottom: molecular Frontier orbitals and corresponding energy levels of 1 calculated in hexane.

The change in spectral properties of 1 and 2 was further investigated in the presence of trifluoroacetic acid (TFA). This was to confirm the origin of the red-emission arising from intramolecular charge transfer. TFA was chosen because of its capacity to fully dissociate in a wide range of organic solvents. It is therefore a sufficiently strong acid to protonate the N,N-dimethylamino terminus in the solvents examined. Protonating the amine disrupts the electronic push–pull character of the fluorophore, yet the overall conjugation remains intact. As a result, only the electron withdrawing group can affect the properties. The absorption of 1 at 450 nm was blue-shifted to 370 nm (4805 cm⁻¹) when the fluorophore was protonated with TFA. The same trend was observed for 2. As seen in Fig. 4, the red-emission of 1 is gradually quenched when titrating with TFA. The same behavior was also observed for 2 in the various solvents. The collective solvent and TFA induced spectral shifts provide evidence that the red-shifted emission is from intramolecular charge transfer, while the blue-shifted emission is from the local emission.

Fig. 4:

Change in fluorescence of 1 in toluene with 0 (○), 90 (

), 170 (

), 260 (

), 350 (■), 430(

), and 520 (

) mM TFA.

Given the interest of using the fluorophores in emitting applications and rheological probes in polymer composites, the emission properties of 1 and 2 were further examined in thin films. The fluorophores were drop-cast (as-cast) on to glass substrates from dichloromethane and the resulting films of 2 were non-fluorescent (Table 1). The quenched fluorescence in thin films, to some extent, was not surprising, given the known intermolecular self-quenching deactivation modes that prevail in the solid-state. The fluorophores were subsequently mixed in different ratios with the poly(methyl methacrylate) (PMMA) matrix to reduce intermolecular self-quenching (Fig. 5). The fluorescence yield increased with increasing PMMA/fluorophore ratio and the resulting Φ_fl values measured were comparable to the solution measurements. Consistent high emission yields were also obtained when mixing the fluorophores in a poly(dimethylsiloxane) matrix. In fact, bendable and flexible free-standing substrates ca. 5 cm × 5 cm were obtained. Consistent fluorescence was observed with repeated bending and stretching of the elastomeric film.

Fig. 5:

Photographs of 1 in a PMMA matrix 1:0 (left), 1:10 (middle), and 1:100 (right) on glass substrates irradiated with a UV lamp.

The electrochemistry of the fluorophores was investigated to determine whether they could be reversibly oxidized. This is a desired property for using the fluorophores as functional materials in plastic electronics. As seen in Fig. 6, both 1 and 2 could be reversibly oxidized. The reversible behavior is in part owing to the terminal electron groups that prevent cross-coupling of the oxidized intermediates, according to known mean [43]. Using ferrocene as an internal reference, the oxidation potentials (E^o) measured for 1 and 2 were 875 and 930 mV vs. SCE, respectively. The difference in the oxidation potentials is owing to the different electron withdrawing strengths of the CN and NO₂ groups. As a result of the reduced electron withdrawing capacity of the nitrile group, the donating effect of the terminal amine of 1 is enhanced compared to 2. This results in the less positive oxidation potential observed for 1 relative to 2.

Fig. 6:

Anodic cyclic voltammograms of 1 (

) and 2 (

) measured in anhydrous and deaerated dichloromethane with 0.1 M TBAPF₆ supporting electrolyte and ferrocene as an internal reference.

To further evaluate the usefulness of the fluorophores, especially as electrochromes, their spectroelectrochemistry was evaluated. This was done by gradually applying a positive potential to generate the oxidized species, while examining the change in either the absorption or fluorescence. As seen in Fig. 7a, the absorption of 1 at 455 nm decreases with increasing potential with the concurrent formation of new peaks at 365, 525, 760, and 840 nm. The isosbestic point at 405 nm confirms the presence of only two species, the neutral and oxidized states, and their independence. The resulting visible color change of 1 with electrochemical oxidation was yellow to mauve. The neutral state could be regenerated by applying a slightly negative potential. The original absorption spectrum was obtained after neutralizing the oxidized state. The same spectral trend with applied potential was also observed with 2. For this fluorophore, the color changed from yellow to pale violet with a positive potential, along with the formation of new absorptions at 370, 505, 760, and 835 nm. Similar to 1, a well-defined isosbestic point was observed at 410 nm for 2.

Fig. 7:

(a) Spectroelectrochemistry of 1 with applied potentials from 0 to 1 300 mV vs. Ag/Ag⁺ held for 30 s per potential in dichloromethane with TBAPF₆ as an electrolyte. (b) Fluorescence change of 1 when excited at 390 nm with applied potentials from 0 to 1 300 mV for 30 s followed by −1000 mV.

Given the intrinsic fluorescence of 1 and 2, their spectroelectrochemical fluorescence was also investigated. This was done by exciting the fluorophores at 390 nm and monitoring their change in fluorescence with increasing potential. The resulting fluorescence change of 1 is shown in Fig. 7b. Qualitatively, the emission color changed from purple to blue upon oxidation. Quantitatively, the emission of the intramolecular charge transfer band at 675 nm decreased when the neutral form was oxidized concomitant with an increase at 480 nm. The two neutral and oxidized states are interdependent, as per the isosbestic point at 615 nm in Fig. 7b. 2 similarly underwent fluorescence changes when oxidized. Its emission at 540 nm increased while the emission at 680 nm decreased with increasing potential, resulting in visible color changes from gray-green to green.

Conclusion

Two electronic push–pull benzothiadiazole fluorophores were prepared and the effect of the terminal electron withdrawing groups on the spectroscopic and electrochemical properties was examined. It was found that both fluorophores had similar properties, notably their solvatochromism and fluorescence quantum yields. Interestingly, the intramolecular charge transfer band could be electrochemically modulated. This resulted in a change in emission wavelength and intensity with electrochemical oxidation. The intrinsic intense fluorescence of the fluorophores was preserved when incorporated into polymer matrices. The collective solvatochromism, intense fluorescence in polymer matrices, and spectroelectrofluorescence make the electronic push–pull benzothiadiazole derivatives interesting universal fluorescence probes. Meanwhile, the similar spectroscopy and electrochemistry of the two fluorophores demonstrate that the electron withdrawing –NO₂ group can be readily replaced with –CN, without affecting adversely the properties.