computed at the CAM-B3LYP/6-311++G\*\* level of theory.

Similar localized orbitals correspond also to the HOMO-2 and LUMO of molecules **5c** and **5d**. Indeed, the absorption predicted at about 594–596 nm for **5c** and **5d** is due to the HOMO-2→LUMO transition. Likewise, the HOMO-1→LUMO transition of the asymmetric compound **DF90**, which corresponds to the absorption maximum at about 645 nm, is characterized by the same ground-state electron density delocalization on the Indigo part of the molecule. In this last case, the limited spatial separation between these frontier molecular orbitals suggests a consequent limited intramolecular charge separation upon photoexcitation of the dyes. The DFT FMOs energies obtained in DCM for tautomer KK of **5a**, **5b**, **5c**, **5d**, and **DF90** are reported in the Supporting Information (Figure S5).

**Figure 3.** B3LYP/6-31G\*\* ground-state electron density distributions in DCM of KK tautomer of compounds **5a**, **5b**, **5c**, **5d** and **DF90**.

#### *2.2. Synthesis of Dyes*

To prepare the symmetrical dyes **5a**–**d** we used an approach (Scheme 1) similar to that already described for the synthesis of 6,6 -dithienylindigo [38]. Tyrian Purple (**10**) was obtained as a purple powder in 71% yield, using the classical indigo-forming protocol [2]. To increase its solubility and simplify its chemical manipulation and processing, Tyrian Purple was protected using (*t*-Boc)2O and DMAP in DMF solution, and the soluble deep red product **11** was obtained in 92% yield. In order to preserve the required but thermolabile *t*-Boc protection, the introduction of the side groups should occur under very mild and chemoselective conditions. For these reasons we decided to take advantage of the mild conditions usually applied in the Stille-Migita cross-coupling, and thus to react intermediate **11** with stannanes **12a**–**d** (Scheme 1).

**Scheme 1.** Stille-Migita cross-coupling of t-Boc-protected Tyrian Purple (**10**) with thienyl- and triarylamino-stannanes 8a**–**d.

We optimized the reaction conditions using commercially available stannane **12a** and found that Pd2(dba)3 and the electron-rich tri(*o*-tolyl)phosphine was the best combination to generate the active catalytic species able to perform the cross-coupling at 50 ◦C. In this way, deprotection of the *t*-Boc group, which might occur when temperatures higher than 90 ◦C are used [44,45] and the consequent precipitation of unprotected starting material was prevented. Thus, using toluene as solvent and two equivalents of stannane **12a**, symmetric indigo derivative **6a** was obtained in good yield in 5 h. The reaction was then repeated using the more electron-rich thienylstannane **12b**, and two triarylamine-containing stannanes **12c** and **12d**, which were prepared as previously reported. [44,46] In all cases, the corresponding coupling compounds **6b**–**d** were recovered with high yields after purification. Unfortunately, compound **6d** appeared unstable in solution ad could not be fully characterized nor used for further deprotection. However, we were able to identify it by ESI/MS and to record UV/visible spectra. Finally, reaction of **6a**–**c** with TFA occurred smoothly at room temperature to give compounds **5a**–**c**, which were soluble in the most common organic solvents**.**

To prepare dye **DF90** a modification of the above-described synthetic approach was necessary in order to obtain a non-symmetrical molecule. Again we used the *t*-Boc protected Tyrian Purple (**11**) as starting material and decided to install the acceptor moiety first (Scheme 2).

To this end, we prepared stannane **12e** [44] carrying a formyl group, which was essential to later establish the required cyanoacrylic moiety (Scheme 2). Desymmetrization of compound **11** with stannane **12e** needed to be carried out using a large excess (five-fold) of the starting material**,** which, opportunely, could be easily recovered at the end of reaction by precipitation from ethyl acetate/hexane mixture. Evaporation of the solvent gave intermediate **13** which was obtained in 84% yield (based on **12e**) after column chromatography. The second coupling, required to introduce the donor group, was performed in essentially the same conditions, albeit using one equivalent of stannane **12c**: pure *t*-Boc protected aldehyde **14** was thus obtained in 79% yield after purification. Knoevenagel condensation of aldehyde **14** with cyanoacetic acid and piperidine in acetonitrile allowed introduction of the desired cyanoacrylate together with the simultaneous deprotection of *t*-Boc groups, affording dye **DF90** in 55% yield.

**Scheme 2.** Synthesis of dye **DF90**.

#### *2.3. Optical and Electrochemical Properties*

The optical properties of all the new dyes were studied and compared with those of the parent compound indigo (Figure 4). It must be considered, however, that when Indigo was suspended in CHCl3 in the reported conditions (0.17 g/mL corresponding to 6.5 <sup>×</sup> 10−<sup>5</sup> M), it was not possible to obtain a completely clear solution, thus the resulting molar extinction coefficient must be taken only as an approximate value.

**Figure 4.** UV-Vis absorption spectrum of indigo dissolved in CHCl3 (**a**) and corresponding Tauc plot (**b**).

In good agreement with previous literature reports [47], we observed a relatively intense transition at 604 nm, corresponding to an *E0-0* of 1.96 eV, due to the so-called "*H*-chromophore" [8,38,48] corresponding to a cross-conjugated donor-acceptor system held together by intramolecular hydrogen bonds (see Figure 1). Computational studies revealed that the H-chromophore absorption is due to a HOMO-LUMO transition with π-π\* character, corresponding to a net electron transfer from the N–H group (acting as a donor) to the C=O group (acting as an acceptor) [9]. The UV-vis spectra of dyes **5a**–**c** were recorded in CH2Cl2 and EtOH solution. Spectra are reported in Figure 4 and compared with those of the corresponding *t*-Boc protected compounds **6a**–**d**. Due to its low solubility in EtOH, the spectrum of compound **6b** could be recorded only in CH2Cl2. All relevant spectroscopic data have been summarized in Table 2.


**Table 2.** Optical properties of dyes **5a**–**d** and **6a**–**c** compared with those of parent indigo compound.

*<sup>a</sup> E0-0* of the lowest energy transition, estimated by means of the corresponding Tauc plot (see Figure S6). *<sup>b</sup>* Spectrum recorded in CHCl3. *<sup>c</sup>* The UV-Vis absorption spectrum of this compound could not be recorded in EtOH due to insufficient solubility.

The absorption maxima for the lowest energy band of **5a**–**5c** and **DF90** (see below) were well predicted by the results of the computational investigation. Indeed, the differences between DFT and experimental vertical excitation energies (Eexc) were 0.05 eV at most. In absence of the experimental values for **5d**, we computed the geometry, orbital energies, and UV-Vis properties for compound **6d** (see Supporting Information, Figure S4, Tables S4 and S6). The presence of the *t*-Boc group causes a slight deviation from the planarity of the central C–C bond of the indigo moiety (dihedral angle changes from 0◦ to 5◦) which in turns lead to a blue-shift of the lowest energy band.

Significant differences were observed between the *N*-*t*-Boc protected and free N–H species (Figure 5). Considering the *t*-Boc-protected compounds **6a**–**d**, while the first two peaks (marked as a,b) can be assigned to localized π-π\* transitions involving different parts of the molecules, the lower energy band is likely due to a charge transfer (ct) transition between the lateral donor groups and the central acceptor unit: this hypothesis is supported by the fact that such band is most red-shifted and intense in the case of compound **6c**, featuring the strongly electron-donating hexyloxy-TPA side groups. As a consequence of the particular absorption profile of the dyes, the corresponding solutions appeared red to purple in color.

**Figure 5.** UV-Vis absorption spectra of dyes **6a–d** (panels (**a**), (**c**), (**e**), (**g**), respectively) and **5a–c** (panels (**b**), (**d**), (**f**), respectively).

No significant difference was observed when passing from CH2Cl2 to EtOH, with all compounds displaying essentially the same spectra, highlighting also their good solubility in the more polar and protic solvent (except for **6b**). Moving from *N*-Boc protected compounds **6a**–**d** to free *N*–H compounds **5a**–**c** the main change observed in the absorption spectra was the activation of the "*H*-chromophore" transition (indicated as *i*). In analogy with the parent indigo compound, such transition, in CH2Cl2

solution, appeared as an intense peak in the 610–614 nm range, corresponding to *E0-0* values of 1.93–1.95 eV. The original *ct* band already observed for compounds **6a**–**d** was still present in the spectra of compounds **5a**–**c**, but appeared only as a shoulder of the more intense indigo transition. Furthermore, in analogy to its precursor **6c**, derivative **5c** had a relatively strong *ct* band, which together with the *i* transition gave rise to an intense panchromatic absorption in the 400–650 nm range.

In EtOH, unprotected dyes (especially compounds **5a** and **5c**) featured a much less intense and broadened spectrum, with a long tail extending in the near-IR region above 750 nm: this was attributed to their reduced solubility in that solvent, leading to the formation of aggregates and observation of light scattering effects. In addition, for most of the compounds, the lowest energy absorption peak was red-shifted in the more polar solvent, and in all cases, smaller *E0-0* values were recorded (1.82–1.87 eV): this bathochromic shift when the dielectric constant of the solvent is increased is well-known also for the parent indigo compound [47] and has been attributed to increased stability of charged-separated structures (for instance, C+–O−) in the excited state rather than in the ground state [9]. As a result of their light absorption profiles, compounds **5a**–**c** gave bright green to dark green-coloured solutions (Figure 6), which were different from the typical blue colour of indigo, demonstrating that the placement of donor moieties on 6,6 -positions of the main chromophore could significantly alter the optical properties of the resulting substances.

**Figure 6.** Solutions of dyes **5a** (2.3 <sup>×</sup> 10−<sup>4</sup> M), **5b** (3.5 <sup>×</sup> 10−<sup>4</sup> M) and **5c** (7.8 <sup>×</sup> 10−<sup>5</sup> M) in CH2Cl2, in comparison with indigo suspension in CHCl3 (0,17 mg/mL <sup>≈</sup> 6.5 <sup>×</sup> <sup>10</sup>−<sup>4</sup> M).

The fluorescence behavior of the new dyes was complicated by the possible occurrence of several different emissive transitions, as illustrated by the emission spectra obtained for compound **5a** at three different excitation wavelengths (Figure 7, top). After excitation at 333 nm (corresponding to an S0–S3 transition), three different fluorescence peaks were observed at 380, 481 and 647 nm, respectively (the latter was partially covered by a peak at 666 nm due to second-order diffraction of the incident radiation). While the first of them was due to the opposite S3–S0 transition, the other two likely originate from S2–S0 and S1–S0 transitions, demonstrating the possibility of non-radiative decay from the S3 state to the lower excited states of the dye. This was confirmed by the fact that irradiating at 415 nm only the peaks at 481 and 647 nm were observed, while only the latter was visible when irradiating at the wavelength of the indigo transition (610 nm).

**Figure 7.** UV−Vis absorption and fluorescence emission spectra of compounds **5a**–**c** in CH2Cl2 solution. (**5a**): in the emission trace after excitation at 333 nm (red line) the peak visible at 666 nm is due to the second-order diffraction of the incident light. (**5b**): in the emission trace after excitation at 334 nm (dark green line) the sharp peak visible at 668 nm is due to the second-order diffraction of the incident light. (**5c**): in the emission traces after excitation at 292 nm (purple line) and 353 nm (blue line), the sharp peaks visible at 584 nm and 706 nm are due to the second-order diffraction of the incident light.

In the case of compounds **5b** (Figure 7, middle) the situation was slightly different. Irradiating a solution of **5b** at 334 nm (thus populating the S3 state) induced only weak emissions at 411 and 493 nm, respectively, while a much more intense peak was observed at 646 nm, indicating that for this compound non-radiative decay to the S1 state was the preferred mean of energy dissipation when excited with higher energy radiation (also in this case a sharp peak at 668 nm was observed due to second-order diffraction of the incident radiation). This supposition was confirmed by the experiments conducted with irradiation at 445 and 610 nm, for which the lowest energy emission band at 646 was the only one observed.

The opposite behaviour was displayed by compound **5c** (Figure 7, bottom). In this instance, irradiation at 292 and 353 nm (whose absorption peaks should correspond to localized π→π\* transitions within the triarylamine moiety) caused a fluorescence of moderate intensity centered at 474 nm, while only a very weak peak was seen at 645 nm. When irradiating at 610 nm, the fluorescence was only barely detectable, as shown by the very noisy normalized spectrum in Figure 7 indicating the prevalence of non-radiative decay from the lowest excited state S1 or the occurrence of extensive reabsorption by the dye.

Finally, the spectra of the unsymmetrical D-A-π-A dye **DF90** (Figure 8) were registered. In solution, the dye exhibited absorption spectra similar to those of the symmetrical compounds, with a significantly red-shifted low energy transition (633–635 nm in CH2Cl2 and EtOH, respectively), whose molar extinction coefficient was however only moderate (0.65–0.70 <sup>×</sup> <sup>10</sup><sup>4</sup> <sup>M</sup>−<sup>1</sup> cm<sup>−</sup>1). The spectrum in CH2Cl2 also presented a shoulder at longer wavelengths relative to the indigo H-chromophore transition, which was attributed to the low solubility of **DF90** with the consequent formation of J (head to tail)-aggregates [49]. Accordingly, Tauc plots for both spectra resulted in estimated E0-0 values of 1.65–1.79 eV, which were smaller than those calculated for indigo and its symmetrical derivatives. When adsorbed on TiO2, **DF90** gave a very broad UV-Vis spectrum with the main peaks at 411 nm (shoulder at 495 nm) and 612 nm, and onset around 720–730 nm. The blue-shift of the spectrum compared to the one in solution could be due both to the formation of H (parallel)-aggregates as well as the deprotonation of the carboxylic function upon anchoring onto the semiconductor [50,51].

**Figure 8.** UV−vis absorption spectra of dye **DF90** in CH2Cl2 and EtOH solution (**a**), and adsorbed on TiO2 (**b**); Tauc plots corresponding to the absorption spectra in solution (**c**); fluorescence emission spectra in CH2Cl2 solution after excitation at different wavelengths (**d**); in the latter scheme, a peak at 582 nm deriving from second-order diffraction of incident light was deleted for clarity (red trace).

Due to the absorption minimum centered at about 550 nm, also in the case of **DF90** the resulting solutions as well as the semiconductor surface assumed a bright green coloration.

The emission behaviour of **DF90** in CH2Cl2 was qualitatively similar to that already observed for the symmetrical compound **5c**, with relatively strong fluorescence peaks corresponding to the S3-S0 and S2-S0 transitions, while the emission peak at approx. 650 nm was very weak (or even visible only as a shoulder of the more intense peak at 530 nm), perhaps due to extensive reabsorption by the wide absorption band between 600 and 750 nm. This is not surprising considering that both **5c** and **DF90** share the same donor group.

Finally, the ground-state oxidation potential (ES+/S) of **DF90** was measured by means of cyclic voltammetry (CV), which was carried out in THF and is reported in Figure 9.

**Figure 9.** Cyclic Voltammetry plot relative to compound **DF90** in CH2Cl2 solution.

The observed curve indicated a reversible oxidation process and the observed potential (0.89 V vs. Ag/AgCl/satd. KCl, corresponding to 0.69 V vs. NHE) was more positive than the redox potential of the iodide/triiodide couple (0.35 V vs NHE), ref. [52] suggesting that regeneration of the sensitizer during operation of a solar cell was possible.

After a small current drop following the first cycle (perhaps due to the consumption of the dye physisorbed on the glassy carbon surface), the current/voltage curve remained practically identical in the following two cycles, indicating that the dye was sufficiently stable upon repeated oxidation/reduction processes. The ES+/<sup>S</sup> and E0-0 values in the same solvent were then used to calculate the excited state oxidation potential (ES+/S\*) by means of the equation ES+/S\* = ES+/<sup>S</sup> − E0-0. ES+/S\* was found to be around −0.96 V vs. NHE, thereby more negative than the conduction band edge of the semiconductor (−0.5 V vs NHE) [53] and therefore appropriate to allow electron injection from the excited state dye to titania.
