From 2,5-Diformyl-1,4-dihydropyrrolo[3,2-b]pyrroles to Quadrupolar, Centrosymmetric Two-Photon-Absorbing A–D–A Dyes

An original approach has been developed for the insertion of formyl substituents at positions 2 and 5 of 1,4-dihydropyrrolo[3,2-b]pyrroles by conversion of thiazol-2-yl substituents. The synthetic utility of these formyl groups was investigated, and a series of centrosymmetric A−π–D−π–A frameworks were constructed. The two-photon absorption of the quadrupolar pyrrolo[3,2-b]pyrrole possessing two dicyanovinylidene flanking groups is attributed to an S0 → (S1) → S4 transition which has a large TPA cross-section (1300 GM) for a molecule of this size.


General information
All chemicals were used as received, unless otherwise noted. All reported NMR spectra were recorded on a 500 MHz spectrometer unless otherwise noted. Chemical shifts (δ; ppm) for 1 H and 13 C NMR were determined with TMS as the internal reference. J values are given in Hz.
Absorption and fluorescence spectra were recorded in cyclohexane, toluene, tetrahydrofuran, dichloromethane and acetonitrile. Mass spectra were obtained with an EI ion source and EBE double focusing geometry mass analyzer or from a spectrometer equipped with an electro-spray ion source with Q-TOF type mass analyzer.

Two-photon absorption spectral measurements.
Two-photon absorption (TPA) spectra were measured by the open-aperture Z-scan method 1 with a femtosecond optical parametric amplifier (Spectra-Physics TOPAS Prime) as the wavelength-tunable light source. The setup for the measurements and the analysis are only described briefly here, because they have been previously reported in detail. 2 where 0 is the on-axis peak intensity of the incident pulse. For the measurements performed using a fixed incident power (i.e., a fixed 0 ), was calculated by assuming the proportionality relation of eq. S2. Practically, artifacts or unwanted processes may cause apparent deviation from proportionality, so we also recorded the incident-power dependence (i.e., by changing 0 ) S3 at some wavelengths to check this relationship. In this case, was determined from the slope of the plot of 0 against 0 . The data points obtained by this method were shown with error bars in Fig. 2 in the main text. Finally, the TPA cross section (2) was obtained from by using the convention (2) = ℎ / , where hν is the photon energy of the excitation pulse and is the number density of the sample molecule calculated from the concentration.
The dyes were dissolved in spectroscopic grade dichloromethane (Fuji Film-Wako). Solutions were then held in 2-mm cuvettes whose path length is short enough against (7-10 mm) to satisfy the thin sample condition for eq. S1. To avoid unwanted effects such as interference by photoproducts etc., the samples were stirred by a micro rotor put in the cell during the measurements. No significant degradation was observed for the samples after Z-scan measurements, which is confirmed by UV-vis measurements. For calibration of day-by-day fluctuation of the measured values, inhouse standard materials were measured at the same time with the samples (MPPBT in dimethyl sulfoxide 3,4 ).
The open aperture z-scan traces at 650 nm or shorter for 7 showed saturable absorption (SA) as shown in Fig. S4. These z-scan traces were analyzed by phenomenological modeling of SA as where is saturation intensity and ( ) = 0 /(1 + 2 ). 5

Quantum chemical calculations.
Molecular structures were optimized at the CAM-B3LYP/6-31+G(d) level of theory. The appropriateness of the obtained structures were checked by the absence of negative frequency in frequency calculations. The optimized structures are shown in Fig. S9. Transition dipole moments and transition energies, and permanent dipole moments were calculated for the lowest 20 excited states using the Tamm-Dancoff Approximation (TDA) 6 with the same level of theory (CAM-B3LYP/6-31+G(d)). All calculations were performed in the Gaussian program suit. 7 Solvent effects were considered by the polarizable continuum model (PCM) for dichloromethane for all calculations In this model, the bulk effect induced by a reaction field is considered. The solvent is modeled as a continuous medium with a dielectric constant, and the solute molecule is placed inside a cavity in the media. The reaction field is formed by polarization of the solvent medium surrounding the cavity, so it is also applied to the solute molecule in the cavity, in addition to the external field of incident light. The shape of the cavity is overlapping S4 spheres in PCM in contrast to the single sphere in the Onsager model. 8 Calculated features of the lowest 10 excited states of 6-8 are listed in Tables S1-S3, respectively.
For postprocessing of the TPA spectrum simulation, we used a homemade program on Igor Pro (Wavemetrics Inc.) based on reported procedures. 9 The spectral simulation by this method gives relatively good agreement of the calculated TPA cross section for the molecules with short πconjugation (containing a few double bonds) but overestimates for longer π-conjugation by one order of magnitude of more. A rigorous explanation of the overestimation is not easy, but it often originates from the overestimation of transition dipole moments. The overestimation was enhanced for the higher-order (multiphoton) absorption process because the fourth power of transition dipole moments is necessary for TPA while only the square is needed for linear optical properties (OPA). The relaxation constants of OPA and TPA were set to 0.07 eV so that the OPA peak in the simulated spectrum matched to the 0-0 peak of the experimental one. The choice of the relaxation constants also directly affects the magnitude of the calculated TPA cross section because the height of the peak is inversely proportional to the square of the relaxation constant.
Therefore, the comparison in absolute value to experiential results is very difficult. We would like to emphasize that the simulation results are useful for comparison of the relative shape of the spectra and clarifying the nature of the transition (allowed/forbidden).

Synthesis of compound 6
A dried Schlenk flask was charged with 5 (41 mg, 120 μmol) and an ACN/THF (3 mL/1.5 mL)  Figure S1. (left) Open-aperture Z-scan traces of 6 in dichloromethane (0.43 mM) measured at 651 nm at different incident powers (symbols) with theoretical fits (grey curves). (right) The corresponding plot of the two-photon absorbance q 0 obtained from the curve fits with eq. S1 against the incident power in the left panel.        Table S4. All data from Table S4. Unity in the horizontal scale means 10-folds difference.