Bisporphyrin 1 was synthesized from 5,15-bis(2-methoxycarbonylethyl)-10- (trimethylsilylpropargyl)-20-(1-methyl-2-imidazolyl)porphyrinatozinc 3[15] and 2,7-diiodo-9,9- bis(3,5,5-trimethylhexyl)fluorene 4[35,36] using a Pd₂(dba)₃/AsPh₃ catalytic system, where deprotection of the TMS group and the coupling reaction were conducted in a one-pot procedure because of the low solubility of the deprotected ethynylimidazolylporphyrin (Scheme 1). Purification by preparative gel permeation chromatography (GPC) with pyridine elution allowed for the isolation of 1 from the crude mixture in 33% yield. Both the MALDI-TOF mass spectral and analytical GPC measurements showed a single species corresponding to the molecular weight of 1. However, broad signals were obtained in the NMR spectrum of 1, even in a coordinating solvent such as pyridine-d₅. Accordingly, the free base 5 was prepared by treatment with HCl to obtain a clearer NMR spectrum (Figure 1). Bisporphyrin 1 could not be obtained when 2,7-dibromofluorene was used as a starting material, due to its low reactivity.

When 1 was dissolved in a nonpolar solvent such as chloroform or 1,1,2,2,-tetrachloroethane after the evaporation of pyridine, the long linear polymer 1P was formed by self-coordination. Polymer 1P changed gradually to cyclic arrays of various sizes in chloroform above 45 °C or in chloroform/methanol (7:3, v/v%) at room temperature. Such cyclic arrays have been reported elsewhere [20]. 1P retains its linear polymer structure in chloroform or in tetrachloroethane without methanol at room temperature sufficiently long for 2PA measurements. The molecular weight of 1P was analyzed by GPC with chloroform elution on a column with an exclusion limit of 5 × 10⁵ Da. Figure 2 shows the GPC chart of 1P monitored at 420 nm by an UV-vis photodiode array. The arrows indicate the peak positions of polystyrene standards. The absorption spectra of 1P at different retention times between 7 and 10 min were identical. The distribution maximum of the elution curve of 1P appeared around Mw 1.5 × 10⁵ Da. The mean molecular weights, M_w and M_n, estimated by comparing the data with those of polystyrene standards [37], were ~2.1 × 10⁵ Da and ~1.6 × 10⁵ Da, respectively, which results in a polydispersity index of M_w/M_n = 1.4. The mean molecular weight of M_n = 1.6 × 10⁵ corresponded to ca. 94 units of 1, and these values were comparable with linear self-assembled bisporphyrin arrays [14].

The UV-vis absorption spectra of 1P in tetrachloroethane (solid line), 1M in tetrachloroethane/pyridine (7:3, v/v%), and 2P in chloroform (dashed line) are shown in Figure 3. The Soret and Q-bands of 1P were observed at around 485 and 670 nm, respectively. These were blue-shifted by 15 and 70 nm compared with those of 2P (500 nm and 740 nm), suggesting a decrease in interaction due to the long distance and torsional angle of 160° between the two porphyrins. However, the absorption intensity of the Q-band in 1P was still strong. On the other hand, the Q-band of unit 1M was blue-shifted to 655 nm and its intensity was considerably decreased because of the disappearance of excitonic interactions through the zinc-imidazolyl complementary coordination.

The 2PA cross sections of 1P and 1M were measured with an open-aperture Z-scan technique at wavelengths from 740 to 920 nm, using a Nd:YAG nanosecond pulse laser system with optical parametric oscillator. The nanosecond measurement system and analysis method were previously reported [14]. In general, the effective cross section value, ^effσ⁽²⁾, measured using nanosecond pulses, is 2–3 orders of magnitude larger than the pure 2PA cross section, σ ⁽²⁾, obtained with femtosecond pulses, because the excited state absorption (ESA) contributes to the data. In our porphyrin systems, the ^effσ ⁽²⁾ values were 30 to 50 times larger than the σ ⁽²⁾ values, and the 2PA spectra measured with nanosecond and femtosecond pulses were similar. Therefore, ^effσ ⁽²⁾ values obtained by nanosecond measurements were a good measure for the 2PA properties. Moreover, in some 2PA applications, such as 2PA-PDT and 3-D optical memory, nanosecond pulses will be more appropriate because of the ease of operation and availability.

Figure 4 shows typical Z-scan traces for 1P in tetrachloroethane (dotted trace) and theoretically fitted curves (solid line) at 890 nm ((a) at a concentration of 7.66 × 10⁻⁵ M with a pulse energy of 1.6 mJ) and 760 nm ((b) at a concentration of 6.87 × 10⁻⁵ M with a pulse energy of 1.45 mJ). The ^effσ ⁽²⁾ values were obtained from the fitted curves and theoretical equations [14,38]. Figure 5 shows 2PA spectra of 1P (square). Two peaks were observed at 760 and 890 nm with large ^effσ ⁽²⁾ values of 3.4 × 10⁵ and 1.1 × 10⁵ GM (per dimer unit), respectively. 2PA spectra of 1M (Figure 5, cross) also showed two peaks at 770 and 890 nm, with values of 9.1 × 10⁴ and 5.6 × 10⁴ GM, respectively. The values obtained for 1M were almost one-third to one-half of those of 1P. As described for the absorption spectra, 1P showed strong one-photon absorption of the Q-band at 670 nm. On the other hand, that of 1M was blue-shifted to 655 nm and the oscillator strength was almost half that of 1P, indicating that the resonance enhancement for 1M, which comes from the one-photon absorption near the two-photon resonance [38,39], would be expected to be poorer than that for 1P. This may clearly explain the difference in the ^effσ ⁽²⁾ values between 1P and 1M, and we can confirm again that self-organization through complementary coordination is effective for the enhancement [13,14]. The shapes of the 2PA spectra of 1P and 1M were similar, but did not correspond to those of the one-photon spectra in the wavelength range of half of 2PA, i.e., 370 nm to 500 nm, suggesting that the final state by 2PA is different from the one-photon excited state [5,39]. The ^effσ ⁽²⁾ values per dimer unit of 1P in the wavelength range between 800 and 860 nm ranged from 3.9 × 10⁴ to 5.4 × 10⁴ GM. These values were similar to those of macrocycles measured by femtosecond pulses [20], indicating negligible ESA contributions in this wavelength range, considering nanosecond measurement.

In the previous report, the linear polymer 2P exhibited a large ^effσ ⁽²⁾ value of 2.0 × 10⁵ GM (per dimer unit) at 890 nm [14], which was twice of that of 1P at the same wavelength. 2P had a strong one-photon absorption of the Q-band at 740 nm that was longer than 1P by 70 nm, suggesting that 2P undergoes a stronger resonance enhancement effect than 1P. However, the shift of the strong Q-band to the shorter wavelength of 670 nm for 1P allows the Z-scan measurements in the wavelength range of 720 to 820 nm, which is prohibited by one-photon absorption in the case of 2P. Then, we could find another strong 2PA band at around 760 nm with the largest ^effσ⁽²⁾ value of 3.4 × 10⁵ GM. The cross sections increased with the shortening of the incident light wavelength below around 800 nm. Similar behavior was observed in simple porphyrins such as tetraphenylporphyrin, Zn octaethylporphyrin, and a self-assembled dimer measured using a femtosecond fluorescence method [39] or nanosecond Z-scan method [16]. This has been explained as a resonance enhancement near the one-photon absorption of the Q-band.

¹H NMR spectra were obtained in CDCl₃ with Me₄Si as the internal standard (δ 0 ppm) and recorded on either a JEOL JNM EX270 or JEOL ECP600. UV-vis spectra were obtained on either a Shimadzu UV-1650PC or UV-3100PC. MALDI-TOF mass spectra were obtained on Perseptive Biosystems Voyager DE-STR and Shimadzu/KRATOS Axima-CFR Kompact MALDI with dithranol (Aldrich, St. Louis, MO, USA) or 2-[3-(4-tert-butylphenyl)-2-methylprop-2-enylidene]-malononitrile (DCTB, Aldrich) as the matrix. Reactions were monitored on silica gel 60 F254 TLC plates (Merck, Darmstadt, Germany). The silica gel utilized for column chromatography was purchased from Kanto Chemical Co. Inc.: Silica Gel 60N (Spherical, Neutral) 60–210 mm and 40–50 mm (Flash). The alumina used for column chromatography was purchased from Merck: Aluminum oxide 90 active basic. Analytical GPC measurements were performed on a HEWLETT PACKARD 1100 SERIES with a JAIGEL 3H-A or 4H-A column for chloroform eluent. The nanosecond Z-scan measurement system and analysis method were previously reported [14].

Synthesis of 1: Under argon atmosphere, 3[15] (95 mg, 132 μmol) and 4[35,36] (35.4 mg, 52.6 μmol) were dissolved in dry THF (45 mL). Then, tetrabutylammonium fluoride in THF (264 μL, 264 μmol) was added. After 30 min, triethylamine (9 mL), Pd₂(dba)₃ (27.2 mg, 26.3 μmol), and AsPh₃ (32.2 mg, 105 μmol) were added. After stirring 4 h, the reaction solvents were evaporated. The residue was dissolved in chloroform and washed with water. The crude residue after concentration was purified by preparative GPC (Tosoh G2500H_HR, pyridine elution) to give 1 (29 mg, 33%). MALDI-TOF mass (dithranol) m/z 1711.66 (M + H)⁺, calcd for C₉₉H₉₈N₁₂O₈Zn₂, 1710.62; UV-vis λ_max (nm, in CHCl₃) 449, 484, 583, 667; fluorescence λ_max (nm, λ_ex = 484 nm, in CHCl₃) 670, 731.

Synthesis of 5: Under an Ar atmosphere, 1 (4.5 mg, 2.63 μmol) in chloroform (30 mL) was added into a mixture of 12 N hydrochloric acid (8 mL) and methanol (40 mL). After stirring for 30 min at room temperature, the reaction solution was diluted with water and extracted with chloroform. The chloroform layer was washed with saturated NaHCO₃ aqueous solution. The crude residue after concentration was purified with silica gel column chromatography (CHCl₃/MeOH = 20/1) (yield: 53%). ¹H NMR (600 MHz, CDCl₃) δ 9.96 (q, 4H, J = 4.8 Hz, β), 9.58 (q, 4H, J = 4.8 Hz, β), 9.44 (d, 4H, J = 4.8 Hz, β), 8.78 (d, 4H, J = 4.8 Hz, β), 8.17(m, 2H, Ph), 8.12 (s, 2H, Ph), 8.07 (d, 2H, J = 7.8 Hz, Ph), 7.70 (d, 1H, J = 1.8 Hz, im), 7.50 (d, 1H, J = 1.8 Hz, im), 5.32 (t, 8H, J = 7.8 Hz, CH₂), 3.77 (s, 12H, J = 7.8 Hz, OMe), 3.53 (t, 8H, J = 7.8 Hz, CH₂), 3.43 (s, 6H, NMe), 2.36–0.81 (m, 38H, alkyl groups in fluorene), −2.29 (s, 4H, NH); MALDI-TOF mass (dithranol), m/z 1587.75 (M + H)⁺, calcd for C₉₉H₁₀₂N₁₂O₈, 1586.79.