Towards improved halogenated BODIPY photosensitizers: clues on structural designs and heavy atom substitution patterns

Abstract

The singlet oxygen (¹O₂) production quantum yield (Φ_Δ) of 14 halogenated BODIPY dyes has been determined (0.01 < Φ_Δ < 0.99). ¹O₂ production and photostability have been evaluated considering the BODIPY structure, the substitution pattern, and the number and type of heavy atoms and quenching rate constants of ¹O₂ by the sensitizer. In view of the experimental results and principal component analysis (PCA), guidelines for an improved design of efficient and photostable halo-BODIPY sensitizers are proposed.

---

BODIPY dyes have emerged as relevant fluorophores and ¹O₂ photosensitizers for different applications.^1–3 The inclusion of heavy atoms such as Br or I in the BODIPY core is a common strategy to enhance the excited state (S₁ → T₁) intersystem crossing. This modification induces fluorescence attenuation and efficient energy transfer to the ground state molecular oxygen, leading to singlet oxygen production.⁴ However, photobleaching of BODIPYs is often attributed to ¹O₂. Moreover, the influence of heavy atoms on ¹O₂ production by BODIPY dyes has been barely studied. An increment in Φ_Δ values has been reported for polybrominated BODIPYs as the number of Br atoms is increased.⁵ On the other hand, the inclusion of I atoms in positions 3 and/or 5 of polyiodinated derivatives does not produce a significant increment in Φ_Δ.⁶ Therefore, the structural design of halogenated BODIPYs combining efficient ¹O₂ production and high photostability is still under debate.

In order to shed light on this topic, a series of polyhalogenated BODIPY dyes ranging from disubstituted to hexasubstituted derivatives has been synthesized (Fig. 1 and Table 1, ESI†) and characterized concerning ¹O₂ production (Φ_Δ) and quenching (k^Sens_qΔ) in acetonitrile. Their photostability has also been evaluated in terms of initial photodegradation rates (r) and photodegradation quantum yields (Φ_pd), and PCA has been used to find out potential correlations among the parameters. The general structures of the BODIPY dyes studied in this work are polyhalogenated 2-ethyl-4,4-difluoro-1,3-dimethyl-4-bora-3a,4a-diaza-s-indacenes (trialkylside-BODIPYs 1–5) or 4,4- difluoro-8-(4-tolyl)-4-bora-3a,4a-diaza-s-indacenes (8-tolyl-BODIPYs 6–14), which represent two common structural frames often studied in the BODIPY literature. The triplet state lifetimes of these BODIPY structures are usually above 20 μs,^2,3b,5,6 and for long-lived triplets and [O₂] in the mM range (typical of air-equilibrated organic solvents such as acetonitrile), the excited triplet states are quantitatively quenched by molecular oxygen and Φ_Δ generally becomes independent of [O₂].⁷

Fig. 1 General chemical structure of BODIPY sensitizers.

Table 1 Substitution pattern of the BODIPYs studied in this work

	R1	R2	R3	R5	R6	R7	R8
1	Br	Br	H	Me	Et	Me	H
2	H	Br	Br	Me	Et	Me	H
3	I	I	H	Me	Et	Me	H
4	H	I	I	Me	Et	Me	H
5	I	I	I	Me	Et	Me	H
6	H	Br	H	H	Br	H	4-Tolyl
7	H	H	Br	Br	H	H	4-Tolyl
8	H	Br	Br	Br	Br	H	4-Tolyl
9	Br	Br	Br	Br	Br	Br	4-Tolyl
10	Br	Br	CH(CO2Et)2	CH(CO2Et)2	Br	Br	4-Tolyl
11	H	I	CH(CO2Et)2	I	I	H	4-Tolyl
12	H	I	NH(CH2)2CO2H	I	I	H	4-Tolyl
13	H	I	NH-p-(C6H4)OMe	I	I	H	4-Tolyl
14	H	I	I	I	I	H	4-Tolyl

---

For Φ_Δ determination (Table 2), the ¹O₂ phosphorescence decay traces at 1270 nm after laser pulse excitation (532 nm) were recorded at different laser powers (ESI†). Rose bengal was used as the reference sensitizer (Φ_Δ = 0.71).⁶ The quenching rate constants of ¹O₂ deactivation by the BODIPYs (k^Sens_qΔ) were determined by Stern–Volmer analysis from the ¹O₂ emission lifetimes (τ_Δ) measured in the photosensitization experiments.

Table 2 Photophysical and photochemical parameters in air-equilibrated acetonitrile of the BODIPY dyes studied in this work

	λ ^maxabs ^a/nm	ε /M^−1 cm^−1	Φ ^appΔ	τ Δ /μs	k ^SensqΔ/10^8 M^−1 s^−1	Φ Δ	r /10^−8 M s^−1	Φ pd /10^−6
a Absorption maximum in the visible region. Uncertainty ±2 nm from duplicate samples. b Uncertainty ±10%. c Apparent singlet oxygen production quantum yield, uncorrected for ^1O2 quenching by the sensitizer. d Singlet oxygen phosphorescence lifetimes measured for BODIPY solutions with an optical density of 0.1 at 532 nm (dye concentrations in the 10^−4–10^−6 M range). τΔ values reported in the literature for photosensitizers with lower k^SensqΔ than the BODIPYs studied in this work are 83.7 μs^8 and 81.4 μs^9 for 1H-phenalen-1-one and [Ru(dip)3]Cl2, respectively, in acetonitrile. e Singlet oxygen production quantum yield corrected for ^1O2 quenching by the photosensitizer. f Uncertainty ±10%. g Uncertainty ±15%. h Φ Δ = 0.87 ± 6%.^6 i Not determined.
1	488	31 300	0.83 ± 0.03	76.9 ± 3.3	0.3 ± 0.1	0.90 ± 0.04	1.6	5.8
2	509	33 100	0.55 ± 0.05	76.8 ± 2.3	1.0 ± 0.5	0.60 ± 0.05	2.4	8.0
3	503	24 400	0.90 ± 0.06	75.6 ± 2.2	1.0 ± 0.4	0.99 ± 0.06	2.6	8.5
4	521	28 800	0.55 ± 0.06	76.3 ± 1.1	2.6 ± 0.4	0.60 ± 0.06	1.3	—^i
5	525	33 900	0.56 ± 0.04	77.5 ± 2.1	2.5 ± 0.5	0.60 ± 0.04	—^i	—^i
6	531	46 200	0.54 ± 0.11	77.9 ± 1.2	4.3 ± 0.9	0.57 ± 0.11	1.4	1.7
7	515	75 700	0.23 ± 0.02	72.7 ± 3.6	1.5 ± 0.6	0.26 ± 0.02	1.5	2.0
8	546	57 300	0.52 ± 0.07	74.1 ± 2.1	4.4 ± 0.8	0.58 ± 0.07	4.2	5.8
9	544	70 500	0.45 ± 0.06	80.2 ± 0.9	1.7 ± 0.3	0.47 ± 0.06	1.7	—^i
10	534	23 200	0.15 ± 0.02	79.4 ± 3.4	1.3 ± 0.4	0.16 ± 0.02	9.5	—^i
11	559	39 200	0.72 ± 0.09	73.8 ± 1.1	2.6 ± 0.5	0.81 ± 0.09	9.3	12.1
12	497	30 800	0.43 ± 0.04	46.8 ± 0.8	0.8 ± 0.1	0.77 ± 0.05	9.5	—^i
13	518	14 800	0.011 ± 0.001	66.8 ± 3.4	4.7 ± 0.5	0.014 ± 0.001	—^i	—^i
14	573	43 800	0.85 ± 0.04^h	75.4 ± 1.1	1.7 ± 0.5	0.94 ± 0.04	4.0	6.1

---

Since the k^Sens_qΔ values obtained for the BODIPY dyes under study are rather high (10⁷–10⁸ M⁻¹ s⁻¹, compared to ∼10⁴ M⁻¹ s⁻¹ typical of reference ¹O₂ photosensitizers such as phenalenone or rose bengal),^7b the apparent ¹O₂ production quantum yields (Φ^app_Δ) calculated from the phosphorescence decay traces were corrected for the ¹O₂ quenching by the sensitizer itself by using the following equation: Φ^app_Δ/Φ_Δ = k_d/(k_d + k^Sens_qΔ × [Sens]), where k_d is the rate constant of ¹O₂ deactivation by the solvent (1/τ_Δ)^8,9 and [Sens] is the concentration of the BODIPY sensitizer.¹⁰ It has to be noted that this important correction, especially when ¹O₂ is deactivated by the solvent and by the sensitizer in similar timescales, is often neglected in the calculation of Φ_Δ values. In this way, more accurate Φ_Δ values could be determined by taking into account k^Sens_qΔ and the dye concentration (Table 2). According to these results, some general trends in ¹O₂ production depending on the halogen and the substitution pattern could be observed. Firstly, iodinated derivatives tend to show higher Φ_Δ values when compared with their brominated counterparts, as it is evidenced by Φ_Δ of the 1 and 3 or 8 and 14 pairs of trialkylside-BODIPYs or 8-tolyl-BODIPYs, respectively. The higher spin–orbit coupling of iodine vs. bromine may well account for this effect.¹¹ Secondly, when the trialkylside-BODIPY family is considered, a significant role concerning ¹O₂ production ability is played by the presence of halo atoms at position 1 (beneficial) vs. 3 (detrimental). In particular, substitution at position 1 tends to increase Φ_Δ, while substitution at position 3 promotes a substantial decrease of Φ_Δ (cf. Φ_Δ of compounds with Br or I atoms at positions 1 and 2, Φ_Δ(1) = 0.90 and Φ_Δ(3) = 0.99, vs. positions 2 and 3, Φ_Δ(2) = 0.60 and Φ_Δ(4) = 0.60). The predominant detrimental effect on Φ_Δ due to substitution in position 3 can also be observed in the triiodinated compound 5 (Φ_Δ(5) = 0.60, despite its potentially advantageous substitution in position 1 as well, which suggests a major effect of substitution in position 3 vs. 1). Regarding the substitution pattern by bromine in the symmetric 8-tolyl-BODIPYs 6–10, although the presence of Br atoms in positions 3 and 5 seems to support the previous statement in the dibromo derivatives (Φ_Δ(6) = 0.57 and Φ_Δ(7) = 0.26), this trend is not followed in the case of the tetrabromo or hexabromo compounds 8, 9 and 10, which also shows that an increased number of heavy atoms does not necessarily increase Φ_Δ (Φ_Δ(8) = 0.58, Φ_Δ(9) = 0.47 and Φ_Δ(10) = 0.16). Concerning the asymmetric triiodo 8-tolyl-BODIPYs 11–13, all of them show lower Φ_Δ values than their nonalkylated equivalent (R₂, R₅, R₆ = I, R₃ = H, Φ_Δ = 0.86)⁶ or their symmetric tetraiodo counterpart (Φ_Δ(14) = 0.94), and suggest that the type of nonhalogenated substituent can induce strong changes in Φ_Δ when it is alkyl or (amino)alkyl/aryl (Φ_Δ(11) = 0.81 and Φ_Δ(12) = 0.77, while Φ_Δ(13) = 0.014). Overall, when substitution of halo atoms in position 3 by alkyl or N-alkyl/aryl is considered, lower Φ_Δ values are always observed (e.g., Φ_Δ(9) > Φ_Δ(10); Φ_Δ(14) > Φ_Δ(11), Φ_Δ(12) and Φ_Δ(13)). It should be noted that the inclusion of nonhalogenated substituents in the BODIPY structure is important for their functionalization with other subunits or biomolecules, or for their immobilization onto solid supports, in order to develop different photosensitization applications.^1–3

The photodegradation process of a series of BODIPYs studied in this work was monitored by following the optical density changes at λ^max_abs with the irradiation time under green light illumination. However, when the different experimental procedures reported in the literature about the photodegradation of BODIPY dyes are examined,^4a,12–19 a common methodology for the determination of a specific experimental parameter quantitatively related to dye photostability is not easy to find. Among the different approaches considered, the comparison with other well-known photosensitizers,^4a the remaining absorbance,^12–19 the determination of kinetic half-degradation times,¹⁷ and the calculation of the photodegradation quantum yield^18,19 are the most relevant methods. In our case, since the irradiation experiments revealed that most of the studied dyes tend to show an asymptotic behaviour of the remaining optical density after 3 h of illumination, the initial degradation rates (r) and the photodegradation quantum yields (Φ_pd) were calculated from the evolution of the optical density after 30 minutes of irradiation (Fig. 2 and Table 2, ESI†).‡

Fig. 2 Photodegradation plot (normalized O.D. at λ^max_absvs. irradiation time) of the BODIPY sensitizers under green light in air-equilibrated acetonitrile. Compound 1 (●), 2 (○), 3 (▼), 4 (△), 6 (■), 7 (□), 8 (◆), 9 (⋄), 10 (▲), 11 (▽), 12 ([hexagon filled, point down]) and 14 ([hexagon open, flat-side down]).

Φ _Δ should correlate with r and with Φ_pd if the photodegradation of the studied BODIPYs were a bimolecular photooxidation process, as their relatively high k^Sens_qΔ values suggest,§ however, the trialkylside-BODIPYs, with moderate to high Φ_Δ values, seem to suffer much less from photodegradation when compared with the 8-tolyl-BODIPYs (Fig. 2). This result agrees well with the lower r values determined in the case of the trialkylside-BODIPYs 1–4 ((1.3–2.6) × 10⁻⁸ M s⁻¹ range, Table 2), while the 8-tolyl-BODIPYs 8, 11, 12 and 14, also with moderate to high Φ_Δ values (0.6 < Φ_Δ < 1), display higher photodegradation rates ((4.0–9.5) × 10⁻⁸ M s⁻¹ range, Table 2). The weak Φ_Δ–r correlation is also evidenced by 10 (low Φ_Δ and high r) when compared with 8 or 9. From the r values, alkyl/N-alkyl substitution in 10–12 seems to be detrimental concerning photostability if these BODIPYs are compared with 8, 9, or 14. When the series of polybrominated symmetric 8-tolyl-BODIPYs (except compound 10) is considered, their photodegradation rates tend to be low despite their moderate Φ_Δ values. Interestingly, dibromo 8-tolyl-BODIPYs 6 and 7, and hexabromo derivative 9 with moderate to low Φ_Δ, display similar r values to the dihalogenated trialkylside-BODIPYs that are more efficient singlet oxygen photosensitizers. These facts point out that factors such as the hydrocarbon structural frame of the BODIPY, and not Φ_Δ nor the total number and position of the halogens, are influencing the photostability of the BODIPYs. When the amount of photons absorbed by the dye is taken into account and Φ_pd is discussed, a weak correlation between Φ_pd and Φ_Δ is observed, and the influence of other structural features appears to be of little relevance as well.

Therefore, since different structural and photophysical factors might influence the singlet oxygen production and/or photostability of BODIPY dyes, a principal component analysis (PCA) approach was considered in order to qualitatively elucidate, from a statistical point of view, the strongest possible correlations, or lack of correlations, between all the potential variables. The set of prospective variables or parameters used to perform the different PCAs was composed by Φ_Δ, k^Sens_qΔ, Φ_pd, r, the relative positions substituted by halogen (1, 2 and 3), and the type and number of halogens (details on the PCA methodology are summarized in the ESI†).

Among all the PCAs carried out, some of them supported the existence of certain positive or negative correlations, while other PCAs did not provide significant information. The most relevant results are discussed below. The PCA displayed in Fig. 3, describing 61% of the variance, shows the existence of a strong correlation between Φ_Δ and the relative position 1 of the BODIPYs under study (since those more related variables display closer vectors between them). Conversely, an opposite link between Φ_Δ and position 3, and Φ_Δ and k^Sens_qΔ is detected. Interestingly, the data points of compounds 1 and 3 (belonging to the 1,2-dihalo pattern in trialkylside-BODIPYs) are very close to the Φ_Δ vector, supporting the existence of this strong correlation. On the other hand, Fig. 4, accounting for 81% of the variance, shows a lack of correlation between Φ_Δ and the number of halogen atoms, and the influence (in favour of iodine) of the type of halogen atom. Finally, regarding the analysis of the factors that could influence the photostability of the BODIPYs under study, Fig. 5, where 85% of the variance is justified, shows the inverse relationship between the quantum yields (Φ_Δ mainly, and Φ_pd) and k^Sens_qΔ, and only a certain connection between Φ_Δ and Φ_pd, while Φ_Δ displays a weak link with r. From these results, regarding the photostability of the BODIPYs studied in this work, the bimolecular reaction with the photogenerated singlet oxygen does not seem to be the determining factor concerning the photodegradation of these dyes, and it is likely that unimolecular photoprocesses (e.g., a C–X bond fission)²⁰ could also have a strong influence.

Fig. 3 PCA showing the correlations between Φ_Δ, k^Sens_qΔ and the relative positions 1, 2 and 3 of halogen atoms in the BODIPY dyes studied in this work.

Fig. 4 PCA showing the correlations between Φ_Δ and the type and number of heavy atoms in the BODIPY dyes studied in this work.

Fig. 5 PCA showing the correlations between Φ_Δ, Φ_pd, r, and k^Sens_qΔ in the BODIPY dyes studied in this work.

In summary, for the series of polyhalogenated BODIPYs presented in this work (trialkylside- and 8-tolyl-BODIPYs), a relevant influence of the substitution pattern by halogens on Φ_Δ has been observed. In general, the presence of halogens (particularly iodine) in the relative position 1/7 is beneficial for efficient ¹O₂ production, while halogenation of the relative position 3 is detrimental. Moreover, the presence of (N)-alkyl/aryl substituents in the relative position 3/5 tends to decrease ¹O₂ production in comparison with their halogenated counterparts. On the other hand, a high number of halogen atoms in the BODIPY structure are not necessarily linked to higher Φ_Δ values. Regarding the photostability, trialkylside-BODIPYs are, in general, more photostable than 8-tolyl-BODIPYs, and the photodegradation process seems to be more related to intramolecular pathways rather than to the bimolecular reaction with photogenerated ¹O₂. With this facts in mind, an improved design of efficient and photostable polyhalogenated BODIPY photosensitizers is possible via the appropriate selection of the BODIPY structural frame and the introduction of the halogens in suitable positions of the BODIPY structure.

Financial support from the MINECO, Spain, is acknowledged (MAT-2014-51937-C3-2-P and MAT-2015-68837-REDT (to M. J. O.) and CTQ2014-52045-R (to D. G.-F.)). The work of A. J. S.-A. was funded by UCM (predoctoral grant ref. CT4/14).

Notes and references

1. N. Boens, V. Leen and W. Dehaen, Chem. Soc. Rev., 2012, 41, 1130–1172.
2. A. Kamkaew, S. H. Lim, H. B. Lee, L. V. Kiew, L. Y. Chung and K. Burgess, Chem. Soc. Rev., 2013, 42, 77–88.
3. (a) T. Kowada, H. Maeda and K. Kikuchi, Chem. Soc. Rev., 2015, 44, 4953–4972; (b) J. Zhao, K. Xu, W. Yang, Z. Wang and F. Zhong, Chem. Soc. Rev., 2015, 44, 8904–8939.
4. (a) T. Yogo, Y. Urano, Y. Ishitsuka, F. Maniwa and T. Nagano, J. Am. Chem. Soc., 2005, 127, 12162–12163; (b) Y. C. Lai, S. Y. Su and C. C. Chang, ACS Appl. Mater. Interfaces, 2013, 5, 12935–12943.
5. X. F. Zhand and X. Yang, J. Phys. Chem. B, 2013, 117, 5533–5539.
6. M. J. Ortiz, A. R. Agarrabeitia, G. Duran-Sampedro, J. Bañuelos-Prieto, T. Arbeloa-Lopez, W. A. Massad, H. A. Montejano, N. A. García and I. Lopez-Arbeloa, Tetrahedron, 2012, 68, 1153–1162.
7. (a) F. Wilkinson, W. P. Helman and A. B. Ross, J. Phys. Chem. Ref. Data, 1993, 22, 113–263; (b) D. García-Fresnadillo and S. Lacombe, in Singlet Oxygen. Applications in Biosciences and Nanosciences, ed. S. Nonell and C. Flors, Royal Society of Chemistry, Cambridge, 2016, vol. 1, ch. 6, pp. 105–143.
8. O. Shimizu, J. Watanabe, K. Imakubo and S. Naito, Chem. Lett., 1999, 67–68.
9. A. J. Sánchez-Arroyo, Z. D. Pardo, F. Moreno-Jiménez, A. Herrera, N. Martín and D. García-Fresnadillo, J. Org. Chem., 2015, 80, 10575–10584.
10. (a) D. Garcia-Fresnadillo, Y. Georgiadou, G. Orellana, A. M. Braun and E. Oliveros, Helv. Chim. Acta, 1996, 79, 1222–1238; (b) I. Gutiérrez, S. G. Bertolotti, M. A. Biasutti, A. T. Soltermann and N. A. García, Can. J. Chem., 1997, 75, 423–428.
11. N. Epelde-Elezcano, V. Martinez Martinez, E. Peña-Cabrera, C. F. A. Gómez-Durán, I. Lopez Arbeloa and S. Lacombe, RSC Adv., 2016, 6, 41991–41998.
12. C. Jiao, K. W. Huang and J. Wu, Org. Lett., 2011, 13, 632–635.
13. L. Zeng, C. Jiao, X. Huang, K. W. Huang, W. S. Chin and J. Wu, Org. Lett., 2011, 13, 6026–6029.
14. D. Wang, J. Fan, X. Gao, B. Wang, S. Sun and X. Peng, J. Org. Chem., 2009, 74, 7675–7683.
15. T. Terai, K. Hanaoka, Y. Urano, T. Mineno and T. Nagano, Chem. Commun., 2011, 47, 10055–10057.
16. Y. C. Lai, S. Y. Su and C. C. Chang, ACS Appl. Mater. Interfaces, 2013, 5, 12935–12943.
17. S. Banfi, G. Nasini, S. Zaza and E. Caruso, Tetrahedron, 2013, 69, 4845–4856.
18. G. Vives, C. Giansante, R. Bofinger, G. Raffy, A. del Guerzo, B. Kauffmann, P. Batat, G. Jonusauaskas and N. D. McClenaghan, Chem. Commun., 2011, 47, 10425–10427.
19. K. K. Jagtap, N. Shivran, S. Mula, D. B. Naik, S. K. Sarkar, T. Mukherjee, D. K. Maity and A. K. Ray, Chem. – Eur. J., 2013, 19, 702–708.
20. E. Palao, T. Slanina, L. Muchová, T. Šolomek, L. Vítek and P. Klán, J. Am. Chem. Soc., 2016, 138, 126–133.

---

Footnotes

† Electronic supplementary information (ESI) available: Synthesis, characterization data and NMR spectra of the BODIPY dyes, UV-VIS absorption spectra, determination of Φ^app_Δ, k^Sens_qΔ, r and Φ_pd, chemical actinometry and principal component analyses. See DOI: 10.1039/c6cp06448e

‡ Dye samples (3 mL, air-equilibrated acetonitrile) were irradiated in a quartz cell of 1 cm optical path length using a green LED lamp (λ^max_em = 515 nm and FWHM = 33 nm). The incident photon flux, q_p = (2.2 ± 0.2) × 10⁻⁷ einstein s⁻¹, was determined by Reinecke's salt actinometry. For the calculation of photodegradation quantum yields, since the amount of photons absorbed by the dye during the irradiation process progressively decreased due to photobleaching, the total number of absorbed photons was determined by considering the time-dependent absorbance (ESI†).

§ It has to be noted that k^Sens_qΔ represents the rate constant of total ¹O₂ quenching by the sensitizer itself (physical quenching and chemical reaction).^10a

---