Aromatic Difluoroboron β-Diketonate Complexes: Effects of π-Conjugation and Media on Optical Properties

Abstract

Aromatic difluoroboron β-diketonate complexes (BF₂bdks) are classic fluorescent molecules that have been explored as photochemical reagents, two-photon dyes, and oxygen sensors. To gain a better understanding of their emissive properties in both solution and polymer matrices, BF₂bdks with varying aromatic groups were synthesized and their photophysical properties were investigated in both methylene chloride and poly(lactic acid) (PLA). Absorption spectra showed systematic variations that are well correlated with structural features, including the size of the aryl substituent and the presence of a para electron donating methoxy substituent. Computational modeling of the absorption spectra with the TD-B3LYP/6-311+G(d)//B3LYP/6-31G(d) formulation of density functional theory and a polarizable continuum model of dichloromethane solvent shows that all systems show intense π–π* one-electron excitations, usually from one of the highest occupied MOs (HOMO-k, k=0, 1, 2) to the lowest unoccupied MO (LUMO). Emission properties are sensitive to the dye structure and medium. Based on spectroscopic and lifetime studies, BF₂bdks exhibit comparable fluorescence properties in both solutions and polymers when the diketonate group is functionalized with smaller aromatic ring systems such as benzene. For BF₂bdks with larger arene ring systems, such as anthracene, emission from a strong intramolecular charge-transfer (ICT) state was also noted in both solution and in PLA. There are differences in relative intensities of peaks arising from π-π* and ICT excitations depending upon dye loading in PLA. Substituent effects were also observed. Electron-donating methoxyl groups on the aromatic rings lead to enhanced fluorescence quantum yields. For certain dyes, phosphorescence is detected at low temperature or under a nitrogen atmosphere in PLA matrices.

Introduction

Boron-containing materials possess impressive optical properties.^1,2,3,4 Difluoroboron β-diketonate complexes (BF₂bdks) are classic fluorescent molecules with large molar absorptivities and often high quantum yields⁵ that have been explored as photochemical reagents,^{6,7,8,9,10,11} two-photon materials,^12,13 conjugated polymers,^14,15,16 semi-conductors,¹⁷ photochromic materials,¹⁸ near IR probes,¹⁹ oxygen²⁰ and mechanical sensors.²¹ Extensively investigated targets among these aromatic BF₂bdks are difluoroboron dibenzoylmethane (BF₂dbm) and its derivatives, given some of the parent diketones are commercial products, such as dbm²² and avobenzone.²³ After boronation with BF₃·Et₂ in non-carbonyl solvents, most BF₂dbm derivatives can be obtained as air-stable, crystalline compounds. In solutions, they typically exhibit intense blue fluorescence and some give excimer or exciplex emission⁹ or dye polarity²⁴ effects at higher concentrations. When BF₂dbm fluorophores are dissolved in solid-state media, such as poly(lactic acid) (PLA), also known as polylactide, long-lived room-temperature phosphorescence (RTP) is observed in the absence of oxygen.²⁰ Through further heavy atom perturbation, in vivo luminescence oxygen sensing and dynamic hypoxia imaging are possible in tumors,^25,26 the brain and other contexts. BF₂dbm analogues have thus yielded promising preliminary results for cellular^{27, 28} and hypoxia imaging with 2-photon absorbing capability and compatibility with multiphoton methods.²⁸ Building upon these early successes, dyes with emission profiles across the visible region are important for multiplexing, and tissue penetration depth can be increased with redshifted dyes.²⁹ Although the emission wavelength of boron dye-polymer conjugate BF₂dbmPLA may be manipulated to a certain extent by polymer molecular weight, the tuning range is limited.²⁴ Also, this method does not shift the BF₂dbm absorption out of the UV region, which can be damaging to biological systems. Thus, the development of red shifted BF₂bdk derivatives can expand their utility for in vitro cellular studies, ex vivo assays, and in vivo imaging agents.

Although BF₂bdk luminescence has been investigated by many groups,^{5,18,30,31,32,33,34} ours included, we observe interesting emissive behaviors for the boron complexes in PLA compared to solution, which have not been reported in the literature. For example, for the naphthalene derivative, BF₂nbmPLA,³⁵ the fluorescence emissions in both CH₂Cl₂ and in the solid state (~440 nm) were similar to those for the benzene derivative BF₂dbmPLA under the same conditions.^{20, 36} However, the phosphorescence of BF₂nbmPLA showed a significant redshift (544 nm) compared to that of BF₂dbmPLA (509 nm), which suggests that π-conjugation length affects singlet and triplet states differently. Compared to the well-known difluoroboron BODIPYs (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene),^37,38 mechanistic studies of BF₂bdk luminescence are more sparse. Therefore to better understand the optical properties of BF₂bdk complexes, including emissive states, emission color range, and media effects, it is important to conduct systematic structure-property investigations.

Here we synthesized a series of simple BF₂bdk derivatives (1-10) for a luminescence study in combination with computational chemistry. The boron complexes all possess aromatic hydrocarbons of different sizes. Compared to phenyl molecules 1-4, the methoxyphenyl counterparts 5-8 have the same chemical structures except that the latter series has an electron-donating methoxyl group on the benzene ring to explore substituent effects. The dimethoxy-phenyl BF₂bdk 9 and the methyl-naphthyl complex 10 were also included for comparison. Nomenclature for the complexes is as indicated. We will also refer to complexes by the hydrocarbon substituents on the difluoroboron diketonate ring (i.e. Me-Ph = mbm, 1; Ph-Ph = dbm, 2; Ph-Np = nbm, 3; Ph-An = abm, 4; Me-PhOMe = mbmOMe, 5; Ph-PhOMe = dbmOMe, 6; Np-PhOMe = nbmOMe, 7; An-PhOMe = abmOMe, 8; PhOMe-PhOMe = dbm(OMe)₂, 9; Me-Np = mnm, 10). Fluorescence properties of these boron complexes were investigated in CH₂Cl₂ via UV/Vis and fluorescence spectroscopies, and quantum yield and fluorescence lifetime measurements. Computational studies were also performed to support and provide further insight into experimental findings. Also, because many useful photophysical properties of BF₂dbm derivatives arise in a solid-state environment, both fluorescence and phosphorescence were investigated for dye/PLA blends, which can inform future work with dye-polymer conjugates for imaging, sensing and other uses.

Experimental

Materials

Solvents CH₂Cl₂ and THF were dried and purified by passage through alumina columns. Boron trifluoride diethyl etherate (Aldrich, purified, redistilled) and all other reagents and solvents were used as received without further purification. Diketone ligands were prepared by Claisen condensation using NaH and boron complexes were prepared using BF₃ etherate as previously described. The data of complexes 1-4,³⁹ 5,³¹ 6,¹³ 9,¹³ 10 ³¹ are in accord with the literature. The synthesis of complexes 7 and 8 is described below.

Methods

¹H NMR (300 MHz) spectra were recorded on a UnityInova 300/51 instrument in CDCl₃. ¹H NMR spectra were referenced to the signal for residual protio chloroform at 7.26 ppm. UV/Vis spectra were recorded on a Hewlett-Packard 8452A diode-array spectrophotometer. Mass spectra were recorded using an Applied Biosystems 4800 spectrometer with a MALDI TOF/TOF analyzer. Elemental analysis was performed by Atlantic Microlab, Inc., Norcross, GA. Steady-state fluorescence emission spectra were recorded on a Horiba Fluorolog-3 Model FL3-22 spectrofluorometer (double-grating excitation and double-grating emission monochromators). Time-correlated single-photon counting (TCSPC) fluorescence lifetime measurements were performed with a NanoLED-370 (369 nm) excitation source and DataStation Hub as the SPC controller. Lifetime data were analyzed with DataStation v2.4 software from Horiba Jobin Yvon. Fluorescence quantum yields, Φ_F, for BF₂bdks (1-10) in CH₂Cl₂ were calculated versus anthracene in EtOH as a standard as previously described⁴⁰ using the following values: Φ_F anthracene = 0.27,⁴¹ n_D ²⁰ EtOH = 1.360, n_D ²⁰ CH₂Cl₂ = 1.424. Optically dilute CH₂CH₂ solutions of BF₂bdks and EtOH solution of the anthracene standard were prepared in 1 cm path length quartz cuvettes with absorbances < 0.1. Quantum yield measurements were performed with excitation, λ_ex = 350 nm, and an emission integration range = 365–695 nm. Phosphorescence spectra were recorded with the same instrument except that a pulsed xenon lamp (λ_ex = 369 nm; duration < 1 ms) was used for excitation and spectra were collected with a 2 ms delay after excitation. Phosphorescence lifetimes were measured with a 500 ns multi-channel scalar (MCS) excited with a pulsed xenon lamp (λ_ex = 369 nm; duration < 1 ms).

Compounds 1-10 were modeled using the Gaussian 09 suite of programs ⁴² using density functional theory for the ground state S₀ and time-dependent density functional theory for the electronic excited states S_j, j = 1,…. The S₀ geometry optimization was conducted with B3LYP/6-31+G(d) for the ground state in vacuo and with B3LYP/6-31G(d) for the ground state in dichloromethane, modeled as a polarizable continuum according to the default method in Gaussian 09. The optimized ground state geometries are assured to be local minima since all computed vibrational frequencies were real. Two estimates of the triplet emission frequencies were possible; both used the optimized geometry of the lowest-energy triplet state, called T₁(opt), obtained with B3LYP/6-31G(d) in vacuo and in dichloromethane. One estimate is provided by the energy difference between T₁(opt) and the S₀ energy at the same geometry. Another estimate is obtained by TD-B3LYP/6-311+G (d) calculations of vertical S₀-T₁ electronic transitions. Both included the solvent model.^{43 , 44} Graphic representations for SCF orbitals for the ground state were obtained with GaussView 5. Triplet structures were obtained by geometry optimization B3LYP/6-31G(d).

Difluoroboron 2-naphthoyl 4′-methoxybenzoyl methane (BF₂nbmOMe), 7

4-Methoxyacetophenone (150 mg, 3.0 mmol), methyl 2-naphthoate (230 mg, 1.2 mmol) and THF (20 mL) were added sequentially to a 50 mL round bottom flask. After stirring the mixture for 10 min, a suspension containing NaH (40 mg, 1.5 mmol) in THF (10 mL) was added dropwise at room temperature under N₂. The mixture was stirred for 20 h then quenched with saturated aqueous NaHCO₃ (1 mL). THF was removed in vacuo then 1 M HCl (20 mL) was added. The aqueous phase was extracted with CH₂Cl₂ (3 × 20 mL), and combined organic layers were washed with distilled water (2 × 10 mL) and brine (10 mL), and dried over Na₂SO₄ before filtration and concentration in vacuo. The residue was purified by column chromatography on silica gel eluting with hexanes/ethyl acetate (6:1) to give the diketone precursor, naphthoyl-4′-methoxybenzoyl methane (nbmOMe) as a yellow solid (240 mg, 79%). ¹H NMR (300 MHz, CDCl₃) δ 17.10 (s, 1H, COCHCOH), 8.54 (s, 1H, 1′-ArH), 8.06-7.88 (m, 6H, 3′, 4′, 5′, 8′-ArH, 3″, 5″-ArH), 7.61-7.55 (m, 2H, 6′, 7′-ArH), 7.03–6.99 (m, 2H, 2″, 6″-ArH), 6.95 (s, 1H, COCHCO), 3.90 (s, 3H, -ArOMe); MS (MALDI): m/z calcd for C₂₀H₁₆O₃ [M+H]⁺ 305.11, found 305.07. Boron trifluoride diethyl etherate (81 μL, 0.64 mmol) was added to a solution of nbmOMe (130 mg, 0.43 mmol) in CH₂Cl₂ (20 mL) under N₂. The mixture was refluxed for 12 h then the solvent was removed in vacuo. The residue was purified by column chromatography on silica gel eluting with hexanes/ethyl acetate (6:1) to give 7 as a yellow solid (123 mg, 81%). Anal. calcd for C₂₀H₁₅BF₂O₃: C, 68.22; H, 4.29; F, 10.79; Found: C, 68.31; H, 4.42; F, 10.52. ¹H NMR (300 MHz, CDCl₃) δ 8.74 (s, 1H, 1′-ArH), 8.19 (d, 2H, J = 9.0 Hz, 3″, 5″-ArH), 8.08-7.89 (m, 4H, 3′, 4′, 5′, 8′-ArH), 7.66-7.60 (m, 2H, 6′, 7′-ArH), 7.24 (s, 1H, COCHCO), 7.04 (d, 2H, J = 9.0 Hz, 2″, 6″-ArH), 3.94 (s, 3H, -ArOMe); MS (MALDI): m/z calcd for C₂₀H₁₅BF₂O₃Na [M+Na]⁺ 375.10. Found 375.03.

Difluoroboron 2-anthracenoyl 4′-methoxybenzoyl methane (BF₂abmOMe), 8

This complex was prepared as described for 7. Briefly, 2-acetylanthracene (101 mg, 0.45 mmol), methyl benzoate (91 mg, 0.55 mmol) and sodium hydride (46 mg, 1.82 mmol) in THF (20 mL) gave crude diketone anthroyl-4-methoxybenzoyl methane, abmOMe, which was purified by column chromatography on silica gel eluting with hexanes/ethyl acetate (6:1). The diketone was obtained as a crude yellow solid (72 mg, 45%). ¹H NMR (300 MHz, CDCl₃) δ 17.09 (s, 1H, ArCOH), 8.73 (s, 1H, 1′-ArH), 8.58 (s, 1H, 9′-ArH), 8.45 (s, 1H, 10′-ArH), 8.09-7.94 (m, 6H, 3′, 4′, 5′, 8′-ArH. 3″, 5″-ArH), 7.54-7.51 (m, 2H, 6′, 7′-ArH), 7.03-6.96 (m, 3H, 2″, 6″-ArH, COCHCO), 3.88 (s, 3H, -ArOMe). (A second species is suggested by minor peaks at 17.13 and 3.90 ppm.) The boron complex was prepared via the reaction of crude abmOMe (46 mg, 0.14 mmol) and BF₃ ·OEt₂ (36 μL, 0.28 mmol) in CH₂Cl₂ (20 mL) and was purified via column chromatography on silica gel eluting with hexanes/ethyl acetate (1:1) to give 8 (30 mg, 57%) as a red solid. Anal. calcd for C₂₄H₁₇BF₂O₃: C, 71.67; H, 4.26; F, 9.45; Found: C, 71.50; H, 4.20; F, 9.25. ¹H NMR (300 MHz, d⁶-DMSO) δ 9.30 (s, 1H, 1′-ArH), 8.91 (s, 1H, 9′-ArH), 8.70 (s, 1H, 10′-ArH), 8.44 (d, 2H, J = 8.7 Hz, 3″, 5″-ArH), 8.27-8.14 (m, 4H, 3′, 4′, 5′, 8′-ArH), 8.01 (s, 1H, COCHCO), 7.67-7.58 (m, 2H, 6′, 7′-ArH), 7.23 (d, 2H, J = 8.7 Hz, 2″, 6″-ArH), 3.94 (s, 3H, -ArOMe); MS (MALDI): m/z calcd for C₂₄H₁₇BF₂O₃Na [M+Na]⁺ 425.11. Found 425.13.

Results and Discussion

The boron complexes were synthesized by Claisen condensations with NaH to generate the β-diketones, followed by boronation with BF₂·OEt₂ in CH₂Cl₂. The rate of Claisen condensation drops and the product solubility decreases with increasing arene size. All of the complexes were obtained as yellow crystals except for the anthryl compounds 4 and 8, which are dark red powders. The synthesis and solid-state fluorescence and mechanochromic properties for compounds 1-4 have been described elsewhere.³⁹

Absorption Spectra in CH₂Cl₂

The optical properties of compounds 1-10 (Table 1) were first studied in CH₂Cl₂ because these conditions do not result in boron-ligand dissociation, ⁴⁵ exciplex formation ^{46 , 47} or photochemical reactions^7,8 seen for solvents containing Lewis bases, aromatic rings or double bonds. Absorption spectra show that all ten complexes have high extinction coefficients (26900–73000 M⁻¹ cm⁻¹), which is typical of π-π* transitions (Figure 1). Complex 9 with two methoxyl benzene rings has the strongest absorption; whereas, the weakest absorptions occur for complexes containing the anthracene moiety, namely 4 and 8. Compared with complexes 1-4, complexes 5-8 show similar but stronger and red-shifted absorption spectra, since the methoxyl substituent possesses electron donating ability. Another difference between the two sets of compounds is that 5-8 exhibit slightly sharper absorption spectra (i.e. smaller full width at half maximum (fwhm) for major absorption bands), perhaps due to more rigid structures from the electron donating methoxyl groups (inferred from resonance structures). Figure 1C compares BF₂dbm (2) with mono-(6), and di-methoxy-substituted (9) derivatives to explore substituent effects. As for 5-8, in this series too, increasing donor strength correlates with stronger absorption and redshifted emission. The naphthyl-methyl complex 10 is included because 2 and 10 have similar π conjugation lengths; however, the absorption of 10 is different from 2 in two aspects. First, 10 shows a bimodal absorption spectrum (344 and ~380 nm shoulder), as is also noted for complexes 4 (390 and 467 nm) and 8 (404 and 451 nm); secondly, the absorption maximum of 10 is far more blue shifted than the rest. From the absorption spectra in CH₂Cl₂, it is reasonable to believe that complexes 4, 8, and 10 (which also have the smallest extinction coefficient values) have significant intramolecular charge transfer (ICT) character compared to the rest as evidenced by a second structureless, redshifted absorption band in addition to the main absorption peak. One common feature shared by the three molecules (4, 8, and 10) is that they are the most unsymmetrically substituted in terms of the arene sizes on the diketonate. More definitive conclusions may be reached after the emission spectral analysis and computational results are presented below.

Table 1.

Optical Properties of BF₂bdks 1-10 in CH₂Cl₂.^a

	λmax/nmb	ε/M−1cm−1c	λem/nmd	τF/nse	φFg	τradh	Stokes Shift/cm−1
1	330	31400	416	1.04f	0.0031	--	--
2	365 380i	49200 42900i	417 396i	0.47	0.26	1.81	1060
3	376	37100			0.50	9.38
	(398)a,i	(36300)i	(482)	(4.69)			(4380)
4	390	26900	417	too weak	0.30	35	1660
	(467)a	(8100)	(596)	(10.5)			(4630)
5	360	39900	393	1.02	0.65	1.57	2330
6	397	57000	433	2.06	1.00	2.06	2090
7	415	66300	449	1.51	0.68	2.22	1820
8	404	29700	460i	1.96	0.49	21.4	2330
	(451)a	(11700)	(572)	(10.5)			(4590)
9	411	73000	437	1.79	0.78	2.29	1450
10	344	34300			0.63	15.2
	(380)a,i	(9900)	(459)	(9.55)			(4530)

^aValues in parentheses correspond to charge transfer processes.

^bAbsorption maxima.

^cExtinction coefficients calculated at the absorption maxima.

^dFluorescence emission maxima.

^eFluorescence lifetime excited with a 369 nm LED monitored at the emission maxima. All fluorescence lifetimes are fitted with single-exponential decays unless indicated.

^fPossibly multi-exponential decay.

^gRelative quantum yield using anthracene in EtOH as a standard.

^hRadiative lifetime, where τ_rad =τ_F/φ_F.

ⁱBand is present as a shoulder or second maximum.

Figure 1.

Absorption spectra of BF₂bdk complexes 1-10 in CH₂Cl₂.

Fluorescence in CH₂Cl₂

The emission spectra and fluorescence lifetimes of the complexes were also measured. The results are given in Figure 2 and Table 1. As shown in the photograph in Figure 2, under UV illumination, all the complexes exhibit intense fluorescence ranging from violet to orange-red except for 1, which only faintly fluoresces in the violet-blue region. The Stokes shift, the difference in energy between the emission state energy and the energy of its emitting state absorption, are also included. In several cases the lowest energy absorption or highest energy emission bands were estimated from a shoulder or a weaker peak than the absolute maximum (See Table 1, footnote i). For the dual emitting complexes, e.g. 4 and 8, data are reported for both emissions from each complex. The complexes are grouped into two categories. One has large Stokes shifts (~4500 cm⁻¹) with long radiative lifetimes (>9 ns); these include 3, 10 and the low energy emissions of 4 and 8. The other complexes and the high energy emissions of 4 and 8 have small Stokes shifts (1200–2500 cm⁻¹) and short radiative lifetimes (2 ns). As will be shown, these emissions arise from different types of excited states.

Figure 2.

Photographs and steady-state emission spectra of BF₂bdks 1-10 in CH₂Cl₂ (λ_ex = 369 nm).

Theoretically, an increase in π conjugation length typically brings about a red-shift of emission and change in quantum yield.⁴⁸ The spectra roughly follow this rule but exceptions were also observed. For example, the π donating methoxyl substituent extends the conjugation of 8 through π donation compared to 4; however the emissions of 8 are blue-shifted compared with 4. The same trend was observed for the naphthalene-containing BF₂bdks 3 (Ph-Np, 482 nm), 7 (Np-PhOMe, 449 nm), and 10 (Me-Np, 459 nm), where the greatest π conjugation (7) results in the most blue-shifted emission in CH₂Cl₂. This observation indicates that conjugation length is not the only parameter that influences the emissive state energy of the complexes. This signals that other factors such as inductive effects or charge transfer are involved, as we will discuss in greater detail below.

The fluorescence lifetimes of the complexes were also measured and the results are collected in Table 1. The radiative lifetime for each molecule are given by

$${\tau }_{\text{rad}}={\tau }_{\text{F}}/{\phi }_{\text{F}}$$ (1)

where τ_F and φ_F are the observed lifetime and luminescence quantum yield respectively. This value, τ_rad which is also known as the instrinsic lifetime, is the lifetime that would be measured if the there were no radiationless processes competing with the luminescence decay. The radiative lifetime τ_rad is a direct measure of the allowedness of the emission. Optically dilute complexes 2-10 exhibited single exponential decays in CH₂Cl₂. The decay of 1 was very weak (φ_F <0.004) and with an estimated multi-exponential lifetime of ~1 ns. Given the weak emission, the absence of a mirror image relationship between the absorption and emission, and the anomalously long lifetime for the low quantum yield, we attribute the emission to trace impurities, possibly a very low concentration conformer, given it persists from preparation to preparation, and consider 1 to be non-or extremely weakly luminescent. The anthracene-containing complexes 4 (Ph-An) and 8 (An-PhOMe) displayed especially long fluorescence decay (10.5 ns); whereas, BF₂dbm exhibited the shortest decay (0.47 ns) despite its reasonably large fluorescence quantum yield (Φ_F = 0.26). However, the addition of one (6) or two (9) methoxy groups onto BF₂dbm increases the quantum yields of 6 and 9 considerably.

An interesting result was also observed for the fluorescence lifetimes of the most unsymmetric molecules. Complexes 3, 4, 8, and 10, with different diketonate arenes (or Me), have the longest radiative decay times (3, nbm, τ_rad = 9.38 ns; 4, Ph-An, τ_rad = 35.0 ns; 8, An-PhOMe, τ_rad = 21.4 ns; and 10, Me-Np, τ_rad = 15.2 ns); whereas, the remaining complexes including the symmetrically substituted complexes, such as 2 (Ph-Ph) and 9 (PhOMe-PhOMe) all have much shorter radiative decays of about 2 ns. The differences in lifetimes for 3, 4, 8 and 10 coupled with the presence of weaker long wavelength absorption bands for 4, 8, and 10 indicate a fundamentally different type of excited state emission. Though not as distinct in the UV/vis spectrum for 3, it could still be argued that the low energy feature evident at ~400 nm on the absorbance band also arises from albeit weaker ICT. This is also supported by computational results presented below.

As stated above, distinct low intensity red features in the absorption spectra for unsymmetrical complexes 4, 8, and 10 are ascribed to intramolecular charge transfer, ICT. Computational modeling discussed below supports this assertion; these features are dominated by excitations in which charge is transferred from the larger more electron rich arene towards the boron diketonate center. The second and more intense feature lying at higher energy is a π–π* transition where both π and π* are delocalized throughout the molecule. Since both MOs occupy a common space, the transition has little ICT character. In the other more symmetric structures such as 2 or 9 the π–π* transition is at lower energy than any ICT transition and dominates the long-wavelength region of the absorption spectrum. As a result, shorter radiative decay lifetimes are expected.

These observations about absorbance also inform the interpretation of Ph-An (4) and An-PhOMe (8) complex emissions, where 4 has a stronger ICT tendency than 8. From the fluorescence spectra, the emission maximum of 4 is 24 nm (704 cm⁻¹) redshifted compared to 8. Though the π donating methoxy group does not seem to make a difference in terms of the fluorescence decay (τ_F = 10.5 ns for both 4 and 8), the radiative lifetimes follow the expected trend, given the differing quantum yields. For 4, τ_rad is nearly double the value for 8 indicating a large long lived contribution from the ICT state. ICT from anthracene to the diketonate group is significant regardless of whether benzene or methoxybenzene is present as a competing donor group. In this process, the phenyl ring may serve largely as a substituent on the acceptor. The more electron rich MeO-Ph ring (vs Ph) may slightly weaken the acceptor ability of the boron diketonate ring, thus increasing the energy of the transition. In comparison to anthracene dyes 4 and 8, the phenyl derivatives 2 (Ph-Ph, τ_rad = 1.81 ns) and 9 (PhOMe-PhOMe, τ_rad = 2.29 ns) and 6 (Ph-PhOMe, τ_rad = 2.06 ns) are symmetrically substituted molecules with respect to the arenes and they show relatively short fluorescence decay. Within this series where π-π* transitions likely dominate the electronic transitions, the methoxy substituents exert a subtle but measureable effect on fluorescence decay.

Complexes 4 and 8 exhibit dual emissions (Table 1). There is a weak high energy emission and stronger low energy emission. They clearly come from two different ground state species as the excitation spectra (provided in the Supporting Information) are different, including in the UV. Both low and high energy emissions for 8 have a single exponential decay (Table 1). The same is true for 4, though the high energy emission is weak; a lifetime could not be obtained. Further, there is no negative preexponential factor for the long lived emission in either case, which establishes that the low energy emissions are not arising from the state responsible for the high energy emission. We suggest that there are two ground state conformers for these complexes. One has a structure conducive to the π–π* state being lowest energy in the excited state and gives rise a short lived (~2 ns) π–π* fluorescence. The dominant structure gives an ICT excited state lowest in energy and is the source of the much longer lived ICT emission (10.5 ns).

In summary, absorption and fluorescence data indicate moderate to strong ICT character for unsymmetrically substituted difluoroboron diketonates 3, 4, 8 and 10, while π-π* transitions dominate for symmetrically substituted compounds (e.g. 2 and 9). This is consistent with known solvatochromic properties of this family of dyes, as charge separated states are sensitive to the polarity of the surrounding medium.^{13, 49} The electronic transitions of the BF₂bdks depend on their molecular symmetries. A greater disparity in π electron donating ability of the two arene rings seems to result in a stronger ICT state dominated by the more electron rich aryl group. When the bdk arene ring sizes are comparable, in more symmetrical structures, a delocalized π-π* model is proposed. These experimental conclusions are supported by computational studies discussed below.

Computational Studies

The Gaussian 09 software suite⁴² was used for all computational modeling. Optimized geometries for species 1-10 were determined with a density functional theory method, B3LYP/6-31G(d). The absorption spectra were characterized by time-dependent density functional theory, specifically TDB3LYP, in a more flexible basis 6-311+G(d). TD-DFT methods represent excited states as combinations of the single excitations of an electron from an orbital occupied in the SCF ground state to an orbital in the set vacant in the ground state. The wavelength and intensity of vertical absorption transitions correlate well with experimental counterparts, with R²=0.91 between absorption wavelengths computed with no consideration of solvent effects and the band maxima as reported in Table 1. Incorporation of solvent effects by the polarizable continuum model (Tomasi’s PCM) brings predicted absorption wavelengths into close agreement with observed values, increases R² to 0.96 and also slightly improves the estimates of absorption intensities. The calculations, in vacuo or with solvent treatment, capture not only the general form of the absorption curves but track the intensity enhancement and the wavelength shifts attending structural changes. The Supporting Information contains graphs of computed spectra and pictures of relevant orbitals, which may aid in comparison.

Trends in the computed and observed absorption spectra can be described most easily by considering subsets of related molecules. The family 1-4 has a common Ph-R structure and dye group 5-8, PhOMe-R, with R = methyl, phenyl, naphthyl, and anthryl. The subset 1, 5, and 10 (Me-Ph, Me-PhOMe, and Me-Np respectively) allows examination of the relative effects of methoxy substitution and extending the aryl from phenyl to naphthyl. Finally the subset 2, 6, and 9 (Ph-Ph, Ph-PhOMe, and PhOMe-PhOMe respectively) allows discussion of the cumulative effects of methoxy substitution.

The families 1-4 and 5-8 show a red shift in the low-energy edge of the optical spectrum as the aryl substituent increases in size. Extending the size of the arene raises the energy of its highest occupied molecular orbital (HOMO). If there is a reasonable energy match between the HOMO for the aryl fragment and MOs in the boron containing ring there will be substantial mixing between them. This gives rise to delocalization over the entire molecule, elevation of the highest occupied MO, and a redder transition. If the energy match is poor, the aryl MO remains localized, and the HOMO (aryl) to LUMO (delocalized) transition changes character to a charge transfer. The charge transfer band may have relatively weak intensity owing to reduction in Franck-Condon factors, and can be very red. It has been noted that the DFT functional is generally inaccurate in predicting transition energies involving long-range ICT states.^50,51 Nonetheless, this feature of the absorption spectra is captured by the modeling, and is a significant part of what we see in the anthryl-substituted species 4 and 8, where contrast between aryl rings is greatest and the effect on the spectrum is most pronounced. Similar but less evident features are found for 3 and 10. For example, the computed spectrum for complex 3 suggests the presence of a weak absorption with some ICT character, to the red of a stronger π-π* transition. The weaker red absorption computed at 430 nm may not be clearly distinguishable from the stronger blue transition computed at 372 nm owing to the breadth of the absorption bands. As inferred from the experimental observations, compound 4 appears to have the greatest ICT character. See Figure 3 for a visualization of transitions for molecules 2 and 4 determined with the solvent correction. Table 2 contains a more detailed summary of these results.

Figure 3.

Orbitals involved in the absorption spectra for complexes 2 (Ph-Ph) and 4 (Ph-An) in methylene chloride medium.

Table 2.

Comparison of Experimental and Calculated Optical Properties of BF₂bdks 1-10 in CH₂Cl₂ and In Vacuo.

	λmax/nma	ε/M−1cm−1 b	λcalc CH2Cl2/nmc	fd	λcalc in vacuo/nme	ff	Transition Type
1	330	31400	306	0.6111	306	0.4671	π-π*
2	365	49200	364	0.9764	342	0.7761	π-π*
3	376 398i	37100 36300i	372 430	0.9648 0.2791	353	0.8200	π-π* ICT
4	390 467	26900 8100	385 539	0.8325 0.2525	359 498	0.9262 0.1723	π-π* ICT
5	360	39900	338	0.8090	325	0.7202	π-π*
6	397	57000	387	1.0768	361	0.8994	π-π*
7	415	66300	389g,h 422g,h	0.7417 0.6436	367	0.7811	π-π*
8	404 451	29700 11700	403 525	1.2386 0.3155	382 485	0.7828 0.2092	π-π* ICT
9	411	73000	399	1.2893	370	1.1035	π-π*
10	344 380i	34300 9900	338 405	0.7263 0.1541	324 380	0.5522 0.106	π-π* ICT

^aExperimental absorption maxima in dichloromethane.

^bExtinction coefficients calculated at the absorption maxima in dichloromethane.

^cComputed transition wavelength (in dichloromethane medium)

^dComputed oscillator strength (in dichloromethane medium)

^eComputed transition wavelength (in vacuo)

^fComputed oscillator strength (in vacuo)

^gTwo overlapping transitions: ε is proportional to the sum of squares of f’s.

^hComplex 7 exhibits two transitions of roughly comparable energy. These each have some ICT character but are predominately π-π*.

ⁱLow energy ICT band is present as a shoulder.

It is interesting to note that species 4 and 8 display the strongest Stokes shift in the emission. That is entirely consistent with the substantial geometrical rearrangement to be expected in the charge-transfer excited state. It is clear that the modeling is entirely consistent with the interpretation of the experimental absorption spectra, and lends indirect support to the ideas expressed in the discussion of the emission properties of these systems.

When aryl substituents on the difluoroboron diketonate ring are identical (2), spectra are dominated by HOMO to LUMO excitations; both orbitals are delocalized throughout the molecule. When the substituents are very different (4), the spectrum has a broad weak red feature derived from a HOMOLUMO ICT transition, and a higher energy more intense feature in which the HOMO-1 to LUMO and HOMO to LUMO+1 excitations are strongly mixed. Additional orbitals and characterization of the transitions may be found in the Supporting Information.

Modeling permits further remarks on the absorption spectra. In most cases one can discern two strongly overlapping bands in the leading red edge of the optical absorption spectra. The two bands seem to grow more separate and distinct as the size of the aromatic substituent grows, either in the series 1-4 or 5-8, until they become entirely separate in 4 and 8. This is evidently another consequence of the systematic elevation of the highest occupied MO of the aryl moiety, which eventually allows distinct π-π* transitions, first the low-energy ICT from HOMO to LUMO and then the fully delocalized HOMO-1 to LUMO.

The spectra for the subset 1, 5, and 10 (Me-Ph, Me-PhOMe, and Me-Np respectively) suggests that methoxy substitution on phenyl elevates the highest occupied MO without blocking its delocalization, while extending the aryl to naphthyl localizes the highest occupied MO more thoroughly on the aryl fragment. In the computed spectrum a weak red HOMO to LUMO transition distinct from the more intense and higher energy HOMO-1 to LUMO feature is noted. This finds its counterpart as a red shoulder in the first broad absorption band of the experimental spectrum.

Finally the sequence of methoxy substitution in the subset 2, 6, and 9 (Ph-Ph, Ph-PhOMe, and PhOMe-PhOMe respectively) shows successive red shifts and intensity enhancements as first one and then two para-methoxy substituents are introduced. Each methoxy elevates its aryl highest occupied MO without going so far as to impede delocalization. Again, computed results correlate with experiment.

Even approximate treatment of solvent effects has had some positive effects. For example, the excitation wavelength calculated in vacuo is always blue-shifted compared to the experimental value; the PCM version of the polar solvent dichloromethane induces a red shift, toward the experimental values. This behavior is consistent with the observed solvatochromism of this class of compounds.^13,49 the calculation is still approximate however, and absorption intensities are poorly described. In particular, compounds 4 and 8 have much weaker absorption than their relatively high oscillator strengths would predict. It may be that more advanced methods^52,53 can be employed to better estimate the excitation energies of this family of complexes.

Polymer Medium Effects: Fluorescence of BF₂bdks/PLA Blends

The optical properties of the BF₂bdk series were also measured in poly(lactic acid), a solid-state solvent, at both room temperature and 77 K with two different dye loadings. The PLA used in the experiment was synthesized via solvent-free ring-opening polymerization.⁵⁴ Boron complexes 1-10 and PLA were dissolved in CH₂Cl₂ to form a homogenous solution that was used to cast a thin film onto the inner wall of a small vial. The film was then thoroughly dried in vacuo before luminescence measurements. The measured optical properties for these blends at room temperature are presented in Figure 4 and Table 3.

Figure 4.

Photograph and steady state emission spectra of BF₂bdk/PLA blends at 0.05% (A, B, C) and 0.5% (D, E, F) dye loadings at room temperature. (λ_ex = 369 nm except for 1 and 5 where λ_ex = 350 nm)

Table 3.

Optical Properties of BF₂bdk/PLA Blends at Room Temperature

	0.05% BF2bdk/PLA(w/w%)		0.5% BF2bdk/PLA (w/w%)
	λF/nma	τF/nsb	λF/nma	τF/nsb
1	415	n/ac	413	n/ac
2	415	1.82	415	1.67
3	448	2.31	453	3.66
4	419 (532)d	weak (10.1)d	428 (540)d	1.81 (10.8)d
5	389	0.61	392	0.89
6	427	2.36	435	3.79
7	437	1.44	443	1.83
8	430 (514)d	1.22 (9.12)d	430 (531)d	1.98 (13.43)d
9	429	2.44	435	3.71
10	429	5.35	413	6.58

^aλ_ex = 369 nm.

^bPre-exponential weighted lifetimes. All fluorescence lifetimes were fitted with triple-exponential decays.

^cSignal too weak to be detected with current LED excitation.

^dValues in parentheses correspond to charge transfer processes.

Compared to the emission data in CH₂Cl₂, for the more dilute 0.05% (w/w) dye/polymer loading, the fluorescence emission maxima at room temperature are typically blue-shifted in the PLA substrate possibly owing to the fact that PLA is a less polar and a more rigid medium than CH₂Cl₂. A combination of solvatochromic and rigidochromic effects ⁵⁵ may explain the blueshifts, which are greatest for emissions associated with ICT transitions of complexes 4 (Ph-An) and 8 (An-PhOMe) (64 and 58 nm, i.e. 2018 and 1972 cm⁻¹ respectively) with large anthracene substituents. This too, supports the hypothesis that the strongest ICT states are present in these anthracene complexes, because in the solid-state solvent PLA, the presumed ICT “twisted configuration” may be hindered.⁵⁶ The emission spectra from 4 and 8 in PLA are bimodal, with 4 possessing a more conspicuous high energy shoulder at 419 nm and 8, at 430 nm). In fact, even in CH₂Cl₂, both 4 and 8 have additional high energy bands around 417 nm and 460 nm respectively (Figure 2), which are weaker compared to those recorded in PLA. It is not uncommon to observe dual emission bands for donor-acceptor fluorescent molecules in polar solvents. According to twisted intermolecular charge transfer (TICT) theory,²⁹ the two distinct fluorescence emission bands of complexes 4 and 8 might be ascribed to a TICT state (lower energy) and a co-planar locally excited state (higher energy), as discussed above for solution and computational studies. Interestingly, the shoulder emission of complex 4 shares almost the same emission maxima and vibronic features as those for the dbm complex 2; similarly for complex 8, the higher energy shoulder closely matches to the fluorescence of dbmOMe complex 6.

The fluorescence lifetimes measured for the complexes in PLA were all fitted to triple-exponential decays, except for the Me-Ph derivative 1 which did not show strong enough intensity for measurement with the current instrument setup. As has been described previously, the solid-state polymeric environments are heterogeneous and therefore, multi-exponential decays are common for BF₂bdks.²⁰ Compared to fluorescence lifetimes recorded in CH₂Cl₂, the values obtained in PLA did not follow an obvious trend. However, we did observe some reductions in τ_F in PLA versus CH₂Cl₂ for molecules that exhibited strong ICT character (e.g. ICT band in PLA vs CH₂Cl₂: 4: 10.1 vs 10.5 ns; 8: 9.12 vs 10.5 ns; and 10: 5.35 vs 9.55 ns), which again suggests that the ring twisting may be partially inhibited or that there are reduced relaxation effects in the rigid and less polar PLA matrix at room temperature.

From the steady-state fluorescence spectra, most of the polymer films with higher dye loading exhibit slightly red-shifted emission maxima (Table 3 and Figure 4), likely due to the exciton coupling or “fluorophore-fluorophore interaction”²⁴ as previously reported. Methyl dyes 1 and 10 lacking a methoxyl substituent and the dbm complex 2 serve as exceptions. The fluorescence lifetimes, accordingly, are consistently longer for molecules showing red-shifted emission maxima at higher concentrations. The increase in lifetimes for high dye loadings has also been discussed in terms of dye-dye interactions, where excimeric species are likely to form. The propensity of BF₂bdk dyes to form H aggregates has also been advanced as a possibility.⁵⁷

Temperature effects were also studied by collecting steady-state emission spectra at 77 K in liquid nitrogen for all BF₂bdk/PLA blends at two different dye loadings (Figure 5 and Table 4). For dye blends at lower concentrations, most emission maxima are slightly blue-shifted probably due to the rigidochromic effect at liquid nitrogen temperature. The blueshifts are generally small; the low loading of compound 2 is even slightly redshifted, indicating that PLA at room temperature, below the glass transition temperature of ~60°, is already rather rigid. At low temperature, the emission spectra also show sharper vibrational structure for complexes 1-4. The spectra are also narrowed for BF₂bdks 5-8, which have methoxy substitutions, and exhibit less prominent vibrational features. Furthermore, distinct phosphorescence bands are observed for certain complexes at low temperature (e.g. naphthyl complexes 3: 537 nm, 7: 538 nm, 10: 523 nm). Further discussion of the emissive triplet state for these materials is provided below. Low temperature fluorescence lifetimes were also examined for comparison with room temperature data. Most compounds have moderately increased fluorescence lifetimes, as is expected at lower temperatures due to less collisional quenching. However, more significant differences in both emission spectra and lifetimes were observed for the 0.5% dye/PLA blends at 77K. Many of the differences are probably caused by the inhibition of excimer formation at liquid nitrogen temperature and a significant reduction in fluorescence lifetimes (e.g., lifetime change for 9, PhOMe-PhOMe, from 3.71 ns at RT to 1.75 ns at 77 K). In fact, fluorescence lifetimes at 77 K for boron dyes at different loadings are quite close to the lifetimes at room temperature, which confirms that the lengthened lifetimes at the high loadings at room temperature are not likely to be caused by trivial effects such as reabsorption and reemission. The results suggest that even in a rigid polymer matrix such as PLA, aggregate or excimer formation is possible for smaller BF₂bdks at room temperature.

Figure 5.

Fluorescence emission of BF₂bdk/PLA blends at 0.05% (A, B, C) and 0.5% (D, E, F) dye loading at 77 K (λ_ex = 369 nm except for 1 and 2 at 0.5% loading, and 5 for both loadings, where λ_ex = 350 nm).

Table 4.

Optical Properties of BF₂bdk/PLA Blends at 0.05% and 0.5% Dye Loadings at 77 K

	0.05% BF2bdk/PLA (w/w%)		0.5% BF2bdk/PLA (w/w%)
	λF/nma	τF/nsb	λF/nma	τF/nsb
1	412d	n/ac	391d	n/ac
2	421	2.11	413d	2.19
3	441	2.40	449	3.33
4	418 (524)e	weak (9.35)e	418 (531)e	1.54 (10.10)e
5	385	0.64	387	0.84
6	422	2.06	435	2.12
7	434	1.42	425	1.39
8	427 (523)e	1.85 (7.87)e	431 (529)e	2.03 (8.31)e
9	425	1.77	432	1.75
10	426	6.59	409d	7.81

^aExcited with 369 nm LED.

^bPre-exponential weighted lifetimes. All fluorescence lifetimes were fitted with triple-exponential decays.

^cSignal too weak to be detected with current LED excitation.

^dExcited at 350 nm.

^eValues in parentheses correspond to charge transfer processes.

Polymer Medium Effects: Phosphorescence of BF₂bdks/PLA Blends

As described above, in PLA matrices under air most of the complexes exhibit intense blue fluorescence except complex 4 and 8, which exhibit yellow and greenish-yellow emission, as shown in Figure 4. When steady state emission spectra were acquired at 77 K for BF₂bdk/PLA blends, phosphorescence emission spectra were observed for complexes 3 (Ph-Np), 7 (Np-PhOMe) and 10 (Me-Np) (e.g. Figure 5, A–C). As previously reported,²⁰ in the solid state, amorphous PLA serves as a rigid medium which may block thermal decay pathways and at the same time hinder quencher molecules from interacting with the dyes in the excited states.⁵⁸ Thus, in PLA matrices even room temperature phosphorescence (RTP) is observed for BF₂bdks. Because RTP can be quenched by oxygen in ambient air, the BF₂bdk/PLA blends at two different dye loadings (0.05% vs. 0.5%) were also studied under a N₂ atmosphere, collecting both total emission spectra (fluorescence plus phosphorescence), RTP spectra (gated intensity) and low temperature phosphorescence (LTP) spectra (77 K, in liquid nitrogen). Under a nitrogen atomosphere, distinct green or yellow RTP was observed for all BF₂bdks/PLA blends except for complexes 1 (Me-Ph), 4 (Ph-An) and 8 (An-PhOMe). Phosphoresence for 5 was very weak; emission maxima could be determined but lifetimes were not detectable (RT under N₂ in PLA: 0.05% loading: 466 nm; 0.5% loading: 487 nm). (Room and low temperature emission spectra for 5 in PLA are provided in the Supporting Information.)

The emission spectra of 2 (Ph-Ph), 6 (Ph-PhOMe) and 9 (PhOMe-PhOMe) (0.05% and 0.5% dye loading) are provided in Figure 6. Compared to fluorescence, the relative intensity of RTP is quite low; the total emission spectra (both fluorescence and phosphorescence under nitrogen) are nearly identical to the fluorescence spectra (under air) (Figure 4). Phosphorescence spectra, both room temperature and 77 K, were obtained by gated measurement with a delay time of 2 ms after excitation. The shoulder band observed in certain phosphorescence spectra (e.g. 2 at ~420 nm, 6 at 427 nm, 9 at ~430 nm) which matches the fluorescence emission, has been ascribed to delayed fluorescence caused by the thermal repopulation of the excited singlet state from the long lived excited triplet state.^20,24 Complexes 2, 6 and 9 all have phenyl moieties on both sides of the diketone but are non-, mono-, and bis-methoxy substituted, respectively. With the presence of methoxyl substituents, phosphorescence emission maxima for 6 and 9 exhibit distinct red-shifts compared to 2. Whether one or two –OMe groups are present, however, seems to have little effect on emission wavelength. Unlike the more significant red-shift in fluorescence emission in going from low to high dye loading, the phosphorescence merely exhibits a slight red-shift in room temperature spectra (e.g. 6: Δλ = 8 nm (430 cm⁻¹) for fluorescence (Table 3), Δλ = 1 nm (39 cm⁻¹) for phosphorescence (Table 5) from low to high dye loading.) The delayed fluorescence band intensity increases and redshifts for the higher dye loading, consistent with previously reported dye loading effects.²⁴ At 77 K, where thermal back population of the singlet state is inhibited, delayed fluoresence is not evident.^20,35

Figure 6.

Normalized emission spectra for BF₂bdk complexes 2 (Ph-Ph), 6 (Ph-PhOMe) and 9 (MeOPh-PhOMe) under N₂ (left: 0.05% dye loading, right: 0.5% dye loading) at room temperature and 77K. F: total emission spectrum under N₂ (with dominant fluoresence); RTP: room temperature phosphorescence spectrum; LTP: low temperature (77 K) phosphorescence spectrum (λ_ex = 369 nm).

Table 5.

Optical Properties of BF₂bdk Complexes in PLA Matrices Under Nitrogen.

	0.05% BF2bdk/PLA (w/w%)				0.5% BF2bdk/PLA (w/w%)
	λta/nm	λp RTb/nm	λp 77Kb/nm	τpw0c/ms	λta/nm	λp RTb/nm	λp 77Kb/nm	τpw0c/ms
2	414	488	486	513	420	488	488	581
6	422	504	494	440	439	505	510	386
9	426	501	505	402	439	504	513	214
3	439	537	533	440	454	541	542	520
7	431	538	535	463	444	541	544	424
10	427	523	522	603	434	526	527	612

^aTotal emission maxima (i.e. fluorescence peak) under N₂; excited at 369 nm.

^bPhosphorescence emission maxima under N₂; excited at 369 nm.

^cPre-exponential weighted room temperature phosphorescence (RTP) lifetime fit to triple exponential decay. Excitation source: xenon flash lamp.

In the absence of oxygen, the RTP decay of BF₂bdk/PLA blends is complicated and can be fitted with a triple-exponential. Pre-exponential weighted lifetimes, ⁵⁹ τ_pw0, are provided in Table 5. The phosphorescence lifetimes decrease from 2 to 6 and 9 with increasing number of methoxy groups.

Figure 7 shows the total and phosphorescence spectra of 3 (Ph-Np complex), 7 (Np-PhOMe complex) and 10 (Me-Np complex), with 0.05% and 0.5% dye loadings, respectively. Obvious phosphorescence contributions were observed in the total emission spectra for these complexes because of the relatively strong triplet emission. Similar to complexes 2, 6 and 9, the phosphorescence exhibits red-shifted emission as a result of the increase of π conjugation length from complex 10 to complexes 3 and 7 (e.g. 0.05% dye loading: 10: 523 nm, 3: 537 nm, 7: 538 nm). At high dye loading, RTP lifetimes decrease for this series from 10 to 3 to 7 (612, 520, and 424 ms respectively, Table 5) with increased conjugation length. Delayed fluorescence bands are also observed for complexes 3 and 7 at ~450 nm and are slightly stronger at higher loading, given a smaller singlet-triplet splitting, as previously described.²⁴ And again, delayed fluorescence is generally inhibited at low temperature, given decreased thermal back population.^20,35

Figure 7.

Normalized emission spectra for BF₂bdk complexes 3 (Ph-Np), 7 (Np-PhOMe) and 10 (Me-Np) under N₂. (Left: 0.05% dye loading, Right: 0.5% dye loading) at room temperature and 77 K. F: total emission spectrum under N₂ (with dominant fluoresence); RTP: room temperature phosphorescence spectrum under N₂; LTP: low temperature (77K) phosphorescence spectrum. (λ_ex = 369 nm)

Computational studies were also performed for the triplet states. We modeled the phosphorescence with the following simplifying assumptions. Treating the PLA medium as a polarizable continuous medium with the same dielectric constant as dichloromethane, and optimizing the lowest triplet with B3LYP/6-31G(d) to obtain E(T₁), we can approximate the energy of the phosphorescence as either (a) the energy difference between that triplet energy E(T₁) and the E(S₀) energy obtained with B3LYP/6-31G(d) at that same geometry; or (b) the transition energy for the singlet-triplet excitation as computed by TD/6-311+G(d) at the B3LYP/6-31G(d) optimized T₁ geometry. Table 6 shows values of estimates of both types. Linear regression shows a correlation statistic R² = 0.89 between observed phosphorescence wavelengths and values computed by direct difference of B3LYP/6-31G(d) values, and 0.92 between observed values and estimates from TD/6-311+G(d)//B3LYP/6-31G(d) calculations. Though the linearity as measured by the R² statistic is slightly better for the TD calculations than for direct energy differences, the TD estimates are more seriously shifted to the red. The very red results for complexes 4 and 8 suggest that their phosphorescence may lie outside the region monitored experimentally with the present instrumentation and hence, explain why phosphorescence was not observed for these anthryl derivatives.

Table 6.

Phosphorescence Emission Maxima and Computational Estimates of the Singlet Triplet Gaps.

	0.05% in PLA/N2	0.5% in PLA/N2	TD S-T Absorption Wavelengthsa	Wavelengths from the B3LYP S-T Energy Gapb
1	-	-	540	501
2	486	488	564	554
3	533	542	701	621
4	-	-	1105	899
5	466	478	549	510
6	494	505	590	547
7	535	544	691	611
8	-	-	1095	886
9	505	504	590	548
10	522	527	697	616

^aSinglet-triplet excitation energy estimated by TD-B3LYP/6-311+G(d), at the UB3LYP/6-31G(d) optimized T₁ geometry.

^bDirect differences of B3LYP/6-31G(d) for singlet and UB3LYP/6-31G(d) for triplet, all at the UB3LYP/6-31G(d) optimized T₁ geometry.

In summary, when the dye complexes were excited from the ground state to a higher vibrational level of S₁ (or S₂), the majority of the molecules readily relax (10⁻¹² s) to the lowest vibrational level, S₁, through internal conversion. From the S₁ state, excited molecules may decay radiatively in several nanoseconds and exhibit fluorescence. They can also undergo a spin conversion to the triplet state by intersystem crossing and exhibit phosphorescence. Although this and the reverse processes are spin forbidden, under certain conditions, back population can occur by thermal excitation of the singlet state from the triplet state if the thermal energy kT is comparable to the S₁-T₁ energy gap.⁶⁰ Therefore, the energy gap between S₁ and T₁ has an important influence on this reverse conversion. Given the molecular structure of these complexes, we may draw a preliminary conclusion. Complexes with greater π conjugation length (e.g. 3 and 7 versus 10) and donor strength (e.g. 6 and 9 versus 2), are more inclined to undergo reverse intersystem crossing perhaps because a lower energy S₁ level decreases the S₁-T₁ energy gap. For higher dye loading, dye-dye distances are shorter, aggregates may form, and the medium is more polar. Under these circumstances, singlet state energy also decreases while phosphoresence energy is little affected, facilitating intersystem crossing.

Conclusion

In summary, we have synthesized a class of aromatic hydrocarbon substituted difluoroboron diketonate complexes (BF₂bdks) ranging from methyl, phenyl, methoxyphenyl, naphthyl, to anthryl groups on either side of the diketonate ligands. The optical properties of these boron complexes were studied in CH₂Cl₂ and in the solid-state solvent PLA. In CH₂Cl₂, the absorption, fluorescence emission, and fluorescence lifetimes were investigated for all of the BF₂bdks. It was found that almost all complexes exhibit intense fluorescence emission (Φ_F: 0.26–1) except for one molecule, which has a phenyl and methyl groups on the BF₂bdk (1). While one might think that simply extending the conjugation would lead to redshifted dyes, there appear to be two or more factors at play, and properties in PLA environments are different than in solution. The emission color of the series of boron dyes are not only correlated to the length of their π conjugation but also, to the tendency for intramolecular charge transfer (ICT) in solution; larger π conjugation and stronger ICT result in more redshifted emission. In a lifetime study, we discovered that the extent of the ICT processes for these molecules are determined by the electron donating ability of the arene groups on the two sides of the diketonate complexes. A bigger difference in electron donating ability on the two sides correlates with stronger ICT and longer fluorescence lifetime, which is predicted by modeling and calculations. The optical properties of these boron dyes were then investigated in PLA substrates at two different dye loadings, 0.05% and 0.5%, at both room temperature and 77 K. Most of the boron complexes exhibit slightly blue-shifted emission and reduced fluorescence lifetimes in PLA at lower dye loadings, suggesting the hinderance of ICT processes by the polymer medium. At higher dye loadings, however, both spectra and lifetimes representative of aggregate or excimeric species were observed, which is consistent with the previous report²⁴ and has been ascribed to dye-dye interactions in close proximity. These interactions were reduced at 77 K possibly due to the rigidity of the medium, which can limit the configurations required for excimeric species. Delayed emission, namely phosphorescence and delayed fluorescence, are also present for many dyes in PLA environments under nitrogen. Studying the emission properties of these model compounds may inform the rational design of BF₂bdks for practical applications in imaging and sensing.