Boron difluorides with formazanate ligands: redox-switchable fluorescent dyes with large stokes shifts

Abstract

The synthesis of a series of (formazanate)boron difluorides and their 1-electron reduction products is described. The neutral compounds are fluorescent with large Stokes shifts. DFT calculations suggest that a large structural reorganization accompanies photoexictation and accounts for the large Stokes shift. Reduction of the neutral boron difluorides occurs at the ligand and generates the corresponding radical anions. These complexes are non-fluorescent, allowing switching of the emission by changing the ligand oxidation state.

---

Introduction

Fluorescent dyes have attracted increasing interest in various fields of modern research, including biological imaging,¹ molecular probes,² electroluminescent devices³ and photosensitizers.⁴ An ideal fluorescent dye presents high absorption coefficients and emission quantum yield, large Stokes shift, tuneable absorption/emission profiles, as well as high chemical and photochemical stability. Fluorescent dyes with large Stokes shift are more likely to show emission in the long wavelength region (>600 nm), which is desirable for applications in laser printing, information storage, displays and solar power conversion.⁵ Most importantly, fluorescent dyes emitting within the biological window (650–900 nm) are useful in bioimaging applications due to the small light scattering, low background emission and deep penetration into cells and tissues.⁶ Fluorescent dyes with small Stokes shifts usually suffer from self-quenching⁷ which limits the application in high concentration conditions and as solid state materials.

A large class of fluorescent dyes is based on the BF₂ moiety bearing N,N′- or N,O-chelating ligands.⁸ Of these compounds, those with dipyrrin ligands, commonly known as BODIPYs (A in Chart 1),⁹ are the most popular systems due to their stability, high quantum yield and adaptable absorption/emission profiles. The optical properties of the BODIPYs are tuneable by changing the R₁–R₇ substituents of the ligand backbone.^9,10 In addition, the F-substituents of the boron center can also be replaced by aryl, alkyl, and alkoxide groups.¹¹

Chart 1

Even though the largest documented Stokes shift for mono-BODIPYs is 185 nm,¹² for most of these systems the Stokes shift is usually less than 40 nm. Recently, it was shown that incorporation of a triazole moiety on the R₃ position of the BODIPY framework results in a substantial increase in Stokes shift (up to 160 nm).¹³ Also, chelates based on N,O donor atoms lead to dyes with relatively large Stokes shifts.⁸ In an alternative approach, the Stokes shift of fluorescent dyes can be increased by incorporation into an energy transfer fragment in which the energy absorbed by BODIPY unit is transferred to a second fluorophore.^1a,4a A drawback, however, is that this strategy usually requires large synthetic efforts. In order to develop new fluorescent dyes, a variety of other N,N′-chelating ligands have been proposed as shown in Chart 1. Boron difluoride complexes bearing β-diketiminate ligands (B) exhibit strong absorption.¹⁴ In addition, compounds B show larger Stokes shifts (ca. 80 nm) than BODIPYs. A de-symmetrized β-diketiminate analog, the anilido-pyridine (C) framework, was reported by the groups of Piers and Heyne in 2011 and presents large Stokes shifts of 90–120 nm.¹⁵ Further modifications include 1,2-bis(pyrrolylmethylene) hydrazones (BOPHY) (D),¹⁶ and indigo-N,N′-diarylamines (E).¹⁷ Compounds D are highly fluorescent (Φ > 0.99) in solution with Stokes shifts around 40 nm.¹⁶ In addition, they show emission in thin films and as solid powders. Compounds E are a class of redox-active and near-infrared dyes that show Stokes shifts of 30–70 nm.¹⁷

Our group has been interested in the chemistry of formazanate ligands as nitrogen-rich, redox-active analogues of β-diketiminates,^18,19 and we have previously communicated the synthesis of formazanate boron difluoride complexes.²⁰ Concurrently, Gilroy and co-workers reported a series of similar compounds.²¹ In their work, the effect of ligand substituents on the optical properties was evaluated and the application of these compounds in electro-chemiluminescence²² and cell imaging²³ was reported. Recently, Studer and co-workers used verdazyl radicals as “pro-fluorescent” radical probes.²⁴

Here we describe the synthesis and characterization of (formazanate)boron complexes (LBF₂, 1, Scheme 1) as well as the corresponding 1-electron reduction products, the radical anions [LBF₂]˙⁻ (1˙⁻), and discuss the impact of electronic and steric effects on both the optical and redox properties. We show that the neutral compounds 1 are fluorescent with large Stokes shifts, whereas in the radical anions 1˙⁻ the fluorescence is quenched.

Results and discussion

Synthesis and characterisation of formazanate boron complexes

Two synthesis methods for formazanate boron difluoride complexes have been reported by us²⁰ and the Gilroy group,²¹ shown as path A and path B in Scheme 1, respectively. In the present contribution, both methods were used to yield the desired products in moderate to good yields (60–90%). Using these methods, we prepared 9 (formazanate)boron difluoride compounds 1a–i. In addition, we synthesized two compounds in which the (formazanate)BF₂ units are linked by phenylene moiety (1j and 1k). These compounds were recently reported by Gilroy and co-workers while this manuscript was in preparation.²⁵ The formation of complexes 1 was confirmed by NMR spectroscopy, which shows diagnostic 1 : 2 : 1 triplets in the ¹¹B NMR spectra for the 4-coordinate B centres (δ −0.7 to −2.3 ppm).¹⁹ In the ¹⁹F NMR spectra, 1 : 1 : 1 : 1 quartets between −145.6 and −159.4 ppm with J_B–F = 20–30 Hz are observed that are consistent with the presence of a BF₂ moiety.

Scheme 1 Indirect (path A) and direct (path B) methods for the synthesis of (formazanate)boron difluorides (1).

Single crystals suitable for X-ray diffraction analysis were obtained by recrystallization from heptane or hexane (1a,b,d,h,i).† The molecular structure of 1d as a representative example is shown in Fig. 1 whereas Table 1 lists selected geometrical parameters. The solid-state structures of all compounds show four-coordinate boron centres bound to the two terminal N atoms of a formazanate ligand to form a six-membered chelate ring, similar to those observed before.^20,21 In most cases, the observed C–N and N–N bond lengths are in the range of 1.33–1.36 Å and 1.29–1.31 Å, respectively, consistent with a delocalized ligand backbone. For the derivative with electronically dissimilar N–Ar substituents (1d, R₁ = Ph and R₅ = C₆F₅) the data in Table 1 indicate a more localized bonding picture: the N–N bond length adjacent to the electron-withdrawing C₆F₅ group is significantly shorter (N1–N2: 1.299(2) Å) than that vicinal to the Ph substituent (N3–N4: 1.329(2) Å). All the boron centres in 1a, 1b, 1d, 1h and 1i show a distorted tetrahedral geometry, and the boron centre is significantly displaced from the formazanate backbone, with values ranging from 0.17–0.59 Å.

Fig. 1 Molecular structure of 1d (left) and 1hBCF (right). Thermal ellipsoids are shown at 50% probability, and hydrogen atoms are removed for clarity.

Table 1 Selected bond lengths (Å) and bond angles (°) of LBF₂ complexes (1) and their radical anions [1]˙⁻

	1a	1b	1d	1h	1i	1hBCF	[Na(15-c-5)]^+ [1c]˙^− ^d	[Cp2Co]^+ [1f]˙^−	[Cp*2Co]2 [1j]˙^2−
a Displacement of the B atom from the least-squares plane defined by N1, N2, N3 and N4. b Angle between the planes defined by the N–Ar aromatic rings and the ligand backbone. c C10–N2 instead of C7–N2. d Values reported for one of the two independent molecules. e Data taken from ref. 20.
N1–N2	1.308(1)	1.3177(12)	1.2993(18)	1.294(1)	1.2931(15)	1.307(4)	1.376(3)	1.374(2)	1.3599(19)
N3–N4	1.308(1)		1.3289(17)	1.295(1)		1.300(4)	1.370(3)	1.356(2)	1.3620(18)
C7–N2	1.346(1)	1.3394(12)	1.361(2)	1.344(1)	1.3442(13)^c	1.340(5)	1.337(4)	1.329(3)	1.335(2)
C7–N3	1.343(1)		1.329(2)	1.340(2)		1.339(5)	1.337(4)	1.333(2)	1.338(2)
N1–B1	1.559(2)	1.5523(15)	1.577(2)	1.573(2)	1.5665(16)	1.562(5)	1.520(4)	1.533(3)	1.541(2)
N4–B1	1.552(2)		1.548(2)	1.580(2)		1.553(5)	1.510(4)	1.512(3)	1.536(2)
N1–B1–N4	102.40(9)	100.15(11)	101.52(12)	105.87(9)	103.76(14)	104.4(3)	107.4(2)	108.49(17)	108.46(13)
B out-of-plane^a	0.50	0.59	0.51	0.17	0.29	0.49	0.38	0.08	0.05
Dihedral angles^b	47.98	47.97	47.54	24.16	88.11	41.53	27.54	12.64	4.01
	42.00		50.09	22.35		44.39	65.05	83.44	13.23

---

The N-aryl substituents are rotated out of the plane of the ligand backbone with dihedral angles ranging from 23–50° for the compounds with unhindered Ar groups. The increased steric interactions in the N-Mes derivative 1i result, as expected, in a perpendicular orientation (dihedral angle of 88°).

Compounds with cyano substituents (1h,i) offer the possibility to influence the optical and redox properties by coordination of Lewis acids. To probe this, the B(C₆F₅)₃-adduct 1hBCF was prepared by stirring 1h with B(C₆F₅)₃ in toluene solution. Indicative of the formation of 1hBCF is the appearance of 3 additional resonances in the ¹⁹F NMR spectrum at −134, −155, and −163 ppm due to the C₆F₅ rings. An X-ray structure determination (molecular structure in Fig. 1, pertinent bond distances and angles in Table 1) confirms the coordination of the Lewis acidic B(C₆F₅)₃ fragment to the cyano group.

UV-Vis absorption and emission spectroscopy

The optical properties of the formazanate boron complexes 1 were studied by UV-Vis absorption and emission spectroscopy in THF solution (Fig. 2 and Table 2). All compounds show medium to strong absorption bands with extinction coefficients between 9700 and 30 200 L mol⁻¹ cm⁻¹ in the visible range of the spectrum (between 410 and 520 nm). The spectral trends observed here are very similar to those described for bis(formazanate)zinc complexes.¹⁹ Introducing an electron-donating ^tBu-group (1b) instead of a p-tolyl moiety (1a) results in a blueshift of the absorption maximum (517 and 473 nm for 1a and 1b, respectively).

Fig. 2 UV-Vis absorption spectra (top) and normalised emission spectra (bottom) of 1c, 1f, and 1g. Data were collected in 10⁻⁵ M dry THF solution. The excitation wavelength of emission spectra is at 473 nm.

Table 2 Optical properties of formazan boron complexes 1 in THF solution

	λ max (nm)	ε (M^−1 cm^−1)	λ em ^c (nm)	QY (%)	SS (nm)	SS (cm^−1)
a Taken from ref. 21b. b Data collected in toluene solution. c Excitation wavelength = 473 nm.
1a	517	13 736	641	0.2	124	3742
1b	473	13 220	643	<0.1	170	5590
1c	464	21 306	617	<0.1	153	5344
1d	482	12 523	640	<0.1	158	5122
1e	460	12 139	610	0.1	150	5346
1f	431	13 702	612	<0.1	181	6862
1g	414	11 852	650	<0.1	236	8770
1ha	489	25 400	585	5	96	3356
1h ^,	502	30 400	586	15	84	2855
1hBCF ^b	502	20 582	632	0.5	130	4098
1i	428	9662	643	0.2	215	7812
1j	523	30 227	664	0.6	141	4060
1k	508	16 768	631	1.3	123	3837

---

In the case of the addition of an electron-withdrawing C₆F₅-group (1f, 1g) at the R₃ position instead of a p-tolyl moiety (1c, 1e), the absorption maximum undergoes a blueshift as well (431 nm for 1fvs. 464 nm for 1c; 414 nm for 1gvs. 460 m for 1e). Upon substitution of phenyl for mestiyl at the R₁ or R₅ positions, the λ_max is again blueshifted (489 nm for 1hvs. 428 nm for 1i). This likely results from the nearly perpendicular orientation of the mesityl group, which doesn't allow conjugation between the formazanate core and the side mestiyl ring, thereby limiting the size of the π-conjugated system.^21c The introduction of the Lewis acid B(C₆F₅)₃ in 1hBCF surprisingly does not influence λ_max. The UV-Vis absorption spectra of the di-formazanate system (1j and 1k) are very similar to 1a except for the expected ca. two-fold increase in molar extinction coefficients.

Similar to other reported boron difluoride complexes, compounds 1 were shown to be emissive, with emission wavelengths λ_em in the 580 nm to 670 nm range in THF solution (Fig. 2). The reported (formazanate)boron difluoride complexes usually show large Stokes shifts (100–150 nm).^21,23 For most of the compounds reported here, the Stokes shifts are even larger (>150 nm). As an example, 1g shows λ_max at 414 nm and λ_em at 650 nm, a Stokes shift of 236 nm (8770 cm⁻¹), which to the best of our knowledge is the highest value reported to date for this class of compounds. The large Stokes shift of compounds 1c–g might be due to the asymmetry in the substitution pattern of the ligands. The B(C₆F₅)₃ adduct 1hBCF shows a redshift in its emission spectrum (λ_max = 632 nm) relative to 1h (586 nm), which increases the Stokes shift from 84 nm (4098 cm⁻¹) to 130 nm (4098 cm⁻¹) in toluene solution. These results suggest that the emission profile of (3-cyanoformazanate)boron difluoride complexes is tuneable via binding of Lewis acids. The phenylene-linked systems 1j and 1k show emission spectra that are redshifted and blueshifted, respectively, compared to those of the parent compound 1a, whereas the obtained quantum yields are higher.

Cyclic voltammetry

The redox chemistry of compounds 1 was studied by cyclic voltammetry (CV) in tetrahydrofuran (THF) or 1,2-dichloroethane (DCE) solution under nitrogen atmosphere. All complexes show two (quasi)reversible redox processes corresponding to the formation of the radical anions [1]^˙− and dianions [1]²⁻. The results of cyclic voltammetry studies are summarized in Table 3. From these data it is apparent that the redox potentials of (formazanate)BF₂ complexes can be altered over a wide range (up to 530 mV) by changing the substituents on the ligand framework. The measured redox-potentials correlate with the electron-donating ability of the R₃-substituent via inductive effects: an electron-donating t-butyl group (1b) induces a shift to more negative potential, while electron-withdrawing cyano (1h) or C₆F₅ substituents (1f) result in an anodic shift, in comparison to R₃ = p-tolyl (1a). Changing the nature of the aromatic N-substituent(s) from Ph to an electron-withdrawing C₆F₅ group (1avs.1d, 1cvs.1e and 1fvs.1g) shifts the redox potentials in the positive direction by 130–180 mV. As expected, substitution of a phenyl group for an electron-donating mesityl group (1avs.1c, 1dvs.1e and 1hvs.1i) will shift the measured redox potentials to more negative values by around 240 mV. Surprisingly, coordination of the Lewis acid B(C₆F₅)₃ in 1hBCF induces only a marginal variation of the first redox potential from −0.66 V (1h) to −0.67 V (1hBCF). This might be related to the high concentration of electrolyte: the B(C₆F₅)₃ group in 1hBCF does not interact strongly in the high ionic strength solvent system used for collecting the CV data. The LBF₂^0/−1 redox couple in 1j and 1k is observed at potentials similar to those in 1a, but corresponds to two sequential one-electron transfers as evidenced by the somewhat larger peak-to-peak separation in 1j and 1k (0.268–0.287 V, see Fig. 3) than in 1a (0.222 V). At more negative potential, 1j and 1k show two very close one-electron processes (between −2.0 and −2.4 V vs. Fc/Fc⁺). In the case of 1j, these are poorly resolved but for 1k there are clearly two separate events, indicating that the two (formazanate)BF₂ units in these phenylene-linked systems do not behave fully independently.²⁶ Gilroy and co-workers described similar electrochemical data for these compounds in dichloromethane solution. While the 2-electron reduction products appeared unstable in dichloromethane solution,²⁵ our data in THF suggest that in this solvent the products are stable and might be isolable (vide infra).

Fig. 3 Cyclic voltammograms of 1j and 1k recorded at 0.1 V s⁻¹ in 1.5 mM THF solution containing 0.1 M tetrabutylammonium hexafluorophosphate.

Table 3 Electrochemical Data (V vs. Fc/Fc⁺) of LBF₂ complexes^a

	LBF2^0/−1	LBF2^−1/−2
a Cyclic voltammetry experiments collected in THF solution (1.5 mM 1 and 0.1 M [Bu4N][PF6] as supporting electrolyte) at a scan rate if 0.1 V s^−1. The data were referenced internally against the Fc/Fc^+ couple. b Collected in DCE solution.
1a	−0.98	−2.06
1b	−1.08	−2.21
1c	−1.19	−2.34
1d	−0.85	−1.99
1e	−1.01	−2.26
1f	−1.02	−2.25
1g	−0.84	−2.17
1h	−0.66	−1.83
1h	−0.65	−1.76
1hBCF ^b	−0.67	−1.75
1i	−0.90	−2.41
1j	−0.95	−2.0 to −2.3
1k	−0.94	−2.0 to −2.4

---

Synthesis and characterisation of radical anions [1]˙⁻

According to the CV data, the radical anions [1]˙⁻ are relatively stable and could be accessible chemically. Indeed, treatment of compounds 1 with Cp₂Co precipitated the dark green salts [Cp₂Co]⁺[1]˙⁻. For the phenylene-linked bis(formazanate) systems 1j and 1k, treatment with 1 equiv. of reducing agent resulted in precipitation of 0.5 equiv. of the corresponding diradical dianions [1j,k]˙²⁻ and 0.5 equiv. of unreacted starting material instead of the radical anions [1j,k]˙⁻. This is presumably driven by the poor solubility of the dianionic products. In order to obtain crystalline products, a variety of reducing agents was used; a representative series ([1a]˙⁻,²⁰ [1c]˙⁻, [1f]˙⁻ and [1j]˙²⁻) could be obtained in high yield with either Cp₂Co⁺ ([1a]˙⁻ and [1f]˙⁻), Na(15-crown-5)⁺ ([1c]˙⁻) or Cp*₂Co⁺ ([1j]˙²⁻) as countercations. For all compounds, significant changes were observed in the metrical parameters of the reduced compounds in comparison to the neutral precursors. The most obvious change is the elongation of the N–N bonds (from 1.30–1.31 Å in 1 to 1.34–1.37 Å in [1]˙⁻) due to population of a π*-orbital that has N–N antibonding character.^19a,20 The B–N bond lengths decrease upon reduction (1a: 1.559(2)/1.552(2) Å vs. [1a]˙⁻: 1.532(3)/1.536(3) Å), as anticipated. In addition to the changes of the N–N and B–N bond lengths, the formation of [1]˙⁻ is accompanied by a substantial planarization of the system: the B atom is virtually in the plane of the ligand (displacement <0.09 Å for both [1a]˙⁻ and [1f]˙⁻) and the unhindered N-Ph substituents become coplanar with the formazanate backbone. The slightly larger boron displacement in [1c]˙⁻ (0.376 Å) is due to an asymmetric interaction between the Na⁺ cation and the F atoms (2.383(4) and 2.419(4) Å) (Fig. 4).

Fig. 4 Molecular structures of [Na(15-crown-5)]⁺[1c]˙⁻ (left) and [Cp*₂Co]⁺₂[1j]˙²⁻ (right) showing 50% probability ellipsoids; cations in the latter omitted for clarity.

The metrical parameters of each of the (formazanate)boron fragments in the para-phenylene linked compound [1j]˙²⁻ are similar to that in [1a]˙⁻. These data suggest that [1j]˙²⁻ is best represented as two nearly-independent [LBF₂]˙⁻ units, and a significant quinoidal contribution to the structure of the phenylene linker can be excluded.²⁷

EPR spectroscopy for the radical anions [1]˙⁻ shows broad, featureless resonances at g ∼ 2, indicative of ligand-based radicals. While related organic verdazyl radicals show hyperfine interactions with the N nuclei,^24,28 these are absent in the boron systems studied here. DFT calculations (see the ESI†) suggest that the hyperfine coupling constants in [1a]˙⁻ are indeed smaller for the boron compounds reported here than those in organic analogues. Additionally, hyperfine interactions with the boron/fluorine nuclei are present in the compounds [1a]˙⁻, which accounts for the broad signal with unresolved hyperfine interactions. The phenylene-linked systems [1j]˙²⁻ and [1k]˙²⁻ are diradicals similar to linked bis(verdazyl) radicals that have been reported,^26,29 but in contrast to these purely organic systems our boron-based diradicals do not show zero-field splitting in their EPR spectra. This implies that the spin centres behave independently, effectively leading to S = 1/2 behaviour without intramolecular electronic coupling (J ∼ 0).³⁰ EPR spectra were collected in MeTHF/DMSO glass in the range between 5 and 40 K to evaluate the spin ground state of these compounds (Fig. S1 in the ESI†). Even though the absence of zero-field splitting indicates little or no electronic coupling, for the para-phenylene linker the signal intensity decreases upon cooling to 5 K. This suggests the two radical centres to be anti-ferromagnetically coupled (J < 0) leading to a singlet biradical ground-state. For the meta-phenylene linked compound [1k]˙²⁻, the signal broadens upon cooling but its intensity does not change significantly and a conclusive assignment of the electronic ground-state cannot be made. DFT calculations for both systems (see the ESI†) agree with the VT EPR data that [1j]˙²⁻ has a singlet biradical ground state. On the other hand, [1k]˙²⁻ is calculated to have a triplet ground state, which is in line with meta-phenylene linked bis(verdazyl) radicals reported in the literature.²⁶

Optical properties of the radical anions [1]˙⁻

UV/Vis absorption spectroscopy of compounds [1]˙⁻ shows two absorption bands in the visible range of the spectrum (see Table S3 in the ESI†). One of these is shifted to lower energy (up to 716 nm for [1a]˙⁻, Fig. 5) in comparison to the neutral precursors and is indicative of the presence of a formazanate ligand in the radical dianionic form (L˙²⁻).^18–20 We were interested to evaluate the influence of the redox-state of the compounds on their emission properties, and thus emission spectra of [1a]˙⁻ were collected. Regardless of the excitation wavelength, the compound was only very weakly emissive and the observed spectrum indicates that λ_em is identical to that of the neutral precursor 1a. Based on this, we conclude that a small amount of decomposition to 1a is responsible for the observed spectrum, and the radical anions [1]˙⁻ are non-emissive.

Fig. 5 UV-Vis absorption spectrum of compounds 1a (dotted red line) and [1a]˙⁻ (green line) in THF solution.

Theoretical calculations

We have performed calculations in which the ground- and excited-states structures were respectively determined with DFT and Time-Dependent DFT (TD-DFT), whereas the transition energies were obtained with both TD-DFT and SOS-CIS(D) model, corrected for solvent effects (see the ESI† for details). This approach was shown previously to be adequate for fluoroborate complexes.³¹ In order to demonstrate the ability of this selected theoretical level in reproducing the structure, properties and electronic absorption spectra of formazanate boron complexes, systems 1a and [Cp₂Co]⁺[1a]˙⁻ have been chosen as representative examples. Inspection of the theoretically determined structure of 1a, shows that the B atom is displaced with respect to the formazanate backbone, as indicated by the dihedral angle (B1,N1,N2,N3) of 16.29°. Moreover, the DFT-computed C–N, N–N and B–N bond lengths of the neutral 1a system are 1.340 Å, 1.286 Å and 1.562 Å, respectively, showing electron delocalization along the chelating formazanate ligand. These values match well their experimental counterparts, though theory undershoots the N–N bond length (Table 1). We have determined the optimal structure of the lowest singlet excited-state as well. It shows a nearly perfectly planar core with a (B1,N1,N2,N3) dihedral of 0.03°. In the excited-state, the computed C–N, N–N and B–N bond lengths attain 1.345 Å, 1.320 Å and 1.567 Å, respectively, indicating a strong elongation of the N–N bonds. These significant geometrical changes are consistent with the sizeable Stokes shifts that have been measured experimentally (see Table 1). Passing to the theoretically determined ground-state structure of the [1a]˙⁻ radical anion, large structural changes are also observed, in agreement with the experimental trends. Theory correctly predicts a significant planarization of the boron atom with respect to the plane of the formazanate ligand, resulting in a dihedral angle (B1,N1,N2,N3) value of 4.93°, much smaller than in 1a. Moreover, the experimentally observed elongation of the N–N bond and the simultaneous shortening of the B–N bond upon passing from neutral to radical anionic forms is also qualitatively reproduced by the theoretical calculations. Indeed, DFT foresees that the former goes from 1.286 Å to 1.343 Å and the latter from 1.562 Å to 1.543 Å. To investigate the importance of the counterion, we have also optimized the 1 : 1 [Cp₂Co]⁺[1a]˙⁻ complex, and obtained similar trends but quite different values, i.e., the dihedral is 9.34°, the N–N bond lengths are 1.342 Å and 1.347 Å, whereas the B–N distances are 1.539 Å and 1.541 Å. It has to be noted that while in the solid-state it is in principle possible to observe changes in the planarity of the two phenyl substituents with respect to the central ring upon reduction of 1a, in the theoretical liquid-phase calculations the essentially free rotation of the phenyl rings makes analogous comparisons less relevant. Clearly, DFT allows a reasonable determination of the ground-state equilibrium geometries of formazanate boron complexes.

In order to access the predictive capability of our model in reproducing the electronic absorption spectra of the neutral formazanate boron complexes, we have determined the vertical transition energies of all 11 1x [x = a,b…] compounds in Table 1 and the obtained results are listed in Table 4. We note that the chosen theoretical level, SOS-CIS(D), can nicely reproduce both the absolute experimental values and the trends upon substitution whereas TD-DFT overestimates the transition energies, as expected.³¹ For the full set of compounds, the mean signed error is −4.7 nm and the mean absolute error is 15.1 nm when SOS-CIS(D) is applied. Clearly, these values are rather small. At the same time, the linear correlation coefficient relating theoretical and experimental values is large (0.97).

Table 4 Experimental and theoretical wavelengths of maximum intensity for the absorption bands of systems [1x]. We show both the TD-DFT values and their SOS-CIS(D) counterparts

	λ ^TD-DFTmax (nm)	λ ^SOS-CIS(D)max (nm)	λ ^Expmax (nm)
1a	463	543	517
1b	418	485	473
1c	417	462	464
1d	445	501	482
1e	420	451	460
1f	400	434	431
1g	407	400	414
1h	414	486	502
1i	373	412	428
1j	469	545	523
1k	463	535	508

---

Let us now turn to the radical anions [1a]˙⁻, the experimentally recorded UV-Vis spectra of both 1a and its radical anionic form in THF are shown in Fig. 5. One clearly notes the emergence of two bands in the visible range for the latter structure. Indeed, using TD-DFT, we computed two intense bands at 595 nm and 399 nm for [1a]˙⁻ which nicely surround the 463 nm values computed for 1a at the same level of theory. This fits well with the experimental trends. Indeed, the experiment reveals that the lowest-energy band of the neutral precursor 1a is significantly redshifted by ca. 198 nm upon passing to the radical anion, whereas theory underestimates the redshift (132 nm). For the second band, the experimental blueshift compared to the main band of 1a is ca. −59 nm, and theory provides −64 nm. Test calculations performed with the [Cp₂Co]⁺[1a]˙⁻ complex demonstrated that the counter-ion has a rather marginal impact on these variations.

In Fig. 6 the occupied-virtual molecular orbital (MO) pairs that characterise the most intense theoretically determined electronic transitions for both 1a and 1a˙⁻ are shown. For 1a the lowest state can be mainly ascribed to a HOMO–LUMO transition, as expected. The pictures of the electron density redistribution that accompanies photoexcitation show that reduction of the neutral 1a compound does not change the nature of the transition (of HOMO to SOMO nature in 1a˙⁻) involved in the lowest-energy band of the absorption spectrum but merely shifts its position, in agreement with the experimental suggestion. Indeed, for both 1a and 1a˙⁻ the lowest-energy transition is characterized by withdrawal of electron density from the p-tolyl group and its redistribution on the rest of the system. In contrast, the long wavelength band of 1a˙⁻ is characterized by electron density withdrawal from the formazanate-containing ring and two phenyls to the p-tolyl group.

Fig. 6 Occupied-virtual molecular orbital (MO) pairs that characterize the most intense theoretically determined electronic transitions for compounds 1a and 1a˙⁻. The numbering of the bands is based on the experimentally determined absorption spectra for both systems. Contour threshold: 0.03 a.u.

Conclusions

We have prepared a series of (formazanate)boron difluoride complexes and evaluated their structural, electrochemical and luminescent properties. The neutral compounds (1) show an intense absorption band in the visible range, which TD-DFT calculations confirm to be due to electronic transitions (π → π*) within the ligand. Evaluation of their luminescence properties shows large Stokes shifts across the series, albeit with low quantum yields. In case of the cyano-substituted formazans, the emission wavelength may be further increased by coordination of a Lewis acid (as shown for 1hBCF). For a representative series of compounds, the 1-electron reduction products (1˙⁻) were characterised, which show two absorption maxima in the visible that occur at both higher and lower energy than in the neutral precursor. Theoretical calculations indicate that, similar to the experimentally verified structures of 1-electron reduction products, photochemical excitation to the lowest singlet excited state results in large structural reorganization/planarization following the partial population of the N–N π*-orbitals. Due to the non-emissive nature of the radical anions 1˙⁻, the formazanate boron difluoride compounds presented here allow for redox-switching of their luminescent properties, and may find application in sensing and imaging applications.

Acknowledgements

E. O. is grateful to the Netherlands Organisation for Scientific Research (NWO) for a Veni grant. D. J. acknowledges the European Research Council (ERC) for financial support in the framework of a starting grant (Marches-278845). A. C. thanks the ERC (Marches-278845) for his postdoctoral grant. This research used resources of the GENCI-CINES/IDRIS, of the CCIPL, and of a local Troy cluster. We thank Prof. Wesley Browne for access to spectroscopic facilities and help with quantum yield determinations and Prof. Bas de Bruin for VT-EPR data.

Notes and references

1. (a) A. Martin, R. D. Moriarty, C. Long, R. J. Forster and T. E. Keyes, Asian J. Org. Chem., 2013, 2, 763–778; (b) K. Sreenath, Z. Yuan, J. R. Allen, M. W. Davidson and L. Zhu, Chem. – Eur. J., 2015, 21, 867–874; (c) A. Yamada, Y. Hiruta, J. Wang, E. Ayano and H. Kanazawa, Biomacromolecules, 2015, 16, 2356–2362; (d) P. Agarwal, B. J. Beahm, P. Shieh and C. R. Bertozzi, Angew. Chem., Int. Ed., 2015, 54, 11504–11510.
2. (a) T. W. Hudnall and F. P. Gabbai, Chem. Commun., 2008, 4596–4597; (b) M. V. Kuperman, S. V. Chernii, M. Y. Losytskyy, D. V. Kryvorotenko, N. O. Derevyanko, Y. L. Slominskii, V. B. Kovalska and S. M. Yarmoluk, Anal. Biochem., 2015, 484, 9–17.
3. (a) A. M. Prokhorov, T. Hofbeck, R. Czerwieniec, A. F. Suleymanova, D. N. Kozhevnikov and H. Yersin, J. Am. Chem. Soc., 2014, 136, 9637–9642; (b) X. Cao, J. Miao, M. Zhu, C. Zhong, C. Yang, H. Wu, J. Qin and Y. Cao, Chem. Mater., 2015, 27, 96–104; (c) L. Yao, S. Zhang, R. Wang, W. Li, F. Shen, B. Yang and Y. Ma, Angew. Chem., Int. Ed., 2014, 53, 2119–2123.
4. (a) S. D. Topel, G. T. Cin and E. U. Akkaya, Chem. Commun., 2014, 50, 8896–8899; (b) T. Yogo, Y. Urano, Y. Ishitsuka, F. Maniwa and T. Nagano, J. Am. Chem. Soc., 2005, 127, 12162–12163.
5. (a) S. Erten-Ela, M. D. Yilmaz, B. Icli, Y. Dede, S. Icli and E. U. Akkaya, Org. Lett., 2008, 10, 3299–3302; (b) A. Mishra, M. K. R. Fischer and P. Bäuerle, Angew. Chem., Int. Ed., 2009, 48, 2474–2499.
6. V. Ntziachristos, J. Ripoll, L. V. Wang and R. Weissleder, Nat. Biotechnol., 2005, 23, 313–320.
7. (a) S. J. Farley, D. L. Rochester, A. L. Thompson, J. A. K. Howard and J. A. G. Williams, Inorg. Chem., 2005, 44, 9690–9703; (b) W. Qin, M. Baruah, M. Van der Auweraer, F. C. De Schryver and N. Boens, J. Phys. Chem. A, 2005, 109, 7371–7384.
8. D. Frath, J. Massue, G. Ulrich and R. Ziessel, Angew. Chem., Int. Ed., 2014, 53, 2290–2310.
9. (a) A. Loudet and K. Burgess, Chem. Rev., 2007, 107, 4891–4932; (b) G. Ulrich, R. Ziessel and A. Harriman, Angew. Chem., Int. Ed., 2008, 47, 1184–1201; (c) N. Boens, V. Leen and W. Dehaen, Chem. Soc. Rev., 2012, 41, 1130–1172.
10. (a) J. Rosenthal and S. J. Lippard, J. Am. Chem. Soc., 2010, 132, 5536–5537; (b) J.-J. Shie, Y.-C. Liu, Y.-M. Lee, C. Lim, J.-M. Fang and C.-H. Wong, J. Am. Chem. Soc., 2014, 136, 9953–9961; (c) C. Cheng, N. Gao, C. Yu, Z. Wang, J. Wang, E. Hao, Y. Wei, X. Mu, Y. Tian, C. Ran and L. Jiao, Org. Lett., 2015, 17, 278–281.
11. (a) T. Lundrigan, S. M. Crawford, T. S. Cameron and A. Thompson, Chem. Commun., 2012, 48, 1003–1005; (b) G. Duran-Sampedro, I. Esnal, A. R. Agarrabeitia, J. Bañuelos Prieto, L. Cerdán, I. García-Moreno, A. Costela, I. Lopez-Arbeloa and M. J. Ortiz, Chem. – Eur. J., 2014, 20, 2646–2653; (c) T. Lundrigan, T. S. Cameron and A. Thompson, Chem. Commun., 2014, 50, 7028–7031; (d) A. B. More, S. Mula, S. Thakare, N. Sekar, A. K. Ray and S. Chattopadhyay, J. Org. Chem., 2014, 79, 10981–10987.
12. A. Martin, C. Long, R. J. Forster and T. E. Keyes, Chem. Commun., 2012, 48, 5617–5619.
13. J. C. Er, M. K. Tang, C. G. Chia, H. Liew, M. Vendrell and Y.-T. Chang, Chem. Sci., 2013, 4, 2168–2176.
14. (a) F. P. Macedo, C. Gwengo, S. V. Lindeman, M. D. Smith and J. R. Gardinier, Eur. J. Inorg. Chem., 2008, 3200–3211; (b) R. Yoshii, A. Hirose, K. Tanaka and Y. Chujo, J. Am. Chem. Soc., 2014, 136, 18131–18139; (c) R. Yoshii, A. Hirose, K. Tanaka and Y. Chujo, Chem. – Eur. J., 2014, 20, 8320–8324; (d) S. M. Barbon, V. N. Staroverov, P. D. Boyle and J. B. Gilroy, Dalton Trans., 2014, 43, 240–250.
15. J. F. Araneda, W. E. Piers, B. Heyne, M. Parvez and R. McDonald, Angew. Chem., Int. Ed., 2011, 50, 12214–12217.
16. (a) C. Yu, L. Jiao, P. Zhang, Z. Feng, C. Cheng, Y. Wei, X. Mu and E. Hao, Org. Lett., 2014, 16, 3048–3051; (b) I.-S. Tamgho, A. Hasheminasab, J. T. Engle, V. N. Nemykin and C. J. Ziegler, J. Am. Chem. Soc., 2014, 136, 5623–5626; (c) Q. Huaulmé, A. Mirloup, P. Retailleau and R. Ziessel, Org. Lett., 2015, 17, 2246–2249.
17. (a) G. Nawn, R. McDonald and R. G. Hicks, Inorg. Chem., 2013, 52, 10912–10919; (b) G. Nawn, S. R. Oakley, M. B. Majewski, R. McDonald, B. O. Patrick and R. G. Hicks, Chem. Sci., 2013, 4, 612–621.
18. J. B. Gilroy, M. J. Ferguson, R. McDonald, B. O. Patrick and R. G. Hicks, Chem. Commun., 2007, 126–128.
19. (a) M.-C. Chang, T. Dann, D. P. Day, M. Lutz, G. G. Wildgoose and E. Otten, Angew. Chem., Int. Ed., 2014, 53, 4118–4122; (b) M.-C. Chang, P. Roewen, R. Travieso-Puente, M. Lutz and E. Otten, Inorg. Chem., 2015, 54, 379–388.
20. M. C. Chang and E. Otten, Chem. Commun., 2014, 50, 7431–7433.
21. (a) S. M. Barbon, J. T. Price, P. A. Reinkeluers and J. B. Gilroy, Inorg. Chem., 2014, 53, 10585–10593; (b) S. M. Barbon, P. A. Reinkeluers, J. T. Price, V. N. Staroverov and J. B. Gilroy, Chem. – Eur. J., 2014, 20, 11340–11344; (c) S. M. Barbon, V. N. Staroverov and J. B. Gilroy, J. Org. Chem., 2015, 80, 5226–5235.
22. M. Hesari, S. M. Barbon, V. N. Staroverov, Z. Ding and J. B. Gilroy, Chem. Commun., 2015, 51, 3766–3769.
23. R. R. Maar, S. M. Barbon, N. Sharma, H. Groom, L. G. Luyt and J. B. Gilroy, Chem. – Eur. J., 2015, 21, 15589–15599.
24. D. Matuschek, S. Eusterwiemann, L. Stegemann, C. Doerenkamp, B. Wibbeling, C. G. Daniliuc, N. L. Doltsinis, C. A. Strassert, H. Eckert and A. Studer, Chem. Sci., 2015, 6, 4712–4716.
25. S. M. Barbon, J. T. Price, U. Yogarajah and J. B. Gilroy, RSC Adv., 2015, 5, 56316–56324.
26. J. B. Gilroy, S. D. J. McKinnon, P. Kennepohl, M. S. Zsombor, M. J. Ferguson, L. K. Thompson and R. G. Hicks, J. Org. Chem., 2007, 72, 8062–8069.
27. L. K. Montgomery, J. C. Huffman, E. A. Jurczak and M. P. Grendze, J. Am. Chem. Soc., 1986, 108, 6004–6011.
28. R. G. Hicks, in Stable Radicals, John Wiley & Sons, Ltd, 2010, pp. 245–279.
29. (a) K. Mukai, N. Azuma, H. Shikata and K. Ishizu, Bull. Chem. Soc. Jpn., 1970, 43, 3958–3960; (b) R. M. Fico, M. F. Hay, S. Reese, S. Hammond, E. Lambert and M. A. Fox, J. Org. Chem., 1999, 64, 9386–9392.
30. M. Abe, Chem. Rev., 2013, 113, 7011–7088.
31. B. Le Guennic and D. Jacquemin, Acc. Chem. Res., 2015, 48, 530–537.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental section, X-ray data, computational methods. CCDC 1471287–1471289, 1471291–1471293 and 1471295. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c6dt01226d

---