A ratiometric, fluorescent BODIPY-based probe for transition and heavy metal ions

Abstract

A novel metal ion-sensitive fluorescent probe – 4,4-difluoro-8-(4-methylphenyl)-5-(phenylethynyl)-3-[bis(pyridin-2-ylmethyl)amino]-4-bora-3a,4a-diaza-s-indacene – based on the BODIPY platform with di(2-picolyl)amine as chelator has been synthesized and spectroscopically and photophysically characterized. The generalized treatment of the solvent effect shows that solvent dipolarity is primarily responsible for the observed shifts of the absorption and fluorescence emission maxima. Complex formation with various metal ions is investigated in acetonitrile solution by means of spectrophotometric and fluorometric titrations. The BODIPY indicator forms 1 : 1 complexes with several transition metal (Ni²⁺, Cu²⁺, Zn²⁺) and heavy metal (Cd²⁺, Hg²⁺) ions, producing large bathochromic shifts in the absorption and fluorescence spectra and, except for Ni²⁺, cation-induced fluorescence amplifications. The dissociation constants of the metal ion complexes range from 4 μM for Hg²⁺ to 48 μM for Zn²⁺.

---

Introduction

Fluorescent probes for the measurement of concentrations of analytically and biologically important ions are essential tools in numerous fields of modern medicine and materials science.^1,2 The design and synthesis of fluorescent indicators with high selectivity and sensitivity for metal ions remain an exciting field of supramolecular chemistry. Of major importance are fluoroionophores targeting transition and heavy metal ions (such as Ni²⁺, Cu²⁺, Zn²⁺, Cd²⁺, Hg²⁺, …). Ratiometric fluorescent probes which exhibit shifts in the absorption and/or fluorescence spectra upon coordination of the analyte of interest are preferred over probes with on/off switching of fluorescence. Indeed, measuring the ratio of the fluorescence signals at two suitably chosen wavelengths avoids the influence of many extraneous artifacts in fluorescence changes at a single wavelength. Therefore, the design, synthesis and spectroscopic characterization of ratiometric fluorescent chemosensors have emerged as an important area of supramolecular chemistry.

Here we describe a new ratiometric, fluorescent probe for transition and heavy metal ions based on the 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY or boron dipyrrin or boron dipyrromethene)^3,4 platform. Lately there is intense research interest in the synthesis and photophysical/spectroscopic properties of BODIPY-based fluorescent dyes and their application in selective and sensitive fluorescent chemosensors.^4c,d As a result, there are now thousands of different BODIPY dyes described in the literature with exciting structural variations. BODIPY is an outstanding fluorophore characterized by valuable properties including bright fluorescence [due to the combination of a high fluorescence quantum yield Φ with a large molar absorption coefficient ε(λ)] with absorption and fluorescence emission bands in the visible range, robustness towards chemicals and light, and generally a negligible intersystem-crossing.⁴ Moreover, the BODIPY core can be postfunctionalized easily (at the pyrrole carbons, the meso-carbon and the boron atom), leading to dyes with custom-made, fine-tuned spectroscopic properties for use in bioscience and material research.⁵

The transition and heavy metal ions of which the fluorometric/spectrophotometric detection is investigated are Cd²⁺, Hg²⁺, Ni²⁺, Cu²⁺ and Zn²⁺. As chelator for these ions, we chose bis(pyridin-2-ylmethyl)amine [commonly known as di(2-picolyl)amine, DPA], linked to the BODIPY core at the 3-position. Di(2-picolyl)amine is a chelator of several metal ions, including Mn²⁺, Fe²⁺, Co²⁺, Ni²⁺, Cu²⁺, Zn²⁺, Ag⁺, Cd²⁺, Hg²⁺ and Pb²⁺.⁶ Reported values of the dissociation constants K_d of the 1 : 1 complex [Zn(DPA)]²⁺ are 158 nM (ref. 7) and 70 nM.⁸

The literature of BODIPY-based fluorescent indicators, among them those for transition and heavy metal ions, has been reviewed recently by the KU Leuven authors.^4c In addition, there are excellent reviews on fluorescent chemosensors for zinc ions based on different fluorophore platforms and various receptor units, including DPA and related N,N-bis(2-pyridylmethyl)ethane-1,2-diamine.⁹ Of significant importance for this study are reports on fluoroionophores built on the BODIPY platform with DPA as ion receptor. A BODIPY-DPA Zn²⁺ probe with 30-fold fluorescence enhancement was reported by Nagano et al. (Chart 1).¹⁰ A fluorescent sensor for Zn²⁺ (K_d = 1 nM), based on photoinduced electron transfer (PET), utilizing BODIPY linked at the meso-position to the DPA chelator (Chart 1), which displays a significant fluorescence enhancement upon Zn²⁺ binding, was described by Peng and coworkers.¹¹ The same research group also reported a selective sensor for imaging Cd²⁺ in living cells, using the BODIPY scaffold (K_d is ca. 60 μM), but with the DPA chelator attached at the 3-position through a p-styryl spacer (Chart 1).¹² A very similar BODIPY-styryl linked DPA chelator (Chart 1) was described by Akkaya and coworkers, but it was reported as being selective for Zn²⁺ (K_d = 20 μM), though Hg²⁺ and Cd²⁺ also showed some response.¹³ Although the ratiometric probes of Peng¹² and Akkaya¹³ are very alike (see Chart 1), they display a different selectivity (Cd²⁺ vs. Zn²⁺). Usually discrimination between Cd²⁺ and Zn²⁺ is very difficult because they are stereoelectronic isosteres. For Ni²⁺ detection, only one account on a BODIPY-based fluorescent sensor can be found in the literature.¹⁴ This fluorescence turn-on (PET) probe (K_d = 0.2 mM) shows a ca. 25-fold fluorescence increase upon Ni²⁺ binding with no shifts in absorption and emission maxima. Since paramagnetic Cu²⁺ is a notorious fluorescence quencher, few ratiometric fluorescent chemosensors for Cu²⁺ are available in the literature.^4c,15 A colorimetric and near-IR fluorescent turn-on BODIPY-based probe with DPA as a chelator with high selectivity for Cu²⁺ among several transition metal ions has been reported by the KU Leuven authors (Chart 1).^16a Similar BODIPY-DPA derived colorimetric and NIR fluorescent chemosensors for Cu²⁺ (Chart 1) have been described by Yin and coworkers.^16b,c BODIPY-DPA functionalized hydroxyapatite (HA) nanoparticles formed 1 : 1 complexes with Cd²⁺ or Zn²⁺ with large cation-induced fluorescence amplifications,¹⁷ whereas BODIPY-functionalized silica nanoparticles exhibited high specificity for Cu²⁺ over other transition metal ions in aqueous-organic media and resulted in notable fluorescence quenching, as reported by Qin et al.¹⁸ Fluorescent probes for lead, cadmium and mercury¹⁹ ions have been reviewed recently.²⁰

Chart 1 BODIPY-DPA linked probes described in the literature.^10–13,16

In this work, we synthesized the chemosensor 1 with BODIPY as fluorophore coupled at the 3-position to the DPA chelator and to a phenylethynyl subunit at the 5-position. The phenylethynyl moiety at 5-position causes a red shift of the spectra in relation to unsubstituted and common BODIPY dyes.^5,21 The solvent-dependent spectroscopic/photophysical characteristics of 1 were investigated by steady-state UV-vis spectrophotometry and fluorometry. Coordination of 1 to various metal ions (Na⁺, K⁺, Mg²⁺, Ca²⁺, Ni²⁺, Cu²⁺, Zn²⁺, Cd²⁺, Hg²⁺) was spectroscopically studied and discussed. The measured ground-state dissociation constants K_d for the formed complexes, the absorption and fluorescence emission maximal wavelengths in the absence and presence of ions, the fluorescence quantum yields Φ of the apo and ion-bound species of 1 are presented and discussed.

Results and discussion

Synthesis

Fluorescent probe 4,4-difluoro-8-(4-methylphenyl)-5-(phenylethynyl)-3-[bis(pyridin-2-ylmethyl)amino]-4-bora-3a,4a-diaza-s-indacene (1) was synthesized by nucleophilic substitution of 3-chloro-4,4-difluoro-8-(4-methylphenyl)-5-(phenylethynyl)-4-bora-3a,4a-diaza-s-indacene (2) with bis(pyridin-2-ylmethyl)amine (DPA, 3) (Scheme 1). Compound 2 was synthesized according to our published procedure.²²

Scheme 1 Synthesis of compound 1.

UV-vis spectroscopy

A selection of UV-vis absorption and fluorescence emission spectra of 1 dissolved in a series of solvents is depicted in Fig. 1. The spectroscopic/photophysical properties of 1 are compiled in Table 1. The maximum of the main absorption band [λ_abs(max)] of 1, attributed to the 0–0 vibrational band of a strong S₁ ← S₀ transition, is located between 525 nm (in acetonitrile) and 561 nm (in cyclohexane), corresponding to a energy difference of ca. 1200 cm⁻¹. A shoulder in the 510–525 nm range at the short wavelength side of the spectrum is assigned to the 0–1 vibrational band of the same transition. This vibrational fine structure is undetectable in the more polar solvents (i.e. DMSO, acetonitrile, DMF and methanol, see Fig. 1a). The typical, narrow BODIPY-like main absorption band is not observed for 1. The most narrow absorption bands are detected in cyclohexane, toluene, dibutyl ether, pentan-1-ol and octan-1-ol. The full width at half height of the maximum of the absorption band (fwhm_abs) increases from ca. 2700 cm⁻¹ in cyclohexane to ca. 3900 cm⁻¹ in butanenitrile. The broadening of the absorption spectra in the more polar solvents indicates a change in the size or orientation of the permanent dipole moment upon excitation, which could correspond to a partial charge-transfer character for the lowest excited state in the ground-state geometry. To make this observation compatible with the blue shift of the absorption spectra upon increasing solvent dipolarity, one has to assume that the size of the permanent dipole moment probably decreases upon excitation or that excitation changes rather the orientation than the size of the permanent dipole moment. A closer inspection of the data in Table 1 suggests that a red shift related to an increased solvent polarizability and the resulting increased dispersion and van der Waals interactions is accompanied by a blue shift upon increasing the dipolarity of the solvent.

Fig. 1 (a) Normalized absorption spectra of 1 in a selection of solvents. (b) Corresponding normalized fluorescence emission spectra (excitation at 530 nm, except for acetonitrile, DMF and DMSO with excitation at 510 nm).

Table 1 Spectroscopic and photophysical properties of 1 in several solvents

#^a	Solvent	λabs(max) [nm]	λem(max) [nm]	λex(max) [nm]	Δ[small nu, Greek, macron] [cm^−1]	fwhmabs [cm^−1]	fwhmem [cm^−1]	Φ^b
a The solvents are numbered according to increasing dielectric constant.b Φ values were determined vs. cresyl violet as reference (Φρ = 0.55). Excitation wavelength λex was 530 nm, except for DMF, acetonitrile and DMSO with λex = 510 nm.c THF = tetrahydrofuran, DMF = N,N-dimethylformamide, DMSO = dimethyl sulfoxide.
1	Cyclohexane	561	589	561	847	2720	1300	0.18 ± 0.03
2	Toluene	554	594	555	1216	2720	1340	0.180 ± 0.002
3	Dibutyl ether	553	587	556	1047	2770	1310	0.134 ± 0.005
4	Diethyl ether	550	590	552	1233	3000	1390	0.106 ± 0.002
5	Chloroform	552	590	553	1167	3450	1320	0.24 ± 0.03
6	Ethyl acetate	539	584	543	1430	3180	1450	0.12 ± 0.01
7	THF^c	542	585	545	1356	3100	1430	0.156 ± 0.006
8	1-Octanol	554	587	554	1015	2910	1330	0.497 ± 0.004
9	1-Pentanol	552	586	552	1051	2930	1390	0.30 ± 0.01
10	1-Butanol	550	586	552	1117	3040	1530	0.38 ± 0.01
11	2-Propanol	548	586	550	1183	2980	1390	0.313 ± 0.006
12	Acetone	530	582	534	1686	3490	1750	0.103 ± 0.008
13	Butanenitrile	529	582	532	1721	3900	1580	0.131 ± 0.006
14	Ethanol	542	583	543	1298	3220	1410	0.165 ± 0.008
15	Methanol	538	582	541	1405	3760	1360	0.115 ± 0.004
16	DMF^c	527	577	530	1644	3620	1660	0.158 ± 0.001
17	Acetonitrile	525	575	527	1656	3670	1710	0.108 ± 0.006
18	DMSO^c	526	581	530	1800	3660	1580	0.26 ± 0.02

---

The structureless fluorescence emission of 1 also is solvent dependent (Fig. 1b). The maximum of the structureless emission, λ_em(max), shifts from 575 nm in acetonitrile to 594 nm in toluene. This red shift (ca. 560 cm⁻¹) is accompanied by a decrease of the emission bandwidth fwhm_em (from ca. 1700 cm⁻¹ in acetonitrile to ca. 1300 cm⁻¹ in toluene). This suggests an increase of the permanent dipole moment or a change of the orientation of the permanent dipole moment during the return to the Franck–Condon ground state.

It is useful to determine the origin of the solvent-dependent spectral changes because these deliver information on the nature of the ground and excited states. The most recent, comprehensive treatment of the solvent effect (based on a set of four empirical, complementary, mutually independent solvent scales, i.e., dipolarity, polarizability, acidity and basicity of the medium) has been proposed by Catalán.²³ In this method, the polarizability and dipolarity of a particular solvent are characterized by the parameters SP and SdP, respectively, whereas acidity and basicity are described by the scales SA and SB, respectively. The {SA, SB, SP, SdP} parameters for a large number of solvents can be found in ref. 23. Mathematically, the solvent effect on the physicochemical observable y can be expressed by the multilinear eqn (1):

y = y₀ + a_SASA + b_SBSB + c_SPSP + d_SdPSdP (1)

where y₀ denotes the physicochemical property of interest in the gas phase; a_SA, b_SB, c_SP and d_SdP are regression coefficients that describe the sensitivity of the property y to the various solvent–solute interaction mechanisms; {SA, SB, SP, SdP} are independent solvent parameters (indices) accounting for the various types of solvent–solute interactions.

The spectroscopic observables y analyzed in this paper are the absorption maxima [small nu, Greek, macron]_abs [= 1/λ_abs(max)] and the fluorescence emission maxima [small nu, Greek, macron]_em [= 1/λ_em(max)], both expressed in cm⁻¹. The use of {SA, SB, SP, SdP} (eqn (1)) gives high-quality fits of [small nu, Greek, macron]_abs of 1 (for the solvents listed in Table 1), using the correlation coefficient r as goodness-of-fit criterion (r = 0.939, eqn (2a) and , ESI†). Good-quality fits are also obtained for the multilinear analysis of [small nu, Greek, macron]_em according to eqn (1) (r = 0.827, eqn (2b) and ESI†). The extra benefit of the generalized (Catalán) treatment of the solvent effect is that it allows one to separate the relative contributions of dipolarity, polarizability, acidity and basicity of the medium. Therefore, we utilized the new methodology to resolve which solvent property/properties is/are responsible for the observed shifts of [small nu, Greek, macron]_abs and [small nu, Greek, macron]_em. The relative importance of each of the Catalán solvent scales was studied by omitting in turn one, two or three solvent scales from the regression analysis (for details, see ESI†). These analyses clearly identify solvent dipolarity (SdP) as the most critical parameter that accounts for the observed shifts of [small nu, Greek, macron]_abs and [small nu, Greek, macron]_em. Moreover, only for SdP are the estimated d_SdP coefficients significantly larger than their associated standard errors.

[small nu, Greek, macron]_abs = (18.2 ± 0.5) × 10³ + (−841 ± 249) SA + (−54 ± 170) SB + (−625 ± 649) SP + (1.3 ± 0.1) × 10³ SdP (2a)

[small nu, Greek, macron]_em = (17.2 ± 0.3) × 10³ + (−178 ± 142) SA + (31 ± 97) SB + (−474 ± 371) SP + (3.9 ± 0.8) × 10² SdP (2b)

Complex formation between 1 and various metal ions in acetonitrile

To investigate the chelating ability of 1, UV-vis spectrophotometric and fluorometric titrations were carried out in acetonitrile solution as a function of metal ion concentration. Complex formation was examined for 1 with alkali (Na⁺ and K⁺), alkaline-earth (Mg²⁺ and Ca²⁺), transition metal (Ni²⁺, Cu²⁺, Zn²⁺) and heavy metal (Cd²⁺ and Hg²⁺) ions. The results are summarized in Table 2.

Table 2 Spectroscopic and photophysical characteristics of 1 in the absence and presence of transition metal (Ni²⁺, Cu²⁺, Zn²⁺) and heavy metal (Cd²⁺, Hg²⁺) ions in acetonitrile. K_d is the average value determined from all the (direct and ratiometric) analyses of fluorometric and spectrophotometric titrations

Complex	λabs(max) [nm]	λem(max) [nm]	Isosbestic points [nm]	Δ[small nu, Greek, macron] [cm^−1]	Kd [μM]	fwhmabs [cm^−1]	fwhmem [cm^−1]	Φ
1	525	575		1656		3670	1710	0.108 ± 0.006
1–Ni^2+	565	580	540, 443	458	13 ± 2	1120	1800	0.020 ± 0.004
1–Cu^2+	594	616	547, 463	601	9 ± 3	−	950	0.40 ± 0.03
1–Zn^2+	565	584	540, 434	576	48 ± 11	1100	900	0.29 ± 0.05
1–Cd^2+	575	590	540, 438	442	9 ± 2	950	920	0.23 ± 0.02
1–Hg^2+	575	590	537, 438	442	4 ± 1	2000	840	0.27 ± 0.02

---

Upon addition of the alkali ions Na⁺ and K⁺ and the alkaline-earth ions Mg²⁺ and Ca²⁺ to a solution of 1 in acetonitrile, no change in the UV-vis absorption and fluorescence spectra could be detected. Obviously, the interaction between 1 and these ions is too weak to cause any change of either the UV-vis absorption spectra or the vis fluorescence spectra.

Conversely, the transition metal (Ni²⁺, Cu²⁺, Zn²⁺) and heavy metal (Cd²⁺, Hg²⁺) ions produced spectral changes of 1 (Fig. 2). For example, the lowest-energy absorption band of 1 shifts bathochromically by ca. 40 nm, from 525 nm in an ion-free environment to 565 nm upon addition of Zn²⁺ to the acetonitrile solution (Fig. 3a). The relative contributions of the 565/525 nm signals change with varying [Zn²⁺] and the vis absorption spectra show isosbestic points at 434 and 540 nm. Similar changes were observed in the fluorescence excitation spectra (Fig. S1, ESI†). The maximum of the fluorescence emission band shifts bathochromically from 575 nm in ion-free acetonitrile to 584 nm in the presence of Zn²⁺ and is accompanied by an increase in intensity (Fig. 3b). A pseudo-isoemissive point can be observed at 564 nm, which can be used to simplify analysis of ratiometric fluorometric titrations. The fluorescence quantum yield Φ of 1 increases from 0.11 in the absence of Zn²⁺ to 0.29 for the 1–Zn²⁺ complex. These red shifts can be rationalized as follows. Upon complexation of Zn²⁺, the lone pair of N of DPA connected to C3 of BODIPY will become less available for delocalization over the BODIPY moiety. One should note also that, upon complexation of Zn²⁺, the features of the absorption spectra become closer to that of unsubstituted and common BODIPYs showing a narrow band and a hint of vibrational fine structure even in polar solvents.^5,24 This can be attributed again to the blocking of the delocalization of the lone pair of N of DPA in 3-position over the rest of the conjugated system. The same effect will decrease the dipole moments (and their differences between ground and excited state) in the ground and excited state. This also accounts for the strong decrease of the Stokes shift from 1660 cm⁻¹ in 1 to 580 cm⁻¹ in 1–Zn²⁺ and of both fwhm_abs and fwhm_em of 1–Zn²⁺ (Table 2). Often BODIPYs with explicit electron-withdrawing or electron-donating substituents are characterized by low Φ values, especially in polar solvents.²⁵ Upon complexation by Zn²⁺, the electron-donating effect of N in 3-position is significantly decreased which leads to an increase of Φ from 0.11 to 0.29.

Fig. 2 (a) Absorption and (b) fluorescence emission spectra of apo 1 and of 1 in the presence of the transition metal ions Cu²⁺ (60 μM), Zn²⁺ (600 μM), Ni²⁺ (170 μM) and the heavy metal ions Cd²⁺ (235 μM) and Hg²⁺ (50 μM) in acetonitrile.

Fig. 3 Compound 1 as a function of [Zn²⁺] in acetonitrile solution. (a) Absorption spectra. (b) Fluorescence emission spectra (excitation at 510 nm). The full line in the inset shows the best fit to the ratiometric emission (eqn (4) with n = 1) titration data at λ_em¹/λ_em² = 582 nm/564 nm (isoemissive point) as a function of [Zn²⁺].

Analysis of the UV-vis spectrophotometric and vis fluorometric titration data in the presence of varying concentrations of Zn²⁺ allows one to extract the ground-state dissociation constant K_d and the stoichiometry of the 1–Zn²⁺ complex. The values of K_d and the stoichiometry (n) of binding of Zn²⁺ by the DPA ligand of 1 were determined by nonlinear fitting eqn (3) (direct fluorometric titration)²⁶ and eqn (4) (ratiometric fluorometric titration)²⁶ to the fluorescence excitation or emission spectral data F (eqn (3)) and ratios R (eqn (4)), measured as a function of metal ion concentration, i.e., [X] = [Zn²⁺]. Nonlinear least-squares analyses of the corresponding spectrophotometric titrations (direct and ratiometric) as a function of metal ion were also performed.²⁷ In all titrations, no metal ion buffer was used to control the free metal ion concentration. It was assumed that the free metal ion concentration [X] could be approximated by its analytical concentration.

In eqn (3), F stands for the fluorescence signal at free ion concentration [X], whereas F_min and F_max denote the fluorescence signals at minimal and maximal [X], respectively, and n is the number of cations bound per probe molecule (i.e., stoichiometry of binding). Because the fits of eqn (3) to the fluorescence data F with n, K_d, F_min and F_max as freely adjustable parameters always gave values of n close to 1 (indicative of a 1 : 1 complex), n = 1 was used in the final curve fittings, from which the estimated K_d values are reported here. The plot of F against −log[X] has a sigmoidal shape. This plot starts at F_min at very low [X] and changes most in the X-concentration range from 0.1 × K_d to 10 × K_d to asymptotically reach F_max at high [X]. Outside this critical X-concentration range, F does not change much. Therefore, it important that the dissociation constant K_d is well matched with the X-concentration range of interest.

In the excitation ratiometric method, one measures R = F(λ_em,λ_ex¹)/F(λ_em,λ_ex²) at a common emission wavelength, λ_em, and two different excitation wavelengths, λ_ex¹ and λ_ex². R_min is the ratio of the fluorescence intensities at two distinct excitation wavelengths and one emission wavelength of the apo form of the indicator (minimum [X]). R_max represents the ratio of the fluorescence intensities of the bound form of the indicator (maximum [X]). R denotes the ratio of the fluorescence intensities corresponding to intermediate [X] and ξ = F_min(λ_em,λ_ex²)/F_max(λ_em,λ_ex²). Choosing the wavelength of the pseudo-isoemissive point as λ_ex² makes the analysis simpler. In this case we have ξ = 1 and the expression for R (eqn (4)) simplifies to that of F (eqn (3)). Fitting eqn (4) to excitation ratiometric values R as a function of [X] yields values for K_dξ(λ_em,λ_ex²), R_min, R_max and n. Because ξ(λ_em,λ_ex²) – the ratio of the fluorescence signal of the apo form of the indicator over that of the bound form at the indicated wavelengths – is experimentally accessible, a value for K_d can be recovered from ratiometric excitation fluorescence data.

In the emission ratiometric method, one determines R = F(λ_em¹,λ_ex)/F(λ_em²,λ_ex) at the indicated wavelengths as a function of cation concentration [X]. In this case, ξ is defined as ξ = F_min(λ_em²,λ_ex)/F_max(λ_em²,λ_ex). ξ = 1 when λ_ex² is chosen as the wavelength corresponding to the pseudo-isoemissive point and, in that case, the expression for R (eqn (4)) simplifies to that of F (eqn (3)). Fitting eqn (4) to the emission ratiometric fluorescence data R as a function of [X] yields values for K_dξ(λ_em²,λ_ex), R_min, R_max and n. Because ξ(λ_em²,λ_ex) can be determined from the fluorescence signals of the apo and bound forms of the indicator at the indicated wavelengths, a value K_d for the complex can be obtained.

From the analyses of all the direct and ratiometric spectrophotometric and fluorometric titrations of 1 with Zn²⁺ ions it is clear that a 1 : 1 ligand/cation stoichiometry is found with an average K_d value for the 1–Zn²⁺ complex of 48 ± 11 μM.

The vis absorption and fluorescence (excitation/emission) spectral changes observed when Cd²⁺ is added to 1 in acetonitrile solution are comparable to those for Zn²⁺ (Fig. S2, ESI†), although the bathochromic shifts upon complex formation are somewhat larger [λ_abs(max) = 575 nm, λ_em(max) = 590 nm] and Φ of the 1–Cd²⁺ complex is slightly lower than that of 1–Zn²⁺ (Table 2). In the emission spectra as a function of [Cd²⁺] pseudo-isoemissive points can be found at 571 and 550 nm. An average K_d value of 9 ± 2 μM was obtained for the 1–Cd²⁺ complex. This smaller K_d value (compared to K_d of the 1–Zn²⁺ complex) suggests a stronger bond between Cd²⁺ and the N ligands of DPA. Hence, the lone pair of N connected to C3 of the BODIPY framework will be less available for delocalization over the conjugated system, leading to somewhat larger red spectral shifts than those observed for Zn²⁺.

The response of the absorption and fluorescence (excitation/emission) spectra of 1 in acetonitrile upon addition of Hg²⁺ (Fig. 4) is nearly the same as that found for 1 in the presence of Cd²⁺, but Φ of 1–Hg²⁺ is somewhat higher than Φ of 1–Cd²⁺ (Table 2). This Φ value is unexpected taking into account the possible heavy-atom quenching by Hg²⁺. The emission spectra as a function of [Hg²⁺] show two pseudo-isoemissive points, at 569 and 544 nm. The average K_d value of the 1–Hg²⁺ complex was found to be 4 ± 1 μM.

Fig. 4 Compound 1 in acetonitrile solution as a function of [Hg²⁺]. (a) Fluorescence excitation spectra (emission observed at 620 nm). (b) Fluorescence emission spectra (excitation at 510 nm). The full line in the insets of (a) and (b) shows the best fit to (a) the direct fluorometric excitation (eqn (3) with n = 1) titration data at λ_ex = 575 nm, λ_em = 620 nm and (b) the ratiometric emission (eqn (4) with n = 1) at λ_em¹/λ_em² = 590 nm/569 nm (isoemissive point) titration data as a function of [Hg²⁺].

Upon addition of Ni²⁺ to a solution of 1 in acetonitrile, the vis absorption spectra respond analogously to the addition of Zn²⁺: a ca. 40 nm red shift (to 565 nm) is found in addition to the appearance of isosbestic points at 443 and 540 nm. In contrast, addition of Ni²⁺ causes strong fluorescence quenching: Φ of 1–Ni²⁺ is only 0.02 compared to 0.11 for apo 1. Quenching of fluorescence upon Ni²⁺ binding arises from a PET from the BODIPY fluorophore to the metal center. Ni²⁺ has – in contrast to the close-shell ions Zn²⁺, Cd²⁺ and Hg²⁺ – several low-lying dd-type excited states and is moreover paramagnetic. The influence of Ni²⁺ on the spectral maxima and the features of the absorption and emission spectra can be clarified in the same way as for Zn²⁺, Cd²⁺ and Hg²⁺. The absorption and fluorescence emission spectra of 1 in acetonitrile as a function of [Ni²⁺] are given in Fig. S3 (ESI).† The average K_d value of the 1–Ni²⁺ complex amounted to 13 ± 2 μM.

Compared to the transition and heavy metal ions Ni²⁺, Zn²⁺, Cd²⁺ and Hg²⁺, addition of Cu²⁺ triggers a large fluorescence enhancement (Φ = 0.40 for 1–Cu²⁺) and the largest red shifts [λ_abs(max) = 594 nm, λ_em(max) = 616 nm] (Fig. 2 and 5). The average K_d value of 1–Cu²⁺ complex was 9 ± 3 μM. The Stokes shift of the 1–Cu²⁺ complex (600 cm⁻¹) is larger than those of the 1–Cd²⁺ and 1–Hg²⁺ complexes (440 cm⁻¹) and close to that of the 1–Zn²⁺-complex (580 cm⁻¹). Also the large increase in Φ is quite unexpected because Cu²⁺ is paramagnetic and has low-lying dd-states. Both properties are analogous to Ni²⁺, which strongly quenches the BODIPY fluorescence. Although the absorption spectrum of 1 undergoes a red shift upon complex formation with Cu²⁺ (in analogy to what occurs upon complex formation with the other divalent ions), the features of the absorption spectrum of the 1–Cu²⁺ complex differ in several aspects from those of the other complexes. First, the apparent ratio of the 0–1 to the 0–0 band of the S₁ ← S₀ transition decreases from 0.9 for Cu²⁺ to 0.3–0.4 for the other ions. Second, the ratio of the maximum absorbance of the S₂ ← S₀ to that of the S₁ ← S₀ transition decreases from 0.9 for Cu²⁺ to 0.15–0.25 for the other ions. Finally, the long wavelength edge of the 0–0 band of the S₁ ← S₀ transition is considerably shallower. Conversely, the features of the fluorescence spectrum of 1–Cu²⁺, although bathochromically shifted, resembles quite well those of the other complexes. This makes it unlikely that in 1–Cu²⁺ the nature of the excited state differs strongly from that in the complexes of 1 with the other metal ions. In those complexes, the excited state resembles that of a BODIPY fluorophore without strong electron-donating or electron-withdrawing substituents. Taking into account that Cu²⁺ complexes are known to absorb strongly between 550 and 650 nm, the unexpected features of BODIPY in the presence of Cu²⁺ is perhaps attributable to an overlap of the absorption spectrum of BODIPY and that due to dd-transitions of Cu²⁺ in 1–Cu²⁺. The relative short wavelength of the absorption spectrum of Cu²⁺ in 1–Cu²⁺ will lead to a poor overlap with the emission spectrum of 1–Cu²⁺. Hence energy transfer to the dd-type excited states of Cu²⁺ will be rather slow which may account for the (unexpected) absence of quenching of the BODIPY fluorescence by Cu²⁺ in the 1–Cu²⁺ complex.

Fig. 5 Compound 1 in acetonitrile solution as a function of [Cu²⁺]. (a) Absorption spectra. (b) Fluorescence emission spectra (excitation at 510 nm). The full line in the inset of (b) shows the best fit to the direct fluorometric emission titration data (eqn (3) with n = 1) at λ_em = 580 nm as a function of [Cu²⁺].

High selectivity towards the analyte of interest is an important property of any chemosensor. BODIPY indicator 1 forms 1 : 1 complexes with several transition metal (Ni²⁺, Cu²⁺, Zn²⁺) and heavy metal (Cd²⁺, Hg²⁺) ions. The dissociation constants K_d of these formed complexes (Table 2) provide numerical criteria for evaluating the selectivity of 1 for a particular ion. As shown above, the determination of K_d of each metal ion complex requires the laborious execution and analysis of fluorometric titrations.

In the literature, competitive, steady-state fluorescence measurements are frequently used as a quick alternative for assessing the selectivity of a fluorescent indicator for a certain ion. These simple tests are typically carried out as follows. First, one has to establish through fluorometric titration that the probe is sensitive to a specific ion, let's say X. The analysis of the titration data (eqn (3)) yields a value of K_d of the complex with X. To determine if the probe is selective for X, one then measures the steady-state fluorescence signal F₀ (either the whole, integrated fluorescence spectrum or at a single emission wavelength) of the probe in the presence of X at a certain concentration. Subsequently, the competing ion Y at a specific concentration is added to this solution and the fluorescence signal F is measured. The outcome of these competition experiments is often displayed in the form of bar graphs, as presented in Fig. S5 (ESI),† and shows the change/constancy of the steady-state fluorescence upon addition of the competing ion Y to a solution of the probe in the presence of ion X (X = Cu²⁺ or Cd²⁺ in Fig. S5, ESI†). If no change is found upon addition of Y, it is commonly assumed that Y does not compete with X for the probe and hence that the probe is selective for X. We shall show that such fast, easy competition experiments often lead to erroneous conclusions either because the experiments are carried out defectively or for fundamental reasons. For an in-depth investigation of the fluorescence signal F arising from a photophysical system consisting of a probe in the presence of competing ions X and Y, a correct mathematical description of F is needed. Although such analysis is beyond the scope of this paper, it is still possible, without going into mathematical detail, to discuss the present photophysical system (Fig. S5, ESI†) consisting of 1 in the presence of metal ions X and Y. It can be demonstrated mathematically that the fluorescence F of 1 in the presence of X and Y is dependent on several parameters, namely (i) the K_d values of the complexes of 1 (1–X and 1–Y) with the competing ions X and Y, (ii) the stationary fluorescence F_min of the apo form of 1, (iii) the fluorescence F_1–X and F_1–Y of the pure complexes 1–X and 1–Y, respectively and (iv) the concentrations [X] and [Y] of the unbound (free) ions X and Y, respectively.²⁸ F describes a 3D data surface as a function of the independent variables [X] and [Y]. Using the equation of F with the above (i–iv) parameters, one can predict mathematically how addition of ion Y to a solution of 1 in the presence of X will affect its steady-state fluorescence F. The dependence of the steady-state fluorescence on the addition of Y at a specific concentration to a solution of 1 in the presence of certain concentration of X is commonly measured by competition experiments (such as those shown in Fig. S5, ESI†). However, in such experiments only two data points of the full 3D data surface are determined: F₀ and F in Fig. S5 (ESI†) are even fused into the unique data point F/F₀. F₀ stands for the fluorescence signal of 1 in the presence of 10 μM X (= Cu²⁺ or Cd²⁺) and F corresponds to 1 in the presence of 10 μM X (= Cu²⁺ or Cd²⁺) plus a 50 μM solution of other metal ions Y. These two data points and a priori the single data point F/F₀ are insufficient to determine the selectivity of indicator 1 vs. X or Y. Addition of the competitive ion (50 μM) to a solution of 1 in the presence of 10 μM Cu²⁺ (or 10 μM Cd²⁺) shows no significant variation in the fluorescence intensity (Fig. S5, ESI†). This indicates that the fluorescence of 1–Cu²⁺ (or 1–Cd²⁺) is not influenced by other coexisting metal ions Y, at least not at the concentrations used for the original ion X (Cu²⁺ or Cd²⁺) and the competing metal ions Y.

The near-invariability of the fluorescence intensities F₀ and F (i.e., F/F₀ ≈ 1.0) cannot be taken as unambiguous proof that 1 is not selective for Y, for the following reasons. It is crucial to start the experiment beginning with the pure complex 1–X; F₀ should therefore correspond to the fluorescence signal F_1–X of the pure complex 1–X, which is not the case here. At 10 μM Cu²⁺ (or 10 μM Cd²⁺) (Fig. S5, ESI†), the fluorescence intensity of the 1–Cu²⁺ (or 1–Cd²⁺) complex is not reached. This is only the case when the free concentration [Cu²⁺] (or [Cd²⁺]) is equal to at least 10 × K_d (or preferably > 10² × K_d) of the corresponding 1–Cu²⁺ (or 1–Cd²⁺) complex (K_d = 9 μM) (see eqn (3)).²⁷ At [Cu²⁺] (or [Cd²⁺]) = 10 μM and with K_d = 9 mM, we are approximately at the mid-point of the titration. Hence, the fluorescent indicator appears both in its free form 1 and its complexed form 1–X (= 1–Cu²⁺ or 1–Cd²⁺). At low [X] ([Cu²⁺] or [Cd²⁺]), addition of competing metal ion Y to a mixture containing both 1 and 1–X (1–Cu²⁺ or 1–Cd²⁺) may form the complex 1–Y directly from 1 or/and by displacement of X (Cu²⁺ or Cd²⁺) from the 1–X (1–Cu²⁺ or 1–Cd²⁺) complex. Although under those conditions there might be a change of fluorescence, it is impossible to estimate from these two data points the K_d values of the complex 1–Y and/or F_1–Y. Even if only 1–X (1–Cu²⁺ or 1–Cd²⁺ in Fig. S5, ESI†) is present before addition of the competing metal ion Y, constancy of fluorescence upon addition of Y may simply be attributable to the concentration [Y] used that is too low for influencing the fluorescence. Then, a large excess of [Y] might reveal fluorescence change or invariability in relation to the initial experimental condition where only 1–X is present. The fluorescence will vary from F_1–X (only 1–X present) at low [Y] ([Y] → 0) to F_1–Y (only 1–Y present) for [Y] → ∞. However, it is still impossible to estimate a value for K_d of the complex 1–Y. Theoretically, constancy of the fluorescence as a function of [Y] may also be the result of F_1–X and F_1–Y being equal (F_1–X = F_1–Y). In this case, the fluorescence remains constant, irrespective of the concentration [Y]. Finally, the competition between X and Y is a dynamic process: it is possible that decomplexation of the original 1–X complex and/or subsequent formation of the new 1–Y complex is too slow to occur on the timescale used. To summarize, there are several reasons for the invariability of the fluorescence signal upon addition of Y (too low [Y] used, F_1–X = F_1–Y, decomplexation of 1–X and/or succeeding formation of 1–Y is too slow, or the probe is simply not sensitive to Y), so that no reliable decision about the selectivity of the probe can be reached. To conclude, these quick competitive tests are not capable of providing a clear answer to the question of selectivity; only K_d values estimated from fluorometric titrations can do.

Besides selectivity, another important parameter defining fluorescent indicators is sensitivity. Chemosensor 1 forms 1 : 1 complexes with various transition metal (Ni²⁺, Cu²⁺, Zn²⁺) and heavy metal (Cd²⁺, Hg²⁺) ions. An excellent linear correlation (r = 0.994, n = 7) between the steady-state fluorescence F and the Zn²⁺ concentration (log[Zn²⁺]) was obtained over the range 0–186 μM Zn²⁺ (Fig. S6, ESI†). The detection limit (DL) was estimated based on the following equation: DL = 3σ/k, where σ is the standard deviation of 10 blank samples (containing only probe 1 without any metal ions) and k is the slope of the calibration curve (fluorescence intensity plotted against [metal ion] or log[metal ion]).²⁹ The detection limit was calculated to be as low as 1.0 μM for Zn²⁺. The sensitivity of the probe towards each metal ion is compiled in Table 3 and is shown to be 1.0 μM or lower.

Table 3 The sensitivity of probe 1 towards each metal ion

Metal ion	Linear range/μM	F^a	DL/μM	r
a Equation used for the fitting of F as a function of ion concentration. The Zn^2+ and Cd^2+ concentrations are expressed in M (mol L^−1), whereas [Ni^2+], [Cu^2+] and [Hg^2+] are in μM.
Ni^2+	0–25.7	F = 534 − 15.3 × [Ni^2+]	0.1	0.990
Cu^2+	1.5–26	F = 81.6 + 53.7 × [Cu^2+]	0.04	0.992
Zn^2+	0–186	F = 5591 + 879 × log[Zn^2+]	1.0	0.994
Cd^2+	1.3–235.5	F = 876.7 + 122.2 × log[Cd^2+]	0.5	0.994
Hg^2+	0–5.3	F = 389 + 142.7 × [Hg^2+]	0.01	0.996

---

Conclusions

We have synthesized the visible light excitable, fluorescent BODIPY-based sensor 1, which shows solvent-dependent spectroscopic/photophysical properties. The generalized treatment of the solvent effect shows that solvent dipolarity is primarily responsible for the observed shifts of the absorption and fluorescence emission maxima. Fluorescent indicator 1 forms 1 : 1 complexes with several transition metal ions (Ni²⁺, Cu²⁺, Zn²⁺) and heavy metal ions (Cd²⁺, Hg²⁺), but not with alkali-metal ions (Na⁺, K⁺) and earth-alkaline-metal ions (Ca²⁺, Mg²⁺). K_d values of the metal ion complexes range from 4 μM for Hg²⁺ to 48 μM for Zn²⁺. The new compound is an example of a very sensitive fluorescent probe for several metal ions displaying large absorption and fluorescence changes in an analytically interesting wavelength region.

Experimental

The materials used, the steady-state UV-vis absorption and fluorescence spectroscopy, the determination of K_d through direct and ratiometric fluorometric titration are described in the ESI.†

Synthesis of 4,4-difluoro-8-(4-methylphenyl)-5-(phenylethynyl)-3-[bis(pyridin-2-ylmethyl)amino]-4-bora-3a,4a-diaza-s-indacene (1)

3-Chloro-4,4-difluoro-8-(4-methylphenyl)-5-(phenylethynyl)-4-bora-3a,4a-diaza-s-indacene (2) was synthesized following a literature procedure.²² To a solution of 2 (104 mg, 0.25 mmol) in acetonitrile (50 mL) under argon atmosphere was added bis(pyridin-2-ylmethyl)amine 3 (60 mg, 0.30 mmol, 1.2 equiv.). The reaction mixture was stirred at room temperature for 16 h. Afterwards the resulting solution was poured in water and extracted with dichloromethane (2 × 100 mL). The combined organic layer was washed with brine, dried over MgSO₄, filtered and evaporated to dryness under reduced pressure. The crude solid was purified by column chromatography on silica gel using ethyl acetate as eluent to yield 129 mg (89% yield) of a red solid. M.p. 176 °C. ¹H NMR (CDCl₃) δ 8.55 (d, 2H, J = 4.1 Hz), 7.66 (td, 2H, J = 7.9 Hz, J = 1.5 Hz), 7.48 (dd, 2H, J = 7.9 Hz, J = 1.5 Hz), 7.44 (d, 2H, J = 7.9 Hz), 7.34 (d, 2H, J = 7.9 Hz), 7.28 (m, 3H), 7.24 (d, 2H, J = 7.9 Hz), 7.20 (dd, 2H, J = 5.3 Hz, J = 1.8 Hz), 6.82 (d, 1H, J = 4.9 Hz), 6.59 (d, 1H, J = 4.1 Hz), 6.36 (d, 1H, J = 3.8 Hz), 6.28 (d, 1H, J = 5.2 Hz), 5.29 (s, 4H), 2.42 (s, 3H, tolyl). ¹³C NMR (CDCl₃) δ 164.2, 156.6, 149.6, 139.2, 137.1, 136.3, 135.4, 133.7, 132.2, 132.0, 131.6, 130.6, 128.2, 128.0, 126.0, 125.6, 123.8, 122.7, 122.1, 120.3, 119.5, 115.3, 100.1, 95.6, 84.0, 58.1, 22.8. IR (cm⁻¹) 3055 (m), 3014 (m), 2920 and 2852 (CH₂, s), 2202 (C=C, w). LRMS (EI, 70 eV) m/z 579. HRMS: calcd for C₃₆H₂₈BF₂N₅ 579.2406, found 579.23967.

¹H and ¹³C NMR spectra were recorded at room temperature on a Bruker 600 instrument operating at a frequency of 600 MHz for ¹H and 150 MHz for ¹³C. ¹H NMR spectra were referenced to tetramethylsilane (0.00 ppm) as an internal standard. Chemical shift multiplicities are reported as s = singlet, d = doublet and m = multiplet. ¹³C spectra were referenced to the CDCl₃ (77.67 ppm) signal. Mass spectra were recorded on a Hewlett Packard 5989A mass spectrometer (EI mode and CI mode). High-resolution mass data were obtained with a KRATOS MS50TC instrument. The IR spectrum was recorded on a Bruker FT-IR spectrometer Alpha-P with Diamond ATR. Melting points were taken on a Reichert Thermovar and are uncorrected.

Acknowledgements

WD and MVdA are thankful to BELSPO (Belgium) for funding through IAP 06/27 and IAP7/05. WQ thanks the Natural Science Foundation of China (no. 21271094).

References

1. Chemosensors of Ion and Molecule Recognition, ed. J.-P. Desvergne and A. W. Czarnik, Kluwer, Dordrecht, The Netherlands, 1997.
2. New Trends in Fluorescence Spectroscopy. Applications to Chemical and Life Sciences, ed. B. Valeur and J.-C. Brochon, Springer, Berlin, 2002.
3. A. Treibs and F.-H. Kreuzer, Liebigs Ann. Chem., 1968, 718, 208–223.
4. (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; (d) T. Kowada, H. Maeda and K. Kikuchi, Chem. Soc. Rev., 2015, 44, 4953–4972; (e) S. Yin, V. Leen, C. Jackers, M. van der Auweraer, M. Smet, N. Boens and W. Dehaen, Dyes Pigm., 2011, 88, 372–377; (f) S. Yin, V. Leen, C. Jackers, D. Beljonne, B. van Averbeke, M. van der Auweraer, N. Boens and W. Dehaen, Chem.–Eur. J., 2011, 17, 13247–13257.
5. N. Boens, B. Verbelen and W. Dehaen, Eur. J. Org. Chem., 2015, 6577–6595.
6. K. Kyose, H. Kojima, Y. Urano and T. Nagano, J. Am. Chem. Soc., 2006, 128, 6548–6549.
7. D. W. Gruenwedel, Inorg. Chem., 1968, 7, 495–501.
8. G. K. Walkup, S. C. Burdette, S. J. Lippard and R. Y. Tsien, J. Am. Chem. Soc., 2000, 122, 5644–5645.
9. (a) K. Kikuchi, K. Komatsu and T. Nagano, Curr. Opin. Chem. Biol., 2004, 8, 182–191; (b) P. Jiang and Z. Guo, Coord. Chem. Rev., 2004, 248, 205–229; (c) N. C. Lim, H. C. Freake and C. Brückner, Chem.–Eur. J., 2005, 11, 38–49; (d) P. Carol, S. Sreetjith and S. Ajayaghosh, Chem.–Asian J., 2007, 2, 338–348.
10. H. Koutaka, J.-i. Kosuge, N. Fukasaku, T. Hirano, K. Kikuchi, Y. Urano, H. Kojima and T. Nagano, Chem. Pharm. Bull., 2004, 52, 700–703.
11. Y. Wu, X. Peng, B. Guo, J. Fan, Z. Zhang, J. Wang, A. Cui and Y. Gao, Org. Biomol. Chem., 2005, 3, 1387–1392.
12. X. Peng, J. Du, J. Fan, J. Wang, Y. Wu, J. Zhao, S. Sun and T. Xu, J. Am. Chem. Soc., 2007, 129, 1500–1501.
13. S. Atilgan, T. Ozdemir and E. U. Akkaya, Org. Lett., 2008, 10, 4065–4067.
14. S. C. Dodani, Q. He and C. J. Chang, J. Am. Chem. Soc., 2009, 131, 18020–18021.
15. J. A. Cotruvo Jr, A. T. Aron, K. M. Ramos-Torres and C. J. Chang, Chem. Soc. Rev., 2015, 44, 4400–4414.
16. (a) S. Yin, V. Leen, S. van Snick, N. Boens and W. Dehaen, Chem. Commun., 2010, 46, 6329–6331; (b) S. Yin, W. Yuan, J. Huang, D. Xie, B. Liu, K. Jiang and H. Qiu, Spectrochim. Acta, Part A, 2012, 96, 82–88; (c) J. Huang, X. Ma, B. Liu, L. Cai, Q. Li, Y. Zhang, K. Jiang and S. Yin, J. Lumin., 2013, 141, 130–136.
17. X. Jia, X. Yu, G. Zhang, W. Liu and W. Qin, J. Coord. Chem., 2013, 66, 662–670.
18. X. Yu, X. Jia, X. Yang, W. Liu and W. Qin, RSC Adv., 2014, 4, 23571–23579.
19. E. M. Nolan and S. J. Lippard, Chem. Rev., 2008, 108, 3443–3480.
20. H. N. Kim, W. X. Ren, J. S. Kim and Y. Yoon, Chem. Soc. Rev., 2012, 41, 3210–3244.
21. W. Qin, T. Rohand, W. Dehaen, J. N. Clifford, K. Driessen, D. Beljonne, B. van Averbeke, M. van der Auweraer and N. Boens, J. Phys. Chem. A, 2007, 111, 8588–8597.
22. T. Rohand, W. Qin, N. Boens and W. Dehaen, Eur. J. Org. Chem., 2006, 4658–4663.
23. J. Catalán, J. Phys. Chem. B, 2009, 113, 5951–5960.
24. L. Jiao, C. Yu, J. Wang, E. A. Briggs, N. A. Besley, D. Robinson, M. J. Ruedas-Rama, A. Orte, L. Crovetto, E. M. Talavera, J. M. Alvarez-Pez, M. van der Auweraer and N. Boens, RSC Adv., 2015, 5, 89375–89388.
25. (a) W. Qin, M. Baruah, M. Sliwa, M. van der Auweraer, D. Beljonne, B. van Averbeke and N. Boens, J. Phys. Chem. A, 2008, 112, 6104–6114; (b) W. Qin, V. Leen, T. Rohand, W. Dehaen, P. Dedecker, M. van der Auweraer, K. Robeyns, L. van Meervelt, D. Beljonne, B. van Averbeke, J. N. Clifford, K. Driesen, K. Binnemans and N. Boens, J. Phys. Chem. A, 2009, 113, 439–447; (c) I. Esnal, J. Bañuelos, I. López Arbeloa, A. Costela, I. Garcia-Moreno, M. Garzón, A. R. Agarrabeitia and M. J. Ortiz, RSC Adv., 2013, 3, 1547–1556.
26. E. Cielen, A. Tahri, A. Ver Heyen, G. J. Hoornaert, F. C. de Schryver and N. Boens, J. Chem. Soc., Perkin Trans. 2, 1998, 1573–1580.
27. A. Kowalczyk, N. Boens, V. van den Bergh and F. C. de Schryver, J. Phys. Chem., 1994, 98, 8585–8590.
28. N. Boens, unpublished results.
29. W. Qin, W. Liu and M. Tan, Anal. Chim. Acta, 2002, 468, 287–292.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental (materials, steady-state UV-vis absorption and fluorescence spectroscopy, determination of K_d through direct and ratiometric fluorometric titration). The Catalán analyses of the solvent-dependent absorption and fluorescence emission maxima. Absorption, fluorescence excitation and emission spectra of 1 in acetonitrile as a function of [Zn²⁺], [Cd²⁺] and [Ni²⁺]. Cyclic voltammetry. Fluorescence experiments with competing ions. Sensitivity of 1 towards Zn²⁺, Cu²⁺ and Ni²⁺. ¹H NMR and ¹³C NMR spectra of 1 in CDCl₃. See DOI: 10.1039/c5ra23751c

---