Novel acridine-based thiosemicarbazones as 'turn-on' chemosensors for selective recognition of fluoride anion: a spectroscopic and theoretical study.

New thiosemicarbazide-linked acridines 3a-c were prepared and investigated as chemosensors for the detection of biologically and environmentally important anions. The compounds 3a-c were found selective for fluoride (F-) with no affinity for other anions, i.e. -OAc, Br-, I-, HSO4-, SO42-, PO43-, ClO3-, ClO4-, CN- and SCN-. Further, upon the gradual addition of a fluoride anion (F-) source (tetrabutylammonium fluoride), a well-defined change in colour of the solution of probes 3a-c was observed. The anion-sensing process was studied in detail via UV-visible absorption, fluorescence and 1H-NMR experiments. Moreover, during the synthesis of acridine probes 3a-c nickel fluoride (NiF2), a rarely explored transition metal fluoride salt, was used as the catalyst. Theoretical studies via density functional theory were also carried out to further investigate the sensing and anion (F-) selectivity pattern of these probes.

Introduction

Acridine, a tricyclic nitrogen heterocyclic compound, has been frequently used as a dye for dyeing silks, wood and leather. In medicinal chemistry, many acridine derivatives have found applications as drugs such as acriflavine (a topical antiseptic) and quinacrine (an antiprotozoal drug). Most of the acridine derivatives and their metal complexes also act as DNA intercalating agents. However, the use of acridine-based compounds as chemosensors has not yet been explored for anionic analysis. With this interest, we aimed to synthesize acridine-based compounds functionalized with semicarbazide moieties. Previous studies have provided evidence regarding the role of semicarbazide moiety in anion sensing [1-4]. Chemosensing carries unique significance in monitoring biological processes, in clinical diagnostics and in monitoring environmental factors. Anion sensing via colorimetric assay has gained much popularity for qualitative and quantitative ion sensing for the purpose of regulating ion concentration in biological systems, water samples and clinical analysis [5-9]. Many different types of small molecules have been discovered as anion-sensing probes (or chemosensors) for the detection of biologically important anions such as halides, phosphates, acetates, ascorbate, citrates, etc. [4,10-12]. Fluoride carries unique importance among biologically important anions due to its applications in dental care products such as toothpaste, a material used in the daily routine in human life, and in treatment of osteoporosis [13-15].

Traditionally, the known techniques for ion detection are spectroscopy, chromatography and ion-selective electrodes which require expensive instrumentation and extensive and destructive sample preparation, and hence cannot be employed for cell or tissue imaging [16]. In recent times, the use of small molecules as fluorescent probes for ion detection in biological, environmental and industrial systems has rapidly gained in popularity [10,17-29]. Anion detection via small molecules holds enormous potential for biologically important anions, such as fluoride (F−), acetate (AcO−), sulfate, phosphate, pyrophosphate, ascorbate, citrate, etc. [10,30,31]. In this regard, molecules having biologically and environmentally friendly scaffolds are more desirable for anion-sensing process, because they offer least toxicity. In this study, we have functionalized the acridine scaffold with a thiosemicarbazide moiety, which acts as an ion-binding site during the anion-sensing process (figure 1).

Figure 1.

Acridine-based compounds.

Results and discussion

Chemistry

The synthetic layout of targeted thiosemicarbazide-linked acridine is given in scheme 1. The core acridine template was constructed by reacting 2 mol of 5,5-dimethylcyclohexane-1,3-dione (dimedone), 1 mol of benzaldehyde and 1 mol of ammonium acetate in the presence of nickel fluoride tetrahydrate as a catalyst. It is important to note here that NiF2·4H2O is not well explored as a catalyst in organic synthesis, though such type of reactions have been explored with some traditional fluorides such as CsF or tetrabutylammonium fluoride (TBAF) [32-34]. The reaction proceeded smoothly in the presence of NiF2·4H2O (25 mol%) as the catalyst to furnish the acridine scaffold 7 with a simple work-up and washing with n-hexane. The use of transition metal fluoride (NiF2) as a new catalytic source to promote organic reaction could serve as a substitute for common fluoride salts, i.e. CsF, KF, etc. Further, in order to endow the acridine scaffold with a semicarbazide moiety, it was treated with excess of hydrazine hydrate (80%), and subsequently with phenyl isocyanate in acetonitrile at 80–85°C, until the disappearance of the starting materials, as monitored by thin-layer chromatography (TLC). The desired thiosemicarbazide-linked compounds, as anion-sensing probes 3a–c, were obtained in a 55–85% yield (scheme 1).

Scheme 1.

Synthetic layout for acridine-based probes 3a–c.

In the 1H-NMR spectrum of acridine-based thiosemicarbazone 3b (as a selected example), characteristic broad signals for two NH groups were observed at δH 10.32 and 9.23 ppm, while the broad signal for NH of the acridine scaffold appeared at δH 8.80 ppm. The peaks of all (13) aromatic protons appeared between δH 7.65 and 6.91 ppm. A sharp singlet for the H-9 proton appeared at δH 5.27 ppm. Further, the signals for four methylene groups (CH2)4 appeared as doublets at (i) 2.49 ppm (obscured by the dimethylsulfoxide (DMSO) signal), (ii) 2.31 ppm, (iii) 2.23 ppm and (iv) 2.16 ppm. Two singlets for each of two methyl groups at position C-3/6 appeared at δH 1.03 ppm and 0.82 ppm. The structure of 3b was further confirmed with two-dimensional NMR techniques. The heteronuclear single quantum coherence spectrum showed direction correlation of the methine (CH-9) group at 35.6 ppm, while the heteronuclear multiple bond correlation (HMBC) spectrum showed correlations of H-9 with C-1/8, C-2′′/6′′, C-10/13 and C-11/12. The two NHa protons showed correlations with C-1′, C-6′ carbons atoms, while the NHb protons showed correlations with C-1/8 carbon atoms in the HMBC spectrum. The proton of acridine-NH has a correlation with C-10/13 carbon atoms. Further details of NMR signals and correlations have been provided in table 1. The infrared (IR) signals of 3b have been provided in the experimental section. The mass of 3b via electrospray ionization (ESI+) was found at m/z 684.1 (M + H).

Table 1.

Characteristic 1H-NMR, 13C-NMR and HMBC correlation data of 3b. .

atom no.	δH (m, J in Hz)	δC	HMBC correlations
1/8 (C)	—	152.1	—
2/7 (CH2)	2.49 (obscured by DMSO signal) /2.23 (d, J = 16.4)	38.2	NHb, C-1/8, C-4/5, C-3/6, C-10/13, (CH3)2
3/6 (C)	—	30.3	—
4/5 (CH2)	2.31 (d, J = 16.4)/2.16 (d, J = 16.4)	39.3	NH, C-2/7, C-3/6, C-10/13, C-11/12, (CH3)2
9 (CH)	5.27	35.9	C-1/8, C-2′′/6′′, C-10/13, C-11/12
10/13 (C)	—	106.7	—
11/12 (C)	—	140.1	—
1′/1′ (C)	—	126.6/126.5	—
2′/2′ (C)	—	157.2/154.2	—
3′/3′ (CH)	6.92 (m)	115.2/115.1	C-1′, C-2′, C-5′
4′/4′ (CH)	7.22–7.18 (m)	126.9/126.8	C-2′, C-6′
5′/5′ (CH)	7.10 (t, J = 7.6)	123.6	C-1′, C-3′
6′/6′ (CH)	7.64 (t, J = 6.8)	128.2	C-2′, C-4′
1′′ (C)	—	148.4	—
2′′/6′′ (CH)	7.31 (d, J = 7.2)	128.0	C-4′′, C6′′/C-2′′, C-4′′
3′′/4′′/5′′ (CH)	6.96–6.90 (m)	127.2/125.2	C-1′′, C-5′′/C-2′′, C-6′′ /C-1′′, C-3′′
NHa	9.23	—	C-1′, C-6′
NHb	10.32	—	C-1/8, C=S
NH	8.79	—	C-10/13, C-4/5
C=S	—	175.8	—
C(CH3)2	1.03	—	C-2/7, C-4/5
C(CH3)2	0.82	—	C-2/7, C-4/5

Anion sensing

For colorimetric sensing, the pattern of receptors/chemosensors 3a–c for different biologically important anions, i.e. F−, −OAc, Br−, I−, HSO4−, ClO3−, CClO4−, CN− and SCN−, has been studied via UV–visible (UV–Vis) spectroscopy. The anion-sensing study with thiosemicarbazide-linked acridines 3a–c showed high selectivity for fluoride when compared with other anions such as acetate, bromide, iodide, bisulfate, chlorate, perchlorate, cyanide and thiocyanate. An increasing amount of different anions (up to 30 equivalents) was added to the solution of receptors 3a–c (4 × 10−5 M). The change in colour of the receptor solution from yellow to reddish brown, to dark brown (up to 30 equivalents) was only observed upon the addition of fluoride (figure 2), whereas in the case of other anions, no change in colour was observed. The UV spectral profile of a model receptor 3a showed absorption peaks at 320, 415 and 485 nm in DMSO diluted in acetonitrile solvent due to π–π* and n–π* transitions. A detailed record of UV–Vis absorption peaks for all receptors 3a–c is given in electronic supplementary material, table S1.

Figure 2.

Acridine-based probe 3a: selective sensing of F− among different anions; 3b: gradual addition of fluoride (0–30 equiv.) anions (naked eye view).

The titration of thiosemicarbazide–acridine receptor 3a as a model sensor against different anions showed high selectivity for fluoride anion (red line) with a clear shift in the UV spectrum (figure 3a). Further the UV–Vis titration experiments of receptor 3a upon gradual increase of fluoride anions (0–30 equiv.) showed a decrease in intensity at 320, 415 and 480 nm with a concomitant appearance of a new red-shifted band at 550 nm (figure 3b). The formation of three isobestic points at 332, 375 and 455 nm clearly indicates the formation of new species upon the addition of a fluoride anion source, TBAF. The UV–Vis spectra of receptor 3b having mono-fluoride substituent on the phenyl ring were found to be almost similar in selectivity and detection pattern for fluoride anions as that of probe 3a, when titrated with different anion solutions such as F−, −OAc, Br−, I−, HSO4−, SO42−, PO43−, ClO3−, ClO4−, CN−, SCN− and H2PO4− (electronic supplementary material, figure S4). Moreover, the titration of di-fluoride-substituted receptor 3c with different anions also showed it to be selective only for fluoride anion (electronic supplementary material, figure S5). Selectivity of all probes 3a–b towards fluoride anion could be due to the small size and basic nature of fluoride that helps to remove acidic hydrogen atoms from the thiosemicarbazide moiety of receptor 3c, resulting in observed changes in colour and UV–Vis spectrum (electronic supplementary material, figure S5).

Figure 3.

(a) UV absorption spectra of probe 3a, (b) absorption spectra of probe 3a upon gradual increase of fluoride addition, (c) upon dilution in methanol absorbance of 3a is restored and (d) fluorescence spectra of receptor 3a (excitation at 270 nm) upon the gradual increase of fluoride anions by using DMSO/acetonitrile as the solvent.

Moreover, a continuous variation method, known as Job's plot, was used to determine the stoichiometric ratio between receptor 3a and the fluoride anion during titration experiments. The Job's plot of probe 3a and fluoride showed maxima at a mole fraction greater than 0.8 (electronic supplementary material, figure S2), while for 3b and 3c between 0.7 and 0.8 (electronic supplementary material, figure S3 and S5), which indicates the interaction/abstraction of more than one fluoride anion [3]. The binding constant calculated by using the Benesi–Hildebrand equation [35] was found to be 2.66 × 102 M−1 with a limit of detection (LOD) of 6.879 × 10−5 M for probe 3a (electronic supplementary material, figure S2). For probes 3a and 3c, the binding constant and LOD values are given in table 2.

Table 2.

Binding constants and detection limits of receptors 3a–c for fluoride anion detection.

comp. no.	binding constant (M−1)	detection limit (M)
3a	2.66 × 102	6.879 × 10−5
3b	4.48 × 103	9.08 × 10−5
3c	2.86 × 103	6.17 × 10−5

Interestingly, it has been also observed that upon dilution of 3a receptor–anion solution in methanol, the absorption intensity at 360 nm was restored with a concomitant disappearance of the band at 460 nm (figure 3c). These results suggested the degradation of the receptor–anion complex via re-protonation of NH groups in the presence of protic (i.e. MeOH) solvent. The other probes 3b and 3c also showed the same pattern of UV–Vis absorption spectra (electronic supplementary material, figures S4c and S5c). The sensitivity of probes 3a–c was further explored via a fluorescence spectral study. The solutions of probes 3a and 3b were prepared followed by gradual addition of fluoride ions (0 to 10 equiv.), and were then subjected to fluorescence spectroscopic studies. The fluorescence spectra of probe 3b upon excitation at 278 nm showed a gradual decease in the peak at 530 nm with a concomitant appearance of a new peak at 610 nm (electronic supplementary material, figure S4d). The pattern of probe 3a was also observed in the same way, with a little difference in that it showed two peaks at 480 and 560 nm (figure 3d). Moreover, the applicability of the thiosemicarbazone-linked acridine, i.e. probe 3b, has been explored with toothpaste samples (electronic supplementary material, figure S4e). The variations in UV spectra clearly indicate the fluoride-sensing ability of the synthesized linked acridine–thiosemicarbazone receptors.

The interaction of semicarbazide-linked acridines 3a–c with fluoride anions was further investigated by 1H-NMR study. The receptor 3b was treated with a fluoride anion source (TBAF) in DMSO-d6. The fluoride source was added gradually to the NMR sample of receptor 3b in DMSO-d6. 1H-NMR spectra were recorded before fluoride addition and then after adding different concentrations (0, 1, 2 and 4 equiv. of TABF) (figure 4). The area between δ 3 and 10 ppm in the 1H-NMR spectrum of receptor was zoomed to monitor the changes in NH signals of receptor 3b. The spectrum of 3b showed signals for NH groups at δ 10.32 (2NH), δ 9.23 (2NH) and δ 8.79 (1NH) ppm before addition of fluoride anion (TBAF). Upon addition of 1–2 equiv. of fluoride anions to the solution of receptor 3b, a drastic decrease in the intensity of signals for NH protons (figure 4b) was observed. The signals for NH protons completely disappeared after addition of 4 equiv. of fluoride ions to receptor 3b (figure 4c). The disappearance of characteristic signal for NH protons of model receptor 3b upon the gradual addition of fluoride anion strongly suggested the abstraction of more than one NH proton by fluoride anions (scheme 2).

Figure 4.

Fluoride (F−) anion gradual addition to probe 3b: 1H-NMR spectra (a) without F− anion, (b) 1 equiv., (c) 2 equiv. and (d) 4 equiv. of F−source (TBAF) in DMSO-d as the solvent.

Scheme 2.

Fluoride interaction with thiosemicarbazone-linked acridine.

The interaction between a model probe 3a and fluoride anions was also studied by 1H-NMR spectroscopy (electronic supplementary material, figure S1). The amount of TBAF as the fluoride anion source was gradually increased (0‒4 equiv.) to observe the changes in 1H-NMR signals. The signals for all NH groups at δ 10.17, δ 9.48 and δ 8.79 ppm for probe 3a were significantly deceased and completely disappeared after the addition of 4 equiv. of fluoride anions (TBAF). The abstraction of NH protons of 3b occurred due to the strong basic effect of F− with respect to other anions studied. The signal of CH-9 proton at δ 5.26 ppm shifted slightly up-field, which was considered insignificant. The sequential 1H-NMR data showed clearly the selective interaction of the fluoride anion with NH protons due to its strong basic characteristic compared to other anions, i.e. −OAc, Br−, I−, HSO4−, SO42−, PO43−, ClO3−, ClO4−, CN− and SCN−. Moreover, the presence of five NH groups in thiosemicarbazide-linked acridine as receptor 3a bestows the ability to detect more than one fluoride anion during the sensing process.

Computational study

All calculations were performed with a Gaussian 09 suite of programs [36]. Geometries of thiosemicarbazide acridines and their complexes with various anions are optimized without any symmetry constraints by the Coulomb-attenuated method of B3LYP (CAM-B3LYP) at the 6–311G(d,p) basis set [37,38]. Frequency analysis has been performed to confirm these optimized geometries as true minima (no imaginary frequency). The interaction energies of thiosemicarbazide acridines with various anions are calculated by the following equation: where Ecomplex is the energy of complex formed between thiosemicarbazide–acridine and anions, whereas Etsc and Eanion are the energies of isolated thiosemicarbazide–acridine and anions, respectively.

The excitation energies were also calculated with time-dependent density functional theory (DFT) at the (TD)CAM-B3LYP/6–311G(d,p) [37,38] level of theory. The electronic properties were computed with the Coulomb-attenuated method (CAM), a hybrid functional that implements long-range correction of B3LYP for calculation of excitation energies. The CAM-B3LYP method was chosen to study the excited state because of the well-established accuracy of this method for a variety of classes of organic compounds [39-42]. The solvent effect was studied through the polarization continuum model. All calculations were performed in DMSO solvent. A total of 20 states are calculated with 50 : 50 singlet : triplet states.

DFT calculations have been performed to rationalize the selectivity of thiosemicarbazide–acridine-based probes for fluoride ions. In this regard, the calculations have been performed for complexation energies (of anions with the probe molecules) and UV–Vis absorption spectroscopic properties. In the first step, the most stable geometry of the probe molecule 3a is searched. We have previously shown [43] that acridines prefer to adopt E-orientation of the N-group (thiosemicarbazide in our case, figure 5) even when the central ring bears hydrogen atoms. In our case, a phenyl ring is present in the central six-membered ring, which further limits the possibility of Z-orientation. Therefore, we started with the E-orientation of the thiosemicarbazide groups. The thiosemicarbazide group has several single bonds around which conformational scans produce a number of conformers. However, the bonds ‘A' and ‘C' are more important around which conformational scan has been performed to locate the low-energy conformers. The conformation of the lowest energy conformer (obtained through a conformational scan) is shown in figure 6, where the thio group interacts with the hydrogen atoms of the central phenyl ring. Moreover, the terminal phenyl ring is almost perpendicular to the plane of thiosemicarbazide (figure 6). The central C–N bonds (bond B) of the thiocarbazide adopt planar orientation because of the delocalization of electrons between nitrogen and the thio group.

Figure 5.

Description of degree of freedom for conformational and configurational isomerism.

Figure 6.

The optimized geometries of the lowest energy conformer of 3a (a) and its 1 : 3 complex with fluoride ions (b).

In the next step, the interactions of various anions with the probe molecule are studied. The experimental results show that the probe molecule-to-anion ratio is more than three. Therefore, we started with the optimization of the 1 : 3 complex of probe 3a with various anions. Analysis of the probe molecule reveals that there are three positions where anions can interact. These positions include N–H of the central ring, and H–N–C(S)–N–H of both thiosemicarbazide groups. An anion can interact with both N–H group of the H–N–C(S)–N–H moiety simultaneously to form a stable six-membered ring. The computational study is performed for complexes of probes 3a–c with F−, −OAc, Br−, Cl− and CN− anions, whereas other less basic and bulky anions (I−, HSO4−, ClO3−, ClO4−) are not considered. The optimized geometry of the 1 : 3 complex of probe (3a): F− is shown in figure 6b.

Compound 3a shows considerable changes in the geometry after complexation with fluoride ions. Important structural reorganizations during complexation are the changes in the orientations of the central and terminal phenyl rings. The fluoride ion interacts not only with the N–H protons of the thiosemicarbazide but also with the ortho hydrogen atoms of the terminal phenyl ring (figure 6b). Owing to this interaction, the terminal phenyl rings become coplanar with the thiosemicarbazide moiety, which causes the thio group to form loose non-bonding interactions with the central phenyl ring. As a result, the central phenyl ring gets rotated to have favourable interaction through its ortho hydrogen atoms with the nitrogen atom (of thiosemicarbazide) at a distance of 2.72 Å (figure 6b). The fluoride ion has strong proton affinity; therefore, the N–H bond of the central ring is considerably broken. The N–H and F–H bond lengths are 1.21 Å and 1.18 Å, respectively. The weakening of N–H bond (of the central ring) causes the nitrogen atom to become more electron-rich. The increased electron density causes changes in the electronic and spectroscopic properties of the complex (vide infra). The strong proton abstraction ability of F− can be attributed to the high charge density on fluoride ions. The fluoride ion has −1 charge which is completely localized on it, whereas the negative charge on other ions (such as −OAc, Br−, Cl−) is either more delocalized or diffused. The diffused or delocalized negative charges on other anions result in relatively weak interaction with the probe molecule (vide infra for binding energy calculations). These anions (−OAc, Br−, Cl− and CN−) interact only with the N–H of the thiosemicarbazide without any involvement of the ortho hydrogen of the terminal phenyl ring. Therefore, the coplanarity of the terminal phenyl ring with the thiosemicarbazide moiety is not seen for these anions.

The binding energies are also calculated for these complexes, which reflect that the probe 3a has the strongest binding affinity for the fluoride ion. The binding of three fluoride ions with a molecule of probe 3a is exothermic by 383.97 kcal mol−1, which corresponds to an exothermicity of −127.99 kcal mol−1 per fluoride ion (table 3). The binding energy of the fluoride ion with probe 3a is much stronger than that of other anions. The interaction energy of the fluoride ion with the acridine–thiosemicarbazide is comparable to a covalent bond. The higher binding energy of the fluoride ion with the probe molecule is consistent with the experimental observation where maximum response in sensing is seen for the fluoride ion. The binding energies of other anions range from −72 to −87 kcal mol−1 per anion, which is about 40–55 kcal mol−1 lower than that for the fluoride ion. Among other anions, the highest binding energy is calculated for acetate ion which may be attributed to the high electronegativity of oxygen combined with two binding sites (two oxygen atoms) per anion. Nitrile anion has a binding energy of −80.82 kcal mol−1 per anion, which amounts to −242.45 kcal mol−1 for three CN− anions. The least binding energy is calculated for the interaction of bromide with the probe molecule, which may be attributed to the low charge density of bromide due to its large size.

Table 3.

Total binding energies and binding energies per anion of probes 3a–c for F−, Cl−, Br−, OAc− and CN−.

	F−	Cl−	Br−	OAc−	CN−
total binding energies (kcal mol−1)
3a	−383.97	−226.80	−216.86	−259.54	−242.45
3b	−383.25	−227.25	−217.25	−256.63	−242.14
3c	−386.43	−227.50	−217.49	−259.80	−242.24
binding energies per anion (kcal mol−1)
3a	−127.99	−75.60	−72.29	−86.51	−80.82
3b	−127.75	−75.75	−72.42	−85.54	−80.72
3c	−128.81	−75.83	−72.50	−86.60	−80.75

A similar trend in the binding energies of different anions is observed for other probes 3b and 3c. The binding energies of the fluoride ion with probe 3b and 3c are −383.25 and −386.43 kcal mol−1, respectively, which amount to −127.75 and −128.81 kcal mol−1 per fluoride ion. The probe 3c has highest binding energy for the fluoride ion when compared with 3b and 3a, which may be attributed to a greater electron-withdrawing effect of fluoride ions. Probe 3c has relatively higher binding affinity for other anions compared to probe 3b and 3a but the difference is almost negligible. In general, the electron-withdrawing substituent on the terminal aromatic ring increases the binding affinities for different anions compared to the unsubstituted probe 3a.

Finally, the UV–Vis spectra are calculated for probe–anion complexes, and they are compared to the UV–Vis spectrum of a pure probe molecule. The probe 3a has two absorption bands appearing at 355.0 nm and 274.75 nm. The band at 355 nm has a tail which is stretched to 480 nm. The calculated UV–Vis spectrum at CAM-B3LYP is consistent with the experimental UV–Vis spectrum. Complexation of all anions except F− does not cause any shift in the UV–Vis spectrum of the probe (figure 7).

Figure 7.

UV–Vis spectra of probe 3a and its complexes with various anions, calculated at (TD)CAM-B3LYP/6-311G(d,p).

For some anions such as −OAc or CN−, the intensities of some of the absorption peaks are affected; however, the absorption maximum remains almost unaffected. The most significant change in the UV–Vis spectrum is observed for fluoride complexes where the UV–Vis spectrum is red-shifted, which is consistent with the experimental observations. The red-shifting of the UV–Vis spectrum of the 3a–F− complex is due to two factors: (i) the abstraction of proton by the fluoride ion leaves the nitrogen with more electron density which is now delocalized on the entire skeleton and (ii) the complexation of fluoride with the probe causes the planarization of the skeleton, particularly at the terminal phenyl ring (through interaction with ortho hydrogen atoms of the phenyl ring, vide supra). The other probes 3b and 3c also show similar selectivity for fluoride ions compared to other anions (table 4).

Table 4.

Changes in the absorption maxima of 3a–c on complexation with various anions, calculated at CAM-B3LYP/6–311G(d,p).

	λ_max^1	λ_max^2	λ_max^1	λ_max^2	λ_max^1	λ_max^2
X	3a-X− (1 : 3 complexes) λmax (3a) = 355 and 274.75 nm		3b-X− (1 : 3 complexes) λmax (3b) = 356.43 and 274.52 nm		3c-X− (1 : 3 complexes) λmax (3c) = 356.38 and 274.27 nm
F−	391.78	303.32	382.96	295.46	400.27	296.59
Cl−	347.95	276.75	345.36	272.94	345.76	272.82
Br−	346.37	276.76	344.64	272.90	345.93	272.92
OAc−	363.26	287.14	347.80	284.22	359.12	283.79
CN−	358.64	284.24	357.28	282.35	354.15	277.72

Conclusion

In summary, thiosemicarbazide-linked acridines 3a–c have been synthesized and explored as chemosensors for biologically important anions, i.e. F−, −OAc, Br−, I−, HSO4−, SO42−, PO43−, ClO3−, ClO4−, CN− and SCN−. The core acridine was furnished from corresponding precursors, i.e. 5,5-dimethylcyclohexane-1,3-dione and ammonium acetate in the presence of nickel fluoride as the catalyst. The anion-sensing study was demonstrated via naked eye observation, UV–Vis absorption, fluorescence and 1H-NMR spectroscopic experiments. In the anion-sensing study, the receptors 3a–c were found to be highly selective for the fluoride anion. This is possibly due to its small size and high basicity, which makes possible its strong interaction/abstraction of proton from receptors to induce changes in colour and signals/peaks in UV–Vis, fluorescence and 1H-NMR spectra. Upon the gradual addition of TBAF to acridine-based receptors, systemic changes in the UV–Vis and fluorescence spectra have been observed which illustrated the formation of new species during the sensing process. Moreover, the fluoride interaction with new thiosemcarbazide-linked acridines 3a–c was further exploited via DFT.

Materials and methods (chemistry)

All the starting materials including 5,5-dimethyl-1,3-cyclohexanedione (95%), benzaldehyde, ammonium acetate, nickel fluoride (NiF2·4H2O) (99%), hydrazine hydrated (80%) and phenyl isothiocyanate (98%) were purchased from Sigma Aldrich or Merck and used directly for carrying out our reaction unless otherwise stated. Acetonitrile (HPLC grade), DMSO (HPLC grade), methanol (HPLC grade) and distilled or Milli-Q water were used. TLC was carried out with silica gel 60 aluminium-backed plates 0.063–0.200 mm (Merck, Germany). Analytical grade solvents such as ethyl acetate, diethyl ether, hexane and methanol were used as eluents for purification purpose through column chromatography. Round-bottomed flasks (5, 10, 25, 50, 100, 250 ml) were used. TLC visualization was carried out using UV lamp radiation at 254 nm. In addition, different spot test mixtures, such as basic potassium permanganate or vanillin, were also used. IR spectra were recorded with a Bruker Vector-22 spectrometer. The 1H-NMR spectra were recorded using Bruker Avance spectrometers at 300, 400, 500 and 600 MHz, while 13C-NMR spectra were recorded at 75, 100, 125 and 150 MHz in the appropriate deuterated solvent. The chemical shifts were recorded on the δ-scale (ppm) using residual solvents as an internal standard (DMSO: 1H 2.50, 13C 39.43; and CHCl3: 1H 7.26, 13C 77.16). Coupling constants were calculated in hertz (Hz) and multiplicities were labelled as s (singlet), d (doublet), t (triplet), q (quartet) and quint (quintet), and the prefixes br (broad) or app (apparent) were used. Mass spectra (EI+ and FAB) were recorded with a Finnigan MAT-321A, Germany. Melting points of solids were determined using a Stuart™ melting point SMP3 apparatus and are uncorrected.

General procedure for the synthesis of acridin-based thiosemicarbazones (3a–c)

To a 25 ml round-bottomed flask, equipped with a magnetic stirrer, dimedone (1.40 g, 10 mmol), ammonium acetate (0.57 g, 7.5 mmol), benzaldehyde (0.50 ml, 5 mmol), nickel(II) fluoride tetrahydrate (25 mol%) as the catalyst and ethanol (3 ml) were added. The reaction mixture was heated on an oil bath at 80°C with continuous stirring for 80 min until the complete consumption of the starting material, as monitored using TLC analysis. The resulting mixture was dissolved in methanol, and poured onto crushed ice with stirring to obtain precipitates of the product. The precipitates were washed with n-hexane to get rid of unreacted benzaldehyde. Pure corresponding acridine 7 was obtained in 86% yield (1.50 g).

The compound 7 (0.35 g, 1 mmol) was treated with hydrazine hydrate (10 ml) and acetic acid (2 ml) at 120°C for 28 h. Progress of the reaction was monitored by TLC analysis using 30% EtOAc in n-hexane. On completion of the reaction, water was added to get the precipitate of compound 8, which was filtered and dried under vacuum. The yield was 97% (0.36 g) and the colour was golden yellow.

The synthesis of compounds 3a–c was carried out by reacting intermediate 8 (1.0 equiv.) and the appropriate substituted aryl isocyanates or isothiocyanates (2 equiv.), which were taken in 15 ml acetonitrile and heated at 98–100°C for 13 h. The progress of the reaction was monitored through TLC analysis using a 0.20 mm thick precoated silica plate, and spots were visualized through UV light. On completion of the reaction, the mixtures were cooled down to ambient temperature, resulting in the formation of precipitates. The precipitates were filtered, washed with warm acetonitrile and dried to obtain the pure compounds 3a–c in a 55–85% yield (scheme 1).

2,2 (3a). Light yellow solid, yield 60% (over three steps, 192 mg), MP 281–284°C. IR (νmax, cm−1): (KBr disc) 3420, 3272, 3216, 3093, 2956, 2925, 2863, 1645, 1596, 1526, 1479, 1405, 1367, 1316, 1260, 1240, 1193, 1156, 1074, 1001, 755, 697, 661. 1H-NMR (400 MHz, DMSO-d): δH 10.17 (2H, brs, 2NH), 9.48 (2H, brs, 2NH), 8.79 (1H, brs, NH), 7.46–7.38 (6H, m, ArH), 7.23–7.12 (6H, m, ArH), 6.98–6.91 (3H, m, ArH), 5.35 (1H, s, CH), 2.39 (obscured by DMSO signal, H-2/7), 2.32 (2H, d, J = 16.4 Hz, H-4/5), 2.24 (2H, d, J = 16.4 Hz, H-2/7), 2.16 (2H, d, J = 16.4 Hz, H-4/5), 1.04 (6H, s, (CH3)2), 0.81 (6H, s, (CH3)2). C13-NMR (100 MHz, DMSO-d): δC 193.7 (C), 175.0 (C), 151.9 (C), 148.7 (C), 139.9 (C), 137.8 (C), 128.2 (CH), 128.0 (CH), 127.3 (CH), 125.3 (CH), 124.8 (CH), 124.5 (CH), 123.7 (CH), 106.7 (C), 39.6 (CH2 × 2), 38.3 (CH2 × 2), 35.9 (CH), 30.3 (C), 29.4 (CH3 × 2), 26.4 (CH3 × 2). ESI-MS: m/z 648.1 (M + H).

2,2 (3b). Yellow solid, yield 85% (over three steps), MP 275–278°C. IR (νmax, cm−1): (KBr disc) 3749, 3446, 3271, 3222, 3093, 2955, 2926, 1646, 1620, 1527, 1478, 1404, 1367, 1239, 1223, 1157, 1077, 1029, 811, 752, 709, 658. 1H-NMR (400 MHz, DMSO-d): δH 10.32 (2H, brs, 2NH), 9.23 (2H, brs, 2NH), 8.80 (1H, brs, NH), 7.64 (2H, t, J = 6.8 Hz, ArH), 7.31 (2H, d, J = 7.2 Hz, ArH), 7.22–7.18 (2H, m, ArH), 7.10 (2H, t, J = 7.6 Hz, ArH), 6.96–6.91 (5H, m, ArH), 5.27 (1H, s, CH), 2.49 (obscured by DMSO signal, H-2/7), 2.31 (2H, d, J = 16.4 Hz, H-4/5), 2.23 (2H, d, J = 16.4 Hz, H-2/7), 2.16 (2H, d, J = 16.4 Hz, H-4/5), 1.03 (6H, s, (CH3)2), 0.82 (6H, s, (CH3)2). C13-NMR (100 MHz, DMSO-d): δC 193.7 (C), 175.8 (C), 157.2/154.8 (C), 152.1 (C), 148.4 (C), 140.1 (C), 128.2 (CH), 128.0 (CH), 127.2 (CH), 126.9/126.8 (CH), 126.6/126.5 (CH), 125.2 (CH), 123.6 (CH), 115.3/115.1 (CH), 39.6 (CH2 × 2), 29.2 (CH2 × 2), 35.9 (CH), 30.3 (C), 29.2 (CH3 × 2), 26.5 (CH3 × 2). ESI-MS: m/z 684.1 (M + H).

2,2 (3c). Yellow solid, yield 55% (over three steps), MP 279–281°C. IR (νmax, cm−1): (KBr disc) 3440, 3273, 2957, 1644, 1612, 1529, 1478, 1308, 1237, 1146, 1086, 965, 848, 765, 710. 1H-NMR (400 MHz, DMSO-d): δH 10.21 (2H, brs, 2NH), 9.39 (2H, brs, 2NH), 8.68 (1H, brs, NH), 7.34–7.29 (4H, m, ArH), 7.04–6.98 (4H, m, ArH), 6.94 (3H, m, ArH), 5.42 (1H, s, CH), 2.31–2.06 (8H, m, (CH2)4), 1.01 (6H, s, (CH3)2), 0.83 (6H, s, (CH3)2). ESI-MS: m/z 720.1 (M + H).