Novel turn-on fluorescent sensor for cyanide ion based on the charge transfer transition of phenothiazine/indolium compounds

A new fluorescent sensor combining phenothiazine and indolium, which reacts specifically with the cyanide ion with a large Stokes shift and a good fluorescence quantum yield, was prepared. When CN − was added to an ethanol solution containing the synthesized sensor molecules, the solution color changed from purple to colorless, and fluorescence was emitted under 270-nm light irradiation. The mechanism of this luminescence, investigated via proton nuclear magnetic resonance, electrospray ionization mass spectrometry, and computational analysis, was defined as follows: the sensor undergoes nucleophilic addition of cyanide to the carbon of C = N + in its indolium moiety, which limits the intramolecular charge transfer, resulting in a colorless transition and blue fluorescence. The specificity of the proposed sensor for cyanide has been confirmed through tests with other 14 anions (F − , Cl − , Br − , I − , ClO 3 − , ClO 4 − , NO 2 − , NO 3 − , SO 32 − , SO 42 − , S 2 O 32 − , S 2 − , N 3 − , and SCN − ) and the detection limit is 0.02 µM in the fluorescence spectrum. Changes in this sensor color can be detected by the naked eye and fluorescence emission can be induced via black light irradiation. (ESI-MS),


Introduction
Cyanide ion has extremely high toxicity but is relatively easy to obtain because it is used in the metal plating and plastics industries, as well as used as a reagent in most laboratories. 1 Given the concern about the health effects of CN − release into the environment, various technologies have been developed to detect CN − in drinking water and similar, mainly in the environmental field. CN − is also involved in serious crimes; 2 for example, cyanide gas is produced in fires in addition to carbon monoxide. 3 Therefore, CN − detection is extremely important to determine the cause of death in homicides and fires. Thus, many instrumental analytical methods such as gas chromatography, mass spectrometry, liquid chromatography, and ion chromatography have been applied for the detection of CN − ; 4-15 however, these methods require expensive equipment, advanced facilities, and excellent operation skills. Hence, several chemical sensors, including colorimetric sensors and fluorescence sensors that selectively react with CN − , have been recently developed for the rapid detection of CN − in the field. 16 This approach has advantages such as shorter inspection time and lower cost, which allows simple and rapid on-site determinations and is extremely useful for practical applications. Many sensors have been reported and most of them are based on the nucleophilic addition of CN − to their molecules [17][18][19][20] or the coordination of cyano groups to metal reagents. 16,[21][22][23] In this study, we attempted to develop a turn-on fluorescent sensor for CN − , although fluorescence emission generally has a low detection limit; in other words, we designed a molecule that emits fluorescence when reacting with CN − . There have been many recent reports detection reagents based on the mechanism of quenching the fluorescence that should be emitted via intramolecular charge transfer (ICT) and emitting fluorescence by cutting off the charge transfer. [24][25][26] In the molecular design, an electron-donating and an electronwithdrawing group are connected by a carbon-carbon double bond to form a donor-π-acceptor system. Although various compounds are possible, we selected phenothiazine as the electron-donating group and an indolium salt as the electronwithdrawing group. A similar combination has been reported by El-Shishtawy et al. as dimethine cyanine dye (PTZIS), a candidate for nonlinear optical materials. 27 Phenothiazines emit fluorescence and have properties as electron-donating groups, 28-31 while many studies have reported the reaction of the indolium cation, which is a strong electron-withdrawing group, with CN − . 32-44 Therefore, we thought that a phenothiazine/indolium compound could also be applied for the CN − detection. We synthesized a novel phenothiazine/indolium conjugated molecule (PI) as a CN − sensing dye; its color and fluorescence changes in response to CN − , as well as the selectivity to the target analyte, were also investigated. Then, its reaction mechanism was analyzed through proton nuclear magnetic Scheme 1 shows the synthetic route of PI, which was modified from the method reported by El-Shishtawy et al. 27 10-Octyl-10H-phenothiazine (1). Potassium hydroxide (2.08 g, 370 mmol) was added portionwise to a solution of 10Hphenothiazine (2.92 g, 14.6 mmol), 1-bromooctane (3.45 g, 17.8 mmol), and potassium iodide (40.3 mg, 0.24 mmol) in DMSO (50 mL). The reaction mixture was stirred at room temperature for 29 h and, then, poured into water (200 mL). The reaction mixture was extracted with chloroform. The combined chloroform layer was washed with saturated NH4Cl aqueous solution, dried over sodium sulfate, and concentrated in vacuo; the crude product was purified via silica gel flash column chromatography (hexane) to give 4.28 g, 94% yield as a colorless oil. The spectral data of 1 was in good agreement with the reported values. 31 10-Octyl-10H-phenothiazine-3-carbaldehyde (2). Phosphoryl chloride (2.74 mL, 29.3 mmol) was added dropwise at 0 °C to ice-cooled N,N-dimethylformamide (5 mL, 64.6 mmol) under a nitrogen atmosphere. The reaction mixture was warmed up to room temperature, stirred for 10 h, and then cooled down to 0 °C, followed by the addition of a solution of 1 (2.19 g, 7.02 mmol) in N,N-dimethylformamide (2.5 mL). The resulting mixture was heated to 80 °C, stirred for 18 h, and then poured into ice water; afterward, it was basified by adding saturated potassium carbonate aqueous solution and extracted with chloroform. The combined organic phase was dried over sodium sulfate and concentrated in vacuo. The crude product was purified via flash column chromatography (n-hexane/ethyl acetate = 9:1) to give the title compound (2.02 g, 85% yield) as a yellow oil. The NMR and IR spectral data of 2 were in good agreement with the reported values. 31 (E)-2-(2-(10-Hexyl-10H-phenothiazin-3-yl)vinyl)-1,3,3trimethyl-3H-indol-1-ium iodide (PI). A solution of 2 (195.4 mg, 0.576 mmol) and 1,2,3,3-tetramethyl-3H-indolium iodide (190.5 mg, 0.632 mmol) in dry ethanol (5.5 mL) was stirred at 60 °C for 10 h. After the solvent was removed by vacuuming, the residue was diluted with ethyl acetate, followed by the addition of nhexane to precipitate; the resulting precipitate was filtered off to give the title compound (300 mg, 84% yield) as a dark blue solid and, then, was further purified via recrystallization from a methanol-diethyl ether solution for the analysis.

Preparations of sample solutions for the UV-Vis and fluorescence measurements
Please do not adjust margins Please do not adjust margins CN − aqueous solution was added to 3 mL of 10 µM PI solutions and each spectrum was measured. For optimization of the solvents, 6 µL (2 equiv) of a 10 mM CN − aqueous solution, for determination of the quantitative relationship of CN − reacting with PI, 3-75 µL (0.1-2 equiv) of a 1 mM CN − aqueous solution and for selectivity to various anions, 6 µL of 10 mM aqueous anion solution was added. Each aqueous anion solution was prepared by dissolving 100 µmol of one among 15 inorganic salts (KCN, NaF, NaCl, NaBr, KI, KClO3, KClO4, NaNO2, NaNO3, Na2SO3, Na2SO4, Na2S2O3, Na2S, NaN3, and KSCN) in 10 mL of pure water.

UV-Vis and fluorescence measurements
A quartz cuvette with an optical path length of 1 cm was used for all the measurements. The UV-Vis spectra were collected in the 200-800 nm range. The fluorescence intensity was measured at 365 (ex)/483 (em) nm for the solvent effect and the fluorescence spectra were recorded between 300 and 700 nm under 270-nm light irradiation. The fluorescence quantum yield (Ф) was determined via a comparative method by using 15 µM quinine sulfate dihydrate (F = 0.55 in 0.1 M H2SO4, where F is the integrated area under the corrected emission spectrum) as the standard according to the following equation: 45 Фx = Фst (Ast/Ax) (Fx/Fst) (ƞx/ƞst) 2 where A is the absorbance at the excitation wavelength, ƞ is the refractive index of the solution, and the subscripts x and st denote the unknown material and the standard, respectively.

Job's plot measurement
A 20 µM solution of PI was prepared by dissolving in EtOH. Similarly, 20 µM solution of CN − was prepared by dissolving in a 4:1 (v/v) mixture of EtOH/H2O. Mixture solutions (3.0 mL) of PI and CN − at 11 different ratios from 0:3 to 3:0 (v/v) were prepared. After the solutions were allowed to stand for 10 min, the absorbance at 555 nm was measured at room temperature. The difference between initial absorbance (A0) and obtained absorbance value (A) were plotted against [PI]/([PI] + [CN − ]), and the complex formation ratio was estimated from the abscissa at the maximum absorbance.

Computational analysis
The structural optimization and computational calculations for the ground state of PI were conducted via the density functional theory (DFT) with B3LYP/6-311G level of theory using Spartan'10, 46 while those for the excited states of the fluorescent product produced by adding CN − to PI were performed at the B3LYP/6-31G level of theory by using the Gaussian 09 package. 47     27 When adding CN − , the color tone disappeared and all the solutions exhibited fluorescence with large Stokes shifts. The fluorescence intensity of the fluorescent product was weak in chloroform; we attributed this to the large bathochromic shift caused by ICT in that solvent. Among the tested solvents, dimethyl sulfoxide showed the strongest fluorescence intensity, but sufficient fluorescence intensity was observed also in solvents other than chloroform. In the case of chloroform, the dipole moment of the excited state of the fluorescent product was smaller due to the lower polarity of the solvent, maximum absorption wavelength has led to the bathochromic shift. Therefore, we decided to use ethanol, which is more practical and safer, in the subsequent experiments.

Changes in the absorption and fluorescence spectra of PI when adding CN −
The quantitative relation between PI and CN − was studied via UV-Vis and fluorescence spectroscopy analyses. The absorbance at 390 and 555 nm decreased with increasing the CN − concentration (Fig. 3(a)), while the fluorescence increased with it ( Fig. 3(b)). The fluorescence intensity at 483 nm varied linearly with the CN − concentration (R² = 0.9945) (ESI, Fig. S2). The limit of detection (LOD), calculated as 3.3s/a, where s is the standard deviation of the signal in the blank material and a is the slope of the calibration curve near the detection limit, was 0.29 µM by UV-Vis and 0.02 µM by UV-Vis fluorescence spectroscopy. Although many fluorescent sensors for CN − detection have been reported, 48 PI showed relatively good sensitivity compare with these sensors (ESI, Table S1). The CN − concentration in human blood is toxic when it exceeds 19 μM, 49,50 but the World Health Organization (WHO) has set the maximum concentration of CN − in drinking water at 1.9 µM. 51 Therefore, PI could be applied in the environmental field or for forensic analysis because the LOD is much lower than these concentrations. The reaction kinetics between PI and CN − were also investigated. CN − (20 µM, 2 equiv) was added to a 10 µM PI solution and the change in fluorescence intensity was measured over time (ESI, Fig. S3(a)). The reaction started upon the CN − addition and reached a steady state after 100 s. When ln[(Intt − Intmax)], where Intmax is the emission intensity at the steady state and Intt is the intensity at time t, was plotted versus time, a linear fit was obtained, as shown in Fig. S3(b), demonstrating that the reaction follows pseudo-first-order kinetics. The rate constant of this reaction (Kobs), calculated via the above reaction equation, was 0.0307 s −1 (ESI, Fig. S3(b)). Then, the reaction rate was determined in the same way for CN − less than one equivalent solutions containing 10, 7.5, and 5 µM of CN − were added to a 10 µM solution of PI. The fluorescence intensity increased linearly up to 50 s and then slowly; after 20 minutes of measurement, no steady state was reached, but the fluorescence intensity of each solution was proportional to the respective CN − concentration (ESI, Fig. S4).

Selectivity of PI for different anions
To investigate the selectivity of PI for anions other than CN − , 15 species of anions including CN − were added to the PI solution and their behavior was observed. In addition, CN − and other anions were added at the same time, and anti-interference experiments were also performed. As a result, the purple color disappeared and fluorescence was observed only when adding CN − (Fig. 4). The absorbance intensity at 555 nm and the fluorescence intensity at 483 nm excited at 273 nm were also measured (Fig. 5, ESI, Fig. S5 and Fig. S6). Among the other 14 anions, S 2− induced a change in the solution color, which became lighter, and fluorescence emission, but the fluorescence intensity was about 1/7 that obtained for CN − . To investigate the reactivity of PI to S 2− , various S 2− concentrations (1-16 equiv) were added to PI (ESI, Fig. S7); the purple color disappeared with increasing the S 2− concentration and the absorption spectrum became constant at 4 equiv, but the fluorescence did not increase with it. The mass spectrum of the reaction solution was measured, but only the peak of the cationic component of PI was observed, that is, the peak of the Please do not adjust margins Please do not adjust margins PI/S 2− conjugated compound was not observed. These results indicate that S 2− reacted with PI, but its reactivity is low. Besides, the estimated quantum yield of the reaction products was very low. Thus, PI could discriminate between CN − and S 2− and, in terms of fluorescence, it has specificity for CN − .

Identification of the fluorescent compound and sensing mechanism
To determine the structure of the PI/CN conjugated compound, we performed 1 H-NMR and MS measurements after the CN − addition to PI. The ratios of the reactants were determined via Job's plots. A 1 H NMR titration study in DMSO-d6 was conducted to acquire an insight into the proposed reaction pathway of CN − and PI (Fig. 6). A CN − solution was added at 0.25-1.25 equiv to PI solution incrementally. According to the amount of CN − added, the signals at 8.30 and 7.52 ppm corresponding to the protons on the (E)-olefin moiety (J =16 Hz) shifted to 6.86 and 6.33 ppm (J =16 Hz), respectively, and the signal of the methyl protons on the nitrogen atom shifted from 4.11 to 2.71 ppm. These results suggest that the chemical shifts of these hydrogens changed significantly because indolium was converted into a tertiary amine upon the CN − addition and the olefin moiety conjugated to indolium ceased to be conjugated to the resulting tertiary amine. The binding mechanism between PI and CN − was further confirmed via the MS measurements.  (Fig. 7). Moreover, their isotopic ratio was consistent with the theoretical value. This result supports the nucleophilic addition of CN − to the PI molecule, and the detection of the cationic component of PI as a fragment ion peak suggests that the cation is very stable. Moreover, to confirm the generation constants of PI and CN − , we analyzed the generation ratio via Job's plot analysis. The maximum value was observed at 0.5, which means that PI and CN − react at a ratio of 1:1 (ESI, Fig. S8). From these results, we concluded that the reaction between PI and CN − is a CN − addition to the C = N + of the indolium ring, yielding (E)-1,3,3trimethyl-2-(2-(10-octyl-10H-phenothiazin-3-yl)vinyl)indoline-2-carbonitrile (PICN). As reported for many CN − sensors containing indolium, the CN − addition to the carbon of the double bond of indolium breaks the conjugation between phenothiazine and indolium, linked by a double bond, inhibiting the ICT from the electron-donating (phenothiazine) to the electron-withdrawing group (indolium) (Scheme 2). As a result, quenching in the visible part and fluorescence emission should occur. The Stokes shift and Ф of PICN were, respectively, 118 nm and 0.60 relative t to quinine (Ф = 0.55 in 0.1 M H2SO4) (ESI ,  Table S2). Therefore, PI reacts with the cyanide ion with a large Stokes shift and a good Ф.  Please do not adjust margins Please do not adjust margins

Computational analysis
The frontier orbitals were obtained through DFT calculations for a compound where the n-octyl group on the phenothiazine of PI is replaced with a methyl group to reduce the computational cost. The higher occupied molecular orbital (HOMO) of PI was widely distributed from the phenothiazine ring to the C = N + site of the indolium group (Fig. 8); this suggests that ICT is occurring, which may have resulted in the loss of fluorescence. The gap between HOMO and lower unoccupied molecular orbital (LUMO) was 2.10 eV (590 nm), which correlates well with the experimental data. The large coefficient on the carbon of the C = N + moiety in the LUMO also explains the nucleophilic addition of CN − to it. The HOMO-LUMO gaps of both the enantiomers of PICN were also calculated and they showed the same value (3.92 eV), higher than that of PI (Fig. 9). This increment in the  energy gap implies the breakage of the π-conjugation between the phenothiazine and indolium moieties and it implies that the ICT process has been stopped.
Time-dependent DFT calculations for the excited state of PICN were also performed to verify the emission. The fluorescence wavelength of both PICN enantiomers was 481 nm, which is in good agreement with the measured values. This result supports the sensing mechanism proposed in the previous section.

Conclusion
We designed and synthesized a new fluorescent chemical sensor, PI, and investigated its response to CN − . PI showed clear color change and fluorescence when adding CN − , as well as sufficient sensitivity with a LOD of 0.02 µM, which is below the maximum value established by the WHO and the toxic CN − concentration in human blood. Among 15 anions, PI reacted only with CN − ; that is, it emitted fluorescence specifically in response to CN − and its fluorescence intensity increased linearly when adding 0-1.3 equiv CN − .The mechanism of fluorescence emission was analyzed through various spectral measurements and DFT. We found that it was caused by the ICT disruption. Moreover, PI can react with CN − in ethanol and, thus, can be used for on-site determination. Therefore, it is a promising candidate for practical CN − sensors.

Conflicts of interest
There are no conflicts to declare.