Reversal of a Fluorescent Fluoride Chemosensor from Turn-Off to Turn-On Based on Aggregation Induced Emission Properties

Here we present a new approach for the development of fluoride chemosensors taking advantage of aggregation induced emission (AIE) properties. Although AIE-based chemosensors have been described, they rely primarily on the analyte causing aggregation and hence fluorescence. We propose a new concept in the use of AIE for the development of fluorescent sensors. Our hypothesis is based on the fact that a turn-off chemosensor in solution can be transformed into turn-on in the solid state if the properties of ACQ and AIE are properly combined between the fluorescent molecules involved. To demonstrate this hypothesis, we have selected a fluorescent chemosensor for the fluoride anion with a conjugated structure of bis(styryl)pyrimidine that, while showing turn-off behavior in solution, becomes turn-on when it is brought to the solid state. We have also combined it with the advantages of a detection system based on the microfluidic paper-based analytical devices (μPAD). The system is fully characterized spectroscopically both in solution and in the solid state, and quantum mechanical calculations were performed to explain how the sensor works. The prepared device presents a high sensitivity, with no interference and with an LoD and LoQ that allow determination of fluoride concentrations in water 2 orders of magnitude below the maximum allowed by WHO.


Apparatus, instruments and software
1 H NMR and 13 C NMR spectra were acquired at room temperature in VARIAN 400MHz. The chemical shifts () are given in ppm and are referenced to the residual protons of the deuterated solvent or carbon nuclei of chloroform ( 1 H,  = 7.27 ppm; 13 C,  = 77.0 ppm) or DMSO ( 1 H,  = 2.50 ppm; 13 C,  = 39.5 ppm). Acidic impurities in CDCl 3 were removed by treatment with anhydrous K 2 CO 3 . IR spectra were recorded on an FT-IR spectrophotometer equipped with an ATR accessory and the main peaks are given in cm -1 . MALDI-TOF mass spectrometry was carried out on an Autoflex maX (Bruker) and high-resolution mass spectrometry (HR-MS) was carries out on an LCT Premier XE.
The µPAD were fabricated using a craft-cutting technique by cutting the paper using a 12W CO 2 laser engraver (Rayjet, Barcerlona, Spain). Fluorescence spectrums were collected using a Varian Cary Eclipse luminescence spectrometer (Varian Ibérica, Madrid, Spain). For µPAD samples, excitation spectra were recorded on a Varian Cary Eclipse fluorescence spectrophotometer in the range 550 -800 nm with a xenon lamp as excitation source.
Luminescence intensity was measured using the following conditions: λ ex = 530 nm and λ em = 617 nm, the photomultiplier voltage was set at 550 V and excitation and emission slits of 5 nm were used.
UV-Vis absorption spectra were acquired on a Cary 100 (Varian) spectrophotometer at room temperature using a slit width of 0.4 nm and scan rate of 600 nm/min. Steady state fluorescence were recorded on an FLS920 system (Edinburgh Instruments) equipped with a time correlated single photon counting (TCSPC) detector. A TLC 50 temperature-controlled cuvette holder (Quantum Northwest) was used for the measurements (temperature was controlled at 296 S3 K). Quartz cuvettes (Hellma Analytics) of 10 mm were employed for all the spectroscopic measurements. For SSF spectra, a Xe lamp of 450 W was used as the light source and the excitation and emission slits were both fixed at 1 nm. The step and dwell time were 1 nm and 0.1 s, respectively. The quantum yields of the compounds in solution and solid state were recorded on an FS5 system (Edinburgh Instruments) equipped with a time-correlated single-photon counting (TCSPC) detector and an integrating sphere module (SC-30). The quantum yields of solutions were measured using 10 mm quartz cuvettes (Hellma Analytics). In solid, the quantum yields were obtained from the corresponding drop cast samples over quartz slides and directly from the Pad paper and measuring both the direct and the indirect fluorescence emission. In all cases, the step and dwell time were 1 nm and 0.3 s, respectively.
The microstructure was evaluated by Scanning Electron Microscopy (SEM) using a Jeol 6490LV electron microscope equipped with SE and BSE detectors and an Oxford Link EDS Probe. The samples were prepared by placing a spatula tip of the organogel on top of a sampler stub and allowing them to dry for 6 h at room temperature. Specimens for SEM observation required coating (Au-Pt, Emitech) to avoid problems related to the surface charging-up.

Synthesis of SP-OSi
Compounds 1 and 3 were prepared according to literature 1 and compound 4 was prepared as follows: Cl Scheme S1. Synthesis of SP-OSi 4.

Computational details
The molecular geometry of the ground, anionic and first excited states were optimized using the density functional theory (DFT) method implemented in the Gaussian 16 programme package M06-2X functional was used along with the 6-31+G** basis set which has proven to be suitable to estimate the singlet excited state in previous works. 3 The vibrational frequencies were also calculated for both ground and excited stets to check the absence of imaginary frequencies. The energy barrier to rotation of the N-C-C-C dihedral angle was calculated to determine the most stable conformation (see Figure S3). The polarisable continuum model (PCM) was used to consider the effect of the solvent as implemented in the Gaussian package. 4 Time-dependent DFT calculations at the TD-M06-2X/6-31+G** were performed in dimethylsulfide. The reorganisation energy upon excitation, , was calculated using the programme DUSHIN developed by Reimers 5 according to: is the wavenumber associated with the vibrational mode i and S i is the Huang-Rhys  (HR) factor.   S10 Table S1. Calculated lowest-energy transition wavelengths (λ ab calc ), oscillator strengths (f) and orbital contributions for these transitions. Calculations were carried out at the TD-M062X/6-31+G** level of theory in DMSO solution. Calculated wavelengths (λ emis calc ) for the S 1 S 0 transition.    Figure S9 shows the titration of SP-OSi with acid, exhibiting a single equivalence point at pH=5.18. When a similar titration was conducted using an equivalent amount of water to the case of HCl, the quenching of fluorescence was negligible even after adding 7 times the volume usually employed for titrations (2.5 L). This experiment supports the fact that variation of fluorescence is not caused by the increase of polarity due to the addition of water. To study any effect of a polarity, increase due to the presence of ions, the averages of emission intensities for SP-OSi solutions in DMSO before and after the addition (2.5 L) of different inert salts at 1 mM concentrations were compared. Figure  S6b shows the average values and their standard deviations (n = 18). These averages were compared using a two-tailed t-test assay with Welch's correction at a 95% confidence level; obtained p-value was 0.3778, which stated that there were no significant differences between them, pointing to the absence of any polarity influence on the fluorescence using the reported volumes of aqueous solutions. Precision, % 1.5 Table S4. Analytical parameters of µPAD for fluoride ions determination.

Fabrication and preparation of the µPAD
Microfluidic paper-based analytical devices were fabricated using a paper sheet (Whatman nº1) by craft-cutting employed a CO 2 laser engraver. The pattern of the µPAD was first designed using Illustrator software (Adobe Systems) and the design was cut by CO 2 laser.
The µPAD consists of two layers prepared individually, as shown in Figure S7. The first layer is the detection area where the SP-OSi is immobilized. The second layer is made up of two separates areas, one for sampling and the second for transport of the sample. To prepare the µPAD, 4 µL of SP-OSi in 10 M THF solution was dispensed onto the detection area by drop casting and 2 µL of phosphate buffer 100 mM pH 7 was added on the transport channel, both pieces of paper were dried at room temperature for 2 min and stored in a dry environment at 4ºC in the dark.   S18 concentrations from 11.9 -125.0 µM, and the detection limit was evaluated to be 3.9 µM, indicating that SP-OSi has a very good sensitivity to the detection of fluoride. LOQ, µM 11.9 Precision, % 1.3 Table S5. Analytical parameters of µPAD for fluoride ions determination.

S19
In order to evaluate the operating applicability of the sensor in real samples it was carried out the analysis of samples of tap water, brick green tea infusion and commercial mouthwash. All the samples were collected in clean polyethylene bottle and were stored at -18ºC until they were analyzed. The sample of tap water and the commercial green tea were filtered using a filter of 0.45 µm to remove the impurities. Different concentrations of fluoride ions were added to the samples of water and green tea, then these samples were analyzed using the sensor. For the sample of the mouthwash solution, a dilution of 1:100 with distilled water was carried out and the fluoride ions concentration was determined using the PAD sensor such as the analytical protocol indicated. On the other hand, the sample of toothpaste (1g) was dissolved in 100 mL of distilled water heating for 10 minutes in a water bath. Later, 100 µL of propanol was added to reduce foaming and the sample was diluted 100-fold. 6 The concentration average of fluoride for each sample was obtained from three independent measurements and presented with the associated standard deviation.
For fluoride determination, 20 µL of a solution of different concentrations of fluoride is dropped onto the sampling area of the µPAD. After the solution flows by capillarity towards the areas where the reagents are placed, twenty minutes are needed to complete the reaction, when a blue or red colour appears in the white detection area, depends on the concentration of fluoride in the sample. Once the sensing area is allowed to react a fluorescence measurement is carried out using a luminescence spectrometer. For the luminescence measurement, the detection area of the µPAD was excited at 530 nm.