Naphthalimide-Piperazine Derivatives as Multifunctional “On” and “Off” Fluorescent Switches for pH, Hg2+ and Cu2+ Ions

Novel 1,8-naphthalimide-based fluorescent probes NI-1 and NI-2 were designed and screened for use as chemosensors for detection of heavy metal ions. Two moieties, methylpyridine (NI-1) and hydroxyphenyl (NI-2), were attached via piperazine at the C-4 position of the napthalimide core resulting in a notable effect on their spectroscopic properties. NI-1 and NI-2 are pH sensitive and show an increase in fluorescence intensity at around 525 nm (switch “on”) in the acidic environment, with pKa values at 4.98 and 2.91, respectively. Amongst heavy metal ions only Cu2+ and Hg2+ had a significant effect on the spectroscopic properties. The fluorescence of NI-1 is quenched in the presence of either Cu2+ or Hg2+ which is attributed to the formation of 1:1 metal-ligand complexes with binding constants of 3.6 × 105 and 3.9 × 104, respectively. The NI-1 chemosensor can be used for the quantification of Cu2+ ions in sub-micromolar quantities, with a linear range from 250 nM to 4.0 μM and a detection limit of 1.5 × 10−8 M. The linear range for the determination of Hg2+ is from 2 μM to 10 μM, with a detection limit of 8.8 × 10−8 M. Conversely, NI-2 behaves like a typical photoinduced electron transfer (PET) sensor for Hg2+ ions. Here, the formation of a complex with Hg2+ (binding constant 8.3 × 103) turns the green fluorescence of NI-2 into the “on” state. NI-2 showed remarkable selectivity towards Hg2+ ions, allowing for determination of Hg2+ concentration over a linear range of 1.3 μM to 25 μM and a limit of detection of 4.1 × 10−7 M.


Introduction
Due to the well-known impact of heavy metals in the environment and on human health, the development of new, cost-effective, fast and reliable methods for their detection has become urgent [1,2]. Heavy metal detection methods have variously been proposed based on sensing mechanisms that include electrochemical and optical techniques with specifically designed recognition elements, such as molecularly imprinted polymers [3], nanomaterials [4], metal-organic frameworks [5] and N-heterocyclic receptors [6].
Photo-induced electron transfer sensors are particularly popular in molecular fluorescent "switch" designs. In PET systems, a fluorophore is covalently linked with a receptor through a short aliphatic spacer, while both segments maintain their individual properties. The NI-1 and NI-2 compounds each have a typical donor-spacer-acceptor structure (D-π-A), comprising a 1,8-naphthalimide core, which represents the fluorophore, and either a pyridin-2-ylmethyl or hydroxyphenyl moieties attached to the core through piperazine. The moieties act as additional receptor sites for metal ion binding. Due to the presence of the N-substituent containing a 2-hydroxyethylsulfonyl group, the covalent The NI-1 and NI-2 compounds each have a typical donor-spacer-acceptor structure (D-π-A), comprising a 1,8-naphthalimide core, which represents the fluorophore, and either a pyridin-2-ylmethyl or hydroxyphenyl moieties attached to the core through piperazine. The moieties act as additional receptor sites for metal ion binding. Due to the presence of the N-substituent containing a 2-hydroxyethylsulfonyl group, the covalent immobilization of these naphthalimide probes to different substrate materials and to nanoparticles (e.g., cellulose, textiles) is also possible using well-established vinyl sulfonyl immobilization chemistry. This molecular feature opens up a wide range of solid state applications for these probes in the future [44].
Structures of studied naphthalimide derivatives NI-1 and NI-2 are shown in Figure 1. The main structural difference between them is a molecular fragment attached via piperazine moiety at 2-hydroxyethylsulfonyl-functionalised naphthalimide.

Spectral Characteristics of NI-1 and NI-2 in Methanol
Structures of studied naphthalimide derivatives NI-1 and NI-2 are shown in Figure 1. The main structural difference between them is a molecular fragment attached via piperazine moiety at 2-hydroxyethylsulfonyl-functionalised naphthalimide. Absorption and emission spectra of compounds are shown in Figure 2a. The NI-1 and NI-2 compounds each have a typical donor-spacer-acceptor structur (D-π-A), comprising a 1,8-naphthalimide core, which represents the fluorophore, and either a pyridin-2-ylmethyl or hydroxyphenyl moieties attached to the core through pi perazine. The moieties act as additional receptor sites for metal ion binding. Due to th presence of the N-substituent containing a 2-hydroxyethylsulfonyl group, the covalen immobilization of these naphthalimide probes to different substrate materials and t nanoparticles (e.g., cellulose, textiles) is also possible using well-established vinyl sul fonyl immobilization chemistry. This molecular feature opens up a wide range of soli state applications for these probes in the future [44].
Structures of studied naphthalimide derivatives NI-1 and NI-2 are shown in Figur 1. The main structural difference between them is a molecular fragment attached via pi perazine moiety at 2-hydroxyethylsulfonyl-functionalised naphthalimide.

Spectral Characteristics of NI-1 and NI-2 in Methanol
Structures of studied naphthalimide derivatives NI-1 and NI-2 are shown in Figur 1. The main structural difference between them is a molecular fragment attached via pi perazine moiety at 2-hydroxyethylsulfonyl-functionalised naphthalimide.
Absorption and emission spectra of compounds are shown in Figure 2a. Both compounds displayed absorption spectra with the absorption maximum at 410 nm with molar absorption coefficients, ε = 2.41 × 10 4 M −1 cm -1 for NI-1 and ε = 8.52 × 10 3 M −1 cm −1 for NI-2, which can be assigned to π-π* transitions. To the naked eye, NI-1 and NI-2 displayed a very light yellow color (Figure 2b). Considering the fluorescence emission spectra, compounds NI-1 and NI-2 exhibited a single emission band with emission maxima at 523 and 525 nm, respectively, but the intensity was strongly dependent on the attached moiety. NI-1-bearing pyridine functionality is strongly green fluorescence emitting fluorophore, compared to hydroxyphenyl-substituted NI-2. Since the lone electron pair on the oxygen atom is a part of the conjugated system, overall electron density of the hydroxyphenyl moiety is increased, which might repulse the lone electron pair on the piperazine nitrogen. That in turn induces the electron transfer from the piperazine nitrogen to the fluorophore, which lowers the emission rate. On the other hand, the lone electron pair on the pyridine nitrogen is not involved in the conjugated system, making the ring overall electron-deficient. Compounds NI-1 and NI-2 exhibited large Stokes shifts in neutral pH (5558 cm -1 and 5629 cm −1 ) and therefore have great potential for the application as

Effect of pH on Spectral Properties
The spectral properties of NI-1 and NI-2 as a function of pH were investigated using both absorption and fluorescence emission spectroscopy. The potential use of NI-1 and NI-2 as pH-sensing probes is based on the spectral changes caused by the protonation of the piperazine nitrogen atom. Optical pH sensors are characterized by their apparent pK a values in buffer methanol mixtures (2:1), determined by spectrophotometry and spectrofluorimetry. The pH values correspond to the pH values of the aqueous buffers used.

UV-Visible Absorption Spectra
Changes in the pH of solutions caused minor changes in the absorbance spectra of the studied compounds ( Figure S1 (Supplementary Materials)). As the pH decreases from 7.0 to 1.6 the absorption band of NI-1 gradually shifts from 410 nm to 388 nm, generally followed by an increase in absorbance of the maximum. The absorbance ratio at peak values (for protonated and neutral forms) has been used for calculating the apparent pK a value ( Figure S1a,b). NI-2 showed the same behavior upon the pH changes, with hypsochromic shift of absorption maxima from 410 nm to 389 nm upon the protonation along with the increase in the peak intensity, as well as a gradual hyperchromic shift in the pH range 5.0-8.0 ( Figure S1c).

Fluorescence Emission Spectra
The emission spectral responses to pH for NI-1 and NI-2 and the dependence of fluorescence intensity (at 523 nm for NI-1 and 525 nm for NI-2) upon the pH value are shown in Figure 3 and Figure S2. The fluorescence intensity changes of NI-1 and NI-2 upon pH were described well by a sigmoidal function (Boltzmann fitting), Figure 3b.  The effect of pH is more pronounced in the fluorescence spectra than in the absorption spectra of compounds. It is known that the protonation of the nitrogen atom of piperazine stops the PET process from amine donor to the naphthalimide core and therefore, the fluorescence intensity increases gradually with the decrease in pH value [45][46][47]. The fluorescence intensity of both NI-1 and NI-2 were considerably increased upon the protonation, up to 29.6-fold for NI-1 and 162.5-fold for NI-2. Fluorescence turn-ON in acidic solution supports the inhibition of the PET process in the piperazinenaphthalimide derivative upon the protonation of piperazine, as shown in Figure 4. The effect of pH is more pronounced in the fluorescence spectra than in the absorption spectra of compounds. It is known that the protonation of the nitrogen atom of piperazine stops the PET process from amine donor to the naphthalimide core and therefore, the fluorescence intensity increases gradually with the decrease in pH value [45][46][47]. The fluorescence intensity of both NI-1 and NI-2 were considerably increased upon the protonation, up to 29.6-fold for NI-1 and 162.5-fold for NI-2. Fluorescence turn-ON in acidic solution supports the inhibition of the PET process in the piperazine-naphthalimide derivative upon the protonation of piperazine, as shown in Figure 4. orescence intensity at 523 nm vs. pH.
The effect of pH is more pronounced in the fluorescence spectra than in the absorption spectra of compounds. It is known that the protonation of the nitrogen atom of piperazine stops the PET process from amine donor to the naphthalimide core and therefore, the fluorescence intensity increases gradually with the decrease in pH value [45][46][47]. The fluorescence intensity of both NI-1 and NI-2 were considerably increased upon the protonation, up to 29.6-fold for NI-1 and 162.5-fold for NI-2. Fluorescence turn-ON in acidic solution supports the inhibition of the PET process in the piperazinenaphthalimide derivative upon the protonation of piperazine, as shown in Figure 4.  The pH response curves exhibit excellent adjustment and apparent pKa values were calculated to be 4.98 for NI-1 (R 2 = 0.9923) and 2.91 for NI-2 (R 2 = 0.9982).
NI-1 and NI-2 have different substituents on their pH-sensitive amino group. The substituents did not affect the pH sensitivities, as both compounds displayed fluorescence enhancement upon the protonation. The different nature of substituent moieties attached to the protonable piperazine moiety (alkyl or aryl) significantly influenced the photophysical properties of derivatives, showing that the suitable substituents play a crucial role for tuning the pKa and pH sensitivity. Moreover, NI-1 and NI-2 dyes were immobilized onto cellulose layers and the pKa values of were determined to be 6.05 and 2.98, respectively [43]. These results indicate that both the nature of substituents and the immobilization technique have a significant impact on spectroscopic and pH sensing The pH response curves exhibit excellent adjustment and apparent pK a values were calculated to be 4.98 for NI-1 (R 2 = 0.9923) and 2.91 for NI-2 (R 2 = 0.9982).
NI-1 and NI-2 have different substituents on their pH-sensitive amino group. The substituents did not affect the pH sensitivities, as both compounds displayed fluorescence enhancement upon the protonation. The different nature of substituent moieties attached to the protonable piperazine moiety (alkyl or aryl) significantly influenced the photophysical properties of derivatives, showing that the suitable substituents play a crucial role for tuning the pK a and pH sensitivity. Moreover, NI-1 and NI-2 dyes were immobilized onto cellulose layers and the pK a values of were determined to be 6.05 and 2.98, respectively [43]. These results indicate that both the nature of substituents and the immobilization technique have a significant impact on spectroscopic and pH sensing properties and could provide a different application of fluorescent naphthalimide dyes [48][49][50].
The properties of NI-1 and NI-2 in acidic (pH 1.6) and neutral solution (pH 7.0), including absorption and emission wavelength (λ abs , λ emiss ), molar absorption coefficient (ε), fluorescence intensity, Stokes shift and calculated apparent pK a values are summarized in Table 1.

Metal Ion Sensing
The difference in molecular fragments attached via piperazine moiety at 2hydroxyethylsulfonyl-functionalised naphthalimide, affected not only the emission spectra of NI-1 and NI-2, but also their response upon binding with metal cations. Spectroscopic measurements of metal ion sensing were performed in MeOH/buffer (v/v, 1/2, pH 5.5). For the spectroscopic measurements, compound concentration and experimental conditions were adjusted for these two compounds separately, because of their different behavior in solution and response in the presence of metal cations. A pH value of 5.5 for the metal ion studies was chosen as a compromise to maximize sensitivity and to minimize effects of competing equilibrium reactions, i.e., protonation of NI at lower pH values and the metal ion hydrolysis at higher pH values. To evaluate the metal ion sensitivity, preliminary tests were performed under a UV lamp (365 nm) by adding 30.0 µL metal ion stock solution to the compound in MeOH/buffer (v/v, 1/2, pH = 5.5) (c (NI-1) = 6.56 × 10 −6 M) and the results are presented in Figure S3.
NI-1 exhibits bright green fluorescence in MeOH/buffer solution and with the addition of Cu 2+ , fluorescence quenching can be seen by naked eye. Addition of other metal ions did not cause any observable changes in the fluorescence of NI-1;

• Fluorescence titration experiments
The experiments were performed at compound concentration 5.00 × 10 −7 M in MeOH/buffer (1/2, v/v, pH = 5.5) in the presence of 2, 20 and 95 equiv. of metal ions and the results are summarized in Figure 5. A small addition of metal ions (2 equiv.) did not significantly affect the fluorescence intensity of NI-1, while the presence of 20 equiv. of Cu 2+ and Hg 2+ resulted in fluorescence quenching of NI-1 intensity for 46.4% and 22.7%, respectively. Furthermore, 95 equiv. of these two cations induced further quenching of fluorescence intensity of NI-1 for 60.0% upon Cu 2+ and 39.8% upon Hg 2+ addition). Moreover, Al 3+ , Zn 2+ , Ba 2+ , Pb 2+ and Cd 2+ ions present in high concentrations (95 equiv.) induced NI-1 fluorescence intensity increase in the range of 31.1 to 36.0% for Zn 2+ , Ba 2+ , Pb 2+ and Cd 2+ and 47.9% increase in the presence of Al 3+ , as summarized in Figure 5.   Encouraged by these findings, we performed titration experiments with Cu 2+ and Hg 2+ and the results are presented in Figures 6 and 7.
It can be seen from the emission spectra that an increase in the concentration of metal ions leads to a decrease in the fluorescence intensity, and a significant decrease in intensity is achieved at higher concentrations. In the case of Cu 2+ ions, a hypochromic shift occurs, with the maximum of fluorescence intensity at 523 nm, while in the case of Hg 2+ ions, a bathochromic shift occurs with a decrease in intensity. The wavelength of the fluorescence intensity maximum in the presence of Hg 2+ ions shifts from 523 nm to 541 nm. Hypochromic shift is common in naphthalimide derivatives whose fluorescence is quenched in presence of Cu 2+ ions [39]. However, bathochromic shift in the presence of Hg 2+ ions is not common and the possibility of obtaining two different fluorescent responses can potentially be used to discriminate Cu 2+ from Hg 2+ using NI-1.  It can be seen from the emission spectra that an increase in the concentration of metal ions leads to a decrease in the fluorescence intensity, and a significant decrease in intensity is achieved at higher concentrations. In the case of Cu 2+ ions, a hypochromic shift occurs, with the maximum of fluorescence intensity at 523 nm, while in the case of Hg 2+ ions, a bathochromic shift occurs with a decrease in intensity. The wavelength of the fluorescence intensity maximum in the presence of Hg 2+ ions shifts from 523 nm to 541 nm. Hypochromic shift is common in naphthalimide derivatives whose fluorescence is Fluorescence properties of immobilized napthalimide derivatives with (methylpyridine)piperazine receptor in presence of Hg 2+ and Cu 2+ ions have been reported previously. Covalently immobilized probes on a glass carrier were examined at pH 7 in the presence of Hg 2+ ions and have shown an increase in fluorescence with Hg 2+ ions and a slight decrease with Cu 2+ ions [37]. Additionally, Liu et al. [51] synthesized two naphthalimide derivatives, also with (methylpyridine)piperazine as receptor of NI-1, and immobilized them on (poly)styrene substrate. The resulting optodes differ only in the linker (N-substituent of naphthalimide) through which the compounds were attached to the carrier. In both cases, Hg 2+ ions caused a significant increase in fluorescence intensity. It is interesting to notice that, although the receptor part of all four compounds is the same, their response to Hg 2+ and Cu 2+ ions differs significantly. This implies that other factors, such as immobilization method, material of the substrate, length of the linker part on the N-substituent, test conditions-particularly the pH of the solution, solvents and anion composition, etc., have a significant effect on the mechanism of the fluorescence response, possibly causing steric hindrance and influencing binding stoichiometry as well.
Unlike fluorescence emission spectra, UV-Vis absorption spectra of NI-1 were not significantly affected by presence of Hg 2+ and Cu 2+ ions ( Figure S4), showing an absorption maximum at 410 nm which can be assigned to the naphthalimide core; •

Reversibility tests
Reversibility tests were performed with the addition of EDTA to the solution containing NI-1 and metal cations, in five cycles, as shown in Figures S5 and S6 for Cu 2+ and Hg 2+ , respectively. In general, upon addition of EDTA, the fluorescence of NI-1 recovers. In the case of Cu 2+ , it was noticed that the recovered fluorescence intensity of NI-1 increases by approx. 10% upon each addition of EDTA in repeating cycles. This behavior could be attributed to the enhancement of fluorescence of NI-1 caused by other effects, possibly related to the gradually changing composition of the solution with each cycle. In comparison to free NI-1, the solutions containing NI-1, EDTA and excess Hg 2+ ion show fluorescence intensity quenching, varying by approximately 8%, while the maximum shifts from 523 nm to 536 nm (∆λ max = 13 nm);

•
Metal-binding studies The stoichiometric coefficients of Cu 2+ and Hg 2+ ions upon binding with NI-1 were assumed using Job's plot as shown in Figure 8 and a 1:1 binding ratio is suggested for both ions, which is in accordance with previously published results for the same substituent [36,37].

Reversibility tests
Reversibility tests were performed with the addition of EDTA to the solution containing NI-1 and metal cations, in five cycles, as shown in Figures S5 and S6 for Cu 2+ and Hg 2+ , respectively. In general, upon addition of EDTA, the fluorescence of NI-1 recovers. In the case of Cu 2+ , it was noticed that the recovered fluorescence intensity of NI-1 increases by approx. 10% upon each addition of EDTA in repeating cycles. This behavior could be attributed to the enhancement of fluorescence of NI-1 caused by other effects, possibly related to the gradually changing composition of the solution with each cycle. In comparison to free NI-1, the solutions containing NI-1, EDTA and excess Hg 2+ ion show fluorescence intensity quenching, varying by approximately 8%, while the maximum shifts from 523 nm to 536 nm (Δλmax = 13 nm); •

Metal-binding studies
The stoichiometric coefficients of Cu 2+ and Hg 2+ ions upon binding with NI-1 were assumed using Job's plot as shown in Figure 8 and a 1:1 binding ratio is suggested for both ions, which is in accordance with previously published results for the same substituent [36,37]. The equilibrium association (binding) constant for Cu 2+ and Hg 2+ ions and NI-1 binding was determined by Benesi-Hildebrand's plot using the equation: The equilibrium association (binding) constant for Cu 2+ and Hg 2+ ions and NI-1 binding was determined by Benesi-Hildebrand's plot using the equation: where F 0 is the fluorescence intensity in the absence of metal ions; F max -the fluorescence intensity in presence of saturated metal; F-the intensity of the metal-probe complex in the linear range and n-factor of stoichiometric binding (1 for 1:1 ratio). The correlation factor for both ions is R 2 > 0.96 with the binding constant being 3.9 × 10 4 for Hg 2+ and 3.6 × 10 5 for Cu 2+ ( Figure S7); • LOD (limit of detection) The limit of detection (LOD) of NI-1 for Cu 2+ and Hg 2+ ions was determined using the IUPAC-recommended equation: where σ sp is the standard deviation of blank measurements and k is the slope between the fluorescence intensity and metal concentration in linear range as shown in Figure S8. The standard deviation for four blank measurements is σ sp = 0.3295. The detection limit of NI-1 is 1.5 × 10 −8 M and 8.8 × 10 −8 M for Cu 2+ and Hg 2+ ion, respectively. These values are comparable to similar fluorescent quenching chemosensors utilizing the naphthalimide structure with aliphatic amino receptors, reported previously [52][53][54][55]. Wu et al. [37] determined the limit of detection of Hg 2+ ions by naphthalimide-based optode to be 2.00 × 10 −6 M at pH 7.01 in 0.05 M Tris-HCl which is significantly higher than the detection limit of the compound NI-1. Furthermore, Gao et al. determined the detection limit of Cu 2+ ions for three naphthalimide derivatives, which have the same receptor moiety as NI-1, in which fluorescence quenching occurs. The determined detection limits reported are in range from 2.62 × 10 −9 to 1.17 × 10 −8 M [39]. The Stern-Volmer constants for NI-1 were calculated to be 152570 M −1 for Cu 2+ , and 23909 M −1 for Hg 2+ , as determined from the respective Stern-Volmer plots ( Figure S9). Based on this evidence from the literature and the known behavior of Cu(II) as a strong paramagnetic quencher, we assume that the quenching here is also induced by chelation, rather than being collisional. Table 2 shows a comparison between different copper sensors reported in the literature and the napthalimide sensor NI-1 presented in this work.

•
Visual detection of metal ion sensitivity NI-2 did not fluoresce in the MeOH/buffer mixture at pH 5.5. Upon addition of metal cations, a drastic increase in the fluorescence intensity was observed only in the presence of Hg 2+ ions (Figure 9). These initial experiments suggested that NI-2 could be used as selective fluorimetric naked-eye chemosensor for Hg 2+ ; • Visual detection of metal ion sensitivity NI-2 did not fluoresce in the MeOH/buffer mixture at pH 5.5. Upon addition of metal cations, a drastic increase in the fluorescence intensity was observed only in the presence of Hg 2+ ions (Figure 9). These initial experiments suggested that NI-2 could be used as selective fluorimetric naked-eye chemosensor for Hg 2+ ;

•
Titration experiments To investigate the possibility of an NI-2 compound to detect Hg 2+ ions, fluorescence titration was conducted. The fluorescence spectra were recorded at excitation wavelength of 410 nm. As shown in Figure 10, the maximum fluorescence band is centered at 525 nm. The spectrum of the NI-2 compound without Hg 2+ exhibits a low intensity peak around 533 nm, typical for the naphthalimide derivatives containing an amine substituent at the C-4 position of the naphthalene imide ring [61]. The addition of Hg 2+ to the compound NI-2 results in a remarkable enhancement of fluorescence signal but has nearly no effect         To investigate the possibility of an NI-2 compound to detect Hg 2+ ions, fluorescence titration was conducted. The fluorescence spectra were recorded at excitation wavelength of 410 nm. As shown in Figure 10, the maximum fluorescence band is centered at 525 nm. The spectrum of the NI-2 compound without Hg 2+ exhibits a low intensity peak around 533 nm, typical for the naphthalimide derivatives containing an amine substituent at the C-4 position of the naphthalene imide ring [61]. The addition of Hg 2+ to the compound NI-2 results in a remarkable enhancement of fluorescence signal but has nearly no effect on its UV-Vis spectrum ( Figure S10). The fluorescence intensity increases steadily in response to the gradual increase in concentration of added Hg 2+ ion in MeOH/buffer (v/v, 1/2, pH 5.5) at a compound concentration 6.53 × 10 -6 M, while a hypsochromic shift ∆λ F = 6 nm was observed. Such a small shift suggests that upon addition of the metal ions the fluorophore ICT state is not affected and that the complexation takes place via a nitrogen atom on the piperazinyl part and hydroxyphenyl group of the substituent simultaneously, rather than the imide moiety. These results support the potential of the NI-2 compound to be utilized as a highly selective and sensitive fluorescent sensor for Hg 2+ ions;   Compound NI-2 shows a greater selectivity relative to NI-1, which can be attributed to the structural difference of their receptors. Namely, in NI-2 the hydroxyphenyl group is attached directly to piperazine via a nitrogen atom which makes the receptor structure more rigid compared to NI-1 in which pyridine is linked to piperazine by a methylene bond. The stiffer receptor structure in NI-2 prevents it from coordinating various ions as is the case in NI-1;

•
Metal-binding studies Using Job's method, analogous to NI-1, the supposed stoichiometric binding coefficient of Hg 2+ and NI-2 was determined in MeOH/buffer mixture at pH 5.5. As shown in Figure S11, the dependence of fluorescence intensity at 525 nm on the molar content of NI-2 in solution in the presence of Hg 2+ ions might vaguely suggest a 2:1 metal-ligand binding ratio, which, given the structure of ligand, is however unlikely. We could conclude here that Job's plot failed to reveal the real stoichiometry of the complex which is not unusual given the known limitations of the method [62].
The Benesi-Hildebrand's method was employed by plotting the F 0 /(F-F 0 ) against 1/[Hg 2+ ], where F is the fluorescence intensity at 525 nm in the presence of the metal ion at the given concentration, and F 0 is the fluorescence intensity of the free compound ( Figure S12). The association constant was evaluated from the ratio of the intercept and the slope to be 8278 M −1 , which is similar to previously reported results [37]; • LOD (limit of detection) As one of the most important sensor features for useful applications, the limit of detection (LOD) of NI-2 for Hg 2+ ions was determined using the IUPAC-recommended equation. Figure S13 showed good linearity between the emission at 525 nm and concentrations of Hg 2+ in the range from 0 to 25.0 µM, indicating that the compound NI-2 can measure specific concentrations of Hg 2+ quantitatively.
In the linear range between 0-25 µM of the calibration curve, the limit of detection can be obtained as LOD = 3σ/k, where σ is standard deviation of blank measurements and k is the slope between the fluorescence intensity and the metal concentration. The standard deviation for four blank measurements is σ = 0,621 and the limit of detection was calculated to be 4.1 × 10 −7 M, being comparable with similar chemosensors based on naphthalimides [26,37,63]; • Reversibility The reversibility test was performed by the addition of Hg 2+ and EDTA to the NI-2 solution. The addition of Hg 2+ increased fluorescence intensity of NI-2 as shown in Figure  S14, but there was no change in the spectrum after the addition of EDTA to the solution of NI-2-Hg 2+ . Table 3 shows a comparison between the different mercury sensors and napthalimide sensors NI-1 and NI-2 presented in this work.      It is important to note that nearly all the literature data in the area of chemosensors research for determination of complex stoichiometry, binding constants and quenching constants are based on the use of "old" graphical methods, such as Job's plot, Benesi-Hildebrand (BH) or Stern-Volmer (SV) plots. Those methods show significant limitations and are often used non-critically to draw conclusions that cannot be made without more detailed analysis [67][68][69]. In this context, it is important to stress that all constants reported in our work are conditional constants. They were calculated under defined conditions, i.e., certain mixtures of solvents and pH values of buffer solutions. In addition to chelation equilibrium reactions assumed to be happening during metal-ion titrations, there are other possible reactions simultaneously happening in the solution including protolytic equilibrium of the sensor molecules and hydrolysis of metal cations. Calculation of exact binding constants would require more sophisticated calculation tools and different measurements, which for purely analytical applications is usually not necessary. Nevertheless, one has to be aware of the limitations of these graphical methods and use them accordingly.

Reagents and Instruments
Indicator dyes NI-1 and NI-2 were from Joanneum Research Forschungsgesellschaft mbH (Austria) [43]. All chemicals and solvents were purchased from commercial suppliers Acros, Sigma Aldrich or Fluka and were used as received without further purification. Methanol for spectroscopic measurements was of spectroscopic (HPLC) grade. A phosphate buffer was used for the near neutral pH range (5.0-8.0) and citric buffer for the acidic pH range (1. 2-4.8). Buffer solutions were prepared using chemicals of analytical grade following the procedure [70]. The pH values of buffer solutions were monitored using MA 5740 pH meter. The 1.0 × 10 −3 M stock solutions of metal ions (Ag + , Mg 2+ , Ba 2+ , Cu 2+ , Hg 2+ , Zn 2+ , Co 2+ , Fe 2+ , Ni 2+ , Pb 2+ , Mn 2+ , Cd 2+ , Fe 3+ , Al 3+ ) were prepared by dissolving metal salts (AgNO 3 , MgCl 2 , BaCl 2 , CuCl 2 , Hg(NO 3 ) 2 , ZnSO 4 , Co(CH 3 CO 2 ) 2 , FeSO 4 , NiSO 4 , Pb(NO 3 ) 2 , MnCl 2 , (CH 3 CO 2 ) 2 Cd, FeCl 3 , AlCl 3 ) in 0.01 M HNO 3 . The UV-Vis spectra were recorded, against the solvent, at 25 • C, on a Varian Cary spectrophotometer in double-beam mode using quartz cuvette (1 cm). Wavelength scan was performed between 200 nm and 800 nm. Baseline was recorded prior to each set of experiments. Fluorescence measurements were carried out on a Varian Cary Eclipse fluorescence spectrophotometer. Emission spectra were recorded from 380 nm to 800 nm with excitation wavelength at 410 nm, at 25 • C using 1 cm path quartz cells. Relative fluorescence quantum yields were determined according to Miller using Equation (3): where Φ is the emission quantum yield; A is the absorbance at the excitation wavelength; D is the area under the corrected emission curve and n is the refractive index of the solvents used. The subscripts s and x refer to the standard and to the unknown, respectively. The standard employed was quinine sulphate with a published fluorescence quantum yield of 0.54 [71].

UV-Vis Absorbance and Fluorescence Spectra Measurements
Stock solutions of compounds were prepared by dissolving 1.0 mg of compound in 100.0 mL methanol to form a 1.97 × 10 −5 M (NI-1) and 1.96 × 10 −5 M (NI-2) solutions.

The Effect of pH
In order to study the effect of pH on the spectroscopic properties of new compounds, UV-Vis and fluorescence emission spectra were recorded covering the pH range 1.6-8.0 for NI-1 or 1.2-8.0 for NI-2. Testing solutions of compounds were prepared by diluting 1.0 mL of stock solution with 2.0 mL buffer and the final concentrations of NI-1 and NI-2 for the absorbance and fluorescence measurements were 6.56 × 10 −6 M and 6.54 × 10 −6 M, respectively.

Metal Ion Sensing
All measurements of fluorescent properties were obtained in MeOH/buffer pH 5.5 solution (1:2) with excitation wavelength at 410 nm. The visual evaluation of NI-1 and NI-2 sensing properties towards metal cations were performed prior to spectroscopic measurements. The experiment was carried out in 1.0 mL MeOH/buffer mixture (1:2) at a compound concentration 6.56 × 10 −6 M and 6.54 × 10 −6 M, respectively, followed by addition of 30.0 µL (2.00 equiv.) of each metal ion stock solution.
The spectroscopic evaluation of the metal-sensing properties of NI-1 was carried out at a concentration of 5.0 × 10 −7 M. The effect of metal cation on emission spectra was tested by adding 3.0, 30.0 and 150.0 µL of stock solution of each metal cation. For titration experiments, in the prepared solution of compound, small aliquots of stock solutions of metal ions were added. The resulting solution of compound and metal ions were mixed and the fluorescence spectra were recorded after 2 min.
For the evaluation of the sensing properties of NI-2 towards different metal cations, spectroscopic measurements were obtained at a compound concentration of 6.54 × 10 −6 M, followed by addition of 150.0 µL of metal ion stock solution.

Conclusions
In this paper we investigated the photophysical characteristics of two novel 1,8naphthalimide-based fluorescent compounds, NI-1 and NI-2, and their potential use as optical probes for pH and heavy metal ions. Different molecular moieties attached at the C-4 position in NI-1 and NI-2 show a notable effect on the optical properties of the derivatives. NI-1 bearing a methyl-pyridyl substituent exhibits 6-fold higher quantum yield than hydroxyphenyl substituted NI-2. Both NI-1 and NI-2 showed fluorescence intensity increase with decreasing pH. The corresponding pK a values for NI-1 and NI-2 were determined to be 4.98 and 2.91, respectively.
When it comes to metal sensing, NI-1 and NI-2 show opposite behaviors. The compound NI-1 shows high sensitivity and selectivity toward Cu 2+ and Hg 2+ , via fluorescence quenching (switch "off" probe), while the compound NI-2 undergoes significant fluorescence enhancement exclusively by Hg 2+ ions (switch "on" probe).
The fluorescence intensity of NI-1 decreases upon complexation with Cu 2+ and the chemosensor can be applied to the quantification of Cu 2+ ions within a linear range from 2.5 × 10 −7 M to 4.0 × 10 −6 M with a detection limit of 1.5 × 10 −8 M. The linear range for Hg 2+ was determined to be 2.0 × 10 −6 M to 1.0 × 10 −5 M, with a detection limit of 8.8 × 10 −8 M.
Coordination of the receptor in NI-2 with Hg 2+ hinders the PET mechanism from the aliphatic amino group to the naphthalimide core, resulting in fluorescence enhancement. Therefore, the NI-2 compound can be utilized as a turn-on PET sensor for Hg 2+ ions. The results show a remarkable selectivity towards Hg 2+ ions, with a linear detection range from 1.3 × 10 −6 to 2.5 × 10 −5 M and a limit of detection of 4.1 × 10 −7 M.
Due to the presence of a 2-hydroxyethylsulfonyl group, the covalent immobilization of these naphthalimide probes to different substrate materials and to nanoparticles (e.g., cellulose, textiles) is possible, opening up a wide range of solid state applications for these molecules in the future.