The detection of water in organic solvents is of significant interest from a variety of industrial, pharmaceutical, and chemical safety perspectives (Mishra and Singh 2021). From an industrial perspective, even minute quantities of water in organic solvents can rapidly contaminate chemical reactions, leading to deactivation of catalysts, a reduction in reaction efficiencies, and the generation of undesired side products (Lin et al., 2012). From a pharmaceutical perspective, the presence of water in organic solvents can dramatically affect the optimized syntheses of pharmaceutically active agents, with significant financial consequences particularly for large-scale syntheses (Thenrajan et al., 2021). From a chemical safety perspective, the incompatibility of a variety of molecules with water is well-known (Zor et al., 2021), and inadvertent exposure to water found as a contaminant in organic solvents can lead to the violent decomposition of water-incompatible reagents, resulting in chemical explosions and/or other undesired highly exothermic reactions (Yang et al., 2019).

Current methods for the detection of water in organic solvents tend to rely on spectral monitoring of the solvent to detect signals that correspond to the presence of water, and include the use of FTIR (Saleh and Tripp 2021), Raman (Yeung and Chan 2010), UV-visible (Mohar 2019), and fluorescence spectroscopy (Zhou et al., 2019; Kumar et al., 2016). Other methods, including gas chromatography (Kay et al., 2021) and electrochemical methods (Marecek and Samec 2020), have also been reported. These methods, which rely either on the detection of water directly or on the detection of signal changes corresponding to other molecules that interact with water, generally lead to extremely high sensitivity and selectivity (Wang et al., 2021). Nonetheless, ongoing challenges in using such methods include the need for costly and non-portable laboratory-grade instruments to monitor such spectral changes, as well as the fact that many of these methods are not reversible. Moreover, toxicity concerns remain around the use of water-detection materials that contain heavy metals, such as mercury and cadmium (Othong et al., 2020); around the use and responsible disposal of sensors composed of graphene oxide (Chi et al., 2021), carbon dots (Lee et al., 2019), and other carbon-based materials (Liu et al., 2020); and around sensors that rely on relatively weak noncovalent interactions in the sensor construction (Wu et al., 2020), and the potential for the degradation of such sensors to lead to environmental contamination (Gao et al., 2010).

Recent reports from one of our research groups have demonstrated the development of high-performance fluorometric (Haynes and Levine 2020) and colorimetric (Haynes et al., 2019) sensors for a variety of organic and inorganic analytes. Moreover, recent successes in novel sensor development have been reported in joint publications from our research groups, and demonstrate that bimane-based supramolecular constructs act as highly effective sensors for cobalt (II) ions (Pramanik et al., 2020) and for molecular iodine (Pramanik et al., 2021). These sensors, which operate in both solution-state and on filter papers to provide fluorometric and colorimetric analyte detection, have notable practical advantages, including their high sensitivity, ease of access, and non-toxicity of both the bimane transducing element and supramolecular cyclodextrin scaffold. The lack of toxicity of both cyclodextrin (Kfoury et al., 2019) and bimane (MSDS 38369) has been well-established in the literature, and has led to the use of both of these components in a variety of biologically-relevant applications (Barbosa et al., 2019; Lapidot et al., 2016; Lavis and Raines 2008). Moreover, an additional report from one of our research groups has demonstrated a streamlined approach to access bimane derivatives, which facilitates practically attainable access to a variety of these highly fluorescent structures (Szumski et al., 2021). This straightforward access, combined with our already demonstrated ability to use bimanes as components of effective sensors (Roy et al., 2018), prompted us to investigate the use of bimane derivatives as high impact fluorescent sensors for water contamination. Moreover, a recent theoretical report indicated that bimane is expected to demonstrate significant water-induced fluorescence quenching, although experimental evidence in support of these calculations was not provided (Maillard et al., 2021).

Reported herein are the results of our investigations, which demonstrate that bimane 1 responds to the presence of liquid water in organic solvents and water vapor in high humidity environments, and that such a response occurs both in solution of bimane 1 and on filter papers to which bimane 1 had been adsorbed, leading to colorimetric and fluorimetric changes. Although water sensing via a similar boronate hydrolysis mechanism has been demonstrated using other fluorophores (Selvaraj et al., 2019), it has not yet been demonstrated using the bimane scaffold and represents a significant step towards the establishment of bimanes as a viable option in the tool-box of available fluorophores (Kosower et al., 1979). Determining the linear relationship between the sensor response (either colorimetric or fluorometric) and the amount of water provides a method for its straightforward quantitation. Finally, good reversibility of the paper-based sensors was also demonstrated, as were detailed mechanistic investigations that explain the basis of the sensor response.

The fluorescent sensor for water reported herein relies on bimane 1, which was synthesized via the Sonogashira coupling of diiodobimane 3 with two equivalents of acetylene 4 (Figure 1). This diidobimane was in turn accessed from 5-phenyl-4,4-dihydropyrazol-3-one 2 in a one-pot, three-step sequence in 68% overall yield, representing a marked improvement over multi-step, literature-reported methods that require the use of toxic chlorine gas and provide low overall yields (Kosower et al., 1982). The relative quantum yield of bimane 1 (0.17) and its extinction coefficient (1.23 × 10⁴ L M⁻¹ cm⁻¹) are somewhat lower than values reported for other bimane structures (i.e., 0.83 for an unsymmetrically substituted bimane containing one chlorine substituent) (Szumski et al., 2021), but are in line with Kosower’s observation that electron-donating alkyl and aryl substituents on the syn-bimane structure lead to noticeable reductions in the quantum yield (Kosower et al., 1983).

Interestingly, compound 1 displayed mild solvatochromism in its steady-state photophysical measurements, with slight solvent-induced differences in the absorbance (Figure 2A) and emission (Figure 2B) spectra observed. These changes were markedly more pronounced in a fully aqueous solvent, with water leading to a strong decrease in the visible absorbance band of compound 1, and an even more pronounced decrease in the fluorescence emission. Interestingly, compound 1 in the solid-state also displayed a strong fluorescence emission, which decreased markedly upon the addition of water.

This water-induced quenching displayed a strong solvent-dependent response (Figure 3), with the addition of water to acetonitrile displaying the greatest degree of water-induced fluorescence quenching (Figure 3A, 53% decrease in the integrated fluorescence emission compared to the emission in the absence of water). Extremely high sensitivity for low concentrations of water was also demonstrated (Table 1), with the lowest detection limit for water of 0.018% (v/v) calculated in DMSO. This detection limit is in line with the limits of the most sensitive sensors for water in organic solvents reported to date (Huang et al., 2021), and indicates that the sensor reported herein represents a highly sensitive detection method (Kumar et al., 2021). Of note, all aqueous studies were conducted in double deionized water, so that no interference and/or fluorescence quenching occurred due to residual metal ions. Such ions, in particular palladium (Das et al., 2016), sodium (Roy et al., 2018), and cobalt (Pramanik et al., 2020), have been shown to complex effectively to the bimane core, leading to marked changes in the bimane’s photophysical properties. Also of note, the existence of trace amounts of water even in the “0%” water is likely, based on literature precedent (Son et al., 2001), but this trace amount of water is accounted for as part of the baseline for our fluorescence quenching experiments, in accordance with literature precedent (Porter and Markham 1970).

Moreover, this highly sensitive fluorescence-based quenching response was accompanied by noticeable changes in the color of bimane 1 solutions in varying solvents with increasing percentages of water, which were particularly visible using 365 nm excitation of the bimane 1 solutions. Examples of such dramatic color changes are shown for water added to acetonitrile, THF, and DMSO (Figure 4, top). The quantitative blue values changed as a result of the addition of increasing percentages of water in a manner that was linear for a broad range of water percentages (Figure 4, bottom, Table 2); such linearity, in turn, directly enables the ability to detect unknown water concentrations within such solvent systems. Finally, general applicability for the system was also demonstrated, with both purified milli-Q water and aqueous buffer solutions between pH 5 and pH 7.5 inducing identical colorimetric responses (see ESI for more details).

The mechanism by which the addition of water induces the fluorescence quenching of bimane 1 is likely via the water-induced hydrolysis of the boronate ester moieties on compound 1 to form the hydrolyzed boronic acid 5 (Figure 5). The formation of compound 5 was confirmed by ¹H NMR titrations and high-resolution mass spectrometry (see ESI for more details). Moreover, an analogous hydrolysis reaction occurred in methanol, resulting in the formation of methyl ester-substituted boronate moieties in lieu of the pinacol boronates. The fact that changing the hybridization of boron-containing moieties from tetrahedral to trigonal planar induces such marked fluorescence decreases has been previously reported in the literature (Samaniego Lopez et al., 2015; Williams et al., 2021), and is fully consistent with the results reported herein. Notably, this fluorescence quenching occurs rapidly (within seconds), and remains stable (with less than 5% additional quenching) up to 20 min after the initial contact between compound 1 and water.

Gratifyingly, moving from solution-state water sensing to sensing on a solid-support resulted in analogous water-induced photophysical changes, with the addition of water to filter paper onto which bimane 1 had been adsorbed causing a marked change in the color and fluorescence of the system (Figure 6A). Notably, the procedure by which water is added to these paper sensors (dropping via pipette onto the functionalized paper) is done by design to ensure that we control the amount of water that the paper contacts, but we recognize that such a procedure will require further optimization before real-world device development and deployment. These changes were quantified by RGB color analysis (Figure 6B), which shows that the system becomes rapidly saturated with water, resulting in minimal changes in the RGB values after 100 µL of water were added. Even more significantly, bimane 1-functionalized filter papers responded to the presence of water vapor with an analogously strong sensor response, with exposure of the functionalized paper to a high humidity chamber (99% relative humidity) resulting in analogous changes in the sensor’s photophysical (colorimetric and fluorometric) profile (Figure 6C).

Finally, the bimane 1-functionalized filter papers demonstrated strong reversibility in the sensing of water in both the liquid and vapor phase. Liquid water, which induced noticeable photophysical changes of the bimane 1-functionalized paper (see Figure 6A, above), was effectively removed by drying the papers under ambient conditions followed by brief heating (80 °C for 10 min). This drying allowed for the introduction of water again to the paper, followed by re-drying of the paper under the aforementioned conditions. Overall, 11 cycles of sensing followed by drying were demonstrated, albeit with progressive decreases in the performance observed. This reusability was visible by naked eye detection (Figure 7A), and could also be quantified by measuring the saturation values for each of the papers (Figure 7B). Similarly, the reusability of bimane 1-functionalized papers as humidity sensors was also demonstrated, with 7 cycles of exposure to 99% relative humidity, followed by drying and reuse of the same sensor with comparable performance results (Figure 7C). Of note, at this stage we cannot say with full certainty that in the dry state the original bimane 1 is regenerated. Reaction of bimane 5 with free OH groups in the surface of the paper would also lead to fluorescent bimane based boronate esters, which means that the observed partial reversibility may be a result of the generation of new fluorescent bimane-based boronate esters on the paper support.

The development of methods for the rapid, sensitive, and generally applicable detection of water represents a high priority research area due to its significant applications in chemical safety and industry. The work reported herein represents an important step towards addressing this issue, by reporting that a boronate ester-functionalized bimane, compound 1, undergoes rapid and sensitive fluorescence quenching in the presence of trace quantities of water, both in solution and in the vapor phase. The successful access to this structure due to recent advances in synthetic methodology for syn-bimanes represents an additional important advance, and the reusability of paper-based sensors with compound 1 highlights the strong practical applicability of this system. Future work will be directed towards demonstrating the broad applicability of this sensor in real-world samples, and results of these and other investigations will be reported in due course.