A chromo-fluorogenic synthetic “ canary ” for CO detection based on a pyrenylvinyl ruthenium ( II ) complex

The chromo-fluorogenic detection of carbon monoxide in air has been achieved using a simple, inexpensive system based on ruthenium(II). This probe shows exceptional sensitivity and selectivity in its sensing behavior in the solid state. A color response visible to the naked eye is observed at 5 ppb of CO, and a remarkably clear color change occurs from orange to yellow at the onset of toxic CO concentrations (100 ppm) in air. Even greater sensitivity (1 ppb) can be achieved through a substantial increase in turn-on emission fluorescence in the presence of carbon monoxide, both in air and in solution. No response is observed with other gases including water vapor. Immobilization of the probe on a cellulose strip allows the system to be applied in its current form in a simple optoelectronic device to give a numerical reading and/or alarm.

monoxide in air has been achieved using a simple, inexpensive system based on ruthenium(II). This probe shows exceptional sensitivity and selectivity in its sensing behavior in the solid state. A color response visible to the naked eye is observed at 5 ppb of CO and a remarkably clear color change occurs from orange to yellow at the onset of toxic CO concentrations (100 ppm) in air. Even greater sensitivity (1 ppb) can be achieved through a substantial increase in turn-on emission fluorescence in the presence of carbon monoxide, both in air and in solution. No response is observed with other gases including water vapor. Immobilization of the probe on a cellulose strip allows the system to be applied in its current form in a simple optoelectronic device to give a numerical reading and/or alarm.
The development of electronic household detectors for harmful gases dates from the 1980s and 90s. Back in 1911, canaries were traditionally used in coal mines as an early detection system against life threatening gases such as carbon dioxide, carbon monoxide and methane. The canary, normally a very songful bird, would stop singing and eventually die in the presence of these gases, signaling to the miners to exit the mine. Although considered as moderately toxic compared to other highly poisonous gases, carbon monoxide (CO) detection has always been of vital importance as it lacks color, odor or taste and it is present in numerous natural and artificial environments.
1 Carbon monoxide is a temporary atmospheric pollutant in some urban areas, mainly arising from the exhaust of internal combustion engines (such as vehicles, portable generators, lawn mowers and power washers), but also from incomplete combustion of other fuels (including wood, coal, charcoal, oil, paraffin, propane, natural gas, and waste). Many technologies and devices (electrochemical, 2 optical, 3 and those based on metal-oxide semi-conductors 4 ) have been developed to detect, monitor, and alert the leakage of carbon monoxide. These systems suffer from drawbacks such as false alarms arising due to water vapor or airborne particulates. As an alternative to these systems the design of molecular probes for the chromofluorogenic recognition of CO is of importance. In particular, colorimetric methods are undemanding, giving advantages such as real-time monitoring and the use of simple and inexpensive instrumentation. Moreover, certain colorimetric changes can be observed by the naked eye, thus making chromogenic protocols matchless for certain applications.
Optical detection of carbon monoxide dates back to the late twentieth century when the presence of CO was revealed by a chemically infused paper that turned brown when exposed to the gas. However, only in the last few years has the number of chromogenic probes for CO detection based on new sensing concepts increased. In this field, oxoacetatobridged triruthenium cluster complexes, 5 glass-immobilized rhodium complexes, 6 iron porphyrin functionalized polypyrroles 7 and a phosphino diaminopyridine iron complex 8 have been reported to display color modulations in the presence of carbon monoxide. More recently the optical detection of CO using bimetallic rhodium complexes has also been reported by some of us. 9 Regardless of several advantages offered by chromogenic sensors, only a few probes for carbon monoxide detection using emission changes have been reported so far involving the use of iron 10 and palladium 11 complexes. However, in some of these reported systems the modulations caused by the presence of carbon monoxide reveal particular shortcomings, typically involving poor color or emission changes, sensing in solution but not in air and high detection limits which hamper the use of these probes as viable sensing systems.
Within this context, we have taken advantage of the wellknown ability of alkenyl-ruthenium(II) complexes to react with small donor ligands such as carbon monoxide 12 to explore their use as potential colorimetric probes for CO sensing.
The organometallic probe, [Ru(CH=CHPyr-1)Cl(CO)(BTD)(PPh 3 ) 2 ] (1) (Scheme 1), consists of a Ru(II) metal center bonded to a fluorogenic pyrenylvinyl (CH=CH-Pyr-1) ligand trans to a 2,1,3-benzothiadiazole (BTD) molecule that acts as an emission quencher (vide infra). The coordination sphere in 1 is completed by two triphenylphosphine ligands, a chloride and a CO moiety. The probe was designed to express a dual chromo-and fluorogenic response to carbon monoxide. Thus, the presence in the complex of a pyrene donor group and a BTD acceptor was expected to result in a ligand-to-ligand charge transfer band that would inhibit the fluorescence emission from pyrene. The interchange of the labile quencher BTD by CO to give 2 was envisaged to induce both a revival of the pyrene fluorescence and a color change due to the modification of the coordination sphere of the Ru(II) center.
Probe 1 was prepared in 96% yield following the simple and extensively utilized synthesis of alkenyl ruthenium complexes through hydroruthenation of a hydride complex by a suitable terminal alkyne. 13 This process consists of the regioand stereospecific insertion of a terminal alkyne into the Ru-H bond. Thus, the brightly colored σ-alkenyl 18-electron adduct 1 (see Scheme 1) was readily synthesized by reaction of the hydride [RuHCl(CO)(PPh 3 ) 3 ] 14 with 1-ethynylpyrene and BTD 15 in dichloromethane at room temperature.

Scheme 1. Formation of complex [Ru(CH=CHPyr-1)Cl(CO)(BTD) (PPh 3 ) 2 ] 1 and complex [Ru(CH=CHPyr-1)Cl(CO) 2 (PPh 3 ) 2 ] 2.
X-ray quality crystals of 1 were obtained by recrystallization of the complex via vapor diffusion of diethyl ether onto a dichloromethane solution of 1. Figure 1 provides a plot of the complex showing the atomic numbering scheme. In the structure, the Ru(II) center adopts a distorted octahedral coordination environment with two triphenylphosphine ligands in axial positions and four ligands (Cl, CO, BTD and vinylpyrene) occupying the equatorial sites.
The electronic spectrum of a methanol solution of complex 1 in the visible region is dominated by a moderately intense absorption band at 390 nm that is tentatively attributed to a pyrenylvinyl-to-BTD ligand-to-ligand charge transfer transition (LLCT). 16 In a preliminary test, air samples containing CO were bubbled into dichloromethane or methanol solutions of 1 and a remarkable color change from red to yellow was observed. This color modulation is concomitant with the disappearance of the intense and broad band at 390 nm which additionally reveals the typical absorption bands of the pyrene group at 339 and 355 nm. 17 All these changes are consistent with displacement of the BTD ligand by CO and the formation of complex [Ru(CH=CHPyr-1)Cl(CO) 2 (PPh 3 ) 2 ] (2) (Scheme 1). Suitable crystals of 2 for X-ray single-crystal diffraction studies were obtained by diethyl ether diffusion onto a dichloromethane solution of 1 under an atmosphere of CO. Figure 2 shows a plot of 2 which reveals the presence of a CO molecule trans to the pyrenylvinyl ligand due to the displacement of the BTD group, supporting the proposed mechanism.   (6), C20-Ru-C1 178.37 (8), C20-Ru-P2 93.09 (6).
Moreover, the displacement of the BTD ligand by CO also results in the recovery of the fluorescence emission of the pyrene group. Thus, whereas 1 is very weakly fluorescent in methanol (λ exc = 355 nm, λ em = 458 nm) formation of 2 results in a remarkable 36-fold increase in emission. This can be seen in Figure 3 which shows the effect of bubbling increasing volumes of CO through a methanolic solution of compound 1. Further experiments demonstrated that a similar chromo-fluorogenic response from 1 in the presence of CO was observed in methanol:water 1:1 v/v solutions (complex 1 is not soluble in higher proportions of water). Titration experiments with carbon monoxide in methanol:water 1:1 v/v allowed the determination of a limit of detection (LOD) as low as 1 ppb. The turn-on fluorescence is tentatively explained by the fact that BTD displacement eliminates the electron transfer between pyrenylvinyl and BTD ligands in 1 resulting in a revival of the emission. Motivated by the favorable chromo-fluorogenic sensing features of 1 observed in solution, the system was developed with a view to exploiting the potential use of this compound for CO detection in air. With this aim in mind, compound 1 was adsorbed on silica gel resulting in an orange solid. In a typical test, this colored silica support containing the ruthenium probe was exposed to air containing different concentrations of carbon monoxide. An instantaneous and remarkable modulation of color from orange to yellow was observed, consistent with the formation of 2 (Figure 4). From further titration studies a detection limit as low as 0.6 ppb of CO in air was obtained. Furthermore, a visual color change to the naked eye from orange to yellow was observed for CO concentrations in air as low as 5 ppb. At concentrations of ca. 100 ppm, which is the concentration at which CO becomes toxic for short-term exposure to humans, the color change is extremely clear (see Figure 4). Furthermore, the chromogenic response to CO in air using 1 absorbed on silica gel was found to be highly selective over other gases tested. For instance no color changes were found in the presence of Ar, N 2 , O 2 , CO 2 , H 2 S, coordinating gases such as SO 2 and NO x , nor in the presence of common volatile organic compounds such as chloroform, formaldehyde, acetone, ethanol, toluene or hexane. Only acetonitrile induced some evident color change to yellow, but only at concentrations of ca. 5000 ppm. Importantly in terms of potential applications (where steam may be present), exposure of 1 to water-saturated air did not trigger any chromogenic response.
These data show silica to be a simple and effective support for the chromogenic detection of CO in air using complex 1. However, the fluorogenic response observed for 1 in the presence of carbon monoxide was relatively poor when using silica as the support. In contrast, 1 was found to display a clear turn-on emission enhancement at 495 nm in the presence of carbon monoxide when adsorbed on strips of cellulose paper. This support also offers many practical benefits as it avoids the need for silica and is easier to handle. It is also compatible with simple optoelectronic devices which can convert color changes into numerical readings. 9c Using cellulose paper the response of the probe was studied by monitoring the emission changes upon exposure to increasing concentrations of CO and a remarkable LOD of 0.7 ppb was calculated from these experiments. A clear optical response to the naked eye can also be observed for concentrations of ca. 90 ppm (see Figure 4) using a conventional UV lamp, or simply the readily apparent color change. Complex 1 is also highly selective to CO on this support and it was confirmed that no changes in color or in emission were observed in the presence of Ar, N 2 , O 2 , CO 2 , SO 2 , NO x , H 2 S and common volatile organic compounds (e.g. chloroform, formaldehyde, acetone, ethanol, toluene or hexane). In summary, we report the design and use of a ruthenium(II) complex (1) for the simple, selective chromofluorogenic detection of carbon monoxide. To our knowledge, this is the first complex capable of both chromogenic and fluorogenic sensing of CO in air. In addition, the combination of two sensing modalities allows both simple visible detection as well as greater sensitivity when desired. The probe shows a color change visible to the naked eye at CO concentrations of 5 ppb and a remarkably clear change from orange to yellow at CO concentrations of 100 ppm in air. This serves as a clear warning at levels which become toxic on prolonged exposure to the gas. Moreover, 1 also displays a turn-on fluorescence emission in the presence of carbon monoxide. Both the color and turn-on emission modulations observed are highly selective and due to a displacement of the BTD ligand in 1 by CO to yield complex 2. This exceptionally high selectivity for CO over water vapor is of crucial importance in the potential application of this system in a domestic setting where steam is present (a shortcoming of commercial devices). The high-yielding and straightforward synthetic procedure used to prepare 1 in air, coupled with the commercial availability and relatively low cost of ruthenium and other reagents, render compound 1 both accessible and inexpensive. These attributes make this system suitable for implementation in an easy-to-use portable optoelectronic device, 9c allowing 1 to be applied as an efficient chemosensor for the simple chromo-fluorogenic detection of this colorless and odorless killer.

ASSOCIATED CONTENT Supporting Information
General considerations. Instrumentation. Synthesis of [Ru(CH=CHPyr-1)Cl(CO)(BTD)(PPh 3 ) 2 ] (1) and [Ru(CH=CHPyr-1)Cl(CO)2(PPh 3 ) 2 ] (2). Silica gel immobilization of complex 1. Preparation of cellulose probe of 1. Reactivity of 1 with CO. Crystallographic data and structure of 1 and 2. Selected geometric data for compounds 1 and 2 at 173 K. Fluorescence calibration curve for 1 in aqueous methanol solution and response upon addition of CO. Sensitivity studies with 1 adsorbed on silica upon addition of CO. Selectivity studies with 1 adsorbed on silica. Sensitivity studies with 1 deposited on cellulose strip upon addition of CO. This material is available free of charge via the Internet at http://pubs.acs.org.

Supporting Information
A chromo-fluorogenic synthetic "canary" for CO detection based on a pyrenylvinyl ruthenium(II) complex General Considerations. The complex [RuHCl(CO)(PPh 3 ) 3 ] was prepared by a published procedure. S1 The chemicals 1-ethynylpyrene and 2,1,3-benzothiadiazole were provided by Alfa-Aesar and Sigma-Aldrich, respectively. Tris(2,2′-bipyridyl)dichlororuthenium(II) hexahydrate was purchased from Sigma-Aldrich and used as a reference compound for quantum yield calculations. All solvents were of analytical grade. Solvent mixtures are volume/volume mixtures. Solvents used for UV-Vis and fluorescence measurements were thoroughly degassed with N 2 . All experiments and manipulations of compounds were conducted in air, unless otherwise specified. Carbon monoxide was provided by a commercially available CO cylinder. The procedures given provide materials of sufficient purity for synthetic and spectroscopic purposes.
Instrumentation. NMR spectroscopy was performed at 25 °C using a Bruker AV400 spectrometer in CDCl 3 unless stated otherwise. Chemical shifts are reported in ppm and coupling constants (J) are in Hertz. Elemental analysis data were obtained from London Metropolitan University. Mass spectrometry analyses were performed at the Servei d'Espectrometria de Masses of the Universitat de València, Spain, on a Quadrupole time of flight (QqTOF) 5600 system (Applied Biosystems-MDS Sciex) spectrometer for high resolution mass spectra. The HRMS was operated in positive mode under the following conditions: Gas1 35 psi, GS2 35, CUR 25, temperature 450 ºC, ion spray, voltage 5500 V. UV-Vis spectra were recorded using a Jasco V-650 spectrophotometer equipped with a diffuse reflectance sphere (model ISV-722) for measurements on solids. In the latter case, measurements were conducted at room temperature over a wavelength range of 350-800 nm with a wavelength step of 1 nm. Fluorescence measurements were carried out using a Jasco FP-8500 spectrophotometer. Carbon monoxide concentrations were measured using an ambient carbon monoxide analyzer (Testo 315-2 model 0632 0317), properly validated with an ISO calibration certificate issued by Instrumentos Testo, Cabrils, Spain.
2,1,3-benzothiadiazole (BTD, 25 mg, 0.184 mmol) was added to a dichloromethane (10 mL  Silica gel immobilization of complex 1. Complex 1 (5 mg) was dissolved in the minimum amount of chloroform (5 mL). This solution was then added to silica gel (150 mg), yielding 1 as an orange solid after evaporation of chloroform (see Figure S1). Figure S1. From left to right: color of ruthenium compound 1, silica-adsorbed 1, and the silica-adsorbed solid upon reaction with CO (to form 2).
Preparation of cellulose probe bearing 1. Complex 1 was supported on cellulose paper for chromatographic use (Whatman Grade no. 3MM Chromatography Paper). Probes were prepared in a simple fashion by dropping 0.2 mL of a solution of 1 in dichloromethane (5 mg/mL) on the cellulose paper strip and then were dried in air under ambient conditions.

Reactivity of 1 with CO
Scheme S2. Reactivity of 1 with CO to give dicarbonyl complex 2.
Air samples containing different concentrations of CO were bubbled through methanol : water, 12:10 (v/v) solutions of 1 (2 mL, 1.9 x 10 -3 mol dm -3 ) to yield a remarkable color modulation from red to bright yellow. These changes are consistent with coordination of CO and the concomitant displacement of the BTD molecule to give product 2.
The included solvent in the structure of 2 was found to be highly disordered, and the best approach to handling this electron density was found to be the SQUEEZE routine of PLATON. S3 This suggested a total of 36 electrons per unit cell, equivalent to 18 electrons per complex. During the synthesis and crystallization of the compound, both dichloromethane [CH 2 Cl 2 , 42 electrons] and ethanol [C 2 H 6 O, 26 electrons] were employed, but unfortunately before the use of SQUEEZE the electron density distribution did not clearly favor either. However, as dichloromethane was the most recently used, and as the crystals showed some signs of desolvation, dichloromethane was assumed as the solvent present. 50% Dichloromethane equates to 21 electrons, and so this was used as the solvent present. The atom list for the asymmetric unit is thus low by 0.5(CH 2 Cl 2 ), and that for the unit cell is low by one molecule of CH 2 Cl 2 . Table S1. Selected structural data for compound 1 at 173 K.

Selected bond lengths [Å]
Fluorescence calibration curve to 1 aqueous methanol solution response upon addition of CO

Rehm-Weller calculations.
In complex 1, the occurrence of an electron transfer process between vinylpyrene and BTD ligands is thermodynamically favorable as can be demonstrated from electrochemical and photophysical data. The free energy of the process (BTD + vinylpyrene  BTD ─ + vinylpyrene + ) in which the vinylpyrene acts as an electron donor and the BTD as an electron acceptor can be calculated by the  56 v), respectively, and  (458 nm) can be obtained from the emission fluorescence spectrum. Taking into account these data, G is −89.03 kJ mol -1 , indicating that the electron transfer process from the vinylpyrene to the BTD can occur. These calculations supported the fact that the weak emission observed for complex 1 is due to an electron transfer process between vinylpyrene and BTD ligands and also pointed to BTD displacement by CO as the mechanism behind the turn on fluorescence observed.  Figure S7 shows the calculation of the detection limit, based on visual examination of the reflectance measurement at a certain wavelength (474 nm) versus logarithm of CO concentration. The detection limit concurs with the inflection point of the two slopes observed for the concentration range studied. Selectivity studies with 1 adsorbed on silica.

Sensitivity studies with 1 adsorbed on silica upon addition of CO
Silica-adsorbed 1 showed a remarkably selective response to carbon monoxide in air. For instance, no reaction was observed in the presence of gases such as Ar, N 2 , O 2 , CO 2 , SO 2 and NO x (see Figure S8) at very high concentrations (up to 50,000 ppm). The same was true of H 2 S at (lethal) concentrations of 200 ppm. Similarly, no color changes were observed in the presence of volatile organic compounds (see Figure S9) such as chloroform, formaldehyde, acetone, ethanol, toluene or hexane (up to 30,000 ppm). Only acetonitrile (ACN) induced some evidence of a color change to yellow, although only at concentrations of approximately 5000 ppm (see Figure S10). The reaction of 1 with ACN was found to be reversible.  Figure S10. Diffuse reflectance UV-Vis spectra of silica-adsorbed 1 and the changes observed in the presence of air containing increasing quantities of ACN (4. 6,46,460,4600,46,000 ppm). Inset: detection limit calculation from diffuse reflectance measured at 500 nm.