WO2017137086A1

WO2017137086A1 - A device for the highly sensitive detection of a gaseous analyte

Info

Publication number: WO2017137086A1
Application number: PCT/EP2016/052941
Authority: WO
Inventors: Thomas Frederik VAN DIJKMAN; Elisabeth Bouwman; Wangyang FU; Gregory Schneider; Jan-Kees Boerman
Original assignee: Universiteit Leiden; Environmental Monitoring Systems (Ems) B.V.
Priority date: 2016-02-11
Filing date: 2016-02-11
Publication date: 2017-08-17

Abstract

The present invention relates to a device for detection of a gaseous analyte. Specifically, the present invention relates to a device for detection of a gaseous analyte, wherein the device is comprised of a first layer, being a substrate layer, a second layer, being an electrical insulating layer interposed between the first layer and a third layer, being a graphene layer and a fourth layer that is comprised of transition metal complexes, wherein the fourth layer is provided on a top surface of the graphene layer, wherein said transition metal complexes are capable to form a stable complex with said gaseous analyte. Furthermore the device comprises at least two electrodes being in electrical communication with the graphene layer and are capable of measuring a change in conductance of the graphene layer by measuring a change in electric potential between electrodes. Furthermore the present invention relates to methods for the detection of gaseous analyte and the use of the device of present invention to measure the gaseous analyte concentrations.

Description

A DEVICE FOR THE HIGHLY SENSITIVE DETECTION OF A GASEOUS ANALYTE

Description

The present invention relates to a device for detection of a gaseous analyte.

Furthermore the present invention relates to methods for the detection of gaseous analyte and the use of the device of present invention to measure the gaseous analyte concentrations.

The word transistor is a combination of two words: transfer and resistor. Usually a transistor is used to switch or amplify an electronic signal, comparable to a tap-valve that controls the supply and flow of water. A field-effect transistor (FET) is a transistor that uses an electric field to control the electrical conductance of a channel of one type of charge carrier in a semiconductor material. The carrier density, and thus the conductance of the channel is typically modulated by the electric field by gating a highly conductive silicon substrate located underneath an insulating Si0₂ dielectric layer to a range of voltages. The FET controls the flow of electrons from the source to the drain by affecting the size and shape of a conductive channel that is created and influenced by voltage applied across the gate and source terminals. The conductive channel is the stream through which charge carriers flow between the source and drain electrodes. The conductance of a FET is regulated by a voltage applied to a terminal (the gate), which is insulated from the charge carrying part of the device. The applied gate voltage imposes an electric field, which in turn attracts or repels charge carriers to or from the region between a source terminal and a drain terminal. The density of charge carriers in turn influences the conductance between the source and drain. A typical measurement consists of applying a constant bias voltage between the source and the drain of the channel, controlled by the gate and subsequently monitoring the resulting source -drain current.

A chemical conductive sensor consist of a chemical conductive material in which electrons flow in and out by means of a chemical reaction, where the analyte yields the measured signal.

A common use of the FET is typically an amplifier combined with the concept of a chemical sensor, called a FET chemical sensor, where the charge on the gate electrode is affected by a chemical process in proximity of the charge carrier layer. The FET chemical sensor may be used to detect atoms, molecules, ions in liquids and gaseous molecules. The FET chemical sensor is comprised of a conductive material and a sensitizer that interacts with the chemical that must be detected, resulting in the production of a signal that is detected.

Recently, sensitized carbon nanotubes (CNTs) were used as the conductive material in sensors for the detection of ethene in which a sensitivity of 500 ppb was reached. The sensitizer consisted of a hydridotrispyrazolylborate copper (I) complex which is known for its ability to interact strongly with ethene molecules. Trispyrazolylborates (commonly referred to as scorpionate ligands) form a highly versatile class of ligands. The great diversity in steric and electronic properties available in trispyrazolylborate ligands allows for the optimization of complexes for specific purposes such as catalysis and biomimetic structural and functional models.

CNTs network-based sensors, such as for gas detection, have long been considered poorly applicable for practical, large scale gas sensor applications due to their unpredictable characteristics and reliability issues. Researchers identified fundamental problems in carbon nanotube networks, such as unavoidably high contact resistance between carbon nanotubes and other conductors. The resistance of a device comprising a CNT network was found to be unstable over time. The resulting devices gradually degrade in such an unpredictable manner that device performances are fundamentally undermined.

In addition, drop-casting methods also result in irreproducible and variable measurement results and variable baseline resistances. CNT networks when produced by drop- casting of a solution or dispersion of CNTs in a solvent are irreproducible due to a lack of control over the deposition of the CNTs and sensitizer. The major disadvantage of using CNT -based devices is the irreproducibility in detection. The poor control during drop-casting and device-to- device specific variations result in irreproducible and variable measurements and baseline resistances. This is caused by the differences in the amounts of CNT/scorpionate/CNT junctions per device and difficulties involved in the synthetic aspects of CNTs, such as the presence of undesired metal catalyst residues and CNTs of different chirality and diameters, and hence of different electrical properties.

Due to the abovementioned reasons, the CNT-based sensors face the major drawback that their fabrication is highly irreproducible, even in the devices composed of only single nanotubes or arrays of carefully oriented CNTs. Additionally, the performances of the CNT network-based sensors were found to be unstable and their sensitivity to be varied in time because of the intrinsically unstable CNT/scorpionate/CNT junctions, which has been shown to be mechanically and electronically unstable. As a result, the CNT network based FET chemical sensors are unsuitable for straight electronic design, practical use, and commercial applications.

The global population is steadily growing and the need for rapid, secure transportation of fruits and vegetables grows with it. A major factor in the loss of food crops during transportation is the production and subsequent exposure of sensitive products to the gaseous plant- hormone ethene which results in spoilage. Ethene is a gaseous analyte that is particularly important in agri- and horticulture where it serves as a hormone for plants. Ethene gas is a highly diffusive, relatively unreactive gas that induces ageing responses in plants at concentrations as low as parts per billion (ppb) by volume. In order to prevent the untimely exposure of valuable crops to high levels of ethene gas in the production and logistic chain, during cultivation, long- and short term storage and transport, monitoring is necessary to be able to vent or cool the product and thus guarantee the quality of the products to the consumer. The detection of ethene at biologically relevant concentrations, which can be as low as parts per billion (ppb) by volume, requires an extraordinary sensitivity of the sensor. Currently available ethene sensors are based on expensive and large equipment (e.g. gas chromatography, photo acoustic laser systems or electrochemical sensors) and are only used in laboratories, (fruit)storages, and greenhouses. They are not portable or compatible with, for example, the containers used for transportation. At the moment there is no cost effective and suitable commercially available method to monitor ethene concentrations at ppb levels per volume in the production and logistic chain of agricultural fresh products.

Considering the above, there is a need in the art for a device that is capable of a rapid and sensitive detection of a gaseous analyte, wherein the detection characteristics of the device are highly interpretable, predictable, reproducible and reliable.

It is an objective of the present invention, amongst other objects, to address the above need in the art. The object of present invention, amongst other objects, is met by the present invention as outlined in the appended claims.

Specifically, the above object, amongst other objects, is met, according to a first aspect, by the present invention by a device for detection of a gaseous analyte, wherein the device is comprised of

a. a first layer, being a substrate layer (10);

b. a second layer (20), being an electrical insulating layer interposed between said first layer (10) and a third layer (30), being a graphene layer;

c. a fourth layer (40) that is comprised of transition metal complexes, wherein said fourth layer is provided on a top surface of said graphene layer (30), wherein said transition metal complexes are capable to form a stable complex with said gaseous analyte;

d. at least two electrodes (50) being in electrical communication with said graphene layer (30), wherein said electrodes are capable of measuring a change in conductance of said graphene layer (30) by measuring a change in electric potential between electrodes (50).

The device according to present invention is comprised of a field-effect transistor (FET) build up out of a layer of graphene (GFET) on a substrate layer with an insulating layer between the graphene layer and the substrate layer. The graphene layer is provided with a layer comprised of transition metal complexes, preferably a thin layer of transition metal complexes. Attached to the graphene are at least two electrodes, preferably gold electrodes, preferably spaced from 1 μιη to 1 cm apart, most preferably 1 mm. This thin layer (ideally a monolayer) of transition metal complex is formed by dipping the graphene in a solution of the transition metal complex in an organic solvent (e.g. 1-10 mM in dichloromethane) for preferably 1-100 minutes, more preferably 10 minutes, followed by rinsing with clean solvent (pure dichlorome thane) to remove excess complex.

Graphene - at least ideal graphene - is highly chemically inert. The functionalization and chemical alterations of the graphene surface - both covalently and non- covalently - are crucial steps that define the sensitivity of graphene-based sensors. The impact on the physical, electrical and chemical properties of graphene upon gas interaction with a layer of gas -sensitive material deposited on graphene unlocks the selective detection of particular gas and analytes. The physical and electrical properties of graphene coupled with the possibility for significant sensitization offers promise for the detection of gaseous analytes with the required sensitivity and reproducibility. The two-dimensional nature of graphene and the concomitant extremely high surface to volume ratio and its unusually high sensitivity to external electrical fields mean that graphene-based FETs (GFETs) are excellently suited for highly sensitive chemical sensing.

Present invention proposes a well-defined interface comprising graphene, an insulating material and transition metal complexes that upon gas exposure yields a chemical alteration of the transition metal complexes resulting in a rearrangement of the electrical field in and around the graphene layer, which can be monitored as a change in the conductance of the graphene between the source and drain. The electric field rearrangement causes the number of charge carriers in the graphene to vary, which can be detected as a change in the conductance of the graphene; an increase in the number of charge carriers gives rise to more conductance and vice versa, related to the graphene conductance without an applied electrical field or gate voltage. The electric field can be achieved using the conductor layer as the gate on the backside of the sensor (backgate) and/or by changing the chemistry of the transition metal complex. When the conductor layer is held at a fixed voltage the change in conductance between the drain and source must thus be due to chemical and/or physical changes in the transition metal complex. The transition metal complexes used in present invention are capable of reversibly reacting with a gaseous analyte such as ethene, to form a new complex. When the gas flow over the device of present invention is switched back to ambient air (-79% N₂, -21% 0₂) the newly formed complex reverts back to its "gas-free" state and the conductance returns to the baseline.

In contrast to a CNT -based device, the resistance of a graphene-based device according to present invention is based on a continuous graphene film and is well-defined and reproducible. The device of present invention is produced using a dip-coating method to achieve a homogenous self-assembled transition metal complex layer, preferably a thin layer of less than 10 nm thickness, on the graphene. The excess of sensitizing molecules is rinsed off with the same solvent as the one used to form the sensitizing thin layer. The chemical changes in the transition metal complexes due to gas coordination can now be reliably and reproducibly probed due to the effect of such chemical changes on the electric field effect. The present invention exhibits a well- defined and reproducible sensing response with minimal device-to-device variations and offers new opportunities for developing the next generations of label-free chemical sensors using graphene and/or other two-dimensional materials, such as molybdenum disulfide (MoS₂), molybdenum diselenide (MoSe₂), molybdenum ditelluride (MoTe₂), tungsten disulfide (WS₂), tungsten diselenide (WSe₂) and other semiconducting inorganic materials with the cadmium iodide (Cdl₂) crystal structure. Like graphene, these two-dimensional materials are flat and can therefore be integrated into the fabrication of scalable sensors.

According to a preferred embodiment, the present invention relates to the device of present invention, wherein said transition metal complexes are according to formula

wherein,

M is a transition metal, selected from copper, silver, gold, titanium, chromium, manganese, iron, nickel, preferably gold, more preferably silver, most preferably copper; and,

R¹ is selected from the group NMe₂, OMe, N0₂, H, F, CI, CF₃ and R¹ substituents being either identical or different; and,

R² is selected from the group H, alkyl, aryl, acyl or aromatic ring; and, the number of aromatic rings on B is 2, 3 or 4, preferably 3 or 4, most preferably 3. The aromatic ring can be a

heteroaromatic ring, preferably a pyrazole ring.

The transition metal complexes used in present invention have a coordination cavity at the metal ion where the gaseous analyte, such as ethene, can bind. The electronic properties of the metal compounds are varied through substitution on a benzene ring that is placed away from the gas binding pocket. The substituents (labelled "R¹" in formula I) have varying degrees of electron attracting or donating behaviour, indicated using Hammett σ_ρ -parameters); N0₂ (+0.78), CF₃ (+0.54), CI (+0.23), F (+0.06), H (0), NMe₂ (-0.83), OMe (-0.27), wherein a positive value implies electron-withdrawing and a negative value implies electron-donating behaviour. Depending on the substituents in the complex the specific group affects (increase or decrease) "the sensing" of the gaseous analyte of the device of present invention by changing the electron density on the metal center.

The π-backbonding ability of the metal centers of the transition metal complexes is modified through substitution on a benzene ring that is placed away from the gas binding pocket without significantly affecting the steric properties of the binding pocket surrounding the metal centre. In π-backbonding electrons move from an atomic orbital on the metal center to a π* antibonding orbital on a π-acceptor ligand. Electrons from the metal (present in the transition metal complexes) are used to bond to a ligand (gaseous analyte), in the process relieving the metal of some electron density and changing the dipole moment of the transition metal complex.

According to another preferred embodiment of the present invention, the fourth layer (40) has a layer thickness of between 0.1 nm to 100 nm, preferably of between 0.5 nm to 10 nm, more preferably 0.5 nm to 5.0 nm. Preferably, this layer is a homogenous layer.

According to another preferred embodiment of the present invention, said second layer (20) is comprised of, or made of, an insulating material selected from the group of silicon dioxide, silicon oxynitride, silicon nitride, preferably silicon dioxide and has a layer thickness of at most 500 nm, preferably at most 300 nm, most preferably at most 100 nm.

According to the present invention, the substrate layer (10) is comprised of, or made of, a conductor, preferably p-doped or n-doped silicon

According to a preferred embodiment of the present invention the substrate layer (10) is provided on a conductive layer comprised of, or made of, a metal, preferably comprised of a metal selected from the group of gold, platinum, palladium, silver, copper, nickel, preferably gold and the substrate layer (10) is in electrical communication with, or electrical contact with an electrode (60). The conductive layer itself may also serve as the electrode.

According to another preferred embodiment of the present invention the electrodes (50) and the electrode (60) are comprised of, or made of, a conductive material, preferably a metal selected from the group of gold, platinum, palladium, silver, copper, nickel, preferably gold.

According to present invention the gaseous analyte is selected from the group of ethene, carbon monoxide, carbon dioxide, nitrogen monoxide, nitrogen dioxide, acetone, methyl ethyl ketone, benzene, styrene, xylene, tetrahydrofuran, acetonitrile, ammonia, water, ethanal, propene, 1-methylcyclopropene, ethyl ethanoate, methyl ethanoate, ethanoic acid, methanoic acid, propanoic acid, ethanol, methanol, 1-propanol, 2-propanol, 1-butanol, 1-hexanol, hexanal, phenol, H₂S, preferably ethene, carbon monoxide or ethanol, most preferably ethene. Gases such as ammonia, water vapour and carbon monoxide can be detected using bare graphene, typically with sensitivities at the ppm level or lower. The mechanism of detection of these gases lies in the ability of the detected molecules to adsorb on the graphene surface thus doping the material with holes or electrons, which results in a change in the conductance of graphene.

Impurities on the graphene, such as polymeric residual materials from the fabrication and deposition steps, are likely to be responsible for the reactivity with many of the analyte gases. Thoroughly cleaned samples of graphene show little to no response to analytes that before cleaning induced strong electrical responses. This finding indicates not only that the, potentially undesired, chemical reactivity towards contaminant gases such as H₂0, CO and NH₃ can be eliminated by thorough cleaning of the graphene, but also that apparently the graphene can be sensitized greatly to particular analytes by the introduction of 'tailored impurities' . Using this principle, custom-made sensitizers can be designed to modify the graphene layer for the detection of otherwise difficult to detect, relatively unreactive molecules. Aside from gas sensing the use of tailored sensitizing layers for specific target compounds has also been explored in liquid gated GFETs sensitized using biological receptors for charged biomolecules.

According to present invention, a top surface of said fourth layer (40) is coated with a hydrophobic layer and is comprised of a composition that reacts with carbon oxides and/or nitrogen oxides, such as CO, NO_x. This layer removes compounds other than the analyte such that they do not bias the sensing results. This hydrophobic layer isolates the graphene layer from the excess amount of ambient water molecules and other unwanted molecules and prevents ingress and excessive altering of the graphene layer. The hydrophobic layer can be comprised of polymer such as polypropene, polyethene, poly(methyl methacrylate) or polystyrene. The presence of water near the sensor surface may negatively affect the detection of gaseous analyte by the device of present invention. The device is more stable and produces a more reproducible response in the absence of water. To exclude water from the sensor surface a thin layer of a hydrophobic gas-permeable polymer is applied on the sensor surface. The polymers used should be permeable to the desired analyte gas but not to water. Furthermore the hydrophobic layer is comprised of a catalyst that can oxidise contaminants - that negatively influence the sensor performance - to harmless compounds which are then released by the sensor surface and emitted to the ambient air. Typical compounds that are oxidized are CO, NO_x , and volatile organic compounds like ozone and dihydrogen.

Nitrogen oxide, nitrogen dioxide and carbon monoxide are the first major cross sensitivity contributors in ambient air.

According to a preferred embodiment of present invention the hydrophobic layer is comprised of tin oxide with palladium and/or platinum. This hydrophobic layer can exclude further cross-sensitivity of the sensor and prevents gases such as carbon monoxide from adsorbing to the transition metal complex. This exclusion of carbon monoxide interference is for instance necessary to obtain a highly sensitive detection of ethene in ambient conditions. Apart from carbon monoxide also nitrogen monoxide and nitrogen dioxide are effectively removed which prevents environmental interferences. An important property of this hydrophobic layer is that it is functional at room temperature without the need for additional energy, which is not allowed in respect to the small GFET chemical gas sensor design.

According to yet another preferred embodiment of present invention a layer comprised of, or made of silane is interposed between the second layer (20) and the graphene layer (30). This additional silane layer can improve the detection efficiency of the gaseous analyte by the device of present invention by improving the electrical performances of the GFETs, and hence the signal to noise ratio of the signal produced upon exposure of the device to a gaseous analyte

The present invention, according to a second aspect, relates to a method for the detection of gaseous analyte, wherein said method comprises the steps:

a) exposing the device of present invention to a gas analyte;

b) detecting a change in conductance of a graphene layer (30) by measuring a change in electric potential between electrodes (50).

Detecting a change in conductance is performed by means of the field effect induced on the gate channel, for instance by resistance measurement, voltage measurement, current measurement or measurement of a resistance with a frequency optimized lock-in amplifier.

According to a preferred embodiment of present method, the method is capable of measuring a gaseous analyte below 10 ppm, preferably below 1.0 ppm, more preferably below 0.2 ppm, most preferable below 0.1 ppm. The device according to present invention is capable of detecting gas in very low concentrations ranging from 1 ppm to 1 ppb.

According to another preferred embodiment of present method the gaseous analyte is ethene, carbon monoxide, carbon dioxide, nitrogen monoxide, nitrogen dioxide, acetone, methyl ethyl ketone, benzene, styrene, xylene, tetrahydrofuran, acetonitrile, ammonia, water, ethanal, propene, 1-methylcyclopropene, ethyl ethanoate, methyl ethanoate, ethanoic acid, methanoic acid, propanoic acid, ethanol, methanol, 1-propanol, 2-propanol, 1-butanol, 1-hexanol, hexanal, phenol, H₂S, preferably ethene, carbon monoxide or ethanol, most preferably ethene.

The present invention, according to a third aspect, relates to the use of the device of present invention to measure ethene concentrations in agriculture and food related products, preferably fruit, vegetables, flowers, plants, bulbs, cereals, and other agriculture fresh products. Ethene is a gaseous analyte that is particularly important in agri-and horticulture where it serves as a hormone for plants to ripe in a desired state or alter in undesired circumstances of unexpected exposure. Especially those products that produce high ethene concentrations, and are highly sensitive to ethene and are densely packed for transport or storage have a high risk of early deterioration of the product. Ethene gas is a highly diffusive, relatively unreactive gas that induces ageing responses in plants in concentrations as low as parts per billion by volume. Currently available ethene sensors are based on expensive and large gas chromatography, electrochemical or photo acoustic laser equipment, and are used in storages, greenhouses and logistics. At the moment there is no cost effective and suitable, commercially available method to monitor ethene concentrations during the shipping of agricultural and/or fresh products in boxes, vans, containers or reefer containers. The present invention can be used as a sensorial platform (e.g. as a micro sensor) that travels with products through the entire chain from production to in-store display or potentially even until consumption. The sensor can be equipped with a unique ID or tag combined with place, position and timestamp devices or electronics to identify the fresh goods, to track and trace the sensor and to read-out the monitored and temporary stored information. The data can be uploaded to a central database to process calculations with results and disclosure of shelf live and affected quality issues. Information can be provided to customers in the logistic chain like importers, exporters, warehouses, logistics, supermarkets, consumers and end users. The sensor, combined with new insights into the biological effects of ethene could inform customers and suppliers of the expected lifetime of the product in significantly better detail than current visual inspection. Also, detailed knowledge of the life of a product can allow for improved liability coverage for producers and transporters.

According to another preferred embodiment of the present invention the device is used to measure the carbon monoxide concentrations or ethanol concentrations. Carbon monoxide concentrations can be measured using the device of present invention in for instance closed environments, exhaust fumes, and combustion processes or used as breath monitor or as health and safety gas monitor to prevent carbon monoxide poisoning in closed environments.

The present invention will be further detailed in the following examples and figures wherein:

Figure 1: shows the layout of the device according to present invention, (10) is a first substrate layer, (20) is the second layer, preferably Si0_2, (30) is a graphene layer, (40) is the layer comprised of copper complexes, (50) are two electrodes which are the source and the drain and (60) in a third electrode that can be used as a gate electrode.

Figure 2a: shows a layout of the device according to present invention;

Figure 2b: shows π-stacking interactions between the transition metal complexes and the graphene lattice serving to anchor the complexes onto the graphene surface.

Figure 3: shows that when the gas flow over the device of present invention is switched back to ambient air the newly formed complex reverts back to its "gas-free" state and the conductance returns to the baseline. Data processing for the signals obtained from the gas exposure experiments requires a number of steps that are visually outlined in figure 3. In panel a, the signal (black) is shown after normalization using the device transconductance, the signal is convoluted with noise and drift. In panel b, the signal is shown once the baseline drift is subtracted. The signal is still convoluted with noise, in signal I the drift in the baseline is too great to extract useful data, and can not be used. Signals II and III have the expected lineshape and show clear steps when exposure with ethene gas is initiated or halted. When ethene gas was applied a pressure spike is clearly visible. The step-height after the initial sharp peak is taken as the signal intensity. Upon cessation of exposure the signal returns to the baseline. In panel c, the averaged response intensities at different concentrations of ethene gas are shown versus the ethene concentrations. The red line is a fit using a Langmuir Isotherm.

Figure 4: Sensor responses of the complexes [Cu(Tp^{CF3,4 RPh})(MeCN)] (R = NMe₂,

OMe, H, F and CI) when the device according to present invention was exposed to dilute ethene gas. The complexes with electron-donating ligands (R = NMe₂, OMe and H) produce signals with a negative AY_G whereas the ligands with electron-withdrawing substituents (R = F and CI) produce signals with a positive AY_G.

Figure 5: Response intensities at 1.0 ppm (C₂H₄) and 10 ppm (ethanol) versus the

Hammett σρ parameters of the substituents on the hydridotrispyrazolylborate ligands of the copper(I) complexes. The ethene response of the device of present invention functionalized with [Cu(Tp C' F3.4- C^F3Ph)(MeCN)] that is marked with * is the actual signal with its sign reversed.

Figure 6: Shows the Equilibrium dissociation constants (K_D) of ethene and ethanol on the GFET device of present invention. Equilibrium dissociation constants were obtained for a number of the complexes and provide a clear distinction between the ethene and ethanol complexes.

Figure 7: Shows the pre-f actor (Aq/C₀)[on]_max in equation 1 that determines the amplitude of the signal that is generated by the GFET upon exposure to the analyte. Aq is a measure of the change in the local charge perceived by the graphene that results from changes in the magnitude and direction of the dipoles of the complexes, Aq is scaled by the coupling between the graphene surface and the complex molecule. The pre-factors Aq/C₀ vs. the Hammett parameters σ_ρ of the substituents in the complexes [Cu(Tp^CF3'^4~ ^rPh)(MeCN)] that were used to sensitize the GFET devices of present invention.

Figure 8: Dissociation rate constants for the ethanol (black) and ethene (red) complexes vs. the Hammett σρ parameters of the substituents on the hydridotrispyrazolylborate ligands of the transition metal complexes.

Example 1 - Construction of the copper(I) trispyrazolylborate sensitized GFET devices

As a reference the complex [Cu(Tp^(CF3)2)(MeCN)] was included as it has been used previously as the sensitizer in a ethene sensor based on SWCNTs. This compound has a binding "pocket" at the copper(I) ion that is extremely similar to the binding pocket of the copper(I) complexes of present invention.

The GFET devices were prepared by wet-transfer of CVD graphene onto highly p⁺-doped silicon substrates with 285 nm thick silicon dioxide insulator layers. In order to limit the effects of charge puddles on the Si0₂ the substrates were first modified with octadecyltrimethoxysilane ((MeO)₃SiC₁₈H₃7) , resulting in self-assembled monolayers (SAMs) of hydrophobic octadecylsilyl (OTS) chains.

To study the effect of these hydrophobic SAMs a control experiment was performed using a SAM of the much shorter trimethylsilyl (TMS) groups. As expected the shorter spacing between the graphene layer and the Si0₂ substrate resulted in a larger residual effect of charge puddles as was evident from the larger p⁺-doping-induced shift of the charge neutral point of the graphene. Gold source and drain electrodes with a spacing of 1 mm were patterned on top of the graphene using vapour deposition. The acetonitrile adducts of the copper(I) trispyrazolylborate complexes were then applied on top of the graphene layer by immersion of the devices in 10 mM solutions of the complexes in dichloromethane for 10 minutes, after which any non-adsorbed material was rinsed off using a stream of pure solvent. The thickness of the layers obtained were studied using ellipsometry and were found to range from 1.98(10) nm in the case of [Cu(Tp^CF3'^4F~ ^Ph)(MeCN)] to 3.74(10) nm for [Cu(Tp^CF3'^{4CF3 Ph})(MeCN)]. The thin layer of [Cu(Tp^CF3'^4F" ^ph)(MeCN)] may not lead to reasonable comparison, but its solubility in dichloromethane is too low to reach a concentration of 10 mM. If [Cu(Tp^CF3'^{4F Ph})(MeCN)] is excluded from the series the thinnest layer observed was found for [Cu(Tp^{CF3 Ph})(MeCN)] at 2.46(9) nm. The layer thickness does not seem to correlate to the polarity of the complexes; rather it appears that the steric bulk of the complexes is the principal influence, with the thickest layers being obtained for the bulkiest complexes. The layers were found to be homogeneous and smooth with only small variations in thickness.

Based on the crystal structures of copper(I) complexes of the various trispyrazolylborate ligands our initial estimation was that the molecular volume and conformation of the copper(I) complexes on the graphene layer would be approximately constant regardless of the ancillary ligand (ethene, carbon monoxide or acetonitrile). This estimation was corroborated by the X-ray crystal structure of [Cu(Tp^CF3'^{4MeO Ph})(MeCN)], which is indeed nearly identical in structure to its ethene and carbonyl congeners.

The complexes are assumed to adsorb side on to the graphene layer as this allows for optimal π-stacking interactions between the aromatic rings of the ligands and the graphene surface. In such a configuration the height of the complexes vs. the graphene basal plane is approximately 0.9 nm. This indicates that the complexes assemble on the graphene surface as structures of approximately three monolayers. For the reference compound [Cu(Tp^(CF3)2)(MeCN)] a layer thickness of 1.68(9) A was obtained. Presumably the highly fluorinated complex molecules have smaller intermolecular interaction with the graphene substrate leading to thinner layers. Surprisingly the duration of immersion time in the solution has little effect on the layer thickness of [Cu(Tp^(CF3)2)(MeCN)] : incubation times of 1 or 100 minutes resulted in layer thicknesses of 1.44 and 1.37 nm respectively.

In order to better understand the packing of the copper(I) complexes on the graphene surface the samples were studied using atomic force microscopy (AFM). The dip-coated samples showed no clear signs of crystallinity or any other organization on the graphene surface; instead smooth, featureless surfaces were observed.

To investigate whether the multi-layer structure is more ordered directly on the graphene surface, HOPG (highly ordered pyrolytic graphite) samples were prepared by drop casting dilute (approximately 1 μΜ) solutions of the compounds [Cu(Tp )(MeCN)], [Cu(Tp^CF3'^ph)(MeCN)] and [Cu(Tp^CF3'^{4 FPh})(MeCN)]. The concentrations of the drop casting solutions were chosen such that the resulting surface coverage on HOPG would approximate one third of a monolayer, avoiding the formation of multi-layered structures. HOPG was chosen over CVD graphene as freshly cleaved HOPG offers convenient access to large, flat, contiguous surfaces of high quality graphene. As graphene is wetting transparent it is important to consider the possible effects of the underlying substrate. However, it is believed that the presence of the SAM of OTS groups on the surface results in similar wetting properties to multilayer graphene, allowing for reasonable comparison. The compound [Cu(Tp^(CF3)2)(MeCN)] was found to assemble on HOPG into side -by-side assemblies of rod-shaped structures. The rods have lengths of tens to hundreds of nm with diameters of 5 nm. Similar heads-to-tails orientations were observed in the single crystal X-ray structures of many of the complexes [Cu(Tp^{CF3 R})(L)] (L = C₂H₄ or CO).

Example 2- Detection of ethene gas

The GFET devices were mounted on gas-tight epoxy chip carriers and placed in a

Teflon flow cell that was tightly sealed using a PDMS ring. Using two mass flow controllers (MFCs) ethene gas diluted in air was further diluted to reach concentrations ranging from 1.0 ppm to 0.1 ppm, the range within which most climacteric crops respond to ethene exposure. In order to be able to compare the ethene-sensing performance of the GFET devices exposures to ethanol was also included. Ethanol is an agriculturally relevant gas that signals rotting. As ethanol was expected to bind considerably more weakly to the copper(I) center in the hydridotrispyrazolylborate complexes than ethene exposures were performed at 10, 5, 2 and 1 ppm instead of 1, 0.5, 0.2 and 0.1 ppm as was the case for ethene. For an explanation of the way in which the data were processed see the appendix.

When the devices were exposed to ethene all of the sensors except the sensor bearing the reference complex [Cu(Tp^(CF3)2)(MeCN)] showed saturation at 1.0 ppm. In the cases of R = OMe and F saturation was even observed at 0.1 ppm. For ethanol the devices showed signs of saturation at 10 ppm. As the required concentrations were quite low not all devices produced equally high quality data, drift in the baseline and poor signal to noise ratios occasionally obscured the signal.

Nonetheless clear trends emerge when the responses of the sensors are collected. The complexes with electron-donating ligands (R = NMe₂, OMe and H) produce signals with a negative AV_G whereas the ligands with electron-withdrawing substituents (R = F and CI) produce signals with a positive AV_G.

Contrary to expectations when [Cu(Tp C"F"3^ 4 CF"3P^rh^u)(MeCN)] was used the sensor produced a large negative signal (AV_G = -202(9) mV) when exposed to 1.0 ppm ethene gas. The ethanol exposures of [Cu(Tp CF3 ' 4 ^" CF3Ph )(MeCN)] produced the expected positive signals (albeit of poor quality). The signal at 10 ppm ethanol exposure was AV_G = 163(11) mV (the trace of the signal at 10 ppm ethanol exposures is shown in the appendix). The negative signal produced upon ethene exposure is surprising as the ellipsometry and AFM data showed no unusual surfaces for the

[Cu(Tp CF3 ' 4 ^" CF3Ph )(MeCN)] functionalized devices compared to the other devices. Nor were the ethene and carbonyl complexes bearing the [Tp CF3 ' 4 ^" CF3Ph ] ligand found to have unpredictable properties on NMR and infrared spectroscopy in our previous study. Particularly intriguing is the fact that if the sign of the negative signal is reversed the signal lines up very well with the values from the complexes with R = F and CI (see figure 5). It appears plausible therefore, that

[Cu(Tp C—F3 4— CF3Ph ¹)(C₂H₄)] assumes a reversed orientation towards the graphene surface compared with the other ethene complexes resulting in a signal with a reversed (i.e. negative) sign. The reference complex [Cu(Tp^(CF3)2)(MeCN)] was the only complex to result in a linear response in the entire range.

Kinetics of ethene and ethanol sensing

The GFET surface consists of a discrete number of binding sites for analyte molecules (the copper(I) centers of the complexes) in a quasi two-dimensional arrangement, as such it can be described using a Langmuir adsorption isotherm. The results of the gas sensing experiments were therefore interpreted quantitatively by fitting Langmuir adsorption isotherms to the AY_G/p plots (where p is the partial pressure of the analyte gas in ppm). The Langmuir isotherms used to describe the signals generated by the GFETs can be described using equation 1 :

AG_sd Aq p(an. )

—— = &V_G =— [on]_max x

9m Co p(an. ) + K_D

Aq is the difference between the charges induced in the graphene by the dipole of the complexes in their "on"- and "off-states. C₀ is the coupling between the complexes and the graphene, [on]_max is the maximum concentration of complexes in the "on"-state, p(an.) is the partial pressure of the analyte gas (typically in ppm) and K_D is the equilibrium dissociation constant. This equation describes the generation of the signal AY_G as a function of the surface coverage of the graphene with the complexes that constitute the "on"-state of the sensor which are assumed to be the copper(I) ethene or ethanol complexes. The pre -factor Aq/C₀ represents the change in the effective charge on the graphene that is caused by the conversion of the complexes in the "off-state to the complexes in the "on"-state. The first term, when multiplied by the maximum surface coverage with complexes in the "on"-state, thus gives the maximum amplitude of the signal and is effectively a scalar. The second term is the Langmuir adsorption isotherm that dictates the degree to which the graphene surface is covered in complexes in the "on"-state. Equation 1 offers the possibility to extract the equilibrium dissociation constant K_D, and the term (Aq/C₀)[on]_max which can give insight into the importance of the contributions of the imposed charges on the graphene.

The reaction on the surface of the graphene is assumed to consist of the conversion of a complex without associated ethene or ethanol, the "off-state and the ethene or ethanol complex, the "on"-state as described in equation 2:

[CUCTP^1"-^*)] ^: = [C_U(Tp^CF¾R)(C₂H₄)]

"off" -state "oi '-state

Equation 2 shows the equilibrium reaction between a copper(I) complex and a gaseous analyte (in this case ethene). Aside from the equilibrium constant the dissociation rate constant of the desorption of the analyte gases can be extracted from the traces of the gas sensing experiments. In order to be able to model the dissociation of the analyte molecules typically both the association (]¾) and dissociation (k_ reaction rate constants must be considered as re- association may occur during this phase. However, as the dissociation of the analyte molecules occurs in a flow of analyte-free air re-association is effectively impossible, this means the kinetics of the dissociation can be simplified to equation 3. Only the traces of the dissociation phase (immediately following the switch from gas exposure to pure air) could be fitted as the association events were obscured by large peaks that resulted from the MFCs. The devices functionalized with the complexes [Cu(Tp^{CF3 4 RPh})(MeCN)] (R = NMe₂, H (ethene only), F (ethanol only) and CI) showed signals with sufficiently good signal to noise ratios.

V_G = V_one-^k→^t + V_off

Equation 3. V_on is the signal during analyte exposure, V_off is the signal after dissociation of the analyte, k__\ is the dissociation rate constant and t is time.

The equilibrium dissociation constants (K_D, see figure 6) were obtained for a number of the complexes and provide a clear distinction between the ethene and ethanol complexes. The equilibrium dissociation constants of the ethene complexes are 0.11(3) ppm (R¹ = NMe₂), 0.23(3) ppm (R¹ = H) and 0.21(7) ppm (R¹ = CI) while the values obtained for the ethanol complexes were considerably higher at 3.15(0.95) ppm (R¹ = NMe₂), 3.13(0.95) ppm (R¹ = F) and 3.91(0.90) ppm (R¹ = CI). Physically K_D can be interpreted as the concentration of the analyte gas at which half of the copper complexes on the graphene surface are coordinated by analyte molecules. K_D is therefore a useful measure of the ability of the GFET's surface to adsorb analyte molecules; a lower K_D means a GFET generates a signal at lower analyte concentrations and becomes saturated at lower concentrations. Using the equilibrium dissociation constants for ethene and ethanol it is readily apparent that, compared to ethene, the GFETs can be exposed to higher concentrations of ethanol before becoming saturated. There is no clear correlation between K_D and the Hammett σ_ρ parameters of the hydridotrispyrazolylborate ligands in the complexes. It is likely that such a correlation does exist as there is a direct relation between the electron density on a copper ion and its ability to engage in π-backbonding interactions with a ligand such as ethene. The presence of π-backbonding interactions between the copper ions in the complexes and the ethene ligands is also the most plausible explanation for the much lower equilibrium dissociation constants observed for ethene compared to ethanol. Ethanol can coordinate to the copper(I) ion only through σ-donation as there is no n*-orbital available for π-backbonding interactions with the copper ion and is thus bound weakly than ethene by comparison.

The relative lack of π-backbonding interactions can also be invoked to explain the linear response observed in the GFET that was sensitized with [Cu(Tp^(CF3)2)(MeCN)] which contains the most electron-withdrawing hydridotrispyrazolylborate ligand of all the complexes used in this study. The resulting electron-poor copper(I) center binds ethene less strongly than the copper(I) centers in the other complexes as is evident from the linearity of the response. The equilibrium dissociation constant could not be determined but must be considerably higher than those observed in the other complexes as the response curve shows no signs of saturation effects at 1.0 ppm.

The pre-factor (Aq/C₀)[on]_max in equation 1 can be extracted using the same fit as used for the equilibrium dissociation constants. This pre-factor determines the amplitude of the signal that is generated by the GFET upon exposure to the analyte. The pre-factors for the fitted curves are shown in figure 7. Aq is a measure of the change in the local charge perceived by the graphene that results from changes in the magnitude and direction of the dipoles of the complexes, Aq is scaled by the coupling between the graphene surface and the complex molecule.

The induced local charge due to the complexes is assumed to scale linearly with the concentration of the complexes. The observed pre-factors provide a number of interesting insights the most obvious of which is that, while the K_D values showed no correlation with σ_ρ the pre-factors do. The polarity of the complexes (and their orientation on the graphene) is thus apparently the most important factor in the amplitude of the signal.

The second observation is that the pre-factors for the ethene and ethanol sensing experiments are very close, apparently the induced change in the effective charge of the complexes as felt by the graphene is not very different. The second observation can be nuanced by noting that at negative σ_ρ values (and thus more polar complexes) the lines for the ethene and ethanol complexes drift apart which indicates that either the dipole moments of the ethene and ethanol complexes do not scale the same way vs. σ_ρ or that the orientations of the complexes on the surface are not the same for all the complexes. The latter is more plausible as stacking interactions of the complex molecules on the graphene have a larger contribution from complex -complex interactions when the complex has a larger dipole moment and the complexes with the most electron donating substituents are the most polar complexes. Changes in the exact orientation of the complexes on the graphene would not only influence the projection of the dipoles on the graphene, they might also change the coupling between the complex and the graphene which is expected to be strongly influenced by the π-stacking interactions between the aromatic moieties of the complexes and the graphene. The exact contributions of Aq, C₀ and [on]_m!i cannot be extracted from the fitted data but as the π-stacking interactions between the complexes and the graphene are likely highly similar for all the phenyl-substituted complexes C₀ is unlikely to vary strongly between the complexes.

Likewise the steric bulk of the complexes is quite similar which means [on]^ will be similar for all the complexes.

As the equilibrium dissociation constants are approximately equal for different complexes when they are exposed to the same gas it could be concluded that the scaling relation observed between σ_ρ and Aq is the result solely of the difference in the electronic effects of the complexes on the graphene. It is important to note that the complete absence of a correlation between the electron density on the copper(I) center and the bond dissociation energy in the ethene and ethanol complexes is physically impossible. A trend in this regard should be most visible in the ethene complexes as the π-backbonding component of the Cu-C₂H₄ bond should help to amplify the effects of the electronic modification of the copper(I) center. Indeed in the ethene complexes, although the effect is statistically insignificant, it appears to be case that the equilibrium dissociation constant for the complex with the most electron rich copper(I) center (R = NMe₂) is smaller than the others.

The dissociation rate constants (k__! in equations 2 and 3) were obtained from the traces of the dissociation phases of the gas exposure experiments. Both for the ethene and ethanol complexes the dissociation rate constants show a minimum at σ_ρ = -0.27 (R = OMe). For the ethanol complexes the experimental errors obscure whether or not both curves follow the same trend completely (including the NMe₂ substituted complex). The rate of dissociation in the ethanol complexes is higher by a factor 1.5 - 2 in all cases as could be expected from the stronger bonding interactions observed in the equilibrium dissociation constants.

As the MFCs could not be used to reliably produce ethene concentrations lower than 0.1 ppm the lower limit of detection remains elusive. It is possible however, to estimate a limit of detection using the best peak-to-peak noise that was observed (in the OTS modified substrates) of 1 mV_G and to extrapolate the performance of the NMe₂ substituted complex using the Langmuir isotherm fits for the ethene and ethanol detection experiments. Assuming a signal-to-noise ratio of at least 2 is required to identify a signal the extrapolated lower limits of detection are 2.2 ppb for ethene and 35 ppb for ethanol. By using graphene with higher charge carrier mobility these limits can be lowered further. Example 4 - Silanization of the wafer substrates

Si wafers with 288 nm Si0₂ were cleaned by rinsing with 2-propanol and milli-Q water. After being blown dry the substrates were immersed in a warm piranha solution for at least 60 minutes, rinsed with de-ionized water and dried at 150 °C for one hour. Thus cleaned, hydrophyllized and dried the substrates were immersed in a 10% solution of trimethoxyoctadecylsilane (Sigma- Aldich, 90+%) in hexane and incubated at 60 °C overnight. For TMS modification TMSC1 was used instead in combination with a few drops of ethyldiisopropylamine. The following day the substrates were rinsed sequentially using hexane, toluene, ethanol and water before being heated at 110 °C for at least one hour. The quality of the surface modifications was verified by sessile drop contact angle measurements which showed contact angles of -100° for the OTS modified surfaces and -84° for TMS modified surfaces.

Example 5 - Graphene transfer

The transfer of the chemical vapor deposition graphene films on Cu is done by first spin-coating a relatively thick PMMA (Poly(methyl methacrylate)) layer over the graphene film on copper film. After etching away the graphene coverage on the other side by oxygen plasma, the Cu film is dissolved in an ammonium persulfate solution. The solution is then completely exchanged multiple times with deionized water to remove as much of the dissolved salts as possible.

Eventually, the graphene together with the polymer film is left floating in the aqueous phase from which it is carefully scooped up using a Si0₂/Si substrate. The PMMA film is dissolved using acetone, leaving uniform, large area monolayer graphene on the substrate for further processing. In order to remove residues left behind during the final washing step the graphene can be further cleaned by annealing in forming gas (8:2 Ar/H₂, 1-10 mbar, 80 seem flow) at 350 °C for one hour.

In case OTS-modified substrates were used milder annealing conditions (160 °C, one hour) were used to preserve the OTS layers.

Example 6 - Device construction

Silicon substrates with graphene and Au electrodes were immersed for 10 minutes in 10 niM dichloromethane solutions of the copper complexes. The samples were then extensively rinsed using a stream of pure dichloromethane from a syringe before being blown dry in a stream of argon (Linde gas, 4.6 N) filtered through PTFE filter (poresize 0.45 μιη) to exclude dust. The samples were annealed at 50 °C for 10 minutes and then immediately installed in the flowcell and flushed with air (200 seem) for several hours to stabilize drift.

Example 7 - Lock-in technique

When used for ultrasensitive detection, the resistance change of the GFETs might be very small and overwhelmed by noise. In order to recover the very weak (and in our case slow) sensing signal, we employ a lock-in amplifier (HF2LI, Zurich Instruments) to measure with very narrow bandpass filters (~1 Hz). We use the HF2LI to generate a sinusoidal alternating voltage (10 Vpk, frequency 10-100 kHz) to drive a sinusoidal alternating current through a 1 ΜΩ resistor (as a current source, 10 μApk or 7.07 μΑπηβ). This current source is connected to the GFETs (with resistance on the order of 1 kO) in series to drive a constant current lac (7.07 μΑπικ) through the GFETs. The resultant voltage drop VG across the GFETs (proportional to graphene resistance RGr=VGr/Iac) is monitored versus time at a bandwidth of ~1 Hz using the ZiControl (Zurich Instruments) program. A noise frequency sweep is performed before every measurement to identify the testing frequencies with minimum noise power spectrum density (PSD), and thus optimizing the signal-to-noise ratio (SNR). A temperature sensor (PtlOO) is mounted at the outlet of the gas tube and the gas temperature can be read off in-situ (experiments were conducted at room temperature). Example 8 - Synthesis of the Copper(I) hydridotrispyrazolylborate acetonitrile complexes [Cu(Tp^CF3'^4R-^ph)(MeCN)]

The sodium hydridotrispyrazolylborate (1.00 g, 1.14 - 1.38 mmol) was dissolved in dichloromethane (50 mL) or toluene (50 mL, only used in the case of NaTpCF3,4F-Ph) and the solution was purged with argon to remove dissolved oxygen. After bubbling for five minutes [Cu(MeCN)₄]SbF₆ (1.0 equivalent) was added. The solution was left to stir under argon overnight. The following day the stirring was stopped to allow the NaSbF₆ by-product to settle. The supernatant was then siphoned off and filtered through a syringe filter (0.45 μιη, PTFE). The clear solutions were then evaporated to dryness in vacuo to yield the complexes as white solids. The complexes thus obtained were typically of good purity, additional purification could be performed by recrystallization of the complexes using CH₂C1₂/Et₂0, CH₂Cl₂/pentane or storage of a concentrated solution of a complex in CH₂C1₂ at -20 °C.

[Cu(Tp^CF3'^4CF3-^ph)(MeCN)]

Performed at half scale, yield 381 mg (87 ). *H NMR (500 MHz, CD₂C1₂) δ 7.21 (d, / =8.3 Hz, 6H), 6.98 (d, / =8.3 Hz, 6H), 6.61 (s, 3H), 4.37 (bs, 1H), 2.32 (s, 3H). ¹⁹F NMR (471 MHz, CD₂C1₂) δ -61.84, -63.48. ¹³C NMR (126 MHz, CD₂C1₂) δ 148.66, 142.49 (q, / =37.8 Hz), 134.68, 131.02 (q, / =33.1 Hz), 130.46, 124.99, 124.13 (q, / =272.0 Hz), 121.53 (q, / =269.3 Hz), 114.55, 105.10, 1.16. Elemental analysis calculated (%) for C₃₅H₁₉BCuF₁₈N₇ ^»1.5H₂CK).5Et₂C> (found): C 43.66 (43.36), H 2.67 (2.78), N 9.63 (9.36).

[Cu(Tp^CF3'^4C1-^ph)(MeCN)]

Yield 963 mg (95 %). H NMR (500 MHz, CD₂C1₂) δ 7.01 (d, / =8.4 Hz, 6H), 6.79 (d, / =8.4 Hz, 6H), 6.54 (s, 3H), 4.38 (bs, 1H), 2.30 (s, 3H). ¹⁹F NMR (471 MHz, CD₂C1₂) δ - 62.00. ¹³C NMR (126 MHz, CD₂C1₂) δ 148.98, 142.20 (q, / =37.4 Hz), 135.09, 131.40, 129.56, 128.34, 121.61 (q, / =269.4 Hz), 114.41, 104.65 (q, =1.7 Hz), 2.70. Elemental analysis calculated (%) for C₃₂H₁₉BCl₃CuF₉N₇ ^»0.1CH₂Cl₂ (found): C 44.74 (44.48), H 2.25 (2.42), N 11.38 (11.18).

[Cu(Tp^CF3'^4F-^ph)(MeCN)]

Yield 635 mg (63 %). H NMR (500 MHz, CD₂C1₂) δ 6.87 (ddd, / =8.2, 5.2, 2.5

Hz, 6H), 6.73 (t, / =8.7 Hz, 6H), 6.52 (s, 3H), 2.30 (s, 3H). ¹⁹F NMR (471 MHz, CD₂C1₂) δ -61.88, -113.25 (tt, =8.8, 4.7 Hz). ¹³C NMR (126 MHz, CD₂C1₂) δ 131.91 (d, =8.4 Hz), 115.05 (d, / =22.0 Hz), 104.72 (q, =2.0 Hz), 1.15. Elemental analysis calculated ( ) for C₃₂H₁₉BCuF₁₂N₇ ^»0.7CH₂Cl₂ ^»0.5toluene (found): C 47.81 (48.22), H 2.70 (2.52), N 10.78 (10.37).

[Cu(Tp^CF3-^ph)(MeCN)]

Yield 897 mg (87 %). H NMR (500 MHz, CD₂C1₂) δ 7.27 (dd, / =8.1, 6.5 Hz, 3H), 6.95 (t, J =1.6 Hz, 6H), 6.88 (dd, J =1.9, 1.4 Hz, 6H), 6.55 (s, 3H), 4.66 (bs, 1H), 2.30 (s, 3H). ¹⁹F NMR (471 MHz, CD₂C1₂) δ -61.76. ¹³C NMR (126 MHz, CD₂C1₂) δ 150.44, 141.94 (q, =37.3 Hz), 131.07, 130.06, 128.61, 128.18, 121.83 (q, / =269.0 Hz), 114.32, 104.53 (q, / =1.8 Hz), 2.69. Elemental analysis calculated (%) for C₃₂H₂₂BCuF₉N₇ (found): C 51.25 (51.11), H 2.96 (3.11), N 13.07 (12.73).

[Cu(Tp^CF3-^40Me-^ph)(MeCN)]

Yield 997 mg (97 %). H NMR (500 MHz, CD₂C1₂) δ 6.80 (d, / =8.7 Hz, 5H), 6.51 (d, / =8.7 Hz, 6H), 6.49 (s, 3H), 4.53 (bs, 1H), 3.78 (s, 9H), 2.29 (s, 3H). ¹⁹F NMR (471 MHz, CD₂C1₂) δ -61.68. ¹³C NMR (126 MHz, CD₂C1₂) δ 150.99, 150.40, 141.51 (q, / =37.0 Hz), 130.96, 122.00 (q, / =268.6 Hz), 118.83, 114.09, 111.44, 103.29, 40.17, 2.73. Elemental analysis calculated (%) for C₃₅H₂₈BCuF₉N₇0₃ (found): C 50.05 (49.99), H 3.36 (3.31), N 11.67 (11.61).

[Cu(Tp^CF3'^4NMe2-^ph)(MeCN)]

Yield 863 mg (84 %). ^lU NMR (500 MHz, CD₂C1₂) δ 6.74 (d, / =8.8 Hz, 6H), 6.43 (s, 2H), 6.26 (d, =8.8 Hz, 6H), 4.70 (s, 1H), 2.92 (s, 18H), 2.28 (s, 3H). ¹⁹F NMR (471 MHz, CD₂C1₂) δ -61.61. ¹³C NMR (126 MHz, CD₂C1₂) δ 160.07, 150.13, 141.80 (q, =37.3 Hz), 131.46,

123.56, 121.86 (q, / =268.8 Hz), 114.26, 113.41, 103.97, 55.43, 2.70. Elemental analysis calculated (%) for C₃₈H₃₇BCuF₉N₁₀ ^«0.5C₅H₁₂ (found): C 53.15 (53.30), H 4.74 (4.26), N 15.30 (15.59). General considerations

All manipulations of air-sensitive compounds were performed in an atmosphere of purified argon gas using standard Schlenk techniques. The sodium salts of the ligands were synthesized using a procedure we described previously (van Dijkman, T. F.; Siegler, M. A.; Bouwman, E., Dalton Trans. 2015, 44, 21109-21123). All solvents were purchased from commercial sources and reagent grade. The graphene used in this work was purchased from Graphenea Inc. Solvents used for air-sensitive manipulations were dried and deaerated using a PureSolv MD 5 Solvent Purification System and stored on 3 A molecular sieves under argon. When appropriate, glassware was flame dried in vacuo immediately prior to use. NMR spectra were recorded on a Bruker AV500 spectrometer (500 MHz for ¾ 471 MHz for ¹⁹F and 126 MHz for ¹³C). Elemental analyses were performed by the Microanalytical laboratory Kolbe in Germany. IR spectra were recorded on a Perkin Elmer UATR Two FT-IR spectrometer set to a resolution of 1 cm^"1. ESI-MS spectra of compounds in MeCN were recorded on a Thermal Finnigan AQA ESI- MS system. Contact angles were determined using a Rame-Hart goniometer using drops of milli-Q water. Multiple drops were used and the results averaged. EUipsometry was performed using a Vase Ellipsometer from J.A. Woollam Co. Inc. and fitted using a Cauchy model. Data analysis was performed using Origin 9.1 (OriginLab).

Claims

1. A device for detection of a gaseous analyte, wherein the device is comprised of a. a first layer, being a substrate layer (10);

b. a second layer (20), being an electrical insulating layer interposed

between said first layer (10) and a third layer (30), being a graphene layer;

2. A device according to claim 1 , wherein said transition metal compl according to formula

wherein,

M is a transition metal, selected from copper, silver, gold, titanium, chromium, manganese, iron, nickel preferably gold, more preferably silver, most preferably copper; and,

R¹ is selected from the group NMe₂, OMe, N0₂, H, F, CI, CF_3i and R¹ substituents being either identical or different; and,

R² is selected from the group H, alkyl, aryl, acyl or aromatic ring; and, the number of aromatic rings on B is 2, 3 or 4, preferably 3 or 4, most preferably 3.

3. A device according to claim 1 or 2, wherein said fourth layer (40) has a layer thickness of between 0.1 nm to 100 nm, preferably of between 0.5 nm to 10 nm, more preferably 0.5 nm to 5.0 nm.

4. A device according to any of claim 1 to 3, wherein said second layer (20) is comprised of an insulating material selected from the group of silicon dioxide, silicon oxynitride, silicon nitride, preferably silicon dioxide.

5. A device according to any of claim 1 to 4, wherein said second layer (20) has a layer thickness of at most 500 nm, preferably at most 300 nm, most preferably at most 100 nm.

6. A device according to any of claim 1 to 5, wherein first substrate layer (10) is comprised of a conductor, preferably p-doped or n-doped silicon.

7. A device according to any of claim 1 to 6 wherein said first layer (10) is provided on a conductive layer comprised of a metal, preferably comprised of a metal selected from the group of gold, platinum, palladium, silver, copper, nickel, preferably gold.

8. A device according to any of claim 1 to 7, wherein said substrate layer (10) is in electrical communication with an electrode (60).

9. A device according to any of claim 1 to 8, wherein said electrodes (50) and a electrode (60) are comprised of a conductive material, preferably a metal selected from the group of gold, platinum, palladium, silver, copper, nickel, preferably gold.

10. A device according to any of claim 1 to 9, wherein said gaseous analyte is selected from the group of ethene, carbon monoxide, carbon dioxide, nitrogen monoxide, nitrogen dioxide, acetone, methyl ethyl ketone, benzene, styrene, xylene, tetrahydrofuran, acetonitrile, ammonia, water, ethanal, propene, 1-methylcyclopropene, ethyl ethanoate, methyl ethanoate, ethanoic acid, methanoic acid, propanoic acid, ethanol, methanol, 1-propanol, 2-propanol, 1- butanol, 1-hexanol, hexanal, phenol, H₂S, preferably ethene, carbon monoxide or ethanol, most preferably ethene.

11. A device according to any of claim 1 to 10, wherein a top surface of said fourth layer (40) is coated with a hydrophobic layer.

12. A device according to claim 11, wherein said hydrophobic layer is comprised of a composition that reacts with carbon oxides and/or nitrogen oxides.

13. A device according to any of claim 1 to 12, wherein said hydrophobic layer is comprised of tin oxide with palladium and/or platinum.

14. A device according to any of claim 1 to 13, wherein a layer comprised of silane is interposed between said second layer (20) and said graphene layer (30).

15. Method for the detection of gaseous analyte, wherein said method comprises the steps;

a) exposing a device according to any of claim 1 to 14 to a gaseous analyte; b) detecting a change in conductance of a graphene layer (30) by measuring a change in electric potential between electrodes (50).

16. Method according to claim 11, wherein said method is capable of measuring a gaseous analyte below 10 ppm, preferably below 1.0 ppm, more preferably below 0.2 ppm, most preferable below 0.1 ppm.

17. Method according to claim 11 or 12, wherein said gaseous analyte is ethene, carbon monoxide, carbon dioxide, nitrogen monoxide, nitrogen dioxide, acetone, methyl ethyl ketone, benzene, styrene, xylene, tetrahydrofuran, acetonitrile, ammonia, water, ethanal, propene, 1-methylcyclopropene, ethyl ethanoate, methyl ethanoate, ethanoic acid, methanoic acid, propanoic acid, ethanol, methanol, 1-propanol, 2-propanol, 1-butanol, 1-hexanol, hexanal, phenol, H₂S, preferably carbon monoxide or ethanol, preferably ethene.

18. Use of a device according to any of claim 1 to 14 to measure ethene concentrations in agriculture and food related products, preferably fruit, vegetables, flowers, plants, bulbs, cereals, and other agriculture fresh products.

19. Use of a device according to any of claim 1 to 14 to measure the carbon monoxide concentrations, or ethanol concentration.