Sensing method, sensing device and method of manufacture

Abstract

A method of sensing, comprising the steps of exposing a sensing device to a gas phase analyte and subjecting the sensing device to AC excitation and measuring a DC conductance of the sensing device. The sensing device comprises discrete carbon-based electrically conducting regions on a surface of an electrically insulating carbon-based material substrate. Electrical conduction through the sensing device is by percolation between the electrically conducting regions. Interaction between the gas phase analyte and the discrete carbon-based electrically conducting regions affects the DC conductance to enable sensing of the analyte. Sensing devices configured to be used in such methods of sensing, and methods of manufacturing such sensing devices.

Description

[0001] SENSING METHOD, SENSING DEVICE AND METHOD OF MANUFACTURE

[0002] This application claims priority from GB 2313455.4 filed 04 September 2023, the contents and elements of which are herein incorporated by reference for all purposes.

[0003] Field of the Invention

[0004] The present invention relates to a method of sensing gas phase analytes. The present invention also relates to sensing devices configured to be used in such methods of sensing. The present invention also relates to methods of manufacturing such sensing devices.

[0005] Background

[0006] Devices capable of sensing through a combination of a sensor array and a pattern recognition system are known as “electronic noses”. The electronic nose, a device that tries to mimic the function of the biological nose, is needed to support many areas of activity where the use of a human or animal nose would be inappropriate for safety reasons or when long term, continuous coverage is required [1]. Electronic nose technology is particularly promising for use in complex chemical environments. For example, use cases might be determining optimum crop harvest time, disease and pests, monitoring industrial processes and sensing hundreds of compounds in human breath simultaneously, for the diagnosis of diseases or otherwise.

[0007] Since olfaction in animals has evolved as a partly functioning sense at birth, the biological nose requires training to reach its full potential; but even after training cannot be used continuously as this ability is dulled by overuse. For these reasons, the biological nose is therefore an expensive resource to create and to maintain, as well as being unable to record results in a quantitative and meaningful way, under most circumstances.

[0008] The first electronic nose was created by Persaud and Dodd [2], using chemoresistive gas sensors. These devices are used to detect particular chemicals or classes of chemicals through a change in resistance, but can respond to multiple chemicals. This lack of selectivity can be a problem when used as a single sensor with unknown gas mixtures. To solve this problem, Persaud and Dodd exploited the different sensitivities of each type of gas sensor to a range of gases as part of an array of different sensors, in order to determine the mixture of gases within a sample. A chemoresistive sensor can be made using many types of semiconductor (often metal-oxide or polymer based), exploiting either the inherent sensitivity of each semiconductor material to a range of different gases or by using functionalisation to extend that range to gases that would otherwise be undetectable [3]. The number of different gases that this type of electronic nose is able to detect and distinguish between the components in a mixture then depends on the number of different chemoresistive gas sensors within the array. Many different semiconductor material systems have been studied, including 2D and nanowire forms and systems for measurement of arrays of more than 100 sensors have been developed [4]. Despite the complexity of these arrays, it still remains a problem to realise a level of performance comparable to a biological nose. This limitation with chemoresistive arrays has led to the search for alternative approaches to an electronic nose function, these alternatives include acoustic (SAW microbalance, [5]), catalytic (electric field change due to a reactive gate metal in a MOSFET, [6]) and optical methods (eg IR absorbtion spectroscopy of an optically active vapour absorber, [7]). These methods can be divided into two broad groups when considering the problem of identifying the gas being detected. The first group uses a sensor measurement based on a non-chemical structure specific interaction to indirectly identify the molecules being detected. This group includes chemoresistive, acoustic, catalytic and some optical sensors. All produce a single measured parameter from each sensor device and so requires a large number of different sensors to provide sufficient information to distinguish many different chemicals. However, the second group uses a chemical structure specific interaction (such as the measurement of chemical bond energies) to directly identify molecules; this can be done with IR spectroscopy. Here a single measurement can include many parameters eg the light absorbtion at different wavelengths, that is sufficient to distinguish a very large number of different chemicals. However, such spectroscopic methods are much more complex and expensive to implement, as well as having limited portability and sensitivity compared to many chemoresistive or electrochemical sensors

[0009] Existing compact electronic nose (e-nose) technologies, such as those based on chemoresistive sensor arrays, have difficulty in distinguishing large numbers of different chemicals, due to the requirements on the array size and on the limits of measurements based on non-chemical structure specific interactions. Also, many chemoresistive material systems have different operating temperature requirements and complex fabrication routes, so making the production of a large array of different sensors both difficult and expensive. In addition, bulk effect chemoresistors can suffer significant operating speed and poisoning effects, when exposed to incompatible chemicals. SAW and microbalance sensors are susceptible to false readings due to changes in temperature and humidity and often rely on functionalisation for VOC identification. Catalytic sensors are limited to certain types of VOC. These indirect molecule identification systems can work well under controlled circumstances, but have problems when faced with unknown gas mixtures. While IR absorbtion spectroscopy does give direct access to chemical structure specific information, allowing use with complex vapour mixtures, the equipment required is large, expensive and not very portable, which is why the biological nose still has widespread use.

[0010] The widely considered ‘gold standard’ of electronic-nose technologies is high field asymmetric waveform ion mobility spectrometry (FAIMS) in combination with mass spectrometry (MS); an approach which uses mass and shape to identify molecules. The main limitation of this method is that uncertainty increases with molecular size, which poses challenges for identifying larger molecules and those that are also compositionally similar, but structurally different. This includes many volatiles of biological origin to which we fall to nature for a solution e.g. medical and police sniffer dogs, truffle hunting pigs. FAIMS used in combination with MS experiences further barrier to adoption due to its bulk, considerable expense and requirement for a steady inert gas flow.

[0011] The present invention has been devised in light of the above considerations. Summary of the Invention

[0012] The present inventors have realised that it is advantageous to be able to carry out electronic nose type sensing using a sensing device that utilises molecular shape as a characteristic that enables distinction. Many information-carrying volatiles, particularly those from biological origins, reach sizes of 20 / 30 atoms which current technologies are not equipped to identify, (due to size and compositional similarity), but which a molecular shape dependent approach would enable. The present inventors have also recognised the value in a sensing device that is cheap, easily fabricated and portable, and that can be used under ambient conditions. In preferred embodiments, the present invention achieves one or more of these benefits whilst also preferably achieving high levels of chemical discrimination and fouling resilience.

[0013] The invention includes the combination of the aspects and optional features described except where such a combination is clearly impermissible or expressly avoided.

[0014] In a first aspect, the present invention provides a method of sensing, comprising the steps of: exposing a sensing device to a gas phase analyte; and subjecting the sensing device to AC excitation and measuring a DC conductance of the sensing device; wherein: the sensing device comprises discrete carbon-based electrically conducting regions on a surface of an electrically insulating carbon-based material substrate, such that electrical conduction through the sensing device is by percolation between the electrically conducting regions; and interaction between the gas phase analyte and the discrete carbon-based electrically conducting regions affects the DC conductance to enable sensing of the analyte.

[0015] In some embodiments, the method of sensing is able to generate a global response from a gas phase analyte that is unique and can be used (by itself or in combination with other analyses) to identify the components of the analyte. Its mode of sensing is of the electronic nose type.

[0016] The sensing device is exposed to a gas phase analyte and is subjected to AC excitation. Here, AC excitation is an applied AC voltage which excites localised charges in the sensing device, this is described in the detailed description below. The AC excitation encompasses any periodically modulating voltage. This includes sinusoidal waveforms, square waveforms, triangle waveforms, and others. The positive and negative maxima of the AC excitation do not need to have equal magnitude (i.e. there may be a DC offset), as such the AC excitation includes a pulsed DC voltage. The AC voltage is an “excitation” as it delocalises charges trapped in the conducting regions of the device.

[0017] It may be that the AC excitation frequency is greater than or equal to 100 and less than or equal to 600 MHz. It may be that the AC excitation frequency conforms to min < AC frequency < max, where min is 100 MHz, 150 MHz, 200 MHz or 250 MHz and max is 600 MHz, 550 MHz, 500 MHz, or 450 MHz. It may be that the AC excitation frequency is greater than or equal to 250 MHz and less than or equal to 450 MHz.

[0018] The gas phase analyte may be a single analyte, or a mixture of more than one analyte. Notably the present invention works well to identify larger chemicals that known methods struggle to characterize. It is considered that the gas phase analyte adsorbs (physisorbs) to the surface of the sensing device. This may be by van der Waals forces, electrostatic forces or otherwise.

[0019] The sensing device comprises discrete carbon-based electrically conducting regions. Such regions may be comprised of one area of conducting material, or they may be comprised of more than one such region which are closely spaced together. The regions are discrete in the sense that there is no conductive connection between them, and there exists a potential energy barrier to charge movement between them. The individual areas of conducting material may have an average diameter that is greater than or equal to 0.5nm and less than or equal to 2nm. It may be that the individual conducting areas are on average about 1.3nm in diameter. However, the individual areas of conducting material are not particularly limited in size, and may have an average diameter of up to about 600nm. Here, “carbon-based” includes “carbon rich”.

[0020] Regarding the “depth” of a conducting regions; an optical microscope image does not give significant contrast between the conducting regions and any surrounding insulating material (partially degraded carbon-based material, or amorphous carbon, for example), so that it is not possible to measure the depth using optical microscopy. It should also be noted that the surface is not typically flat at this size scale so that defining a depth is problematical. Generally, it may be that the depth of the conducting regions of the sensing device are such that close proximity of the percolation pathway to adsorbed gas phase analytes is still possible, so that the percolation pathway can be affected by their binding. It is also noted that it may be that close proximity between percolation pathway and adsorbed gas phase analytes is ensured through the lack of an oxide layer forming on the surface of the device.

[0021] The conducting regions are produced from the substrate over a thin surface layer. It may be that the substrate is substantially amorphous. Electrical conduction through the sensing device is by percolation between the conducting regions. Percolation conduction is a known phenomenon in the technical field and describes conduction with spatial limitations. Here the percolation comprises charge hopping events between adjacent conducting regions. Here the percolation may be dominated by charge hopping events between adjacent conducting regions, “nearest neighbour hopping”, in the best performing devices.

[0022] The conducting regions encompasses regions which are quasi-conducting or partially conducting. The carbon-based material substrate and any partially thermally or photolytically degraded material in between the conducting regions is electrically insulative.

[0023] In a second aspect the present invention provides a sensing device configured to be used in the method of the first aspect of the present invention.

[0024] In a third aspect the present invention provides a method of manufacturing a sensing device according to the second aspect of the present invention, wherein the conducting regions are formed using localised heating on the surface of the electrically insulating carbon-based material substrate.

[0025] In some embodiments the conducting regions comprises graphitic carbon. Graphitic carbon is a region of graphite-like carbon, in the sense that the region comprises of sp2 bonded carbon planes that may be one or more atomic layers thick. The term has overlap with that of graphene which is also used in literature to describe graphitic carbon up to several atomic layers thick. Carbon in this form allows conduction of charge via extended electronic orbitals.

[0026] It may be that the sensing device, in between the conducting regions, comprises partially thermally or photolytically degraded carbon-based material substrate that is electrically insulative. Such material may be amorphous carbon (a-C).

[0027] In some embodiments the electrically insulating carbon-based material substrate (here, the substrate was a precursor material before the sensing device is formed, and may be described as such) comprises (or in some cases, consists of) an aromatic polymer (that is, a polymer containing an aromatic ring in each monomer unit). The carbon-based material comprising aromatic carbon rings promotes the formation of conducting regions when used in the sensing device.

[0028] In some embodiments the sensing device demonstrates a room temperature non-linear current-voltage relationship which is demonstrative of hopping conduction. As such, it may be that the current-voltage relationship is S-shaped.

[0029] It may be that the electrically insulating carbon-based material substrate comprises (or in some cases, consists of) a polyimide. It may be that the electrically insulating carbon-based material substrate comprises (or in some cases, consists of) an aromatic polyimide. It may be that the electrically insulating carbon-based material substrate comprises (or in some cases, consists of) Kapton®.

[0030] In some embodiments the sensing device has an aspect ratio of length:width, wherein: the length of the sensing device is a dimension that is parallel to a conduction direction; and the width of the sensing device is a dimension that is perpendicular to the length and is parallel to a plane that is tangential to the surface that comprises the electrically conducting regions; wherein the aspect ratio has a value greater than or equal to 2:1 . In this discussion, the relevant dimensions are those of an active region of the sensing device, i.e. not taking into account the parts of the device that merely support the active region. As such, the aspect ratio may be described as a patterned aspect ratio.

[0031] Here, conduction direction is the net conduction direction, and as such is the average direction of the percolation pathway through which conduction occurs. Here, the width is defined relative to a plane that is tangential to the surface that comprises the electrically conducting regions; whilst the micro and nanostructure of the device surface might be un-even and have a complex structure, the plane here is perpendicular to the macroscopic surface of the sensing device.

[0032] When the sensing device has an aspect ratio greater than or equal to 2:1 , it is found that there is a significant number of sites of interaction with the gas phase analyte to encompass all possible structural arrangements. It may be that the aspect ratio has a value greater than or equal to 10:1 .

[0033] It may be that the sensing device has an aspect ratio with a maximum value of 100:1 . It may be that the sensing device has an aspect ratio with a maximum value of 75:1 . It may be that the sensing device has an aspect ratio with a maximum value of 50:1 . An aspect ratio that is too high may lead to a sensing device resistance that is too high, so that a resultant signal is too difficult to measure due to noise. The aspect ratio in part depends on the laser power and the modulation conditions.

[0034] Note that “hopping” is defined as charge transport via the phonon-induced tunnelling of electrons between localised states which are randomly distributed in energy and position and it describes the mode of conduction taking place in disordered materials. Hopping happens along percolation pathways, and here it may be that the device dimensions are such that the complete pathways are limited and many deadend pathways exist. By choosing a high aspect ratio, the number of complete percolation pathways is diminished, and the spread of pathway resistivities is narrowed, so that percolation pathway switching is able to take place. The laser conditions may be chosen to reduce the thickness of the channel material and increase the probability that conduction takes place at the very surface.

[0035] It may be that the conducting regions of the sensing device are contained within a channel. It may be that the channel is a trench. Here, a trench is understood to mean an elongate depression or groove in the sensing device. The trench may be a protective trench, such that the conducting regions are protected from damage (for example, scratching).

[0036] In some embodiments, the density of conducting regions in the channel is highest at the centre widthwise. Here, the density refers to the density of conducting region, p = — , where Acis the total area of conducting region in the trench, and ATis the total area of the trench. That is, it may be that the average spacing between conducting regions is smallest at the centre of the channel width-wise.

[0037] In some embodiments the sensing device comprises amorphous carbon between the conducting regions. It may be that between the conducting regions the sensing device comprises partially thermolytically or photolytically degraded material.

[0038] In some embodiments the conducting regions are heterogeneously distributed. When a channel is present, it may be that the conducting regions are heterogeneously distributed across the channel. These features promote a molecular shape dependent affinity to the device.

[0039] In some embodiments the sensing device comprises heterogeneously distributed attachment points for analyte binding. This feature also promotes a molecular shape dependent affinity to the device.

[0040] In some embodiments, the AC excitation voltage is in the range 0.1 to 10V. It may be that the lower value of the range of AC excitation voltage is 0.25V, 0.5V, 0.75V or 1V. It may be that the upper value of the range of AC excitation voltage is 7.5V, 5V, 4V, 3V or 2V. It may be that these voltages are those applied indirectly to the device. It may be that these voltages are those applied to the AC branch of a bias tee which precedes the sensing device.

[0041] The waveform of the AC excitation voltage is not particularly limited. The key feature is that the device is subject to a periodic voltage modulation. Note that as discussed herein, the AC excitation may be describing the voltage applied to the sensing device (i.e. that applied directly to the device), or it may be the voltage as applied indirectly (i.e. via some other component, for example, a bias tee).

[0042] It may be that the waveform of the AC excitation is sinusoidal, square, triangular, or any other form. It may be that the maximum negative amplitude and maximum positive amplitude of the AC excitation have the same magnitude. It may be that the maximum negative amplitude and maximum positive amplitude of the AC excitation are not the same (i.e. there is a DC offset). It may be that the maximum positive amplitude is greater in magnitude that the maximum negative amplitude (positive DC offset), or it may be that the maximum negative amplitude is greater in magnitude than the maximum positive amplitude (negative DC offset). In some embodiments, the waveform of the AC excitation is an important parameter for optimising the response generated by the method of the present invention.

[0043] In some embodiments the DC voltage is in the range 0.1 to 10 mV. It may be that the lower value of the range of DC voltage is 0.25 mV, 0.5mV or 0.75mV. It may be that the upper value of the range of DC voltage is 7.5 mV, 5mV or 0.25mV. It may be that the DC voltage is approximately 1 mV.

[0044] It may be that the AC voltage is greater, e.g. substantially greater, than the DC voltage. A large difference in voltages between AC and DC ensures that charge transport in the device is maintained in the non-equilibrium regime by the AC excitation. It may be that the AC voltage is at least 10 times greater than the DC voltage. It may be that the AC voltage is at least 100 times greater than the DC voltage. It may be that the AC voltage is at least 1000 times greater than the DC voltage.

[0045] In some embodiments the method of sensing comprises repeating the measurement of DC conductance as the AC excitation frequency is varied to generate a spectrum. It may be that the spectrum plots DC conductivity vs AC frequency. It may be that any number of measurements of DC conductance are taken at different AC frequencies.

[0046] In some embodiments the method of sensing comprises a step of analysing the spectrum. Using this measurement approach gives several options for data analysis. One approach may be to use discrete measurements at different frequencies, which could be treated in the same way as an array of different sensor devices. Since expanding the number of frequencies sampled has no upper limit, realising an array size of greater than 50 is easily done. Another approach may involve recognising the appearance and location of particular peaks and associating them with particular atomic arrangements.

[0047] It may be that information is encoded within other spectral differences which appear in the form of peak ratios, shifts, FWHM, Q-factor and ‘noise’. It may be that the analysis step involves machine learning techniques. In this context, machine learning is particularly useful as a method for extracting meaning from less clear relationships in the data. The method of analysis ultimately aims to result in identification of one or more gas phase analytes.

[0048] In some embodiments the method of sensing further comprises a calibration step, wherein the DC conductance of an individual sensing device when exposed to a known gas phase analyte under AC excitation is measured. Each device will have a difference micro- and nanoscale structure, and so for a given analyte vapour under test, a different DC conductance vs AC frequency response will be generated for each sensing device, even if such a difference is very slight. As such, each sensing device produced may benefit from calibration with a known analyte, set of analytes, or mixture of analytes.

[0049] Without wishing to be bound by theory, the present inventors consider that the constituent graphitic content, which is distributed heterogeneously in an insulating matrix, is a significant factor in the operation of the sensing in the present invention.

[0050] In some embodiments the method of sensing is performed at ambient temperatures. It is an advantage of the present invention that localised charge transport can be used at ambient temperatures. It is particularly useful for practical applications where in situ monitoring and sensing is preferable. Here, “ambient” encompasses moderate temperatures and pressures, and those that are found in environments in which the device might be used. Room temperature and pressure is encompassed by, and representative of, ambient conditions.

[0051] In some embodiments the temperature and pressure in the method of sensing are modulated to adjust a degree of adsorption of the analyte to the sensing device. It may be that changing the temperature and / or pressure promotes greater or lower adsorption amounts, which may be useful for optimising the response generated, and ultimately optimising the method of sensing.

[0052] In some embodiments of the method of manufacturing the sensing device of the present invention, the conducting regions are formed by localised heating on the surface of the electrically insulating carbonbased material substrate using laser light of wavelength that is greater than or equal to 360nm and less than or equal to 480nm. That is in some embodiments the wavelength (A) of the laser light conforms to 360nm < K < 480nm. All laser induced change produces a gradient with depth (where maximal change is seen at the surface), however to ensure percolation conduction at the surface a short wavelength with low penetration depth is chosen. However, this value should not be so energetic that the laser-driven reactions become photolytically driven to such a degree that, for example, stable carbon rings are broken, disrupting the formation of the graphitic material that might, in some embodiments, make up the conducting regions. Around 350nm the process of photolytic breakdown might begin to emerge.

[0053] More generally, in such embodiments, a number of temperature and exposure time driven thresholds may exist driving the thermolytic transformations of the insulating material to form conducting regions. The choice of laser beam properties is used to engineer the chemical and micro composition, the width and depth of the conducting regions on the sensing device. For example, 375nm irradiation, previously unexplored, has a short penetration depth in polyimide. The attenuation happens over approximately an order of magnitude lower thickness than what has been done before and consequently should be expected to result in a thinner layer of degraded insulating material. A laser wavelength of 360nm < K < 480nm means that the polymer decomposition is predominantly photothermic in nature and hence less likely to result in breakage of the aromatic rings. The beam intensity at initial exposure decreases by Beer’s Law with depth and due to beam profile and heat flow, this means that the temperature profile reaches a peak at the central point. The temperature thresholds for thermolytic driven changes are passed, and the greatest density of graphitic clusters are formed at the very surface and around the central point. At the point of highest density, there is the lowest spacing between graphitic islands, therefore hopping, which has distance dependency, is most probable. That is, it may be that the invention provides a material with a nonlinearly changing composition across its depth and width.

[0054] Regarding the “width” of the channel, the electrical properties of the sensing device arise from a small region (the current is carried along one sequence of hops or percolation pathway, however the terminating branches will also contribute to the electrical behaviour). When ‘width’ is discussed, this is the average breadth of the region giving a darkening under the microscope, but it is not the total breadth of the laser-induced degradation of the polymer or a measure of the region through which the charge is being moved.

[0055] It may be that the wavelength of the laser light conforms to min < A < max, wherein min is 365nm or 370nm and max is 470nm, 450nm, 430nm, 41 Onm, 390nm or 380nm. It may be that the wavelength of the laser light is about 375nm.

[0056] It may be that the laser light used herein are of a continuous-wave type. This enables finer control of the temperature profile and produces a denser, more uniform conducting regions. Greater control of exposure is achieved and low exposure conditions can be investigated using a continuous type wave.

[0057] In some embodiments of the method of manufacturing the sensing device of the present invention, the conducting regions are formed by localised heating on the surface of the electrically insulating carbonbased material substrate using laser light of power that is greater than or equal to 14mW and less than or equal to 70mW. That is in some embodiments the power of the laser light conforms to 14mW < power < 70mW. It may be that the laser light conforms to 14mW < power < max, wherein max is 60mW, 50mW, 40mW or 30mW. It may be that the laser light conforms to 14mW < power < 28mW.

[0058] In some embodiments of the method of manufacturing the sensing device of the present invention, the conducting regions are formed by localised heating on the surface of the electrically insulating carbonbased material substrate using laser light with a scan rate that is greater than or equal to 25 mms1and less than or equal to 175 mms1. It may be that the scan rate is greater than or equal to 2mms1and less than or equal to 300 mms1. The formation of the conducting regions is a temperature driven process, and so the combination of scan rate and power is important to the extent of formation of the conducting regions.

[0059] In some embodiments of the method of manufacturing the sensing device of the present invention, the conducting regions are formed by localised heating on the surface of the electrically insulating carbonbased material substrate using laser light with a duty ratio greater than or equal to 1 :2 and less than or equal to 2:1 . That is, the laser light may conform to 1 :2 < duty ratio < 2:1 . It may be that the laser light conforms to min < duty ratio < max, wherein min is 1 :2, 2:3 or 3:4, and max is 2:1 , 3:2, or 4:3. It may be that the duty ratio is about 1 :1 . The duty ratio describes the ratio between the time that the laser light is on compared to the time that the laser light is off. It may be that the duty ratio is such that the laser light is turned on and off at a frequency of greater than or equal to 180kHz and less than or equal to 250 kHz.

[0060] It may be that the sensing device further comprises highly conductive regions at both ends, and metal contacts connected thereto, through which electrical interrogation (DC conductance measurement under AC excitation) can be carried out. It may be that the metal contacts are indium metal contacts. It may be that the highly conducting regions are formed by controlling the laser light parameters (power, scan rate and duty ratio), in order to control the chemical composition, microstructure and macrostructure of the channel.

[0061] Summary of the Figures

[0062] Embodiments and experiments illustrating the principles of the invention will now be discussed with reference to the accompanying figures in which:

[0063] Fig. 1 illustrates a vertical cross-sectional diagram of the sensing device of an embodiment of the present invention (not to scale);

[0064] Fig. 2 illustrates a Raman spectrum of a laser carbonised conductive channel, produced by a method of an embodiment of the present invention;

[0065] Fig. 3 illustrates optical microscopy images (x50 magnification) of tracks fabricated at lOOmms1at a range of power and duty ratios;

[0066] Fig. 4 illustrates three devices produced identically using the same channel irradiation conditions of 21 mW at 75 mms1and measured at 293K;

[0067] Fig. 5 illustrates device current-voltage characteristics over temperature range (channel irradiation conditions of 15mW at 75 mms1);

[0068] Fig. 6 illustrates typical initial current-voltage sweeps demonstrating percolation pathway switching.

[0069] Fig. 7 illustrates a representation of sp2 bonded carbon cluster in an insulating matrix, possible percolation pathways are demonstrated through the lines connecting darker grey portions and the single headed arrows indicate the single pathway responsible for conduction. The small section of percolation pathway across the disordered carbon surface is influenced by the barriers (insulating matrix) between conducting regions (sp2 bonded carbon) and by trapped charge (coulombic interactions). At each AC frequency, the distribution of charge along the channel changes, such that a nearby island or branching pathway may enable current to flow more easily. This distribution of charge results from the difference in time constants that govern charge movement between islands and along branches. The applied AC voltage drives a DC current along the favoured sequence of conductive islands at each frequency. This situation can be complicated by the charging and discharging of terminating branches into this dominant pathway if the branch time constant, hopping distances and charge distribution favours this. Due to the spatial distribution of channel attachment points, each molecule should have a characteristic affinity, decided by its own molecular attachment points, to different regions / sites along the channel length. The adsorption of the molecules changes the local dielectric permittivity, changes the potential barrier energy and provides coulombic interactions which changes the time constants governing charge movement and produces the resultant DC conductance spectra.

[0070] Fig. 8 illustrates a schematic plan view (top) of a laser processed Kapton film. Unexposed film, strongly conducting region (heavily exposed) and weakly conducting region (weakly exposed) are labelled. Schematic cross-sectional view (bottom) of a laser processed Kapton film mounted for electrical measurement as the device under test (DUT). The Kapton film is clamped between two unreactive plastic plates. The upper plate has an aperture to allow ambient gas to reach the weakly conducting region, in addition to two copper wires terminating in Indium metal pads to contact the strongly conducting regions.

[0071] Fig. 9 illustrates measurement setup of AC source (E8257D) and DC source (KE237), the bias tee (HP33150A) and sample enclosure containing the DUT, with provision for controlling the gas ambient and device temperature;

[0072] Fig. 10 illustrates a plot of low field conductivity measured across temperature range (blue points). Exponential fit (red line) to the data indicates a relationship captured by the generalised hopping equation

[0073] Fig. 11 illustrates reference characteristic for device with no gas and using a fixed DC bias of 1 mV and an AC power of +15dBm at the capacitively coupled termination of the bias tee

[0074] Fig. 12 illustrates the same reference characteristic as in Fig. 11 , but now including the device responses to an acetone environment and an I PA environment saturated at 660mBar

[0075] Fig. 13 illustrates ball and stick model for acetone and IPA molecules, showing the limited differences between them that results in a large difference in the frequency response;

[0076] Fig. 14 illustrates (top) sensing device characteristics taken in dichlorobenzene, toluene and turpentine atmospheres; (bottom) sensing device characteristics taken in sulpholane, nitromethane, water, dimethyl sulphoxide and hexamethyldisilane saturated at 660mBar;

[0077] Fig. 15 illustrates how changes in laser processing conditions results in systematic changes in pattern morphology. These changes influence the electrical conduction behaviour via controlling the factors that govern disordered transport and the dimensions over which this can take place. The table shows that changes happening to the polyimide are driven by a temperature profile and that these are controlled via the laser scan rate and laser power. Temperature thresholds exist for different degradation processes to take place which means that you can control the chemical composition as well as the micro structure (e.g.graphitic cluster spacing and surface topography) and macro structure (channel dimensions) using this process. The time sustained at each temperature changes the types of reactions taking place and the rate. Detailed Description of the Invention

[0078] Aspects and embodiments of the present invention will now be discussed with reference to the accompanying figures. Further aspects and embodiments will be apparent to those skilled in the art. All documents mentioned in this text are incorporated herein by reference.

[0079] Percolation Pathway Switching in Laser Graphitized Polyimide Conducting Tracks

[0080] Below it is described how laser processing has been used to create weakly conducting tracks in polyimide film (the electrically insulating carbon-based material that forms the precursor material and subsequently, substrate). Raman spectroscopy shows that these tracks consist of nanometre sized graphitic regions (discrete sp2 bonded carbon electrically conducting regions) contained in a insulating matrix. The measured temperature dependent and electric field dependent conduction characteristics show an activated characteristic that is consistent with nearest neighbour hopping. In addition, discrete percolation pathway switching events are seen when the system is subjected to significant disturbance from equilibrium. The following discussion provides a detailed description of the theoretical context of the present invention, as understood by the present inventors at the time of writing.

[0081] Disordered electron transport is of interest for a range of device applications, including thermometry [8], ovonic memory [9], sensors

[0010] and information processing

[0011] . Many semi-conductors and granular metal films show disordered transport behaviour, but engineering the transport properties requires very precise control over material composition and structure. A common feature of these disordered systems is the presence of highly conducting islands within a non-conducting matrix, where electron transport occurs through hopping over small gaps between these islands. Hopping is a tunnelling process that requires these gaps to be of very small size since the tunnelling probability decreases exponentially with distance. Control over the gap distance is probably the most important factor in engineering disordered transport behaviour.

[0082] Herein, the inventors investigate the electron transport characteristics of laser graphitized tracks in polyimide (Kapton) fabricated under conditions designed to realise very weak electron conduction. Previous work [12-21] has concentrated on realising strongly conducting tracks made using high power laser light to degrade the polyimide at high temperature. This process results in a rapid loss of material through the liberation of gases resulting from polymer chain breakup

[0022] . The remaining material is usually a porous layer consisting of a mixture of (highly conductive) sp2 bonded and amorphous (highly insulating) carbon-rich regions, as shown by Raman spectroscopy. A low resistance track is created when the interconnections between the sp2 bonded regions dominate the structure, as is the case with these thicker porous tracks.

[0083] Here, control of the laser processing has enabled a thin non-porous carbon-rich layer to be formed, which allows greater control over the interconnections between the sp2 bonded regions. This is done by using a short wavelength laser to limit the depth of material in which the light is absorbed (see the absorbance spectra

[0023] ). The chosen wavelength of 375nm lies just above the point at which degradation transitions from being primarily photolytic to thermolytic, hence preserving the carbon rings which are known to align to form the graphitic content [24-27]. By adjusting the scan speed and laser power, and hence controlling the temperature profile it is possible to systematically create tracks from a highly conductive state through to highly insulating.

[0084] Exposures were carried out using an L3PM tool, manufactured by 3TTechnologies

[0028] ; this has a 375nm laser with nominal focus diameter of 12pm. Polyimide film from a range of sources have been tested and all behave similarly. A range of laser beam powers were used with the beam turned on and off at a frequency of 30kHz to realise controlled modulation (mark-to-space or duty ratio) ratios at various scanning speeds in order to change the exposure conditions. Higher exposure conditions (70mW, 50mms“1, no modulation) realise highly conducting tracks similar to those seen in previous work. These conditions were used to form connecting leads between the channel (a single pass exposure track at 15mW, 75mms-1) with length 70pm crossing two highly conducting tracks that were contacted using indium metal away from the device under test for electrical characterisation. The laser exposed track has a carbon-rich deposit left at the base of a shallow trench a few microns in depth, as shown schematically in Fig. 1 .

[0085] The width of this trench depends on the laser exposure conditions, but the depth remains similar across this range. The thickness of the deposit varies across the width of the track and is accompanied by significant texture differences, as seen in optical micrographs, see Fig. 3. Raman spectroscopy shows a ratio of peaks consistent with sp2 bonded graphitic regions of size ~ 1 .4nm, calculated via the method given in

[0029] , see Fig. 2. Briefly, the ratio of the D and G peaks allows the region size to be found by: where C=0.707

[0086] This equation relates the ratio of D and G peak intensities to the graphitic cluster diameter. Several equations exist for the relationship between this ratio and the graphitic cluster size. The reasoning for using this equation is that 1) the ratio can relate to two positions on the Ferrari-Robertson ‘amorphisation trajectory’ but what is know about the material from previous studies indicates that it would be the more disordered stage. 2) Features of the Raman spectra across the range of conditions (including the FWHM, intensity and position of all peaks) almost all lie within the range that appear for graphene oxide to which this equation has been successfully applied. 3) A less important factor is that the spectra indicate the type of impurities and defects which would be what can be expected for this material and which are understood to influence which equation is used.

[0087] Assuming that the remaining material in the carbon-rich deposit (is in an amorphous form and so) is nonconducting, then the electrical conduction mechanism will depend on the strength of coupling between the conducting sp2 bonded regions. Conduction will be confined to pathways where the gaps between these sp2 bonded regions are smallest, as shown schematically in Fig. 7. The conduction route with the most favourable combination of gaps for charge movement is known as a percolation pathway. Since the sp2 bonded regions will be distributed vertically as well as in the plane, the percolation pathway may be very complex. However, due to the thermal gradient with depth present during laser processing resulting from the beam exposure, the greatest density of graphitic clusters can be expected at the channel surface. The width and thickness of the channel as well as the topographic and chemical composition of the carbon-rich deposit can be systematically controlled by changing the exposure conditions, as shown the optical images in Fig. 3.

[0088] Both the presence of temperature thresholds enabling thermolytic degradation, and the duration at which they are sustained, influence the resulting chemical composition and macromolecular structure. This may be observed as considerable colour and texture differences across 1) the width of the laser exposed region and 2) between optical micro-graphs demonstrating tracks created with different exposure scan rate, power and duty ratio, even at the same fluence. Across the conditions tested, the graphitic cluster size remains the same, but inter-cluster spacing and channel dimensions change; in this way, providing a method through which the electrical characteristics may be engineered - and weakly conducting conditions achieved.

[0089] As the extent of the laser exposure is reduced (by a combination of reducing the continuous laser power, laser scan rate and by modulation) the electrical behaviour of the device becomes more non-linear, see Fig. 3. Under these conditions alternative conduction pathways are cut off until electrical conduction is limited to a single dominant percolation pathway, where the spacing between the sp2 bonded regions is expected to be smallest. Such a weak conductor shows a significant IV non-linearity and temperature dependence, see Fig. 4 and Fig. 5 respectively.

[0090] In order to understand this temperature dependence, the different electron transport mechanisms in disordered systems must be considered. Dasgupta et al

[0030] have suggested that the electrical conductivity in amorphous carbon layers with different sp2and sp3content may be due to either variablerange hopping of electrons at the Fermi level near room temperature or due to hopping in band tails, at higher temperatures. They also point out that the sp2:sp3ratio determines the dominant transport mechanism near room temperature. A high ratio allows conduction directly by percolation just between sp2regions; but a lower value ratio would lead to conduction being controlled by hopping through sp3and other non-conducting materials between nearby sp2islands. In order to distinguish between the different transport mechanisms in amorphous carbon films Godet et al [31 , 32] have studied the temperature and electric field characteristics of amorphous carbon layers. They conclude that the temperature dependence of the low field conductivity and the field dependence together may distinguish most mechanisms.

[0091] The low electric field conductivity for the data shown in Fig. 6 is given in Fig. 5, where the data is consistent with a simple exponential relationship with temperature i.e. an activated behaviour, which is consistent with nearest neighbour hopping. A thin film of low conductivity material will be very susceptible to trapped charge, since screening will be poor and trap lifetimes long. The characteristic temperature is 30K (for the fitted continuous curve) which is much smaller than would be expected if this temperature was due to single nm sized sp2regions, but would be consistent with a cluster of sp2regions together acting as an island. Such clusters would be an expected stage in the evolution towards the continuous conducting pathway as found in the low resistance wires. Hopping conduction in an organic semiconductor system has been studied by Nardes et al

[0033] , who used a film formed by quasimetallic poly3,4-ethylenedioxythiophene (PEDOT)-rich grains embedded in continuous insulating polystyrenesulfonate (PSS)lamellas. The spin process used to create the film resulted in a wider PEDOT- rich grain separation perpendicular to the film compared to the in-plane separation. The measured temperature dependent conductivity for the in-plane direction was consistent with 3D variable range hopping, whereas the conductivity in the perpendicular direction was consistent with nearest neighbour hopping. This suggests that the spacing between the sp2regions in our laser graphitized tracks is larger than in the other sp2:sp3systems studied by Dasgupta et al and by Godet et al.

[0092] In order to further understand the transport mechanism the electric field dependence of the conductivity must be considered. Nardes et al

[0033] also derive the expected electric field dependence for their nearest neighbour hopping model, obtaining a modified conductivity (Equation 1) that includes an additional term proportional to the square of the product between the electric field and the localisation length for the perpendicular direction.

[0093] Here the conductivity, a F,T), may be related to L, the average hopping length, through change with F, the applied electric field and T, the temperature.

[0094] The data shown in Fig. 4 and Fig. 5 are the steady state readings. A typical feature observable in the initial (<5) sweeps is shown in Fig. 6. In this example, the initial conductivity is high at the start voltage of - 20V, but follows an erratic pathway until about -10V, where the current jumps to a well defined s-shaped characteristic. This characteristic was maintained until +10V was reached, when the current again became erratic. As the voltage reached +20V and the sweep went back towards -20V, the current traced out a new s-shaped characteristic with much lower conductivity. In subsequent sweeps the lower conductivity characteristic was maintained and this later behaviour was used to characterise the temperature dependence.

[0095] Most models for the electric field dependence of hopping and related conduction mechanisms do not allow for the discontinuous IV behaviour shown in Fig. 6. This behaviour shows almost discrete switching events similar in character to the random telegraph events observed in low electron number semiconductor devices. In such semiconductor devices, a switching event is seen when a trap randomly fills or empties; with a fixed voltage the current through the device switches discretely between two levels provided that the time interval between events is long compared to the current measurement time. If the switching is faster than the measurement time then the current value takes a weighted average bounded by the same two levels. In the data shown in Fig. 6, both of the non-erratic parts of the characteristic are consistent with the equations from the work of Nardes et al

[0033] given earlier. By fitting these two s-shaped regions separately it is seen that the ohmic part of the conductivity (7?0) changes by a factor of almost 60 times and that the electric field constant (E!:) changes by a factor of ~ 4. In the erratic part of the sweep, the measured current is largely bounded by these the two s-shaped characteristics, as would be expected if the switching time was much shorter than the measurement time.

[0096] The origin of this behaviour seems to be linked with the disturbance of the system at the start of the sweep. The abrupt application of the -20V at the start of the sweep seems to cause a significant disturbance from the equilibrium state, which is largely relaxed during the first sweep. This disturbance is likely to change trap occupancies, not only along the percolation pathway, but also in nearby traps. As a result, it is possible that some of the resistances along the pathway are changed or that the route of the pathway is switched. Since at least some of this behaviour appears abrupt, a switching of the pathway seems to be the more likely explanation. One possible explanation for this switching behaviour is given by Patmiou et al

[0034] , who consider the effects of a Poole-Frenkel mechanism combined with percolating conduction in a numerical study. They conclude that such systems will exhibit multi-valued conductivities resulting from structural modification of the percolation pathway due to the local electric field which they call ‘percolation with plasticity’. Such structural changes are the basis for operation of some types of nonvolatile memory, but require significant time and energy to be dissipated in order to break and make chemical bonds, which is much less likely in carbon based materials compared to the materials considered by Patmiou et al. However, a similar change in percolation pathway may occur in low electron number systems just by the re-arrangement of trapped charge. If free electrons were confine to the sp2regions in these laser graphitized tracks, the small size of each region (about 1 nm) would limit excess charge to a single electron per region due to Coulomb blockade (even at room temperature). Even as a cluster of sp2regions acting as a single island, the number of free electrons would be significantly limited compared to a similar sized metal island, justifying a low electron number assumption.

[0097] In conclusion, this simple method of fabrication and the ability to engineer the conduction properties in these laser written structures allows easier implementation of devices and circuits designed to exploit this conduction mechanism. The existence of percolation pathway switching with a significant accompanying change in device resistance will enable their exploitation in novel logic and memory functions. Electronic Nose Function

[0098] Electronic Noses

[0099] In order to solve the problem of molecule identification in a simple and compact manner, a chemoresistive technology that is confined to surface operation, responds to a wide range of chemicals and provides an electrical signal that changes in a recognisable way between different chemical structures (i.e. in the direct group of molecule identifying methods) is needed. Such a device must use a thin film technology (so that transport is confined to the surface or nearby), that is chemically unreactive (to avoid poisoning effects) and has a low free carrier density (to enable the very high charge sensitivity needed to detect surface attached molecules). The thin film also needs to discriminate between different detected chemicals by allowing molecule attachment (physisorbtion) at widely separated sites for different chemicals, but close enough to the conducting channel so that a measurable change in resistance results from the occupation of a specific attachment site. The process of this molecular attachment must have an energy scale such that the site occupancy is of order ~0.5 when the surface is in thermal equilibrium with a saturated vapour of the chemical to be detected.

[0100] The present inventors have realised that many information-carrying volatiles, particularly those from biological origins, reach sizes of 20 / 30 atoms which current technologies are not equipped to identify. They have developed a new electronic device able to detect molecules based on features including their shape, in a similar way to how a biological nose works and which has functional and vast price advantages to the current ‘gold standard’, (FAIMS in combination with MS). It comprises a device, made simply using a laser treated polymer sheet, that with an applied signal is capable of giving a distinctive response in the environment of different volatile compounds across the range of biomarker sizes, where competing technologies fail. The device capabilities indicate that it could be well suited in breath analysis where it could function as a diagnostic tool, but also it holds immense possibility, due to the pervasive source of information carrying volatiles, in agri-tech fields, the food industry, animal husbandry among many others. The sensor has functional advantages but through price alone it is competitive with other technologies.

[0101] Percolation Pathway Sensing

[0102] The fabrication of these devices has been described above. In summary, conducting tracks are made using UV laser treatment of a Kapton film (the electrically insulating carbon-based material substrate) to form a two terminal resistor structure (a narrow weakly conducting section connecting two wide strongly conducting sections that are contacted using indium metal), above, Raman analysis has shown that the laser exposure decomposes the surface of the Kapton and leaves behind small (~1 .3nm) sp2regions of carbon (discrete carbon-based electrically conducting regions) buried in an amorphous carbon matrix in the form of a very thin film. The sp2regions are electrically conducting and the amorphous carbon (and the remaining Kapton) are both non-conducting, so that the resistance can be controlled through the fabrication process by choosing laser exposure conditions such that the space between the sp2regions allows for easy (highly conducting) or difficult (weakly conducting) electron transport. The laser exposure conditions can be controlled through the scan rate, the laser power and by modulation. Devices are mounted on a solid substrate with Indium metal contacts to strongly conducting regions for electrical characterisation, see Fig. 8. The weakly conducting region (which dominates the resistance of the two terminal structure) is exposed to the gas ambient via an aperture in the top of the solid encapsulation mount.

[0103] The measurement setup is shown in Fig. 9, where the mounted device under test (DUT) is contained in an enclosure, which may be under vacuum, filled with ambient air or with a controlled volatile compound vapour. Electrical connection to the DUT (2 terminals) is made through a bias Tee (HP 3315A), using a KE237 SMU to supply a DC voltage and measure the resulting current via the inductively coupled arm and an Agilent E8257D to supply an AC voltage via the capacitively coupled arm. Using only a small DC voltage (~1 mV) the device resistance was measured in a vacuum ambient over a range of temperatures (see Fig. 10) showing an exponential rise in resistance with decreasing temperature, with a characteristic temperature of 42K. This could be interpreted as either simple Arrhenius behaviour of band conduction or nearest neighbour hopping (NNH) in a strongly disordered system

[0035] . However, the unusual electric field behaviour characteristic of nearest neighbour hopping (NNH) as described by Nardes et al

[0020] was found in these devices, see above. Above described are large discrete changes in resistance (x4) when the DUT was exposed to rapid large changes in the value of the DC voltage and this behaviour was ascribed to percolation pathway switching. For this reason, great care was taken to avoid exposing the DUT to large DC voltages changes during these measurements.

[0104] The change in resistance of this DUT (when exposed to different volatile compound vapours - the gas phase analyte) is given in Table 1 . Each compound tested gave a small change in resistance (shown as a percentage with reference to the nitrogen ambient value). The small size of this resistance change suggests that the interaction with different compounds did not result in any percolation pathway switching. This is consistent with the percolation pathway being the most stable configuration except when the system is significantly disturbed, say by the sudden application of a large DC voltage. Although this approach allows the detection of a wide range of compounds (with modest changes in sensitivity between different chemicals) this would not satisfy the requirement of being a direct molecule identifying method.

[0105] Table 1 Percentage change in DC resistance for different test vapours vs the reference resistance under vacuum.

[0106] Clearly, this structure can detect a wide range of chemicals (all that have been tested), but is unable to distinguish between them when using a simple DC resistance measurement. However, the unusual transport behaviour of a hopping system when subjected to an AC excitation can be exploited in order to solve this difficulty. The theory and experimental AC behaviour of hopping systems has been widely studied previously

[0036] and suggest that the AC conductance increases with frequency in a simple manner (as expected due to the displacement currents resulting from rapid charge re-arrangement). This behaviour is characterised by a collective average of many different local charge re-arrangements with individual time constants determined by cascades of charge transfers between conducting islands, in response to the rapidly changing local potential difference. In hopping systems, these time constants are not fixed by geometry and material properties, since the determining factors, the tunnel barrier resistance and local capacitance are strongly influenced by the local potential in such nanoscale systems. Previous experimental work has reported time constants extending over the range from as short a time that is possible to measure up to hours

[0037] . This report also indicated that the relaxation of the hopping system to a rapid change in the DC applied voltage may be accelerated by an AC excitation. This method of measurement has been extensively applied to the characterisation of trapped charge in nanoscale localised electron transport devices through their resonant response [38, 39]. In a dilute localised electron transport system, a trapped charge can be treated as a 2 level system at sufficiently low temperature. Such a system may be driven to undergo a spatial Rabi oscillation when driven resonantly

[0040] , which can be detected through the change in DC conductance. But such quantum mechanical effects are washed out at high temperature (hw<kT) so that these effects will not influence the behaviour of these devices under the conditions described here.

[0107] However, the relaxation of a hopping system to a sudden change in applied voltage is a resource that can be exploited at any temperature (as long as localisation is maintained). The measurement setup is similar to that used previously

[0041] except that: the environment around the DUT (gas ambient / vacuum, temperature, pressure) can be controlled; both the DC and the AC excitation is applied to the resistor rather than separately to the source / drain and gate in a transistor structure; and the excitation waveform is not limited to a sine wave (since only the period matters when manipulating the relaxation of a hopping system, not the frequency content, as would matter for a resonant system). The devices used here are shown above to exhibit a switching behaviour similar in appearance to random telegraph switching (RTS). RTS is normally associated with low electron number systems, where single events (such as a trap nearby the conducting channel being emptied or filled due to interaction with the thermal background) results in a measurable change in resistance. As shown above, the switching occurred during a DC voltage sweep at a consistent value, whether applied positively or negatively. This must result from a single event (as required for RTS) to explain the discrete nature of the change in resistance, but this single event must be limited to only a very small part of this device. If the influence of this event were limited to the local region then the change in resistance would not be measurable in such a long device. However, if the event resulted in a re-routing of the percolation pathway, then the influence of the event is extended to a much greater fraction of the device length and so can become measurable, despite the total electron number not being low. The electron number in the section of the percolation pathway within the local region responsible for the event must be low for the event to be discrete. Low electron densities are expected at many sites within a percolation pathway undergoing hopping conduction, so that this condition is not significantly restrictive.

[0108] The behaviour of the system is changed when the frequency of the AC excitation is varied due to the wide range of time constants characterising charge re-arrangements. The time taken to relax the trapped charge is subject to a wide range of local timescales owing to the disordered nature of the material. Charge re-arrangements with time constants much shorter than the period of the AC excitation will follow the applied voltage and this effect averages to zero for the DC behaviour for a symmetric waveform. Charge re-arrangements with time constants much longer than the period of the AC excitation will not respond to the applied voltage and also have no effect. But charge re-arrangements with time constants similar to the period of the AC excitation will be enhanced and will result in a frequency dependent internally generated DC voltage due to net charge imbalances (after averaging). By choosing an applied AC amplitude that internally generates a DC voltage that is large compared to the externally applied DC voltage, the DC measured current becomes dominated by this generated DC voltage. The effect of these net charge imbalances in the changes in DC current as the AC voltage frequency changes can be seen in Fig. 11.

[0109] It is known (from the switching characteristics observed on a DC sweep, see Fig. 6) that the system can be driven out of equilibrium by an applied DC voltage, resulting in a significant (measurable) change in conductivity. During such a sweep each step change in the applied voltage (apart from the initial step) is made relatively slowly so that the system is able to relax (minimise free energy through trapped charge rearrangement) before a current measurement is taken. Fig. 7 is a schematic diagram showing a small section of percolation pathway. This shows the influence of the relaxed trapped charge, on the route of the pathway before the system was driven out of equilibrium. When the system is driven out of equilibrium, the trapped charge may be displaced, so affecting the percolation pathway and measured conductivity. A possible configuration of the (temporarily) relaxed trapped charge after the application of a large voltage is shown in Fig. 7, with a corresponding change in the percolation pathway and resistance.

[0110] As an AC frequency is applied and changed, the population of delocalised sites will change too, with a corresponding change to the measured conductivity. This effect gives rise to the change in apparent conductivity with excitation frequency.

[0111] The time needed for trapped charge re-arrangement to relax the system depends on the barrier heights to each local re-arrangement, as well as the time needed for the complex sequence of local rearrangements required to reach the new equilibrium state. It is expected that the time foreach local re-arrangement to be distributed over a very wide range as the barrier heights are widely distributed due to the disorder. A low barrier height will correspond to a high conductivity and will allow charge transport to occur on a timescale much shorter than any experimental timescale (eg the period of the AC excitation). However, as the timescale for charge transport is related exponentially to the barrier height many transfers will be delayed or blocked for larger barriers.

[0112] When the system is suddenly deviated by a large extent from the relaxed state (such as the case of the initial sweep shown in Fig. 6) this relaxation process can take seconds to complete; as seen in the timescale for the switching events. The relaxation process due to the large change in static voltage (at the start of the sweep the applied voltage jumps from 0 to 20V) comprises a complex cascade of trapped charge movements with the overall timescale dominated by the sequence of slower transitions.

[0113] In measurements using AC excitation, the DC and AC voltages are kept in a range where the AC frequency effects are able to affect a significant portion of the charge relaxation transfers as the frequency is swept. The DC voltage is kept close to zero to avoid the electric field enhanced tunnelling that leads to the s-shaped characteristic seen in Fig. 5. The AC voltage is chosen to be large enough that trapped charge transfers can take place (but would lead to a net zero current in a symmetric IV device). This leads to a different quasi-static relaxed trapped charge arrangement at each AC frequency as different charge transfers are allowed or disallowed. As a result, the percolation pathway deviates from that taken in a (small) DC only measurement, leading to a change in the measured conductivity. As the frequency changes in a combined AC and DC excitation measurement, each frequency corresponds to a new pathway and therefore a new conductivity, due to the modifications in trapped charge layout.

[0114] The DC measurement is clearly affected by the AC excitation frequency, despite the fact that the KE237 cannot measure changes in voltage at this rate (the practical DC current measurement bandwidth under these conditions is no more than a few Hz). It must be considered whether the measurement setup can contribute to these results. Substitution of a resistor of the same value as measured for the device gives a flat response (unaffected by excitation frequency). Moreover, the symmetry of the IV characteristics eliminates rectification as being the origin of this behaviour. Heating by the AC excitation is a possible cause as the device exhibits a strong temperature dependence, but the frequency dependence is then limited to simple behaviours resulting from the coupling efficiency through the bias tee. This coupling efficiency could be affected by poor impedance matching resulting in standing wave effects (regularly spaced peaks and troughs in the measured current) or by the frequency dependent coupling characteristics of the bias tee itself (which has lower and upper frequency cut-offs). But, both of these explanations are not consistent with the observed response (of irregular peaks) or to changes in RF power (not shown).

[0115] Influence of Physisorbed Molecules on the Percolation Pathway Sensor

[0116] Solvents ranging in properties and functional groups were exposed to the sensor in order to determine its capabilities. These molecules ranged in features such as MR, dipole moment, number of electron pairs, protophillic and non protophillic, organic and inorganic nature. The lack of a single, clear relationship indicates that adsorption to the channel is a multifactorial process or that a characterising feature of the molecule / moiety such as shape could be involved. By considering the response to these solvents, a number of tentative comments may be made on processes responsible, which involve the structural and chemical landscape at the surface and how it induces percolation pathway switching.

[0117] The uneven charge distribution changes conditions that drive hopping and is itself responsible for driving charge redistribution. This is complicated by the charging of dead-end branches. In addition to capacitance between islands in each pathway, between traps and branches, there also exists a capacitance between adsorbed molecules. These adsorbed molecules can exert their influence both through adsorption to the conductive island, invoking a gating effect or by changing the local permittivity, and through adsorption to the insulating barrier which may raise or lower the energy required by an electron to surpass it. A response to the applied signal itself may also influence the adsorption. A potential difference across terminals of the device results in concentrations of electric field in locations along the percolation pathway. The consequent electric field will affect the ability of the weakly bonded solvent molecules to stick or to detach from the site i.e. the adsorption process which is the phenomena to be detected itself will have some frequency dependency. These mechanisms are all in combination likely responsible for charge redistribution in the channel.

[0118] Factors governing adsorption indicated by Raman spectra and optical micrographs

[0119] The channel exhibits a high surface area which possesses cavities of complimentary shapes to the gas phase molecules. These structures would be present at a nano-scale since it is known that outgassing is a feature of the fabrication process, yet this is not visible under the optical microscope for the threshold carbonisation conditions. Dangling bonds at the periphery of the sp2bonded carbon cluster, impurities in the island and groups from the insulating gaps such as amino, phenolic and phthalimide groups, which may be polar or non-polar, may provide points at which bonding between the solvent may occur. This is supported by literature and Raman spectra which possess features characteristic of graphitic carbon in stage 2 of Andrea Ferrari’s amorphization trajectory but with the presence of a small quantity of impurities e.g. The FWHM(G), which is used as an indicator of both strain arising from doping and disorder, is ~ 80 cm1, almost double that of graphite oxide (GO) and up to 10 x larger than that of pure graphite. The Raman spectra also display a small peak present at 1787 cnr1that belongs to the C-0 stretching vibration, the C-O-C stretching vibration of the ether bridging group at 1272cm7, C-N-C stretching vibration of the imide group (along the axial direction) at 1394 cnr1(and along the transverse direction) at 1124 cnr1and 1167 cnr1, also a peak is present between 2900-2940 cnr1corresponding to CH2 groups.

[0120] Factors governing adsorption indicated by DC conductance spectra

[0121] The responses to acetone and to isopropyl alcohol are shown in Fig. 12; these two very similar molecules (the ball and stick structural diagrams are shown in in Fig. 13) give very different responses, giving larger currents and a different pattern of peaks than for the reference condition (reduced pressure air). Attachment of molecules to the amorphous carbon surface resulted in only very small changes in resistance under DC applied voltage conditions, but the changes seen under AC excitation are very considerable in the regions covered by peaks. More importantly, the peak pattern also differs significantly between that for acetone and that for isopropyl alcohol, despite their molecules’ structural similarities. In particular, six out of the nine surface atoms in isopropyl alcohol and six out of the seven surface atoms in acetone are equivalent, so that the differences between these responses must result from the three inequivalent surface atoms on the isopropyl alcohol molecule and the one inequivalent surface atom on the acetone molecule. In particular, the manner in which the inequivalent parts of the two molecules interact with the device surface must lead to the differences in the response. In the cases of acetone and IPA shown in Fig. 12, a series of ~15 peaks and dips for each can be seen in the frequency range of 200MHz. There is a close match up to ~290MHz where acetone shows a peak and IPA shows a dip (a similar dip appears for toluene). A much more prominent peak appears at ~320MHz for IPA but is almost absent for acetone. The responses for a range of molecules with a ring structure are shown in Fig. 14 (top), where some features are in common with acetone and IPA, but most are different. The responses to a further range of molecules is shown in Fig. 14 (bottom).

[0122] It is known that higher electric field gradient surfaces are best adapted for molecules containing an uneven distribution of charge

[0042] . The highly localised charge distribution existing at points along the percolation pathway might therefore be responsible for invoking a response to molecules which have a large charge distribution. Another argument towards the significance of molecule shape as a determining feature arises. Since the density of graphitic clusters in the insulating channel should decrease with distance from the central point, so the distribution and magnitudes of charge distributed across the surface should change. Thus, a range of different sized and shaped molecules are accommodated for via this sensing process.

[0123] The laser used to generate the conducting regions in the tracks causes heating which drives chemical changes that vary across the channel width, since at the central point, temperature is highest. Based upon this, it may be argued that a different chemical and dangling bond landscape will appear across this axis which may also influence the size and type of molecules able to adsorb to different points. The irradiation conditions chosen retain charge distributions within a heterogeneous region chemically and structurally. By this logic, fabrication conditions may be optimised to certain types of molecules.

[0124] Hexamethyl disilane displays one of the most positive peaks (between 250MHz and 350MHz). The Si-Si bond is known for high polarisability and the molecule has no polarity,

[0043] . It is known that a response to VOCs possessive of these two qualities indicates a high specific surface area of the channel,

[0042] . The response to molecules across the range of polarisability in the sample is not immediately clear which indicates that specific surface area is far from being the most significant determining factor towards the spectra

[0042] .

[0125] Dimethyl sulfoxide, which creates the greatest change in DC current-voltage plot through a reduced conductivity, causes the highest peak at 330MHz. This pronounced response to dimethyl sulfoxide and sulfolane could be related to the affinity of sulphur to sp2-bonded carbon, which may lead to a greater density of coverage across the channel.

[0126] Non-sulphur containing compounds seem also to be associated with increased noise as does the lower molecular weight solvents (i.e. making the comparison between water or toluene relative to hexamethyl disilane or dichlorobenzene). This may be due to rapid changes in coverage since these molecules may have a faster rate of adsorption and desorption. The increased noise also has some frequency selectivity, demonstrating that the AC signal may be promoting this process. For some VOCs, it was found that for a DC-operated sensor based on graphene channel, that noise followed different Lorenzian distributions with characteristic frequencies centred between 5Hz and 1 kHz

[0044] . Changes to the noise spectrum of the DC conductance at different AC excitation frequency may pose a method in which the volatile may be identified.

[0127] Modelling of the energy required for physisorption of the molecule to its constituent and primary materials is of interest to 1) explore the suitability of the sensor to detecting different compounds and 2) to identify and draw meaning from divergences from the these conditions. Modelling suggests that an amorphous carbon (a-C) surface is dominated by sp3sites which are chemically unreactive

[0045] . The most reactive sites are isolated C dangling bonds, but these are likely to be naturally terminated. The remaining sites are characterised by bond angles or lengths that deviate from the natural graphitic or amorphous carbon structure; as this deviation increases the chemical reactivity of the site correspondingly. This modelling shows a wide range of interaction energies (+1 to -4eV) for a range of molecular terminations, without accounting for all of the possible contributions. In practice, only a very small fraction of these sites will participate in the capture and release of molecules when in thermal equilibrium with a vapour, defined by a small range of energies of width a few kT, say of order 50meV. The dynamics of these capture and release sites depends on the local temperature and can be manipulated to some extent by using a Peltier heater / cooler unit. No direct correlation is expected between the route of the percolation pathway and the distribution of sp, sp2and sp3structures at the surface of the device. For this reason, the device needs to be long enough that sufficient ‘capture and release’ sites to encompass all of the possible structural arrangements. However, making the device too long would result in a very high resistance, which could be difficult to measure and be very noisy. The choice of device length is a compromise between these two competing requirements.

[0128] Without wishing to be bound by theory, the charge landscape and microstructure of the channel surface may be complementary to the charge landscape across an analyte molecule. Depending on this distribution of charge across space, analyte molecules will experience a different degree of affinity to the attachment sites. Adsorption may happen to dangling bonds on the sp2 bonded islands, from the amorphous carbon and from non-carbon impurities. This may alter the current flow by changing the local dielectric constant, via coulombic interactions and by changing the barrier energy.

[0129] The characteristic molecular affinity will depend on the number and position of anchor points that a molecule is able to make to the channel surface.

[0130] Any particular molecule has a well-defined shape with very specific locations on its outer surface in which these bonds can form. The carbon at the sensing device surface is highly disordered so that the dangling bonds are distributed randomly. Any one of these dangling bonds can act as a single anchor point to any shape of molecule. But a single anchor point will result in a very weak bond, allowing the molecule to detach again very quickly. Two bonds may be formed if a pair of carbon dangling bonds match the spacing of terminations on the vapour molecule. Finding a pair of carbon dangling bonds that matches one of the spacings that appear in a molecule between peripheral atoms considerably reduces the number of possible anchor sites on the amorphous carbon surface. Estimation of the number of dangling bond separations that will conform to one of the vapour molecule termination spacings is complicated by the roughness of the amorphous carbon surface. But anchoring the molecule at a pair of sites results in a much stronger bond compared to the case when only a single anchor site is formed.

[0131] When in thermal equilibrium with a given partial pressure of a particular vapour molecule, the average occupancy of a suitably spaced carbon dangling bond pair will be very much higher than for vapour molecules bonded at single sites alone. The average occupancy increases as the number of bonds formed to a single molecule increases, but suitable sites become very rare. With three or more bonds made, the molecule is much more constrained than for fewer bonds. As a result, the electrical influence of the presence of the molecule is much greater than when fewer bonds are made. This increased electrical influence and occupancy means that one can mostly disregard the many single and double bond sites in determining the effect of molecule attachment.

[0132] Finally, only attachment sites that are close to the percolation pathway and trapped charges that are susceptible to the range of AC excitation frequencies used will result in a change in the electrical measurement as a result of a change in molecule occupancy. These sites are specific to certain bond configurations and so certain combinations of sites correspond to particular molecules. As the site combinations can be addressed in frequency space of the AC excitation, it is possible to identify individual molecules by this method.

[0133] Sensing

[0134] The DC conductance spectra differences meet the information requirement for a direct method of chemical identification, or at least the inequivalent part. This is equivalent to the problem of chemical identification from measuring bond energies determined from IR absorption. Using this measurement approach gives several options for data analysis. A first approach would be to use discrete measurements at different frequencies, which could be treated in the same way as an array of different sensor devices. Since expanding the number of frequencies sampled has no upper limit, realising an array size of greater than 50 is easily done. A second approach would involve recognising the appearance and location of particular peaks and associating them with particular atomic arrangements (such as the two inequivalent sections seen in acetone and isopropyl alcohol). This second approach much more closely resembles the way in which a biological nose must work.

[0135] Information is likely also encoded within other spectral differences which appear in the form of peak ratios, shifts, FWHM, Q-factor and ‘noise’. Machine learning techniques offer a method to extract meaning from less clear relationships in the data.

[0136] Without wishing to be bound by theory, the present inventors consider that the presence of sp2 bonded carbon clusters used in the present invention is a significant factor in the operation of the sensing. Each device will have a difference micro- and nanoscale structure, and so for a given analyte vapour under test, a different DC conductance vs AC frequency response will be generated for each sensing device, even if such a difference is very slight. As such, each sensing device produced may benefit from calibration with a known analyte, set of analytes, or mixture of analytes. * * *

[0137] The features disclosed in the foregoing description, or in the following claims, or in the accompanying drawings, expressed in their specific forms or in terms of a means for performing the disclosed function, or a method or process for obtaining the disclosed results, as appropriate, may, separately, or in any combination of such features, be utilised for realising the invention in diverse forms thereof.

[0138] While the invention has been described in conjunction with the exemplary embodiments described above, many equivalent modifications and variations will be apparent to those skilled in the art when given this disclosure. Accordingly, the exemplary embodiments of the invention set forth above are considered to be illustrative and not limiting. Various changes to the described embodiments may be made without departing from the spirit and scope of the invention.

[0139] For the avoidance of any doubt, any theoretical explanations provided herein are provided for the purposes of improving the understanding of a reader. The inventors do not wish to be bound by any of these theoretical explanations.

[0140] Any section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.

[0141] Throughout this specification, including the claims which follow, unless the context requires otherwise, the word “comprise” and “include”, and variations such as “comprises”, “comprising”, and “including” will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

[0142] It must be noted that, as used in the specification and the appended claims, the singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from “about” one particular value, and / or to “about” another particular value. When such a range is expressed, another embodiment includes from the one particular value and / or to the other particular value. Similarly, when values are expressed as approximations, by the use of the antecedent “about,” it will be understood that the particular value forms another embodiment. The term “about” in relation to a numerical value is optional and means for example + / - 10%.

[0143] References

[0144] A number of publications are cited above in order to more fully describe and disclose the invention and the state of the art to which the invention pertains. Full citations for these references are provided below. The entirety of each of these references is incorporated herein.

[0145] [1] A. C. Illahi, et al, Electronic Nose Technology and Application: A Review, 2021 IEEE 13thInternational Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM), Manila, Philippines, 2021 , pp. 1-5, doi: 10.1 109 / HNICEM54116.2021 .9731890. [2] Persaud, K. & Dodd, G. Analysis of discrimination mechanisms in the mammalian olfactory system using a model nose. Nature 299, 352-355 (1982).

[0146] [3] Park, S. Y. et al. Chemoresistive materials for electronic nose: Progress, perspectives, and challenges. InfoMat 1, 289-316 (2019).

[0147] [4] Lakshminarayana, S., Park, Y., Park, H. & Jung, S. A Readout System for High Speed Interface of Wide Range Chemiresistive Sensor Array. IEEE Access 10, 45726-45735 (2022).

[0148] [5] Matatagui, D., Bahos, F. A., Gracia, I. & Horrillo, M. del C. Portable Low-Cost Electronic Nose Based on Surface Acoustic Wave Sensors for the Detection of BTX Vapors in Air. Sensors 19, 5406 (2019).

[0149] [6] Baik, J. M. et al. Tin-Oxide-Nanowire-Based Electronic Nose Using Heterogeneous Catalysis as a Functionalization Strategy. ACS Nano 4, 3117-3122 (2010).

[0150] [7] Rakow, N. A. & Suslick, K. S. A colorimetric sensor array for odour visualization. Nature 406, 710-713 (2000).

[0151] [8] Dann, J. R. A. et al. Au-Ge alloys for wide-range low-temperature on-chip thermometry. Phys. Rev. Applied 12, 034024 (2019).

[0152] [9] Noe, P. et al. Toward ultimate nonvolatile resistive memories: The mechanism behind ovonic threshold switching revealed. Science Advances 6, eaay2830 (2020).

[0153]

[0010] Snow, A. W. et al. Disordered Nanomaterials for Chemielectric Vapor Sensing: A Review. IEEE Sensors Journal 15, 1301-1320 (2015).

[0154]

[0011] Karpov, V. G., Serpen, G. & Patmiou, M. Percolation with plasticity for neuromorphic systems. J. Phys. Complex. 1 , 035009 (2020).

[0155]

[0012] Lin, J. et al. Laser-induced porous graphene films from commercial polymers. Nat Commun 5, 5714 (2014).

[0156]

[0013] Schumann, M., Sauerbrey, R. & Smayling, M. C. Permanent increase of the electrical conductivity of polymers induced by ultraviolet laser radiation. Appl. Phys. Lett. 58, 428-430 (1991).

[0157]

[0014] Phillips, H. M., Feurer, T., Blanc, S. P. L., Callahan, D. L. & Sauerbrey, R. A. Excimer-laser- induced permanent electrical conductivity and nanostructures in polymers, in Laser Radiation Photophysics vol. 1856 143-154 (SPIE, 1993).

[0158]

[0015] Romero, F. J. et al. In-Depth Study of Laser Diode Ablation of Kapton Polyimide for Flexible Conductive Substrates. Nanomaterials 8, 517 (2018).

[0159]

[0016] Srinivasan, R., Hall, R. R., Wilson, W. D., Loehle, W. D. & Allbee, D. C. Ultraviolet laser irradiation of the polyimide, PMDA-ODA (Kapton™), to yield a patternable, porous, electrically conducting carbon network. Synthetic Metals 66, 301-307 (1994).

[0017] Biswas, R. K., Farid, N., O’Connor, G. & Scully, P. Improved conductivity of carbonized polyimide by CO2 laser graphitization. J. Mater. Chem. C 8, 4493-4501 (2020).

[0160]

[0018] Davenas, J. Laser and ion beam processing of conductive polyimide. Applied Surface Science 36, 539-544 (1989).

[0161]

[0019] Raffel, J. I., Freidin, J. F. & Chapman, G. H. Laser-formed connections using polyimide. Applied Physics Letters 42, 705 (1998).

[0162]

[0020] Srinivasan, R., Hall, R. R. & Allbee, D. C. Generation of electrically conducting features in polyimide (KaptonTM) films with continuous wave, ultraviolet laser radiation. Appl. Phys. Lett. 63, 3382-3383 (1993).

[0163]

[0021] Decker, C. An original method for producing polymer micro-circuits by laser irradiation. Makromolekulare Chemie. Macromolecular Symposia 24, 253-263 (1989).

[0164]

[0022] Vivaldi, F. M. et al. Three-Dimensional (3D) Laser-Induced Graphene: Structure, Properties, and Application to Chemical Sensing. ACS Appl. Mater. Interfaces 13, 30245-30260 (2021).

[0165]

[0023] Virk, H. S., Chandi, P. S. & Srivastava, A. K. Physical and chemical response of 70 MeV carbon ion irradiated Kapton-H polymer. Bull Mater Sci 24, 529-534 (2001).

[0166]

[0024] Vashisth, A. et al. ReaxFF Simulations of Laser-Induced Graphene (LIG) Formation for Multifunctional Polymer Nanocomposites. ACS Appl. Nano Mater. 3, 1881-1890 (2020).

[0167]

[0025] Plis, E. A. et al. Review of Radiation-Induced Effects in Polyimide. Applied Sciences 9, 1999 (2019).

[0168]

[0026] Zhao, P. etal. Replacing the Metal Electrodes in Triboelectric Nanogenerators: High- Performance Laser-Induced Graphene Electrodes. Nano Energy 75, (2020).

[0169]

[0027] Wang, L., Wang, Z., Bakhtiyari, A. N. & Zheng, H. A Comparative Study of Laser-Induced Graphene by CO2 Infrared Laser and 355 nm Ultraviolet (UV) Laser. Micromachines 11 , 1094 (2020).

[0170]

[0028] 3ttechnologies.com

[0171]

[0029] Canpado, L. G. et al. General equation for the determination of the crystallite size La of nanographite by Raman spectroscopy. Appl. Phys. Lett. 88, 163106 (2006).

[0172]

[0030] Dasgupta, D., Demichelis, F. & Tagliaferro, A. Electrical conductivity of amorphous carbon and amorphous hydrogenated carbon. Philosophical Magazine B 63, 1255-1266 (1991).

[0173]

[0031] Godet, C. Hopping model for charge transport in amorphous carbon. Philosophical Magazine B 81 , 205-222 (2001).

[0174]

[0032] Godet, C., KumarJ, S. & Chu, V. Field-enhanced electrical transport mechanisms in amorphous carbon films. Philosophical Magazine 83, 3351-3365 (2003).

[0033] Nardes, A. M., Kemerink, M. & Janssen, R. A. J. Anisotropic hopping conduction in spin-coated PEDOT:PSS thin films. Phys. Rev. B 76, 085208 (2007).

[0175]

[0034] Patmiou, M., Niraula, D. & Karpov, V. G. The Poole-Frenkel laws and a pathway to multi-valued memory. Appl. Phys. Lett. 115, 083507 (2019).

[0176]

[0035] Stallinga, P. Electronic Transport in Organic Materials: Comparison of Band Theory with Percolation / (Variable Range) Hopping Theory. Advanced Materials 23, 3356-3362 (2011).

[0177]

[0036] Schroder, T. B. & Dyre, J. C. ac Hopping Conduction at Extreme Disorder Takes Place on the Percolating Cluster. Phys. Rev. Lett. 101 , 025901 (2008).

[0178]

[0037] Ovadyahu, Z. Long-range influence of manipulating disordered-insulators locally. Phys. Rev. B 101 , 094202 (2020).

[0179]

[0038] Bian, X., Milne, W. I. & Hasko, D. G. Fast and sensitive detection of an oscillating charge. Applied Physics Letters 107, 022108 (2015).

[0180]

[0039] Tenorio-Pearl, J. O., Herbschleb, E. D., Creatore, C., Chin, A. & Oda, S. Coherent control of trapped-charge induced resonances in a field-effect transistor, in 2016 IEEE Silicon Nanoelectronics Workshop (SNW) 28-29 (2016). doi:10.1109 / SNW.2016.7577969.

[0181]

[0040] Stafford, C. A. & Wingreen, N. S. Resonant Photon-Assisted Tunneling through a Double Quantum Dot: An Electron Pump from Spatial Rabi Oscillations. Phys. Rev. Lett. 76, 1916-1919 (1996).

[0182]

[0041] Creswell, L. A., Hasko, D. G. & Williams, D. A. Microwave excitation of localized electrons in phosphorus-doped silicon single electron transistors. Journal of Applied Physics 105, 104508 (2009).

[0183]

[0042] Gadipelli, S. & Guo, Z. X. Graphene-based materials: Synthesis and gas sorption, storage and separation. Progress in Materials Science 69, 1-60 (2015)

[0184]

[0043] Razuvaev, G. A., Brevnova, T. N. & Semenov, V. V. The Oxidative Cleavage of the Silicon- Silicon Bond. Russ. Chem. Rev. 55, 606-621 (1986).

[0185]

[0044] Rumyantsev, S., Liu, G., Shur, M. S., Potyrailo, R. A. & Balandin, A. A. Selective Gas Sensing with a Single Pristine Graphene Transistor. Nano Lett. 12, 2294-2298 (2012).

[0186]

[0045] Caro, M. A., Aarva, A., Deringer, V. L., Csanyi, G. & Laurila, T. Reactivity of Amorphous Carbon Surfaces: Rationalizing the Role of Structural Motifs in Functionalization Using Machine Learning. Chem Mater 30, 7446-7455 (2018).

Claims

Claims:1 . A method of sensing, comprising the steps of: exposing a sensing device to a gas phase analyte; and subjecting the sensing device to AC excitation and measuring a DC conductance of the sensing device; wherein: the sensing device comprises discrete carbon-based electrically conducting regions on a surface of an electrically insulating carbon-based material substrate, such that electrical conduction through the sensing device is by percolation between the electrically conducting regions; and interaction between the gas phase analyte and the discrete carbon-based electrically conducting regions affects the DC conductance to enable sensing of the analyte.

2. The method of claim 1 , wherein the conducting regions comprise graphitic carbon.

3. The method of claim 1 or claim 2, wherein the electrically insulating carbon-based material substrate comprises an aromatic polymer.

4. The method of any one of claims 1 to 3, wherein the sensing devices demonstrates a room temperature non-linear current-voltage relationship.

5. The method of any one of claims 1 to 4, wherein the sensing device has an aspect ratio of length:width, wherein: the length of the sensing device is a dimension that is parallel to a conduction direction; and the width of the sensing device is a dimension that is perpendicular to the length and is parallel to a plane that is tangential to the surface that comprises the electrically conducting regions; wherein the aspect ratio has a value greater than or equal to 2:1 .

6. The method of any one of claims 1 to 5, wherein the conducting regions are contained in a channel.

7. The method of claim 6, wherein the density of conducting regions in the channel is highest at a centre of the channel width-wise.

8. The method of any one of claims 1 to 7, wherein the sensing device comprises amorphous carbon between the conducting regions.

9. The method of any one of claims 1 to 8, wherein the conducting regions are heterogeneously distributed.

10. The method of any one of claims 1 to 9, wherein the sensing device comprises heterogeneously distributed attachment points for analyte binding.

11. The method of any one of claims 1 to 10, wherein the AC excitation voltage is in the range 0.1 to 10 V.

12. The method of any one of claims 1 to 11 , wherein the DC voltage is in the range 0.1 to 10mV.

13. The method of any one of claims 1 to 12, further comprising repeating the measurement of DC conductance as the AC excitation frequency is varied to generate a spectrum.

14. The method of claim 13, further comprising the step of analysing the spectrum.

15. The method of any one of claims 1 to 14, further comprising a calibration step, wherein the DC conductance of an individual sensing device when exposed to a known gas phase analyte under AC excitation is measured.

16. The method of any one of claims 1 to 15, wherein the sensing is performed at ambient temperatures.

17. The method of any one of claims 1 to 16, wherein the temperature and / or pressure are modulated to adjust a degree of adsorption of the analyte to the sensing device.

18. A sensing device configured to be used in the method of any one of claims 1 to 10.

19. A method of manufacturing the sensing device of claim 18, wherein the conducting regions are formed using localised heating on the surface of the electrically insulating carbon-based material substrate.

20. The method of claim 19, wherein the localised heating is by laser light.

21. The method of claim 19 or claim 20, wherein the conducting regions are formed by localised heating on the surface of the electrically insulating carbon-based material substrate using laser light of wavelength that is greater than or equal to 360nm and less than or equal to 480nm.

22. The method of any one of claims 19 to 21 , wherein the conducting regions are formed by localised heating on the surface of the electrically insulating carbon-based material substrate using laser light of power that is greater than or equal to 14mW and less than or equal to 70mW.

23. The method of any one of claims 19 to 22, wherein the conducting regions are formed by localised heating on the surface of the electrically insulating carbon-based material substrate using laser light with a scan rate that is greater than or equal to 25 mms1and less than or equal to 175 mms'1.

24. The method of any one of claims 19 to 23, wherein the conducting regions are formed by localised heating on the surface of the electrically insulating carbon-based material substrate using laser light with a duty ratio that is greater than or equal to 1 :2 and less than or equal to 1 :1.

Images