Method for detecting volatile compounds and a universal multisensor platform "omniolfactor" based on a functionalized graphene framework

Abstract

FIELD: chemistry.SUBSTANCE: invention is intended for detection and identification of volatile compounds. A sensor platform for detecting and identifying volatile compounds comprises a housing with inlet and outlet gas channels and a measuring circuit, wherein the sensor platform comprises a sensor matrix made in the form of an array of individual cells, each of which is an independent sensitive element based on a monolithic block of a graphene framework with a highly developed specific surface area, wherein the matrix cells are functionalized with various receptor groups selective to different classes of target molecules, and the measuring circuit is configured for multiplexed reading of the electrical parameters of each cell, wherein the monolithic block is made with through channels that provide forced pumping of the gas medium through its volume.EFFECT: creating a portable platform that combines laboratory sensitivity and selectivity with a rapid response, as well as long-term stability.11 cl, 3 dwg, 5 tbl

Description

[0001] The present invention relates to the field of analytical instrumentation and nanotechnology, specifically to means for detecting and identifying trace amounts of explosives, toxic gases, volatile organic compounds, and biomarkers in gas environments. The invention can be used in security systems, including airports, customs terminals, and emergency response units; in environmental monitoring for air pollution control; in medical diagnostics for the creation of non-invasive exhaled air analyzers; in the food industry for monitoring product freshness; in forensic examination; and in the creation of odor and emotional state recognition systems known as electronic noses.

[0002] The state of the art in the field of gas analytics is characterized by the presence of many methods and devices, each of which has its own advantages and disadvantages, but none of the existing solutions simultaneously provides high sensitivity, selectivity, speed, portability and long-term stability in the field.

[0003] Chromatograph-mass spectrometry, known from US Patent 7,123,345 B2, is a laboratory method based on a combination of chromatographic separation of gas mixture components followed by mass spectrometric identification and quantitative analysis. This method provides high sensitivity down to 0.1-1 ppb and exceptional selectivity. However, its main drawbacks include its bulkiness, high cost, and significant energy consumption, as well as the need for a qualified operator and laboratory conditions. This makes this method completely unsuitable for the creation of portable rapid systems or networked sensors for field use.

[0004] The ion mobility method described in US Patent 6,495,823 B1 is based on the separation of ionized molecules in an electric field based on their mobility in a carrier gas, enabling the identification of substances with a relatively fast response time of 2 to 10 seconds. The disadvantages of this method include the need for an ionization source, often radioactive or corona discharge-based, which complicates the design and creates regulatory restrictions. It also has moderate selectivity and high sensitivity to humidity and other interfering factors, reducing the reliability of measurements under real-world operating conditions.

[0005] Optical methods, particularly infrared spectroscopy, known from US Patent 8,247,761 B2, are based on the detection and identification of molecules by their characteristic absorption bands in the infrared range. These methods provide instantaneous response and high selectivity for functional groups, but they have low sensitivity to trace concentrations at the ppm level, which is orders of magnitude lower than that required for detecting explosives and biomarkers. They are critically dependent on atmospheric transparency and require complex and expensive optomechanical systems that are difficult to miniaturize.

[0006] Chemical resistive sensors based on carbon nanotubes, known from US Patent No. 8,748,143 B2, use a network or layer of carbon nanotubes as a sensing element, detecting changes in conductivity upon gas adsorption. These sensors have a sensitivity of 1-10 ppb but require high operating temperatures of 150 to 300 degrees Celsius to activate and accelerate kinetics, which increases energy consumption. They exhibit low selectivity and incomplete reversibility, are highly susceptible to atmospheric moisture, and the process of manufacturing reproducible sensors is complex and expensive.

[0007] Graphene oxide-based sensors, described in patent CN 105548152 A, are designed to detect peroxide explosives by reducing graphene oxide, altering its conductivity. The disadvantages of these sensors include low chemical and thermal stability of the material, susceptibility to uncontrolled structural changes, slow process kinetics, and often an irreversible or partially reversible response, which precludes repeated use without replacing the sensing element.

[0008] Multisensor systems, known as electronic noses and described in US patents 6,085,576 A and 7,132,124 B2, use a combination of sensors of different types, including metal oxide, polymer, and electromechanical, to form a unique fingerprint of a complex gas mixture and then process the data. The main drawback of such systems is that they consist of technologically incompatible elements, leading to difficult calibration, low reproducibility, and significant reading drift. The lack of a unified technological platform increases cost, size, and reduces reliability.

[0009] Metal oxide semiconductor sensors, known from US Patent 6,085,576 A, are based on changes in the conductivity of a metal oxide layer, such as SnO2 or WO3, during gas chemisorption and surface catalytic reactions and are widely used for detecting CO, CH4, and ethanol. Their disadvantages include the need for constant high power consumption to maintain an operating temperature of 200 to 400 degrees Celsius, very low selectivity, and susceptibility to significant baseline and sensitivity drift under the influence of atmospheric conditions and time.

[0010] The electrochemical sensors described in US Patent 5,145,645 A are designed to detect gases such as CO, H2S, and NOₓ by generating a measurable current through an electrochemical reaction at electrodes in a liquid or solid electrolyte. A disadvantage of these sensors is the presence of a liquid electrolyte, which limits their service life, spatial orientation, and resistance to temperature fluctuations, as well as their limited service life and gradual degradation due to drying out or degrading of the electrolyte.

[0011] Antibody- or enzyme-based biosensors, known from US Patent No. 8,114,622 B2, use highly selective biological receptors immobilized on the transducer surface to detect specific biomarkers. The extremely low stability of biological components outside strictly controlled conditions of temperature, humidity, and pH makes them unsuitable for long-term use in field devices. Furthermore, they are characterized by slow response and difficulty in regeneration.

[0012] Quartz crystal microbalances with polymer coatings, known from US Patent 4,742,437 A, measure the change in the resonant frequency of a quartz crystal during the sorption of analyte vapor into a selective polymer coating. These devices have very low sensitivity in the gas phase, are extremely sensitive to changes in temperature and humidity, and the polymer coatings are short-lived and susceptible to nonspecific sorption.

[0013] Sensors based on conductive polymers, described in US Patent 7,132,124 B2, detect changes in the conductivity of polymer films such as polyaniline when exposed to gases such as NH3 or H2S. Polymers are subject to irreversible chemical and physical degradation during long-term operation or when exposed to environmental conditions, including oxygen and ultraviolet radiation, leading to rapid drift and loss of sensitivity, and are also characterized by low reproducibility from sample to sample.

[0014] A special place in the state of the art is occupied by an artificial composition known as graphene pumice, described in patent RU 2550176 C2. Graphene pumice is a composite of parallel-oriented graphene sheets bound at the edges by amorphous carbon, with a density of 0.008 to 0.3 grams per cubic centimeter and a specific surface area reaching 2000 square meters per gram or more. A key feature of this material is contact chaos, i.e., random but strong adhesion of clusters at contact points without blocking the main pore channels. This provides a combination of high specific surface area, open porosity for ion transport, and mechanical and thermal stability.The material's specific surface area is formed by hierarchical porosity, which includes three main contributions: the surface area from the parallel sheets, the additional area from developed micropores smaller than 2 nanometers in size within and between clusters, and the area from the nanopores of the binders and the corrugation of the graphene sheets themselves. This material is synthesized using low-temperature self-propagating high-temperature synthesis directly as a monolith of a given shape, eliminating the need for binders or additional molding. It features high mechanical strength, hydrophobicity, electrical conductivity, chemical and thermal resistance up to 900 degrees Celsius in air, and low cost.

[0015] The closest analogue to the proposed invention are chemical resistive sensors based on graphene nanosheets, known from US Patent 9,291,667 B2, which record changes in the electrical resistance of a thin layer of graphene nanosheets upon adsorption of gas molecules. Such sensors are potentially miniaturized, but have insufficient sensitivity at the 10-50 ppb level due to the limited specific surface area of ​​the material (less than 500 square meters per gram) and the lack of optimal hierarchical porosity. Furthermore, they are characterized by low selectivity, slow response and desorption kinetics, strong susceptibility to parameter drift, and irreversible poisoning of active sites, leading to instability and a short service life.

[0016] Thus, the unsolved technical problem remains of the lack of a universal analytical platform that would combine laboratory sensitivity and selectivity with portability, long-term stability, the ability to detect a wide range of substances and low operating costs.

[0017] The technical problem solved by this invention is to overcome the fundamental contradiction of modern gas analytics between high analytical characteristics, including sensitivity, selectivity and stability, inherent in stationary laboratory instruments such as chromatograph mass spectrometers and optical spectrometers, and the requirements of portability, fast response, autonomy and low operating costs imposed on field and embedded systems for security, medicine and environmental monitoring.The invention aims to address key shortcomings of existing portable chemical sensors, including insufficient sensitivity due to a small contact area with the analyte, irreversible degradation and poisoning of active sites, low selectivity in complex gas mixtures, and the lack of a universal technological platform capable of flexibly adapting to the detection of various classes of substances, from explosive compounds and toxic gases to disease biomarkers in exhaled air.

[0018] These goals are achieved by creating a universal multisensor platform, the core of which is an intelligent matrix of individual monolithic cells of functionalized graphene pumice with a built-in thermal regeneration system and data processing based on artificial intelligence.

[0019] The proposed invention is a sensor platform for the detection and identification of volatile organic compounds, including explosives, toxic gases, and biomarkers. It comprises a housing with inlet and outlet gas channels and a measuring circuit. The platform is based on a sensor matrix consisting of an array of individual cells, each of which is an independent sensing element based on a monolithic graphene framework block with a highly developed specific surface area. The matrix cells are functionalized with various receptor groups selective for different classes of target molecules. The measuring circuit enables multiplexed reading of the electrical parameters of each cell. The monolithic graphene framework block is designed as a three-dimensional structure with through-channels that ensure forced circulation of the gaseous medium throughout its entire volume.

[0020] The graphene framework in cells is a composite of parallel-oriented graphene sheets bound at the edges by amorphous carbon, with a density of 0.008 to 0.3 grams per cubic centimeter. The density of the various cells in the matrix can be varied to optimize for specific applications. The monolithic block is designed as a three-dimensional framework with through channels with diameters ranging from 0.1 to 2 millimeters, which ensures forced gas flow throughout its entire volume and dramatically reduces response time. The through channels are formed directly during the synthesis stage using removable cores in a mold, allowing for the creation of a regular, strictly controlled system of macrochannels, unlike the chaotic macroporosity typical of known materials.

[0021] To ensure selectivity, the matrix cells are functionalized with one or more receptor types from a wide range of receptors, including metal complexes, amino and thiol groups, metal and metal oxide nanoparticles, conductive polymers, antibodies, enzymes, molecular mimics, macrocyclic compounds, and lipid membranes. This diverse functionalization creates a unique multidimensional response to various gases and gas mixtures, enabling the formation of characteristic fingerprints for each analyte.

[0022] The platform's key feature is a heating control unit capable of passing an electric current through the electrodes of selected cells sufficient to resistively heat the graphene framework to temperatures ranging from 300 to 880 degrees Celsius for thermal cleaning or thermal cycling. The heating control unit and the measuring circuit use the same electrode structures, alternately switching between measuring and heating modes. The control unit is capable of programmable thermal cycling according to specified time and temperature profiles, including pulsed heating to 880 degrees Celsius for deep cleaning and maintaining an operating temperature of 80 to 120 degrees Celsius for optimal sorption.

[0023] To ensure correct operation of the platform when using the same electrode structures in both measurement and heating modes, time-division multiplexing is used, taking into account the thermal inertia of the graphene framework. The heating control unit and measurement circuit are connected to the electrodes via high-speed analog switches, which provide alternating switching at a frequency of 10 to 1000 Hz, depending on the required temperature control accuracy.

[0024] During the heating phase (lasting from 1 ms to 1 s), a current pulse of up to 10 A is passed through the electrodes, resistively heating the graphene monolith to the set temperature. During the measurement phase (lasting from 1 to 100 ms), the heating circuit is completely disconnected, and a precision measuring circuit is connected to the same electrodes. This circuit records the cell resistance in the absence of heating current, eliminating the influence of voltage drop across parasitic resistances and thermal EMF. Due to the high thermal conductivity of the graphene framework and its significant heat capacity, the monolith temperature does not change significantly during the measurement pause (milliseconds) (the temperature drop does not exceed 1-5°C), allowing the set temperature profile to be maintained with high accuracy while simultaneously obtaining reliable data on the electrical parameters of the cells.This approach also makes it possible to implement a pulsed heating mode for thermolabile receptors, when short-term heating of the surface occurs faster than heat transfer into the depth of the sensitive layer.

[0025] Data processing unit and artificial intelligence architecture

[0026] The data processing unit is the computing core of the multisensor platform and is implemented on a low-power 32-bit microcontroller with ARM Cortex-M4F or RISC-V architecture, equipped with a floating-point unit and, depending on the configuration, a hardware neural network accelerator. Unlike traditional approaches based on threshold comparison of individual sensor signals, the proposed solution implements a multi-level digital processing pipeline, including preprocessing, dimensionality reduction, and machine learning stages.

[0027] The multidimensional signal is generated by multiplexing all cells of the sensor matrix at a frequency of 10 to 100 Hz. For each cell, not only the absolute value of electrical resistance is recorded, but also a vector of kinetic parameters, including the rate of resistance change during adsorption and desorption, as well as the time to reach equilibrium. This approach allows for the formation of a dynamic portrait of the gas mixture with a dimension of 32 to 64 features, updated every 10-100 milliseconds.

[0028] The preprocessing stage includes digital filtering of high-frequency noise using median and moving average filters, compensation for temperature and humidity drift based on reference and empty cell signals, and data normalization to a uniform scale. Additionally, corrections are made based on readings from the built-in temperature, pressure, and relative humidity sensors.

[0029] Feature space dimensionality reduction is achieved using two alternative methods. The classic approach uses a principal component algorithm, which allows real-time projection of multidimensional data onto a plane of two or three principal components for visualization and preliminary evaluation. A more advanced method is based on the use of a neural network autoencoder, which nonlinearly compresses the input vector into a lower-dimensional latent space and then restores it, thereby extracting the most informative features.

[0030] Classification and identification of gas mixtures is performed using an ensemble of machine learning methods. A multilayer perceptron with two or three hidden layers or a support vector machine with a radial basis kernel is used to detect known substances included in the reference library. For anomaly detection and unknown compound detection, a single-class support vector machine or a Gaussian mixture model is used, which captures the deviation of the current chemical fingerprint from the reference state of a clean environment. The classification result is a substance identifier, its estimated concentration in the range from 0.01 to 100 ppb, and a confidence score as a percentage.

[0031] The platform supports two main operating modes for the artificial intelligence unit. In inference mode, the primary operating mode, the trained model performs only forward signal propagation with fixed weighting coefficients, ensuring minimal power consumption. In calibration mode, activated upon connection to the base station or by operator command, the model is retrained using a set of reference gas mixtures, allowing the platform to be adapted to specific operating conditions and compensate for the natural aging of sensor cells.

[0032] The resulting data is transmitted to the user interface, recorded in an internal non-volatile log with timestamps, and can be sent wirelessly to centralized monitoring systems for further analysis and decision making.

[0033] The detection method using this platform includes passing the analyzed gaseous medium through a sensor matrix, simultaneously or sequentially reading the electrical responses of all cells, generating a multidimensional response vector and comparing it with a library of reference materials or classifying it using a trained machine learning model to determine the presence and concentration of target compounds. The method also includes a step of periodic or program-controlled thermal cleaning of the sensor matrix by resistively heating the graphene framework to temperatures ranging from 500 to 880 degrees Celsius for 1 to 30 seconds. The measurement and heating modes are performed cyclically, with the measurement phase being conducted at a temperature of 80 to 120 degrees Celsius and the cleaning phase at a temperature of 500 to 880 degrees Celsius.

[0034] The method for manufacturing a sensor cell includes forming a graphene framework monolith of a given shape and density by self-propagating high-temperature synthesis in a press mold with removable cores to create through channels, subsequent mechanical processing of the monolith to create fastening elements and seats for electrodes, applying electrode structures to the surface or inside the monolith and functionalizing the surface of the monolith by impregnation with solutions of receptor compounds, followed by drying and heat fixation.

[0035] The fundamental difference between the proposed platform and existing solutions, including graphene pumice according to patent RU 2550176 C2, is the presence of a regular system of through-channels in each monolithic block, formed during the synthesis stage using removable cores in a mold. In the known material, macropores form randomly as a result of gas evolution during synthesis and cannot be precisely controlled in diameter, direction, or quantity. This structure is suitable for static adsorption but is unable to ensure forced gas flow through the entire volume with low pneumatic resistance and predictable contact time. In contrast, the proposed system of through-channels with a predetermined diameter of 0.1 to 2 mm transforms the monolith from a passive porous medium into a highly efficient flow reactor.Gas is forced through the channels by a pressure differential, delivering analyte molecules directly into the bulk of the material, where they interact with the functionalized surface. This ensures the use of the entire available surface area of ​​the graphene framework, not just the surface layer, and reduces the response time to 0.5-5 seconds, which is unachievable for existing porous materials without a system of regular macrochannels.

[0036] Functionalizing additives selected from a number of metal salts, organic ligands with functional groups, macrocyclic compounds or nanodispersed metal oxides can be additionally introduced into the reaction mixture for synthesizing the graphene framework, which during the synthesis process form active centers uniformly distributed throughout the volume of the monolith that are selective to the target molecules.

[0037] For thermolabile receptor groups unstable at the temperature of self-propagating high-temperature synthesis, functionalization is performed after monolith formation by immersion impregnation, spraying, or coating with a solution containing the specified groups, followed by drying and fixation at a temperature not exceeding 80 degrees Celsius, which ensures the preservation of their chemical or biological activity. This combined approach enables basic selectivity to be achieved during the synthesis stage by introducing thermostable additives, while fine-tuning the cells for specific analytes is achieved by the subsequent gentle application of specialized receptors.

[0038] The platform also implements the principle of indirect detection of ionizing radiation sources by analyzing volatile chemical markers formed through environmental radiolysis processes or characteristic of specific chemical forms of radioactive elements. When atmospheric air is exposed to ionizing radiation, active radicals are formed, which quickly recombine to form stable marker molecules, including ozone and nitrogen oxides, which are detected by matrix cells functionalized with metal oxide nanoparticles such as SnO2 or WO3. To detect specific volatile forms of radionuclides, such as iodides, cells with silver nanoparticles are used. These form insoluble iodides and alter electrical resistance.The combination of cells sensitive to different classes of markers in a single matrix allows not only to record the presence of a radiation anomaly, but also to identify it, distinguishing neutron flux from gamma radiation or detecting the leakage of fissile materials.

[0039] The technical result of the invention is the creation of a universal multisensor platform that provides an unprecedented combination of extremely high sensitivity at the level of 0.01-0.1 ppb for most target compounds due to the use of graphene framework monoliths with a highly developed specific surface area and volumetric sorption, as well as a fast response time of 0.5-5 seconds due to the hierarchical porosity of the material and a system of through macro-channels for forced gas pumping.This achieves universal and adaptive selectivity, enabling the simultaneous detection and identification of dozens of classes of substances, including explosives, toxic gases, volatile organic compounds, and biomarkers, through the combination of cells with various targeted functionalizations in a matrix and subsequent intelligent processing of the multidimensional response using machine learning algorithms. High long-term stability with a drift of less than one percent per month and a service life of over five years is ensured by a built-in thermal regeneration system, using Joule heating up to 880 degrees Celsius through the measuring electrodes for complete desorption of the analyte and restoration of active sites.The platform is portable and autonomous, weighing between 0.1 and 2 kilograms and consuming between 0.1 and 5 watts in standby mode. It is cost-effective, with analysis costs orders of magnitude lower than those of laboratory methods. Its modular design with replaceable cartridges also provides the unique ability to detect unknown or unaccounted-for anomalies by detecting the slightest deviations in the integral chemical fingerprint of the monitored environment from its reference state.

[0040] The present invention is illustrated by graphical materials featuring three-dimensional models created in the OpenSCAD environment and visualizing the platform design in a section for a clear display of the internal structure, as well as demonstrating a full eight-second operating cycle consisting of three phases: gas compound detection for three seconds, thermal regeneration for three seconds, and an annotated circuit for two seconds. The main body of the device is a rectangular block measuring 120 by 60 by 80 millimeters, made of a metal alloy, with a removable cartridge containing a 4 by 4 sensor matrix consisting of sixteen individual cells, each of which is a monolithic block of a graphene framework with a system of through channels. The electrodes are designed as interdigitated structures integrated into the monoliths.The device is connected to gas channels, including an inlet and outlet, as well as wiring for measurements and heating power. The model is animated and visualizes key processes, including the flow of the analyzed gas through the channels, the adsorption of molecules on the functionalized surface, changes in the electrical parameters of the cells, Joule heating for thermal regeneration, and data processing by a processing unit with machine learning algorithms. Display elements are located around the model, including three LEDs for various operating modes, a legend with dynamically updated parameters, and technical specifications.

[0041] In a static, neutral state, the model is shown in cross-section with animation effects disabled and supplemented with numbered markers indicating key design components. The left side features a legend describing the main metal body, removable cartridge with sensor matrix, individual cells based on a monolithic graphene framework, a system of through gas channels, comb electrodes for measurements and Joule heating, inlet and outlet gas channels, contact pads for electrical measurements, signal and power wiring, a processing unit with machine learning algorithms for data processing, and a thermal regeneration system. The right side of the legend lists key technical specifications, including sensitivity, response time, regeneration temperature, operating temperature, power consumption, and weight.

[0042] At various points in the working cycle, the model demonstrates changes in the state of the cells, displaying the beginning of detection with the appearance of a weak blue glow and the first gas molecules, active detection with a pulsating bright blue glow and an intense flow of molecules, the end of detection with a decrease in the intensity of the glow and a slowdown in the movement of molecules, the beginning of thermal regeneration with the appearance of an orange-red glow and the activation of heating elements, active regeneration with a bright glow and thermal waves, the end of regeneration and cooling with an uneven dimming glow and the appearance of a green ready LED.

[0043] A key feature of the platform's implementation is a differentiated approach to cell thermal cycling depending on the type of their functionalization. Cells functionalized with thermally stable receptors, such as metal complexes, metal nanoparticles, and metal oxides, are designed to withstand repeated high-temperature exposure and undergo a full regeneration cycle, including heating to 880°C to remove the most persistent contaminants. A different strategy is used for cells containing thermolabile biological receptors, including enzymes, antibodies, and lipid membranes, which remain active only at temperatures up to 80°C. These cells are either housed in a separate replaceable cartridge that is not subjected to high-temperature annealing and is replaced when their service life is exhausted, or they are protected from thermal exposure by utilizing the thermal inertia effect.In the latter case, a short current pulse lasting 1-3 seconds predominantly heats the graphene framework, which has high thermal conductivity, while the thin surface layer of the bioreceptors, due to its small thickness and low thermal conductivity, does not have time to reach the critical degradation temperature. These modes are implemented programmatically by the heating control unit based on preset profiles for each cell type.

[0044] The stated average power consumption range of 0.1 to 5 watts using pulse currents up to 10 amps is achieved by an asymmetrical cyclic operating mode, in which energy-intensive heating phases occupy a vanishingly small fraction of the device's overall operating time. The relationship between peak and average parameters is quantitatively described by the fundamental relationship for pulse processes:

[0045]

[0046] where D is the duty cycle, defined as the ratio of the pulse duration to the repetition period:

[0047]

[0048] Let's consider the most energy-intensive mode of those presented in Table 2, namely, deep regeneration mode. According to the given parameters, a current of up to 10 amps is passed through the electrodes of a cell with a typical resistance of 2-5 ohms for 3-10 seconds. According to the Joule-Lenz law, the peak power released in a single cell reaches:

[0049]

[0050] However, according to the same table, this mode is activated only once every 50–100 measurement cycles. With a detection cycle duration of \tau_{\text{det}} = 3 seconds, the time interval between deep regeneration pulses is:

[0051]

[0052] The energy spent on one regeneration pulse with a duration of t_{\text{imp}} = 5 seconds is equal to:

[0053]

[0054] Then the average power consumed by one cell over the entire period between regenerations is:

[0055]

[0056] The duty cycle for this mode is D = 0.0167\ (1.67%), which is fully consistent with the fundamental relationship P_{\text{avg}} = P_{\text{peak}} \times D.

[0057] It's important to emphasize that this calculation represents the maximum estimate for a single cell operating under the most severe thermal conditions. Thanks to the implementation of smart thermal cycling principles, heating pulses are distributed over time and addressed only to those cells that actually require regeneration at that moment. Simultaneous heating of all 32 matrix cells is algorithmically excluded. Taking these factors into account, the total average power consumption of the entire platform is confidently within the range of 0.1-5 W.

[0058] The physical feasibility of implementing such pulsed modes is ensured by the unique properties of the graphene framework. The extremely low density of the material (\rho = 0.008\text{–}0.3\ \text{g / cm}^3) results in negligible thermal mass of the cells. The amount of heat required to heat a cell with a mass m \approx 10^{-4}\ \text{kg} from room temperature to 800\ \text{°C} is:

[0059]

[0060] which is only 4% of the total pulse energy, confirming the high efficiency of direct Joule heating.

[0061] This differentiated approach allows hybridization of the advantages of inorganic and biological receptors in a single platform, significantly expanding the range of detectable analytes.

[0062] Justification of the unity of invention

[0063] The claimed group of inventions is united by a single inventive concept, since each invention of the group is aimed at solving a common technical problem - ensuring ultra-sensitive, selective and regenerable detection of volatile compounds in the gas phase - and is based on the use of one key design and technological solution: a monolithic graphene frame with through channels, which simultaneously performs the functions of a sensitive element, a heater and a carrier of receptor groups.

[0064] The device according to item 1 defines the basic architecture of the platform, in which through channels provide forced gas flow through the volume of material, and the integration of measuring and heating functions in the same electrodes creates the prerequisites for cyclic operation with thermal regeneration.

[0065] The detection method described in paragraphs 6 and 7 is inextricably linked to this architecture, as it implements a cyclic "measurement-heating" mode, which is only possible due to the low thermal inertia of the graphene framework and the high rate of its resistive heating via the same electrodes. Without this design solution, the method would either require a separate heating system (increasing mass, energy consumption, and complexity) or would not ensure complete desorption of the analyte.

[0066] The manufacturing method according to paragraphs 8-10 directly determines the possibility of obtaining a device with the desired properties, namely, the formation of through channels of a given diameter directly during the SHS synthesis stage using removable cores. This technological approach allows for the production of a monolith with regular macroporosity, which distinguishes the claimed device from known chaotically porous graphene materials. Furthermore, combined functionalization (thermostable additives during the synthesis stage + thermolabile receptors afterward) enables the creation of cells with a broad selectivity spectrum without destroying sensitive biological components.

[0067] The indirect radiation detection method described in paragraph 11 is a specific but critically important application of the same platform, leveraging its ability to simultaneously detect different classes of compounds (ozone, nitrogen oxides, volatile iodides) in a single matrix. This capability stems from the device's multidimensional architecture (a matrix of cells with different functionalizations) and the intelligent signal processing embodied in paragraph 5. Without a device capable of generating a multidimensional response, such a radiation reconnaissance method would be impossible or would require a set of heterogeneous sensors.

[0068] Thus, all the inventions of the group are so closely interconnected that they form a single technological complex, where:

[0069] • the device creates a physical basis,

[0070] • the manufacturing method ensures its implementation,

[0071] • The detection method reveals its operating modes,

[0072] • The radiation detection method demonstrates one of the key applications.

[0073] A separate consideration of any of these objects outside the context of the others does not allow us to fully reveal the essence of the invention and achieve the stated technical result - the creation of a universal analytical platform that combines laboratory sensitivity with portability, autonomy and long-term stability.

[0074] The essence of the invention is explained by illustrations, which

[0075] Fig. 1 shows an annotated diagram of the sensor platform in a static neutral state with the main structural elements indicated.

[0076] Fig. 2 shows the active detection phase, characterized by the blue glow of the cells, the flow of molecules of the analyzed gas through the through channels, and the processing of data by the processing unit with machine learning algorithms.

[0077] Fig. 3 shows the thermal regeneration phase, demonstrating the orange-red glow of the cells when heated to 700–800°C, activation of the heating elements, and removal of gas molecules.

[0078] The essence of the invention is explained by examples of specific implementation and application.

[0079] Example 1. Work on explosive detection: When pumping air containing 10 ppb of TNT vapor, cells functionalized with a Cu(II) complex show a 15-20% increase in resistance after 2-5 s. The responses of other cells are minimal. The AI ​​processing unit (17), analyzing this pattern, identifies TNT with a probability of >95%. After 50 measurement cycles, a regeneration cycle is automatically initiated: a 7 A current pulse for 5 s heats the cells to ~700°C, fully restoring their sensitivity.

[0080] Example 2. Implementation of a detection method for medical diagnostics (non-invasive analysis of exhaled air).

[0081] The patient is asked to exhale into the platform's gas inlet channel. The sample volume is 500 ml, and the flow rate is 100 ml / s. The analyzed gas mixture passes through a sensor matrix containing 16 cells, functionalized as follows:

[0082] • cells 1-4: acetone oxidase enzyme (acetone selectivity);

[0083] • cells 5-8: molecular imitators (MIPs) to isoprene;

[0084] • cells 9-12: lipid membranes (ammonia selectivity);

[0085] • cells 13-16: empty (background control).

[0086] The electrical resistance of each cell is measured for 3 seconds at an operating temperature of 80°C. Absolute resistance values ​​are recorded, as well as kinetic parameters: the rate of change in resistance and the time to reach equilibrium.

[0087] The resulting multidimensional vector (16 static + 32 kinetic features) is fed to the input of a trained neural network (a multilayer perceptron with two hidden layers of 64 neurons). The output is the substance identifier and its concentration.

[0088] An experimental study showed that the acetone concentration in exhaled breath of a patient with a confirmed diagnosis of type 2 diabetes mellitus was 1.8 ppm (the normal range is less than 0.8 ppm), which was confirmed by a parallel chromatograph mass spectrometric analysis. The platform's response time was 4 seconds. After completing the measurements, a gentle thermal cleaning (300°C, 10 seconds) was performed to remove residual organic compounds.

[0089] Example 3 (new, technical)

[0090] Example 3. Implementation of a detection method for environmental monitoring (analysis of atmospheric air in an industrial zone).

[0091] The platform is located at a stationary monitoring post near an industrial facility. Atmospheric air is continuously sampled through the gas inlet duct using a built-in micro-blower (flow rate 200 ml / s). The sensor matrix contains 24 cells, functionalized as follows:

[0092] • cells 1-6: SnO2 nanoparticles (CO detection);

[0093] • cells 7-12: WO3 nanoparticles (NO2 detection);

[0094] • cells 13-18: Ag nanoparticles (H2S detection);

[0095] • cells 19-24: polythiophene (SO2 detection).

[0096] Measurements are performed cyclically: 5 seconds of detection at an operating temperature of 150°C, followed by a 2-second pause for stabilization. Every 100 cycles (approximately every 12 minutes), a deep regeneration cycle is automatically initiated: heating to 650°C for 5 seconds, then cooling to operating temperature.

[0097] During tests over 72 hours of continuous operation, the following exceedances of maximum permissible concentrations were recorded:

[0098] Substance Fixed concentration MAC Exceeding time CO 12 mg / m³ 5 mg / m³ 08:15-10:30 NO2 0.25 mg / m³ 0.2 mg / m³ 09:00-11:00

[0099] Data was transmitted wirelessly (LoRa, 868 MHz frequency) to the central monitoring server every 5 minutes. Sensitivity drift over 72 hours was less than 2%, confirming the effectiveness of the thermal regeneration system.

[0100] Tables and parameters

[0101] To ensure a complete disclosure of the essence of the invention, detailed tables are provided below that characterize the main parameters of graphene pumice, thermal cycling modes, examples of configurations of sensor matrices for various tasks, options for targeted functionalization of cells, as well as the correspondence of receptor types to classes of target substances.

[0102] The first table describes the parameters of the graphene pumice monoliths used as the base of the sensing elements. The specific surface area of ​​the material, determined by the BET method using nitrogen as the adsorbent gas, is 1500 square meters per gram, which approaches the theoretical limit for graphene materials and ensures an exceptionally high contact area with the analyzed molecules. The density of the monoliths can vary widely, from 0.008 to 0.3 grams per cubic centimeter, depending on the synthesis conditions and the required characteristics. This ability to vary the density allows for optimization of the cells for specific applications, such as maximum sensitivity or operation under high flow rates.The diameter of the through channels, formed using removable cores in the mold during the synthesis stage, ranges from 0.1 to 2 millimeters, enabling precise control of the cell's pneumatic resistance and ensuring efficient gas flow throughout the monolith's volume. The average mesopore size, which determines the material's transport properties and the accessibility of the internal surface to gas molecules, ranges from 3 to 5 nanometers, providing an optimal balance between a high specific surface area and rapid diffusion. The electrical conductivity of graphene pumice ranges from 10 to 100 siemens per meter and depends on the material's density, with denser monoliths exhibiting higher conductivity, which is important for both resistance measurements and efficient Joule heating. The material's thermal stability in air reaches 900 degrees Celsius, meaning no oxidation or structural degradation occurs at temperatures used for thermal regeneration.The mechanical compressive strength ranges from 5 to 50 kilopascals and also correlates with the density of the monolith, providing sufficient rigidity to maintain integrity during installation in a cartridge and during operation under conditions of vibration and pressure changes.

[0103] The second table details the thermal cycling modes implemented by the heating control unit to support various functions of the sensor platform. The operating mode maintains a constant temperature between 80 and 120 degrees Celsius, providing optimal conditions for the sorption of target molecules on the functionalized surface without undesirable desorption or thermal degradation of the receptors. The soft cleaning mode removes loosely bound water molecules and light volatile organic compounds and is performed at a temperature of 300 to 400 degrees Celsius for 10-30 seconds, every 10-20 cycles, thereby maintaining a stable baseline without significant thermal stress.Deep regeneration mode is used to desorb explosives and other stable compounds firmly attached to active sites. It is performed at temperatures ranging from 500 to 700 degrees Celsius for 3-10 seconds every 50-100 measurement cycles, ensuring full restoration of cell sensitivity. Burnout mode, or extreme cleaning, is used when active sites are contaminated with irreversibly sorbed substances or when complete surface restoration is necessary after extended use. It involves short-term heating to 750-880 degrees Celsius for 1-5 seconds every 1,000 cycles or as needed, allowing for the removal of even coke deposits and other stubborn contaminants.

[0104] The third table presents example sensor array configurations for various practical applications, demonstrating the platform's versatility and adaptability to specific application areas. For counterterrorism security applications, a 32-cell array is proposed with functionalization including copper complexes for detecting nitroaromatic explosives, thiol groups for detecting peroxide explosives, polyaniline for detecting ammonia, tin oxide nanoparticles for detecting carbon monoxide, and empty, unfunctionalized cells for background signal control and accounting for the influence of external factors. For medical diagnostics, including noninvasive exhaled breath analysis, a 16-cell array is proposed, functionalized with enzymes selective for metabolic markers, molecular mimics for recognizing specific biomolecules, and lipid membranes for detecting volatile compounds associated with various pathologies.For environmental monitoring, a 24-well matrix with tin and tungsten oxide nanoparticles is proposed for detecting nitrogen oxides and ozone, polythiophene for detecting hydrogen sulfide, silver nanoparticles for detecting halogenated compounds, and polydimethylsiloxane for adsorption of a wide range of volatile organic compounds. For food safety applications, a 16-well matrix with molecular mimics selective for ethylene and ammonia as markers of ripening and spoilage is used, as well as with various polymer coatings for detecting mercaptans and other compounds indicating deterioration of food freshness. For studies related to emotional state and stress, a 16-well matrix with lipid membranes and molecular mimics selective for cortisol, isovaleric acid, and ammonia as markers of stress conditions and metabolic changes is used.

[0105] Table 4 contains a detailed description of the targeted functionalization of individual cells within a typical matrix for complex analysis. Cells 1 through 4 are functionalized with copper complexes and exhibit selectivity for nitroaromatic explosives, such as TNT, at an operating temperature of 100 degrees Celsius. Cells 5 through 8 contain thiol groups and are selective for peroxide explosives, including acetone peroxide, also at an operating temperature of 100 degrees Celsius. Cells 9 through 12 are functionalized with tin oxide nanoparticles and are designed to detect carbon monoxide and methane at an elevated operating temperature of 300 degrees Celsius, necessary to activate the oxide's catalytic properties. Cells 13 through 16 contain polyaniline and provide selective detection of ammonia at an operating temperature of 120 degrees Celsius.Cells 17 through 20 are functionalized with acetone oxidase for selective detection of acetone at a gentle operating temperature of 80 degrees Celsius, which maintains enzyme activity. Cells 21 through 24 contain silver nanoparticles and are designed to detect hydrogen sulfide at an operating temperature of 150 degrees Celsius through the formation of silver sulfide and changes in conductivity. Cells 25 through 28 are coated with polydimethylsiloxane for non-selective sorption of a wide range of volatile organic compounds at an operating temperature of 80 degrees Celsius, allowing them to be used for integrated pollution assessment. Cells 29 through 32 are left empty, i.e., unfunctionalized, to record background signal and account for the influence of temperature, humidity, and other external factors at the same operating temperature of 80 degrees Celsius.Additionally, for radiation-chemical reconnaissance tasks, a block of 8-16 cells functionalized with iodine sorbents can be used, including silver nanoparticles for chemisorption of iodides, tin oxides for detecting ozone and nitrogen oxides as radiolysis markers, as well as acid-base indicators for detecting acid vapors and solvents formed during the decomposition of materials under the influence of ionizing radiation.

[0106] The fifth table establishes a correspondence between the types of receptor groups and the classes of target substances, justifying the choice of functionalization for various analytical tasks. Nitroaromatic explosives, including TNT and RDX, are effectively detected using copper and amino group complexes through pi-pi stacking mechanisms and coordination with nitro groups. Peroxide explosives, such as acetone peroxide and hexamethylene triperoxide diamine, are detected using thiol groups and acid sites that undergo redox reactions with the peroxide moiety. Toxic gases and heavy metal vapors, including mercury vapor, are detected using gold nanoparticles that form an amalgam, selenides and thiol groups that chemisorb mercury to form stable sulfides, and iodine-containing complexes that specifically interact with metal vapors.Toxic and flammable gases, such as carbon monoxide and methane, are detected using tin and palladium oxide nanoparticles, which catalytically oxidize these molecules on the surface and alter the sensor's conductivity. Industrial gases, including ammonia and hydrogen sulfide, are detected using polyaniline, which alters conductivity upon doping, and silver nanoparticles, which form silver sulfide. Breath biomarkers, such as acetone and ammonia, are detected using enzymes that carry out a specific enzymatic reaction and molecular mimics that provide steric recognition of target molecules. A wide range of volatile organic compounds, including benzene, toluene, and formaldehyde, are detected using polydimethylsiloxane, which provides hydrophobic interactions and physical adsorption, and using non-functionalized graphene, which has intrinsic sorption activity toward aromatic compounds.

[0107] TABLE 1: Parameters of graphene pumice monoliths

[0108] Parameter Meaning Comment Specific surface area 1500–2000 m² / year BET method, nitrogen Density 0.008–0.3 g / cm³ Depends on the synthesis conditions Diameter of through channels 0.1–2 mm It is determined by the mold core Average mesopore size 3–5 nm Optimal for gas diffusion Electrical conductivity 10–100 cm / m Depends on the density Thermal stability in air up to 900°C Does not oxidize Mechanical compressive strength 5–50 kPa Depends on the density

[0109] TABLE 2: Thermal Cycling Modes

[0110] Mode Temperature Time Frequency Purpose Worker 80–120°C Constantly — Optimal sorption Soft cleaning 300–400°C 10–30 s Every 10–20 cycles Removal of water, light VOCs Deep regeneration 500–700°C 3–10 s Every 50-100 cycles Desorption of explosives and persistent compounds Burning 750–880°C 1–5 s In case of poisoning, once every 1000 cycles Coke removal, complete restoration

[0111] TABLE 3: Examples of sensor array configurations for various tasks

[0112] Task Number of cells Functionalization (example) Target compounds Anti-terrorism 32 Cu(II), thiols, polyaniline, SnO2, empty TNT, TATP, NH3, CO, background Medical diagnostics 16 Enzymes, MIPs, lipid membranes Acetone, ammonia, isoprene, cortisol Environmental monitoring 24 SnO2, WO3, polythiophene, Ag, PDMS CO, NO2, O3, H2S, VOCs Food safety 16 MIP (ethylene, ammonia), polymers Ethylene, ammonia, mercaptans Emotions / stress 16 Lipid membranes, MIP for acids Cortisol, isovaleric acid, ammonia

[0113] Table 4 (corrected): Sensor matrix architecture

[0114] In this table, we describe each group of cells by separating the concepts.

[0115] Cell number Chemical nature of the receptor Technology / Form Selectivity example Operating temperature, °C 1–4 Transition metal complexes (Cu²+) A molecular complex immobilized on a surface Nitroaromatic explosives (TNT) 100 5–8 Organic ligands Thiol (-SH) groups grafted to the surface Peroxide explosives (TATP) 100 9–12 Inorganic semiconductors Metal oxide nanoparticles (SnO2) CO, CH4 300 13–16 Conducting polymers Polyaniline film NH3 120 17–20 Protein enzymes Acetone oxidase layer Acetone 80 21–24 Precious metals Silver nanoparticles (Ag0) H2S (formation of Ag2S) 150 25–28 Synthetic polymers Hydrophobic coating (PDMS) VOC (non-selective sorption) 80 29–32 - (No) Empty (non-functional graphene) Background signal 80 RX block Organics / metals Sorbents for iodine / acid-base indicators Volatile iodides, radiolysis markers *80–150*

[0116] Table 5 (corrected): Principles of selectivity

[0117] Here we show why this combination works.

[0118] Target class of substances Examples Receptor type (Chemical nature) Detection mechanism Nitroaromatics TNT, RDX Metal complexes (Cu, Co), π-acceptors Host-guest complexation, charge transfer Peroxides TATP, HMTD Thiols, Brønsted acid sites Oxidation of thiols, catalytic decomposition Hg0 and heavy metals Mercury vapor Au, Ag, sulfur- and selenium-containing ligands Amalgamation, chemisorption with a change in conductivity Toxic gases CO, NO2 Semiconductor oxides (SnO2, WO3) Change in conductivity during adsorption / oxidation Industrial gases NH3, H2S Conducting polymers (PANI), Ag nanoparticles Polymer doping / de-doping, sulfidization Biomarkers Acetone, isoprene Enzymes (oxidases), MIP Enzymatic reaction products, steric recognition VOC (general background) Benzene, toluene PDMS, carbon materials Hydrophobic adsorption (van der Waals)

Claims

1. A sensor platform for detecting and identifying volatile compounds, comprising a housing with inlet and outlet gas channels and a measuring circuit, characterized in that it contains a sensor matrix made in the form of an array of individual cells, each of which is an independent sensitive element based on a monolithic block of a graphene framework with a highly developed specific surface area, wherein the matrix cells are functionalized with various receptor groups selective to different classes of target molecules, and the measuring circuit is designed with the possibility of multiplexed reading of the electrical parameters of each cell, wherein the monolithic block is made with through channels that ensure forced pumping of the gas medium through its volume.

2. The platform according to claim 1, characterized in that it contains a heating control unit configured to pass through the electrodes of the selected cells an electric current sufficient for resistive heating of the graphene framework to 300-880°C, with the possibility of programmable thermal cycling, wherein the heating control unit and the measuring circuit use the same electrode structures.

3. The platform according to claim 1, characterized in that the matrix cells are functionalized with one or more types of receptors from the following set: metal complex compounds, amino groups, thiol groups, metal nanoparticles, metal oxides, conducting polymers, antibodies, enzymes, molecular imitators (MIP), macrocyclic compounds, lipid membranes.

4. The platform according to paragraph 1, characterized in that the sensor matrix is ​​made in the form of a removable cartridge with individual fastenings for each monolithic block, while the electrodes are made in the form of mesh, comb or needle structures applied to the surface of the monolith or pressed into it.

5. The platform according to paragraph 1, characterized in that it contains a data processing unit with machine learning algorithms implementing stages of signal preprocessing, including digital filtering and drift compensation, reduction of the dimensionality of the feature space using principal component methods or a neural network autoencoder and classification of gas mixtures based on a set of static and kinetic parameters of the cell response using a multilayer perceptron, a support vector machine or ensemble algorithms, with the formation of a substance identifier, its concentration in the range from 0.01 to 100 ppb and an assessment of the reliability of the result, as well as a wireless communication module for data transmission and updating libraries of reference fingerprints.

6. A method for detecting and identifying volatile compounds using the platform according to claim 1, comprising passing the analyzed gas medium through a sensor matrix made in the form of an array of individual cells, each of which is an independent sensitive element based on a monolithic block of a graphene framework with a highly developed specific surface area, wherein the matrix cells are functionalized with various receptor groups selective to different classes of target molecules, after which the electrical responses of the cells are read, a multidimensional response vector is formed and compared with a library of standards or classification is performed using a machine learning model.

7. The method according to paragraph 6, characterized in that the sensor matrix is ​​periodically thermally cleaned by resistively heating the graphene frame to 500-880°C for 1-30 seconds, while the measurement and heating modes are carried out cyclically, where the measurement phase is carried out at 80-120°C, and the cleaning phase is carried out at 500-880°C.

8. A method for manufacturing a sensor cell for a platform according to claim 1, comprising forming a monolith of a graphene framework of a given shape by the SHS synthesis method in a press mold with removable cores for creating through channels, applying electrode structures and functionalizing the surface by impregnation with solutions of receptor compounds, followed by drying and heat fixation.

9. The method according to claim 8, characterized in that functionalizing additives selected from the following series are additionally introduced into the reaction mixture for SHS synthesis: metal salts, organic ligands, macrocyclic compounds or nanodispersed metal oxides, which form active centers uniformly distributed throughout the volume of the monolith during the synthesis process.

10. The method according to paragraph 8, characterized in that for thermolabile receptor groups that are unstable to the temperature of SHS synthesis, functionalization is carried out after the formation of the monolith by immersion impregnation with a solution containing the said groups, followed by drying and fixation at a temperature not exceeding 80°C.

11. The method according to paragraph 6, characterized in that it additionally carries out indirect detection of sources of ionizing radiation by analyzing the gas environment for the presence of radiolysis markers or specific volatile compounds of radionuclides using matrix cells functionalized for the detection of said markers.

Images