Patent Application
20250198960
The present technology relates to the use of sensors based on functionalized carbon nanotubes for the detection of CO2 in gaseous environments. Nanotubes have multiple layers and are functionalized with —OH and —COOH groups. The process to determine the concentration of CO2 measures the electric current that passes through the sensor in the presence of different gases, the data obtained is treated and the concentration of CO2 is obtained by methods involving artificial intelligence. The technology simultaneously has the following advantages: it identifies CO2 in natural gas, uses short sensor exposure time ranges, allows similar results to be obtained even with small variations in the device due to the production process, uses optimized parameters and models to find the concentration of CO2, the sensor can be miniaturized, is easily produced on large scale, is versatile, can be adapted to different substrates or analysis conditions, provides real-time results and is durable, maintaining accuracy even after several cycles in a row or after a long time of use. The technology can be used for identification of CO2 in heterogeneous gaseous environments, such as natural gas.
A challenge in the field of exploration of oil or natural gas wells is the high content of contaminants, such as CO2 present in the wells. CO2 is the main cause of the greenhouse effect and corrosion of installations. Thus, it is essential to know its concentration for the formulation of solutions in its management, as large concentrations of CO2 cannot be discharged directly into the atmosphere. It is important to monitor pipeline leaks or have greater control over the causes of corrosion in installations (J. J. M. OLIVEIRA. O PROBLEMA DA CORROSÃO POR CO2 NOS TUBOS DE PRODUÇÃO DE POÇOS LOCALIZADOS NA PROVÍNCIA DO PRÉ-SAL. Monograph presented to the Postgraduate Course in Specialization in Oil and Natural Gas Engineering, at Faculdade do Centro Leste, 2015. https://www.researchgate.net/publication/308765198_O_Problema_da_Corrosao_por_CO2_nos_Tubos_de_Producao_de_Pocos_Localizados_na_Provincia_do_Pre-Sal).
In general, to identify the concentration of CO2 and other components present in natural gas, some techniques are used, such as gas chromatography, electrochemical sensors, and semiconductor oxide or infrared sensors. Due to the high cost, short useful life, need for qualified labor, difficulty in withstanding the chemical and physical conditions of a heterogeneous gaseous environment, such as natural gas, and difficult integration into the production chain, these techniques do not fully meet the required demand. Thus, there is a need for CO2 sensing using devices that are selective, small, portable, easy to integrate into the production and distribution chain, and can perform measurements in situ and in real time. In addition to the device, suitable methods are necessary for its use, including short and appropriate time ranges for measurement, calculations that are not affected by small differences due to large-scale production, the use of optimized parameters, and models capable of determining the concentration of CO2 (Vinícius Ornelas da Silva). DESENVOLVIMENTO DE DISPOSITIVOS A BASE DE NANOMATERIAIS DE CARBONO PARA SENSORIAMENTO DE CO2. Dissertation presented to the Postgraduate Program in Physics, Federal University of Minas Gerais, 2023).
Carbon nanotubes are allotropes of carbon with a cylindrical nanostructure. They can have one layer or multiple concentric layers. Their functionalization happens when groups of atoms bind together on its surface, allowing the carbon nanotube to better interact with the medium in which it is inserted. Carbon nanotubes have enormous potential in sensor applications due to their excellent mechanical and electrical properties, large surface area and resistance to corrosive environments. Such characteristics make these materials natural candidates for a novel nanosensor platform. The properties of nanotubes depend on their structure or chirality, and may have, e.g., metallic or semiconductor characteristics. When there are several nanotubes of different chiralities deposited in a medium forming a nanotube film, this film can exhibit a metallic behavior. When gases come into contact with the nanotube film, their transport properties will change, and it is possible to associate this change with the nature of the gas.
In this sense, there are some documents in the state of the art that disclose devices and methods aimed at identifying molecules in gaseous environments.
The patent document US2023121903A1, with priority date Oct. 17, 2022, entitled “Device and analysis method for appreciating and identifying smells”, describes a device comprising an airflow system, a chamber with an array of sensors of molecules present in the air, a configurable sensor interface, a microcontroller, a memory, and a metal mesh filter. The device collects data from molecules present in the air, processes, stores and transfers the data to the user in real time. The data collected by the device is linked to a methodology involving artificial intelligence to identify the aroma of the air sample. However, the device described in the document US2023121903A1 is designed to work in the presence of atmospheric air, not demonstrating to withstand corrosive gaseous environments, and may not maintain its accuracy after several consecutive cycles or after a long time of use in corrosive environments with natural gas, uses an array of sensors to distinguish several molecules, increasing the complexity of the device or data processing method, uses air as a desorption gas for sensor molecules, and the air may have impurities that react with the sensor, reducing its lifespan or its accuracy in identifying some gases, uses non-optimized input parameters in the artificial intelligence model for the detection of CO2, such as the area or the maximum value of the normalized gain or electrical current curve passing through the sensor, which reduces the accuracy in identifying CO2 concentration or does not favor a time range that contributes to the response speed of the sensor array, and the device and the process are not optimized for the detection of gases like CO2 in the presence of heterogeneous gases, such as natural gas.
The patent document EP2912454A1, with priority date Oct. 29, 2012, entitled “Functionalized nanotube sensors and related methods”, describes a sensor and method based on nanotubes with metal oxides for identification of volatile organic compounds and biomarkers in a fluid environment. The sensor has a power source configured to apply an electrical voltage to the nanotube array and a current sensor configured to monitor and detect changes in a response current that varies after binding with the composite target. However, the technology discussed in the document EP2912454A1 focuses on the identification of molecules that can be used as biomarkers, not presenting properties that withstand corrosive gaseous environments, and may not maintain precision even after several consecutive cycles or after a long time of use, it requires an array of sensors to distinguish several molecules, increasing the complexity of the device or data processing method, uses input parameters not optimized for CO2 detection in the artificial intelligence model, such as maximum gain area or value or normalized electric current, decreasing the precision in identifying the concentration of CO2 or not favoring a time range that contributes to the actuation speed of the sensor array, uses non-normalized measured parameters, making the results obtained vary depending on the method of large-scale sensor production or small variations in its use, in addition to the device in the process not being optimized for the detection of gases such as CO2 in the presence of heterogeneous gases, such as natural gas.
The use of sensors based on functionalized carbon nanotubes for the detection of CO2 in gaseous environments was not found in the state of the art. The present disclosure relates to the use of sensors based on multilayer carbon nanotubes functionalized with —OH and —COOH groups for the detection of CO2 in gaseous environments. The technology uses methods, such as data processing and artificial intelligence, that optimize the identification of the concentration of CO2 in heterogeneous environments such as natural gas, presenting the following advantages together: identification of CO2 in heterogeneous gases, such as natural gas, use of optimized input parameters in the artificial intelligence model for the detection of CO2, such as the area or the maximum value of the normalized gain or electric current curve passing through the sensor, increase of the precision in identifying CO2, allows the use of short sensor exposure time ranges, allowing the obtainment of similar results even with small variations of the device due to the production process or its handling, is fast, can be miniaturized, is easy to produce on a large scale, is versatile, can adapt to different substrates or analysis conditions, and is durable, maintaining accuracy even after several consecutive cycles or after a long time of use.
The present technology relates to the use of sensors based on functionalized carbon nanotubes for the detection of CO2 in gaseous environments. Nanotubes have multiple layers and are functionalized with —OH and —COOH groups. The process to determine the concentration of CO2 measures the electric current that passes through the sensor in the presence of different gases, the data obtained is treated and the concentration of CO2 is obtained by methods involving artificial intelligence. The technology simultaneously has the following advantages: it identifies CO2 in natural gas, uses short sensor exposure time ranges, allows similar results to be obtained even with small variations in the device due to the production process, uses optimized parameters and models to find the concentration of CO2, the sensor can be miniaturized, is easily produced on large scale, is versatile, can be adapted to different substrates or analysis conditions, provides real-time results and is durable, maintaining accuracy even after several cycles in a row or after a long time of use. The technology can be used for identification of CO2 in heterogeneous gaseous environments, such as natural gas.
Sensors based on functionalized carbon nanotubes can be used for the detection of CO2 in gaseous environments. The process for detecting CO2 using the sensors of this technology can be carried out through the following steps:
In step (a), the film (1), present in sensor (0), has functionalized carbon nanotubes with —OH and —COOH groups, where all the nanotubes are leaning against and entangled with each other, ensuring that there is always a pathway for electric current to flow between the metal contacts (3), as described in FIG. 2(b) and
In steps (b) and (c), a voltage source V allows the passage of current Ibase(t) on the sensor (0), which is measured by the ammeter (A) and recorded in the computer (9). The change in current Ibase(t) recorded in the computer (9) will indicate that different gases have passed through the sensor.
In step (d), exposing the sensor (0) to an inert gas or low pressure environment within the gas passage chamber (4) removes unwanted impurities that can interact with the functionalized carbon nanotubes and impair the sensor measurement (0). The value of the current Ibase(t) becoming constant over time t is an indication that the sensor (0) is ready to be used to identify other gases. The flow of inert gas or the generation of low pressure can be maintained if new measurements with the sensor (0) are necessary. In this case, the control of the density of the gas containing CO2 is favored with the injection of inert gases or the generation of low pressure inside the gas passage chamber (4).
In step (e), the sensor (0) is subjected to a gas containing CO2, such as natural gas, and the current ICO2(t) passing through the sensor (0) is recorded. Current ICO2(t) will tend to decrease over time in the presence of CO2, demonstrating that the gas is interacting with the carbon nanotubes.
The current passing through the sensor (0) may fluctuate according to its production and deterioration over time, requiring the use of the gain G(t) described in step (f), favoring the method of identification of CO2 in heterogeneous gases. The minimum time texp of exposure of the sensor (0) to the gas flow containing CO2, preferably between texp=60 s and texp=300 s, is chosen in a way that favors the distinction of G(t) for each value of the concentration of CO2 contributing to the increase of the speed in the identification of gases.
The graph of G(t) as a function of time t, plotted in step (g), is used to calculate parameters that facilitate data analysis, such as the area AG under the gain graph G(t) and the maximum gain value Gmax, as shown in step (h). With the area AG under the gain graph G(t) and the maximum gain value Gmax, it is possible to obtain the concentration of CO2 through trained models designed to determine concentrations of CO2.
In step “b”, the difference in electrical potential through a voltage source (V) is between 1 mV and 100 V.
The segment of the gain curve G(t) with the greatest relevance to obtain the concentration of CO2 can be obtained by considering only the values of the gain G(t) at the moments when the sensor (0) comes into contact with the gas flow containing CO2. This procedure favors the obtainment of concentrations of CO2 providing greater precision in the identification of the concentration of CO2 by using different methods, such as artificial intelligence.
The model trained to obtain concentrations is based on the features area AG under the gain G(t) and the maximum gain value Gmax where each predicted class is labeled with the concentration of CO2 and can be selected from the following:
After the concentration of CO2 has been detected, the sensor (0) can be recovered by desorption of unwanted elements from the sensor (0) by following the steps below:
In steps (I), (II) and (III), the sensor (0) is placed in the chamber (4), a voltage is applied, and the current is recorded. In step (IV), impurities that are trapped among the functionalized carbon nanotubes can be removed. When the current value Ides (t) is constant over time, the sensor is recovered and ready for new use.
Exposure of the sensor (0) to an atmosphere with inert gas flow, such as argon, nitrogen, or helium, occurs by allowing inert gas to exit a compartment for different gases (10) through control valves (11), enter the gas passage chamber (4) through the inlet for gases (6), and exit through an outlet for gases (7).
The formation of a vacuum or a low-pressure environment occurs by reducing the gas inflow at the inlet for gases (6) and allowing gas to exit through the outlet for gases (7) with the aid of a vacuum pump (12).
The present technology can be better understood through the following not limiting examples.
Multi-walled carbon nanotubes are produced by the chemical vapor phase deposition (CVD) technique. To perform the synthesis of carbon nanotubes by CVD growth, there is a system consisting of a hot-wall furnace with a quartz tube in its interior. This tube rotates around its axis of rotation at a temperature of approximately 730° C. A constant flow of argon (2 l/min) creates an inert atmosphere under atmospheric pressure, and carries impurities generated by the growth process inside the tube to the exhaust. A constant flow of ethylene gas (1.5 l/min) is also used as a carbon source for the synthesis of carbon nanotubes. The system is continuously fed with a catalyst consisting of Fe and Co oxide nanoparticles, supported by particles of Al2O3, through a feeder connected to the tube inlet. The quartz tube is kept inclined 10° to horizontal and rotates at a speed of 6 RPM to promote the transport of the catalyst through the furnace hot zone. The carbon nanotubes are formed in the catalyst and transported for storage in a container attached to the end of the tube. After the synthesis, the NTC are functionalized. During the functionalization, —OH and —COOH groups are added to the walls of carbon nanotubes through an acid (HNO3/H2SO4) treatment process, which is performed in an ultrasound bath under controlled temperature. The functionalization process is necessary to dilute the carbon nanotube in water or organic solvents. After that, the material is washed and centrifuged several times, and then dried in a kiln (100° C. for 24 hours) and ground in an automated pistil mill.
The carbon nanotubes are dispersed in deionized water at an initial concentration of 2% w/v, in an ultrasound bath for 4 hours. The obtained suspension is centrifuged at 300 RPM for 30 minutes, and the supernatant material is separated from the precipitate. Then, the concentration of the supernatant is measured. In general, after this process, the ink has a concentration of around 1.2-1.3% w/v. Finally, the ink is diluted in isopropyl alcohol until achieving the standard concentration of 0.6% w/v.
Once the process of making the standard nanotube ink is finished, the process of producing the sensors begins. To achieve this, a mask was cut using a laser printer based on a design created on a computer, for the evaporation of the electrical contacts (length=1 mm and width=3 mm). Another mask was made to subsequently paint the conductive channel (length of the carbon nanotube channel=2 mm and width of the carbon nanotube channel=1 mm). Using the thermal evaporation method, electrical contacts (metal contacts (3) from
After painting, the substrate is inserted into a chip, where the contact between the conductive channel and the pins that make up the chip is established, constructing the sensor (0). Once the sensor (0) is completed, it is inserted into the measurement system.
As an example of heterogeneous gas, natural gas was used. The difference in gas flows during the measurement generates a current pattern I(t) as a function of time t similar to the one shown in
With the value of the currents Ibase(t) and ICO2(t), the gain G(t)=|(ICO2(t)−Ibase(t))|/Ibase(t) is calculated.
With gain G(t), the features area of the gain AG and the maximum gain value Gmax are obtained. For the concentration of 10% of CO2 and the exposure time of the sensor (0) to natural gas of t=60 s, we have the area AG 23 s and the maximum gain value Gmax Of 0.135. The values AG and Gmax are submitted as input data to the trained decision tree and Random Forest models, as well as to the linear regression model, where each predicted class is labeled with the concentration of CO2.
To train the decision tree and Random Forest models, the dataset is initially split into training and testing data using the cross-validation technique. Cross-validation is a resampling method that uses different pieces of data to test and train a model in different iterations. The training data is used to teach the model how to identify the outputs when using a new set of inputs. The test data is used to validate the system. The training data is divided into 5 parts. Initially, the first part is used to train and the others to validate the method, and a hit rate is calculated. Then, the second part is used to train and the others to validate the algorithm, and a new hit rate is calculated. This process is done repeatedly up to the fifth part of the training data, calibrating the parameters of the method. In the end, a mean of the hit rate of the processes carried out from the first to the fifth part is calculated, and the best parameter is determined where the least error is obtained. It is observed that the main characteristic of the cross-validation method is to always train and test the algorithm on different dataset configurations, i.e., it is the equivalent of testing and training the algorithm with new datasets in each process performed from the first part of the data to the fifth part; this is what makes the cross-validation method a reliable measure of the method performance. Then, once the best parameter for this particular method is selected, it is retrained, and finally evaluated with new test data being determined as the final model.
When we perform multiple measurements, unwanted molecules can get stuck in the nanotubes present in the film (1), and it is necessary to recover the sensor (0) in order to make a new measurement. Sensor recovery (0) was performed by subjecting the sensor to 110° C. for 8 hours under argon atmosphere for moisture removal. It turns out that, after recovery, the resistance of the sensor (0) decreased from 2.6 kΩ to 2.4 kΩ. This may be associated with improved quality of electrical contacts, as well as desorption of contaminant molecules from the sensor surface (0).
Operating range: in order to verify the operating range of the sensor, gain curves G(t) over time t were collected under a condition where the sensor was exposed to random concentrations ranging from 0% CO2 (pure natural gas) to 100% CO2 over a period of 60 seconds.
Lifespan: an important feature of sensors is the lifespan. Sensors that need constant replacement and maintenance may not be very interesting due to costs, repair downtime or handling by qualified labor. Thus, it is necessary to evaluate the sensor lifespan. To do this, different rounds of experiments were carried out that add up to more than 5,000 measurements on the same device over 8 months.
Reproducibility of the sensors: to test the reproducibility of the production of sensors, four different devices were manufactured. In each round of measurements for a given sample, a total of 30 curves were collected, with 5 measurements for each concentration, from 0% to 12%. As can be seen in
Saturation effect: sensors can present saturation conditions after repeated exposure to the working gas, directly interfering with the precision of the measurements.
In summary, the use of the sensor (0) has several properties such as a lifespan of more than eight months without losing efficiency, has good reproducibility, does not present a saturation over time that prevents its use in the identification of concentration of CO2 and has a good operating range identifying from 0% to 100% of CO2 in natural gas (Vinícius Ornelas da Silva. DESENVOLVIMENTO DE DISPOSITIVOS A BASE DE NANOMATERIAIS DE CARBONO PARA SENSORIAMENTO DE CO2. Dissertation presented to the Postgraduate Program in Physics, Federal University of Minas Gerais, 2023).
| Number | Date | Country | Kind |
|---|---|---|---|
| 1020230264832 | Dec 2023 | BR | national |
