Ultra-low power consumption nano-electromechanical resonant NO2 gas sensor based on audio frequency comb and method

Abstract

The invention belongs to the technical field of gas sensors, and particularly relates to an ultra-low power consumption nano-electromechanical resonant NO2 gas sensor based on an audio frequency comb and a method. Wherein the drain electrode and the source electrode are respectively arranged on the silicon dioxide insulating layer, an I-shaped groove is embedded in the silicon dioxide insulating layer between the drain electrode and the source electrode, an I-shaped gate electrode is arranged in the I-shaped groove, and the adsorption layer is laid on the silicon dioxide insulating layer and covers the surfaces of the drain electrode and the source electrode. The middle part of the adsorption layer is suspended above the I-shaped gate electrode, and the silicon dioxide insulating layer is arranged on the monocrystalline silicon substrate; specific detection of NO2 is achieved through single-layer graphene, the internal resonance excitation audio frequency comb phenomenon is introduced, the gas detection sensitivity and resolution are greatly improved, and sub-ppb-level NO2 gas detection is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of gas sensors, and particularly relates to an ultra-low power nanoelectromechanical resonant NO 2 gas sensor based on an acoustic frequency comb and a method thereof. Background Art

[0002] Nitrogen dioxide (NO 2 ) is an important air pollutant, mainly derived from combustion processes and motor vehicle exhaust emissions. NO 2 not only causes serious environmental pollution, but also poses a great threat to human health, especially to the respiratory system. Research shows that short-term exposure to 0.1 - 0.5 ppm (100 - 500 ppb) of NO2 can cause mild eye and respiratory irritation, and sensitive populations (such as asthma patients) may experience difficulty breathing. 0.5 - 1 ppm (500 - 1000 ppb) can cause obvious respiratory discomfort, including coughing, sore throat and nasal discomfort. Asthma patients may experience acute exacerbation symptoms. Short-term exposure to higher concentrations of 1 - 5 ppm (1000 - 5000 ppb) can lead to acute lung inflammation, difficulty breathing and insufficient oxygen intake, especially for children and the elderly. Long-term exposure in the concentration range of 0.02 - 0.05 ppm (20 - 50 ppb), even for healthy people, may show mild respiratory inflammatory reactions; the symptoms of patients with respiratory diseases such as asthma and allergic rhinitis may worsen. Exposure above 0.05 ppm (50 ppb) and for a long time will exacerbate respiratory inflammation and increase the risk of chronic respiratory diseases (such as chronic bronchitis, asthma) and cardiovascular diseases. NO 2 above 0.1 ppm (100 ppb) will significantly increase the risk of respiratory diseases, especially harmful to children and the elderly. Therefore, real-time monitoring of NO 2 gas is particularly important.

[0003] Currently, the detection methods of NO 2 gas mainly include chemiluminescence method, electrochemical sensor method, optical absorption spectroscopy method, etc. However, these detection methods generally have problems such as high power consumption, large volume, insufficient sensitivity, poor selectivity, etc., and cannot meet the requirements of high-precision real-time monitoring in complex environments.

[0004] Sensors based on two-dimensional materials have received extensive attention in recent years. Among them, graphene, as a two-dimensional material with a unique structure and excellent physical and chemical properties, has become a research hotspot in the field of gas sensors. Graphene materials have a high specific surface area, excellent electrical conductivity and mechanical strength, which can effectively improve the sensitivity and response speed of gas sensors. However, how to further reduce the power consumption of graphene sensors while improving their sensitivity to NO 2The detection selectivity and accuracy of gases remain a technical problem that urgently needs to be solved.

[0005] Resonant sensors can achieve highly sensitive gas detection through the interaction between the vibration modes of resonators and gas molecules. In particular, graphene-based resonant sensors utilize the change in the adsorption characteristics of NO 2 gases under resonant conditions to achieve ultra-low power consumption and high-precision detection of NO 2 gases. By regulating the acoustic frequency comb within the graphene resonance band, the performance of the sensor can be further optimized, enhancing its detection sensitivity and selectivity for NO 2 gases, thereby achieving efficient monitoring of NO 2 gases.

[0006] Against this background, the research and development of ultra-low power nanoelectromechanical resonant NO 2 gas sensors and methods based on acoustic frequency combs have important theoretical and practical significance. This new type of sensor can achieve highly sensitive and real-time monitoring of NO 2 gases in complex environments, and has the advantages of low power consumption, good stability, and low cost. It is expected to be widely used in environmental monitoring, industrial process control and other fields. Summary of the Invention

[0007] To overcome the above problems, the present invention provides an ultra-low power nanoelectromechanical resonant NO 2 gas sensor and method based on an acoustic frequency comb, which realizes the specific detection of NO 2 through single-layer graphene, and introduces the phenomenon of internal resonance excitation of the acoustic frequency comb to greatly improve the gas detection sensitivity and resolution, realizing the detection of NO 2 gases at the sub-ppb level.

[0008] The object of the present invention is achieved by the following technical solutions:

[0009] An ultra-low power nanoelectromechanical resonant NO 2 gas sensor based on an acoustic frequency comb, comprising a drain electrode 11, an I-shaped gate electrode 12, a source electrode 13, a silicon dioxide insulating layer 3, a single-crystalline silicon substrate 4, and an adsorption layer 5; wherein the drain electrode 11 and the source electrode 13 are respectively disposed on the silicon dioxide insulating layer 3, and an I-shaped groove 2 is embedded in the silicon dioxide insulating layer 3 between the drain electrode 11 and the source electrode 13. The I-shaped gate electrode 12 is disposed in the I-shaped groove 2. The adsorption layer 5 is laid above the front end of the silicon dioxide insulating layer 3 and covers the front-end surfaces of the drain electrode 11 and the source electrode 13, and the middle part of the adsorption layer 5 is simultaneously suspended above the front end of the I-shaped gate electrode 12. The silicon dioxide insulating layer 3 is disposed on the single-crystalline silicon substrate 4.

[0010] The adsorption layer 5 is a single-layer graphene layer.

[0011] The adsorption layer 5 is a double-ended clamped beam structure. The drain electrode 11 and the source electrode 13 have the same shape and size, and are symmetrically arranged on both sides of the silicon dioxide insulating layer 3.

[0012] The single-crystalline silicon substrate layer 4, the silicon dioxide insulating layer 3, the drain electrode 11, the I-shaped gate electrode 12, and the source electrode 13 are an integral part.

[0013] The adsorption layer 5 is a single-layer graphene specific sensitive film.

[0014] An ultra-low power consumption nanoelectromechanical resonant NO 2 gas sensor based on an acoustic frequency comb is used to measure NO 2 gas concentration. The specific steps are as follows:

[0015] Step 1: Calibrate the first-order and second-order modes of the gas sensor vibration using the optical method

[0016] Step 1.1: Conduct all measurements in a vacuum test chamber. Select a 633 nm red laser as the detection light source and irradiate the laser beam onto the adsorption layer 5.

[0017] Step 1.2: Apply a DC voltage to the I-shaped gate electrode 12 and gradually increase it from 0 V. When the adsorption layer 5 vibrates, the phase difference between the reflected light and the transmitted light of the red laser will change, and interference will occur when they meet in space, forming interference fringes.

[0018] Step 1.3: Use a photodetector to collect the signal of the light intensity change of the interference fringes generated by the vibration of the adsorption layer 5 over time; through the photodetector, convert the light intensity signal into an electrical signal and send it to a lock-in amplifier for spectral analysis; the electrical signal is subjected to Fourier analysis by the lock-in amplifier to obtain an amplitude-frequency curve. This curve has two peaks. The frequency corresponding to the first peak point is the first-order mode frequency ω 1 of the adsorption layer 5 when vibrating, and the frequency corresponding to the second peak point is the second-order mode frequency ω 2 of the adsorption layer 5 when vibrating; during the process of adjusting the DC voltage until it is observed that the first-order mode frequency ω 1 of the adsorption layer 5 gradually approaches the frequency doubling relationship ω 2 of the second-order mode frequency ω 2 ≈2ω 1 of the adsorption layer 5 when vibrating, complete the adjustment of the DC voltage.

[0019] Step 1.3: Then apply an AC voltage of magnitude V g ac cos(Ωt) to the gate electrode 12 and adjust V g acThe cos(Ωt) generates a non - linear electrostatic force inside the I - shaped gate electrode 12 and the adsorption layer 5, introducing the electrostatic non - linearity into the adsorption layer 5, breaking the structural symmetry, and driving the entire sensor; among them, the driving frequency Ω of the AC voltage needs to match the internal resonance mode of the system, that is, the following relationship should be satisfied: Ω≈ω 1 ; where the control V g ac The amplitude of cos(Ωt) gradually increases between 0V and 1V, and the change of the amplitude - frequency curve is observed. When the adsorption layer 5 is about to enter the non - linear internal resonance region, part of the energy of the first - order mode will transfer to the second - order mode, manifested as a decrease in the amplitude of the first - order mode. At this time, the first peak point does not bifurcate from a single peak to an M - type double peak, and the peak value of the second - order mode increases, completing the adjustment of V g ac cos(Ωt), and record the AC voltage V in this state g ac cos(Ωt);

[0020] Step two, obtain the reference acoustic frequency comb spectrum diagram under vacuum

[0021] Step 2.1, in the absence of NO 2 gas, based on step one, continue to increase the AC voltage V g ac applied to the I - shaped gate electrode 12, observe the amplitude - frequency curve collected by the lock - in amplifier until it is observed that the first peak point bifurcates from a single peak to an M - type double peak. Due to the non - linear behavior of monolayer graphene, an internal resonance interval will be formed under the action of the I - shaped gate voltage 12. The vibration signal of the adsorption layer 5 is measured by a photodetector and a lock - in amplifier to obtain the amplitude curve x(t) changing with time; the collected x(t) is a mixed signal, which contains multiple frequency components. The signal x(t) is subjected to Fourier transform to transfer the signal from the time domain to the frequency domain, and the discrete frequency components in the signal are extracted, manifested as multiple discrete frequency components, that is, the frequency comb spectrum diagram, also known as the acoustic frequency comb. This spectrum diagram consists of n comb teeth. The comb teeth are manifested as a series of discrete frequency components, rather than a continuous frequency band. The magnitude of the amplitude decreases with the increase of the frequency; its abscissa is the frequency f, representing the frequency range of the system response; the ordinate is the amplitude, representing the vibration response intensity of the adsorption layer 5 at different frequencies;

[0022] Step 2.2, observe the main resonance frequency in the acoustic frequency comb, that is, the central frequency f 0 and its higher - order harmonic components f n , in the acoustic frequency comb obtained after Fourier transform, the strongest spectral line, that is, the spectral line with the largest amplitude, corresponds to the main resonance frequency, which is the central comb tooth f 0 ;

[0023] Step 3: Similar to Step 2, except that NO with an unknown concentration is introduced into the test cavity 2 gas, obtaining the acoustic frequency comb spectrogram at this time, and acquiring the main resonance frequency f 0' at this time, and calculating the frequency offset Δf n of the nth-order sub-comb tooth, where Δf n = n·Δf 0 , and the offset Δf 0 of the central frequency f 0 = f 0' - f 0 ;

[0024] Step 4: Calculate and solve for the unknown NO 2 gas concentration according to the following formula

[0025]

[0026] The calculation process of the NO 2 gas concentration in Step 4 is as follows:

[0027] Step 1.1: Inject NO with concentrations ranging from 0 ppb to 10 ppb into the test cavity in ascending order 2 , divide this range into N equally spaced concentrations, and record the corresponding acoustic frequency comb spectrograms at each concentration;

[0028] Step 1.2: Extract the acoustic frequency comb interval: Since NO 2 gas is adsorbed on the surface of the adsorption layer 5, the equivalent mass m eff and the equivalent stiffness k eff of the adsorption layer 5 change, resulting in a shift in the main resonance frequency: Among them, since the adsorption of NO 2 molecules increases the equivalent mass m eff of the adsorption layer 5, the natural resonance frequency decreases; the interaction between NO 2 molecules and graphene affects the nonlinear stiffness term, thereby affecting the higher harmonic frequencies, resulting in a frequency shift. At this time, the main resonance frequency f 0' and its higher harmonic components f n' in the amplitude-frequency curve are collected and observed through a lock-in amplifier, and the offset Δf 0 of the central frequency f 0 = f 0' - f 0 is calculated; when the gas concentration is extremely low, the frequency offset of the nth-order sub-comb tooth, that is, the frequency offset of the sub-comb tooth relative to its original position, is n times the offset of the central frequency comb tooth, that is, Δf n = n·Δf 0 ​, then utilize this amplification effect to improve the detection sensitivity so that it can detect sub-ppb level NO 2 concentration;

[0029] Step 1.3, establish a mathematical model of the NO 2 concentration and the acoustic frequency comb interval

[0030] At each concentration , where i corresponds to the NO at the i-th concentration 2 , i is a positive integer, and its value range is 1 to N. Measure the offset Δf of the main resonance frequency at this gas concentration 0,i , and obtain a set of data points Utilize the amplification effect of the high-order sub-comb teeth to perform mathematical fitting on the NO 2 concentration and Δf n . The relationship between the center frequency offset Δf 0 and the NO 2 gas concentration is as follows:

[0031]

[0032] where k 1 , k 2 , α is obtained by using the nonlinear least squares method to fit the entire data set according to the calibration experiment, and the optimal parameters k 1 , k 2 , α are extracted. The specific fitting method is as follows:

[0033] a. Substitute the collected data into the mathematical model:

[0034]

[0035] b. Construct an objective function to minimize the sum of the squares of the residuals between the fitting curve and the experimental data:

[0036]

[0037] c. Use a numerical optimization algorithm to fit k 1 , k 2 and α. Therefore, the frequency offset of the n-th order sub-comb tooth is:

[0038]

[0039] From this, it can be sorted out as:

[0040]

[0041] The present invention also provides an ultra-low power nanoelectromechanical resonant NO based on an acoustic frequency comb 2A method for preparing a gas sensor, comprising the following process steps:

[0042] Step a, substrate cleaning

[0043] Select a single-sided polished silicon wafer for processing. Use an ultrasonic cleaner to place the single-sided polished silicon wafer in acetone for ultrasonic cleaning for 10 minutes and then take it out, and then place it in isopropyl alcohol for ultrasonic cleaning for 10 minutes. Finally, perform plasma treatment to obtain a single-crystal silicon substrate layer 4. Deposit a 1.5-μm-thick silicon dioxide layer insulating layer 3 on the polished surface of the single-crystal silicon substrate layer 4 by low-pressure chemical vapor deposition to complete the preparation of the oxidized wafer;

[0044] Step b, spin coating

[0045] Perform pretreatment on the oxidized wafer prepared in step a: bake it in a high-temperature furnace at 120 °C for 10 minutes to remove surface moisture;

[0046] Use a spin coater for spin coating: adsorb the non-polished surface of the oxidized wafer on the spin coating tray, drop AR80 photoresist on the silicon dioxide layer insulating layer 3, and the spin coating parameters are 500 r / 5 s, 4000 r / 30 s, and bake at 95 °C / 90 S; use a Nikon I12 lithography machine for exposure, with an exposure time of 270 ms, use a 2.38% TMAH developer for development for 40 s, and develop to form a lithography pattern, so that the photoresist in the area where the I-shaped trench 2 is located after lithography is removed;

[0047] Step c, use an AOE etcher, place the sample prepared in step b on the etching table, set the power to 1000 W, the gas CHF3: 30 sccm, and the etching time to 40 S. Etch a 1-μm-thick I-shaped trench 2 on the silicon dioxide layer insulating layer 3 by dry etching, and then remove the remaining photoresist on the silicon wafer;

[0048] Step d, use a spin coater for spin coating: adsorb the non-polished surface of the oxidized wafer on the spin coating tray, drop AR80 photoresist on the silicon dioxide layer insulating layer 3, and the spin coating parameters are 500 r / 5 s, 4000 r / 30 s, and bake at 95 °C / 90 S; use a Nikon I12 lithography machine for exposure, with an exposure time of 270 ms, use a 2.38% TMAH developer for development for 40 s, and develop to form a lithography pattern, so that the photoresist in the area where the gate electrode 12 is located after lithography is removed, and the photoresist in the remaining areas is retained as a mask layer;

[0049] Step e, deposit a gate electrode 12 with a thickness of 800 nm on the silicon dioxide layer insulating layer 3 by magnetron sputtering deposition to complete the preparation of the gate electrode 12, and then remove the photoresist retained in step d;

[0050] Step f, spin coating to even the photoresist: Adsorb the non-polished surface of the oxidized wafer on the spin coating tray, drop AR80 photoresist on the silicon dioxide layer insulating layer 3, with spin coating parameters of 500 r / 5 s, 4000 r / 30 s, and bake at 95 °C for 90 S; Use a Nikon I12 lithography machine for exposure, with an exposure time of 270 ms, use a 2.38% TMAH developer, and develop for 40 s to form a lithography pattern, such that the photoresist in the areas where the drain electrode 11 and source electrode 13 are located after lithography is removed;

[0051] Step g, deposit a 100-nm-thick drain electrode 11 and source electrode 13 on the silicon dioxide layer insulating layer 3 respectively by magnetron sputtering deposition method to complete the preparation of the drain electrode 11 and source electrode 13, and then remove the remaining photoresist;

[0052] Step h, transfer the single-layer graphene to PMMA by dry stripping method to prepare for dry transfer of the single-layer graphene;

[0053] Step i, transfer the single-layer graphene on PMMA to the corresponding position on the silicon dioxide layer insulating layer 3 to form an adsorption layer 5, and finally use a DISCO cutting machine to perform dicing on the single-crystalline silicon substrate layer 4 at a rotational speed of 20,000 revolutions and a cutting rate of 2 mm / s to complete the preparation of the NO 2 preparation of the gas sensor.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0055] 1. The present invention adopts a resonant sensor, and by using a single-layer graphene film, it realizes specific recognition of NO 2 gas.

[0056] 2. The resonant sensor used in the present invention, by adopting fixed-frequency driving to replace the traditional frequency-sweeping driving or feedback driving, can greatly simplify the driving circuit and reduce the sensing power consumption to below the μW level;

[0057] 3. Utilize the in-resonance excitation frequency comb of the first mode and the second mode, and at the same time utilize the frequency locking and frequency doubling characteristics of the acoustic frequency comb to accurately amplify the frequency response change caused by gas adsorption, and multiply the detection sensitivity to achieve ppb-level gas detection. Description of the Drawings

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings to be used in the description of the embodiments of the present invention. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the content of the embodiments of the present invention and these drawings.

[0059] Figure 1 Schematic structural diagram of the resonant gas sensor of the present invention;

[0060] Figure 2 Top view of the structure of the resonant gas sensor of the present invention;

[0061] Figure 3 Schematic diagram of the second-order mode of the present invention;

[0062] Figure 4 Schematic diagram of the amplitude-frequency curve of the present invention;

[0063] Figure 5 Flow chart of the micro-nano processing technology of the resonant gas sensor according to the embodiment of the present invention;

[0064] Figure 6 Principle diagram of gas concentration detection according to the embodiment of the present invention; Figure 7 Flow chart of the measurement method according to the embodiment of the present invention;

[0065] Wherein: 11 drain electrode, 12 gate electrode, 13 source electrode, 2 I-shaped groove, 3 silicon dioxide insulating layer, 4 single-crystalline silicon substrate, 5 single-layer graphene adsorption layer. Detailed implementation manners

[0066] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. In addition, it should be noted that for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all the structures.

[0067] Embodiment 1

[0068] Ultra-low-power nano-electromechanical resonant NO 2 gas sensor based on acoustic frequency comb, including three gold electrodes (drain electrode 11, I-shaped gate electrode 12, source electrode 13), silicon dioxide insulating layer 3, single-crystalline silicon substrate 4 and adsorption layer 5; wherein the drain electrode 11 and the source electrode 13 are respectively arranged on the silicon dioxide insulating layer 3, and an I-shaped groove 2 is embedded in the silicon dioxide insulating layer 3 between the drain electrode 11 and the source electrode 13, and an I-shaped gate electrode 12 is arranged in the I-shaped groove 2, and the adsorption layer 5 is laid above the front end of the silicon dioxide insulating layer 3 and covers the front-end surfaces of the drain electrode 11 and the source electrode 13, and the middle part of the adsorption layer 5 is simultaneously suspended above the front end of the I-shaped gate electrode 12 (that is, the middle part of the adsorption layer 5 is suspended above the I-shaped groove 2), and the silicon dioxide insulating layer 3 is arranged on the single-crystalline silicon substrate 4.

[0069] The adsorption layer 5 is a single-layer graphene layer.

[0070] The adsorption layer 5 is a double-ended fixed beam structure. The drain electrode 11 and the source electrode 13 have the same shape and size, and are symmetrically arranged on both sides of the silicon dioxide insulating layer 3.

[0071] The single-crystalline silicon substrate layer 4, the silicon dioxide insulating layer 3, the drain electrode 11, the I-shaped gate electrode 12, and the source electrode 13 are integrated components.

[0072] The adsorption layer 5 is a monolayer graphene specific sensitive film, which is preferably obtained from the NO 2 gas to be detected.

[0073] A method for measuring the concentration of NO 2 gas using an ultra-low power nanoelectromechanical resonant NO 2 gas sensor based on an acoustic frequency comb specifically includes the following steps:

[0074] Step 1: Calibrate the first and second order modes of the vibration of the gas sensor using an optical method.

[0075] Step 1.1: Conduct all measurements in a vacuum test chamber at 370K and 1×10 -5 Torr. Select a 633nm red laser as the detection light source and irradiate the laser beam onto the adsorption layer 5.

[0076] Step 1.2: Apply a DC voltage to the I-shaped gate electrode 12, starting from 0V and gradually increasing. The gate electrode 12 and the adsorption layer 5 form a capacitor. Due to the action of the DC voltage of the gate electrode 12, an electrostatic field is formed between the gate electrode 12 and the adsorption layer 5. This electric field acts on the adsorption layer 5 through electrostatic attraction, and the adsorption layer 5 is stretched by the electrostatic force and its initial tension and deformation state are changed. The resonance frequency of the adsorption layer 5 is modulated accordingly. Thus, by adjusting the magnitude of V g dc the resonance frequency of the monolayer graphene adsorption layer 5 can be effectively adjusted. When the adsorption layer 5 vibrates, the phase difference between the reflected light and the transmitted light of the 633nm red laser changes, and they interfere with each other in space to form interference fringes. The vibration amplitude and frequency of the adsorption layer 5 are measured through the change of the interference fringes. Connect the photodetector to the lock-in amplifier, and use the input end of the lock-in amplifier to receive the signal from the photodetector.

[0077] Step 1.3: Use the photodetector to collect the signal of the change of the light intensity of the interference fringes generated by the vibration of the adsorption layer 5 over time, the I(t) curve. Through the photodetector, the light intensity signal is converted into an electrical signal V(t) and sent to the lock-in amplifier for spectral analysis. The electrical signal V(t) is subjected to Fourier analysis by the lock-in amplifier to obtain an amplitude-frequency curve. This curve has two peaks, and the frequency corresponding to the first peak point is the first order mode frequency ω 1, the frequency corresponding to the second peak point is the second-order modal frequency ω of the adsorption layer 5 during vibration 2 ; During the process of adjusting the DC voltage until the first-order modal frequency ω of the adsorption layer 5 is observed 1 gradually approaches the second-order modal frequency ω 2 of the multiple frequency relationship ω 2 ≈ 2ω 1 When, complete the adjustment of the DC voltage, record the DC voltage at this time, denoted as V g dc ;

[0078] Step 1.3, then apply an AC voltage of V g ac cos(Ωt) to the gate electrode 12 (Ω: excitation frequency, t is time), adjust V g ac cos(Ωt) to generate a non-linear electrostatic force inside the I-shaped gate electrode 12 and the adsorption layer 5, introduce the electrostatic non-linearity into the adsorption layer 5, break the structural symmetry, and drive the entire sensor; among them, the driving frequency Ω of the AC voltage needs to match the internal resonance mode of the system, that is, the following relationship needs to be satisfied: Ω ≈ ω 1 ; Among them, control V g ac The amplitude of cos(Ωt) gradually increases between 0V and 1V, observe the change of the amplitude-frequency curve. When the adsorption layer 5 is about to enter the non-linear internal resonance region, part of the energy of the first-order mode will be transferred to the second-order mode, manifested as a decrease in the amplitude of the first-order mode (part of the energy is transferred). At this time, the first-order mode does not split, and the first peak point does not bifurcate from a single peak to an M-shaped double peak. The peak value of the second-order mode increases (energy is coupled to the higher-order mode), complete the adjustment of V g ac cos(Ωt), and record the AC voltage V g ac cos(Ωt) in this state; The AC component is mainly used for dynamic excitation. The applied AC voltage periodically excites the single-layer graphene under the action of the electric field, so as to generate vibration near its natural frequency. By adjusting Ω (excitation frequency) and V g ac The amplitude of cos(Ωt), the single-layer graphene can be accurately driven into the resonance state; By static regulation, adjust the magnitude of V g ac cos(Ωt) to observe the change of the resonance frequency of the sensor, making the subsequent dynamic response of the sensor more sensitive and stable;

[0079] Step 1.4, collect the amplitude-frequency curve at this time, and record the DC voltage V g dcand the AC voltage V applied to the gate electrode 12 in Step 1.3 g ac cos(Ωt), as the reference voltage for the next step; record the first-order modal frequency ω in Step 1.2 1 , the second-order modal frequency ω 2 Complete the calibration of the first-order and second-order modes of the gas sensor vibration.

[0080] Step 2: Obtain the reference audio frequency comb spectrogram under vacuum

[0081] Verify that the experimental environment of the gas test cavity is stable (temperature, humidity), and perform background correction using zero gas; Measure the initial state: in the absence of NO 2 gas, obtain the frequency comb signal, and extract the initial frequency comb interval; The specific content is as follows:

[0082] Step 2.1, in the absence of NO 2 gas, based on Step 1, continue to increase the AC voltage V applied to the I-shaped gate electrode 12 g ac cos(Ωt), observe the amplitude-frequency curve collected by the lock-in amplifier until it is observed that the first peak point (first-order mode) bifurcates from a single peak to an M-shaped double peak. Due to the nonlinear behavior of single-layer graphene, an internal resonance interval will be formed under the action of the I-shaped gate voltage 12. Measure the vibration signal of the adsorption layer 5 through a photodetector and a lock-in amplifier to obtain the amplitude curve x(t) that changes with time, and its amplitude is manifested as the magnitude of the voltage amplitude; The collected x(t) is a mixed signal that contains multiple frequency components. Perform a Fourier transform on the signal x(t) to transfer the signal from the time domain to the frequency domain, and extract the discrete frequency components in the signal, which are manifested as multiple discrete frequency components, that is, the frequency comb tooth spectrogram, also known as the audio frequency comb. This spectrogram consists of n comb teeth, and the comb teeth are manifested as a series of discrete frequency components, rather than a continuous frequency band. The magnitude of the amplitude decreases as the frequency increases; Its abscissa (X-axis) is the frequency f (Hz), representing the frequency range of the system response; The ordinate (Y-axis) is the amplitude (V), representing the vibration response intensity of the adsorption layer 5 at different frequencies;

[0083] Step 2.2, observe the main resonance frequency in the audio frequency comb, that is, the central frequency f 0 and its higher-order harmonic components f n . In the audio frequency comb obtained after Fourier transform, the strongest spectral line, that is, the spectral line with the largest amplitude, corresponds to the main resonance frequency, which is the central comb tooth f 0 , and on both sides of the central comb tooth f 0 are the sub-comb teeth f n and f -n (here the negative sign is only for the higher-order harmonic components with respect to f 0The left - right positional relationship), and the interval between any two adjacent sub - comb teeth is f s = f n -f n-1 , thus obtaining the reference audio frequency comb spectrum and the reference audio frequency comb interval;

[0084] Step 3: Introduce NO with different known concentrations 2 into the gas environment to calibrate the sensor, and establish the calculation formula between and △f;

[0085] Confirm that the experimental environment in the gas test chamber is stable (temperature, humidity). Use a gas mass flow controller (MFC) to generate NO with different concentrations 2 and perform background correction with zero gas; The specific content is as follows:

[0086] Same as step 2, the difference is that NO with an unknown concentration is introduced into the test chamber 2 to obtain the audio frequency comb spectrum at this time, and obtain the main resonance frequency f 0' at this time, and calculate the frequency offset △f of the n - th order sub - comb teeth n , where △f n = n·△f 0 , and the offset △f of the center frequency f 0 is 0 = f 0' -f 0 ;

[0087] Step 4: Calculate and solve the unknown NO 2 gas concentration according to the following formula

[0088]

[0089] The calculation process of the NO 2 gas concentration in step 4 is as follows:

[0090] Step 1.1: Inject NO with concentrations ranging from 0 ppb to 10 ppb into the test chamber in ascending order 2 2 , divide this range into N equally spaced concentrations (i.e., N concentration gradients), and record the corresponding audio frequency comb spectra at each concentration;

[0091] Step 1.2: Extract the audio frequency comb interval: Since NO 2 gas is adsorbed on the surface of the adsorption layer 5, the equivalent mass m eff and the equivalent stiffness k eff of the adsorption layer 5 change, resulting in a shift in the main resonance frequency: Among them, since NO 2The adsorption of molecules increases the equivalent mass m of the adsorption layer 5 eff , resulting in a decrease in the natural resonance frequency; NO 2 The interaction between NO molecules and graphene affects the nonlinear stiffness term, and thus affects the high-order harmonic frequency, resulting in a frequency shift. At this time, the main resonance frequency f in the amplitude-frequency curve is collected and observed through a lock-in amplifier 0' and its high-order harmonic components f n' , and the offset △f of the center frequency f 0 is calculated; when the gas concentration is extremely low, the frequency offset of the nth-order sub-comb tooth, that is, the frequency offset of the sub-comb tooth relative to its original position, is n times the frequency offset of the center frequency comb tooth, that is, △f 0 = f 0' - f 0 ; when the gas concentration is extremely low, the frequency offset of the nth-order sub-comb tooth, that is, the frequency offset of the sub-comb tooth relative to its original position, is n times the frequency offset of the center frequency comb tooth, that is, △f n = n·△f 0 , then using this amplification effect, the detection sensitivity is improved so that it can detect sub-ppb-level NO 2 concentration;

[0092] Step 1.3, establish a mathematical model of the NO 2 concentration and the acoustic frequency comb interval

[0093] At each concentration where i corresponds to NO at the ith concentration 2 , i is a positive integer, and the value range is 1 to N (N concentration gradients), measure the offset △f of the main resonance frequency at this gas concentration 0,i , and obtain a set of data points Using the amplification effect of the high-order sub-comb teeth, the NO 2 concentration and △f n are mathematically fitted. The relationship between the center frequency offset △f 0 and the NO 2 gas concentration is as follows:

[0094]

[0095] where k 1 , k 2 , α is fitted to the entire data set using the nonlinear least squares method according to the calibration experiment, and the optimal parameters k 1 , k 2 , α are extracted, where k 1 (sensitivity coefficient) represents the sensitivity of the frequency offset amplitude of the corresponding center frequency comb tooth (main resonance frequency) when the NO 2 concentration changes, and k 1 measures the contribution of the NO 2 concentration to the change in the acoustic frequency comb interval; k2 (Baseline shift or zero drift) reflects the inherent frequency shift that may still exist in the system when the NO 2 concentration is zero (e.g., due to temperature drift, internal imbalance of the device, noise, etc.). k 2 provides a baseline correction for the entire model, enabling the model to more accurately match the experimental data. α (nonlinear index) reflects the nonlinear relationship between the NO 2 concentration and the change in the acoustic frequency comb interval. If α = 1, it indicates a linear relationship; if α ≠ 1, it indicates that the sensor has a power-law type of nonlinear characteristic in its response to NO 2 . This index usually reveals the influence of factors such as the adsorption process, equivalent mass change, or nonlinear electrostatic interaction on the sensor response. The specific fitting method is as follows:

[0096] a. Substitute the collected data into the mathematical model:

[0097]

[0098] b. Construct an objective function to minimize the sum of the squares of the residuals between the fitting curve and the experimental data:

[0099]

[0100] c. Use a numerical optimization algorithm to fit k 1 , k 2 and α. Therefore, the frequency shift of the nth-order sub-comb tooth is:

[0101]

[0102] After organizing, we can obtain:

[0103]

[0104] The present invention also provides a preparation method for an ultra-low-power nanoelectromechanical resonant NO 2 gas sensor based on an acoustic frequency comb, including the following process steps:

[0105] Step a, substrate cleaning

[0106] Select a single-side polished silicon wafer for processing. Use an ultrasonic cleaning machine to place the single-side polished silicon wafer in acetone for ultrasonic cleaning for 10 minutes and then take it out. Then place it in isopropanol for ultrasonic cleaning for 10 minutes. Finally, perform plasma treatment to obtain a single-crystalline silicon base layer 4. Deposit a 1.5-μm silicon dioxide layer insulating layer 3 on the polished surface of the single-crystalline silicon base layer 4 through low-pressure chemical vapor deposition to complete the preparation of the oxidized wafer;

[0107] Step b, coating with glue

[0108] Pre-treat the oxidized wafer prepared in step a: bake it in a high-temperature furnace at 120 °C for 10 min to remove the surface moisture;

[0109] Spin-coat the photoresist: adsorb the non-polished surface of the oxidized wafer on the spin-coating tray, drop the AR80 photoresist on the silicon dioxide layer insulating layer 3, and the spin-coating parameters are 500 r / 5 s, 4000 r / 30 s, and bake at 95 °C / 90 S; expose it using a Nikon I12 lithography machine (1# mask), with an exposure time of 270 ms, use a 2.38% TMAH developer solution, develop for 40 s, and develop to form a lithography pattern, so that the photoresist in the area where the I-shaped trench 2 is located after lithography is removed;

[0110] In step c, use an AOE etching machine, place the sample prepared in step b on the etching table, set the power to 1000 W, the gas CHF3: 30 sccm, and the etching time to 40 S. Dry-etch an I-shaped trench 2 with a thickness of 1 μm on the silicon dioxide layer insulating layer 3, and then remove the remaining photoresist on the silicon wafer;

[0111] Spin-coat the photoresist in step d: adsorb the non-polished surface of the oxidized wafer on the spin-coating tray, drop the AR80 photoresist on the silicon dioxide layer insulating layer 3, and the spin-coating parameters are 500 r / 5 s, 4000 r / 30 s, and bake at 95 °C / 90 S; expose it using a Nikon I12 lithography machine (2# mask), with an exposure time of 270 ms, use a 2.38% TMAH developer solution, develop for 40 s, and develop to form a lithography pattern, so that the photoresist in the area where the gate electrode 12 is located after lithography is removed, and the photoresist in the remaining areas is retained as a mask layer;

[0112] In step e, deposit a gate electrode 12 with a thickness of 800 nm on the silicon dioxide layer insulating layer 3 by magnetron sputtering deposition. The gate electrode 12 consists of a 80-nm-thick chromium deposited first as an adhesion layer, and then a 720-nm-thick gold deposited on the adhesion layer as a conductive layer. Chromium is used as the adhesion layer and gold is used as the conductive layer to complete the preparation of the gate electrode 12. Then remove the photoresist retained in step d;

[0113] Spin-coat the photoresist in step f: adsorb the non-polished surface of the oxidized wafer on the spin-coating tray, drop the AR80 photoresist on the silicon dioxide layer insulating layer 3, and the spin-coating parameters are 500 r / 5 s, 4000 r / 30 s, and bake at 95 °C / 90 S; expose it using a Nikon I12 lithography machine (3# mask), with an exposure time of 270 ms, use a 2.38% TMAH developer solution, develop for 40 s, and develop to form a lithography pattern, so that the photoresist in the area where the drain electrode 11 and the source electrode 13 are located after lithography is removed;

[0114] Step g, depositing a 100 nm thick drain electrode 11 and a source electrode 13 on the silicon dioxide insulating layer 3 by magnetron sputtering deposition, wherein both electrodes are composed of 10 nm chromium and 90 nm gold, wherein chromium is first sputtered as an adhesion layer, and then gold is sputtered as a conductive layer, to complete the preparation of the drain electrode 11 and the source electrode 13, and then removing the remaining photoresist;

[0115] Step h, transferring the single-layer graphene to PMMA by dry exfoliation to prepare for dry transfer of the single-layer graphene;

[0116] Step i, transfer the single-layer graphene on PMMA to the corresponding position on the silicon dioxide insulating layer 3 to form an adsorption layer 5, and finally use a DISCO cutting machine to scribing on the single-crystalline silicon base layer 4, with a rotation speed of 20,000 rpm and a cutting rate of 2 mm / s to complete NO 2 Preparation of gas sensors.

[0117] Example 2

[0118] See also Figure 1 , Figure 2 and Figure 3 , the embodiment of the present invention includes:

[0119] like Figure 1 As shown, an ultra-low power NOx based on the regulation of the resonant acoustic frequency comb in the resonance band of two-dimensional graphene 2 The gas sensor and its sensing mechanism include 11 a drain electrode, 12 a gate electrode, 13 a source electrode, 2 an I-shaped groove, 3 a silicon dioxide insulating layer, 4 a single crystal silicon substrate, and 5 a single-layer graphene adsorption layer, wherein the silicon dioxide insulating layer 3 is on the single crystal silicon substrate 4, the I-shaped groove 2 is on the silicon dioxide insulating layer 3, and a gate electrode 12 is deposited inside the silicon dioxide insulating layer 3, the drain electrode 11 and the source electrode 13 are located on the top layer of the silicon dioxide insulating layer 3 and distributed on both sides of the I-shaped groove, and the single-layer graphene adsorption layer 5 is located above the gate electrode, suspended on the I-shaped groove, and the two ends are respectively fixed to the drain electrode and the source electrode.

[0120] The drain electrode 11 and the source electrode 13 are both chromium-gold electrodes and have the same shape and size.

[0121] The drain electrode 11 , the gate electrode 12 , the source electrode 13 , the I-shaped groove 2 , the silicon dioxide insulating layer 3 , and the single crystal silicon substrate 4 are an integrated part and are manufactured by MEMS technology.

[0122] The single-layer graphene adsorption layer 5 is a specific sensitive film, which is required to detect NO 2 Gas is preferably obtained.

[0123] According to relevant research, single-layer graphene has a significant effect on NO 2The specific adsorption of gases mainly involves their unique physical and chemical properties. Graphene is a two-dimensional single-layer honeycomb structure formed by carbon atoms arranged in sp 2 hybrid orbitals, with a high specific surface area, electron mobility, and chemical stability, which make it a good gas sensing material. The principle of specific adsorption of NO 2 mainly includes the following aspects:

[0124] π-π interaction: π-electron clouds on the surface of graphene can interact with the unpaired electrons of NO 2 molecules to produce π-π interactions. Since NO 2 is an electron acceptor (electron-withdrawing molecule), it will interact with the π-electrons on the surface of graphene, resulting in the redistribution of electrons in graphene.

[0125] Charge transfer: NO 2 molecules have a strong electron-withdrawing ability. After adsorbing on the surface of graphene, NO 2 will extract electrons from graphene, causing charge transfer. This charge transfer will change the conductivity of graphene, making its resistance change significantly, thus realizing the detection of NO 2 . This conductivity change characteristic enables graphene to sensitively detect NO 2 molecules.

[0126] Surface active sites: Although the ideal graphene surface is inert, the actual graphene surface often has defects or edge sites (such as the edges, vacancies, or wrinkles of graphene), and these sites can provide more adsorption sites for NO 2 , thus enhancing the adsorption capacity for NO 2 .

[0127] Adsorption kinetics: Compared with other gases (such as O 2 , CO 2 , etc.), the adsorption of NO 2 on graphene is a rapid process with a relatively high adsorption energy. This makes the adsorption of NO 2 by graphene more specific.

[0128] Reversibility of response: The adsorption of NO 2 by graphene is usually reversible. By heating or introducing an inert gas, the adsorbed NO 2 molecules can be desorbed, thus restoring the original state of graphene.

[0129] The present invention provides a low-power consumption NO 2 gas sensor processing technology based on the regulation of resonance acoustic frequency comb in the two-dimensional graphene resonance band, as Figure 3 shown, including the following process steps:

[0130] Step a, substrate cleaning. Using an ultrasonic cleaner, place the silicon substrate in acetone and ultrasonically clean it for 10 minutes, then take it out and place it in isopropyl alcohol for ultrasonic cleaning for 10 minutes, and finally perform plasma treatment. Deposit a 1.5-μm silicon dioxide layer insulating layer 3 on the upper surface of the single-crystalline silicon substrate layer 4 by low-pressure chemical vapor deposition to complete the preparation of the oxidation wafer.

[0131] Step b, spin coating. Pretreat the oxidation wafer prepared in step a: bake it in a high-temperature furnace at 120 °C for 10 minutes to remove surface moisture; use a spin coater for spin coating: adsorb the oxidation wafer on the spin coating tray, drop AR80 photoresist on the sample surface (silicon dioxide layer insulating layer 3), spin coating parameters: 500 r / 5 s, 4000 r / 30 s, bake at 95 °C / 90 S; use a Nikon I12 lithography machine for exposure (1# mask), exposure time: 270 ms, use 2.38% TMAH developer, develop for 40 s, and develop to form a lithography pattern, so that the photoresist in the area where the I-shaped trench 2 is located after lithography is removed;

[0132] Step c, use an AOE etching machine, place the sample prepared in step b on the etching table, set the power to 1000 W, the gas CHF3: 30 sccm, and the etching time: 40 S. Etch a 1-μm I-shaped trench 2 on the silicon dioxide layer insulating layer 3 by dry etching, and then remove the remaining photoresist on the silicon wafer;

[0133] Step d, spin coating using a spin coater: adsorb the oxidation wafer on the spin coating tray, drop AR80 photoresist on the sample surface (silicon dioxide layer insulating layer 3), spin coating parameters: 500 r / 5 s, 4000 r / 30 s, bake at 95 °C / 90 S; use a Nikon I12 lithography machine for exposure (2# mask), exposure time: 270 ms, use 2.38% TMAH developer, develop for 40 s, and develop to form a lithography pattern, so that the photoresist in the area where the gate electrode 12 is located after lithography is removed;

[0134] Step e, deposit an 800-nm gate electrode 12 on the silicon dioxide layer insulating layer 3 by magnetron sputtering. The gate electrode 12 is composed of a first deposited 80-nm chromium as an adhesion layer and a second deposited 720-nm gold as a conductive layer. Among them, chromium is used as the adhesion layer and gold is used as the conductive layer to complete the preparation of the gate electrode 12, and then remove the remaining photoresist;

[0135] Step f, spin coating: Adsorb the oxidized wafer on the spin coating tray, drop AR80 photoresist on the surface of the sample (silicon dioxide layer insulating layer 3). Spin coating parameters: 500 r / 5 s, 4000 r / 30 s, bake at 95 °C for 90 S; Use Nikon I12 lithography machine for exposure (3# mask), exposure time: 270 ms, use 2.38% TMAH developer, develop for 40 s, develop to form a lithography pattern, so that the photoresist in the areas where the drain electrode 11 and source electrode 13 are located after lithography is removed;

[0136] Step g, deposit a 100-nm drain electrode 11 and source electrode 13 on the silicon dioxide layer insulating layer 3 by magnetron sputtering. The electrodes are composed of 10 nm of chromium and 90 nm of gold. First, sputter chromium as an adhesion layer, and then sputter gold as a conductive layer to complete the preparation of the drain electrode 11 and source electrode 13. Subsequently, remove the remaining photoresist;

[0137] Step h, transfer the single-layer graphene to PMMA by dry peeling to prepare for dry transfer of the single-layer graphene;

[0138] Step i, transfer the single-layer graphene on PMMA to the chip unit to form a single-layer graphene adsorption layer 5. Finally, use a DISCO cutter to slice the NO 2 gas sensor, rotation speed 20000 rpm, cutting rate 2 mm / s, complete the preparation of the NO 2 gas sensor.

[0139] An ultra-low-power NO gas sensing mechanism based on the regulation of the resonant acoustic frequency comb within the two-dimensional graphene resonance band: 2 Place the sensor of this embodiment in the environment of the gas to be measured, and select the specific sensitive film category on the adsorption layer 5 according to the gas to be measured NO. As

[0140] shown, drive the two-dimensional resonance band at a fixed frequency Ω within the acoustic frequency comb frequency range. Adsorb the NO gas equivalent to a mass perturbation m 2 and a stiffness change k Figure 4 . The interaction between NO molecules and graphene affects the non-linear stiffness term, thereby affecting the high-order harmonic frequency and causing a frequency shift. At this time, the amplitude-frequency curve can be collected and observed through a lock-in amplifier, and the evolution law of the acoustic frequency comb spacing Δf under the mass and stiffness perturbations can be obtained through Fourier transform (FFT); the main resonance frequency f 2 (center frequency f eff ) and its high-order harmonic components f eff , and calculate the offset of the center frequency f 2 as Δf 0' (center frequency f 0 ) and its high-order harmonic components f n' , and calculate the offset of the center frequency f 0 as Δf 0= f 0' -f 0 。Analyze the response amplification mechanism of the frequency locking and frequency doubling characteristics of the audio frequency comb to gas perturbations. For the NO 2 sensor based on audio frequency comb detection, the nth sub-comb tooth of the audio frequency comb corresponds to the frequency doubling component of different orders n. When the gas concentration is extremely low, the offset of the nth order sub-comb tooth is n times the offset of the central frequency comb tooth, that is, △f n = n·△f 0 , then this amplification effect can be utilized to improve the detection sensitivity so that it can detect sub-ppb level NO 2 concentration.

[0141] The working steps of the present invention are as follows:

[0142] Adopt an ultra-low power nanoelectromechanical resonant NO 2 gas sensor and method based on audio frequency comb, and the steps are as follows:

[0143] Step 1, use the optical method to calibrate the first-order and second-order modes of vibration of the resonant NO 2 gas sensor:

[0144] The optical interference measurement method uses the interference phenomenon of 633 nm red laser to detect the vibration of the adsorption layer 5. All measurements are carried out under a vacuum of 370 K and 1×10 -5 Torr. Select 633 nm red laser as the detection light source. Irradiate the laser beam onto the adsorption layer 5. Part of the light will be reflected and part of the light will be transmitted. Use a beam splitter to guide the reflected light and the transmitted light to the same convergence position respectively, so that the two beams of light coincide in space to form an interference field. When the adsorption layer 5 vibrates, the phase difference between the reflected light and the transmitted light of the 633 nm red laser will change, and interference will occur when they meet in space, and interference fringes will be formed; by the change of the interference fringes, the vibration amplitude and frequency of the adsorption layer 5 can be measured; the vibration of the adsorption layer 5 is detected by a photodetector, and the signal of the light intensity of the interference fringes changing with time is collected. The photodetector converts the light intensity signal into an electrical signal, and the electrical signal is swept by a lock-in amplifier to obtain an amplitude-frequency curve, and there are two peak points on this curve. Specific method:

[0145] Apply a DC voltage of magnitude V g dc to the gate electrode 12. The DC voltage generates a stable static electric field on the gate. This electric field acts on the single-layer graphene through electrostatic attraction, changing its initial tension and deformation state. By adjusting the magnitude of V g dc , the reference resonance frequency of the single-layer graphene can be effectively adjusted to achieve preset frequency control. Then apply a DC voltage of magnitude V g acAn AC voltage of cos(Ωt) (Ω: excitation frequency), and the AC component is mainly used for dynamic excitation. The applied AC voltage periodically excites the single-layer graphene under the action of an electric field, thereby generating vibrations near its natural frequency. By adjusting Ω (excitation frequency) and V g ac the amplitude of cos(Ωt), the single-layer graphene can be accurately driven into the resonance state. By static regulation, adjust V g dc Observe the change of the resonance frequency of the sensor by adjusting the magnitude of cos(Ωt), making the subsequent dynamic response of the sensor more sensitive and stable. Adjust the alternating voltage to generate a non-linear electrostatic force inside the adsorption layer 5 of the gate electrode 12, introduce the electrostatic non-linearity into the adsorption layer 5, break the structural symmetry, and at the same time drive the entire resonator into the non-linear internal resonance region; when the adsorption layer 5 vibrates, the phase difference between the reflected light and the transmitted light of the 633 nm red laser will change, meet in space and interfere, and form interference fringes; through the change of the interference fringes, the vibration amplitude and frequency of the adsorption layer 5 are measured; the vibration of the adsorption layer 5 is detected by a photodetector. Due to the vibration of the adsorption layer 5, the optical path of the reflected light and the transmitted light changes periodically, resulting in a modulation phenomenon of the interference fringes changing with time. Collect the signal of the light intensity of the interference fringes changing with time, that is, the I(t) curve. Use a high-speed photodetector to convert the light intensity signal into an electrical signal V(t), and send it to a lock-in amplifier for spectral analysis. The electrical signal V(t) undergoes Fourier analysis by the lock-in amplifier A(f) = F[V(t)] (F represents the Fourier transform) to obtain the amplitude-frequency curve (amplitude-frequency curve A(f)). This curve has two peaks. The frequency corresponding to the first peak point is the frequency ω of the first-order mode when the adsorption layer 5 vibrates 1 , and the frequency corresponding to the second peak point is the second-order mode frequency ω of the adsorption layer 5 when it vibrates 2 , completing the calibration of the first-order and second-order modes.

[0146] Step 2: Obtain the reference acoustic frequency comb spectrogram under vacuum.

[0147] The gas test cavity verifies that the experimental environment is stable (temperature, humidity), and zero gas is used for background correction. Measurement initial state: In the absence of NO 2 gas, obtain the frequency comb signal and extract the initial frequency comb interval. Specific method:

[0148] Increase the voltage applied to the enhanced gate electrode 12 and observe the amplitude-frequency curve collected by the lock-in amplifier until the first peak point is observed to bifurcate from a single peak to an M-shaped double peak. Due to the nonlinear behavior of monolayer graphene, an internal resonance interval will be formed under the action of the gate voltage. Perform Fourier analysis on the amplitude-frequency curve at this time, which is manifested as multiple discrete frequency components (frequency comb spectrum, also known as acoustic frequency comb: this spectrum consists of n comb teeth, abscissa (X-axis): frequency f (Hz) represents the frequency range of the system response; ordinate (Y-axis): amplitude (V) represents the vibration response intensity of the graphene film at different frequencies.). Observe the main resonance frequency f 0 (central frequency f 0 ) and its higher harmonic components f n . The highest among them is the central comb tooth f 0 , and on both sides are f n and f -n (here the negative sign only represents the left-right position relationship of the higher harmonic component with respect to f 0 ). At this time, the interval between any two adjacent acoustic frequency combs is f s = f n - f n-1 . Thus, obtain the reference acoustic frequency comb spectrum and the reference acoustic frequency comb interval.

[0149] Step 3: Introduce NO 2 gas to calibrate the sensor and establish the calculation formula between and △f.

[0150] Confirm the stability of the experimental environment (temperature, humidity) in the gas test cavity. Use a gas mass flow controller (MFC) to generate different concentrations of NO 2 and perform background correction with zero gas.

[0151] 1. Measure the change in the acoustic frequency comb after the action of NO 2 : Control the concentration of NO 2 gas in the cavity (from 0 ppb to 10 ppb, increasing by 0.5 ppb each time), and inject it in sequence from low to high, and record the corresponding acoustic frequency comb spectrum at each concentration.

[0152] 2. Extract the acoustic frequency comb interval: Since NO 2 gas is adsorbed on the surface of monolayer graphene, the equivalent mass m eff and the equivalent stiffness k eff of graphene change, resulting in a shift in the resonance frequency: Among them, the adsorption of NO 2 molecules increases m eff , resulting in a decrease in the natural resonance frequency. NO 2The interaction between the molecule and graphene affects the non - linear stiffness term, which in turn affects the high - order harmonic frequency, resulting in a frequency shift. At this time, the lock - in amplifier can be used to collect and observe the main resonance frequency f in the amplitude - frequency curve 0' (central frequency f 0 ) and its high - order harmonic components f n' , and calculate the offset of the central frequency f 0 as △f 0 = f 0' - f 0 . For the NO 2 sensor based on the acoustic frequency comb detection, the n - th sub - comb tooth of the acoustic frequency comb corresponds to the harmonic component of different orders n. When the gas concentration is extremely low, the offset of the n - th order sub - comb tooth is n times the offset of the central frequency comb tooth, that is, △f n = n·△f 0 , then this amplification effect can be utilized to improve the detection sensitivity so that it can detect sub - ppb - level NO 2 concentration.

[0153] 3. Establish a mathematical model between NO 2 concentration and the acoustic frequency comb interval

[0154] Using the amplification effect of the high - order sub - comb teeth, perform a mathematical fitting between the NO 2 concentration and △f n . The relationship between the central frequency offset △f 0 and the NO 2 gas concentration is as follows:

[0155]

[0156] where k 1 , k 2 , α are coefficients obtained from the calibration experiment. Therefore, the offset of the n - th order acoustic frequency comb tooth can be expressed as:

[0157]

[0158] From this, it can be sorted out as:

[0159]

[0160] Step 4. Measurement and calculation of NO 2 gas with unknown concentration.

[0161] Confirm that the experimental environment of the gas test chamber is stable (temperature, humidity). Use a gas mass flow controller (MFC) to introduce NO 2 gas with unknown concentration, and use the sensor for detection.

[0162] 1. Using optical interferometry, collect the vibration signal of a single-layer graphene resonator through a photodetector, and use a lock-in amplifier to obtain the amplitude-frequency curve at an unknown concentration at this position, and record the main resonance frequency f at this unknown concentration 0' (center frequency f 0' ) and its higher harmonic components f n' .

[0163] 2. Calculate the frequency offset Δf of the nth sub-comb tooth of the frequency comb n .

[0164] 3. Substitute into the gas concentration calculation formula in step three to solve the gas concentration at this time.

[0165] The preferred embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the protection scope of the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and these simple modifications all belong to the protection scope of the present invention.

[0166] In addition, it should be noted that, among the various specific technical features described in the above specific embodiments, without conflict, they can be combined in any appropriate manner. To avoid unnecessary repetition, the present invention will not separately describe various possible combination methods.

[0167] In addition, any combination can be made between various different embodiments of the present invention, as long as it does not violate the idea of the present invention, and it should also be regarded as the content disclosed by the present invention.

Claims

1. An ultra-low power consumption nanoelectromechanical resonant NO2 gas sensor based on an audio comb, characterized in that: The invention comprises a drain electrode (11), an I-shaped gate electrode (12), a source electrode (13), a silicon dioxide insulating layer (3), a single crystal silicon substrate (4) and an adsorption layer (5); wherein the drain electrode (11) and the source electrode (13) are respectively arranged on the silicon dioxide insulating layer (3), and an I-shaped groove (2) is embedded on the silicon dioxide insulating layer (3) between the drain electrode (11) and the source electrode (13), and an I-shaped gate electrode (12) is arranged in the I-shaped groove (2); the adsorption layer (5) is laid above the front end of the silicon dioxide insulating layer (3) and covers the front end surfaces of the drain electrode (11) and the source electrode (13), and the middle part of the adsorption layer (5) is suspended above the front end of the I-shaped gate electrode (12); and the silicon dioxide insulating layer (3) is arranged on the single crystal silicon substrate (4).

2. The ultra-low power consumption nanoelectromechanical resonance NO2 gas sensor based on an audio comb according to claim (1), characterized in that: The adsorption layer (5) is a single-layer graphene layer.

3. The ultra-low power consumption nanoelectromechanical resonant NO2 gas sensor based on an acoustic frequency comb according to claim (1), characterized in that: The adsorption layer (5) is a double-ended fixed beam structure; the drain electrode (11) and the source electrode (13) have the same shape and size and are symmetrically arranged on both sides of the silicon dioxide insulating layer (3).

4. The ultra-low power consumption nanoelectromechanical resonant NO2 gas sensor based on an acoustic frequency comb according to claim (1), characterized in that: The single crystal silicon base layer (4), the silicon dioxide insulating layer (3), the drain electrode (11), the I-shaped gate electrode (12), and the source electrode (13) are an integrated part.

5. The ultra-low power consumption nanoelectromechanical resonant NO2 gas sensor based on an audio comb according to claim (1), characterized in that: The adsorption layer (5) is a single-layer graphene specific sensitive film.

6. A method for measuring NO2 gas concentration using the ultra-low power consumption nanoelectromechanical resonance NO2 gas sensor based on an audio comb as described in any one of claims 1 to 5, characterized in that: The specific contents include the following: Step 1: Use optical method to calibrate the first and second order vibration modes of gas sensor Step 1.1, all measurements are performed in a vacuum test chamber, a 633 nm red laser is selected as the detection light source, and the laser beam is irradiated onto the adsorption layer (5); Step 1.2, applying a DC voltage to the I-shaped gate electrode (12), starting from 0V and gradually increasing, when the adsorption layer (5) vibrates, the phase difference between the reflected light and the transmitted light of the red laser will change, meet in space and interfere, and form interference fringes; Step 1.3, using a photoelectric detector to collect a signal of the light intensity of the interference fringes generated by the vibration of the adsorption layer (5) changing with time; using the photoelectric detector, converting the light intensity signal into an electrical signal, and sending it to a phase-locked amplifier for spectrum analysis; performing Fourier analysis on the electrical signal through the phase-locked amplifier to obtain an amplitude-frequency curve, the curve having two peaks, the frequency corresponding to the first peak point is the first-order modal frequency ω1 when the adsorption layer (5) vibrates, and the frequency corresponding to the second peak point is the second-order modal frequency ω2 when the adsorption layer (5) vibrates; in the process of adjusting the DC voltage, when it is observed that the first-order modal frequency ω1 of the adsorption layer (5) gradually approaches the frequency multiple relationship ω2≈2ω1 of the second-order modal frequency ω2, the DC voltage is adjusted; Step 1.4, then apply a voltage V to the gate electrode (12). g ac cos(Ωt) AC voltage, adjust V g ac cos(Ωt) generates nonlinear electrostatic force inside the I-shaped gate electrode (12) and the adsorption layer (5), introduces electrostatic nonlinearity into the adsorption layer (5), breaks the structural symmetry, and drives the entire sensor; wherein the driving frequency Ω of the AC voltage needs to match the internal resonance mode of the system, that is, it must satisfy the following relationship: Ω≈ω1; wherein the control V g ac The amplitude of cos(Ωt) gradually increases between 0V and 1V. By observing the change of the amplitude-frequency curve, when the adsorption layer (5) is about to enter the nonlinear internal resonance region, part of the energy of the first-order mode will be transferred to the second-order mode, which is manifested as a decrease in the amplitude of the first-order mode. At this time, the first peak point does not fork from a single peak to an M-shaped double peak, and the peak value of the second-order mode increases, completing the V g ac Adjust cos(Ωt) and record the AC voltage V in this state g ac cos(Ωt); Step 2: Obtain the reference audio frequency comb spectrum under vacuum Step 2.1, in the absence of NO2 gas, based on step 1, continue to increase the AC voltage V applied to the I-shaped gate electrode (12) g ac cos(Ωt), observe the amplitude-frequency curve collected by the phase-locked amplifier until the first peak is observed to be bifurcated from a single peak to an M-shaped double peak. Due to the nonlinear behavior of the single-layer graphene, an internal resonance interval will be formed under the action of the I-shaped gate voltage (12). The vibration signal of the adsorption layer (5) is measured by the photodetector and the phase-locked amplifier to obtain an amplitude curve x(t) that changes with time; the collected x(t) is a mixed signal, which contains multiple frequency components. The signal x(t) is Fourier transformed to convert the signal from the time domain to the frequency domain, and the discrete frequency components in the signal are extracted, which are expressed as multiple discrete frequency components, namely, a frequency comb spectrum, also known as an audio comb. The spectrum consists of n comb teeth, and the comb teeth are expressed as a series of discrete frequency components rather than a continuous frequency band. The amplitude decreases as the frequency increases; its horizontal axis is the frequency f, which represents the frequency range of the system response; the vertical axis is the amplitude, which represents the vibration response intensity of the adsorption layer (5) at different frequencies; Step 2.2, observe the main resonant frequency in the audio comb, i.e. the center frequency f0 and its high-order harmonic components f n , in the audio frequency comb obtained after Fourier transformation, the strongest spectral line, that is, the spectral line with the largest amplitude, corresponds to the main resonance frequency, which is the center comb tooth f0; Step 3 is the same as step 2, except that NO2 gas of unknown concentration is introduced into the test chamber to obtain the audio frequency comb spectrum at this time and obtain the main resonance frequency f at this time. 0' , and calculate the frequency offset △f of the nth order sub-comb n , where △f n =n·△f0, the offset of the center frequency f0△f0=f 0' -f0; Step 4: Calculate the unknown NO2 gas concentration according to the following formula 7. The method for measuring NO2 gas concentration according to claim 6, characterized in that: The NO2 gas concentration in step 4 The calculation process is as follows: Step 1.1, inject NO2 from 0 ppb to 10 ppb into the test chamber in order from low to high, divide this range into N equally spaced concentrations, and record the corresponding sound frequency comb spectrogram at each concentration; Step 1.2, extracting the audio comb interval: Since NO2 gas is adsorbed on the surface of the adsorption layer (5), the equivalent mass m of the adsorption layer (5) is eff and equivalent stiffness k eff changes, causing the main resonant frequency to shift: in , due to the adsorption of NO2 molecules, the equivalent mass m of the adsorption layer (5) increases eff , resulting in a decrease in the natural resonance frequency; the interaction between NO2 molecules and graphene affects the nonlinear stiffness term, which in turn affects the high-order harmonic frequency, resulting in a frequency shift. At this time, the main resonance frequency f in the amplitude-frequency curve is collected and observed by the phase-locked amplifier. 0' and its higher-order harmonic components f n' , and calculate the offset of the center frequency f0 △f0 = f 0' -f0; When the gas concentration is extremely low, the frequency offset of the nth order sub-comb tooth, that is, the frequency offset of the sub-comb tooth relative to its original position, is n times the offset of the center frequency comb tooth, that is, △f n =n·△f0, the amplification effect is used to improve the detection sensitivity, making it possible to detect sub-ppb NO2 concentration; Step 1.3, establish a mathematical model of NO2 concentration and acoustic comb interval In each concentration Under this gas concentration, i corresponds to the i-th concentration of NO2, i is a positive integer ranging from 1 to N, and the offset △f of the main resonance frequency under this gas concentration is measured. 0,i , and get a set of data points Using the amplification effect of high-order sub-comb teeth, the NO2 concentration With △f n Perform mathematical fitting, the center frequency offset △f0 and NO2 gas concentration The relationship between them is: Among them, k1, k2, α are fitted to the entire data set using the nonlinear least squares method based on the calibration experiment to extract the optimal parameters k1, k2, α. The specific fitting method is as follows: a. Substitute the collected data into the mathematical model: b. Construct the objective function to minimize the sum of squares of the residuals between the fitting curve and the experimental data: c. Use numerical optimization algorithm to optimize k 1) , k2 and α are fitted, so the frequency offset of the n-th order sub-comb is: From this we can get:

8. A method for preparing an ultra-low power consumption nanoelectromechanical resonant NO2 gas sensor based on an acoustic comb, characterized in that: The process steps include: Step a: substrate cleaning A single-side polished silicon wafer is selected for processing. The single-side polished silicon wafer is placed in acetone for ultrasonic cleaning for 10 minutes using an ultrasonic cleaning machine, then taken out, and then placed in isopropanol for ultrasonic cleaning for 10 minutes. Finally, a plasma treatment is performed to obtain a single-crystal silicon base layer (4); a 1.5 μm silicon dioxide insulating layer (3) is deposited on the polished surface of the single-crystal silicon base layer (4) by a low-pressure chemical vapor deposition method to complete the preparation of the oxide wafer; Step b, glue application The oxide sheet prepared in step a is pretreated by baking in a high temperature furnace at 120 degrees for 10 minutes to remove surface moisture; Use a spin coater to perform coating: adsorb the non-polished surface of the oxide wafer on a coating tray, drop AR (80) photoresist on the silicon dioxide insulating layer (3), the coating parameters are 500r / 5s, 4000r / 30s, and bake at 95°C / 90s; use a Nikon I12 photolithography machine for exposure, the exposure time is 270ms, use 2.38% TMAH developer, develop for 40s, develop to form a photolithography pattern, so that the photoresist in the area where the I-shaped groove (2) is located after photolithography is removed; Step c, using an AOE etcher, placing the sample prepared in step b on an etching table, setting the power to 1000 W, the gas CHF3: 30 sccm, and the etching time to 40 s, etching a 1 μm thick I-shaped groove (2) on the silicon dioxide insulating layer (3) by dry etching, and then removing the remaining photoresist on the silicon wafer; Step d, using a spin coater to perform photoresist coating: adsorbing the non-polished surface of the oxide wafer on a photoresist coating tray, dripping AR80 photoresist on the silicon dioxide insulating layer (3), the photoresist coating parameters are 500r / 5s, 4000r / 30s, baking at 95°C, 90s; using a Nikon I12 photolithography machine for exposure, the exposure time is 270ms, using 2.38% TMAH developer, developing for 40s, developing to form a photolithography pattern, so that the photoresist in the area where the gate electrode (12) is located is removed after photolithography, and the photoresist in the remaining area is retained as a mask layer; Step e, depositing a gate electrode (12) with a thickness of 800 nm on the silicon dioxide insulating layer (3) by magnetron sputtering deposition to complete the preparation of the gate electrode (12), and then removing the photoresist retained in step d; Step f, using a spin coater to perform photoresist coating: adsorbing the non-polished surface of the oxide wafer on a photoresist coating tray, dripping AR80 photoresist on the silicon dioxide insulating layer (3), the photoresist coating parameters are 500r / 5s, 4000r / 30s, and baking at 95°C / 90s; using a Nikon I12 photolithography machine for exposure, the exposure time is 270ms, using a 2.38% TMAH developer, developing for 40s, and developing to form a photolithography pattern, so that the photoresist in the area where the drain electrode (11) and the source electrode (13) are located after photolithography is removed; Step g, depositing a drain electrode (11) and a source electrode (13) with a thickness of 100 nm on the silicon dioxide insulating layer (3) by magnetron sputtering deposition, completing the preparation of the drain electrode (11) and the source electrode (13), and then removing the remaining photoresist; Step h, transferring the single-layer graphene to PMMA by dry exfoliation to prepare for dry transfer of the single-layer graphene; Step i, transfer the single-layer graphene on PMMA to the corresponding position on the silicon dioxide insulating layer (3) to form an adsorption layer (5), and finally use a DISCO cutter to scribing on the single-crystalline silicon base layer (4) at a rotation speed of 20,000 rpm and a cutting rate of 2 mm / s to complete the preparation of the NO2 gas sensor.

Images