Artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device and method

Abstract

The invention relates to the technical field of terahertz wave detection imaging, in particular to an artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device and method. Comprising a femtosecond laser, a beam splitter, a chopper, a laser spot beam expanding module, a spinning terahertz emitter, a transmission and reflection integrated automatic scanning detection module, a terahertz detection device, an artificial intelligence enhanced thickness measurement module and an artificial intelligence enhanced defect detection imaging module. Laser emitted by the femtosecond laser generates pump light and detection light through the beam splitter, and the pump light sequentially passes through the laser spot beam expanding module, the spinning terahertz emitter and the transmission and reflection integrated automatic scanning detection module to enter the terahertz detection device. The detection light enters the terahertz detection device through the transmission and reflection integrated automatic scanning detection module, the terahertz detection device outputs a terahertz pulse shape, and finally the thickness of a sample or a defect contour is obtained; the sample defect detection efficiency can be improved, and the cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of terahertz wave detection and imaging, and in particular to an artificial intelligence-enhanced strong-field terahertz spectroscopy detection and imaging device and method. Background Art

[0002] The terahertz band lies in the frequency range of 0.1-10 THz. As an electromagnetic band between microwaves and infrared, it combines the strong penetration of microwaves with the high resolution of infrared. Compared to traditional detection technologies, terahertz imaging exhibits unique advantages in the detection of metal defects, coatings, and packaging materials. Its penetration depth can reach the centimeter level, and its spatial resolution reaches the submillimeter level, enabling detailed characterization of the internal structure of materials. Compared with microwave detection, terahertz electromagnetic waves have a shorter wavelength, which can significantly improve the ability to identify interface defects. Compared with infrared detection, terahertz waves can penetrate infrared shielding media such as composite materials and avoid the effects of thermal radiation and high energy on the detection object. Currently, terahertz wave non-destructive testing and spectral imaging technology have great application potential and broad development prospects in basic scientific research such as security screening, cultural relic protection, paint thickness measurement, internal structure detection of composite materials, and condensed matter nanoscale imaging.

[0003] With the next generation of technological revolutions and industrial upgrades, the automotive manufacturing, aerospace, and other fields are booming, leading to a growing demand for thickness measurement of non-metallic materials such as paint and thermal barrier coatings, as well as for metal surface defect detection. Thanks to terahertz's ability to penetrate many non-polar materials and its high spatial resolution, terahertz spectroscopy and spectral imaging have become common tools for thickness measurement and non-destructive testing of composite materials. Time-of-flight (TOF) is a traditional method for penetrating material thickness measurement and CT imaging, offering advantages such as convenience and intuitiveness. However, this method requires accurate knowledge of the refractive index of the material being measured to achieve more accurate measurements. Furthermore, TOF thickness measurement typically relies on the time difference between reflected waves from the front and back surfaces. For samples of extremely thin thickness, multiple reflection peaks can be aliased, making direct calculation difficult. Research has demonstrated the use of artificial intelligence methods, using a database of thickness measurement signals for iteration. The training results can effectively identify the time differences between reflection peaks corresponding to aliased signals. Current research based on convolutional neural networks (CNNs) has enabled thickness measurements on the order of 20 microns. Improving the effectiveness and stability of training results, as well as their generalizability across diverse scenarios, is crucial. Furthermore, imaging material defects is also a crucial component of non-destructive testing. Single-point scanning is a common method for defect testing, which can effectively scan detailed information of various parts of the material. However, for larger samples, point-by-point scanning is somewhat time-consuming.

[0004] In summary, existing terahertz nondestructive testing (NTD) technologies face challenges such as unknown material refractive index parameters, excessively large pulse widths leading to echo signal aliasing, low efficiency in detecting defects in large samples, and high costs, hindering further development and breakthroughs in terahertz NDT and spectroscopy imaging. Therefore, research is crucial to developing technologies that can accurately measure the thickness of micron-level samples and image defects in large samples. This also places demands on the flexibility and efficiency of the detection system. Summary of the Invention

[0005] In view of the above problems, the present invention provides an artificial intelligence-enhanced strong-field terahertz spectroscopy detection imaging device and method, which solves the technical problems of low efficiency and high cost of sample defect detection in the prior art.

[0006] On the one hand, the present invention provides an artificial intelligence-enhanced strong-field terahertz spectroscopy detection and imaging device, a femtosecond laser, a beam splitter, a chopper, a laser spot beam expansion module, a spin terahertz emitter, a transmission and reflection integrated automated scanning detection module, a terahertz detection device, an artificial intelligence-enhanced thickness measurement module, and an artificial intelligence-enhanced defect detection imaging module; The laser emitted by the femtosecond laser generates pump light and probe light through a beam splitter. The pump light passes through a chopper, a laser spot beam expansion module, a spin terahertz emitter, and a transflective integrated automated scanning detection module in sequence to enter the terahertz detection device. The probe light passes through the transflective integrated automated scanning detection module to enter the terahertz detection device, and the terahertz detection device outputs a terahertz pulse shape. The artificial intelligence-enhanced thickness measurement module and the artificial intelligence-enhanced defect detection imaging module perform processing based on the terahertz pulse shape to obtain the sample thickness or defect profile.

[0007] Preferably, the laser spot beam expansion module includes a concave lens and a convex lens, and the pump light passes through the concave lens and the convex lens in sequence to complete the beam expansion.

[0008] Preferably, the spin terahertz emitter comprises a one-dimensional photonic crystal enhanced nano-ferromagnetic film, and the expanded pump light is injected into the nano-ferromagnetic film to generate terahertz waves.

[0009] Preferably, the nano-ferromagnetic film is composed of W, CoFeB and Pt layers respectively, and the thickness of each single layer of W, CoFeB and Pt layer is 2 nm; The nano-ferromagnetic film is grown on a Si substrate by magnetron sputtering. Between the Si substrate and W, there are 40 groups of periodically alternating SiO2 with a thickness of 136nm and HfO2 with a thickness of 92nm, forming a one-dimensional photonic crystal with a total thickness of 9μm.

[0010] Preferably, the transmission-reflection integrated automated scanning detection module comprises a placement table, a two-dimensional translation stage, a perforated metal plate fixed on the two-dimensional translation stage, an off-axis parabolic mirror, a high-density polyethylene plate, and a one-dimensional delay line. The placement table is a reflector placement table with a transmission detection function or a sample placement table with a reflection detection function. The terahertz wave generated by the terahertz transmitter passes through the placement table, the perforated metal plate, the off-axis parabolic mirror and the high-density polyethylene plate in sequence to complete the transmission or reflection of the sample under test; the detection light is delayed by the one-dimensional delay line.

[0011] Preferably, the terahertz detection device comprises a detection light focusing lens, ITO glass, electro-optical crystal zinc telluride, a detection light collimating lens, a quarter wave plate, a Wollaston prism, and a balanced detector connected in sequence; The detection light output by the one-dimensional delay line passes through the detection light focusing lens and is incident on the ITO glass together with the pump light output by the high-density polyethylene plate. It then passes through the electro-optical crystal zinc telluride, the detection light collimating lens, the 1 / 4 wave plate, the Wollaston prism, and the balanced detector to finally output a terahertz waveform signal.

[0012] In one aspect, the present invention provides a detection method for a strong-field terahertz spectroscopy detection imaging device based on artificial intelligence enhancement, comprising the following steps: When performing thickness detection, reflection detection is used. If the front surface reflected wave and the back surface reflected wave of the terahertz waveform signal are mixed, the terahertz waveform signal is pre-processed and input into the artificial intelligence enhanced thickness measurement module, and the trained thickness measurement neural network model is used to output the predicted thickness result; When performing defect detection on large-scale samples, reflective detection is used. The sample to be tested reflects the terahertz echo signal, and the presence of defects is determined based on the terahertz waveform signal. When performing defect contour extraction imaging, transmission detection is used to input the terahertz waveform into the artificial intelligence enhanced defect detection imaging module to output the defect contour of the sample.

[0013] Preferably, the thickness detection further includes the following steps: If the front surface reflected wave and the back surface reflected wave of the terahertz waveform signal are not aliased, the sample thickness is calculated based on the distance between the reflected waves and the refractive index. The calculation expression of the refractive index is: (1) (2) (3) in, 、 、 They represent the electric field intensities of the front surface reflected wave, the rear surface reflected wave, and the rear surface secondary reflected wave, respectively. is the incident angle on the sample surface, represents the ratio of the three electric field intensities, represents the field ratio enhancement factor, Represents the refractive index.

[0014] Preferably, the trained thickness measurement neural network is obtained by: A terahertz signal dataset is established based on the measured terahertz echo signal and the simulated terahertz echo signal obtained by applying Snell's law and Fresnel equation; The terahertz signal data set is used to train a thickness measurement neural network model. During the training process, Gaussian noise is added to the data in the data set, and a periodic learning rate is used to finally obtain a trained thickness measurement neural network.

[0015] Compared with the prior art, the present invention has at least the following beneficial effects: (1) The nondestructive testing function of the present invention is comprehensive and flexible. The optical path design of the device combines both transmission and reflection detection functions, which can simultaneously meet the detection needs of various materials such as metal surface defect detection and non-conductive material internal structure detection and imaging, and has a wide range of application scenarios. This technology can also realize the determination of sample parameters such as refractive index calculation and coating thickness measurement. It is scalable for different types of materials, and the detection function can be switched according to the detection needs; (2) The terahertz emission source used in the present invention has excellent performance and is easy to prepare. The single-photonic crystal enhanced spin terahertz radiation source has the advantages of high emission efficiency, wide bandwidth, controllable polarization, stability and reliability, and simple structure. It can emit high-intensity, ultra-short pulse width terahertz signals. Compared with traditional methods, it can achieve direct thickness measurement of thinner coating samples and higher signal-to-noise ratio and higher resolution defect detection imaging; the terahertz emission source can be realized using various commonly used devices and relatively mature magnetron sputtering devices, without the need for complex and high-cost manufacturing processes similar to various metamaterials, thus overcoming the shortcomings of complex structures and high costs of many terahertz emission sources; (3) The terahertz emission source used in the present invention is larger in size and can emit terahertz radiation with a beam diameter of 4 inches. Compared with the point-by-point scanning of the sample by a small-sized terahertz spot, the present invention can cover a circular area with a diameter of 4 inches in a single scan, thereby improving the scanning efficiency per unit area and realizing the rapid detection of the existence of decimeter-level sample defects; (4) The present invention can use the reflection detection function to calculate the refractive index of a single layer of unknown material in the terahertz frequency band under small-angle incidence conditions using the front surface reflection wave, back surface reflection wave, and back surface secondary reflection wave of the sample; this breaks through the normal incidence condition of the existing calculation method and improves the scalability of the present invention for detecting different materials; (5) Based on Snell's law and Fresnel's equation, the present invention establishes a data set and builds a network model based on the Conformer architecture, which realizes the thickness prediction of ultra-thin sample aliasing signals of 1~1000 micron samples, with an accuracy of 99.8% in the range of ±8 micron. Compared with the existing convolutional neural network model for thickness measurement of samples of the order of 20 micron, the effect is better and the accuracy is higher; (6) The present invention is low-cost. The optical path design does not require the use of a parabolic mirror combination. The characteristics of spectral imaging make it unnecessary to use expensive terahertz cameras and other equipment for detection. Compared with traditional terahertz imaging solutions, the cost of device construction, use and maintenance is reduced. BRIEF DESCRIPTION OF THE DRAWINGS The drawings are only for purposes of illustrating particular embodiments and are not to be considered limiting of the invention.

[0016] Figure 1 Schematic diagram of the artificial intelligence-enhanced strong-field terahertz spectroscopy detection imaging device provided by the present invention.

[0017] Figure 2 This is a schematic diagram of the transmission detection function and reflection detection function provided by the present invention.

[0018] Figure 3 Schematic diagram of the artificial intelligence enhanced thickness measurement neural network model provided by the present invention Figure 1: Femtosecond laser, 2: beam splitter, 3: chopper, 4: concave lens, 5: convex lens, 6: spin terahertz emitter, 7: placement table, 8: metal plate with holes, 9: off-axis parabolic mirror, 10: high-density polyethylene plate, 11: one-dimensional delay line, 12: detection light focusing lens, 13: ITO glass (indium tin oxide glass), 14: electro-optical crystal zinc telluride, 15: detection light collimating lens, 16: 1 / 4 wave plate, 17: Wollaston prism, 18: balanced detector. DETAILED DESCRIPTION

[0019] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. In addition, the present invention can also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited by the specific embodiments disclosed below.

[0020] This invention, based on a spintronic terahertz emitter and an integrated transflective automated scanning detection device, cleverly integrates an ultra-wideband, large-scale, high-signal-to-noise ratio terahertz radiation source, a spectroscopy detection and imaging device, and deep learning neural network technology to achieve thickness measurement of micron-scale samples and efficient defect detection and contour extraction imaging of large-scale targets. The invention is low-cost, simple to build, robust and reliable, scalable for different types of materials, and detection functions can be switched according to detection needs, resulting in high efficiency and flexibility.

[0021] In order to illustrate the effectiveness of the method proposed by the present invention, the above technical solution of the present invention is described in detail below through a specific embodiment. Figure 1 As shown, an artificial intelligence-enhanced strong-field terahertz spectroscopy detection and imaging device is disclosed, including: a femtosecond laser 1, a beam splitter 2, a chopper 3, a laser spot beam expansion module, a spin terahertz emitter 6, an integrated transmission and reflection automatic scanning detection module, a terahertz detection device, an artificial intelligence-enhanced thickness measurement module, and an artificial intelligence-enhanced defect detection imaging module.

[0022] The laser emitted by the femtosecond laser 1 generates pump light and probe light through the beam splitter 2. The pump light passes through the chopper 3, the laser spot beam expansion module, the spin terahertz emitter 6, and the transmissive-reflective integrated automatic scanning detection module to enter the terahertz detection device; the probe light passes through the transmissive-reflective integrated automatic scanning detection module to enter the terahertz detection device, and the terahertz detection device outputs a terahertz pulse shape. The artificial intelligence-enhanced thickness measurement module and the artificial intelligence-enhanced defect detection imaging module perform thickness measurement, defect detection, and contour extraction based on terahertz pulse shapes.

[0023] The laser spot beam expansion module of the present invention comprises a concave lens 4 and a convex lens 5 , and the pump light passes through the concave lens 4 and the convex lens 5 in sequence to complete the beam expansion.

[0024] The spin terahertz emitter 6 of the present invention includes a one-dimensional photonic crystal enhanced nano-ferromagnetic film and is provided with an external magnetic field. The expanded pump light is incident on the one-dimensional photonic crystal enhanced nano-ferromagnetic film to generate terahertz waves.

[0025] In some embodiments, the nano-ferromagnetic film of the present invention is a heterojunction film having a multilayer film structure composed of W / CoFeB / Pt layers, wherein the thickness of the W, CoFeB, and Pt single layers is 2 nm, and the total thickness of the nano-ferromagnetic film is 6 nm. The nano-ferromagnetic film is grown on a Si substrate by magnetron sputtering, and the Si substrate thickness is 136 nm. The spin nano-film can emit broadband terahertz waves with an effective bandwidth greater than 5 THz under laser pumping. As a new terahertz emission source technology, it can significantly improve the system's thickness measurement performance for micron-level ultra-thin samples compared to traditional emission sources.

[0026] In some embodiments, between the Si substrate and the heavy metal W, 40 groups of periodically alternating SiO2 with a thickness of 136 nm and HfO2 with a thickness of 92 nm form a multilayer film one-dimensional photonic crystal structure with a total thickness of approximately 9 μm. This can improve the sample's utilization of pump laser energy, achieve coherent enhancement of terahertz waves, increase the field intensity of terahertz radiation, and improve the signal-to-noise ratio of the radar system, thereby improving defect detection performance and contour extraction imaging quality.

[0027] In some embodiments, the W / CoFeB / Pt heterojunction film serving as the emission source has a diameter of 4 inches, and can directly generate large-scale terahertz wave radiation with a diameter of 4 inches under the pumping of a laser with a diameter of 4 inches. There is no need to use an off-axis parabolic mirror combination to expand the terahertz wave. This reduces the complexity of the system while achieving efficient detection of defects in large-scale samples, making it easier to integrate.

[0028] The transmission-reflection integrated automatic scanning detection module includes a placement table 7, a two-dimensional translation stage, a perforated metal plate 8 mounted on the two-dimensional translation stage, an off-axis parabolic mirror 9, a high-density polyethylene plate 10 and a one-dimensional delay line 11.

[0029] The terahertz wave generated by the terahertz transmitter 6 passes through a platform 7, a perforated metal plate 8, an off-axis parabolic mirror 9, and a high-density polyethylene plate 10, transmitting or reflecting the sample under test. The probe light generated by the beam splitter 2 is delayed by a one-dimensional delay line 11. The platform 7 can be used as a reflector platform for transmission detection or a sample platform for reflection detection.

[0030] Specifically, if Figure 2 As shown, the transmission detection function or the reflection detection function is selected according to the detection requirements. The reflection detection function is used when measuring the thickness of micron-level samples. The one-dimensional delay line moves under program control, and the sample to be tested reflects the terahertz echo signal. The reflection detection function is used when detecting defects in large-scale samples. The one-dimensional delay line moves under program control, and the sample to be tested reflects the terahertz echo signal.

[0031] In some embodiments, the one-dimensional delay line and the two-dimensional translation stage move in coordination under program control, and are controlled by a connected computer to achieve automated scanning without manual operation under multiple imaging functions.

[0032] The terahertz detection device includes a detection light focusing lens 12, ITO glass 13 (indium tin oxide glass), electro-optical crystal zinc telluride 14, detection light collimating lens 15, quarter-wave plate 16, Wollaston prism 17, and balanced detector 18, which are connected in sequence. The detection light output by the one-dimensional delay line 11 passes through the detection light focusing lens 12 and is incident on the ITO glass 13 together with the pump light output by the high-density polyethylene plate 10. It then passes through the electro-optical crystal zinc telluride 14, the detection light collimating lens 15, quarter-wave plate 16, Wollaston prism 17, and the balanced detector 18, ultimately outputting a terahertz waveform signal.

[0033] The artificial intelligence-enhanced thickness measurement module and the artificial intelligence-enhanced defect detection imaging module can process the terahertz waveform signal based on the trained thickness measurement neural network model to obtain the sample thickness; the artificial intelligence-enhanced thickness measurement module and the artificial intelligence-enhanced defect detection imaging module can obtain the defect contour based on the contour extraction neural network model.

[0034] In some embodiments, the artificial intelligence-enhanced thickness measurement module and the artificial intelligence-enhanced defect detection imaging module are computers having a processor, a memory, and input and output devices. The trained thickness measurement neural network model and the trained contour extraction neural network model are recorded in the memory. The computer reads the terahertz waveform signal through the input and output devices, and the processor reads and uses the neural network model to process the terahertz waveform signal to obtain the sample thickness or defect contour.

[0035] The present invention also provides a detection method for an artificial intelligence enhanced strong-field terahertz spectroscopy detection imaging device, comprising the following steps.

[0036] When measuring the thickness of micron-level samples, a reflective detection function is used. The terahertz waveform signal is pre-processed and input into the artificial intelligence-enhanced thickness measurement module, and the trained thickness measurement neural network model is used to output the predicted thickness result. When detecting defects in large-size samples, reflective detection is used. The sample to be tested reflects the terahertz echo signal, and the terahertz detection device detects and outputs the time domain waveform. The existence of the defect can be determined based on the waveform. When performing defect contour extraction and imaging, transmission detection is used. The terahertz waveform is input into the artificial intelligence-enhanced defect detection imaging module, and the defect contour of the sample is output.

[0037] In some embodiments, the present invention uses Snell's law and Fresnel's equations to simulate terahertz echo signals based on measured terahertz echo signals to establish a terahertz signal dataset for training and testing. The R-squared value is then used to verify the similarity between the simulated signals and the actual signals, confirming the usefulness of the simulated signals as a dataset. This method ensures a sound and solid theoretical foundation while reducing the workload and cost of establishing a dataset through a large number of repetitive experiments to collect measured terahertz signals.

[0038] In some embodiments, the thickness measurement neural network model of the present invention is as follows Figure 3 As shown, the thickness measurement neural network model can accept terahertz waveform signals belonging to multiple thickness categories as input. The model consists of two sequentially connected sets of convolutional layers, activation functions, and maximum pooling layers. The second set of maximum pooling layers is fed into a feature coupling unit through one channel, and another channel through a position encoding and Transformer layer. The feature coupling unit includes a linear layer. The results of the linear layer are fused through multi-layer perception, residual connections, and normalization. The final predicted thickness result is output through a fully connected layer.

[0039] In some embodiments, the present invention uses the terahertz signal dataset to train the thickness measurement neural network model. The neural network training program adds Gaussian noise to enhance the network's generalization capabilities. A periodic learning rate is used to avoid training falling into local optima, instead seeking a global optimum or better solution, improving final performance. The trained thickness measurement neural network model uses residual connections and layer normalization, which helps improve the model's training efficiency and stability, as well as its generalization capabilities.

[0040] The following describes the various devices on the optical path of the present invention, such as Figure 1 The AI-enhanced high-field terahertz spectroscopy imaging device shown includes a femtosecond laser 1, a beam splitter 2, a chopper 3, a concave lens 4, a convex lens 5, a spin terahertz emitter 6, a placement platform 7, a perforated metal plate 8 mounted on a two-dimensional translation stage, an off-axis parabolic mirror 9, a high-density polyethylene plate 10, a one-dimensional delay line 11, a probe light focusing lens 12, ITO glass 13 (indium tin oxide glass), an electro-optical crystal zinc telluride 14, a probe light collimating lens 15, a quarter-wave plate 16, a Wollaston prism 17, and a balanced detector 18. The two-dimensional translation stage is electrically controlled and can move in both horizontal and vertical directions under computer program control, with the perforated metal plate 8 mounted on the stage moving accordingly. The one-dimensional delay line 11 is also electrically controlled and can move in a single axis under computer program control.

[0041] The transmission and reflection integrated automatic scanning detection device is as follows: Figure 2As shown in the figure, the two sub-figures describe the two functions of the system, namely, transmission detection and reflection detection. In transmission detection, the sample to be detected is placed in the optical path, the terahertz is incident on the sample by the transmitter 6, and the terahertz is reflected to the receiver by the plane mirror on the placement table 7 (the terahertz receiving module is Figure 1 Marked 14-18). Reflection detection uses terahertz radiation directly reflected from the sample. After emanating from transmitter 6, the terahertz radiation enters the sample on stage 7 at a 15-degree angle. It is then reflected by the sample or substrate at a 15-degree angle to the receiver. Both detection methods utilize a perforated metal plate 8 in the optical path for two-dimensional movement to achieve single-point scanning. Different samples require different detection methods: non-conductive composite materials and low-density materials are suitable for transmission detection, while materials with metal substrate coatings are more suitable for reflection detection.

[0042] The 0.5-inch diameter beam emitted by femtosecond laser 1 enters the system from one side and is divided into two paths by beam splitter 2. 90% of the energy of the femtosecond laser enters the pump path, and the remaining 10% of the energy of the femtosecond laser enters the detection path.

[0043] A chopper 3, connected and synchronized with the laser pulses, is placed in the pump circuit and operates at 500 Hz, suppressing noise and improving the system's signal-to-noise ratio. The pump laser then passes through a beam expansion system consisting of a concave lens 4 with a focal length of -50 mm and a convex lens 5 with a focal length of 500 mm. It then strikes a 4-inch spin terahertz emitter 6, generating a large, ultra-broadband, high-field terahertz wave with a diameter of 4 inches. This terahertz wave is incident on an integrated transmission and reflection automated scanning detection device. After detection using a transmission / reflection detection method, the emitted signal is collected by an off-axis parabolic mirror 9 with a focal length of 203.2 mm and focused onto an electro-optical zinc telluride crystal 14. The high-density polyethylene sheet 10 serves to transmit the terahertz wave while blocking laser propagation.

[0044] After passing through a one-dimensional delay line 11 and several reflectors, the probe laser is focused by a probe light focusing lens 12 with a focal length of 150 mm, and overlaps with the terahertz wave in the electro-optical crystal zinc telluride 14. Based on the electro-optic effect, the terahertz electric field changes the refractive index of the electro-optical crystal zinc telluride 14, thereby changing the polarization state of the probe laser. The beam state is then adjusted again by the probe light collimating lens 15. A quarter-wave plate 16 and a Wollaston prism 17 are used to adjust the two orthogonal polarization components of the probe laser, and a balanced detector 18 is used to collect them separately. After being processed by a phase-locked amplifier, the terahertz waveform signal is plotted in self-written software. A computer controls the movement of the delay line to change the optical path of the probe laser, which is used to map the terahertz wave electric field intensity at each moment and plot the shape of the terahertz pulse.

[0045] When measuring the thickness of micron-level samples, reflective detection is used. The signals obtained include the front surface reflection wave, the back surface reflection wave and the back surface secondary reflection wave. The electric field strength of the echo pulse is respectively 、 、 ,set up is the incident angle on the sample surface, the refractive index of the unknown material in the terahertz band can be calculated using the following formula .

[0046]

[0047]

[0048]

[0049] in, 、 、 They represent the electric field intensities of the front surface reflected wave, the rear surface reflected wave, and the rear surface secondary reflected wave, respectively. represents the ratio of the three electric field intensities, represents the field ratio enhancement factor, Represents the refractive index.

[0050] For signals without aliasing of the front surface reflection wave and the back surface reflection wave, the sample thickness can be directly calculated from the distance between the two reflection waves and the refractive index. For signals with aliasing of the front surface reflection wave and the back surface reflection wave, the signals can be input into the trained thickness measurement neural network model, and the output value can be proportionally calculated with the refractive index to obtain the sample thickness.

[0051] When performing defect detection and contour extraction imaging of large-scale targets, transmission detection or reflection detection is used to directly observe the number of signal pulses to determine whether there are defects. A two-dimensional scan is performed using a perforated metal plate 8 mounted on a two-dimensional translation stage to obtain the terahertz signal at each scanning point for peak imaging. The imaging results are sequentially input into the spectral imaging and optimization processing program and the mature contour extraction neural network model to obtain the defect contour extraction results.

[0052] In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connect," "fixed," etc. should be understood broadly. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0053] In the present invention, unless otherwise expressly specified or limited, a first feature being "above" or "below" a second feature may include the first and second features being in direct contact, or may include the first and second features being in contact not directly but through another feature between them. Furthermore, a first feature being "above," "above," and "above" a second feature may include the first feature being directly above or obliquely above the second feature, or may simply mean that the first feature is higher in level than the second feature. A first feature being "below," "below," and "below" a second feature may include the first feature being directly below or obliquely below the second feature, or may simply mean that the first feature is lower in level than the second feature.

[0054] In the present invention, the terms "first", "second", "third", and "fourth" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. The term "plurality" refers to two or more, unless otherwise clearly defined.

[0055] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. An artificial intelligence-enhanced strong-field terahertz spectroscopy detection and imaging device, characterized in that: include: Femtosecond laser (1), beam splitter (2), chopper (3), laser spot beam expansion module, spin terahertz emitter (6), transmission and reflection integrated automatic scanning detection module, terahertz detection device, and artificial intelligence enhanced thickness measurement module and artificial intelligence enhanced defect detection imaging module; The laser emitted by the femtosecond laser (1) generates pump light and detection light through a beam splitter (2); the pump light sequentially passes through a chopper (3), a laser spot beam expansion module, a spin terahertz emitter (6), and a transmissive-reflective integrated automatic scanning detection module to enter a terahertz detection device; the detection light passes through the transmissive-reflective integrated automatic scanning detection module to enter a terahertz detection device, and the terahertz detection device outputs a terahertz pulse shape; The artificial intelligence-enhanced thickness measurement module and the artificial intelligence-enhanced defect detection imaging module perform processing based on the terahertz pulse shape to obtain the sample thickness or defect profile.

2. The artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device according to claim 1, characterized in that: The laser spot beam expansion module comprises a concave lens (4) and a convex lens (5), and the pump light passes through the concave lens (4) and the convex lens (5) in sequence to complete beam expansion.

3. The artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device according to claim 2, characterized in that: The spin terahertz emitter (6) comprises a one-dimensional photonic crystal enhanced nano-ferromagnetic film, and the expanded pump light is injected into the nano-ferromagnetic film to generate terahertz waves.

4. The artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device according to claim 3, characterized in that: The nano-ferromagnetic film is composed of W, CoFeB and Pt layers respectively, and the single layer thickness of W, CoFeB and Pt layers is 2nm; The nano-ferromagnetic film is grown on a Si substrate by magnetron sputtering. Between the Si substrate and W, there are 40 groups of periodically alternating SiO2 with a thickness of 136nm and HfO2 with a thickness of 92nm, forming a one-dimensional photonic crystal with a total thickness of 9μm.

5. The artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device according to claim 3, characterized in that: The transmission-reflection integrated automatic scanning detection module comprises a placement table (7), a two-dimensional translation table, a metal plate with holes (8) fixed on the two-dimensional translation table, an off-axis parabolic mirror (9), a high-density polyethylene plate (10) and a one-dimensional delay line (11), wherein the placement table (7) is a reflector placement table with a transmission detection function or a sample placement table with a reflection detection function; The terahertz wave generated by the spin terahertz emitter (6) passes through a placement table (7), a metal plate with holes (8), an off-axis parabolic mirror (9) and a high-density polyethylene plate (10) in sequence to complete transmission or reflection on the sample to be tested; the detection light is delayed by a one-dimensional delay line (11).

6. The artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device according to claim 5, characterized in that: The terahertz detection device comprises a detection light focusing lens (12), ITO glass (13), electro-optical crystal zinc telluride (14), a detection light collimating lens (15), a quarter wave plate (16), a Wollaston prism (17), and a balanced detector (18) connected in sequence; The detection light output by the one-dimensional delay line (11) passes through the detection light focusing lens (12) and is incident on the ITO glass (13) together with the pump light output by the high-density polyethylene plate (10). The detection light then passes through the electro-optical crystal zinc telluride (14), the detection light collimating lens (15), the 1 / 4 wave plate (16), the Wollaston prism (17), and the balanced detector (18) to finally output a terahertz waveform signal.

7. A detection method for an artificial intelligence enhanced strong field terahertz spectroscopy detection imaging device according to any one of claims 1 to 6, characterized in that: The following steps are involved: When performing thickness detection, reflection detection is used. If the front surface reflected wave and the back surface reflected wave of the terahertz waveform signal are mixed, the terahertz waveform signal is pre-processed and input into the artificial intelligence enhanced thickness measurement module, and the trained thickness measurement neural network model is used to output the predicted thickness result; When inspecting defects on large-scale samples, reflective detection is used. The sample to be tested reflects the terahertz echo signal, and the presence of defects is determined based on the terahertz waveform signal. When performing defect contour extraction imaging, transmission detection is used to input the terahertz waveform signal into the artificial intelligence enhanced defect detection imaging module to output the defect contour of the sample.

8. The detection method according to claim 7, characterized in that The thickness test also includes the following steps: If the front surface reflected wave and the back surface reflected wave of the terahertz waveform signal are not aliased, the sample thickness is calculated based on the distance between the reflected waves and the refractive index. The calculation expression of the refractive index is: in, 、 、 They represent the electric field intensities of the front surface reflected wave, the rear surface reflected wave, and the rear surface secondary reflected wave, respectively. is the incident angle on the sample surface, represents the ratio of the three electric field intensities, represents the field ratio enhancement factor, Represents the refractive index.

9. The detection method according to claim 8, characterized in that The method for obtaining the trained thickness measurement neural network model is as follows: A terahertz signal dataset is established based on the measured terahertz echo signal and the simulated terahertz echo signal obtained by applying Snell's law and Fresnel equation; The terahertz signal data set is used to train a thickness measurement neural network model. During the training process, Gaussian noise is added to the data of the terahertz signal data set, and a periodic learning rate is used to finally obtain a trained thickness measurement neural network model.

Images