System and method for laser processing carbon-based chips

Abstract

The embodiment of the specification provides a system and method for laser processing carbon-based chips, wherein the system comprises: a multi-wavelength laser emission module for emitting laser beams suitable for carbon-based chips with different crystal structures according to laser parameters; an intelligent control device for dynamically adjusting the laser parameters according to the thermal and optical characteristics of the carbon-based chips based on a processing strategy, and sending the laser parameters to the multi-wavelength laser emission module; an adaptive machine learning module for optimizing the processing strategy for the processing characteristics of the carbon-based chips according to monitoring data, and sending the optimization results to the intelligent control device; a micro-nano laser polishing device for processing the surface of the carbon-based chips; a cutting mechanism for performing hidden cutting of the carbon-based chips; a multi-modal sensing and analysis system for monitoring the structural integrity of the carbon-based chips during processing, and sending monitoring data to the adaptive machine learning module.

Description

TECHNICAL FIELD

[0001] The present document relates to the field of laser processing technology, and in particular to a system and method for laser processing carbon-based chips. BACKGROUND

[0002] Chip dicing is one of the key steps in the manufacturing process of integrated circuits. Traditional chip dicing methods usually use mechanical cutting or chemical etching techniques, but these methods have some limitations. Mechanical cutting is prone to mechanical stress, which can cause chip damage or cracks. Chemical etching can cause residues or chemical contamination, which negatively affects the performance and reliability of the chip. Especially for carbon-based chips, these problems are more pronounced. Carbon-based materials such as graphene, carbon nanotubes or diamond have unique physical and chemical properties, and are more sensitive to traditional cutting methods. Mechanical cutting can damage the structural integrity of carbon-based materials, while chemical etching can change their electronic properties.

[0003] To solve these problems, laser processing technology has been gradually introduced into the field of chip dicing in recent years. Laser cutting has the characteristics of non-contact, high precision and high efficiency, and can achieve precise cutting of chips. However, traditional laser cutting methods still have some challenges. First, laser cutting is prone to generate heat-affected zones, which can cause chip surface roughness to increase, residues to be generated, or materials to splash, etc. Second, when processing carbon-based chips, the laser cutting process must fully consider their unique material properties and complex structures to ensure the accuracy and consistency of the cutting. This requires fine tuning of laser parameters and the development of cutting strategies specifically for carbon-based materials.

[0004] As a new type of cutting technology, laser hidden cutting is a cutting method that reduces the heat-affected zone by controlling the parameters and processing conditions of the laser. By reasonably adjusting parameters such as laser power, pulse width and scanning speed, local melting and rapid cooling during chip cutting can be achieved, reducing heat conduction and heat diffusion, thereby reducing the size of the heat-affected zone. This can effectively reduce the roughness of the carbon-based chip surface and the possibility of residue generation.

[0005] Secondly, laser polishing is a method that combines laser irradiation and chemical reaction to modify and smooth the surface of carbon-based chip cutting. After cutting is completed, the laser beam can be directed at the cutting surface for targeted irradiation, through thermal excitation and surface reaction, causing changes in the microstructure of the material surface, thereby further improving the smoothness and quality of the surface. Laser polishing can flexibly adjust parameters such as laser power, irradiation time and chemical reaction conditions according to the characteristics of the chip material and the cutting requirements, to achieve precise control and optimization of the surface of carbon-based chips.

[0006] Machine learning is an artificial intelligence technology that uses deep neural network models to learn and infer complex tasks, which can be applied to the control and optimization of chip cutting processes. By training and learning on a large amount of cutting data and polishing data, an accurate model can be established to predict and adjust laser parameters, processing conditions, and cutting / polishing paths, etc., to achieve adaptive control and optimization of the cutting process. Deep learning can improve the accuracy and stability of chip cutting, reduce the need for manual intervention and optimization, and improve cutting efficiency and consistency.

[0007] Therefore, it is an urgent problem for researchers to develop a chip cutting system and method that can combine laser hidden cutting, laser polishing, and machine learning to solve the problems of increased surface roughness, residue generation, or material splashing in carbon-based chip cutting methods, and to achieve the prediction and adjustment of laser parameters, processing conditions, and cutting paths, etc. to improve the quality and reliability of carbon-based chip cutting after cutting. SUMMARY

[0008] The present application aims to provide a system and method for laser processing of carbon-based chips to solve the above problems in the prior art.

[0009] The present application provides a system for laser processing of carbon-based chips, comprising:

[0010] A multi-wavelength laser emission module for emitting laser beams suitable for different crystal structure carbon-based chips according to laser parameters;

[0011] An intelligent control device connected to the multi-wavelength laser emission module for dynamically adjusting laser parameters based on processing strategies according to the thermal and optical properties of carbon-based chips, and sending the laser parameters to the multi-wavelength laser emission module;

[0012] An adaptive machine learning module connected to the intelligent control device for optimizing processing strategies based on monitoring data for the processing characteristics of carbon-based chips, and sending the optimization results to the intelligent control device;

[0013] A micro-nano laser polishing device for processing the surface of carbon-based chips;

[0014] A cutting mechanism for hidden cutting of carbon-based chips;

[0015] A multi-modal sensing and analysis system connected to the adaptive machine learning module for monitoring the structural integrity of carbon-based chips during processing and sending monitoring data to the adaptive machine learning module.

[0016] The present application provides a system and method for laser processing of carbon-based chips, wherein the system for laser processing of carbon-based chips comprises:

[0017] adjusting laser parameters based on the thermal and optical characteristics of the carbon-based chip according to the processing strategy through the intelligent control device, and sending the laser parameters to the multi-wavelength laser emission module;

[0018] emitting laser beams suitable for carbon-based chips with different crystal structures through the multi-wavelength laser emission module according to the laser parameters;

[0019] performing chip hidden cutting by the cutting mechanism using laser-induced stress wave technology;

[0020] performing atomic-level polishing processing on the cutting surface by the micro-nano laser polishing device using femtosecond laser technology;

[0021] monitoring the structural integrity of the carbon-based chip during processing through the multi-modal sensing and analysis system, and sending the monitoring data to the adaptive machine learning module;

[0022] optimizing the processing strategy according to the monitoring data for the processing characteristics of the carbon-based chip through the adaptive machine learning module, and sending the optimization results to the intelligent control device.

[0023] By combining laser hidden cutting, laser polishing and machine learning, the problems existing in the traditional chip cutting method are solved, and the process of predicting and adjusting laser parameters, processing conditions and cutting paths is realized. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the one or more embodiments of the present specification or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the specification, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0025] Figure 1 is a schematic diagram of a laser processing carbon-based chip system according to an embodiment of the present application;

[0026] Figure 2 is a flowchart of a laser processing carbon-based chip method according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order for those skilled in the art to better understand the technical solutions in the one or more embodiments of the present specification, the technical solutions in the one or more embodiments of the present specification will be clearly and completely described in the following with reference to the drawings in the one or more embodiments of the present specification. Obviously, the described embodiments are only a part of the embodiments of the present specification, rather than all the embodiments. Based on the one or more embodiments of the present specification, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present document.

[0028] System embodiment

[0029] According to the embodiments of the present application, a system for laser processing carbon-based chips is provided, Figure 1 is a schematic diagram of the system for laser processing carbon-based chips according to the embodiments of the present application, as Figure 1 shown, the system for laser processing carbon-based chips according to the embodiments of the present application specifically comprises:

[0030] A multi-wavelength laser emission module 10 is used to emit laser beams suitable for carbon-based chips with different crystal structures according to laser parameters; specifically for:

[0031] According to the different allotropes of carbon-based chips, the quantum confinement effect of carbon-based materials, the energy band structure and the optical response characteristics, the optimal laser wavelength is selected, and the specific layer of the multi-layer carbon-based chip structure is selectively processed.

[0032] An intelligent control device 11 is connected with the multi-wavelength laser emission module, used to dynamically adjust the laser parameters based on the processing strategy according to the thermal and optical characteristics of the carbon-based chip, and send the laser parameters to the multi-wavelength laser emission module; specifically for:

[0033] According to the thermal conductivity, thermal expansion coefficient and light absorption characteristics of the carbon-based chip, the material thickness, density or component changes, the power density, pulse width and repetition frequency of the laser are adjusted in real time.

[0034] For graphene, it can use its high absorption in the visible to infrared spectral range to select appropriate wavelengths for precise processing; for carbon nanotubes, considering its chiral and diameter-dependent optical absorption characteristics, select wavelengths matching specific chirality; for diamond thin film, use its wide bandgap characteristics to select ultraviolet band for efficient processing.

[0035] An adaptive machine learning module 12 is connected with the intelligent control device, used to optimize the processing strategy according to the monitoring data for the processing characteristics of the carbon-based chip, and send the optimization results to the intelligent control device; specifically for:

[0036] Using reinforcement learning algorithms, the behavior of different carbon-based materials under various processing conditions is predicted, and real-time quality control is performed to identify and repair potential defects while detecting and handling abnormal situations. By simulating the response of carbon-based chips under different laser parameters, the processing strategy is optimized to improve cutting precision and efficiency.

[0037] Adaptive machine learning algorithms can form unique responses based on the unique crystal structure and physical and chemical properties of different carbon-based chip materials, accurately regulate the interaction between laser and material, and automatically optimize laser power, pulse width, repetition frequency, scanning speed, and cutting path.

[0038] A micro-nano laser polishing device 13 is used to process the surface of a carbon-based chip. Specifically, it is used for:

[0039] Using femtosecond laser technology, the surface is atomically smooth without damaging the sp2 or sp3 hybrid structure of the carbon-based chip. By precisely controlling the laser pulse energy and duration, the hexagonal lattice structure of graphene or the tubular structure of carbon nanotubes is not damaged. By using the ultra-short pulse characteristics of femtosecond lasers, the strong bonding between carbon atoms is not affected, and the surface defects or impurities are selectively acted upon. Considering the high thermal conductivity and low heat capacity of carbon-based materials, special scanning strategies and thermal management techniques are used to prevent local overheating and structural deformation. While maintaining the integrity of the basic structure, functional modification of specific areas of the surface is achieved through laser-induced local chemical reactions. Different types of carbon-based chips are processed from nanometers to microns in size. By using a special laser scanning mode, residual stress introduced during processing is minimized, maintaining the intrinsic electrical and mechanical properties of carbon-based materials. Using the ultra-high peak power and ultra-short pulse width of femtosecond lasers, precise manipulation of individual or a few atoms is achieved, achieving atomic-level surface smoothness.

[0040] By using low-energy, high-repetition-rate femtosecond laser pulses, precise manipulation of carbon atoms is achieved. For graphene, the two-dimensional structure is used to selectively remove atoms along a specific lattice direction by adjusting the polarization direction of the laser, maintaining the integrity of the hexagonal lattice. For carbon nanotubes, considering their cylindrical structure, a spiral scanning mode is used to uniformly remove surface defects without damaging the tube wall structure. For diamond films, their wide bandgap characteristics are used to select appropriate photon energies to achieve precise surface atom detachment. This method precisely controls the pulse duration in the femtosecond range, ensuring that thermal diffusion is limited to the nanoscale, effectively avoiding thermal accumulation effects and maximizing the preservation of sp2 or sp3 hybrid structures of carbon materials. At the same time, by adjusting the laser parameters and scanning strategy in real time, precise layer-by-layer removal of surface atoms can be achieved, not only achieving sub-nanometer smoothness, but also selectively retaining or removing specific surface functional groups to achieve surface chemical modification.

[0041] Cutting mechanism 14 for performing the stealth cutting of carbon-based chips; in particular for:

[0042] Using laser-induced stress wave technology, the direction and intensity of the laser-induced stress wave are adjusted according to the structural anisotropy of the carbon-based material; and the high strength and toughness of the carbon-based material are utilized to achieve clear fracture without edge micro-cracks; at the same time, the atomic configuration of the cutting edge is precisely controlled, and the edge electronic state is optimized, realizing selective and precise segmentation of multi-layer carbon-based structures.

[0043] By precisely controlling the laser pulse energy, repetition frequency, wavelength and spatial distribution, a directional local stress field is generated inside the carbon-based chip, realizing high-precision stealth cutting; for graphene, the anisotropy of its two-dimensional structure is utilized to induce stress wave propagation along a specific lattice direction, realizing atomic-level precision cutting; for carbon nanotubes, considering the mechanical properties of its tubular structure, precise segmentation is achieved without damaging the tube wall structure by adjusting the stress wave intensity; for diamond thin film, its high Young's modulus characteristics are utilized to achieve precise cutting through the cumulative effect of high-frequency low-energy pulses; this method minimizes the heat-affected zone, avoids sp2-sp3 hybridization transition and defect formation, and at the same time utilizes the high thermal conductivity of carbon materials to quickly dissipate heat, preventing local thermal damage; by adjusting the laser parameters and scanning path in real time, the atomic configuration of the cutting edge can be controlled, the edge electronic state can be optimized, and even functional edge structures can be realized.

[0044] Multi-modal sensing and analysis system 15 connected to the adaptive machine learning module for monitoring the structural integrity of the carbon-based chip during processing and sending the monitoring data to the adaptive machine learning module.

[0045] The multi-modal sensing and analysis system is specifically a Raman spectroscopy analyzer;

[0046] The Raman spectroscopy analyzer is specifically used for: real-time monitoring of structural changes and defect formation in carbon-based chips; monitoring changes in the graphitization degree of carbon-based materials; plotting stress distribution around the cutting area; distinguishing and quantifying different types of defects; real-time monitoring of the number of layers of multi-layer carbon-based materials; identifying possible impurities or contaminants; combining with a miniature electrical measurement device, real-time evaluation of the impact of the cutting process on the electrical properties of the material; monitoring local temperature changes during processing.

[0047] Through high-time-resolution Raman spectrum acquisition, real-time analysis of changes in characteristic peaks, including but not limited to G peak (~ 1580 cm -1 ), D peak (~ 1350 cm -1 ) and 2D peak (~ 2700 cm -1). For graphene, by monitoring the intensity ratio of G and 2D peaks and the shape of 2D peak, the layer number change and stress state are evaluated in real-time; for carbon nanotubes, by analyzing the shift of radial breathing mode (RBM) peak, the change of tube diameter and chirality is determined; for diamond thin film, by monitoring the intensity and position of sp3 carbon characteristic peak (~1332 cm -1 ), the purity and stress state are evaluated; by analyzing the intensity ratio of G and D peaks, the change of defect density is quantified in real-time, where the increase of D peak intensity indicates the increase of defects or disorder; meanwhile, by monitoring the position shift of G peak, the stress state introduced by processing is evaluated; based on these real-time data, the system can timely adjust the laser parameters, such as power, pulse width, scanning speed, etc., to minimize the structural damage; in addition, by analyzing the shape and intensity of G' peak (also known as 2D peak), the change of interlayer coupling strength can be evaluated, providing guidance for precise processing of multi-layer carbon-based structures.

[0048] The technical solutions of the present application will be described in detail below in conjunction with specific embodiments. The examples given are only used to explain the present application and do not limit the scope of the present application.

[0049] Embodiment 1:

[0050] This embodiment takes cutting graphene chips as an example, and the working process is as follows:

[0051] Laser wavelength selection:

[0052] A multi-wavelength laser emission module is used to select the best laser wavelength for the unique optical properties of graphene. Considering the high absorption of graphene in the visible to near-infrared spectral range (400-1100 nm), we choose a wavelength of 800 nm femtosecond laser. This wavelength not only can be effectively absorbed by graphene, but also can minimize the impact on the substrate material. At the same time, the ultra-short pulse characteristics of femtosecond laser can significantly reduce the heat affected zone, maintaining the lattice integrity of graphene.

[0053] Parameter optimization:

[0054] The intelligent control device optimizes the laser parameters according to the physical properties of graphene. Considering the high thermal conductivity (~5000 W / mK) and low heat capacity (~700 J / kgK) of graphene, we use high repetition frequency (80 MHz) and low single pulse energy (~1 nJ) laser parameter settings. This configuration can achieve effective processing while maintaining low heat accumulation effect through multi-pulse superposition. The laser power density is set to 10-12 W / cm 2 , which is sufficient to initiate a multi-photon absorption process to achieve precise material removal while avoiding excessive thermal damage.

[0055] Hidden cutting process:

[0056] We use linearly polarized laser, with the polarization direction aligned with the graphene lattice direction. This method can take advantage of the structural anisotropy of graphene, inducing stress wave propagation along specific lattice directions (e.g., zigzag or armchair edges), achieving atomic-level precision cutting. The laser scanning speed is set to 10 mm / s, which ensures sufficient energy deposition for cutting while avoiding excessive heat accumulation.

[0057] Surface polishing:

[0058] We use femtosecond laser technology for surface polishing. We use circularly polarized laser, with the pulse width adjusted to 30 fs and the single-pulse energy reduced to 0.5 nJ. Circular polarization can achieve more uniform energy distribution, avoiding preferential etching directions. By precisely controlling the laser energy density (~0.1 J / cm 2 ), we can achieve single-atomic layer material removal. We use a grating scanning mode, with a 50% overlap rate for each scan to ensure uniform polishing effects.

[0059] Real-time monitoring:

[0060] We use a high-time-resolution (~1 ms) Raman spectrometer to monitor the processing in real time. We mainly focus on the following characteristic peaks:

[0061] G peak (~1580 cm -1 ): Reflects the in-plane vibration mode of sp2 carbon atoms;

[0062] 2D peak (~2700 cm -1 ): Used to determine the number of layers and stress state;

[0063] D peak (~1350 cm -1 ): Indicates defects or edge effects.

[0064] By analyzing the intensity ratio of G peak and 2D peak (2D / G), we can evaluate the number of layers of graphene in real time. The ideal single-layer graphene 2D / G ratio should be greater than 2. At the same time, monitor the full width at half maximum (FWHM) of the 2D peak, the 2D peak FWHM of single-layer graphene should be less than 30 cm -1 .

[0065] In addition, we also focus on the position shift of G peak, every 1 cm -1 red shift corresponds to about 0.003 strain. In this way, we can accurately control the stress introduced during processing, ensuring that it does not exceed the fracture limit of graphene (~25%).

[0066] Adaptive optimization:

[0067] The adaptive machine learning algorithm dynamically optimizes the processing parameters based on real-time monitoring data. We employ a reinforcement learning algorithm, taking Raman spectrum data as state input and laser parameters (power, scanning speed, repetition rate, etc.) as action output. The reward function design considers the following factors:

[0068] Minimizing D peak intensity while maintaining low defect density;

[0069] Maintaining ideal I2D / IG ratio and 2D peak FWHM to ensure graphene quality;

[0070] Controlling G peak position shift within a reasonable range to avoid excessive stress.

[0071] The algorithm learns through multiple iterations to optimize processing efficiency while ensuring quality. For example, when detecting an increase in D peak intensity, the algorithm automatically reduces laser power or increases scanning speed; when finding a multi-layer region (2D / G < 2), it appropriately increases laser energy density to achieve layer-by-layer peeling.

[0072] Through this fine parameter control and real-time optimization, we can achieve high-precision processing of graphene chips while maximizing their unique electrical, thermal, and mechanical properties. This method not only applies to single-layer graphene but also extends to multi-layer graphene and other two-dimensional material precision processing, providing important technical support for the development of new carbon-based electronic devices and quantum devices.

[0073] Example 2:

[0074] This example takes cutting carbon nanotube chips as an example, the working process is as follows:

[0075] Laser wavelength selection:

[0076] Using a multi-wavelength laser emission module, we select the best laser wavelength based on the unique optical properties of carbon nanotubes. Considering that the optical absorption properties of carbon nanotubes are strongly dependent on their chirality and diameter, we use a tunable femtosecond laser system with a wavelength range of 400-1600 nm. This system can select the best absorption wavelength for different types of carbon nanotubes (metallic or semiconducting). For example, for semiconducting carbon nanotubes with a diameter of 1.2 nm, we choose a laser with a wavelength of about 980 nm, which matches the first exciton transition energy level; for metallic carbon nanotubes, we choose a wavelength of about 700 nm, which corresponds to the plasmon resonance.

[0077] Parameter optimization:

[0078] The smart control device optimizes the laser parameters according to the physical properties of carbon nanotubes. Considering the one-dimensional structure and high thermal conductivity (~3000 W / mK along the axial direction) of carbon nanotubes, we use a low repetition rate (1 kHz) and high single-pulse energy (~100 μJ) laser parameter setting. This configuration can allow sufficient heat dissipation between each pulse to prevent the destruction of the tube wall structure caused by the thermal accumulation effect. The laser power density is set to 10-13 W / cm 2 This range is sufficient to initiate multi-photon absorption and tunneling ionization processes, achieving precise material removal while minimizing the heat-affected zone.

[0079] Blind cutting process:

[0080] Blind cutting of the chip using laser-induced stress wave technology. We use circularly polarized laser light to ensure uniform action on all orientations of carbon nanotubes. The laser focal diameter is adjusted to 5 μm, slightly larger than the typical length of carbon nanotubes (1-3 μm) to ensure complete cutting. The laser scanning speed is set to 0.1 mm / s, which is a relatively slow speed that can ensure sufficient energy deposition to achieve cutting while controlling the heat-affected zone. We use a multi-scan strategy, gradually increasing the energy density of each scan, which can gradually weaken the tube wall structure of carbon nanotubes and ultimately achieve precise segmentation without causing large-scale thermal damage.

[0081] Surface polishing:

[0082] Surface polishing using femtosecond laser technology. We use a spiral scanning mode, with the pulse width adjusted to 50 fs and the single-pulse energy reduced to 10 μJ. The spiral scanning mode can uniformly process carbon nanotubes of different orientations, avoiding the creation of preferential etching directions. By precisely controlling the laser energy density (~0.5 J / cm 2 ), we can achieve selective removal of surface defects and amorphous carbon without damaging the intrinsic tubular structure of carbon nanotubes. The scanning overlap rate is set to 75% to ensure uniform polishing effect.

[0083] Real-time monitoring:

[0084] Real-time monitoring of the processing process through a high-time-resolution (~10 ms) Raman spectrometer. We mainly focus on the following characteristic peaks:

[0085] G peak (~1590 cm -1 ): reflects the tangential vibration mode of sp2 carbon atoms;

[0086] D peak (~1350 cm -1 ): indicates defects or disorder degree;

[0087] RBM peak (100-350 cm -1): Reflecting the radial breathing mode of carbon nanotubes, used to determine tube diameter and chirality.

[0088] By analyzing the positions of the RBM peaks, we can assess the diameter changes of the carbon nanotubes in real-time. The RBM frequency is inversely proportional to the tube diameter, with an approximate relationship of ωRBM= 248 / d (d is the tube diameter in nm). Meanwhile, we monitor the intensity ratio of the G peak and the D peak (G / D) to assess the defect density introduced during processing. Ideally, G / D should be greater than 100, indicating that the high-quality tubular structure is maintained.

[0089] Additionally, we pay attention to the splitting of the G peak (G+ and G- peaks), which can be used to distinguish between metallic and semiconducting carbon nanotubes, ensuring that the processing process does not selectively destroy a certain type of nanotube.

[0090] Adaptive optimization:

[0091] The adaptive machine learning algorithm dynamically optimizes the processing parameters based on real-time monitoring data. We use a deep reinforcement learning algorithm, taking Raman spectrum data and electrical measurement data (such as conductivity changes) of the processing area as state inputs, and laser parameters (wavelength, power, scanning speed, repetition frequency, etc.) as action outputs. The function design takes the following factors into account:

[0092] Minimize the intensity of the D peak to maintain low defect density;

[0093] Maintain the intensity and position of the RBM peak to ensure the integrity of the carbon nanotube structure;

[0094] Control the degree of G peak splitting to maintain the balance between metallic and semiconducting carbon nanotubes;

[0095] Optimize electrical performance, such as maintaining or improving conductivity.

[0096] The algorithm learns through multiple iterations to optimize processing efficiency while ensuring processing quality. For example, when detecting an increase in D peak intensity, the algorithm will automatically adjust the laser wavelength or reduce the power density; when certain RBM peaks disappear, it will appropriately adjust the scanning strategy to avoid selectively destroying nanotubes of a specific diameter.

[0097] Through this fine parameter control and real-time optimization, we can achieve high-precision processing of carbon nanotube chips while maximizing their unique one-dimensional quantum confinement effect and excellent electrical and thermal performance. This method not only applies to single-walled carbon nanotubes but can also be extended to the precise processing of multi-walled carbon nanotubes and carbon nanotube arrays, providing important technical support for the development of new carbon-based electronic devices, sensors, and quantum devices.

[0098] Example 3

[0099] This example takes cutting diamond thin film chips as an example, the working process is as follows:

[0100] Laser wavelength selection:

[0101] Utilize multi-wavelength laser emission module, select the best laser wavelength according to the unique optical properties of diamond. Considering the wide bandgap properties of diamond (~ 5.5eV), we choose deep ultraviolet femtosecond laser, wavelength is 213nm (fifth harmonic Nd:YAG). The corresponding photon energy (5.8eV) of this wavelength is slightly higher than the bandgap of diamond, which can realize efficient single photon absorption, and minimize the thermal influence zone. In addition, short wavelength laser can also provide smaller focused spot, which is beneficial to improve the machining precision.

[0102] Parameter optimization:

[0103] Intelligent control device according to the physical properties of diamond laser parameter optimization. Considering the high thermal conductivity of diamond (~2200W / mK), high Young's modulus (~1050GPa) and high melting point (~3550℃), we use high repetition frequency (100kHz), medium single pulse energy (~10μJ) laser parameter setting. The pulse duration is set to 100fs, which can complete energy deposition before material heat diffusion, effectively reducing the thermal influence zone. Laser power density is set to 10-14W / cm 2 , this range is enough to trigger multi-photon ionization and avalanche ionization process, realize efficient material removal.

[0104] Hidden cutting process:

[0105] Use laser-induced stress wave technology to cut the chip. We use linearly polarized laser, make the polarization direction parallel to the {111} crystal plane of diamond, which can use the cleavage properties of diamond, realize accurate cutting along the specific crystal plane. Laser focusing uses high numerical aperture (NA=0.8) ultraviolet objective, the focal spot diameter is controlled at about 300nm. Adopt multiple scanning strategy, the energy density of each scan is gradually increased, the initial energy density is set to 2J / cm 2 , finally increased to 5J / cm 2 . The scanning speed is set to 1mm / s, which can ensure enough energy deposition to realize cutting, while controlling the heat accumulation effect.

[0106] Surface polishing:

[0107] Use femtosecond laser technology for surface polishing. We use random scanning mode, pulse width is reduced to 50fs, single pulse energy is reduced to 1μJ. Random scanning mode can avoid the generation of periodic surface structure, realize more uniform polishing effect. By accurately controlling the laser energy (~0.8J / cm 2), we can achieve precise desorption of surface atoms without causing damage to the underlying material. The scan overlap ratio is set to 90% to ensure full coverage. The entire polishing process is divided into two stages: coarse polishing and fine polishing, gradually reducing the pulse energy and increasing the scan speed, ultimately achieving atomic-level flatness.

[0108] Real-time monitoring:

[0109] The processing is monitored in real-time by a high temporal resolution (~1ms) Raman spectrometer. We focus on the following characteristic peaks:

[0110] Diamond characteristic peak (~1332 cm -1 ): Reflects the vibration mode of sp3 bonded carbon atoms;

[0111] G peak (~1580 cm -1 ): Indicates the formation of sp2 carbon structure;

[0112] D peak (~1350 cm -1 ): Represents disordered carbon structure or defects.

[0113] By analyzing the intensity, position and full width at half maximum (FWHM) of the diamond characteristic peak, we can evaluate the diamond quality and stress state in real time. Ideally, the peak position should be at 1332±0.5 cm -1 , and the FWHM should be less than 2 cm -1 . At the same time, monitor the appearance and intensity changes of G and D peaks to evaluate the possible graphitization or amorphous carbon during processing.

[0114] In addition, we also introduce photoluminescence (PL) spectrum analysis to monitor the formation of NV centers (zero phonon line at 637 nm) in real time, which is crucial for developing quantum sensing and quantum computing applications.

[0115] Adaptive optimization:

[0116] The adaptive machine learning algorithm dynamically optimizes the processing parameters based on real-time monitoring data. We use a deep reinforcement learning algorithm, taking Raman spectrum data, PL spectrum data and electrical measurement data (such as breakdown voltage) of the processing area as state input, and laser parameters (power, scan speed, repetition frequency, polarization direction, etc.) as action output. The reward function design takes the following factors into account:

[0117] Maximize diamond characteristic peak intensity, maintain high-quality sp3 structure;

[0118] Minimize G and D peak intensity to avoid graphitization and defect formation;

[0119] Control the position deviation of the diamond characteristic peak within ±0.5 cm -1 , manage residual stress;

[0120] Optimize the formation of NV centers if needed;

[0121] Maintain or improve the breakdown voltage to ensure device performance.

[0122] The algorithm learns through multiple iterations to optimize processing efficiency while ensuring processing quality. For example, when detecting an increase in G peak intensity, the algorithm automatically reduces single pulse energy or increases scanning speed; when observing diamond characteristic peak broadening, adjust the laser polarization direction to optimize stress distribution.

[0123] Through this fine parameter control and real-time optimization, we can achieve high-precision processing of diamond thin film chips while maximizing their unique wide bandgap characteristics, high breakdown voltage and excellent thermal performance. This method not only applies to ordinary diamond thin films, but also extends to the precise processing of doped diamond, single crystal diamond and nanostructured diamond, providing important technical support for the development of new high-power electronic devices, quantum sensors and quantum computing platforms.

[0124] Compared with the prior art, the embodiment of the present application has the following beneficial effects:

[0125] (1) Precise control and optimization: The present application uses machine learning technology to predict and adjust the chip cutting process, which can realize precise control and optimization of laser parameters, processing conditions and cutting path, etc. Compared with traditional mechanical cutting and chemical etching methods, it can more accurately control the cutting process and improve the consistency and accuracy of cutting.

[0126] (2) Surface quality improvement: The application of laser hidden cutting technology and laser polishing technology can significantly improve the surface quality of the cut chip. Laser hidden cutting technology reduces the size of the heat-affected zone, reduces the surface roughness and the possibility of residue generation. Laser polishing technology further improves the smoothness and quality of the surface through photoexcitation and chemical reaction. Precise and controllable hidden cutting technology not only maintains the intrinsic electrical, thermal and mechanical properties of carbon-based materials, but also provides a micro-nano processing means for designing new carbon-based devices, opening up new ways for performance optimization and function expansion of carbon-based chips. Compared with traditional methods, a smoother and smoother chip surface can be obtained.

[0127] (3) High efficiency and reliability: The present application uses laser technology for cutting, which has the characteristics of high speed and high efficiency. The laser cutting process is fast and accurate, which can improve production efficiency. Highly precise atomic-level polishing technology not only improves the surface quality of carbon-based chips, but also provides a key processing means for regulating its electronic structure, optimizing interface characteristics and designing new quantum devices, opening up new possibilities for the application of carbon-based chips in nanoelectronics, optoelectronics and quantum computing.

[0128] (4) Flexibility and adjustability: The laser parameters, processing conditions, and cutting paths in the invention can be flexibly adjusted according to different chip materials and cutting requirements. With the support of machine learning technology, adaptive control and optimization can be achieved based on real-time data and feedback. This makes the invention more adaptable and flexible for a variety of chip materials and different cutting requirements.

[0129] (5) Real-time monitoring: The highly sensitive and real-time monitoring method not only ensures the integrity of the structure and performance of carbon-based materials during processing, but also provides precise control means for optimizing processing strategies and achieving specific functional structures (such as controlling edge states and inducing specific defects), further promoting the application and development of carbon-based chips in frontier fields such as quantum devices, high-performance sensors, and new electronic devices.

[0130] (6) Programmability: The invention uses machine learning technology to automatically and programmably control the laser cutting process. Through training models and real-time feedback, optimization and adjustment can be made according to specific cutting requirements and material characteristics. Not only does it consider the anisotropy and thermal conductivity of materials, but it also adapts to the phase transition behavior of carbon-based materials at high temperatures, achieving optimal processing precision and efficiency while maximizing the unique performance of carbon-based materials.

[0131] (7) Cost savings: Compared with traditional cutting methods, the invention can bring cost savings in some aspects. First, laser cutting technology can achieve high-speed and efficient cutting, improving production efficiency and reducing production time and labor costs. Second, since laser cutting is non-contact, it reduces damage and waste production, reduces yield and material waste, and helps reduce production costs.

[0132] (8) Sustainability: Compared with some traditional cutting methods, the invention does not require the use of chemicals during laser cutting, reducing environmental pollution and waste disposal requirements. In addition, since laser cutting can achieve precise control and optimization, it can reduce material waste and energy consumption, improve resource utilization, and have a positive impact on sustainable development.

[0133] Method embodiment

[0134] According to the embodiment of the invention, a method for laser processing of carbon-based chips is provided, which is used in the system for laser processing of carbon-based chips, Figure 2 is the flowchart of the method for laser processing of carbon-based chips according to the embodiment of the invention, as Figure 2 shown, the method for laser processing of carbon-based chips according to the embodiment of the invention specifically includes:

[0135] Step S201, dynamically adjust the laser parameters based on the thermal and optical properties of the carbon-based chip through the intelligent control device according to the processing strategy, and send the laser parameters to the multi-wavelength laser emission module; specifically including:

[0136] According to the changes of the thermal conductivity, thermal expansion coefficient and optical absorption characteristics, material thickness, density or composition of the carbon-based chip, the power density, pulse width and repetition frequency of the laser are adjusted in real time;

[0137] Step S202, emitting laser beams suitable for different crystal structure carbon-based chips through the multi-wavelength laser emission module according to the laser parameters; specifically including:

[0138] According to the different allotropes of carbon-based chips, the quantum confinement effect, energy band structure and optical response characteristics of carbon-based materials, the best laser wavelength is selected, and the specific layer of the multi-layer carbon-based chip structure is selectively processed;

[0139] Among them, for graphene, it can use its high absorption in the visible to infrared spectrum range to select appropriate wavelength to achieve accurate processing; for carbon nanotubes, considering its chiral and diameter-dependent optical absorption characteristics, select the wavelength matching the specific chirality; for diamond film, use its wide bandgap characteristics to select ultraviolet band to achieve efficient processing.

[0140] Step S203, using laser-induced stress wave technology to perform chip hidden cutting through the cutting mechanism; specifically including:

[0141] Using laser-induced stress wave technology, adjust the direction and intensity of laser-induced stress wave according to the structural anisotropy of carbon-based materials; and use the high strength and toughness of carbon-based materials to achieve clear fracture without edge micro-cracks; at the same time, accurately control the atomic configuration of the cutting edge, optimize the edge electronic state, and realize selective and accurate segmentation of multi-layer carbon-based structure.

[0142] By accurately controlling the laser pulse energy, repetition frequency, wavelength and spatial distribution, a directional local stress field is generated inside the carbon-based chip to achieve high-precision hidden cutting; for graphene, use the anisotropy of its two-dimensional structure to induce stress wave propagation along a specific lattice direction to achieve atomic-level precision cutting; for carbon nanotubes, considering the mechanical properties of its tubular structure, precise segmentation is achieved without damaging the tube wall structure by adjusting the stress wave intensity; for diamond film, use its high Young's modulus characteristics to achieve precise cutting through the cumulative effect of high-frequency low-energy pulses; this method minimizes the heat affected zone, avoids sp2-sp3 hybridization transition and defect formation, and at the same time uses the high thermal conductivity of carbon materials to quickly dissipate heat and prevent local thermal damage; by adjusting the laser parameters and scanning path in real time, the atomic configuration of the cutting edge can be controlled, the edge electronic state can be optimized, and even a functional edge structure can be realized.

[0143] Step S204, using femtosecond laser technology, the micro-nano laser polishing device is used for atomic level polishing treatment on the cutting surface; Specifically, it includes:

[0144] Using femtosecond laser technology, the surface is atomically flat without destroying the sp2 or sp3 hybrid structure of the carbon-based chip; By precisely controlling the laser pulse energy and duration, the hexagonal lattice structure of graphene or the tubular structure of carbon nanotubes is avoided; By using the ultra-short pulse characteristics of femtosecond laser, the strong bonding between carbon atoms is not affected, and the surface defects or impurities are selectively acted on; Considering the high thermal conductivity and low heat capacity characteristics of carbon-based materials, special scanning strategies and thermal management techniques are used to prevent local overheating caused by structural deformation; While maintaining the integrity of the basic structure, functional modification of specific areas on the surface is achieved through laser-induced local chemical reactions; According to the needs of different types of carbon-based chips, surface features from nanometers to microns are simultaneously processed; By using a special laser scanning mode, the residual stress introduced during processing is minimized, and the intrinsic electrical and mechanical properties of carbon-based materials are maintained; By using the ultra-high peak power and ultra-short pulse width of femtosecond laser, precise manipulation of single or few atoms is achieved, and atomic-level surface flatness is achieved;

[0145] By using low-energy, high-repetition-rate femtosecond laser pulses, precise manipulation of carbon atoms is achieved; For graphene, by adjusting the polarization direction of the laser, atoms are selectively removed along a specific lattice direction to maintain the integrity of the hexagonal lattice; For carbon nanotubes, considering its cylindrical structure, a spiral scanning mode is adopted to uniformly remove surface defects without damaging the tube wall structure; For diamond film, by using its wide bandgap characteristics, appropriate photon energy is selected to achieve precise surface atom detachment; This method precisely controls the pulse duration in femtosecond order, ensures that thermal diffusion is limited to nanoscale, effectively avoids thermal accumulation effect, and maximizes the sp2 or sp3 hybrid structure of carbon materials; At the same time, by adjusting the laser parameters and scanning strategy in real time, precise layer-by-layer removal of surface atoms can be achieved, not only achieving sub-nanometer flatness, but also selectively retaining or removing specific surface functional groups to achieve surface chemical modification.

[0146] Step S205, through the multi-modal sensing and analysis system, the structural integrity of the carbon-based chip during processing is monitored, and the monitoring data is sent to the adaptive machine learning module; Specifically, it includes:

[0147] Real-time monitoring of structural changes and defect formation in carbon-based chips using a Raman spectroscopy analyzer; monitoring changes in the degree of graphitization of carbon-based materials; mapping stress distribution around the cutting area; distinguishing and quantifying different types of defects; real-time monitoring of the number of layers in multi-layer carbon-based materials; identifying possible impurities or contaminants; combining with a miniature electrical measurement device, real-time evaluation of the impact of the cutting process on the electrical properties of the material; monitoring local temperature changes during processing;

[0148] Through high time resolution Raman spectrum acquisition, real-time analysis of characteristic peak changes, including but not limited to G peak (~ 1580 cm -1 ), D peak (~ 1350 cm -1 ) and 2D peak (~ 2700 cm -1 ). For graphene, by monitoring the intensity ratio of G peak and 2D peak and the shape of 2D peak, real-time evaluation of layer number change and stress state; for carbon nanotubes, analyze the displacement of radial breathing mode (RBM) peak to judge the change of tube diameter and chirality; for diamond thin film, monitor the intensity and position of sp3 carbon characteristic peak (~ 1332 cm -1 ), evaluate purity and stress state; by analyzing the intensity ratio of G peak and D peak, real-time quantification of defect density change, where D peak intensity increase indicates defect or disorder degree increase; at the same time, by monitoring the position offset of G peak, evaluate the stress state introduced by processing; based on these real-time data, the system can timely adjust laser parameters such as power, pulse width, scanning speed, etc. to minimize structural damage; in addition, by analyzing the shape and intensity of G' peak (also known as 2D peak), the change of interlayer coupling strength can be evaluated to provide guidance for precise processing of multi-layer carbon-based structures.

[0149] Step S206, through the adaptive machine learning module, according to the monitoring data, optimize the processing strategy for the processing characteristics of the carbon-based chip, and send the optimization result to the intelligent control device. Specifically includes:

[0150] Using reinforcement learning algorithm, predicting the behavior of different carbon-based materials under various processing conditions, and performing real-time quality control, identifying and repairing potential defects, detecting and handling abnormal situations, optimizing processing strategy by simulating the response of carbon-based chips under different laser parameters, improving cutting precision and efficiency.

[0151] The adaptive machine learning algorithm can form a unique response according to the unique crystal structure and physical and chemical properties of different carbon-based chip materials, accurately regulate the interaction between laser and material, and automatically optimize laser power, pulse width, repetition frequency, scanning speed and cutting path.

[0152] The embodiment of the present application is a method embodiment corresponding to the above system embodiment, and the specific operation of each step can be understood with reference to the description of the method embodiment, which will not be repeated here.

[0153] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A system for laser processing carbon-based chips, characterized in that, The application relates to a multi-wavelength laser emission module for emitting laser beams suitable for different carbon-based chip structures according to laser parameters; specifically for: selecting the optimal laser wavelength according to the different allotropes of the carbon-based chip, the quantum confinement effect of the carbon-based material, the energy band structure and the optical response characteristics, and selectively processing specific layers of the multi-layer carbon-based chip structure; an intelligent control device connected with the multi-wavelength laser emission module for dynamically adjusting the laser parameters based on the processing strategy according to the thermal and optical characteristics of the carbon-based chip and sending the laser parameters to the multi-wavelength laser emission module; specifically for: adjusting the power density, pulse width and repetition frequency of the laser in real time according to the changes of the thermal conductivity, thermal expansion coefficient and light absorption characteristics of the carbon-based chip, the material thickness, density or composition; an adaptive machine learning module connected with the intelligent control device for optimizing the processing strategy according to the monitoring data and the processing characteristics of the carbon-based chip and sending the optimization results to the intelligent control device; a micro-nano laser polishing device for processing the surface of the carbon-based chip; a cutting mechanism for hidden cutting of the carbon-based chip; a multi-modal sensing and analysis system connected with the adaptive machine learning module for monitoring the structural integrity of the carbon-based chip during the processing and sending the monitoring data to the adaptive machine learning module. The adaptive machine learning module is specifically used for:

2. The system of claim 1, wherein, adopting a reinforcement learning algorithm to predict the behavior of different carbon-based materials under various processing conditions, perform real-time quality control, identify and repair potential defects, detect and handle abnormal situations, optimize the processing strategy by simulating the response of the carbon-based chip under different laser parameters, and improve the cutting precision and efficiency. The micro-nano laser polishing device is specifically used for:

3. The system of claim 1, wherein, adopting femtosecond laser technology to realize atomic-level surface flatness without damaging the sp2 or sp3 hybrid structure of the carbon-based chip, avoiding damage to the hexagonal lattice structure of graphene or the tubular structure of carbon nanotubes by accurately controlling the laser pulse energy and duration; selectively acting on surface defects or impurities without affecting the strong bonding between carbon atoms by using the ultra-short pulse characteristics of femtosecond laser; considering the high thermal conductivity and low heat capacity characteristics of carbon-based materials, adopting special scanning strategies and heat management technology to prevent local overheating from causing structural deformation; realizing functional modification of specific areas on the surface by laser-induced local chemical reaction while maintaining the integrity of the basic structure; simultaneously processing surface features from nanometers to microns according to the needs of different types of carbon-based chips; minimizing the residual stress introduced during the processing by using a special laser scanning mode to maintain the intrinsic electrical and mechanical properties of the carbon-based material; using the ultra-high peak power and ultra-short pulse width of femtosecond laser to realize accurate manipulation of single or few atoms and achieve atomic-level surface flatness. The cutting mechanism is specifically used for:

4. The system of claim 1, wherein, ​ The laser-induced stress wave technology is adopted to adjust the direction and intensity of the laser-induced stress wave according to the structural anisotropy of the carbon-based material; and the high strength and toughness of the carbon-based material are utilized to realize clear fracture without edge micro-cracks; meanwhile, the atomic configuration of the cutting edge is accurately controlled, and the edge electronic state is optimized, so that the multi-layer carbon-based structure is selectively and accurately divided.

5. The system of claim 1, wherein, The multi-modal sensing and analysis system is specifically a Raman spectrum analyzer. The Raman spectrum analyzer is specifically used for monitoring the structural changes and defect formation of the carbon-based chip in real time, monitoring the graphitization degree changes of the carbon-based material, drawing a stress distribution map around the cutting area, distinguishing and quantifying different types of defects, monitoring the layer number changes of the multi-layer carbon-based material in real time, identifying possible impurities or pollutants, combining a micro electrical measurement device to evaluate the influence of the cutting process on the electrical properties of the material in real time, and monitoring the local temperature changes in the processing process.

6. A method for laser processing a carbon-based chip, used in the system for laser processing a carbon-based chip according to any one of claims 1 to 5, and specifically comprising the following steps: The intelligent control device dynamically adjusts the laser parameters based on the processing strategy according to the thermal and optical properties of the carbon-based chip, and sends the laser parameters to the multi-wavelength laser emission module; specifically including: According to the changes of the thermal conductivity, thermal expansion coefficient and optical absorption properties of the carbon-based chip, the material thickness, density or composition, the power density, pulse width and repetition frequency of the laser are adjusted in real time; The multi-wavelength laser emission module emits laser beams suitable for different crystal structure carbon-based chips according to the laser parameters; specifically including: According to the different allotropes of the carbon-based chip, the quantum confinement effect, the band structure and the optical response characteristics of the carbon-based material, the optimal laser wavelength is selected, and the specific layer of the multi-layer carbon-based chip structure is selectively processed; The cutting mechanism uses laser-induced stress wave technology to perform chip hidden cutting; The micro-nano laser polishing device uses femtosecond laser technology to perform atomic-level polishing treatment on the cutting surface; The multi-modal sensing and analysis system monitors the structural integrity of the carbon-based chip during the processing process, and sends the monitoring data to the adaptive machine learning module; The adaptive machine learning module optimizes the processing strategy according to the monitoring data based on the processing characteristics of the carbon-based chip, and sends the optimization results to the intelligent control device.

7. The method according to claim 6, wherein The cutting mechanism uses laser-induced stress wave technology to perform chip hidden cutting, specifically including: The laser-induced stress wave technology is adopted to adjust the direction and intensity of the laser-induced stress wave according to the structural anisotropy of the carbon-based material; and the high strength and toughness of the carbon-based material are utilized to realize clear fracture without edge micro-cracks; meanwhile, the atomic configuration of the cutting edge is accurately controlled, and the edge electronic state is optimized, so that the multi-layer carbon-based structure is selectively and accurately divided.

8. The method according to claim 6, wherein The micro-nano laser polishing device uses femtosecond laser technology to perform atomic-level polishing treatment on the cutting surface, specifically including: Using femtosecond laser technology, the surface is atomically smooth without damaging the sp2 or sp3 hybrid structure of the carbon-based chip. By precisely controlling the laser pulse energy and duration, the hexagonal lattice structure of graphene or the tubular structure of carbon nanotubes is avoided. Using the ultra-short pulse characteristics of femtosecond laser, the strong bonding between carbon atoms is not affected, and the surface defects or impurities are selectively acted on. Considering the high thermal conductivity and low heat capacity characteristics of carbon-based materials, special scanning strategies and thermal management techniques are used to prevent local overheating and structural deformation. While maintaining the integrity of the basic structure, the functional modification of the surface of specific areas is realized through laser-induced local chemical reactions. For different types of carbon-based chip requirements, surface features from nanometers to microns are simultaneously processed. Through special laser scanning mode, the residual stress introduced in the processing process is minimized, and the intrinsic electrical and mechanical properties of carbon-based materials are maintained. Using the ultra-high peak power and ultra-short pulse width of femtosecond laser, precise manipulation of single or few atoms is realized, and atomic-level surface flatness is achieved. Through a multi-modal sensing and analysis system, the structural integrity of the carbon-based chip during processing is monitored, including: Through a Raman spectroscopy analyzer, real-time monitoring of structural changes and defect formation in carbon-based chips is performed. The graphitization degree of carbon-based materials is monitored. Stress distribution maps around the cutting area are drawn. Different types of defects are distinguished and quantified. The number of layers of multi-layer carbon-based materials is monitored in real time. Possible impurities or contaminants are identified. Combined with a miniature electrical measurement device, the impact of the cutting process on the electrical properties of the material is evaluated in real time. Local temperature changes during processing are monitored. Through an adaptive machine learning module, based on monitoring data, the processing strategy is optimized for the processing characteristics of carbon-based chips, including: Using reinforcement learning algorithms, the behavior of different carbon-based materials under various processing conditions is predicted, and real-time quality control is performed. Potential defects are identified and repaired, and abnormal situations are detected and handled. By simulating the response of carbon-based chips under different laser parameters, the processing strategy is optimized, and the cutting precision and efficiency are improved.

Images