Method and system for non-destructive removal of quantum chip metal oxide layer based on femtosecond laser

Abstract

This invention discloses a method and system for non-destructive removal of metal oxide layers in quantum chips based on femtosecond lasers, relating to the field of laser processing technology. The system includes a laser device comprising a femtosecond laser, a continuous laser, a beam shaper, a pulse delay line, and a laser scanning module; a processing platform comprising an optical motion platform, an active focusing module, a wafer vacuum chuck, and a pump; a control system for controlling laser parameters, processing path, and processing speed; and a detection feedback system comprising a laser confocal microscope. The laser confocal microscope is used for three-dimensional quantitative measurement of the residual oxide layer quality, real-time detection of the processing process, and optimization of processing parameters based on the detection results. Through continuous laser thermal activation and femtosecond laser selective ablation, high-precision non-destructive removal of the oxide layer is achieved based on the differences in material thermophysical properties. Closed-loop control of the processing platform and detection feedback enables precise non-destructive removal of the oxide layer in micro-areas.

Description

Technical Field

[0001] This invention belongs to the field of laser processing technology, specifically relating to a method and system for non-destructive removal of the metal oxide layer of quantum chip qubit electrodes based on femtosecond lasers. Background Technology

[0002] In the manufacturing process of precision electronic devices, especially quantum computing devices, oxide layers easily form on the surface of the metal electrodes of quantum chip qubits. These oxide layers reduce electrode conductivity, affect device performance, and can even lead to device failure. Meanwhile, core components in quantum computing devices, such as Josephson junctions, have extremely high surface quality requirements. Traditional oxide layer removal methods, such as acid pickling and mechanical polishing, easily damage the metal substrate and leave residual impurities, failing to meet the demands of high-precision processing. Existing laser processing technology can now remove oxide layers from metal surfaces and has a certain application basis in the manufacturing of micro-nano electronics and quantum devices.

[0003] However, in scenarios involving the precise removal of nanoscale oxide layers from multi-point discrete arrays, conventional path planning algorithms used in existing technologies are prone to getting trapped in local optima, failing to achieve optimal scheduling of the global processing path. This results in redundant processing strokes and low overall efficiency. At the same time, single laser processing modes struggle to precisely control the energy boundary. Furthermore, existing femtosecond processing systems generally suffer from insufficient power stability, beam directivity susceptible to temperature influences, and delayed online detection feedback. This makes it easy for laser energy to penetrate and damage the metal substrate, causing deviations in key electrical parameters such as device resistance. Moreover, the precision of oxide layer stripping depth control is insufficient, and the surface residual oxidation rate is too high. This makes it impossible to achieve integrated processing of non-destructive oxide layer removal and device performance optimization, and it is difficult to meet the needs of mass production of high-precision metal devices. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of easy substrate damage, inefficient path planning and high residual oxidation rate when precisely removing nanoscale oxide layers for multi-site discrete arrays, and to propose a non-destructive removal method and system for quantum chip metal oxide layers based on femtosecond laser.

[0005] In a first aspect of this invention, a non-destructive removal system for metal oxide layers on quantum chips based on femtosecond lasers is proposed. The system comprises a laser device, a processing platform, a control system, and a detection feedback system; wherein: The laser device includes a femtosecond laser, a continuous laser, a beam shaper, a pulse delay line, and a laser scanning module; the continuous laser is used to defocus and heat the oxide layer on the surface of the metal device to be treated, and to perform thermal activation treatment on the oxide layer; the femtosecond laser is used to perform pulsed laser ablation on the thermally activated oxide layer to remove the oxide layer on the surface of the metal device to be treated. The processing platform includes an optical motion platform, an active focusing module, a wafer vacuum chuck, and a pump. The control system is used to control laser parameters, processing path, and processing speed; The detection feedback system includes a laser confocal microscope; the laser confocal microscope is used for three-dimensional quantitative measurement of the residual oxide layer quality, real-time detection of the processing process, and optimization of processing parameters based on the detection results.

[0006] Optionally, the beam shaper is used to adjust the spatial beam distribution of the femtosecond laser to form a laser beam that matches the area to be processed; the pulse delay line is used to adjust the pulse timing of the femtosecond laser to change the time delay relationship between femtosecond laser pulses.

[0007] Optionally, the output power of the continuous laser is less than that of the femtosecond laser; the output power of the continuous laser is in the range of 0.1W to 10W, and the output power of the femtosecond laser is at least greater than 200W.

[0008] Optionally, the optical motion platform is used to carry and move the metal device to be processed within a preset processing range; the wafer vacuum chuck is disposed on the optical motion platform and connected to a vacuum pump, and is used to fix the metal device to be processed by vacuum adsorption; the active focusing module is disposed above the optical motion platform and located on one side of the laser processing optical path, and is used to detect the surface height of the metal device to be processed and adjust the relative position of the laser processing focus and the surface of the metal device to be processed according to the detection result.

[0009] Optionally, the laser scanning module includes a Galvo galvanometer, a scanning lens, a sleeve lens, and an objective lens; the Galvo galvanometer, scanning lens, sleeve lens, and objective lens are sequentially arranged in the laser processing optical path along the laser propagation direction; the Galvo galvanometer is located on the incident side of the laser beam; the scanning lens, sleeve lens, and objective lens are sequentially located between the Galvo galvanometer and the metal device to be processed, so that the laser beam deflected by the Galvo galvanometer is transmitted through the scanning lens, sleeve lens, and objective lens and focused onto the oxide layer surface of the metal device to be processed.

[0010] In a second aspect of this invention, a method for non-destructive removal of metal oxide layers from quantum chips based on femtosecond lasers is proposed, the method comprising: Obtain the distribution data corresponding to the oxidation area of ​​the metal device to be processed, determine the planning data of the processing path of the processing area based on the distribution data, and adjust the planning data according to the preset path optimization rules to determine the optimal path for processing execution; According to the optimal path control processing platform, the metal device to be processed is moved to the corresponding processing position, and the first laser beam output by the continuous fiber laser acts on the oxide layer at the processing position, so that the temperature of the oxide layer rises and enters the thermal activation state. When the oxide layer is in a thermally activated state, the femtosecond laser outputs a second laser beam to act on the oxide layer, causing the oxide layer to reach the removal state and obtain the target oxide layer. Surface detection data of the target oxide layer is collected, and the first and second laser beams are adjusted according to the surface detection data to complete the non-destructive removal of the metal oxide layer.

[0011] Optionally, the optimal processing path is determined by adjusting the planning data according to preset path optimization rules. The process is as follows: Based on the processing location and internal processing trajectory of each processing area in the planning data, each processing area is divided into corresponding local path units, and a global path is constructed based on the movement relationship between adjacent local path units. Processing is performed according to the global path and the actual execution data of each local path unit is obtained. The actual execution data is compared with the preset execution data of the corresponding local path unit to determine the target local path unit where the actual execution efficiency deviates. Based on the preceding and following path units of the target local path unit in the global path, the path connection unit associated with the target local path unit is determined, and the local path units in the path connection unit are recombined to generate alternative path units. The optimal processing path is obtained by performing a preset operation on the alternative path unit.

[0012] Optionally, performing a preset operation on the alternative path unit to obtain the optimal processing path includes: The alternative path unit is embedded in the corresponding path connection position in the global path, the alternative path unit replaces the original path connection unit, and the local path units located before and after the original path connection unit in the global path and their processing order remain unchanged to obtain the target global path; The alternative path units are screened based on the actual execution efficiency corresponding to the target global path, and the global path corresponding to the alternative path unit that meets the preset processing efficiency condition is determined as the optimal processing path.

[0013] Optionally, a first laser beam output from a continuous fiber laser is applied to the oxide layer at the processing location, including: The focusing position of the first laser beam is adjusted along the optical axis to be 0.5mm to 1mm above the surface of the metal material to be processed, so that the laser is in a defocused state when it acts on the oxide layer area, and the oxide layer area is raised and enters a thermally activated state by preset adjustment parameters; the center wavelength of the first laser beam is 976nm.

[0014] Optionally, the femtosecond laser outputs a second laser beam that acts on the oxide layer, including: Based on the thermal activation state of the oxide layer, the second laser beam is subjected to spatial and temporal shaping processing, and the spot energy distribution and pulse action parameters corresponding to the second laser beam are adjusted. The controlled and adjusted second laser beam acts on the oxide layer region along the optimal path to selectively strip the oxide layer; the center wavelength of the second laser beam is 1035nm; The second laser beam spatial shaping is achieved by the beam shaper, and the temporal shaping is achieved by the pulse delay line.

[0015] The beneficial effects of this invention are: This invention dynamically optimizes the processing path using the moment-to-moment method. By coordinating the connection between local path units and the global path, it avoids the ineffective movement and repetitive back-and-forth caused by local optima in traditional path planning, thereby improving the overall execution efficiency of continuous processing in multiple processing areas. It employs a composite processing method combining defocused continuous laser thermal activation and femtosecond laser selective ablation, fully utilizing the thermal property differences between the oxide layer and the metal substrate to achieve high-precision, non-destructive removal of the oxide layer, effectively reducing the thermal impact on the metal substrate. Simultaneously, through femtosecond laser spatial and temporal domain co-shaping, it improves laser energy utilization and micro-area processing accuracy. Combined with a high-precision motion platform, real-time detection feedback, and automatic adjustment, it achieves closed-loop control of the processing process, meeting the application requirements of mass production and non-destructive processing of high-precision metal devices. Attached Figure Description

[0016] The invention will now be further described with reference to the accompanying drawings.

[0017] Figure 1 A flowchart illustrating a non-destructive removal method for metal oxide layers in a quantum chip based on femtosecond laser, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram showing the result of removing the oxide layer according to an embodiment of the present invention; Figure 3 This is an experimental diagram showing the oxide layer before removal, provided in an embodiment of the present invention. Figure 4 This is an experimental diagram showing the oxide layer after removal, provided in an embodiment of the present invention. Figure 5 This is a resistance fluctuation diagram after oxide layer removal provided in an embodiment of the present invention; Figure 6 A cross-sectional microscopic morphology image of the region where the femtosecond laser oxide layer was removed, provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and B can represent: A alone, A and B simultaneously, and B alone. Furthermore, descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" can explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0019] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] This invention provides a non-destructive removal system for metal oxide layers on quantum chips based on femtosecond lasers, comprising a laser device, a processing platform, a control system, and a detection feedback system; wherein: The laser device includes a femtosecond laser, a continuous laser, a beam shaper, a pulse delay line, and a laser scanning module; The processing platform includes an optical motion platform, an active focusing module, a wafer vacuum chuck, and a pump; The control system is used to control laser parameters, processing path, and processing speed; The detection feedback system includes a laser confocal microscope; the laser confocal microscope is used for three-dimensional quantitative measurement of the residual quality of the oxide layer, real-time detection of the processing process, and optimization of processing parameters based on the detection results.

[0021] Continuous lasers are used to defocus and heat the oxide layer on the surface of the metal device to be treated, thereby thermally activating the oxide layer; femtosecond lasers are used to perform pulsed laser ablation on the thermally activated oxide layer to remove the oxide layer from the surface of the metal device to be treated.

[0022] In one implementation, the laser device further includes a laser beam expander and a laser notch filter. The laser beam expander is used to adjust the beam diameter and divergence angle of the second laser beam to improve the focusing quality of the second laser beam after it enters the laser scanning module. The laser beam expander is disposed between the pulse delay line and the laser scanning module to transmit the second laser beam, after spatial and temporal shaping, to the laser scanning module. The laser notch filter is used to block the laser processing light from entering the detection system to protect the detection device and improve the detection signal quality. It is also disposed in the incident light path of the laser confocal microscope to block the processing laser from entering the laser confocal microscope.

[0023] In one implementation, the carrier platform of the machining platform has an effective travel of 600mm on the X and Y axes and an effective travel of 100mm on the Z axis. The drive motors of the X and Y axes are Parker linear motors, the Z axis is a Panasonic servo motor, the guide rail is IKO-P grade, the feedback specification is a linear grating ruler, and the reading head is a Renishaw incremental grating with a resolution of 0.1µm.

[0024] In one implementation, an intelligent control system is developed to achieve precise control of laser parameters, processing path (based on path optimization), processing speed, etc., to support the processing of complex micro-nano structures and automated production, and to achieve XYZW four-axis collaborative control. The inspection feedback system includes one Keyence VK-X3000 laser confocal microscope with a height resolution of 0.1nm, a width resolution of 0.1nm, and a high repeatability accuracy σ of 20nm under a 50X objective lens; it can monitor the processing in real time and automatically optimize processing parameters based on the inspection results.

[0025] In one embodiment, a beam shaper is used to adjust the spatial beam distribution of the femtosecond laser to form a laser beam that matches the area to be processed; a pulse delay line is used to adjust the pulse timing of the femtosecond laser to change the time delay relationship between femtosecond laser pulses.

[0026] In one implementation, a beam shaper spatially modulates the cross-section of the femtosecond laser beam, altering the shape of the laser spot and its internal energy distribution, thereby redistributing the laser energy according to the regional morphology of the oxide layer to be processed. A pulse delay line alters the propagation distance of the femtosecond laser pulses in different optical paths, creating a controllable time delay in the separated pulses, thus adjusting the time interval between adjacent laser pulses reaching the processing area. The femtosecond laser has a pulse width of less than 240 fs (each laser pulse duration does not exceed 240 femtoseconds), a single pulse energy greater than 40 μJ, and a repetition frequency of 1 kHz to 1 MHz (the femtosecond laser can adjust its pulse emission rate, firing 1000 to 1000000 independent femtosecond pulses per second). By coordinating the spatial energy distribution of the spot with the pulse time interval, the energy input of the femtosecond laser is matched to the local structure and thermally activated state of the oxide layer.

[0027] In one embodiment, the output power of the continuous laser is less than that of the femtosecond laser; the output power of the continuous laser ranges from 0.1W to 10W, and the output power of the femtosecond laser is at least greater than 200W.

[0028] In one implementation, a continuous fiber laser is used as the continuous laser. The continuous laser is used to preheat and thermally activate the oxide layer on the metal surface. It needs to gradually heat the oxide layer with a low and stable energy input to improve the activity of the oxide layer and reduce the difficulty of subsequent stripping. Excessive power cannot be used to cause continuous heating of the metal substrate, resulting in thermal impact. The femtosecond laser needs to concentrate energy input to the oxide layer in a very short time through ultrashort pulses to achieve rapid and precise stripping of the oxide layer. Therefore, it needs to have higher output power and instantaneous energy density. By classifying the power of the two, the continuous laser is responsible for gentle thermal activation, and the femtosecond laser is responsible for high-energy-density precise removal. This reduces the oxide layer stripping threshold while avoiding excessive thermal impact on the metal substrate caused by continuous heating, effectively improving the oxide layer removal efficiency and the integrity of the metal substrate.

[0029] In one embodiment, an optical motion platform is used to carry and move the metal device to be processed within a preset processing range; a wafer vacuum chuck is disposed on the optical motion platform and connected to a vacuum pump, and is used to fix the metal device to be processed by vacuum adsorption; an active focusing module is disposed above the optical motion platform and located on one side of the laser processing optical path, and is used to detect the surface height of the metal device to be processed and adjust the relative position of the laser processing focus and the surface of the metal device to be processed according to the detection result.

[0030] In one implementation, the metal device to be processed is first placed on the surface of a wafer vacuum chuck on an optical motion platform. A vacuum pump extracts air from the inside of the chuck to create negative pressure, stabilizing the metal device in a predetermined position. After processing begins, the active focusing module detects the surface height of the area to be processed on the metal device in real time and feeds the detected surface height data back to the control system. The control system compares the current surface height with the preset processing focal plane and calculates the height deviation between the laser focus and the surface of the metal device. When the height deviation reaches the preset focusing error threshold, the Z-axis of the optical motion platform is controlled to rise and fall, so that the surface of the metal device returns to the position of the laser focus. At the same time, the X-axis and Y-axis of the optical motion platform are controlled according to the preset processing path to move the metal device to different areas to be processed. This maintains the relative position of the laser focus and the surface of the metal device stable during processing, and allows the laser to act on each area to be processed sequentially according to the planned optimal path.

[0031] In one implementation, the optical motion platform has an XY-axis worktable size of 300mm×300mm, an XY-axis positioning accuracy of ±5μm (no matter which point it moves to within the XY range of the worktable, the error between the actual position and the program-set position will not exceed ±5 micrometers), a repeatability of ±0.25μm (the deviation between the positions of each stop when moving back and forth to the same processing point multiple times will not exceed ±0.25 micrometers), and a Z-axis repeatability of 1μm; the active focusing module has an autofocus accuracy of 1 / 3 of the objective lens depth of field, an autofocus response calculation time of <15ms, and a light source module of 10mW@780nm LED (the illumination source for the active focusing module is a 780nm near-infrared LED with a rated output power of 10 milliwatts) and an operating temperature of 5℃~50℃.

[0032] In one embodiment, the laser scanning module includes a Galvo galvanometer, a scanning lens, a sleeve lens, and an objective lens; the Galvo galvanometer, scanning lens, sleeve lens, and objective lens are sequentially arranged in the laser processing optical path along the laser propagation direction; the Galvo galvanometer is located on the incident side of the laser beam; the scanning lens, sleeve lens, and objective lens are sequentially located between the Galvo galvanometer and the metal device to be processed, so that the laser beam deflected by the Galvo galvanometer is transmitted through the scanning lens, sleeve lens, and objective lens and focused onto the oxide layer surface of the metal device to be processed.

[0033] In one implementation, after the femtosecond laser is incident through a Galvo galvanometer, the galvanometer controls the exit direction of the laser beam by changing the deflection angle of the reflecting mirror, so that the laser beam is deflected along the optimal path within the scanning range. The deflected laser beam enters the scanning lens, which converts the beams with different incident angles into light spots at different positions in the processing plane, achieving large-area rapid scanning. Then, the incident angle, beam diameter, and divergence state of the beam are matched by a sleeve lens to make the beam meet the incident conditions of the objective lens. Finally, the objective lens focuses the laser, so that the laser energy is concentrated on the surface of the metal oxide layer.

[0034] In one implementation, the scanning angle range of the Galvo galvanometer is ≤ ±10°, the galvanometer accuracy is ±2μrad (meaning the upper limit of the error of the galvanometer lens deflection angle is ±2 microradians; the slight deflection of the lens directly changes the position of the laser spot, the angle error is extremely small, ensuring that the spot will not deviate slightly and that nanoscale peeling within the micro-area will not go astray), the frequency is 1kHz~55kHz, the scanning speed is set to 100~2000mm / s, and the galvanometer repeatability is ±1μm (when processing the same electrode multiple times and processing multiple batches of wafers, the scanning position is highly overlapping each time, the oxide layer removal depth is consistent, and there will be no local residue or over-etching).

[0035] In one implementation, the focal length of the scanning lens is set to 100mm; the focal length of the sleeve lens is set to 150-250mm; the objective lens has a magnification of 10× to 50×; the objective lens focal length is 4-100mm (short focal length 4mm: high magnification, extremely small focused spot, used for ultra-fine micro-area nanoscale peeling such as Josephson junctions; long focal length 100mm: low magnification, larger scanning range, used for large-area processing of whole electrode arrays), and the focusing objective lens has a numerical aperture (NA) of 0.25-0.8, with a focused spot size of 1-12μm (the diameter of the laser spot acting on the workpiece surface ranges from 1 to 12 micrometers; a small spot (1μm) achieves ultra-fine windowing, with a peeling resolution of up to 0.5nm; a large spot (12μm) covers a larger area in a single scan, improving the efficiency of large-area oxide layer removal); thus enabling the laser spot to move rapidly along the processing path within the processing area while maintaining a small focused size.

[0036] Based on the same inventive concept, embodiments of the present invention also provide a method for non-destructive removal of metal oxide layers from quantum chips using femtosecond lasers. See also Figure 1 The methods include: S1. Obtain the distribution data corresponding to the oxide area of ​​the metal device to be processed, determine the planning data of the processing path of the processing area based on the distribution data, and adjust the planning data according to the preset path optimization rules to determine the optimal path for processing execution. S2, according to the optimal path control processing platform, the metal device to be processed is moved to the corresponding processing position, and the first laser beam output by the continuous fiber laser acts on the oxide layer at the processing position, so that the temperature of the oxide layer rises and enters the thermal activation state. S3, when the oxide layer is in a thermally activated state, the femtosecond laser outputs a second laser beam to act on the oxide layer, so that the oxide layer reaches the removal state and the target oxide layer is obtained; S4. Collect surface detection data of the target oxide layer, and adjust the first laser beam and the second laser beam according to the surface detection data to complete the non-destructive removal of the metal oxide layer.

[0037] In one implementation, the distribution data is data that characterizes the location, boundary, area, and spatial spacing between the oxide regions on the surface of the metal device to be processed, and is used to determine the multiple oxide regions that need to be laser-processed and their spatial locations. Surface inspection data is data obtained by inspecting the target area after oxide layer removal. It includes information on oxide layer residue, surface morphology of the removed area, peeling depth, and whether damage occurs on the metal substrate surface. It is used to determine whether the current processing result meets the oxide layer removal requirements and adjust the thermal activation parameters of the first laser beam, the peeling parameters of the second laser beam, and the subsequent processing path accordingly.

[0038] In one implementation, the planning data determination includes: determining the boundary positions corresponding to each oxidation region based on distribution data, and using the non-oxidized regions between adjacent oxidation regions as inter-region transfer segments; dividing each oxidation region into a preset number of processing regions; determining the scanning direction and scanning range of the laser within the corresponding region based on the boundary positions of each processing region; and connecting the processing end position of the previous processing region with the processing start position of the next processing region according to the spatial adjacency relationship between adjacent processing regions to form an initial processing path composed of multiple processing segments and region transfer segments; merging the initial processing path according to the spatial distribution of oxidation regions, combining processing regions with continuous or adjacent boundaries into the same processing segment, so that the laser can directly enter the adjacent region after completing the processing of one region, reducing the repeated movement of the laser processing head or processing platform in non-processing regions, thereby obtaining the planning data; so that the processing path is no longer executed one by one according to the independent number of oxidation regions, but rather the processing sequence is organized according to the actual spatial adjacency relationship, thereby reducing cross-region back-and-forth movement and invalid paths, and improving the execution efficiency of continuous processing of multiple oxidation regions.

[0039] In one implementation, the oxide layer peeling depth resolution is ≤0.5nm (the thickness of oxide layer removal can be precisely controlled in a single femtosecond laser scan, with a minimum controllable step of 0.5 nanometers), and the residual oxide rate on the surface of the windowed micro-area is <1%; the windowed micro-area refers to the area or content of residual oxides in the designated area formed by laser removal of the oxide layer that is less than 1%.

[0040] In one implementation, the first laser beam and the second laser beam are adjusted based on surface detection data, and the process is as follows: Surface inspection of the target oxide layer yields the location, extent, and surface condition of residual oxide layers in each processed area. This data is compared with preset oxide layer removal requirements. If the residual oxide layer area ratio in any processed area exceeds a preset residual threshold, or the residual thickness exceeds a preset residual thickness threshold, the location of the residual area is determined as the corresponding reprocessing area. The thermal activation parameters of the first laser beam are adjusted based on the residual extent to bring the reprocessing area back to a thermally activated state suitable for peeling. The spot energy distribution, pulse parameters, and scanning path of the second laser beam are adjusted based on the residual distribution in the reprocessing area and the surface condition of the metal substrate, ensuring that the second laser beam only peels off the reprocessing area again. If the height difference or roughness change on the metal substrate surface exceeds a preset substrate damage threshold, the thermal input of the first laser beam or the unit area energy input of the second laser beam is reduced, and areas already removed are avoided. Based on the inspection results, an adjustment method for inspection, positioning, and graded reprocessing is formed to avoid repeated processing of already removed areas, ensuring the integrity of oxide layer removal while reducing the probability of damage to the metal substrate.

[0041] In one embodiment, the optimal processing path is determined by adjusting the planning data according to preset path optimization rules. The process is as follows: Based on the processing location and internal processing trajectory of each processing area in the planning data, each processing area is divided into corresponding local path units, and a global path is constructed based on the movement relationship between adjacent local path units. Processing is performed according to the global path and the actual execution data of each local path unit is obtained. The actual execution data is compared with the preset execution data of the corresponding local path unit to determine the target local path unit where the actual execution efficiency deviates. Based on the preceding and following path units of the target local path unit in the global path, determine the path connection unit associated with the target local path unit, and recombine the local path units in the path connection unit to generate alternative path units. The optimal processing path is obtained by performing preset operations on the alternative path unit.

[0042] In one implementation, constructing a global path based on the movement relationship between adjacent local path units includes: calculating the actual movement direction and distance between adjacent local path units based on the processing start point and processing end point corresponding to each local path unit; using the processing start point of the next local path unit as the next connection target; connecting the processing end point of the previous local path unit with the processing start point of the next local path unit segment by segment; merging directly connectable local path units into continuous processing segments; and generating corresponding regional transfer paths between adjacent local path units that cannot be directly connected, so that each local path unit forms a continuous overall path according to the processing order, thereby obtaining a global path that includes local processing paths and inter-regional transfer paths. Specifically, the judgment is made based on the straight-line distance between the processing endpoint and the processing start point of adjacent local path units, the direction of movement, and the regional relationship between the two path units. When the straight-line distance between the processing endpoint and the processing start point is less than a preset distance threshold, and the straight-line movement path from the processing endpoint to the processing start point does not pass through other areas to be processed or areas that have been processed, the corresponding two local path units are determined to meet the direct connection condition.

[0043] In one implementation, instead of planning each oxidation region separately and then simply splicing them together, the spatial connection relationship between local paths is directly incorporated into the global path construction process. This allows the laser to enter the next processing region along a predetermined transfer path after completing the processing of a local region, reducing repeated round trips and ineffective movements between local paths, while ensuring that each region to be processed is processed in a predetermined order.

[0044] In one implementation, the actual execution data corresponding to each local path unit is divided according to its execution order in the global path. The actual execution time, actual movement distance, and actual processing completion amount of each local path unit are extracted and compared with the preset execution time, preset movement distance, and preset processing completion amount of the corresponding local path unit. First, the correspondence between the actual processing completion amount and the preset processing completion amount is used to determine whether the local path unit has completed the predetermined processing task. Based on the confirmation that the processing task has been completed, the time difference between the actual execution time and the preset execution time, and the distance difference between the actual movement distance and the preset movement distance are calculated. When the time difference or distance difference exceeds the corresponding preset allowable range, the local path unit is identified as the target local path unit. Based on the connection relationship between the target local path unit and the global path, it is determined whether the execution efficiency deviation is caused by the transfer process between adjacent path units, thereby distinguishing the execution deviation caused by the local processing process and the path connection process, and avoiding misjudging the extra movement time caused by path connection as an abnormal processing efficiency of a single local path unit.

[0045] In one implementation, adjacent local path units in the path connection unit are reordered according to the spatial distance between the processing endpoint and the next processing starting point and the direction of movement. Local path units that can be processed directly and continuously are merged into continuous path segments. At the same time, the transfer order between each path segment is re-determined to generate alternative path units that reduce round-trip movement.

[0046] In one implementation, during the laser removal of metal oxide layers, the thickness, distribution density, and local surface morphology of the oxide layers in different areas to be processed vary. This can lead to discrepancies between the laser scanning speed, the distance the processing platform moves, and the transfer time between areas during the actual processing, which may differ from the preset parameters when following the initially planned path. In particular, the connection method between adjacent local path units can affect the frequent start-stop, cross-area movement, and laser idle scanning time of the processing platform. As a result, although the processing parameters of the local path unit itself have not changed, the actual execution efficiency is reduced due to unreasonable connection between the preceding and following paths. Therefore, after dividing each area to be processed into local path units and constructing a global path, the target local path unit whose execution efficiency exceeds the allowable deviation is identified through actual execution data. Then, the associated path connection units are determined by combining its preceding and following path units, and the local path units are recombined to generate alternative path units. This allows path optimization to no longer be an isolated adjustment for a single abnormal path, but rather to reorganize the processing sequence and the transfer relationship between areas based on the overall connection relationship between the preceding and following paths. This reduces unnecessary back-and-forth movements, repeated start-stops, and idle travel, making the optimized path more consistent with the dynamic execution characteristics of the actual equipment. This ensures that each oxide area is processed while improving the overall processing efficiency and path execution stability.

[0047] In one embodiment, performing a preset operation on the alternative path unit to obtain the optimal processing path includes: The alternative path unit is embedded in the corresponding path connection position in the global path, the alternative path unit replaces the original path connection unit, and the local path units in the global path before and after the original path connection unit and their processing order remain unchanged to obtain the target global path. Based on the actual execution efficiency corresponding to the target global path, alternative path units are screened, and the global path corresponding to the alternative path unit that meets the preset processing efficiency condition is determined as the optimal processing path.

[0048] In one implementation, the process of obtaining the target global path is as follows: First, locate the start and end positions of the original path connection unit in the global path. Extract the continuous path segment corresponding to the original path connection unit from the global path and replace the continuous path segment as a whole using the generated alternative path unit. During the replacement, connect the processing start point of the alternative path unit with the processing end point of the previous local path unit of the original path connection unit, and connect the processing end point of the alternative path unit with the processing start point of the next local path unit of the original path connection unit. At the same time, do not change the order of the local path units other than the original path connection unit in the global path and their corresponding processing paths, thereby obtaining the target global path that only performs local optimization on the original path connection unit.

[0049] In one implementation, by replacing only the original path connection unit and keeping the processing order and processing path of the local path units before and after it unchanged, the path optimization is limited to the local area where the actual execution efficiency deviates, avoiding unnecessary changes to the verified effective processing path caused by replanning the entire global path; at the same time, by directly connecting the replacement path unit to the existing paths before and after it, the impact of path adjustment on the overall processing order can be reduced, and the amount of computation and the risk of processing interruption caused by replanning can be reduced.

[0050] In one implementation, the optimal path is determined as follows: based on the actual processing completion time, processing platform movement time, and actual oxide layer removal amount after the target global path is actually executed, the actual processing efficiency corresponding to the target global path is calculated and compared with the preset processing efficiency conditions; when the actual processing efficiency meets the preset processing efficiency requirements and the oxide layer removal of each processing area is completed, the target global path is determined as a candidate path that meets the requirements; when multiple alternative path units have target global paths that meet the processing efficiency conditions, the actual total processing time and total movement distance of each target global path are further compared, and the target global path with the shorter actual total processing time and smaller total movement distance is selected as the optimal path for final processing. The alternative paths are then verified through actual execution results, rather than determining the optimal path solely based on theoretical planning results. This ensures that the finally selected path can adapt to the actual motion characteristics of the processing platform and the laser processing process, improving the consistency between path planning results and actual processing efficiency.

[0051] In one embodiment, a first laser beam output from a continuous fiber laser is applied to the oxide layer at the processing location, including: The focusing position of the first laser beam is adjusted along the optical axis to be 0.5mm to 1mm above the surface of the metal material to be processed, so that the laser is in a defocused state when it acts on the oxide layer area, and the oxide layer area is raised and enters the thermal activation state by preset adjustment parameters; the center wavelength of the first laser beam is 976nm.

[0052] In one implementation, a first laser beam with a center wavelength of 976 nm is output from a continuous fiber laser and transmitted to the surface of the metal material to be processed via a beam expander and focusing optical component. The actual focal point of the first laser beam is positioned 0.5 mm to 1 mm above the surface of the metal material to be processed along the optical axis by adjusting the focusing component, so that the first laser beam acts on the oxide layer region in a defocused state, thereby expanding the laser spot and reducing the energy density per unit area.

[0053] In one implementation, the continuous output power, scanning speed, and action time of the first laser beam are set according to the area and thickness of the oxide layer to be processed. The continuous output power can be set to 0.1W to 10W, and the scanning speed can be set to 10 to 1000mm / s. The surface temperature of the oxide layer area is obtained by real-time temperature detection. When the detected temperature does not reach the preset thermal activation temperature range, the output power of the first laser beam is increased or the scanning speed is decreased. When the detected temperature reaches the preset thermal activation temperature range, the current parameters are maintained or the output power is decreased, so that the oxide layer remains in a thermally activated state without directly melting the metal substrate, thus providing a suitable thermal state for subsequent femtosecond laser oxide layer stripping.

[0054] In one implementation, the laser power is adjustable from 0.1 to 10 W, the temperature of the oxide layer region can be controlled within the range of room temperature to 300°C, the temperature control accuracy is ±1°C, and the temperature change rate ranges from 10 to 100°C. 6 ~10 12 K / s, thereby utilizing the uniform heat input of the defocused continuous laser to preheat and thermally activate the oxide layer, and reducing the energy input required for subsequent femtosecond laser ablation.

[0055] In one embodiment, the femtosecond laser outputs a second laser beam that acts on the oxide layer, including: Based on the thermal activation state of the oxide layer, the second laser beam is shaped in both the spatial and temporal domains, and the spot energy distribution and pulse parameters corresponding to the second laser beam are adjusted. The controlled and adjusted second laser beam acts on the oxide layer region along the optimal path to selectively strip the oxide layer; the center wavelength of the second laser beam is 1035nm. The second laser beam spatial shaping is achieved through a beam shaper, while temporal shaping is achieved through a pulse delay line.

[0056] In one implementation, the energy input method of the femtosecond laser is determined based on the thermal activation state of the oxide layer after the action of the first laser beam. According to the spatial distribution of the oxide layer and the morphology of the area to be processed, the spot shape and energy distribution within the second laser beam are adjusted by a beam shaper, and the time delay between adjacent pulses is changed by a pulse delay line to match the laser energy with the local state of the oxide layer. The shaped second laser beam with a center wavelength of 1035 nm is input into a laser scanning module, and the oxide layer is scanned and stripped along an optimal path using a galvanometer. The pulse width can be less than 240 fs, the single pulse energy can be set to 40–200 μJ, and the repetition frequency is 1 kHz–1 MHz. The pulse energy and pulse interval are adjusted according to the degree of thermal activation of the oxide layer, so that the femtosecond laser mainly acts on the interface region between the oxide layer and the metal substrate, achieving selective removal of the oxide layer while reducing the possibility of thermal impact on the metal substrate.

[0057] In one implementation, the spatial shaping parameters of the second laser beam are determined based on the spatial distribution characteristics of the oxide layer corresponding to the area to be processed, and the temporal shaping parameters of the second laser beam are determined based on the thermal activation state of the oxide layer after the action of the first laser beam; the beam shaper is controlled to adjust the spot energy distribution of the second laser beam according to the spatial shaping parameters, and the pulse delay line is controlled to adjust the pulse action parameters of the second laser beam according to the temporal shaping parameters.

[0058] In one implementation, the pulse operation parameters are: pulse width < 240 fs, output power > 40 W, single pulse energy range of 40–200 μJ, repetition frequency of 1 kHz–1 MHz, and beam quality M. 2 ≤1.2 (M) 2 The factor is an index that measures how close the laser beam is to the ideal diffraction-limited spot. The closer the value is to 1, the more perfect the beam is. The beam directivity is <20μrad / ℃ (meaning that for every 1℃ change in ambient temperature, the deviation angle of the laser beam propagation direction is less than 20 microradians), and the power stability is <1.0%RMS@24h (RMS is the root mean square error, which represents the root mean square value of the laser output power fluctuation within 24 hours of continuous operation is less than 1%).

[0059] In one implementation, the femtosecond laser uses a femtosecond fiber oscillator to generate an initial femtosecond pulse, which is then time-stretched by a chirped pulse stretcher to reduce peak power during amplification and avoid nonlinear effects from the amplifying crystal. The stretched pulse is then sequentially regenerated and amplified by a Yb:CALGO crystal and a traveling-wave amplified by a Yb:YAG crystal to increase the single-pulse energy. Finally, the pulse duration is shortened by a grating multi-pass pulse compressor, resulting in a femtosecond laser with a center wavelength of approximately 1035 nm, a pulse width of less than 240 fs, a single-pulse energy of greater than 200 μJ, and a repetition frequency of 1 kHz to 1 MHz. However, in actual processing, the single-pulse energy of the second laser beam is adjusted according to the thermal activation state and removal degree of the oxide layer, so that the actual single-pulse energy acting on the oxide layer is set to 40–200 μJ.

[0060] In one implementation, see [link to implementation details]. Figure 2 , Figure 2 Figure (a) shows the 61x61μm area to be processed by heating and activating with a continuous laser. Figure 2 Figure (b) shows the effect of removing the oxide layer from the area to be processed by a femtosecond laser. 14.3 μm is the diameter of the second laser beam. Figure 2 Figure (c) shows the effect analysis; the average peeling depth is 0.050 μm, the maximum edge depth is 0.053 μm, and the negative deviation is 6%. Figure 2 As can be seen, the surface detection curve within the processing area fluctuates slightly around the baseline, and the variation at different positions is small. This indicates that the oxide layer removal area after laser treatment has good surface uniformity and consistency. For oxide layer removal, it can reduce the situation of insufficient or excessive removal in local areas, and enable the oxide layer to be removed more uniformly within the target area.

[0061] In one implementation, see [link to implementation details]. Figure 3 and Figure 4 The femtosecond laser effectively removed the oxide layer of the kilobit (1050) tantalum electrode in 31.5s, and the single electrode removal time was 0.015s, meeting the index of single-point removal time ≤2s; the resistance of the tantalum electrode before peeling was 199.6MΩ (A51), and the resistance after peeling was 8.637KΩ (A51), with the resistance change range within -14 to +18Ω, the electrode was undamaged, and the conductivity was significantly improved.

[0062] In one implementation, see [link to implementation details]. Figure 5After the oxide layer was removed, the resistance values ​​at different test locations of the metal device remained around 8.6 kΩ. The resistance fluctuations between test points were small and there were no obvious abrupt changes, indicating that the conductivity of each region was well consistent after the oxide layer was removed by laser. This shows that the degree of oxide layer removal during the processing was relatively uniform, and the metal substrate did not suffer significant changes in conductivity due to excessively high local laser energy. This figure reflects the consistency of electrical properties of each region of the metal substrate after oxide layer removal and the stability of the processing process.

[0063] In one implementation, see [link to implementation details]. Figure 6 As can be seen from the cross-sectional microstructure, the oxide layer removal area after processing by the composite processing method of the present invention has regular sidewalls and uniform depth. The metal substrate has no obvious heat-affected zone or over-etching damage, and the processing boundary is clear and sharp, which verifies the high-precision non-destructive removal capability of the continuous laser thermal activation + femtosecond laser selective ablation scheme.

[0064] The foregoing has described one embodiment of the present invention in detail, but this content is merely a preferred embodiment and should not be considered as limiting the scope of the present invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the scope of the claims of this invention.

Claims

1. A non-destructive removal system for metal oxide layers on quantum chips based on femtosecond lasers, characterized in that, It includes a laser device, a processing platform, a control system, and a detection feedback system; among which: The laser device includes a femtosecond laser, a continuous laser, a beam shaper, a pulse delay line, and a laser scanning module; the continuous laser is used to defocus and heat the oxide layer on the surface of the metal device to be treated, and to perform thermal activation treatment on the oxide layer; the femtosecond laser is used to perform pulsed laser ablation on the thermally activated oxide layer to remove the oxide layer on the surface of the metal device to be treated. The processing platform includes an optical motion platform, an active focusing module, a wafer vacuum chuck, and a pump. The control system is used to control laser parameters, processing path, and processing speed; The detection feedback system includes a laser confocal microscope; the laser confocal microscope is used for three-dimensional quantitative measurement of the residual oxide layer quality, real-time detection of the processing process, and optimization of processing parameters based on the detection results.

2. The femtosecond laser-based non-destructive removal system for metal oxide layers in quantum chips according to claim 1, characterized in that, The beam shaper is used to adjust the spatial beam distribution of the femtosecond laser to form a laser beam that matches the area to be processed; the pulse delay line is used to adjust the pulse timing of the femtosecond laser to change the time delay relationship between femtosecond laser pulses.

3. The femtosecond laser-based non-destructive removal system for metal oxide layers in quantum chips according to claim 1, characterized in that, The output power of the continuous laser is less than that of the femtosecond laser; the output power of the continuous laser is in the range of 0.1W to 10W, and the output power of the femtosecond laser is at least greater than 200W.

4. The femtosecond laser-based non-destructive removal system for metal oxide layers in quantum chips according to claim 1, characterized in that, The optical motion platform is used to carry and move the metal device to be processed within a preset processing range; the wafer vacuum chuck is set on the optical motion platform and connected to a vacuum pump, and is used to fix the metal device to be processed by vacuum adsorption; the active focusing module is set above the optical motion platform and located on one side of the laser processing optical path, and is used to detect the surface height of the metal device to be processed and adjust the relative position of the laser processing focus and the surface of the metal device to be processed according to the detection result.

5. The femtosecond laser-based non-destructive removal system for metal oxide layers in quantum chips according to claim 1, characterized in that, The laser scanning module includes a Galvo galvanometer, a scanning lens, a sleeve lens, and an objective lens. The Galvo galvanometer, scanning lens, sleeve lens, and objective lens are sequentially arranged in the laser processing optical path along the laser propagation direction. The Galvo galvanometer is located on the incident side of the laser beam. The scanning lens, sleeve lens, and objective lens are sequentially located between the Galvo galvanometer and the metal device to be processed, so that the laser beam deflected by the Galvo galvanometer is transmitted through the scanning lens, sleeve lens, and objective lens and focused onto the oxide layer surface of the metal device to be processed.

6. A method for non-destructive removal of metal oxide layers from quantum chips based on femtosecond lasers as described in claims 1-5, characterized in that, The methods include: Obtain the distribution data corresponding to the oxidation area of ​​the metal device to be processed, determine the planning data of the processing path of the processing area based on the distribution data, and adjust the planning data according to the preset path optimization rules to determine the optimal path for processing execution; According to the optimal path control processing platform, the metal device to be processed is moved to the corresponding processing position, and the first laser beam output by the continuous fiber laser acts on the oxide layer at the processing position, so that the temperature of the oxide layer rises and enters the thermal activation state. When the oxide layer is in a thermally activated state, the femtosecond laser outputs a second laser beam to act on the oxide layer, causing the oxide layer to reach the removal state and obtain the target oxide layer; Surface detection data of the target oxide layer is collected, and the first and second laser beams are adjusted according to the surface detection data to complete the non-destructive removal of the metal oxide layer.

7. The non-destructive removal method for metal oxide layers in quantum chips based on femtosecond lasers according to claim 6, wherein the optimal processing path is determined by adjusting the planning data according to a preset path optimization rule, the process being as follows: Based on the processing location and internal processing trajectory of each processing area in the planning data, each processing area is divided into corresponding local path units, and a global path is constructed based on the movement relationship between adjacent local path units. Processing is performed according to the global path and the actual execution data of each local path unit is obtained. The actual execution data is compared with the preset execution data of the corresponding local path unit to determine the target local path unit where the actual execution efficiency deviates. Based on the preceding and following path units of the target local path unit in the global path, the path connection unit associated with the target local path unit is determined, and the local path units in the path connection unit are recombined to generate alternative path units. The optimal processing path is obtained by performing a preset operation on the alternative path unit.

8. The method for non-destructive removal of metal oxide layers from quantum chips based on femtosecond lasers according to claim 7, characterized in that, Performing a preset operation on the alternative path unit to obtain the optimal processing path includes: The alternative path unit is embedded in the corresponding path connection position in the global path, the alternative path unit replaces the original path connection unit, and the local path units located before and after the original path connection unit in the global path and their processing order remain unchanged to obtain the target global path; The alternative path units are screened based on the actual execution efficiency corresponding to the target global path, and the global path corresponding to the alternative path unit that meets the preset processing efficiency condition is determined as the optimal processing path.

9. The method for non-destructive removal of metal oxide layers from quantum chips based on femtosecond lasers according to claim 6, characterized in that, A first laser beam, output from a continuous fiber laser, is applied to the oxide layer at the processing location, including: The focusing position of the first laser beam is adjusted along the optical axis to be 0.5mm to 1mm above the surface of the metal material to be processed, so that the laser is in a defocused state when it acts on the oxide layer area, and the oxide layer area is raised and enters a thermally activated state by preset adjustment parameters; the center wavelength of the first laser beam is 976nm.

10. The method for non-destructive removal of metal oxide layers from quantum chips based on femtosecond lasers according to claim 6, characterized in that, The femtosecond laser outputs a second laser beam that acts on the oxide layer, including: Based on the thermal activation state of the oxide layer, the second laser beam is subjected to spatial and temporal shaping processing, and the spot energy distribution and pulse action parameters corresponding to the second laser beam are adjusted. The controlled and adjusted second laser beam acts on the oxide layer region along the optimal path to selectively strip the oxide layer; the center wavelength of the second laser beam is 1035nm; The second laser beam spatial shaping is achieved by the beam shaper, and the temporal shaping is achieved by the pulse delay line.

Images