A method, apparatus, device and medium for scheduling a probe bias according to a milling depth

Abstract

The present application relates to the technical field of detection bias scheduling, and discloses a method, device, equipment and medium for scheduling detection bias according to milling depth. The method comprises the following steps: obtaining the total milling depth of the milling depth of the current milling sequence in the target interval period of the milling sequence; indexing the pre-calibrated bias scheduling table based on the total milling depth to determine the target detection bias matched with the total milling depth; switching the grid bias voltage collected by the detector from the current value to the target detection bias within the target interval period; after the switching of the target detection bias is completed and the target detection bias is reached, entering the next electron beam cross-section imaging stage, and keeping the target detection bias unchanged during the cross-section imaging stage to complete the electron beam cross-section imaging. The present application can be deployed in delivered equipment, and can adaptively adjust the detection bias according to the milling progress, thereby improving the signal collection efficiency and imaging quality in the continuous milling imaging process.

Description

Technical Field

[0001] This invention belongs to the field of detection bias scheduling technology, specifically relating to a method, apparatus, equipment and medium for scheduling detection bias according to milling depth. Background Technology

[0002] In deep trench / hole milling, FIB-SEM dual-beam systems commonly employ off-axis Everhart-Thornley (ET) type secondary electron detectors to acquire topographic signals of the milled cross-section / pit bottom. As the milling depth increases, the pit walls and pit edges geometrically obstruct the detector's line of sight, causing a non-linear decrease in the detector's signal collection efficiency with depth. Changes in the local pit wall inclination further exacerbate or alleviate this obstruction effect. A detector with a fixed bias cannot adapt to this depth-dependent attenuation of collection efficiency, resulting in a significantly worse signal-to-noise ratio for deep cross-sections compared to shallow ones. Operators often need to manually and repeatedly adjust the detector gain or contrast, leading to low efficiency and poor consistency. This problem is particularly pronounced in continuous unattended 3D tomography (Slice & View, hundreds to thousands of slices): the systematic degradation of image quality in later slices directly affects the accuracy of 3D reconstruction.

[0003] The patent application with publication number KR20250105457A describes in its specification that the collection efficiency is determined by the geometry and decreases with increasing trench depth. Existing mitigation methods require increasing the lateral dimension of the trench at the cost of greater structural damage. The orthogonal dual-column scheme with publication number JP2025504731A reduces occlusion at the hardware level, but requires changes to the overall structure and cannot be used for upgrading existing equipment. In summary, a novel adaptive scheduling method for detection bias is needed to solve the problems existing in the prior art. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method, apparatus, device, and medium for scheduling detection bias voltage according to milling depth.

[0005] According to one aspect of this application, a method for scheduling detection bias voltage according to milling depth is disclosed, the method comprising: During the target interval of the milling sequence, the cumulative value of the milling depth already performed in the current milling sequence is obtained. The target interval is the interval between the completion of the previous adjacent milling action and the start of the next electron beam cross-sectional imaging in the milling sequence. The cumulative value represents the total milling depth after the end of the previous milling action. Based on the total milling depth, a pre-calibrated bias scheduling table is indexed to determine the target detection bias that matches the total milling depth. The bias scheduling table is a pre-calibrated correspondence table between milling depth and detection bias. During the target interval period, the detector collection gate bias voltage is switched from the current value to the target detection bias voltage; After the target detection bias voltage switching is completed and the target detection bias voltage is reached, the next electron beam cross-section imaging stage begins, and the target detection bias voltage is kept constant during this cross-section imaging stage to complete the electron beam cross-section imaging.

[0006] In some embodiments, obtaining the cumulative value of the milling depth already performed in the current milling sequence during the target interval of the milling sequence includes: During the target interval of the milling sequence, the cumulative value of the milling depth performed in the current milling sequence is determined based on the number of slices recorded by the instrument's slice counter, or calculated based on the axial position record of the sample stage or beam spot of the calibration sample, or estimated based on the product of the known milling rate and the consumed milling time.

[0007] In some embodiments, determining the target detection bias that matches the total milling depth by indexing a pre-calibrated bias scheduling table includes: The local pit wall inclination angle of the current milled groove is obtained. The local pit wall inclination angle is obtained based on the edge gradient analysis of the previous round of slice image and is used to characterize the geometric state of the current cross section to be imaged. The bias scheduling table is indexed based on the cumulative value and the local pit wall inclination to determine the target detection bias.

[0008] In some embodiments, switching the detector collection gate bias from its current value to the target detection bias during the target interval period includes: The target detection bias voltage is used as the sole setting value of the detector collection grid in the next electron beam cross-sectional imaging stage; During the target interval period, the programmable detection bias power supply is controlled to adjust the detector collection gate bias voltage to the target detection bias voltage in one go; During the next electron beam cross-section imaging phase, the detector collection grid bias is maintained at the target detection bias.

[0009] In some embodiments, the bias scheduling table is obtained through electron or ion trajectory simulation calibration, including: Trajectory simulations were performed on multiple sets of joint parameters consisting of total milling depth, local pit wall inclination angle, and detection bias, and the collection efficiency corresponding to each set of joint parameters was statistically analyzed. Based on the correspondence between the obtained collection efficiency and the preset collection efficiency target, the target detection bias is determined under each combination of total milling depth and local pit wall inclination angle.

[0010] In some embodiments, the bias scheduling table is obtained through experimental calibration, including: A calibration structure with various combinations of milling depths and pit wall inclination angles was prepared on the calibration sample; For each set of calibration structures, the collection efficiency was measured under multiple probe bias voltages. Based on the relationship between the intensity of the detection signal or the collection efficiency and the detection bias voltage under each calibration structure, the target detection bias voltage that enables the collection efficiency to reach the preset target under each combination of milling depth and pit wall inclination is determined.

[0011] In some embodiments, the method further includes: Obtain the material category of the current calibration sample; The pre-constructed joint bias scheduling table is invoked, which uses the sample material category, total milling depth and local pit wall inclination as joint index dimensions. The joint bias scheduling table is divided into multiple material category branches according to the sample material category. Within each material category branch, a two-dimensional lookup table grid is formed by the total milling depth and local pit wall inclination. Each grid node is associated with the corresponding target detection bias. Based on the current sample material category, select the corresponding material category branch, and obtain the target detection bias by indexing within that material category branch according to the current total milling depth and the current local pit wall inclination.

[0012] According to another aspect of this application, a device for scheduling detection bias voltage according to milling depth is also disclosed, the device comprising: The cumulative value acquisition module is used to acquire the cumulative value of the milling depth performed in the current milling sequence during the target interval period of the milling sequence. The target interval period is the interval period after the completion of the previous adjacent milling action and before the start of the next electron beam cross-section imaging in the milling sequence. The cumulative value represents the total milling depth after the end of the previous milling action. The target detection bias determination module, based on the total milling depth, indexes a pre-calibrated bias scheduling table to determine the target detection bias that matches the total milling depth. The bias scheduling table is a pre-calibrated correspondence table between milling depth and detection bias. The target detection bias switching module is used to switch the detector collection gate bias from the current value to the target detection bias during the target interval period. The electron beam cross-sectional imaging completion module is used to enter the next electron beam cross-sectional imaging stage after the target detection bias voltage switching is completed and the target detection bias voltage is reached, and to keep the target detection bias voltage unchanged during the cross-sectional imaging stage in order to complete the electron beam cross-sectional imaging.

[0013] According to another aspect of this application, an electronic device is also disclosed, the electronic device including a memory and at least one processor, the memory storing instructions; the at least one processor invokes the instructions in the memory to cause the electronic device to perform various steps of a method for scheduling probe bias by milling depth as described in any of the preceding claims.

[0014] According to a computer-readable storage medium, the computer-readable storage medium stores instructions that, when executed by a processor, implement the steps of a method for scheduling probe bias by milling depth as described in any of the preceding claims.

[0015] The present invention includes, but is not limited to, the following beneficial effects: This solution does not rely on modifications to the existing ET detector hardware structure, and can be deployed on delivered equipment using only control layer software and lookup table data. During the intervals between milling steps, the detection bias can be adjusted, which can optimize the deployment on delivered equipment. Moreover, the detection bias can be adaptively adjusted according to the milling progress. Even when the change in milling depth causes changes in the cross-sectional geometry, a relatively stable and clear cross-sectional image can still be obtained, thus improving the signal collection efficiency and imaging quality in the continuous milling imaging process. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0017] Figure 1 is a flowchart of the method for scheduling detection bias voltage according to milling depth according to an embodiment of this application; Figure 2 is a structural block diagram of the device for scheduling detection bias voltage according to milling depth according to an embodiment of this application; Figure 3 is a schematic diagram of the principle of geometric shielding and bias compensation of milling pit; where d is the milling depth, θ is the local pit wall inclination angle, 11-ion column, 12-electron column, 13-sample stage, 14-sample, 15-milling pit, 16-secondary electron detector, 17-collection grid, 26-auxiliary bias electrode, 27-structural support. Figure 4 shows the curve of the collection efficiency η as a function of milling depth d obtained from simulation calibration. Figure 5 shows the curve of collection efficiency η as a function of probe bias voltage V; Figure 6 shows the curve of the field strength extracted from the bottom of the pit as a function of the pit depth. Figure 7 shows the electronic trajectory simulation diagrams of representative working conditions in the initial screening calibration: Figure 7a shows a flat surface with d=0 and zero bias voltage; Figure 7b shows a shallow pit with d=2 and θ=30° and zero bias voltage; Figure 7c shows a shallow pit with bias voltage applied to V*=100V, and the efficiency is fully recovered; Figure 7d shows a deep pit with d=10 and θ=30° and zero bias voltage; Figure 7e shows a deep pit with bias voltage applied to 200V; Figure 7f shows a deep pit with bias voltage applied to 400V. Figure 8 shows the electronic trajectory simulation diagrams for representative working conditions in high statistical calibration: Figure 8a shows a flat surface with zero bias; Figure 8b shows a flat surface with a bias of 100V; Figure 8c shows a shallow pit (d=2, θ=30°) with a bias of 100V, which completely recovers and surpasses the flat reference; Figure 8d shows a deep pit (d=10, θ=30°) with a bias of 100V; Figure 8e shows a deep pit with a bias of 400V. Figure 9 shows the electrostatic field distribution simulation diagram. Figure 9a shows the electrostatic field distribution simulation diagram of a shallow pit with a fixed detection bias voltage of 300V and d=2mm. Figure 9b shows the electrostatic field distribution simulation diagram of a deep pit with a fixed detection bias voltage of 300V and d=18mm. Figure 10 is a schematic diagram of the cloud map. Figure 10a is the equivalent stress cloud map of the electrode support under a conservative load of 5N. Figure 10b is the total displacement cloud map under the same working condition. Figure 11 is a schematic diagram of the structure of the electronic device provided in an embodiment of this application. Detailed Implementation

[0018] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] For ease of understanding, the specific process of the embodiments of the present invention is described below. Specifically, Figure 1 is a flowchart of the method for scheduling detection bias according to milling depth according to the embodiments of this application. Referring to Figure 1, the method includes the following steps: S100. During the target interval of the milling sequence, obtain the cumulative value of the milling depth already performed in the current milling sequence.

[0020] The target interval is the time interval between the completion of the previous milling action and the start of the next electron beam cross-sectional imaging in the milling sequence, and the cumulative value represents the total milling depth after the completion of the previous milling action.

[0021] For example, the cumulative value of the milling depth performed in the current milling sequence can be determined based on the number of slices recorded by the instrument's slice counter during the target interval of the milling sequence; or, the cumulative value of the milling depth performed in the current milling sequence can be calculated based on the axial position record of the sample stage or beam spot of the calibration sample; or, the cumulative value of the milling depth performed in the current milling sequence can be estimated based on the product of the known milling rate and the consumed milling time.

[0022] S102. Based on the total milling depth, index the pre-calibrated bias scheduling table to determine the target detection bias that matches the total milling depth.

[0023] The bias scheduling table is a pre-calibrated correspondence table between milling depth and detection bias, which can be established through offline experimental calibration, simulation calculation, or regression of historical process data. Specifically, at different milling depths, the detector's collection efficiency, signal-to-noise ratio, or image contrast for signals such as secondary electrons or backscattered electrons can be measured to determine the detection bias that meets preset conditions (such as optimal collection efficiency, signal strength within the target range, or highest image clarity), thus forming a depth-bias mapping relationship.

[0024] For example, based on the total milling depth, a pre-calibrated bias scheduling table is indexed to determine the target detection bias that matches the total milling depth. Specifically, this can be: First, the local pit wall inclination angle of the current milled groove is obtained. This local pit wall inclination angle can be obtained based on edge gradient analysis of the previous slice image. For example, it can be calculated by performing edge detection, gradient extraction, or contour fitting on the most recent electron beam cross-sectional imaging image to determine the angle between the cross-sectional edge and a preset reference surface (such as the initial sample surface or the electron beam incident reference surface). The local pit wall inclination angle characterizes the geometric state of the current imaging cross-section, reflecting the changes in the potential field collection environment formed by the residual walls around the cross-section and the electron emission / collection path as milling progresses.

[0025] Then, the bias scheduling table is indexed based on the cumulative value and the local pit wall inclination angle to determine the target detection bias. Accordingly, the bias scheduling table can be constructed as a two-dimensional lookup table grid with the total milling depth and the local pit wall inclination angle as the joint index dimensions, or it can be associated with bias correction amounts at different inclination angles for each depth node. When the total milling depth and / or the local pit wall inclination angle do not accurately match the calibration nodes in the bias scheduling table, linear interpolation, two-dimensional interpolation, or nearest neighbor selection can be performed based on adjacent calibration nodes to obtain the target detection bias that adapts to the current actual cross-sectional state.

[0026] By combining the local geometric features (pit wall inclination) obtained from the previous round of image analysis, the detector bias can be jointly optimized or corrected. This ensures that the detector collection grid bias matches the electron collection conditions of the current cross-section even when the milling groove morphology changes with the processing progress, thereby improving the signal stability and image quality of subsequent cross-section imaging.

[0027] Furthermore, in another feasible approach, the pressure scheduling table is obtained through electron or ion trajectory simulation calibration. Specifically, a simulation model corresponding to the sample to be tested and the detection system can be constructed first. The simulation model can include the sample stage, the geometry of the milled grooves or pits, the local cross-sectional morphology, the detector collection grid, and the geometric arrangement and material properties of the surrounding electrodes. Among them, the geometry of the pits can be parametrically modeled according to the total milling depth and the local pit wall inclination angle, so that different simulation conditions can reflect the changes in the potential field space around the cross-section during the milling sequence.

[0028] Subsequently, electron or ion trajectory simulations were performed on multiple sets of joint parameters consisting of total milling depth, local pit wall inclination angle, and probe bias. The total milling depth can be selected from multiple discrete calibration points or continuous scanning intervals; the local pit wall inclination angle can also be set with multiple typical values ​​according to preset angle steps, for example, reflecting the degree of sidewall tilt generated at different milling stages; the probe bias can traverse or sample multiple candidate bias values ​​within the expected operating voltage range. For each set of joint parameters, an electron or ion beam with a set energy can be injected into the simulation model, and its trajectory under the influence of an electric field near the sample surface and inside the pit can be simulated.

[0029] During the simulation of each joint parameter group, the corresponding collection efficiency is statistically analyzed. Collection efficiency can be defined as the ratio of the number of electron or ion signals successfully received by the detector's collection grid to the total number of signals emitted or generated in the simulation. It can also take into account factors such as the proportion of trajectories reaching the detector's sensitive area, signal weight, secondary electron yield distribution, or detector solid angle coverage. By traversing different detector bias voltages, simulation curves or datasets showing the change in collection efficiency with detector bias voltage under the same total milling depth and the same local pit wall inclination angle can be obtained.

[0030] Subsequently, based on the correspondence between the obtained collection efficiency and the preset collection efficiency target, the target detection bias voltage for each combination of total milling depth and local pit wall inclination angle is determined. The preset collection efficiency target can be an absolute efficiency threshold, an optimal efficiency value, an efficiency range that meets the image signal-to-noise ratio requirements, or an operating point that keeps the collection efficiency in a plateau region and avoids signal distortion caused by excessively high / low bias voltages. For example, the detection bias voltage that maximizes the collection efficiency or reaches the preset target threshold under a certain depth-inclination angle combination can be determined as the target detection bias voltage corresponding to that combination; if multiple candidate bias voltages meet the requirements, the value with better stability, closer to historical process bias voltages, or easier to achieve power supply adjustment granularity can be selected.

[0031] Furthermore, the above calibration results can be organized into a bias scheduling table indexed by parameters such as total milling depth, local pit wall inclination angle, and sample material type. For intermediate depths or inclination angles that were not precisely calibrated in the simulation, linear interpolation, two-dimensional interpolation, or fitting extrapolation can be performed based on the collection efficiency of adjacent calibration nodes or target detection bias to complete the scheduling table entries.

[0032] For example, in this example, a simplified two-dimensional planar symmetric sample-pit-detector electrostatic model can be established (this embodiment uses SIMION software, 101×81 grid): the sample is a grounded conductor, with the top surface located at y=25 (grid units); when the milling depth d>0, a pit is cut out on the top surface, with the pit bottom located at y=25−d, and the pit walls offset by an inclination angle θ; when d=0, it degenerates into a flat baseline. The detector electrodes are arranged off-axis (above the pit opening) to simulate the oblique angle of a real ET detector; the electrodes are built as a box with a certain height along the height direction rather than a thin surface to avoid missing fast electrons accelerated by a strong field from passing over the top of the electrodes.

[0033] The calibration was conducted in two rounds: the first round used 54 particles per point (6 emission points × 9 uniformly oriented sectors) for initial trend screening; the second round upgraded to 12 emission points × 50 cosine (Lambert) distributed random directions × 3 sets of independent random sampling, with a total of N=1800 particles per point, emission kinetic energy of 3 eV, and statistical standard error. The 95% confidence interval half-width is approximately ±2.3 percentage points. All quantitative values ​​in this manual are based on the second round of testing. Scan range: d∈{0,2,5,10,20} (grid units), θ∈{0°,15°,30°,45°,60°}, V∈{0,100,200,300,400}V, totaling 105 calibration points. The calibration results for the flat baseline (d=0) are shown in Table 1, and the calibration results for the representative deep pit (d=10, θ=30°) are shown in Table 2.

[0034] Table 1. Flat baseline calibration results

[0035] Table 2. Results of representative deep pit calibration

[0036] The main calibration results are as follows: (1) Under zero bias, η decreases monotonically with depth and the decrease is much greater than the confidence interval (as shown in Figure 4). Under the same depth, η increases with the increase of the pit wall inclination angle (the inclined wall opens the exit line of sight); (2) The flat baseline is a statistical plateau (saturation) at V=100–400V. The slight non-monotonic fluctuations observed by the initial screening wheel are confirmed as sampling noise (the electronic trajectories of representative working conditions are shown in Figures 7 and 8); (3) In all 21 (d,θ) combinations, any bias setting is a significant improvement in the separation of the confidence interval compared to zero bias; (4) The angles of the shallow pit d=2 recover at V*=100V and η=0.72–0.76 significantly exceed the peak value of the flat baseline of 0.525—the geometry of the shallow pit is better than that of the flat surface under bias extraction; (5) The optimal setting for the deep pit d=10 / θ=30° is V=100 (η=0.287) and V=400. Instead, it dropped to 0.264, indicating a slight over-biasing effect in the deep pit. Results (4) and (5) together demonstrate that the relationship between η(d,θ,V) is non-monotonic and non-trivial, and the simple formula that the greater the depth, the higher the bias is not optimal—this is precisely the basis for the necessity of using a pre-calibrated lookup table (rather than an analytical formula or ramp rule).

[0037] The calibration results are presented graphically in Figures 4 and 5. Figure 4 shows the curves of the collection efficiency η as a function of milling depth d under zero bias (V=0) (grouped by pit wall inclination angle θ, with N=180° for each point and error bars representing the 95% confidence interval): the curves generally decrease with depth (the curve at θ=60° shows a slight rise within the confidence interval at d=2), for example, the curve at θ=0° decreases from 0.266 on a flat surface to 0.119 at d=10 and 0.104 at d=20, with the decrease far exceeding the confidence interval; at the same depth, the curves are arranged from bottom to top according to the inclination angle from smallest to largest, visually showing the angle effect of the inclined wall opening the exit line of sight. Figure 5 shows the curves of collection efficiency η as a function of detection bias voltage V (θ=30° represents the working condition, grouped by milling depth, the horizontal dashed line is the flat reference peak value of 0.525): the d=2 curve crosses the reference line at V=100V and remains near 0.75; the d=5 curve crosses the reference line near V=300V; the d=10 and d=20 curves are below the reference line in the entire range of 0–400V, among which the d=10 curve reaches its maximum value at V=100V and then slowly falls back as the bias voltage increases, intuitively showing the upper limit of deep pit compensation and the slight over-bias effect.

[0038] Figure 7 shows the electron trajectory simulation diagrams for representative working conditions in the first round of initial screening calibration (54 particles per point), used to qualitatively demonstrate the compensation mechanism. The arrival / loss counts for each diagram are consistent with the original data from the first round: Figure 7a shows electrons exiting at an open solid angle on a flat surface (d=0, V=0), with some reaching the detector directly along the line of sight (η=0.222); Figure 7b shows a large number of trajectories being truncated by the pit wall at zero bias in a shallow pit (d=2, θ=30°) (η=0.185); Figure 7c shows the same shallow pit with a bias voltage of V*=100V, the trajectories collectively bending towards the detector side under the action of the extraction field, the collection efficiency fully recovering and exceeding the flat baseline (η=0.722); Figure 7d shows the vast majority of trajectories terminating at the pit wall at zero bias in a deep pit (d=10, θ=30°) (η=0.148); Figure 7e shows some trajectories being pulled out of the pit after a bias voltage of 200V is applied to the deep pit (η=0.315); Figure 7f shows… Even with a bias voltage applied to the deep pit up to the test limit of 400V, the recovery is still limited (η=0.259), which visually demonstrates the geometric upper limit of deep pit compensation.

[0039] Figure 8 shows the electronic trajectory simulation diagrams of the representative working conditions of the second round of high statistical calibration (600 trajectories with a single random seed, whose arrival / loss counts are checked one by one with the original data of the second round): Figure 8a: flat surface with zero bias (reaching 162 / 600, η=0.270); Figure 8b: flat surface with bias of 100V (reaching 309, η=0.515, entering the saturation plateau); Figure 8c: shallow pit (d=2, θ=30°) with bias of 100V (reaching 450, η=0.750, fully recovered and surpassing the flat baseline); Figure 8d: deep pit (d=10, θ=30°) with bias of 100V, which is the optimal level of this combination (reaching 178, η=0.297, partially recovered); Figure 8e: deep pit with bias of 400V (reaching 156, η=0.260), lower than the 100V level, confirming the slight over-bias effect of the deep pit.

[0040] Bias schedule (all 21 (d,θ) combinations): peak values ​​within the test range with a flat baseline. (Appearing at V=400) as the baseline, V* is the lowest rated bias voltage that makes η reach or exceed the baseline; if it is not reached, the voltage that maximizes η is taken according to the rules of claim 7. It should be noted that the flat baseline itself is a statistical plateau in V=100–400V (the confidence intervals of η at the four levels overlap significantly). In Table 3, the 400V in the d=0 row is only a mechanical result of reaching the peak criterion. In engineering, the flat section can be adjusted to any level within the plateau (such as 100V).

[0041] Table 3. Peak baseline judgment table within the test range for flat baselines.

[0042] Table 3 above shows an example of the scheduling table fixed in the lookup table storage unit 23: shallow pits (d=2) are scheduled to 100V at each angle; medium depths (d=5) with θ≥30° are scheduled to 100–300V to reach the benchmark, while θ=0° / 15° do not reach the benchmark and are assigned the optimal setting; deep pits (d=10, d=20) do not reach the benchmark at any angle, and according to claim 7, their respective optimal settings are assigned—the optimal setting distribution for the 10 deep pit combinations is 100V×2 (d=10 with θ=30° / 45°), 200V×1, 300V×1, and 400V×6. It can be seen that higher bias voltage is not always better, and the optimal setting must be calibrated for each combination. Based on this, the present invention does not advocate complete recovery for deep pits, but only advocates significant compensation within the verified range (compared to zero bias voltage, d=10 / θ=30° is improved by 76%, and η increases from 0.163 to 0.287).

[0043] The above simulation calibration method can obtain the optimal detector collection grid bias configuration under different cross-sectional geometric conditions before physical machining, reduce the time of repeated experimental calibration and sample loss, and provide a data basis for quickly indexing the target detection bias according to the current total milling depth and local pit wall inclination within the target interval period.

[0044] Furthermore, to eliminate the systematic bias of a single simulation tool, an independent second solver can be used to cross-validate the root causes of the electric field attenuation of the collection efficiency with depth. In this embodiment, a two-dimensional simplified electrostatic model is established using COMSOL software: a straight-walled rectangular groove is opened on the grounded conductive sample, the position of the detection electrode and the detection bias voltage are fixed (300V), and the extraction field strength at the bottom of the pit is solved for five working conditions with pit depths d=2 / 6 / 10 / 14 / 18mm (model scale). The results are shown in Figure 6: the electric field strength at the bottom of the pit in the d≥6mm range decreases monotonically with depth (1586→1539→1410→1236 V / m), and the electric field strength at the bottom of the pit in the d=18mm range is about 83.4% of that in the d=2mm range (1482 V / m). This means that the electric field available for extracting secondary electrons at the bottom of the deep pit is significantly weaker than that in the shallow pit, which is consistent with the trend of collection efficiency decreasing with depth in the trajectory simulation, constituting cross-validation across solvers. The deviation of the five working conditions was less than 0.4% when the mesh was refined, and the results have converged to the mesh. Figure 9 shows the electrostatic field distribution simulation diagram. Figures 9a and 9b show the electrostatic field distribution of the shallow pit (d=2mm) and deep pit (d=18mm) working conditions, respectively. It can be seen that the penetration of the extraction field into the bottom of the deep pit is significantly weakened. It is also stated that the field strength at the bottom of the pit at d=2mm (1482 V / m) is slightly lower than the local maximum at d=6mm (1586 V / m) by about 7%. The refined mesh verification proves that this feature is not discrete noise, but the true behavior of the simplified geometry in the extremely shallow pit region. This echoes the previous observation that the collection efficiency after applying bias voltage to the shallow pit exceeds that of the flat benchmark, further illustrating that the electric field and collection behavior are non-monotonic in the shallow pit region, supporting the use of lookup tables instead of monotonic formulas for scheduling.

[0045] Regarding model scale: The above trajectory simulations use grid units, and the electrostatic field simulations use millimeters (model length units) to express depth. These are relative scales of the model, not physical calibration values ​​for a specific instrument. The potential distribution shape remains unchanged under geometric similarity scaling in electrostatic field problems, and electron trajectories maintain geometric similarity under the same scaling. Therefore, trend conclusions obtained at relative scales can be transferred to real milling pits at the micrometer scale. However, the absolute value of collection efficiency depends on the solid angle, working distance, and tilt geometry of the specific detector. Before actual deployment, the bias scheduling table should be recalibrated for the specific instrument according to the aforementioned trajectory simulation calibration or experimental calibration methods.

[0046] Furthermore, in another feasible approach, the bias scheduling table is obtained through experimental calibration. First, calibration structures with various combinations of milling depths and pit wall inclinations can be prepared on the calibration sample. The calibration sample can be made of the same or similar material system as the target sample to be processed, in order to reproduce the secondary electron / backscattered electron yield and surface conductivity characteristics in the actual process as much as possible. The calibration structure can be formed by controlling parameters such as the number of milling operations, single-step milling thickness, incident angle, scanning pattern, and sample stage tilt angle of the focused ion beam or electron beam, so that different regions or different calibration structures have different total milling depths and local pit wall inclinations. For example, a series of stepped, sloping, or grooved structures can be sequentially processed along different positions on the sample surface, with each structure corresponding to different milling depths and sidewall inclinations; alternatively, the same area can be progressively milled, and the corresponding depth and inclination information can be recorded at different milling stages.

[0047] Subsequently, for each calibration structure, the collection efficiency or detection signal strength was measured under multiple detection bias voltages. Specifically, the detector collection grid bias voltage can be adjusted according to a preset voltage step size or a preset candidate bias voltage set, and electron beam cross-sectional imaging or fixed-point scanning detection is performed at each bias voltage setting to obtain the corresponding detection signal response. The collection efficiency can be characterized by the count of the measured signal, the average gray value of the image, the signal integral intensity in the region of interest, the signal-to-noise ratio, or the signal ratio with that of the unobstructed planar reference structure. To avoid bias voltage drift or charge accumulation effects caused by the electron beam scanning itself, a short stabilization wait can be performed before and after each bias voltage measurement, or multiple acquisitions and averaging can be used to improve the reliability of the measured data.

[0048] Subsequently, based on the relationship between the detection signal strength or collection efficiency and the detection bias voltage under each calibration structure, the target detection bias voltage for each combination of milling depth and pit wall inclination angle is determined to achieve the preset target collection efficiency. The preset target can be a preset absolute collection efficiency threshold, a preset detection signal strength threshold, the bias voltage point corresponding to the maximum collection efficiency, the bias voltage value in the stable plateau region of the collection efficiency curve, or the operating point that makes the image contrast / signal-to-noise ratio meet the process requirements. For example, for a specific combination of milling depth and pit wall inclination angle, if the collection efficiency first increases and then tends to saturate or decreases as the detection bias voltage increases, then the bias voltage near the peak of the curve or corresponding to the target efficiency threshold can be determined as the target detection bias voltage for this combination; if multiple bias voltages meet the preset target, then a value with better adjustment stability, closer to the commonly used operating range of the system, or easier to achieve with the power supply resolution can be selected.

[0049] Furthermore, the relationships between the total milling depth, pit wall inclination angle, and target detection bias obtained from the above experimental calibration can be compiled into a bias scheduling table. For intermediate depths or intermediate inclination angles that were not directly prepared in the experiment, interpolation, fitting, or extrapolation can be performed based on the measured data of adjacent calibration structures to complete the scheduling table entries for quick indexing during subsequent online scheduling.

[0050] The above-mentioned experimental calibration method enables the bias scheduling table to directly reflect the signal collection characteristics under the actual equipment conditions, improves the matching accuracy between the determined target detection bias and the actual cross-sectional imaging conditions, and thus improves the image brightness consistency, signal-to-noise ratio and cross-sectional detail discernibility in the subsequent continuous milling-imaging process.

[0051] Furthermore, in other feasible solutions, the target detection bias can be obtained in the following manner: First, the material category of the current calibration sample is obtained. The material category can be obtained in various ways, such as reading user-inputted sample information, parsing from process formulations or experimental task configuration files, automatically associating with database records based on the sample stage's identification code, or determining it through in-situ / offline detection methods such as energy dispersive spectroscopy (EDS) analysis and image contrast recognition. The material category can be used to distinguish different substrates or functional layer systems, such as semiconductor materials, metallic materials, dielectric materials, organic materials, composite materials, or material subcategories with different doping / conductivity properties. Because different materials differ in secondary electron yield, backscattered electron energy distribution, surface potential field establishment characteristics, and charge accumulation behavior under electron or ion beam irradiation, even with the same total milling depth and local pit wall inclination, the optimal bias setting for the detector collection grid may differ.

[0052] Subsequently, a pre-constructed joint bias scheduling table is invoked. This table uses sample material type, total milling depth, and local pit wall inclination angle as joint index dimensions. The joint bias scheduling table can be divided into multiple material category branches based on sample material type. Within each material category branch, a two-dimensional lookup table grid is constructed using the total milling depth and local pit wall inclination angle, with each grid node associated with the corresponding target detection bias. In other words, the scheduling table can adopt a hierarchical structure: the top layer is the material category index, used to quickly locate the bias data subset corresponding to the current sample; under each material category branch, a discrete or interpolated two-dimensional mapping relationship is further established along the dimensions of total milling depth and local pit wall inclination angle. The joint bias scheduling table can be constructed through the aforementioned trajectory simulation calibration, experimental calibration, or a combination of both, and stored in local memory, a process server, or the readable data area of ​​a programmable bias power supply.

[0053] Next, based on the current sample material category, the corresponding material category branch is selected, and the target detection bias is obtained by indexing within that material category branch according to the current total milling depth and the current local pit wall inclination angle. Specifically, the system can use the total milling depth (cumulative value) obtained in step S100 and the local pit wall inclination angle obtained based on the edge gradient analysis of the previous round of slice images as input, and match them in the two-dimensional lookup table grid of the selected material branch. If the current depth and inclination angle exactly correspond to a certain calibration grid node, the target detection bias associated with that node is directly output; if the node is not accurately hit, linear interpolation, two-dimensional interpolation, spline fitting, or nearest neighbor selection can be performed based on the bias values ​​of adjacent nodes to obtain a target detection bias that is adapted to the current actual state.

[0054] For example, Monte Carlo simulation of ion-solid interactions (SRIM software in this embodiment) was used to calculate the sputtering yield (5000 incident ions per material) of four common sample materials under Ga+ 30 keV normal incidence: Cu 10.26, W 6.07, SiO2 4.53, Si 2.20 atoms / ion, with the maximum and minimum differing by approximately 4.7 times; the average projected range also differed by approximately The result quantitatively supports the design of a scheduling table that indexes materials by material type, given that sputtering yield and induced secondary electron yield are usually positively correlated. This is because the detectable signal intensity of different materials under the same beam current and bias conditions can vary by several times. Sputtering yield is a proxy indicator of secondary electron yield rather than a direct measurement, and only one incident angle (normal incidence) was calculated.

[0055] In this way, bias scheduling not only takes into account the geometric changes (total milling depth, pit wall inclination) brought about by the milling process, but also introduces material properties, a key factor affecting signal generation and collection. This allows for a bias setting that is more in line with actual physical conditions to be provided for the detector collection grid in continuous milling-imaging cycles of different material systems, thereby further improving the signal strength stability, signal-to-noise ratio and detail fidelity of cross-sectional images.

[0056] S104. During the target interval period, switch the detector collection grid bias voltage from the current value to the target detection bias voltage.

[0057] S106. After the target detection bias voltage switching is completed and the target detection bias voltage is reached, the next electron beam cross-section imaging stage is entered, and the target detection bias voltage is kept constant during the cross-section imaging stage to complete the electron beam cross-section imaging.

[0058] For example, this step may include the following steps: First, the target detection bias voltage is set as the sole value for the detector collection grid during the next electron beam cross-sectional imaging stage. Then, during the target interval, the programmable detection bias voltage power supply is controlled to adjust the detector collection grid bias voltage to the target detection bias voltage in one go; and during the next electron beam cross-sectional imaging stage, the detector collection grid bias voltage is maintained at the target detection bias voltage.

[0059] Specifically, after the bias switching is completed, no additional bias optimization, scan compensation, or dynamic adjustment is performed in the next electron beam cross-sectional imaging stage. Instead, the target detection bias is used as a fixed operating point to ensure that the entire frame of cross-sectional images is obtained under uniform electric field collection conditions. For example, within the target interval, the programmable detection bias power supply can be controlled to adjust the detector collection grid bias to the target detection bias in one go. This one-time adjustment refers to directly outputting the target voltage value through the programmable power supply, rather than gradually approximating or changing it in segments during imaging; this helps shorten the settling time required for bias stabilization and allows the electric field around the collection grid to reach a steady state as quickly as possible. The programmable detection bias power supply can be a DC regulated power supply with fast response capability, a digital-to-analog converter (DAC) driver power supply, or other remotely settable bias supply module, which can generate corresponding analog or digital control signals based on the pre-determined target detection bias value.

[0060] By means of the above method, the pre-adjustment and stabilization of the bias voltage can be completed in the gap between non-machining and non-imaging, and the subsequent cross-sectional imaging can be performed by detecting the bias voltage with a fixed target, thereby improving the stability and consistency of image quality in the continuous milling-imaging cycle.

[0061] Furthermore, Figure 2 is a structural block diagram of a device for scheduling detection bias voltage according to milling depth according to an embodiment of the application. As shown in Figure 2, the device includes: The cumulative value acquisition module is used to acquire the cumulative value of the milling depth performed in the current milling sequence during the target interval period of the milling sequence. The target interval period is the interval period after the completion of the previous adjacent milling action and before the start of the next electron beam cross-section imaging in the milling sequence. The cumulative value represents the total milling depth after the end of the previous milling action. The target detection bias determination module, based on the total milling depth, indexes a pre-calibrated bias scheduling table to determine the target detection bias that matches the total milling depth. The bias scheduling table is a pre-calibrated correspondence table between milling depth and detection bias. The target detection bias switching module is used to switch the detector collection grid bias from the current value to the target detection bias during the target interval period. The electron beam cross-sectional imaging completion module is used to enter the next electron beam cross-sectional imaging stage after the target detection bias voltage switching is completed and the target detection bias voltage is reached, and to keep the target detection bias voltage unchanged during the cross-sectional imaging stage in order to complete the electron beam cross-sectional imaging.

[0062] The application of the relevant modules of the system in this example can be found in the above introduction to the principles of the method, and will not be repeated here.

[0063] To facilitate understanding of the physical basis of this scheme, the principle of geometric shielding and bias compensation of the milled pit is explained below with reference to Figure 3. In Figure 3, a milled pit 15 is formed on sample 14 by ion beam milling, where d is the milling depth and θ is the local pit wall inclination angle; the secondary electron detector 16 and its collection grid 17 adopt the off-axis arrangement of the ET-type detector and are located above the pit opening. It should be noted that the detection bias, detector collection grid bias, and collection bias mentioned in this specification all refer to the same collection voltage parameter applied to the collection grid 17 or the optional auxiliary bias electrode 26; the bias scheduling table, scheduling table, and lookup table refer to the same data object; and the milled pit and milled groove refer to the same recessed structure formed by milling.

[0064] The root cause of geometric obstruction: The kinetic energy of secondary electrons detected by electron beam cross-section imaging is only on the order of several electron volts. Whether they can reach the detector depends heavily on the detector's geometric line of sight relative to the emission point and the collecting electric field. In the early stages of milling, the pit is relatively shallow, and the secondary electrons emitted from the pit bottom / section have a large open solid angle, which can be directly collected by the detector along the line of sight. As the milling depth d increases, the pit wall and the edge of the pit opening gradually cut off the detector's geometric line of sight, and the emission solid angle narrows with depth, resulting in a nonlinear decrease in collection efficiency. Increasing the local pit wall inclination angle θ (making the sidewalls more gentle) partially opens the emission line of sight and alleviates obstruction, while decreasing θ exacerbates obstruction. This is the geometric root cause of the collection efficiency decaying with depth as described in the background art, which is consistent with the results of the aforementioned simulation calibration where η decreases monotonically with depth and increases with increasing inclination angle under zero bias (see Figure 4).

[0065] Mechanism and boundary of bias compensation: The detection bias applied to the collection grid establishes an extraction electric field above the pit opening. This electric field penetrates into the space inside the pit and guides low-energy secondary electrons near the pit opening. The secondary electrons, which originally flew in a straight line and would hit the pit wall or pass by the detector, have their trajectories bent towards the detector side under the action of the extraction field. They are pulled out of the pit and guided to the collection grid, thus compensating for the loss caused by geometric obstruction at the physical source of signal collection. This trajectory bending effect can be visually observed in the electron trajectory simulation diagrams in Figures 7 and 8. The compensation capability has a physical upper limit with depth: the penetration of the extraction electric field into the pit weakens with increasing pit depth (independent electrostatic field simulations show that the field strength at the bottom of a deep pit is about 83.4% of that of a shallow pit under a fixed bias voltage of 300V, see Figures 6 and 9), and the exit solid angle of the deep pit itself is smaller, so the deep pit can only obtain partial compensation; the shallow pit can actually exceed the collection efficiency of a flat surface under the action of the extraction field (the pit geometry and the extraction field together form favorable convergence guidance conditions); when the bias voltage is too high, a slight over-bias effect will also occur (some trajectories are excessively bent and deviate from the detector). The above three aspects together determine that the relationship between the collection efficiency η and the depth d, tilt angle θ and bias voltage V is non-monotonic and non-trivial. A simple fixed formula that the greater the depth, the higher the bias voltage is is not optimal. Therefore, this scheme uses a pre-calibrated bias voltage scheduling table for lookup scheduling and limits the bias voltage switching to the gap between milling steps.

[0066] Furthermore, this solution also provides a schematic diagram of a detection bias adaptive scheduling device. This device replaces / modifies the existing ET detector's collection bias power supply architecture with a programmable power supply; the control unit interfaces with the instrument's existing milling progress recording interface; and the scheduling table data is stored in the storage unit 23. If an auxiliary bias electrode 26 (Figure 3) needs to be added, the matching structural support 27 is designed as a cantilever beam type 316L stainless steel support (15mm×6mm×1.5mm), with one end fixed to the existing detector housing, bearing gram-level electrode and lead assembly.

[0067] The support was statically verified using finite element analysis (ANSYS software was used in this embodiment, with a hexahedral mesh of 0.5 mm; one end was fully fixed to simulate screw clamping, and a conservative concentrated load of 5 N was applied to the free end; stress / displacement contour plots are shown in Figures 10a and 10b): the maximum total displacement was approximately 16.9 µm; the maximum equivalent stress nodal solution was approximately 29.6 MPa, the upper bound of the estimated value including discrete error was approximately 37.1 MPa, and the safety factor relative to the yield strength of 316L was approximately 205 MPa. to The above values ​​should be understood as conservative upper limits: the stress peak is located in the stress concentration region of the ideal fixed support boundary, and the verification is based on a single static condition. Simulation verification shows that, under the conservative assumptions, the support does not yield, and the safety factor is approximately 5–7 times.

[0068] Figure 2 above describes in detail a device for scheduling detection bias voltage according to milling depth in an embodiment of the present invention from the perspective of modular functional entities. The following describes in detail the electronic device in an embodiment of the present invention from the perspective of hardware processing.

[0069] Figure 11 is a schematic diagram of an electronic device according to an embodiment of the present invention. The electronic device 1100 can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 1110 (e.g., one or more processors) and a memory 1120, and one or more storage media 1130 (e.g., one or more mass storage devices) for storing application programs 1133 or data 1132. The memory 1120 and storage media 1130 may be temporary or persistent storage. The program stored in the storage media 1130 may include one or more modules (not shown in the figure), each module including a series of instruction operations on the electronic device 1100. Furthermore, the processor 1110 may be configured to communicate with the storage media 1130 and execute the series of instruction operations in the storage media 1130 on the electronic device 1100.

[0070] Electronic device 1100 may also include one or more power supplies 1140, one or more wired or wireless network interfaces 1150, one or more input / output interfaces 1160, and / or one or more operating systems 1131, such as Windows Server, MacOSX, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that the electronic device structure shown in FIG1 does not constitute a limitation on electronic devices and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0071] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of any of the above-described methods for scheduling detection bias according to milling depth.

[0072] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0073] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0074] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for scheduling detection bias voltage according to milling depth, characterized in that, The method includes: During the target interval of the milling sequence, the cumulative value of the milling depth already performed in the current milling sequence is obtained. The target interval is the interval between the completion of the previous adjacent milling action and the start of the next electron beam cross-sectional imaging in the milling sequence. The cumulative value represents the total milling depth after the end of the previous milling action. Based on the total milling depth, a pre-calibrated bias scheduling table is indexed to determine the target detection bias that matches the total milling depth. The bias scheduling table is a pre-calibrated correspondence table between milling depth and detection bias. During the target interval period, the detector collection gate bias voltage is switched from the current value to the target detection bias voltage; After the target detection bias voltage switching is completed and the target detection bias voltage is reached, the next electron beam cross-section imaging stage begins, and the target detection bias voltage is kept constant during this cross-section imaging stage to complete the electron beam cross-section imaging.

2. The method for scheduling detection bias voltage according to milling depth as described in claim 1, characterized in that, The step of obtaining the cumulative value of the milling depth already performed in the current milling sequence during the target interval of the milling sequence includes: During the target interval of the milling sequence, the cumulative value of the milling depth performed in the current milling sequence is determined based on the number of slices recorded by the instrument's slice counter, or calculated based on the axial position record of the sample stage or beam spot of the calibration sample, or estimated based on the product of the known milling rate and the consumed milling time.

3. The method for scheduling detection bias voltage according to milling depth according to claim 1, characterized in that, The step of determining the target detection bias that matches the total milling depth by indexing a pre-calibrated bias scheduling table includes: The local pit wall inclination angle of the current milled groove is obtained. The local pit wall inclination angle is obtained based on the edge gradient analysis of the previous round of slice image and is used to characterize the geometric state of the current cross section to be imaged. The bias scheduling table is indexed based on the cumulative value and the local pit wall inclination to determine the target detection bias.

4. The method for scheduling detection bias voltage according to milling depth as described in claim 1, characterized in that, The step of switching the detector collection gate bias voltage from its current value to the target detection bias voltage during the target interval period includes: The target detection bias voltage is used as the sole setting value of the detector collection grid in the next electron beam cross-sectional imaging stage; During the target interval period, the programmable probe bias power supply is controlled to adjust the detector collection gate bias voltage to the target probe bias voltage in one go.

5. The method for scheduling detection bias voltage according to milling depth according to claim 3, characterized in that, The bias scheduling table is obtained through electron or ion trajectory simulation calibration, including: Trajectory simulations were performed on multiple sets of joint parameters consisting of total milling depth, local pit wall inclination angle, and detection bias, and the collection efficiency corresponding to each set of joint parameters was statistically analyzed. Based on the correspondence between the obtained collection efficiency and the preset collection efficiency target, the target detection bias is determined under each combination of total milling depth and local pit wall inclination angle.

6. The method for scheduling detection bias voltage according to milling depth according to claim 3, characterized in that, The bias scheduling table was obtained through experimental calibration and includes: A calibration structure with various combinations of milling depths and pit wall inclination angles was prepared on the calibration sample; For each set of calibration structures, the collection efficiency was measured under multiple probe bias voltages. Based on the relationship between the intensity of the detection signal or the collection efficiency and the detection bias voltage under each calibration structure, the target detection bias voltage that enables the collection efficiency to reach the preset target under each combination of milling depth and pit wall inclination is determined.

7. The method for scheduling detection bias voltage according to milling depth according to claim 1, characterized in that, The method further includes: Obtain the material category of the current calibration sample; The pre-constructed joint bias scheduling table is invoked, which uses the sample material category, total milling depth and local pit wall inclination as joint index dimensions. The joint bias scheduling table is divided into multiple material category branches according to the sample material category. Within each material category branch, a two-dimensional lookup table grid is formed by the total milling depth and local pit wall inclination. Each grid node is associated with the corresponding target detection bias. Based on the current sample material category, select the corresponding material category branch, and obtain the target detection bias by indexing within that material category branch according to the current total milling depth and the current local pit wall inclination.

8. A device for scheduling detection bias voltage according to milling depth, characterized in that, The device includes: The cumulative value acquisition module is used to acquire the cumulative value of the milling depth performed in the current milling sequence during the target interval period of the milling sequence. The target interval period is the interval period after the completion of the previous adjacent milling action and before the start of the next electron beam cross-section imaging in the milling sequence. The cumulative value represents the total milling depth after the end of the previous milling action. The target detection bias determination module, based on the total milling depth, indexes a pre-calibrated bias scheduling table to determine the target detection bias that matches the total milling depth. The bias scheduling table is a pre-calibrated correspondence table between milling depth and detection bias. The target detection bias switching module is used to switch the detector collection gate bias from the current value to the target detection bias during the target interval period. The electron beam cross-sectional imaging completion module is used to enter the next electron beam cross-sectional imaging stage after the target detection bias voltage switching is completed and the target detection bias voltage is reached, and to keep the target detection bias voltage unchanged during the cross-sectional imaging stage in order to complete the electron beam cross-sectional imaging.

9. An electronic device, characterized in that, The electronic device includes a memory and at least one processor, the memory storing instructions; the at least one processor invokes the instructions in the memory to cause the electronic device to perform the various steps of scheduling the probe bias according to the milling depth as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the steps of scheduling the probe bias by milling depth as described in any one of claims 1-7.

Images