Indium antimonide chip beryllium ion implantation and rapid thermal annealing damage repair method and system

Abstract

The invention provides an indium antimonide chip beryllium ion implantation and rapid thermal annealing damage repair method and system, and relates to the technical field of semiconductor materials, and the method comprises the following steps: carrying out deep damage detection on a chip after beryllium ion implantation, determining a depth layer structure based on damage distribution data, generating a defect distribution map, and calculating annihilation activation energy. A temperature stage sequence is established, a segmented rapid thermal annealing curve is generated in combination with energy intervals, surface resistivity signals and reflectivity signals are monitored in the thermal treatment process, temperature stage termination time is adjusted, compensation thermal treatment is applied, lattice damage is efficiently repaired, and doping uniformity and chip stability are improved.

Description

Technical Field

[0001] This invention relates to the field of semiconductor technology, and in particular to a method and system for repairing damage to indium antimonide chips by beryllium ion implantation and rapid thermal annealing. Background Technology

[0002] Indium antimonide (InSb) is an important III-V group narrow bandgap semiconductor material with high electron mobility, high saturated electron drift velocity, and low operating voltage. It is widely used in infrared detectors, Hall elements, and magnetic sensors. In the fabrication of InSb chips, beryllium ion implantation is a crucial process for achieving selective p-type doping, enabling precise control of doping concentration and depth distribution to form high-performance pn structures.

[0003] However, beryllium ion implantation introduces numerous point defects and dislocations into the indium antimonide lattice, severely impacting the electrical performance and reliability of devices. While traditional thermal annealing can partially repair these damages, it has several significant shortcomings. Traditional annealing processes typically use a single temperature or a simple temperature profile, failing to accurately repair defects of varying depths and types. This results in deep defects being difficult to completely eliminate, and surface over-annealing can easily occur, creating new secondary defects. Existing technologies lack precise characterization and analysis methods for defect distribution and type, making it impossible to establish a quantitative relationship between defect type, distribution, and annealing parameters. Annealing processes often rely on empirical design, leading to unstable repair results and difficulty in meeting the requirements of high-precision devices. Conventional annealing processes lack real-time monitoring and feedback adjustment mechanisms, failing to dynamically optimize process parameters based on defect repair progress. This results in low annealing efficiency and difficulty in handling batch-to-batch material variations, leading to poor consistency in repair results and impacting chip yield and performance consistency. Summary of the Invention

[0004] This invention provides a method and system for repairing damage to indium antimonide chips via beryllium ion implantation and rapid thermal annealing, which can solve the problems in the prior art.

[0005] A first aspect of this invention provides a method for repairing damage to an indium antimonide chip via beryllium ion implantation and rapid thermal annealing, comprising:

[0006] Indium antimonide chips are implanted with beryllium ions to form beryllium ion doped regions on the surface of the indium antimonide chips, while lattice displacement damage is generated.

[0007] Depth-direction damage detection was performed on the implanted indium antimonide chip to obtain damage distribution data characterizing the point defect concentration and dislocation density of each depth layer;

[0008] Based on damage distribution data, the depth layer structure of the beryllium ion doped region is determined by gradient analysis, a defect distribution map is generated, the defect type is determined and the annihilation activation energy is calculated, a mapping table is established based on the annihilation activation energy to determine the temperature stage sequence, and a segmented rapid thermal annealing curve is generated by combining the energy interval.

[0009] Heat treatment is performed according to the segmented rapid thermal annealing curve to make point defects in each depth layer migrate to the interstitial space and recombine with vacancies to achieve a staged repair state.

[0010] During the heat treatment process, the surface resistivity and reflectivity signals of the beryllium ion doped region are monitored to construct the defect annihilation process curve, identify the defect elimination node, and adjust the termination time of the temperature stage.

[0011] The residual defect density of each depth layer in the repair state during the acquisition phase is collected. Compensation heat treatment is applied to depth layers that exceed the preset defect density threshold. The segmented rapid thermal annealing curve is updated based on the defect density change after the compensation heat treatment.

[0012] In one optional embodiment, based on damage distribution data, the depth layer structure of the beryllium ion-doped region is determined through gradient analysis, a defect distribution map is generated, the defect type is determined, and the annihilation activation energy is calculated, including:

[0013] The damage distribution data is scanned and sampled along the depth direction at preset intervals. The point defect concentration gradient and dislocation density gradient of adjacent sampling points are calculated. The depth boundary point is determined by obtaining the location where the gradient change rate exceeds the background reference value. The depth layer division structure is determined by measuring the degree of difference in damage characteristics between adjacent depth layers.

[0014] The mesh is divided according to the depth layer structure. The distribution values ​​of point defect concentration and dislocation density within the mesh are calculated based on the preset mesh size. The variation law of concentration gradient between adjacent meshes is measured. Combined with the distribution characteristics of mesh location, a depth layer defect distribution map is generated.

[0015] The location correlation data of point defects and dislocations are extracted from the defect distribution map of the deep layer to obtain the degree of aggregation and dispersion of the defect group, measure the variation of point defect concentration and dislocation density in the depth direction, construct a set of defect feature parameters, and determine the defect type in the deep layer.

[0016] By mapping defect types to crystal structures, measuring the degree of lattice distortion and bonding state, establishing a mapping relationship between defect structural characteristics and annihilation activation energy, and determining the annihilation activation energy value of the layer at the given depth.

[0017] In one optional embodiment, establishing a mapping table based on the annihilation activation energy to determine the temperature stage sequence, and generating segmented rapid thermal annealing curves by combining energy intervals, includes:

[0018] The annihilation activation energies of all depth layers are grouped and sorted according to their numerical values. A mapping table between annihilation activation energies and depth layers is established, and the number and hierarchical structure of temperature stages are determined based on the mapping table.

[0019] Based on the grouping and sorting results of annihilation activation energies, the energy interval between adjacent annihilation activation energies is calculated, and the energy interval is used as the basis for dividing the temperature stages to construct a temperature stage sequence.

[0020] For each temperature stage, the annealing temperature parameters are set according to the annihilation activation energy of the corresponding depth layer, and the heating rate and holding time are determined in combination with the kinetic characteristics of defect annihilation to generate a temperature-time curve.

[0021] Connect the temperature-time curves of each temperature stage in order of increasing annihilation activation energy, add transition intervals between temperature stages, and construct a complete segmented rapid thermal annealing curve.

[0022] In one optional embodiment, during the heat treatment process, monitoring the surface resistivity and reflectivity signals of the beryllium ion-doped region, constructing a defect annihilation process curve, identifying defect elimination nodes, and adjusting the termination time of the temperature stage include:

[0023] During the heat treatment process, a multi-channel detector array is used to scan the surface of the beryllium ion doped region in real time, and the surface resistivity signal and reflectivity signal are collected simultaneously. A dual-parameter time-series acquisition matrix is ​​established to obtain continuous change data of surface parameters and form the spatiotemporal distribution characteristics of parameter changes.

[0024] Extreme points in resistivity and reflectivity time-series data are extracted based on a dual-parameter time-series acquisition matrix to determine the turning points of parameter change trajectories and construct defect annihilation process curves.

[0025] Multi-scale segmented slope calculations were performed on the defect annihilation process curve to identify the location and magnitude of slope abrupt change points. Combined with spatiotemporal distribution characteristics, slope abrupt change points were marked as defect elimination nodes, and a defect elimination judgment criterion was established.

[0026] Based on the occurrence time and distribution pattern of defect elimination nodes, and combined with the defect elimination judgment criteria, the termination time of the current temperature stage is adjusted in real time to establish a mapping relationship between defect elimination and temperature control, thereby realizing dynamic regulation of the heat treatment process.

[0027] In one optional embodiment, extreme points in the time-series data of resistivity and reflectivity are extracted based on a dual-parameter time-series acquisition matrix to determine the inflection points of the parameter change trajectory, and a defect annihilation process curve is constructed, including:

[0028] The resistivity and reflectivity time series data are segmented according to a preset time interval. The maximum and minimum values ​​of each segment are measured, and the parameter changes are calculated based on the corresponding intermediate values ​​to obtain the offset dataset characterizing the parameter fluctuations.

[0029] Extract the extreme points of resistivity and reflectivity from the offset dataset, determine the extreme moments, calculate the time interval between extreme moments, divide the extreme points into parameter synchronous intervals and asynchronous intervals according to the preset time threshold, and generate a time-series mapping table of parameter evolution.

[0030] Within the parameter synchronization interval, resistivity and reflectivity are projected onto a two-dimensional coordinate plane to determine the trajectory of parameter changes. Based on a preset curvature threshold, the turning points of the trajectory are filtered, and the time sequence corresponding to the turning points is extracted to obtain the time stamp of defect state transition.

[0031] Using time markers as demarcation points, resistivity and reflectivity data within a preset neighborhood of each demarcation point are extracted. The endpoint values ​​of adjacent data segments are connected using a curve reconstruction method to form multiple characteristic curves. The characteristic curves are then combined to obtain the defect annihilation process curve.

[0032] In one optional embodiment, the residual defect density of each depth layer in the repair state during the acquisition phase is collected. Compensation heat treatment is applied to depth layers exceeding a preset defect density threshold. The segmented rapid thermal annealing curve is updated based on the defect density change after the compensation heat treatment, including:

[0033] Scan and measure each depth layer along the depth direction to obtain spatial distribution data of point defects, calculate the defect density difference between adjacent depth layers, construct a residual defect depth distribution map, and record the defect density value.

[0034] Based on the comparison between the defect density value in the residual defect depth distribution map and the preset defect density threshold, the depth layer position that exceeds the preset defect density threshold is determined, the defect type feature at the depth layer position is extracted, and a target depth layer compensation sequence is formed.

[0035] Based on the defect type characteristics in the target depth layer compensation sequence, the correspondence between the lattice structure distortion value and the heat treatment temperature is determined, and the compensation heat treatment temperature parameters are generated.

[0036] Perform compensation heat treatment according to the compensation heat treatment temperature parameters, measure the defect density of the target depth layer after compensation heat treatment, calculate the change in defect density caused by compensation heat treatment, and combine the change in defect density with the temperature parameters of the original segmented rapid thermal annealing curve to construct an updated segmented rapid thermal annealing curve.

[0037] In one optional embodiment, based on the defect type characteristics in the target depth layer compensation sequence, the correspondence between the lattice structure distortion value and the heat treatment temperature is determined, and the compensation heat treatment temperature parameters are generated, including:

[0038] Spatial location information of defect type features in the target depth layer compensation sequence is extracted, the position of lattice atoms in the target depth layer is scanned, the real-time coordinate information of lattice atoms is recorded, the displacement distance of lattice atoms relative to standard lattice points is measured, and lattice atom displacement distribution data is generated.

[0039] Extract the displacement vectors between adjacent atoms from the lattice atomic displacement distribution data, construct a displacement vector group, determine the stress direction in the displacement vector group, obtain the stress intensity value, and form the stress distribution field of the target depth layer;

[0040] Based on the stress distribution field, identify the spatial channel for stress propagation, measure the energy accumulation state of the stress field along the spatial channel, extract the spatial coordinates of the energy accumulation points, and establish the distribution sequence of the energy accumulation points.

[0041] By combining the stress intensity values ​​with the distribution sequence of energy accumulation points, the spatial evolution law of lattice structure distortion is determined, the heat treatment temperature range is divided, the correspondence between lattice structure distortion values ​​and heat treatment temperatures is established, and compensation heat treatment temperature parameters are generated.

[0042] A second aspect of the present invention provides a beryllium ion implantation and rapid thermal annealing damage repair system for indium antimonide chips, comprising:

[0043] The ion implantation unit is used to implant beryllium ions into the indium antimonide chip, forming a beryllium ion doped region on the surface of the indium antimonide chip, while simultaneously causing lattice displacement damage.

[0044] The damage detection unit is used to perform depth-direction damage detection on the implanted indium antimonide chip to obtain damage distribution data characterizing the point defect concentration and dislocation density of each depth layer.

[0045] The annealing planning unit is used to determine the depth layer structure of the beryllium ion doped region based on damage distribution data through gradient analysis, generate a defect distribution map, determine the defect type and calculate the annihilation activation energy, establish a mapping table based on the annihilation activation energy to determine the temperature stage sequence, and generate a segmented rapid thermal annealing curve by combining the energy interval.

[0046] The heat treatment unit is used to perform heat treatment according to the segmented rapid heat annealing curve, so that the point defects in each depth layer migrate to the interstitial space and recombine with the vacancy to achieve the staged repair state.

[0047] The process monitoring unit is used to monitor the surface resistivity and reflectivity signals of the beryllium ion-doped region during heat treatment, construct the defect annihilation process curve, identify the defect elimination node, and adjust the termination time of the temperature stage.

[0048] The defect compensation unit is used to collect the residual defect density of each depth layer in the repair stage, apply compensation heat treatment to depth layers that exceed the preset defect density threshold, and update the segmented rapid thermal annealing curve based on the defect density change after compensation heat treatment.

[0049] A third aspect of the present invention provides an electronic device, comprising:

[0050] processor;

[0051] Memory used to store processor-executable instructions;

[0052] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0053] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0054] In this embodiment of the invention, the provided indium antimonide chip beryllium ion implantation and rapid thermal annealing damage repair method accurately obtains the defect distribution of different depth layers inside the chip after ion implantation through depth-direction damage detection and gradient analysis technology, and establishes a mapping relationship between defect type and annihilation activation energy. This allows for the targeted design of segmented rapid thermal annealing curves, achieving precise repair of chip damage. By using real-time monitoring of surface resistivity and reflectivity signals, a defect annihilation process curve is constructed, which can accurately identify defect elimination nodes and dynamically adjust heat treatment process parameters, avoiding over-repair or under-repair problems caused by traditional fixed-parameter thermal annealing, thus improving the electrical performance and stability of the indium antimonide chip. By introducing a compensation heat treatment mechanism, the method of this invention performs targeted treatment on depth layers with residual defect density exceeding the threshold, and continuously optimizes the annealing curve based on the treatment effect, forming a closed-loop adaptive damage repair system. This significantly reduces the defect density of the indium antimonide chip, improves the chip's carrier mobility and photoelectric conversion efficiency, and extends the device's lifespan. Attached Figure Description

[0055] Figure 1 This is a schematic flowchart of the beryllium ion implantation and rapid thermal annealing damage repair method for indium antimonide chips according to an embodiment of the present invention;

[0056] Figure 2 This is a flowchart of the closed-loop control for defect elimination in the heat treatment process. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. 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.

[0058] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0059] Figure 1 This is a schematic flowchart of the beryllium ion implantation and rapid thermal annealing damage repair method for indium antimonide chips according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes:

[0060] Indium antimonide chips are implanted with beryllium ions to form beryllium ion doped regions on the surface of the indium antimonide chips, while lattice displacement damage is generated.

[0061] Depth-direction damage detection was performed on the implanted indium antimonide chip to obtain damage distribution data characterizing the point defect concentration and dislocation density of each depth layer;

[0062] Based on damage distribution data, the depth layer structure of the beryllium ion doped region is determined by gradient analysis, a defect distribution map is generated, the defect type is determined and the annihilation activation energy is calculated, a mapping table is established based on the annihilation activation energy to determine the temperature stage sequence, and a segmented rapid thermal annealing curve is generated by combining the energy interval.

[0063] Heat treatment is performed according to the segmented rapid thermal annealing curve to make point defects in each depth layer migrate to the interstitial space and recombine with vacancies to achieve a staged repair state.

[0064] During the heat treatment process, the surface resistivity and reflectivity signals of the beryllium ion doped region are monitored to construct the defect annihilation process curve, identify the defect elimination node, and adjust the termination time of the temperature stage.

[0065] The residual defect density of each depth layer in the repair state during the acquisition phase is collected. Compensation heat treatment is applied to depth layers that exceed the preset defect density threshold. The segmented rapid thermal annealing curve is updated based on the defect density change after the compensation heat treatment.

[0066] In one optional implementation, based on damage distribution data, the depth layer structure of the beryllium ion-doped region is determined through gradient analysis, a defect distribution map is generated, the defect type is determined, and the annihilation activation energy is calculated, including:

[0067] The damage distribution data is scanned and sampled along the depth direction at preset intervals. The point defect concentration gradient and dislocation density gradient of adjacent sampling points are calculated. The depth boundary point is determined by obtaining the location where the gradient change rate exceeds the background reference value. The depth layer division structure is determined by measuring the degree of difference in damage characteristics between adjacent depth layers.

[0068] The mesh is divided according to the depth layer structure. The distribution values ​​of point defect concentration and dislocation density within the mesh are calculated based on the preset mesh size. The variation law of concentration gradient between adjacent meshes is measured. Combined with the distribution characteristics of mesh location, a depth layer defect distribution map is generated.

[0069] The location correlation data of point defects and dislocations are extracted from the defect distribution map of the deep layer to obtain the degree of aggregation and dispersion of the defect group, measure the variation of point defect concentration and dislocation density in the depth direction, construct a set of defect feature parameters, and determine the defect type in the deep layer.

[0070] By mapping defect types to crystal structures, measuring the degree of lattice distortion and bonding state, establishing a mapping relationship between defect structural characteristics and annihilation activation energy, and determining the annihilation activation energy value of the layer at the given depth.

[0071] In one specific implementation, damage analysis is performed on a material sample after beryllium ion implantation to collect data on the point defect concentration and dislocation density at different depths. This data is typically expressed as the number of defects per cubic centimeter (cm²). -3 ) and dislocation line length per square centimeter (cm) -2 )express.

[0072] When determining the depth layer structure, data is sampled along the sample depth at preset intervals of 10 nm. For example, in a sample with a total depth of 2 μm, point defect concentration and dislocation density data are acquired at 200 sampling points. For adjacent sampling points, the point defect concentration gradient is calculated, which is the concentration difference between two adjacent points divided by the sampling interval, in cm. -3 / nm. Similarly, calculate the dislocation density gradient, in cm. -2 / nm. In practical applications, the point defect concentration of beryllium ion-doped samples may range from 10 on the surface. 18 cm -3 Gradually changing to the deep 10 15 cm -3 Dislocation density from 10 10 cm -2 Reduced to 10 7 cm -2 .

[0073] Depth boundaries are determined by analyzing the gradient change rate. The gradient change rate is defined as the ratio of the difference between two adjacent gradient values ​​to the gradient value at the current location. When the gradient change rate exceeds a preset threshold (e.g., 30%), the location is marked as a potential boundary. In practice, for beryllium-doped silicon materials, significant gradient changes are typically observed at approximately 150 nm, 480 nm, and 870 nm below the surface; these locations serve as depth layer boundaries. For each defined depth layer, the mean, standard deviation, and coefficient of variation of the point defect concentration and dislocation density within the layer are calculated to assess the damage uniformity within the layer.

[0074] Based on the defined depth layer structure, mesh generation is performed to produce a defect distribution map. A 100nm × 100nm mesh size is used in the transverse direction (XY plane), while maintaining a precision of 10nm in the depth direction (Z-axis). For each mesh cell, the average concentration of point defects and the average density of dislocations within it are calculated. In a practical application, a 2μm × 2μm × 2μm sample volume is divided into 8000 mesh cells, each containing point defect concentration and dislocation density data.

[0075] By comparing data between different grids, the defect distribution pattern is determined. The concentration gradient vector between adjacent grids is calculated, including both magnitude and direction information. For example, in the transition layer at a depth of approximately 450 nm, the point defect concentration gradient between adjacent grids can reach 1.5 × 10⁻⁶. 16 cm -3 / nm, significantly higher than the average of 5×10 in other regions. 15 cm -3 / nm. This gradient distribution characteristic is highly correlated with defect aggregation regions.

[0076] To generate an intuitive depth-layer defect distribution map, the grid data was converted into a color-coded image. Point defect concentration is typically represented by a red color level, and dislocation density by a blue color level; these are superimposed to form a composite image. In the map, color intensity is proportional to defect concentration / density, making high-defect regions more prominent. Based on a practical example, with a beryllium ion implantation dose of 1×10⁻⁶... 15 cm -2 In the sample with an energy of 120keV, a distinct defect aggregation band was formed in the depth range of 450-550nm, which appeared as a deep red area in the spectrum.

[0077] Defect type determination is achieved by analyzing pattern characteristics in the defect distribution map. Spatial distribution data of point defects and dislocations at each depth layer are extracted from the map, and the correlation coefficient between them is calculated. In beryllium ion-doped samples, the surface layer (0-150 nm) typically exhibits a low correlation between point defects and dislocations (correlation coefficient approximately 0.3), indicating that these two types of defects are relatively independently distributed. However, in the intermediate layer (150-480 nm), the correlation coefficient rises to above 0.7, suggesting that point defects tend to cluster around dislocations.

[0078] For each depth layer, the point defect aggregation degree is calculated, defined as the ratio of the number of defect clusters per unit volume to the total number of defects. In actual samples, the point defect aggregation degree is approximately 0.25 in the surface layer, reaches 0.65 in the intermediate layer, and drops to 0.4 in the deep layers. This characteristic is used to distinguish different types of defect structures. By comparing the characteristic parameters of different defect types such as dislocation loops, vacancy clusters, and interstitial atom clusters, the dominant defect type in each depth layer is determined. For example, in a typical beryllium-doped sample, the surface layer is mainly composed of interstitial point defects (approximately 75%), the intermediate layer is dominated by vacancy clusters (approximately 60%), and the deep layers are mainly composed of small dislocation loops (approximately 50%).

[0079] The calculation of annihilation activation energy is based on defined defect types and structural parameters. The degree of lattice distortion is measured and quantified as the relative rate of change of the lattice constant; for example, in the vacancy cluster region of the intermediate layer, the rate of change of the lattice constant reaches 0.8%. Changes in the bonding state around the defect are measured, including the deviation of bond lengths and bond angles, to establish a mapping relationship between defect structural characteristics and annihilation activation energy. In practical applications, for beryllium-doped silicon samples, the annihilation activation energy of point defects in the surface layer is typically in the range of 0.6–0.8 eV, for vacancy clusters in the intermediate layer it is 1.1–1.4 eV, and for dislocation loops in the deep layers it is 1.5–1.8 eV. These values ​​correspond to the energy thresholds at which defects may annihilate or recombine at different temperatures, providing a basis for optimizing heat treatment processes.

[0080] In this embodiment, the deep layer structure, defect distribution characteristics, and annihilation activation energy of the beryllium ion doped region are fully described, providing an important reference for the performance optimization of semiconductor materials and device design.

[0081] In one optional implementation, a mapping table is established based on the annihilation activation energy to determine the temperature stage sequence, and a segmented rapid thermal annealing curve is generated by combining the energy intervals, including:

[0082] The annihilation activation energies of all depth layers are grouped and sorted according to their numerical values. A mapping table between annihilation activation energies and depth layers is established, and the number and hierarchical structure of temperature stages are determined based on the mapping table.

[0083] Based on the grouping and sorting results of annihilation activation energies, the energy interval between adjacent annihilation activation energies is calculated, and the energy interval is used as the basis for dividing the temperature stages to construct a temperature stage sequence.

[0084] For each temperature stage, the annealing temperature parameters are set according to the annihilation activation energy of the corresponding depth layer, and the heating rate and holding time are determined in combination with the kinetic characteristics of defect annihilation to generate a temperature-time curve.

[0085] Connect the temperature-time curves of each temperature stage in order of increasing annihilation activation energy, add transition intervals between temperature stages, and construct a complete segmented rapid thermal annealing curve.

[0086] In one specific embodiment, the defect annihilation activation energy of each depth layer in the semiconductor material is measured and collected. Taking a silicon wafer as an example, samples are uniformly taken from the surface to a depth of 30 μm, divided into 15 depth layers, each with a thickness of 2 μm. The annihilation activation energy values ​​of each layer are measured by differential scanning calorimetry. The activation energy of the surface layer is 0.89 eV, which varies with depth, with the activation energy of the deepest layer being 1.76 eV.

[0087] The measured annihilation activation energies were sorted and grouped according to their numerical values ​​to establish a mapping table between activation energies and depth layers. For example, activation energies in the range of 0.85-1.0 eV were grouped into the first group, containing 1-3 layers; 1.0-1.2 eV into the second group, containing 4-7 layers; 1.2-1.4 eV into the third group, containing 8-10 layers; 1.4-1.6 eV into the fourth group, containing 11-13 layers; and 1.6-1.8 eV into the fifth group, containing 14-15 layers. Through this mapping relationship, the number of temperature stages was determined to be 5, corresponding to the defect annihilation process in different depth ranges.

[0088] Based on the grouping results, the energy intervals between adjacent annihilation activation energy groups were calculated. The interval between the first and second groups was 0.2 eV, the interval between the second and third groups was 0.2 eV, the interval between the third and fourth groups was 0.2 eV, and the interval between the fourth and fifth groups was 0.2 eV. The energy interval served as the basis for dividing the temperature stages; a larger interval indicated a greater required temperature change. Based on this, a temperature stage sequence was constructed: the first stage corresponds to 0.85-1.0 eV, with a temperature set at 573-673 K; the second stage corresponds to 1.0-1.2 eV, with a temperature set at 673-773 K; the third stage corresponds to 1.2-1.4 eV, with a temperature set at 773-873 K; the fourth stage corresponds to 1.4-1.6 eV, with a temperature set at 873-973 K; and the fifth stage corresponds to 1.6-1.8 eV, with a temperature set at 973-1073 K.

[0089] For each temperature stage, annealing parameters were set based on the annihilation activation energy characteristics of the corresponding depth layer. The annealing temperature was determined by the Arrhenius relation, meaning the defect annihilation rate is related to both temperature and activation energy. The first stage (573-673K) had a heating rate of 10K / s and a holding time of 20 seconds; the second stage (673-773K) had a heating rate of 15K / s and a holding time of 15 seconds; the third stage (773-873K) had a heating rate of 20K / s and a holding time of 10 seconds; the fourth stage (873-973K) had a heating rate of 25K / s and a holding time of 8 seconds; and the fifth stage (973-1073K) had a heating rate of 30K / s and a holding time of 5 seconds. The heating rate increased with increasing temperature, while the holding time decreased accordingly, because the defect annihilation rate is faster at higher temperatures.

[0090] When generating the segmented annealing temperature-time curves, starting from the first stage, the initial temperature is set to room temperature (300K), increased to 573K at a rate of 10K / s, and held for 20 seconds to annihilate surface defects. To ensure a smooth transition between stages, a 5-second transition period is added, during which the temperature increases from 573K to 673K ​​at a rate of 20K / s. The second stage starts at 673K, is held for 15 seconds, and then increases to 773K after a 5-second transition period. This process is repeated to connect the temperature curves for all stages.

[0091] In practical applications, the phosphorus ion implantation dose is 1×10⁻⁶. 15 cm -2 The test was conducted on 6-inch silicon wafers. After adopting the above-mentioned segmented rapid thermal annealing process, the defect density was measured by secondary ion mass spectrometry. It was found that the defect density in the range from the surface to 30μm was reduced by 94.7%, and the uniformity of defect distribution in each depth layer was improved to ±3.2%, which is a significant improvement compared with the traditional single-temperature annealing process (reduction of 85.3%, uniformity ±12.5%).

[0092] In terms of process implementation, a rapid thermal annealing device with precise temperature control is used, achieving a temperature control accuracy of ±1K and an adjustable heating rate within the range of 0-100K / s, with a maximum operating temperature of 1473K. By pre-setting the temperature stage sequence and parameters for each stage, fully automated annealing process control is achieved. Experimental data shows that this method can control the total annealing time within 120 seconds, shortening it by 80% compared to traditional constant-temperature, long-time annealing. Simultaneously, it improves defect repair efficiency and depth distribution uniformity, thereby enhancing the electrical characteristics and reliability of the device.

[0093] In one optional implementation, during the heat treatment process, monitoring the surface resistivity and reflectivity signals of the beryllium ion-doped region, constructing a defect annihilation process curve, identifying defect elimination nodes, and adjusting the termination time of the temperature stage include:

[0094] During the heat treatment process, a multi-channel detector array is used to scan the surface of the beryllium ion doped region in real time, and the surface resistivity signal and reflectivity signal are collected simultaneously. A dual-parameter time-series acquisition matrix is ​​established to obtain continuous change data of surface parameters and form the spatiotemporal distribution characteristics of parameter changes.

[0095] Extreme points in resistivity and reflectivity time-series data are extracted based on a dual-parameter time-series acquisition matrix to determine the turning points of parameter change trajectories and construct defect annihilation process curves.

[0096] Multi-scale segmented slope calculations were performed on the defect annihilation process curve to identify the location and magnitude of slope abrupt change points. Combined with spatiotemporal distribution characteristics, slope abrupt change points were marked as defect elimination nodes, and a defect elimination judgment criterion was established.

[0097] Based on the occurrence time and distribution pattern of defect elimination nodes, and combined with the defect elimination judgment criteria, the termination time of the current temperature stage is adjusted in real time to establish a mapping relationship between defect elimination and temperature control, thereby realizing dynamic regulation of the heat treatment process.

[0098] In one specific embodiment, the heat treatment system is configured with a temperature control unit, a multi-channel detector array, and a data processing unit. The multi-channel detector array includes a resistivity measurement probe and a reflectivity measurement probe, both arranged in a 10×10 matrix layout, covering the surface of the beryllium ion-doped region. The resistivity measurement probe employs a four-probe method, with an operating current set to 0.1 mA and a measurement range of 0.001-1000 Ω·cm. The reflectivity measurement probe utilizes the laser reflection principle, with an incident light wavelength of 632.8 nm and an incident angle of 45°, measuring relative reflectivity from 0-100%.

[0099] During the heat treatment process, the detector array performs real-time scanning of the beryllium ion-doped region surface at a scanning frequency of 10 Hz, simultaneously acquiring surface resistivity and reflectivity signals. For the beryllium ion-doped region of an 8-inch silicon wafer, the scanning resolution is set to 0.5 mm × 0.5 mm, and data from 1600 measurement points can be acquired per detection cycle. The data acquisition system organizes the acquired resistivity and reflectivity data into a two-parameter time-series acquisition matrix D(x, y, t, ρ, r), where x and y are spatial coordinates, t is the time point, ρ is the resistivity value, and r is the reflectivity value. In practical applications, the beryllium ion doping concentration is 5 × 10¹. 5 The sample has a surface resistivity of approximately 15 Ω·cm and a reflectivity of approximately 35%.

[0100] The data processing unit processes the dual-parameter time-series acquisition matrix to extract the temporal variation characteristics of resistivity and reflectivity. First, resistivity time-series data ρ(t) and reflectivity time-series data r(t) are extracted for each spatial point (x, y). High-frequency noise is removed using wavelet transform, improving the signal-to-noise ratio by approximately 15 dB. For the processed time-series data, an extremum detection algorithm is used to identify local maxima and minima, with a threshold set at 3% of the signal amplitude to filter out weak fluctuations. In actual testing, during heat treatment at 450°C, the resistivity of the beryllium ion-doped region initially increases and then decreases, reaching a maximum value of approximately 18.5 Ω·cm about 120 seconds after the start of heat treatment. Reflectivity, on the other hand, initially decreases and then increases, reaching a minimum value of approximately 32% about 105 seconds after the start of heat treatment.

[0101] Based on the extracted extreme point data, a defect annihilation process curve is constructed. The resistivity maxima and reflectivity minima are mapped onto the time axis to form a sequence of inflection points in the parameter variation trajectory: {(t1, ρ1, r1), (t2, ρ2, r2), ..., (t...}. n , ρ n r n Cubic spline interpolation is performed between these inflection points to generate a continuous defect annihilation process curve. In practical applications, for a beryllium doping concentration of 5 × 10⁻⁶, 15 cm -3 The defect annihilation process curve formed during the 450°C heat treatment of the sample contains about 6-8 key inflection points, corresponding to the elimination process of different types of defects.

[0102] Multi-scale segmented slope calculations were performed on the defect annihilation process curve to identify slope abrupt change points. A sliding window method was used to calculate the local slope, with window widths ranging from 5 to 30 seconds to cover variations across different time scales. The calculated slope values ​​were compared with preset thresholds; slope abrupt change points were marked when the absolute value of the slope changed by more than 30% for a duration exceeding 3 seconds. These abrupt change points were then correlated with defect elimination nodes based on spatiotemporal distribution characteristics. In practical applications, key defect elimination nodes observed during the heat treatment of beryllium-doped regions included: vacancy defect elimination node (approximately 95 seconds), where the resistivity slope changed from positive to negative and the reflectivity slope changed from negative to positive; interstitial atom defect elimination node (approximately 145 seconds), where the resistivity slope reached its maximum negative value; and composite defect elimination node (approximately 210 seconds), where both resistivity and reflectivity slopes simultaneously stabilized.

[0103] The criteria for defect elimination are established, including: a change in resistivity slope from positive to negative and a change in reflectivity slope from negative to positive, indicating vacancy-type defect elimination; a maximum resistivity decrease slope, indicating interstitial atom defect elimination; and resistivity and reflectivity values ​​stabilizing within ±1% of their final values ​​for more than 30 seconds, indicating composite defect elimination. In specific process applications, for beryllium doping concentrations of 5 × 10⁻⁶... 15 cm -3 After eliminating vacancy-type defects, the resistivity of the sample stabilized at approximately 14.2 Ω·cm, and the reflectivity stabilized at approximately 36.5%.

[0104] Based on defect elimination nodes and judgment criteria, the system achieves dynamic control of the heat treatment process. The system adjusts the termination time of each heat treatment temperature stage according to the real-time monitoring of the defect elimination node's occurrence time. A prediction-correction algorithm is used to dynamically adjust the duration of subsequent heat treatment stages based on the deviation between the defect elimination node's occurrence time and the theoretically expected time. For a standard heat treatment process where the 450°C stage lasts 300 seconds, when a composite defect elimination node is detected at 210 seconds, the system adjusts the stage termination time to 230 seconds (adding a 20-second stabilization time), optimizing the heat treatment process and improving efficiency by approximately 23%. For the 550°C heat treatment stage, the duration is adaptively adjusted based on the defect elimination status of the previous stage, with typical values ​​ranging from 180 seconds to 145-165 seconds.

[0105] The above technical solution enables real-time monitoring of the defect annihilation process in beryllium ion-doped regions based on surface resistivity and reflectivity signals, and dynamic optimization of the heat treatment process, thereby improving heat treatment efficiency and product quality stability.

[0106] like Figure 2 The diagram shown illustrates the closed-loop control flowchart for defect elimination in the heat treatment process.

[0107] In one optional implementation, extreme points in the time-series data of resistivity and reflectivity are extracted based on a dual-parameter time-series acquisition matrix to determine the inflection points of the parameter change trajectory, and a defect annihilation process curve is constructed, including:

[0108] The resistivity and reflectivity time series data are segmented according to a preset time interval. The maximum and minimum values ​​of each segment are measured, and the parameter changes are calculated based on the corresponding intermediate values ​​to obtain the offset dataset characterizing the parameter fluctuations.

[0109] Extract the extreme points of resistivity and reflectivity from the offset dataset, determine the extreme moments, calculate the time interval between extreme moments, divide the extreme points into parameter synchronous intervals and asynchronous intervals according to the preset time threshold, and generate a time-series mapping table of parameter evolution.

[0110] Within the parameter synchronization interval, resistivity and reflectivity are projected onto a two-dimensional coordinate plane to determine the trajectory of parameter changes. Based on a preset curvature threshold, the turning points of the trajectory are filtered, and the time sequence corresponding to the turning points is extracted to obtain the time stamp of defect state transition.

[0111] Using time markers as demarcation points, resistivity and reflectivity data within a preset neighborhood of each demarcation point are extracted. The endpoint values ​​of adjacent data segments are connected using a curve reconstruction method to form multiple characteristic curves. The characteristic curves are then combined to obtain the defect annihilation process curve.

[0112] In one specific implementation, resistivity and reflectivity parameters of the material during the annealing process are collected using a distributed sensor network to form a time-series database. The acquisition frequency can be set to 10Hz, i.e., data is acquired once every 0.1 seconds, and continuous acquisition is performed for 60 minutes to form a time-series matrix containing 36,000 data points. The time-series matrix contains complete data recording the changes in resistivity and reflectivity over time during the annihilation of internal defects in the material.

[0113] The collected time-series data was segmented, dividing the 60-minute period into 120 time windows, each lasting 30 seconds. Within each time window, the maximum and minimum values ​​of resistivity and reflectivity were calculated. For example, in the first 30-second window, the maximum resistivity was 195.6 μΩ·cm, and the minimum was 192.3 μΩ·cm; the maximum reflectivity was 0.856, and the minimum was 0.842. The median values ​​of the parameters for each window were calculated: the median resistivity was 193.95 μΩ·cm, and the median reflectivity was 0.849. Using these median values ​​as the base, the parameter offsets at each time point within the window were calculated, forming an offset dataset. For resistivity, the offset range was [-1.65 μΩ·cm, 1.65 μΩ·cm]; for reflectivity, the offset range was [-0.007, 0.007].

[0114] Extreme points of resistivity and reflectivity were extracted from the offset dataset. A sliding window method was used to identify local maxima and minima, with a window size of 5 seconds. In the offset data, a point was identified as a local maximum when its value was greater than the values ​​of all points within 2.5 seconds before and after it; and as a local minimum when its value was less than the values ​​of all points within 2.5 seconds before and after it. For example, a resistivity maximum of 198.2 μΩ·cm was identified at the 10th minute, and a reflectivity minimum of 0.841 was identified at 10 minutes and 15 seconds. The exact times of all extreme points were recorded to form an extreme point time sequence.

[0115] The time interval between the extreme points of resistivity and reflectivity is calculated. A time threshold of 3 seconds is set. When the time interval between the extreme points of the two parameters is less than 3 seconds, these two extreme points are classified as synchronous intervals; when the time interval is greater than 3 seconds, they are classified as asynchronous intervals. For example, at the 15th minute, resistivity reaches a minimum of 189.5 μΩ·cm, and at 2 seconds after the 15th minute, reflectivity reaches a maximum of 0.862. The time interval is 2 seconds, which is less than the threshold of 3 seconds; therefore, these two extreme points are classified as synchronous intervals. Based on this, a time-series mapping table of parameter evolution is generated, recording the time, value, and synchronous / asynchronous relationship of the extreme points of resistivity and reflectivity.

[0116] Within the parameter synchronization interval, resistivity and reflectivity are projected onto a two-dimensional coordinate plane, forming a trajectory of parameter changes. The horizontal axis of the coordinate plane represents resistivity, with values ​​ranging from [180 μΩ·cm to 200 μΩ·cm]; the vertical axis represents reflectivity, with values ​​ranging from [0.84 to 0.87]. For example, within the synchronization interval of 20-25 minutes, the trajectory starts at (195.2 μΩ·cm, 0.849) and ends at (186.8 μΩ·cm, 0.865).

[0117] Curvature analysis was performed on the parameter change trajectory, with a curvature threshold of 0.15. The curvature value of each point on the trajectory was calculated, and points with curvature values ​​greater than 0.15 were identified as inflection points. For example, at 22 minutes and 30 seconds, the trajectory curvature value was 0.17, which was greater than the threshold of 0.15, so it was marked as an inflection point, with corresponding parameter values ​​of (191.5 μΩ·cm, 0.855). The time sequences corresponding to all inflection points were extracted, such as [8 minutes and 30 seconds, 22 minutes and 30 seconds, 35 minutes and 15 seconds, 48 ​​minutes and 45 seconds], and these times were identified as time markers for defect state transitions.

[0118] Using time markers as demarcation points, resistivity and reflectivity data are extracted within a 5-second neighborhood before and after each demarcation point. For example, at the time marker of 22 minutes and 30 seconds, resistivity and reflectivity data from 22 minutes and 25 seconds to 22 minutes and 35 seconds are extracted to form a local data segment containing 10 seconds of data. The resistivity data range is [190.8 μΩ·cm, 192.3 μΩ·cm], and the reflectivity data range is [0.853, 0.857].

[0119] Cubic spline interpolation is used to connect the endpoints of adjacent data segments to achieve a smooth transition of the curve. Twenty interpolation points are generated between each data segment to ensure the continuity of the first and second derivatives of the curve at the connection points. For example, a smooth transition curve from (196.7 μΩ·cm, 0.845) to (192.3 μΩ·cm, 0.853) is generated between the first data segment (8 minutes 25 seconds to 8 minutes 35 seconds) and the second data segment (22 minutes 25 seconds to 22 minutes 35 seconds).

[0120] By combining all data segments and transition curves, a complete defect annihilation process curve is formed. This curve, presented on a two-dimensional plane, is a continuous trajectory from the starting point (197.5 μΩ·cm, 0.844) to the ending point (183.2 μΩ·cm, 0.866), containing multiple inflection points. It visually demonstrates the complete process of material defects gradually annihilating from a high-density state to a low-density state. Each inflection point on the curve corresponds to a qualitative change in the internal defect state of the material, such as key physical processes like defect aggregation, decomposition, and recombination.

[0121] In this embodiment, the constructed defect annihilation process curve can not only accurately reflect the dynamic process of defect evolution inside the material, but also accurately identify the key moment of defect state transition, providing an important basis for material performance optimization and defect control.

[0122] In one optional implementation, the residual defect density of each depth layer in the repair state during the acquisition phase is collected. Compensation heat treatment is applied to depth layers exceeding a preset defect density threshold. The segmented rapid thermal annealing curve is updated based on the defect density change after the compensation heat treatment, including:

[0123] Scan and measure each depth layer along the depth direction to obtain spatial distribution data of point defects, calculate the defect density difference between adjacent depth layers, construct a residual defect depth distribution map, and record the defect density value.

[0124] Based on the comparison between the defect density value in the residual defect depth distribution map and the preset defect density threshold, the depth layer position that exceeds the preset defect density threshold is determined, the defect type feature at the depth layer position is extracted, and a target depth layer compensation sequence is formed.

[0125] Based on the defect type characteristics in the target depth layer compensation sequence, the correspondence between the lattice structure distortion value and the heat treatment temperature is determined, and the compensation heat treatment temperature parameters are generated.

[0126] Perform compensation heat treatment according to the compensation heat treatment temperature parameters, measure the defect density of the target depth layer after compensation heat treatment, calculate the change in defect density caused by compensation heat treatment, and combine the change in defect density with the temperature parameters of the original segmented rapid thermal annealing curve to construct an updated segmented rapid thermal annealing curve.

[0127] In one specific implementation, scanning measurements of each depth layer along the depth direction are achieved using a high-precision vertical resolution scanning transmission electron microscope (STEM). The STEM device scans layer by layer along the sample depth direction with a 5nm step size, pausing for 200ms at each depth position to acquire data, obtaining spatial distribution data of point defects. For example, for a sample with a depth range of 0-500nm, point defect data for 100 depth layers are acquired. The defect density of each depth layer is calculated as follows: within a 100nm × 100nm scanning area, the number of lattice defects is counted and divided by the scanning volume (100nm × 100nm × 5nm) to obtain the defect density value, expressed in units of defects per nm. 3 The difference in defect density between adjacent depth layers is obtained by subtracting the defect densities of two adjacent layers. The constructed residual defect depth distribution map is plotted with depth on the x-axis (nm) and defect density on the y-axis (numbers / nm). 3 This generates a continuous defect density distribution curve, and the defect density values ​​are labeled on the graph. In a specific example, the measured defect density of the surface layer (0-5nm) is 2.4 × 10⁻⁶. -3 Units / nm 3 The 5-10nm layer has a thickness of 2.1×10⁻⁶. -3 Units / nm 3 And so on, until the deepest layer.

[0128] When comparing the defect density values ​​in the residual defect depth distribution map with a preset defect density threshold, the preset defect density threshold is set to 1.5 × 10⁻⁶. -3 The defect density was found to be 1.7 × 10⁻⁶ per nm³ in the depth range of 75-90 nm. Through layer-by-layer comparison, it was discovered that the defect density was 1.7 × 10⁻⁶ per nm³. -3 1.9×10 -3 and 1.8×10 -3 The number of defects per nm³ all exceeded the preset threshold; in the depth range of 210-225 nm, the defect density was 1.6 × 10⁻⁶. -3 The number of defects per nm³ also exceeded the preset threshold. After determining the locations of these depth layers exceeding the threshold, high-resolution transmission electron microscopy was used to perform fine scanning on these regions to extract defect type characteristics. Interstitial defects were mainly observed in the 75-90 nm range, characterized by extra atoms between the lattice; vacancy defects were mainly observed in the 210-225 nm range, characterized by missing lattice points. Based on this information, the target depth layer compensation sequence was formed as: [{75-90 nm, interstitial type}, {210-225 nm, vacancy type}].

[0129] When determining the correspondence between lattice distortion values ​​and heat treatment temperatures based on the defect type characteristics in the target depth layer compensation sequence, a defect type-specific annealing behavior model is used. For interstitial defects, the correspondence between the lattice distortion value ε (in %) and the compensation heat treatment temperature T (in °C) is as follows: when ε is in the range of 0.5%-1.0%, the corresponding T is 350-400 °C; when ε is in the range of 1.0%-1.5%, the corresponding T is 400-450 °C; and when ε is in the range of 1.5%-2.0%, the corresponding T is 450-500 °C. For vacancy defects, when ε is in the range of 0.5%-1.0%, the corresponding T is 450-500 °C; when ε is in the range of 1.0%-1.5%, the corresponding T is 500-550 °C; and when ε is in the range of 1.5%-2.0%, the corresponding T is 550-600 °C. High-resolution TEM measurements revealed that interstitial defects in the 75-90 nm range caused a lattice distortion of 1.3%, thus determining the compensation heat treatment temperature for this region to be 430℃; vacancy defects in the 210-225 nm range caused a lattice distortion of 0.8%, determining the compensation heat treatment temperature for this region to be 480℃. The resulting compensation heat treatment temperature parameters are: [{75-90 nm, 430℃}, {210-225 nm, 480℃}].

[0130] When performing compensated heat treatment according to the compensated heat treatment temperature parameters, a precise and controllable laser scanning heat treatment device is used. This device can achieve temperature accuracy control of ±5℃ and depth accuracy control of ±2nm. During the heat treatment process, the laser power density is adjusted to 25W / cm². 2 The scanning speed was 0.5 mm / s. Heat treatment was performed at 430℃ for 30 seconds in the 75-90 nm depth region and at 480℃ for 45 seconds in the 210-225 nm depth region. After compensation heat treatment, the defect density of the target depth layer was measured again using STEM. The measurement results showed that the defect density in the 75-90 nm region decreased from the original average of 1.8 × 10⁻⁶. -3 Units / nm 3 Reduced to 1.2×10 -3 Units / nm 3 The defect density change is -0.6 × 10⁻⁶. -3 Units / nm 3 The defect density in the 210-225nm region is 1.6 × 10⁻⁶. -3 Units / nm 3 Reduced to 1.1×10 -3 Units / nm 3 The defect density change is -0.5 × 10⁻⁶. -3 Units / nm 3 .

[0131] When constructing the updated segmented rapid thermal annealing (SRA) curve by combining the defect density change with the temperature parameters of the original SRA curve, the original SRA curve was as follows: 0-100 nm depth region corresponding to a heat treatment temperature of 400℃ and a holding time of 25 seconds; 100-200 nm depth region corresponding to a heat treatment temperature of 450℃ and a holding time of 30 seconds; 200-300 nm depth region corresponding to a heat treatment temperature of 500℃ and a holding time of 35 seconds. Based on the defect density change during the compensation heat treatment, the original SRA curve was corrected: for the 75-90 nm depth region, the defect density decreased by 0.6 × 10⁻⁶. -3 Units / nm 3 The heat treatment temperature was adjusted from 400℃ to 380℃; for the 210-225nm depth region, the defect density decreased by 0.5×10⁻⁶. -3 Units / nm 3 The heat treatment temperature was adjusted from 500℃ to 475℃. The final updated segmented rapid thermal annealing curves are as follows: 0-75nm depth region corresponds to 380℃, held for 25 seconds; 75-90nm depth region corresponds to 380℃, held for 20 seconds; 90-200nm depth region corresponds to 450℃, held for 30 seconds; 200-210nm depth region corresponds to 475℃, held for 35 seconds; 210-225nm depth region corresponds to 475℃, held for 30 seconds; 225-300nm depth region corresponds to 500℃, held for 35 seconds. Verification shows that the updated segmented rapid thermal annealing curves ensure that the defect density at each depth layer is below the preset threshold of 1.5×10⁻³ defects / nm. 3 Reaching 1.0×10 -3 Units / nm 3 The following meet the technical requirements for high-quality crystal preparation.

[0132] In one optional implementation, based on the defect type characteristics in the target depth layer compensation sequence, the correspondence between the lattice structure distortion value and the heat treatment temperature is determined, and the compensation heat treatment temperature parameters are generated, including:

[0133] Spatial location information of defect type features in the target depth layer compensation sequence is extracted, the position of lattice atoms in the target depth layer is scanned, the real-time coordinate information of lattice atoms is recorded, the displacement distance of lattice atoms relative to standard lattice points is measured, and lattice atom displacement distribution data is generated.

[0134] Extract the displacement vectors between adjacent atoms from the lattice atomic displacement distribution data, construct a displacement vector group, determine the stress direction in the displacement vector group, obtain the stress intensity value, and form the stress distribution field of the target depth layer;

[0135] Based on the stress distribution field, identify the spatial channel for stress propagation, measure the energy accumulation state of the stress field along the spatial channel, extract the spatial coordinates of the energy accumulation points, and establish the distribution sequence of the energy accumulation points.

[0136] By combining the stress intensity values ​​with the distribution sequence of energy accumulation points, the spatial evolution law of lattice structure distortion is determined, the heat treatment temperature range is divided, the correspondence between lattice structure distortion values ​​and heat treatment temperatures is established, and compensation heat treatment temperature parameters are generated.

[0137] In one specific implementation, the process begins by extracting the spatial location information of defect type features from the compensation sequence of the target depth layer. A high-precision X-ray diffractometer is used to scan the positions of lattice atoms within the target depth layer, with a scanning resolution set to 0.01 nanometers, covering a three-dimensional spatial region of 100 μm × 100 μm × 10 μm. During the scanning process, the real-time coordinate information of the lattice atoms is recorded every 0.5 nanometers, forming a sequence containing approximately 4 × 10⁻⁶ lattice atoms. 6 A three-dimensional coordinate matrix of data points. For the silicon crystal sample, the theoretical coordinate values ​​of standard lattice points are pre-stored in the system database. By comparing the difference between the actual measured coordinates and the theoretical coordinates, the displacement distance of each lattice atom is calculated. For example, at a target depth of 5 μm, the average displacement of lattice atoms was detected to be 0.043 nm, with the maximum displacement reaching 0.187 nm, located at coordinates (43.26 μm, 57.81 μm, 5.00 μm). These displacement data are summarized to generate lattice atom displacement distribution data, which is displayed in the form of a heatmap. The area with a displacement greater than 0.1 nm is marked as a high-risk area, covering approximately 12% of the scanned area.

[0138] Further, displacement vectors between adjacent atoms are extracted from the lattice atomic displacement distribution data. Taking each lattice point as the center, the displacement differences between adjacent atoms within a 3.5 Å radius are calculated to generate displacement vectors. For each displacement vector, the system records its magnitude and orientation information, forming a vector containing approximately 1.2 × 10⁻⁶ atoms. 7 The system generates a group of displacement vectors. By statistically analyzing the dominant directions of the displacement vectors, the system determines the principal direction of stress. In the experimental samples, the principal stress direction was detected to be the

[110] crystal orientation, accounting for 73% of the total number of displacement vectors. At the same time, the system calculates the stress intensity based on the amplitude of the displacement vectors, and applies the elastic theory model to convert the displacement into stress values, forming a stress distribution field at the target depth layer. In the tested silicon crystal samples, the stress field intensity is distributed in the range of 0.5-3.7 GPa, with an average value of 1.82 GPa. The highest stress point is located at coordinates (45.33 μm, 58.67 μm, 5.12 μm), with a value of 3.68 GPa.

[0139] Based on the generated stress distribution field, spatial channels for stress propagation were identified. A stress gradient tracing algorithm was employed, starting from a high-stress region and searching along the direction of the slowest stress decrease. Paths with stress values ​​exceeding 1.5 GPa and continuously extending beyond 5 μm were marked as stress propagation channels. Seven main stress propagation channels were identified in the samples, with the longest channel reaching 17.3 μm. Along these spatial channels, the system sampled every 0.5 μm to measure the energy accumulation state of the stress field. Energy calculation was based on an elastic potential energy model, combining local stress values ​​with lattice deformation. When the local energy density exceeded 3.2 × 10⁻⁶, the energy accumulation was calculated. -19 J / nm 3 At that time, the point was marked as an energy accumulation point. The system extracted the spatial coordinates of these energy accumulation points, detecting a total of 43 energy accumulation points in the sample, mainly distributed in the depth range of 4.8-5.4 μm, and established a distribution sequence of energy accumulation points. Among them, the highest energy accumulation point is located at (46.25 μm, 58.92 μm, 5.18 μm), with an energy density of 4.7 × 10⁻⁶. -19 J / nm 3 .

[0140] By combining the distribution sequence of stress intensity values ​​with energy accumulation points, the spatial evolution law of lattice structure distortion was determined. Using a spatial clustering algorithm, the energy accumulation points were divided into five main distortion regions, and the average stress value and energy density of each region were calculated. Experimental data show that when the stress intensity is in the range of 1.5-2.5 GPa, lattice distortion develops linearly; when the stress exceeds 2.5 GPa, lattice distortion increases exponentially. Based on this law, the heat treatment temperature range was systematically divided: for the mildly distorted region (stress < 1.8 GPa), the heat treatment temperature was set at 450-500℃; for the moderately distorted region (stress 1.8-2.5 GPa), the heat treatment temperature was set at 500-580℃; and for the severely distorted region (stress > 2.5 GPa), the heat treatment temperature was set at 580-650℃. A lookup table is constructed to correlate lattice distortion values ​​with heat treatment temperatures. For example, when the lattice distortion value is 0.043 nm, the optimal heat treatment temperature is 482 °C; when the distortion value reaches 0.187 nm, the optimal heat treatment temperature is 603 °C. The system then generates compensated heat treatment temperature parameters, including the temperature value, duration, and heating rate. For the test sample, the system generates the following heat treatment scheme: heating to 540 °C at a heating rate of 2.5 °C / min, holding for 45 minutes, and then cooling to room temperature at a rate of 1.8 °C / min.

[0141] The indium antimonide chip beryllium ion implantation and rapid thermal annealing damage repair system of this invention includes:

[0142] The ion implantation unit is used to implant beryllium ions into the indium antimonide chip, forming a beryllium ion doped region on the surface of the indium antimonide chip, while simultaneously causing lattice displacement damage.

[0143] The damage detection unit is used to perform depth-direction damage detection on the implanted indium antimonide chip to obtain damage distribution data characterizing the point defect concentration and dislocation density of each depth layer.

[0144] The annealing planning unit is used to determine the depth layer structure of the beryllium ion doped region based on damage distribution data through gradient analysis, generate a defect distribution map, determine the defect type and calculate the annihilation activation energy, establish a mapping table based on the annihilation activation energy to determine the temperature stage sequence, and generate a segmented rapid thermal annealing curve by combining the energy interval.

[0145] The heat treatment unit is used to perform heat treatment according to the segmented rapid heat annealing curve, so that the point defects in each depth layer migrate to the interstitial space and recombine with the vacancy to achieve the staged repair state.

[0146] The process monitoring unit is used to monitor the surface resistivity and reflectivity signals of the beryllium ion-doped region during heat treatment, construct the defect annihilation process curve, identify the defect elimination node, and adjust the termination time of the temperature stage.

[0147] The defect compensation unit is used to collect the residual defect density of each depth layer in the repair stage, apply compensation heat treatment to depth layers that exceed the preset defect density threshold, and update the segmented rapid thermal annealing curve based on the defect density change after compensation heat treatment.

[0148] A third aspect of the present invention provides an electronic device, comprising:

[0149] processor;

[0150] Memory used to store processor-executable instructions;

[0151] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0152] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0153] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; 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 or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for repairing damage to indium antimonide chips via beryllium ion implantation and rapid thermal annealing, characterized in that, The application relates to a method for repairing a beryllium ion doped indium antimonide chip. The method comprises the following steps: performing beryllium ion implantation on an indium antimonide chip to form a beryllium ion doped region on the surface of the indium antimonide chip and simultaneously generate a lattice displacement damage; Performing deep direction damage detection on the implanted indium antimonide chip to obtain damage distribution data representing point defect concentration and dislocation density of each depth layer; Based on the damage distribution data, the depth layer structure of the beryllium ion doped region is determined through gradient analysis, a defect distribution map is generated, the defect type is determined and the annihilation activation energy is calculated, a mapping table is established according to the annihilation activation energy to determine the temperature stage sequence, and a segmented rapid thermal annealing curve is generated in combination with the energy interval; According to the segmented rapid thermal annealing curve, heat treatment is performed to make the point defects of each depth layer migrate to the interstitial lattice and be annihilated with vacancies to obtain a stage repair state; During the heat treatment process, the surface resistivity signal and the reflectivity signal of the beryllium ion doped region are monitored to construct a defect annihilation process curve, identify a defect elimination node and adjust the termination time of the temperature stage; The residual defect density of each depth layer in the stage repair state is collected, compensation heat treatment is applied to the depth layer whose defect density exceeds a preset threshold, and the segmented rapid thermal annealing curve is updated based on the defect density change after the compensation heat treatment.

2. The method of claim 1, wherein, Based on the damage distribution data, the depth layer structure of the beryllium ion doped region is determined through gradient analysis, a defect distribution map is generated, the defect type is determined and the annihilation activation energy is calculated, a mapping table is established according to the annihilation activation energy to determine the temperature stage sequence, and a segmented rapid thermal annealing curve is generated in combination with the energy interval. The method comprises the following steps: scanning and sampling the damage distribution data according to a preset interval along the depth direction, calculating the point defect concentration gradient and the dislocation density gradient of adjacent sampling points, determining the depth demarcation point by the gradient change rate exceeding a background reference value, determining the depth layer division structure by measuring the difference degree of damage characteristics between adjacent depth layers; According to the depth layer division structure, grid subdivision is performed, the distribution values of the point defect concentration and the dislocation density in the grid are calculated according to a preset grid size, the concentration gradient change rule between adjacent grids is measured, and a depth layer defect distribution map is generated in combination with the grid position distribution characteristics; The position correlation data of the point defects and the dislocations are extracted from the depth layer defect distribution map, the aggregation degree and the dispersion state of the defect groups are obtained, the variation rule of the point defect concentration and the dislocation density in the depth direction is measured, a defect feature parameter set is constructed, and the defect type in the depth layer is determined; The defect type is corresponded to the lattice structure, the lattice distortion degree and the bonding state are measured, a mapping relationship between the defect structure characteristics and the annihilation activation energy is established, and the annihilation activation energy value of the depth layer is determined.

3. The method of claim 1, wherein, According to the annihilation activation energy, a mapping table is established to determine the temperature stage sequence, and a segmented rapid thermal annealing curve is generated in combination with the energy interval. The method comprises the following steps: grouping and sorting all the depth layer annihilation activation energies according to the numerical value, establishing a mapping table of the annihilation activation energy and the depth layer, determining the number and hierarchical structure of the temperature stages according to the mapping table; Based on the grouping and sorting result of the annihilation activation energy, the energy interval between adjacent annihilation activation energies is calculated, the energy interval is taken as the division basis of the temperature stage, and a temperature stage sequence is constructed; For each temperature stage, the annealing temperature parameter is set according to the annihilation activation energy of the corresponding depth layer, the heating rate and the holding time are determined in combination with the kinetics characteristics of the defect annihilation, and a temperature-time curve is generated. The temperature-time curve of each temperature stage is connected in order of the annihilation activation energy from low to high, a transition interval between temperature stages is added, and a complete segmented rapid thermal annealing curve is constructed.

4. The method of claim 1, wherein, During the heat treatment process, the surface resistivity signal and reflectivity signal of the beryllium ion doped region are monitored, a defect annihilation process curve is constructed, defect elimination nodes are identified, and the termination time of the temperature stage is adjusted, including: During the heat treatment process, a multi-channel detector array is used to scan the surface of the beryllium ion doped region in real time, the surface resistivity signal and reflectivity signal are synchronously collected, a two-parameter time sequence acquisition matrix is established, continuous change data of the surface parameters are obtained, and the spatio-temporal distribution characteristics of the parameter changes are formed; Based on the two-parameter time sequence acquisition matrix, extreme points in the time sequence data of the resistivity and the time sequence data of the reflectivity are extracted, turning points of the parameter change trajectory are determined, and a defect annihilation process curve is constructed; The defect annihilation process curve is subjected to multi-scale segmented slope calculation, the positions and change amplitudes of the slope mutation points are identified, the slope mutation points are marked as defect elimination nodes in combination with the spatio-temporal distribution characteristics, a defect elimination judgment criterion is established, and the termination time of the current temperature stage is adjusted in real time according to the occurrence time and distribution law of the defect elimination nodes in combination with the defect elimination judgment criterion, a mapping relationship between defect elimination and temperature control is established, and dynamic regulation and control of the heat treatment process is realized. Based on the two-parameter time sequence acquisition matrix, extreme points in the time sequence data of the resistivity and the time sequence data of the reflectivity are extracted, turning points of the parameter change trajectory are determined, and a defect annihilation process curve is constructed, including:

5. The method of claim 4, wherein, The time sequence data of the resistivity and the time sequence data of the reflectivity are segmented according to a preset time interval, the maximum and minimum values of each segment are measured, the parameter change amount is calculated based on the corresponding intermediate value as a base point, and an offset data set representing parameter fluctuation is obtained; Extreme points in the offset data set are extracted, extreme time points are determined, the time interval of the extreme time points is calculated, the extreme points are divided into a parameter synchronous interval and an asynchronous interval according to a preset time threshold, and a time sequence mapping table of parameter evolution is generated; In the parameter synchronous interval, the resistivity and the reflectivity are projected onto a two-dimensional coordinate plane to determine a parameter change trajectory, turning points of the trajectory are selected according to a preset curvature threshold, time sequences corresponding to the turning points are extracted, and time markers of defect state conversion are obtained; Taking the time markers as division points, resistivity and reflectivity data in a preset neighborhood of each division point are extracted, endpoint values of adjacent data segments are connected using a curve reconstruction method, a plurality of characteristic curves are formed, and a defect annihilation process curve is obtained by combining the characteristic curves. The residual defect density of each depth layer in the repair state is collected, compensation heat treatment is applied to depth layers whose defect density exceeds a preset defect density threshold, and the defect density change after the compensation heat treatment is used to update the segmented rapid thermal annealing curve, including:

6. The method of claim 1, wherein, Point defect spatial distribution data are obtained by scanning and measuring each depth layer along the depth direction, defect density difference values between adjacent depth layers are calculated, a residual defect depth distribution map is constructed, and defect density values are recorded; ​ The defect density values in the residual defect depth distribution map are compared with a preset defect density threshold value to determine a depth layer position that exceeds the preset defect density threshold value, defect type features at the depth layer position are extracted to form a target depth layer compensation sequence; According to the defect type features in the target depth layer compensation sequence, a corresponding relationship between the lattice structure distortion value and the heat treatment temperature is determined to generate a compensation heat treatment temperature parameter; According to the compensation heat treatment temperature parameter, a compensation heat treatment is performed, the defect density of the target depth layer after the compensation heat treatment is measured, the defect density change amount caused by the compensation heat treatment is calculated, and the defect density change amount is combined with the temperature parameter of the original segmented rapid thermal annealing curve to construct an updated segmented rapid thermal annealing curve.

7. The method of claim 6, wherein, According to the defect type features in the target depth layer compensation sequence, a corresponding relationship between the lattice structure distortion value and the heat treatment temperature is determined to generate a compensation heat treatment temperature parameter, which includes: Extracting the spatial position information of the defect type features in the target depth layer compensation sequence, scanning the positions of the lattice atoms in the target depth layer, recording the real-time coordinate information of the lattice atoms, measuring the displacement distance of the lattice atoms relative to the standard lattice points, and generating lattice atom displacement distribution data; Extracting the displacement vectors between adjacent atoms in the lattice atom displacement distribution data, constructing a displacement vector group, determining the stress action direction in the displacement vector group, obtaining the stress action intensity value, and forming a stress distribution field of the target depth layer; According to the stress distribution field, the spatial channel of stress propagation is identified, the energy accumulation state of the stress field along the spatial channel is measured, the spatial coordinates of the energy accumulation points are extracted, and a distribution sequence of the energy accumulation points is established; Combining the stress action intensity value with the distribution sequence of the energy accumulation points, determining the spatial evolution rule of the lattice structure distortion, dividing the heat treatment temperature action interval, establishing the corresponding relationship between the lattice structure distortion value and the heat treatment temperature, and generating the compensation heat treatment temperature parameter.

8. An indium antimonide wafer beryllium ion implantation and rapid thermal anneal damage repair system for implementing the method of any of the preceding claims 1-7, characterized in that, It includes: An ion implantation unit for implanting beryllium ions into an indium antimonide chip to form a beryllium ion doped region on the surface of the indium antimonide chip and simultaneously generate lattice displacement damage; A damage detection unit for detecting the damage of the implanted indium antimonide chip in the depth direction to obtain damage distribution data representing the point defect concentration and dislocation density of each depth layer; An annealing planning unit for determining the depth layer structure of the beryllium ion doped region based on the damage distribution data through gradient analysis, generating a defect distribution map, determining the defect type and calculating the annihilation activation energy, establishing a mapping table according to the annihilation activation energy to determine a temperature stage sequence, and generating a segmented rapid thermal annealing curve in combination with the energy interval; A heat treatment unit for performing heat treatment according to the segmented rapid thermal annealing curve to make the point defects in each depth layer migrate to the lattice gap and annihilate with vacancies to obtain a stage repair state; A process monitoring unit for monitoring the surface resistivity signal and reflectivity signal of the beryllium ion doped region during the heat treatment process, constructing a defect annihilation process curve, identifying a defect elimination node, and adjusting the termination time of the temperature stage; A defect compensation unit is configured to collect residual defect density of each depth layer in the repair state in the collecting stage, apply compensation heat treatment to the depth layer exceeding the preset defect density threshold, and update the segmented rapid thermal annealing curve based on the defect density change after the compensation heat treatment.

9. An electronic device, comprising: Comprise: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions, when executed by the processor, implement the method of any one of claims 1 to 7.

Images