A method for detecting self-interstitial dislocations in heavily arsenic-doped substrates

Abstract

The application discloses a method for detecting self-interstitial atom type dislocations in heavily arsenic-doped substrates. The method is based on the fact that self-interstitial silicon atoms in heavily arsenic-doped substrates have a thermal activation threshold effect during high-temperature epitaxy. When the effective thermal treatment time of epitaxy reaches a critical aggregation time τ c , the dispersed point defect state self-interstitial atoms can be excited and aggregated into an extended dislocation state that can be detected. The method comprises the following steps: pretreating the heavily arsenic-doped substrate; performing epitaxial growth under process conditions that satisfy the thermal activation threshold, so that the latent self-interstitial atom defects are made dominant; identifying characteristic signals by using SP1 surface scattering scanning; and optionally performing microscope observation, chemical corrosion verification and substrate trace detection. The application realizes effective detection of latent self-interstitial atom type dislocations in heavily arsenic-doped substrates, provides a quantifiable process window, and has the advantages of high detection sensitivity, good reliability and easy industrial integration.

Description

Technical Field

[0001] This invention relates to the field of semiconductor material testing technology, specifically to a method for detecting interstitial atomic dislocations in heavily arsenic-doped substrates. Background Technology

[0002] Heavy arsenic-doped silicon substrate (N + Arsenic-doped silicon single crystals are a key material for manufacturing power semiconductor devices (such as IGBTs, VDMOS, and Schottky diodes) due to their extremely low resistivity (typically 0.001 Ω·cm to 0.005 Ω·cm). The crystal quality of the substrate directly determines the performance and reliability of the epitaxial layer and the final device.

[0003] During the growth of heavily arsenic-doped silicon single crystals, the difference in atomic radius between arsenic and silicon leads to localized shrinkage stress in the lattice when a large number of arsenic atoms substitute for silicon atoms. To compensate for this stress, an excess of self-interstitial silicon atoms is generated in the crystal. These self-interstitial silicon atoms tend to aggregate and nucleate during crystal cooling or subsequent thermal processes. When the aggregation reaches a critical size, it forms extended defects such as dislocation loops or dislocation networks. These defects severely degrade device performance, such as increasing leakage current and reducing breakdown voltage. Therefore, accurate detection of self-interstitial atomic defects in heavily arsenic-doped substrates is crucial for quality control.

[0004] However, these defects are highly concealed in the substrate state: interstitial atoms often exist in the substrate as dispersed point defects, not yet aggregated to form extended defects that can be effectively detected by conventional methods (such as surface light scattering scanning SP1 and ordinary microscopy). In other words, they are "potential defects" that are difficult to detect with existing conventional detection techniques. Direct etching observation using sampling methods is also difficult because, for heavily arsenic-doped substrates, traditional selective etching methods such as Secco and Wright etching solutions alter the electrochemical properties of silicon due to high concentrations of arsenic doping, resulting in blurred dislocation etch pit morphologies and poor contrast, making effective detection difficult. Currently, only X-ray morphology (XRT) can detect substrate lattice distortion, but this method is expensive, complex to operate, and has limited sensitivity to interstitial atom clusters in their early stages of formation and not yet fully aggregated.

[0005] Several methods for detecting substrate defects using epitaxial layers have been disclosed. For example, Japanese Patent Document JP2006108151A discloses a method for manufacturing a silicon epitaxial wafer, which mentions that by growing an "inspection silicon epitaxial layer" on the substrate, the originally tiny crystal defects in the substrate are enlarged in size due to the epitaxial growth mechanism, thus becoming easier to observe and detect. Similarly, Chinese Patent Document CN114018930A also discloses a method for detecting native defects in silicon crystals, which uses an epitaxial layer to extend native defects into epitaxial defects, and then identifies them through dual-channel light scattering.

[0006] However, both JP2006108151A and CN114018930A follow the same technical path: "epitaxygdral defect extension / amplification." This method is effective only if extended defects with "nuclei" (such as native dislocations, oxygen precipitates, etc.) already exist in the substrate. The epitaxial layer merely amplifies and reveals these "existing" defects. However, for "potential" interstitial atoms in dispersed point defect states in heavily arsenic-doped substrates, since they have not yet formed extendable defect nuclei, the aforementioned "epitaxygdiation" method cannot convert them into detectable signals.

[0007] In summary, existing technologies lack effective methods for making the "latent" self-interstitial atoms in heavily arsenic-doped substrates, which exist as scattered point defects and cannot be directly detected, explicit. Current understanding in this field primarily relies on epitaxial layers to "amplify" and detect existing defects. Therefore, establishing a method to effectively excite and accurately detect the unique, latent self-interstitial atom defects in heavily arsenic-doped substrates is a pressing technical problem that needs to be solved. Summary of the Invention

[0008] This invention aims to overcome the shortcomings of the prior art and provide a method for detecting interstitial atomic dislocations in heavily arsenic-doped substrates, thereby solving the technical problem that the prior art cannot effectively make explicit the detection of potential interstitial atoms existing in the form of scattered point defects in heavily arsenic-doped substrates.

[0009] During their long-term research on the detection technology of interstitial atomic dislocations in heavily arsenic-doped substrates, the inventors unexpectedly discovered that a distinct "thermal activation threshold" exists when performing high-temperature epitaxial processing on heavily arsenic-doped silicon substrates. Specifically, when the thermal budget of the epitaxial process (determined by both heat treatment temperature and time) is below a certain critical value, the interstitial silicon atoms existing as scattered point defects in the substrate cannot obtain sufficient energy and time for long-range diffusion and effective aggregation. At this point, conventional detection methods such as SP1 surface scanning and chemical etching cannot detect defect signals related to interstitial atoms on the epitaxial layer surface; that is, the defects are in an "undetectable latent state." When the thermal budget of the epitaxial process reaches or exceeds this "thermal activation threshold," the originally scattered interstitial silicon atoms are fully activated, undergoing significant diffusion, nucleation, and aggregation, ultimately forming extended defects such as dislocation loops or dislocation clusters. These extended defects project characteristic signals with specific spatial distribution and size ranges onto the epitaxial layer surface, thus allowing for effective detection by conventional methods such as SP1 scanning and optical microscopy.

[0010] This discovery shows that "potential" self-interstitial atomic point defects can be "excited" into "detectable" extended defects by precisely controlling the epitaxial thermal process, and that the excitation process has a quantifiable and repeatable threshold condition, which this invention refers to as the "epitaxythermal activation threshold effect".

[0011] Based on the above findings, the inventors further analyzed the underlying physical mechanism, and theoretically analyzed and formed a quantifiable detection scheme.

[0012] (1) Initial state of interstitial atoms in heavily arsenic-doped substrates: During the growth of heavily arsenic-doped silicon single crystals, due to the larger size of arsenic atoms (radius approximately 1.19 Å) compared to silicon atoms (radius approximately 1.11 Å), a large number of substituted arsenic atoms introduce local compressive stress. To relax this stress, the lattice generates excess interstitial silicon atoms. The mobility of these interstitial silicon atoms decreases sharply during crystal cooling, and most of them are "frozen" in the substrate as isolated point defects or metastable small clusters (size only a few nanometers), without forming extended defects that can be detected by conventional optical methods. Their average spacing is related to the arsenic doping concentration: For heavily arsenic-doped substrates with resistivity from 0.001 Ω·cm to 0.005 Ω·cm (arsenic concentration approximately 10 Å), 19 cm -3 Up to 10 20 cm -3 The average estimated spacing between interstitial atoms is approximately 5 μm to 15 μm.

[0013] (2) Thermally activated diffusion at high temperature in epitaxy When the substrate is heated to the epitaxial temperature T (typically 1050°C to 1150°C), the interstitial silicon atoms gain sufficient thermal energy, and their diffusion coefficient D increases significantly. According to the Arrhenius relation, the diffusion coefficient of interstitial atoms in silicon satisfies:

[0014] Where D0 is the pre-exponential factor (approximately 0.1 cm). 2 / s to 1cm 2 / s), E a The activation energy (increased to approximately 2.3 eV to 2.5 eV under heavily doped arsenic conditions due to the electrostatic pinning effect of arsenic impurities and the binding effect of arsenic-interstitial atom complexes) is given by k, where k is the Boltzmann constant, and T is the epitaxial temperature, taken as its absolute value. At epitaxial temperatures of 1050–1150 °C, the diffusion coefficient can reach 10-1. -8 cm 2 / s to 10 -7 cm 2 On the order of / s.

[0015] The diffusion length L of interstitial atoms and the effective heat treatment time t eff The relationship is:

[0016] (3) Critical aggregation time model (quantification of thermal activation threshold) For dispersed interstitial atoms to effectively collide, nucleate, and aggregate into detectable dislocation loops, their diffusion length L must be at least on the same order of magnitude as their initial average spacing, i.e., reaching the critical diffusion length Lc (approximately 8 μm to 10 μm). From this, the critical aggregation time τ can be derived. c :

[0017] Reduce Lc≈8μm to 10μm, D≈5×10 −8 cm 2 Substituting / s (1100℃) into the equation, we can calculate τ. c The time is approximately 1280 to 2000 seconds, or about 25 to 33 minutes. Under a typical growth rate of 2 μm / min, the corresponding epitaxial layer thickness is approximately 42 μm to 66 μm.

[0018] (4) Threshold jump characteristics Experiments have shown that when t eff <τ c At that time, the diffusion of interstitial atoms is insufficient, the size of the aggregated state does not reach the detectable threshold, and the number of defects detected by SP1 is extremely small or even zero; when t eff Approaching and exceeding τ cAt that time, the number of defects detected increased by an order of magnitude. This characteristic of "threshold jump" rather than "gradual change" is the core of this invention.

[0019] Based on the above findings and mechanisms, this invention provides a method for detecting interstitial atomic dislocations in heavily arsenic-doped substrates based on the epitaxial thermal activation threshold effect, specifically including the following steps: Step 1, Substrate Pretreatment: The heavily arsenic-doped silicon substrate to be tested undergoes standard cleaning to remove surface particles, metal ion contaminants, and the native oxide layer. More specifically, the standard cleaning steps include: SC-1 cleaning to remove particulate contaminants; SC-2 cleaning to remove metal ion contaminants; and finally, immersion in dilute HF solution to remove the native oxide layer, followed by nitrogen drying.

[0020] Step 2, Thermally Activated Epitaxial Growth: The pretreated heavily arsenic-doped substrate is placed in a CVD epitaxial reaction chamber to grow a lightly doped silicon epitaxial layer. The process conditions must be set within the thermal activation threshold window, specifically including: an epitaxial temperature of 1050℃ to 1150℃, preferably 1080℃ to 1120℃; a growth rate controlled at 1μm / min to 3μm / min, preferably 2μm / min; and an effective thermal treatment time t. eff ≥τ c In practical operation, for example, taking a temperature of 1100℃ and a growth rate of 2 μm / min as an example, the epitaxial growth time is ≥50 minutes, preferably 60 to 90 minutes; the corresponding epitaxial layer thickness is ≥100 μm, preferably 120 μm to 180 μm; when t eff Within 25 minutes, interstitial atoms cannot form detectable dislocations, and this condition should be avoided for detection.

[0021] This step differs from the epitaxial layer in existing technologies, which is "only used to amplify existing defects": the thermal process in this step also undertakes the specific function of "exciting the transformation of potential point defects in the substrate into extended defects", which is an irreplaceable core link in the entire detection scheme.

[0022] Step 3, First-level detection—SP1 surface scattering scan for feature signal identification: Using a surface particle detector such as the KLA-Tencor SP1 series, a full-wave scan of the epitaxial wafer surface is performed to obtain information on the spatial distribution, size, and quantity of defects. The identification feature signals from interstitial atomic dislocations in the substrate include: (a) Signals with a central cluster distribution detectable in both DCN (narrow channel) and DCO (wide channel) modes; (b) Signal sizes concentrated in the range of 50 nm to 2 μm. These characteristic signals are direct observational counterparts of dislocations formed by the thermal activation-induced aggregation of interstitial atoms, and can be distinguished from other defects such as general particulate contamination and epitaxial stacking faults by spatial distribution and dual-mode comparison.

[0023] Step 4, Second-level detection – Microscopic morphology observation: For epitaxial wafers with characteristic signals detected by SP1, surface morphology observation can be further performed using an optical microscope and / or electron beam defect verification (eDR) to verify whether there are convex dislocation lines or irregular protrusion groups corresponding to the SP1 signal.

[0024] Step 5, Third-Level Detection – Specific Chemical Etching: Selective chemical etching is performed on the epitaxial wafers requiring further confirmation of dislocation properties. A selective chemical etching solution is used, preferably chromic acid, with a formulation of a 0.5 mol / L to 1.0 mol / L CrO3 aqueous solution. The epitaxial wafers are subjected to static immersion etching for 3 to 10 minutes, preferably 5 minutes. After etching, thermally activated interstitial atomic dislocations exhibit clustered elliptical or quasi-triangular etching pits, their spatial distribution corresponding one-to-one with the SP1 signal in Step 2.

[0025] Step Six, Level Four Inspection – Substrate Traceability and Defect Structure Confirmation: For cases where significant defects are found during epitaxial wafer inspection, further traceability verification of the corresponding substrate can be performed. Specifically, for substrates in the same batch that are confirmed to be interstitial atomic dislocation defects, traceability inspection is performed using X-ray morphology (XRT) and / or transmission electron microscopy (TEM) to directly observe the lattice distortion or dislocation core structure in the substrate, confirm that the nature of the defect is a dislocation formed by the aggregation of interstitial atoms, and complete the closed-loop analysis of the defect cause.

[0026] XRT inspection: X-ray morphology is used to inspect substrates of the same batch or adjacent numbers to confirm whether there are corresponding lattice distortions or dislocation signals in the substrate.

[0027] TEM analysis: High-resolution imaging and structural analysis were performed using transmission electron microscopy. The dislocation core structure was directly observed through inverse Fourier transform (IFFT) images to confirm the Burgers vector characteristics.

[0028] Compared with the prior art, the present invention has the following beneficial effects: For the first time, the explicit detection of potential self-interstitial atomic defects has been achieved: Based on the discovery of the "thermal activation threshold effect", this invention breaks through the limitation of existing technologies that can only detect "existing extended defects" and provides for the first time a complete technical solution to actively excite "potential" self-interstitial atoms existing in the form of scattered point defects in heavily arsenic-doped substrates into detectable extended defects and detect them.

[0029] Providing a quantifiable process window to ensure detection reliability: This invention not only qualitatively identifies the thermal activation effect, but also quantitatively reveals the "critical aggregation time τ". cThe threshold value and its theoretical calculation model are proposed, and the optimal process parameters (t) verified by experiments are given. eff (≥50 minutes, thickness ≥100μm). This quantifiable window makes the detection process reproducible and standardized, avoiding false negatives caused by insufficient thermal budget.

[0030] Compared to the existing technology's "epitaxical amplification" approach of "existing defects in the substrate," this invention does not rely on existing defect seeds but actively creates defect signals.

[0031] The detection features unique characteristics and strong anti-interference ability: thermally activated interstitial atomic dislocations on the epitaxial layer surface exhibit multi-dimensional characteristic signals such as "central cluster spatial distribution", "size range of 50nm-2μm", "detectable in both DCN and DCO modes", and "clustered elliptical / quasi-triangular corrosion pit morphology". The combination of these features is extremely difficult to imitate by other common defects (such as particle contamination, epitaxial stacking faults, and slip lines), resulting in a low false positive rate and high reliability of detection results.

[0032] The method boasts excellent process compatibility and is easy to promote in industry: the epitaxial growth, SP1 scanning, chemical etching, and optical microscopy observation employed in this invention are all mature technologies and standard equipment in the semiconductor manufacturing field, eliminating the need to purchase expensive or specialized testing instruments. XRT / TEM is used solely for traceability. This method can be directly integrated into the epitaxial process of power device production lines for sampling inspection, offering excellent economic efficiency and operability. Attached Figure Description

[0033] Figure 1 This is a schematic diagram illustrating the principle of the transformation of interstitial atoms in a heavily arsenic-doped substrate from point defect states to dislocation states during epitaxial thermal activation.

[0034] Figure 2 This is a scattering scan result of the SP1 surface of the epitaxial wafer after thermally activated epitaxial growth in Embodiment 1 of the present invention, wherein... Figure 2 In the image, 'a' represents the DCN mode scan diagram. Figure 2 In the image, b represents the DCO mode scan.

[0035] Figure 3 This is an optical microscope image of the epitaxial wafer in Embodiment 1 of the present invention.

[0036] Figure 4 This is a MAP image of the electron beam defect verification in the abnormal region of the epitaxial wafer in Embodiment 1 of the present invention.

[0037] Figure 5 This is an observation image of the epitaxial wafer after chromic acid etching in Embodiment 1 of the present invention, wherein... Figure 5 (a) and Figure 5(b) in the image shows optical microscope images at different magnifications.

[0038] Figure 6 This is an X-ray morphology (XRT) image of the corresponding batch of substrates in Embodiment 1 of the present invention.

[0039] Figure 7 This is a transmission electron microscope (TEM) image of the defect region in Embodiment 1 of the present invention, wherein... Figure 7 In the image, 'a' represents the TEM image. Figure 7 In the diagram, b represents the Burgers vector analysis plot of the inverse Fourier transform image.

[0040] Figure 8 This is a scattering scan result of the SP1 surface of the epitaxial wafer in Comparative Example 1 of the present invention, wherein... Figure 8 In the image, 'a' represents the DCN mode scan diagram. Figure 8 In the image, b represents the DCO mode scan.

[0041] Figure 9 This is an optical microscope image of the epitaxial wafer in Comparative Example 1 of the present invention.

[0042] Figure 10 This is a graph showing the correlation between the extrapolation time and the number of SP1 defects detected in Embodiment 2 of the present invention. Detailed Implementation

[0043] To make the present invention clearer, it will be described in further detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0044] This invention provides a method for detecting interstitial atomic dislocations in heavily arsenic-doped substrates based on the epitaxial thermal activation threshold effect. It utilizes the critical aggregation time τ of interstitial silicon atoms in the heavily arsenic-doped substrate during high-temperature epitaxy. c By increasing the effective heat treatment time t of the epitaxial layer eff Set to greater than or equal to τ c This process allows previously dispersed and undetectable interstitial atomic point defects to fully diffuse, nucleate, and aggregate to form detectable extended dislocation defects, which can then be identified and confirmed through multi-level detection methods.

[0045] like Figure 1 This is a schematic diagram illustrating the principle of the transformation of interstitial atoms in a heavily arsenic-doped substrate through "point defect state - nucleation state - dislocation state" during epitaxial thermal activation. As shown in the figure, in the initial state (e.g.... Figure 1 In (a) of the heavily arsenic-doped substrate, arsenic atoms enter the silicon lattice in a substitutional manner. Figure 1(Identified in purple) Substituted arsenic atoms cause local lattice distortion due to differences in atomic radius, thereby generating interstitial silicon atoms in the form of scattered point defects. Figure 1 (Identified in red). When the thermal budget of the epitaxial process is below the threshold, the atomic diffusion distance is limited, and interstitial silicon atoms cannot effectively aggregate; however, when the epitaxial thermal budget reaches or exceeds the critical threshold, atoms obtain sufficient thermal activation energy, the diffusion distance increases, they collide and aggregate, and eventually form detectable dislocation loops or dislocation clusters (e.g., ...). Figure 1 (b) in the middle.

[0046] The following detailed embodiments and comparative examples illustrate the implementation process and the technical effects achieved by the present invention.

[0047] Example 1 This embodiment describes the process of detecting interstitial atomic dislocations on a heavily arsenic-doped substrate according to the complete detection procedure of the present invention.

[0048] This embodiment uses an 8-inch (200 mm) heavily doped arsenic silicon substrate with a crystal orientation of (100) and a resistivity of 0.002 Ω·cm. The detection process is as follows.

[0049] Step 1, Substrate Pretreatment: The heavily arsenic-doped silicon substrate underwent the following cleaning steps in sequence: SC-1 cleaning: The substrate was placed in a mixed solution of NH4OH:H2O2:H2O = 1:1:5 (volume ratio) and cleaned at 75°C for 10 minutes to remove surface particulate contaminants. Then it was rinsed thoroughly with ultrapure water.

[0050] SC-2 cleaning: The substrate was placed in a mixed solution of HCl:H2O2:H2O = 1:1:5 (volume ratio) and cleaned at 75°C for 10 minutes to remove metal ion contaminants. Then it was rinsed thoroughly with ultrapure water.

[0051] Dilute HF treatment: Immerse the substrate in a dilute hydrofluoric acid solution of HF:H2O = 1:50 (volume ratio) for 30 seconds to remove the natural oxide layer on the surface; then dry it with nitrogen gas.

[0052] Step 2, Thermally Activated Epitaxial Growth: Epitaxial growth was performed using chemical vapor deposition (CVD). The process conditions were as follows: the silicon source gas was trichlorosilane (SiHCl3), the carrier gas was high-purity hydrogen with a purity ≥99.9999%, the doping gas was phosphine (PH3) dilution gas, and the epitaxial layer was controlled to be lightly doped N-type; the growth temperature was 1100℃; the growth rate was 2 μm / min; and the epitaxial time was 75 minutes. After growth, the epitaxial layer thickness was measured to be approximately 150 μm, and the resistivity was 12 Ω·cm.

[0053] According to the thermal activation threshold model of the present invention, the critical aggregation time τ is at 1100°C. c Approximately 25-30 minutes. In this embodiment, the effective heat treatment time t... eff It lasts 75 minutes, much longer than τ. c Therefore, the thermal activation condition is met. It is predicted that interstitial atoms in the substrate can fully diffuse, aggregate, and form detectable dislocation defects.

[0054] Step 3, First-level detection – SP1 surface scattering scan: The epitaxial wafer is scanned using a KLA-Tencor SP1 surface particle detector. Detection modes include DCN and DCO modes. DCN mode is a narrow channel, sensitive to small-angle scattered light; DCO mode is a wide channel, sensitive to large-angle scattered light.

[0055] Test results as follows Figure 2 As shown. Figure 2 In this context, 'a' represents the scan pattern in DCN mode. Figure 2 In this diagram, 'b' represents the scan image in DCO mode. From... Figure 2 It can be clearly observed that in the central region of the epitaxial wafer, obvious clustered defect signals were detected in both DCN and DCO modes (the area within the red box in the figure); and the spatial distribution of the defect signals in both modes is highly consistent, exhibiting centrosymmetry; the size of individual defect signals, statistically analyzed by SP1 software, is concentrated in the range of 50 nm to 2 μm. These characteristics fully conform to the characteristic signals of self-interstitial atomic dislocations of this invention, and are significantly different from conventional particulate contamination or epitaxial stacking faults. Conventional particulate contamination typically exhibits a random, discrete distribution, while epitaxial stacking faults (typically exhibit a line-like distribution with specific crystal orientations) are also present.

[0056] Step 4, Second-level inspection – Microscopic morphological observation: The epitaxial wafers found to have abnormalities in the SP1 inspection were observed under an optical microscope. The results are as follows: Figure 3 As shown. Figure 3 The surface morphology of the epitaxial wafer at 50X magnification is shown, clearly revealing irregularly distributed, raised defect features that correspond to the SP1 signal region. Further high-resolution imaging of key anomalous areas was performed using electron beam defect verification (eDR), with results as follows... Figure 4 As shown. eDR MAP ( Figure 4 The fine morphology of the defect is clearly shown, exhibiting a raised feature, which further verifies the existence of self-interstitial atomic dislocations.

[0057] Step 5, Third-level detection – Specific chemical etching display: Prepare a 0.75 mol / L chromic acid (CrO3) aqueous solution as the etching solution. Immerse the epitaxial wafer in the etching solution at room temperature (23±2℃) and perform static etching for 3 minutes. After etching, immediately rinse with plenty of ultrapure water and dry with nitrogen gas.

[0058] The observation results of the etched epitaxial wafer are as follows: Figure 5 As shown in the figure, the etched epitaxial wafer exhibits obvious defects in its central region under fluorescent light; these defects become even more apparent under spotlight illumination, appearing as diffuse reflection spots. Figure 5 Images (a) and (b) show the morphology of corrosion pits observed under an optical microscope at different magnifications. Clusters of elliptical or quasi-triangular corrosion pits can be clearly seen, their spatial distribution consistent with... Figure 2 The SP1 MAP signals shown correspond one-to-one. The morphology of this corrosion pit is a characteristic manifestation of dislocations formed by the aggregation of interstitial atoms, further confirming the nature of the defect.

[0059] Step Six, Level Four Inspection – Substrate Traceability and Defect Structure Confirmation: To confirm substrate quality from the source, adjacent numbered substrate wafers produced in the same batch as in Example 1 were directly subjected to X-ray morphology (XRT) inspection without epitaxial processing. The XRT inspection results are as follows: Figure 6 As shown. From Figure 6 As can be clearly seen in the XRT image, a significant lattice distortion signal appears at the location corresponding to the central region of the epitaxial wafer (the area within the red box in the image). This directly confirms that this batch of substrates already possesses a lattice strain region caused by the aggregation of interstitial atoms in its original state.

[0060] Finally, TEM samples were extracted from the defect region of the epitaxial wafer in Example 1 using focused ion beam (FIB), and observed by high-resolution transmission electron microscopy at an accelerating voltage of 200 kV. The images were then processed using inverse Fourier transform (IFFT), and the results are as follows: Figure 7 As shown, where Figure 7 In the image, 'a' represents the TEM image. Figure 7 In the image, b is the Burgers vector analysis diagram of the inverse Fourier transform image. The lattice distortion characteristics at the dislocation core can be directly observed from the image, which confirms that the interstitial atoms have fully aggregated to form a complete dislocation loop structure. Through this Burgers vector analysis, it is further confirmed that the nature of the defect is an interstitial atom type dislocation.

[0061] Based on the results of the above multi-level tests, and confirmed through traceability testing of the substrate, the heavily arsenic-doped substrate was found to have dislocation defects formed by the aggregation of interstitial atoms in the central region, and was therefore determined to be a defective product.

[0062] Comparative Example 1

[0063] This comparative example uses an epitaxial growth process with insufficient effective heat treatment time, i.e., t. eff <τ c This confirms that under these conditions, potential interstitial atomic defects in the substrate cannot be effectively detected, resulting in a "false negative" phenomenon. Details are as follows.

[0064] An 8-inch heavily doped silicon arsenide substrate from the same batch and with an adjacent serial number as in Example 1 was used. TEM analysis of Example 1 confirmed that this batch of substrates did indeed contain interstitial atomic defects.

[0065] Except for the epitaxial time, the process conditions in this comparative example are exactly the same as in Example 1. Specifically, the epitaxial time in this comparative example is 5 minutes, corresponding to an epitaxial layer thickness of approximately 10 μm (measured at 10.2 μm) and an epitaxial layer resistivity of approximately 11 Ω·cm. In this comparative example, the effective heat treatment time t... eff The time is 5 minutes, which is far lower than the critical aggregation time τ of about 25-30 minutes determined in Example 1. c .

[0066] The epitaxial wafer formed in this comparative example was tested using the same SP1 equipment and mode as in Example 1, and the results are as follows. Figure 8 As shown. Figure 8 In this context, 'a' represents the DCN mode. Figure 8 In this context, 'b' represents the DCO mode. From... Figure 8 It can be seen that only extremely weak center defect signals can be detected in DCN mode, such as Figure 8 Within the red box area in example a, the signal strength is much lower than in Example 1; and in DCO mode, the central defect area was not detected, as the defect signal was completely submerged in the background noise. Figure 8 No obvious signal was observed in b.

[0067] The epitaxial wafer of this comparative example was observed using an optical microscope, and the results are as follows: Figure 9 As shown. In DIC (Differential Interference Differential) modes at different magnifications, Figure 9 In the left and right images, at magnifications of 50X and 100X respectively, the defects are very slight, with blurred morphology, making it difficult to confirm the presence of dislocation features; compared with Example 1... Figure 3 Significant differences were formed.

[0068] This comparative example demonstrates that when the epitaxial time is as short as 5 minutes, even if potential self-interstitial atomic defects do exist in the substrate, the self-interstitial atoms fail to obtain sufficient thermal activation time for diffusion and aggregation, thus failing to form extended defects that can be effectively identified by SP1 and optical microscopy. Therefore, all levels of detection methods struggle to detect them. This also confirms the objective existence of the "thermal activation threshold effect" of this invention and its decisive role in detection. Using the epitaxial conditions of this comparative example for detection would lead to severe false negatives.

[0069] Example 2

[0070] This embodiment verifies the critical aggregation time τ by systematically changing the extrapolation time. c Threshold jump characteristics.

[0071] The same batch of 8-inch heavily doped arsenic substrates as in Example 1 were used. Under the condition of maintaining an epitaxial temperature of 1100℃ and a growth rate of 2 μm / min, four groups of experiments with different epitaxial times were set up: Group 1: t1 = 20 minutes (corresponding to an epitaxial layer thickness of approximately 40 μm); Group 2: t2 = 30 minutes (corresponding to an epitaxial layer thickness of approximately 60 μm); Group 3: t3 = 50 minutes (corresponding to an epitaxial layer thickness of approximately 100 μm); Group 4: t4 = 80 minutes (corresponding to an epitaxial layer thickness of approximately 160 μm); Group 5: t5 = 100 minutes (corresponding to an epitaxial layer thickness of approximately 200 μm). At least three substrates were used in each group of experiments for epitaxy, and the average number of defects detected by SP1 was taken as the statistical result.

[0072] The statistical results of the number of SP1 defects detected are as follows: Figure 10 As shown. From Figure 10 A clear threshold jump trend can be observed: When the epitaxial time t is 20 minutes: the average number of defects detected by SP1 is less than 5 per piece, which cannot form an identifiable feature signal, and the signal is basically submerged in the background noise.

[0073] At an epitaxial time t of 30 minutes (critical transition group): the number of detected defects significantly increased to approximately 45-60 per piece, and the signal began to exhibit identifiable cluster distribution characteristics. This result directly confirms that at the critical aggregation time τ... c A significant threshold jump occurred around 25-33 minutes, which corresponds to the τ predicted by the theoretical model of this invention. c The numerical ranges match.

[0074] With an epitaxial time t of 50 minutes, the number of defects detected increases to approximately 70-80 per piece, and the signal strength is enhanced.

[0075] With an epitaxial time t of 80 minutes, the number of defects detected further increases to approximately 90-100 per piece, the characteristic signals are clear and stable, and the consistency between pieces is good.

[0076] When the epitaxial time t is 100 minutes: the number of defects detected increases to about 100-120 per piece, but compared with 80 minutes, the growth rate tends to be flat, indicating that the defect signal has basically saturated.

[0077] Figure 10 The curve showing the change in the number of defects detected over time exhibits a typical "threshold jump" characteristic: between 20 and 30 minutes, the number of defects increases by nearly an order of magnitude, rather than changing gradually in a nonlinear manner. This is the critical aggregation time τ. c The existence of [the epitaxial process] provides experimental evidence. According to the experimental data, under conditions of 1100℃ and 2μm / min, only by controlling the epitaxial process at [the specified temperature]... eff ≥τ c Only within a specific window can self-interstitial atomic latent dislocation defects in heavily arsenic-doped substrates be effectively excited and detected; the effective thermal activation time should be at least greater than 50 minutes (corresponding to a thickness ≥100 μm) to obtain stable and reliable detection results. This data is consistent with the theoretical model prediction of this invention (τ). c (≈25-33 minutes) matches, but considering the safety margin, ≥50 minutes is preferred in engineering practice.

[0078] Only when the effective heat treatment time of the epitaxial layer is t eff Reaching or exceeding the critical aggregation time τ c Only when t is reached can the dispersed interstitial atoms in the heavily arsenic-doped substrate be fully excited, aggregated, and form detectable extended dislocation defects; when t eff Much smaller than τ c At that time, even if potential defects exist in the substrate, they cannot be effectively detected. Under the condition of meeting the thermal activation threshold, such as t in Example 1... eff =75 minutes, the extension time t in Example 2 eff =50 minutes, SP1, optical microscopy, chemical etching, XRT, TEM and other detection methods can clearly identify the characteristic signals of interstitial atomic dislocations, and the detection results have good self-consistency and correspondence.

[0079] The detection method of this invention can be widely applied to quality inspection and defect screening of heavily arsenic-doped silicon substrates for power semiconductor devices. In actual production, after the epitaxial process is completed, the epitaxial wafers can be sampled for inspection. The sampled epitaxial wafers are then subjected to epitaxial growth according to the thermal activation window provided by this invention or directly using the product epitaxial process that meets the threshold conditions. Then, SP1 rapid scanning and chemical etching verification are performed to efficiently and accurately determine whether there are interstitial atomic dislocation defects in the substrate. This method has the advantages of simple operation, controllable cost, and high reliability of detection results, and is suitable for widespread use in the semiconductor manufacturing industry.

[0080] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art can make various improvements and modifications without departing from the spirit and principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate, characterized in that, The detection method is based on the epitaxial thermal activation threshold effect and includes the following steps: Step 1, Substrate pretreatment: The heavily arsenic-doped silicon substrate to be tested is cleaned. Step 2, thermally activated epitaxial growth: The pretreated heavily arsenic-doped substrate is placed in the epitaxial reaction chamber to grow a lightly doped N-type silicon epitaxial layer. The epitaxial process is set to meet the thermal activation threshold, which refers to the minimum thermal budget required for the interstitial silicon atoms in the substrate to transform from dispersed point defect states to detectable extended dislocation states. The epitaxial process is configured to meet the thermal activation threshold, which is defined as follows: the epitaxial temperature is set to T, and t eff ≥ τ c ; where t eff The effective heat treatment time for epitaxy refers to the heat treatment time at the epitaxial temperature T; τ c The critical aggregation time refers to the time when the diffusion length of interstitial silicon atoms at the epitaxial temperature T reaches the critical diffusion length L. c The required time; the critical diffusion length L c This corresponds to the average initial spacing of interstitial silicon atoms at epitaxial temperature T; Step 3, First-level detection – Feature signal identification: A surface particle detector is used to perform light scattering scanning on the surface of the epitaxial wafer to obtain the spatial distribution and size information of defects; among them, the feature signals identified as interstitial atomic dislocations of the substrate include: a group of signals that can be detected in both DCN mode and DCO mode, and are distributed in a central cluster, and the size of a single signal in the group is concentrated in the range of 50nm to 2μm.

2. The method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 1, characterized in that, The critical aggregation time τ c Determined by the following formula: Where D is the diffusion coefficient of interstitial silicon atoms at epitaxial temperature T; the diffusion coefficient D satisfies: D0 is the pre-exponential factor, with a value of 0.1cm. 2 / s to 1cm 2 / s;E a The activation energy is set to 2.3 eV to 2.5 eV, where k is the Boltzmann constant, and the epitaxial temperature T is its absolute temperature value.

3. The method for detecting interstitial atomic dislocations on a heavily arsenic-doped substrate according to claim 2, characterized in that, The resistivity of the heavily arsenic-doped substrate is from 0.001 Ω·cm to 0.005 Ω·cm; the epitaxial temperature T is from 1050 °C to 1150 °C, corresponding to the critical diffusion length L. c The value ranges from 8 μm to 10 μm, and the diffusion coefficient D is 10. -8 cm 2 / s to 10 -7 cm 2 / s; the τ c The time is 25-33 minutes.

4. The method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 3, characterized in that: In step two, the epitaxial temperature is 1080℃ to 1120℃, the epitaxial growth rate is 1μm / min to 3μm / min, and the heat treatment time is t. eff The time is 50 to 100 minutes.

5. The method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 1, characterized in that, In step two, the epitaxial growth is carried out using chemical vapor deposition, with trichlorosilane as the silicon source gas, high-purity hydrogen as the carrier gas, and phosphine as the dopant gas.

6. The method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 1, characterized in that, The substrate pretreatment in step one includes: sequentially performing SC-1 cleaning to remove particulate contamination, SC-2 cleaning to remove metal ion contamination, then soaking in dilute HF solution to remove the natural oxide layer on the surface, and then drying with nitrogen.

7. The method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 1, characterized in that, It also includes step four, the second-level detection, namely microscopic morphology observation: for the epitaxial wafers with characteristic signals detected by SP1, the surface morphology is observed using an optical microscope and / or electron beam defect verification to verify whether there are convex dislocation lines or irregular protrusion groups corresponding to the SP1 signal.

8. The method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 1, characterized in that, It also includes step five, the third level of detection, namely specific chemical etching display: for epitaxial wafers that have characteristic signals detected by SP1, selective chemical etching solution is used for etching treatment. If clustered elliptical or quasi-triangular etching pits that are consistent with the spatial distribution of the characteristic signals appear on the surface of the epitaxial layer after etching, it is confirmed that the characteristic signals originate from interstitial atomic dislocations in the substrate.

9. A method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 8, characterized in that, The selective chemical etching solution is a chromic acid etching solution with a concentration of 0.5 mol / L to 1.0 mol / L CrO3 aqueous solution; the etching method is static immersion, and the etching time is 3 to 10 minutes.

10. The method for detecting interstitial atomic dislocations in a heavily arsenic-doped substrate according to claim 1, characterized in that, It also includes step six, the fourth level of detection, namely substrate traceability and defect structure confirmation: for substrates in the same batch that are confirmed to be interstitial atomic dislocation defects, X-ray morphology and / or transmission electron microscopy are used for traceability detection to observe lattice distortion or dislocation core structure in the substrate.

Images