Etching control method for low-loss waveguide of laser chip and related equipment

Abstract

The invention provides an etching control method and related equipment for a low-loss waveguide of a laser chip, and the method comprises the steps: synchronously forming a main waveguide ridge and monitoring ridges with different widths on a semiconductor substrate through etching, and enabling the monitoring ridges to comprise a narrow-width monitoring ridge and a wide-width monitoring ridge; in the etching process, optical excitation is applied to the monitoring ridges to collect scattered light intensity signals, difference processing is carried out on scattered light intensity of the narrow-width monitoring ridges and scattered light intensity of the wide-width monitoring ridges to judge the roughness of the side wall, and when the roughness of the side wall deviates from a preset interval, etching process parameters are adjusted; after the passivation layer is deposited, optical excitation is applied to the monitoring ridges to collect guided mode light intensity signals, the guided mode light intensity of the narrow-width monitoring ridge and the guided mode light intensity of the wide-width monitoring ridge are compared to judge the coverage uniformity of the passivation layer, and when the coverage uniformity deviates from a preset threshold value, passivation process parameters are adjusted. In-situ monitoring of the roughness of the side wall and the uniformity of the passivation layer and real-time adjustment of process parameters are achieved, and the problem that the batch consistency is poor due to the fact that real-time feedback cannot be achieved through off-line detection is solved.

Description

Technical Field

[0001] This invention relates to the field of semiconductor device manufacturing technology, and in particular to an etching control method and related equipment for low-loss waveguides of laser chips. Background Technology

[0002] A laser chip is a semiconductor device capable of generating and amplifying optical signals on a microscale. Its basic structure typically begins with a semiconductor substrate on which multiple functional layers are grown, including doped layers to provide charge carriers, active regions to generate light, and cladding layers to confine the light field distribution. On the chip surface, an etching process forms a ridge waveguide structure to constrain the propagation path of light within the chip. The quality of the waveguide sidewalls directly determines the magnitude of optical transmission loss; the smoother the sidewalls, the lower the scattering loss. To protect the waveguide sidewalls and reduce losses caused by surface recombination, a passivation layer, such as a silicon nitride film, is typically deposited after etching to passivate surface dangling bonds and reduce interface defects.

[0003] During mass production, waveguide transmission loss must be kept low to ensure laser output power and efficiency. Sidewall roughness is a major factor contributing to scattering loss, and even variations in root-mean-square roughness on the order of nanometers can significantly impact device performance. However, due to the high aspect ratio of waveguide structures, the plasma interaction is uneven at different locations during etching, and the deposition rate of the passivation layer differs between narrow and wide ridge regions. These factors easily lead to inconsistencies in sidewall quality and passivation layer coverage. Existing technologies primarily rely on scanning electron microscopy or atomic force microscopy after etching to assess sidewall roughness, or indirectly infer passivation layer quality through optical loss testing. However, these methods are offline inspections and cannot determine the evolution of sidewall quality and passivation layer coverage uniformity in real time during the process. Therefore, problems arise such as difficulty in precisely controlling the process window and ensuring batch consistency, which are key challenges in the mass production of high-performance laser chips. Summary of the Invention

[0004] The main objective of this invention is to solve the technical problem that existing laser chip waveguide manufacturing processes cannot monitor sidewall roughness and passivation layer quality in real time, resulting in insufficient process control precision and poor batch consistency.

[0005] The first aspect of this invention provides an etching control method for a low-loss waveguide of a laser chip, the etching control method for the low-loss waveguide of the laser chip comprising: A main waveguide ridge and at least two monitoring ridges of different widths are simultaneously formed on a semiconductor substrate by etching, wherein the at least two monitoring ridges include a narrow-width monitoring ridge and a wide-width monitoring ridge. Photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the scattered light intensity signal during the etching process. The scattered light intensity signal of the narrow-width monitoring ridge and the scattered light intensity signal of the wide-width monitoring ridge are differentially processed to determine the sidewall roughness. When the sidewall roughness deviates from the preset roughness range, the etching process parameters are adjusted. After depositing the passivation layer, photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the guided mode light intensity signal. The guided mode light intensity signal of the narrow-width monitoring ridge is compared with the guided mode light intensity signal of the wide-width monitoring ridge to determine the uniformity of the passivation layer coverage. When the uniformity of the passivation layer coverage deviates from the preset uniformity threshold, the passivation process parameters are adjusted.

[0006] A second aspect of the present invention provides an etching control device for a low-loss waveguide of a laser chip, the etching control device comprising: A structural building block for simultaneously forming a main waveguide ridge and at least two monitoring ridges of different widths on a semiconductor substrate by etching, wherein the at least two monitoring ridges include a narrow-width monitoring ridge and a wide-width monitoring ridge; The roughness monitoring unit is used to apply light excitation to the narrow width monitoring ridge and the wide width monitoring ridge, obtain the scattered light intensity signal during the etching process, and perform differential processing on the scattered light intensity signal of the narrow width monitoring ridge and the scattered light intensity signal of the wide width monitoring ridge to determine the sidewall roughness. When the sidewall roughness deviates from the preset roughness range, the etching process parameters are adjusted. The uniformity monitoring unit is used to apply photoexcitation to the narrow-width monitoring ridge and the wide-width monitoring ridge after the passivation layer is deposited, obtain the guided mode light intensity signal, and compare the guided mode light intensity signal of the narrow-width monitoring ridge with the guided mode light intensity signal of the wide-width monitoring ridge to determine the uniformity of the passivation layer coverage. When the uniformity of the passivation layer coverage deviates from the preset uniformity threshold, the passivation process parameters are adjusted.

[0007] The aforementioned etching control method and related equipment for low-loss waveguides in laser chips simultaneously form a main waveguide ridge and monitoring ridges of different widths on a semiconductor substrate through etching. These monitoring ridges include narrow-width and wide-width monitoring ridges. During etching, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and scattered light intensity signals are collected. The scattered light intensity signals are then differentially processed to determine the sidewall roughness. When the sidewall roughness deviates from a preset roughness range, the etching process parameters are adjusted. After passivation layer deposition, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and guided mode light intensity signals are collected. The guided mode light intensity signals are then compared to determine the passivation layer coverage uniformity. When the passivation layer coverage uniformity deviates from a preset uniformity threshold, the passivation process parameters are adjusted. This solution enables in-situ monitoring and adaptive adjustment of the sidewall quality and passivation layer coverage state of waveguide structures with different widths during etching and passivation processes, improving the uniformity and yield of the laser chip waveguide transmission performance and increasing mass production yield.

[0008] Beneficial Effects: This invention achieves in-situ monitoring and process control of sidewall roughness and passivation layer uniformity during the manufacturing of optoelectronic devices such as laser chips by simultaneously forming monitoring ridges of different widths on a semiconductor substrate. The differential measurement strategy effectively eliminates common-mode interference such as light source fluctuations and material absorption, extracting pure signals directly related to sidewall roughness. The response sensitivity of the narrow-width monitoring ridge to sidewall roughness is 16 times that of the wide-width monitoring ridge. Compared to traditional offline detection using scanning electron microscopy, in-situ monitoring reduces measurement time from several hours to several minutes, enabling rapid feedback and real-time adjustment of process parameters. The quantitative adjustment strategy based on comprehensive roughness indices and transmittance ratios avoids the subjectivity and lag of traditional experience-based adjustments, transforming process optimization from qualitative judgment to quantitative control, shortening the process window convergence cycle, and reducing material and time costs caused by multiple batches of trial and error. This solution improves batch consistency of sidewall and passivation layer quality during laser chip waveguide manufacturing, enhancing the stability of waveguide transmission performance and mass production yield.

[0009] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0010] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the first embodiment of the etching control method for low-loss waveguides of laser chips in this invention. Figure 2 This is a schematic diagram of a second embodiment of the etching control method for low-loss waveguides of laser chips in this invention. Figure 3 A schematic diagram of an embodiment of the etching control device for the low-loss waveguide of the laser chip in this invention. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions 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, 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.

[0013] The terms "comprising" and "having," and any variations thereof, used in the embodiments of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0014] To facilitate understanding of this embodiment, a detailed description of the etching control method for a low-loss waveguide of a laser chip disclosed in this embodiment of the invention will be provided first. For example... Figure 1 As shown, this method includes the following steps: 101. A main waveguide ridge and at least two monitoring ridges of different widths are simultaneously formed on a semiconductor substrate by etching, wherein the at least two monitoring ridges include a narrow-width monitoring ridge and a wide-width monitoring ridge; In this embodiment, a pattern is defined on the surface of the semiconductor substrate using photolithography. After photoresist forms a mask pattern including the main waveguide ridge pattern and the monitoring ridge pattern, an inductively coupled plasma etching (ICP-CPE) system is used to vertically etch the areas not protected by the photoresist to a depth of 500 nanometers to 1 micrometer, forming a ridge structure. After etching, the photoresist is removed. The main waveguide ridge is a ridge-shaped waveguide structure in the laser chip used to constrain and transmit optical signals, while the monitoring ridge is an auxiliary ridge structure specifically used for process monitoring. "Synchronous formation" means that in the same etching process, the main waveguide ridge and the monitoring ridge undergo identical etching parameters and plasma environment. Their surface treatment conditions and material defect distribution are consistent, with only a design difference in ridge width, ensuring that the sidewall quality of the monitoring ridge can represent the actual state of the main waveguide ridge. The monitoring ridge is located in a non-device area or scribe line at the wafer edge, with a distance greater than 500 micrometers from the main waveguide ridge to avoid optical field coupling interference. The narrow-width monitoring ridge has a ridge width of 2 micrometers, the wide-width monitoring ridge has a ridge width of 4 micrometers, and the monitoring ridge length is 500 micrometers.

[0015] Different widths of monitoring ridges are used because waveguide sidewall scattering loss is affected by two factors: scattering caused by sidewall roughness and waveguide geometry. Narrower waveguides are more sensitive to sidewall roughness. When narrow and wide ridges are formed simultaneously, their surface roughness, material absorption, and bulk defect density are the same; only the sidewall scattering contribution differs due to the width. By subtracting the scattered light intensity signals from the narrow and wide ridges, the same background factors (light source intensity fluctuations, material absorption, surface scattering) cancel each other out in the differential operation, retaining only the difference signal related to sidewall roughness, thus achieving high-sensitivity extraction of sidewall roughness. Furthermore, the bombardment effect of plasma varies in different width regions during etching, and the step coverage capability of the passivation layer also differs with ridge widths. Simultaneous monitoring of both widths comprehensively reflects the quality status of the waveguide under different geometries, avoiding the bias caused by monitoring only one width.

[0016] 102. Apply light excitation to the narrow width monitoring ridge and the wide width monitoring ridge to obtain the scattered light intensity signal during the etching process, and perform differential processing on the scattered light intensity signal of the narrow width monitoring ridge and the scattered light intensity signal of the wide width monitoring ridge to determine the sidewall roughness. When the sidewall roughness deviates from the preset roughness range, adjust the etching process parameters. In this embodiment, before the passivation layer is deposited after the etching process, a confocal laser scanning microscope is used to perform high-resolution scanning measurements on the monitoring ridge sidewall. The confocal laser scanning microscope includes a 633 nm laser source, a 0.9 numerical aperture objective lens, a confocal pinhole, a photodetector, and a precision piezoelectric scanning platform. The objective lens focuses the laser to form a spot with a diameter of approximately 400 nm, illuminating the sidewall surface from the monitoring ridge side at a 60-degree incident angle. "Photoexcitation" refers to the interaction between the laser light and the nanoscale roughness of the sidewall when the laser shines on it, resulting in partial reflection and scattering of the light. The confocal pinhole filters out defocused light signals, achieving a lateral spatial resolution of approximately 50 nm. The detector receives the reflected and scattered light intensity from the focal point of the sidewall and converts the optical signal into an electrical signal. The "scattered light intensity signal" refers to the light intensity value recorded by the detector at each measurement position. This value is related to the sidewall roughness at that position. When the sidewall roughness is large, the light is scattered in different directions by the undulating structure, and the light intensity in the specular reflection direction decreases. When the sidewall is relatively smooth, the specular reflection is strong and the scattering is weak, resulting in a higher light intensity received by the detector. The piezoelectric scanning platform controls the laser spot to move along the length of the monitoring ridge, recording the light intensity signal every 50 nanometers. Light intensity data at 200 sampling positions are obtained within a selected 10-micrometer length region of the monitoring ridge. The 50-nanometer sampling interval is set according to the sampling theorem. When the correlation length of the sidewall roughness is on the order of 100 nanometers, the sampling interval must be less than half of the correlation length to accurately characterize the spatial distribution characteristics of the roughness. The "synchronous application" refers to the piezoelectric platform first aligning the scanning area with the sidewall of the narrow-width monitoring ridge. After scanning 200 sampling positions, the platform moves laterally by approximately 3 micrometers to align the scanning area with the same axial position on the wide-width monitoring ridge. Measurements are then performed using the same laser power of 0.5 milliwatts, an incident angle of 60 degrees, and a scanning step size of 50 nanometers, ensuring that the measurement conditions for both monitoring ridges are completely identical. For example, 200 light intensity data points are obtained from the narrow-width monitoring ridge. to 200 data points were obtained from wide-width ridge monitoring. to The nanoscale spatial resolution achieved by using confocal laser scanning can acquire detailed spatial distribution information of sidewall roughness. The spatial filtering effect of the confocal pinhole effectively suppresses background light interference from the top surface of the monitoring ridge and the material body, improving the correspondence between the light intensity signal and the local roughness of the sidewall.

[0017] The light intensity data of the narrow-width and wide-width monitoring ridges at corresponding sampling positions are then subtracted point by point to obtain a differential intensity sequence. Specifically, the differential intensity at the i-th sampling position is equal to the light intensity of the narrow ridge minus the light intensity of the wide ridge, and so on, resulting in 200 differential intensity values. The "differential processing" is a data processing method that subtracts two sets of measurement data to eliminate common interference factors and extract differential information. Since the narrow-width and wide-width monitoring ridges are formed synchronously in the same etching process, they undergo the same plasma environment, etching time, and surface treatment process, and their surface top roughness, material bulk defect density, and light absorption coefficient remain consistent. When irradiated by laser, factors such as scattering from the surface top roughness, scattering from internal material defects, laser power fluctuations, and detector response drift have the same amplitude of influence on the light intensity measurement of the narrow and wide ridges. These identical background signals are eliminated in the differential operation. However, the scattering effect caused by sidewall roughness is closely related to the waveguide width. According to waveguide scattering loss theory, scattering loss is inversely proportional to the fourth power of the waveguide width. A narrow ridge with a width of 2 micrometers is 16 times more sensitive to sidewall roughness than a wide ridge with a width of 4 micrometers. Therefore, the scattering effect produced by the same sidewall roughness on the narrow ridge is significantly stronger than that on the wide ridge, resulting in lower specular reflection intensity at the sidewall of the narrow ridge compared to the wide ridge. Differential operations retain this difference signal, which is directly related to the sidewall roughness and varies with the width, achieving effective separation of sidewall roughness information. Furthermore, statistical analysis is performed on the differential intensity sequence to extract characteristic parameters representing the sidewall roughness. The analysis process includes calculating the root mean square value of the differential intensity to characterize the fluctuation amplitude of the roughness, and calculating the spatial correlation of the differential intensity sequence to characterize the spatial periodicity of the roughness. According to the statistical theory of sidewall roughness and the waveguide scattering loss theory, the scattering loss caused by sidewall roughness depends simultaneously on the fluctuation amplitude of the roughness and the spatial correlation length. The larger the fluctuation amplitude and the shorter the correlation length, the more severe the scattering loss. Therefore, dividing the parameter characterizing the fluctuation amplitude by the parameter characterizing the correlation length yields the comprehensive roughness index. This ratio-based index increases when the fluctuation is large or the correlation length is short, consistent with the trend of scattering loss. The "preset roughness range" is a pre-defined allowable range of the comprehensive index based on the waveguide transmission loss target. During the process development stage, multiple sets of standard samples with known sidewall roughness are prepared. The root mean square roughness of the sidewalls of the standard samples is precisely measured offline using scanning electron microscopy combined with image processing. The roughness range covers 5 nm to 12 nm, the correlation length ranges from 80 nm to 150 nm, and the transmission loss is measured to be 0.08 dB / cm to 0.7 dB / cm using optical testing. The standard samples are then subjected to confocal laser scanning measurements according to this scheme, and the comprehensive roughness index is calculated to establish a quantitative correspondence between the comprehensive index and the transmission loss.When the transmission loss requirement is less than 0.5 dB per centimeter, the corresponding sidewall roughness must be less than 8 nanometers and the relevant length greater than 100 nanometers. The preset roughness range is determined to be 0.045 to 0.070 microwatts per nanometer. The comprehensive roughness index obtained from real-time measurement and calculation is compared with the preset roughness range to determine whether the sidewall quality meets the requirements.

[0018] When the overall roughness index exceeds the upper limit of the preset roughness range, it is determined that the sidewall roughness exceeds the standard. The etching process parameters for the next batch of wafers are adjusted based on the degree of deviation. Etching process parameters include RF power, bias power, etching gas flow rate, and process chamber pressure. Excessive sidewall roughness indicates that the physical bombardment of the sidewall by ions in the plasma is too strong or that the chemical etching reaction is not uniform enough. Reducing the RF power can decrease plasma density and ion energy, weakening the bombardment of the sidewall; increasing the process chamber pressure can increase the number of ion collisions and reduce the vertical energy of ions, making the etching process gentler; increasing the proportion of chemical etching gas flow rate can increase the ratio of chemical etching to physical bombardment, improving sidewall smoothness. The adjustment amount is determined based on the deviation of the measured overall roughness index from the preset range. The relative deviation is obtained by calculating the difference between the measured overall index and the upper limit of the preset range, and dividing this difference by the upper limit of the preset range. The adjustment of etching process parameters adopts a proportional control strategy. For RF power, the adjustment amount equals the negative response coefficient multiplied by the relative deviation and then multiplied by the current RF power setting. The response coefficient is obtained through process experiment calibration. The negative sign indicates that the RF power is reduced when the roughness exceeds the limit. For process chamber pressure, the adjustment amount equals the response coefficient multiplied by the relative deviation and then multiplied by the current pressure setting. The positive sign indicates that the pressure is increased when the roughness exceeds the limit. For example, when the measured comprehensive roughness index is 0.085 μW / nm, exceeding the upper limit of 0.070 μW / nm, the relative deviation is 0.015 divided by 0.070 equals 0.214. If the current RF power is 280 W and the response coefficient is 0.12, then the RF power adjustment amount is -0.12 multiplied by 0.214 multiplied by 280 equals -7.2 W. The RF power for the next batch is adjusted to 272.8 W. Based on the calculated adjustment amount, the etching process parameters for the next batch are modified. After etching is completed, the monitoring ridge is measured and analyzed again to verify whether the sidewall roughness has been reduced to within the preset range.

[0019] By performing in-situ high-resolution optical monitoring of sidewall quality immediately after etching, the measurement time is reduced from several hours to several minutes compared to traditional offline scanning electron microscopy (SEM) inspection, enabling rapid feedback. The differential measurement strategy eliminates common interferences unrelated to the sidewalls, extracting pure signal features of sidewall roughness. Quantitative adjustment of process parameters based on comprehensive roughness indices transforms traditional experience-based batch-to-batch iteration into precise control based on quantitative data. This shortens the process window convergence cycle, reduces batch-to-batch sidewall quality fluctuations, and improves the consistency of laser chip waveguide transmission performance and mass production yield.

[0020] 103. After depositing the passivation layer, photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the guided mode light intensity signal. The guided mode light intensity signal of the narrow-width monitoring ridge and the guided mode light intensity signal of the wide-width monitoring ridge are compared and processed to determine the uniformity of the passivation layer coverage. When the uniformity of the passivation layer coverage deviates from the preset uniformity threshold, the passivation process parameters are adjusted.

[0021] In this embodiment, after depositing the passivation layer, photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the guided mode light intensity signal. The comparison processing of the guided mode light intensity signal of the narrow-width monitoring ridge and the guided mode light intensity signal of the wide-width monitoring ridge includes: applying photoexcitation to the narrow-width monitoring ridge and the wide-width monitoring ridge respectively, and collecting the corresponding guided mode transmitted light intensity; calculating the ratio of each guided mode transmitted light intensity to the corresponding incident light intensity to obtain the transmittance of the narrow-width monitoring ridge and the transmittance of the wide-width monitoring ridge respectively; and calculating the ratio of the transmittance of the narrow ridge to the transmittance of the wide ridge to obtain the transmittance ratio. The determination of passivation layer coverage uniformity includes: comparing the transmittance ratio with a preset standard transmittance ratio to calculate the transmittance ratio deviation; comparing the narrow ridge transmittance with a preset narrow ridge standard transmittance, and comparing the wide ridge transmittance with a preset wide ridge standard transmittance to calculate the narrow ridge transmittance deviation and the wide ridge transmittance deviation; based on the transmittance ratio deviation, the narrow ridge transmittance deviation, and the wide ridge transmittance deviation, when the transmittance ratio deviation exceeds a preset uniformity threshold, or when either the narrow ridge transmittance deviation or the wide ridge transmittance deviation exceeds the preset uniformity threshold, it is determined that the passivation layer coverage uniformity is abnormal. The passivation process parameters include deposition process parameters and annealing process parameters. When the passivation layer coverage uniformity deviates from the preset uniformity threshold, adjusting the passivation process parameters includes: when the transmittance ratio deviation is positive, decreasing the deposition process parameter value for the next batch; when the transmittance ratio deviation is negative, increasing the deposition process parameter value for the next batch; when both the narrow ridge transmittance deviation and the wide ridge transmittance deviation are negative, increasing the annealing process parameter value for the next batch. Furthermore, the method includes: forming multiple sets of monitoring ridges at different locations on the semiconductor substrate, each set of monitoring ridges including narrow-width monitoring ridges and wide-width monitoring ridges, the different locations including the substrate center region and the substrate edge region; performing differential processing of scattered light intensity signals and comparative processing of guided mode light intensity signals on the monitoring ridges at each location to obtain the sidewall roughness determination results and passivation layer coverage uniformity determination results for each location; statistically analyzing the distribution data of sidewall roughness and passivation layer coverage uniformity for each location; adjusting the process parameters affecting uniformity when the number of abnormal monitoring ridges in the edge region exceeds a preset ratio; and adjusting the process parameters affecting stability when the abnormal monitoring ridges are randomly distributed.

[0022] Specifically, after the passivation layer is deposited, an optical testing system is used to measure guided mode transmission on the monitoring ridge. The monitoring ridge is designed with a grating coupler structure, with a shallowly etched grating at each of the ridge's starting and ending points. The "grating coupler" is a shallow groove structure periodically arranged on the waveguide surface, with a period of approximately 400 nanometers, a groove depth of approximately 80 nanometers, and a grating region length of approximately 50 micrometers. The optical testing system includes a tunable laser source, a polarization controller, a focusing objective, a precision three-dimensional displacement stage, a photodetector, and a data acquisition module. The "guided mode" refers to an optical mode that is constrained by the sidewalls and cladding within the waveguide structure and propagates along the waveguide's length; the light energy is mainly concentrated in the core region of the waveguide ridge. "Optical excitation" refers to using a laser to couple light from the top of the monitoring ridge through the grating coupler into the waveguide, exciting the guided mode transmission. During measurement, a tunable laser source emits a 1550 nm wavelength laser. After being adjusted to a TE polarization state by a polarization controller, the laser is focused into a spot with a diameter of approximately 10 micrometers by a focusing objective lens, which perpendicularly illuminates the grating coupler region at the beginning of the monitoring ridge. The periodic structure of the grating alters the direction of light propagation, coupling some of the perpendicularly incident light into a guided mode propagating laterally along the waveguide. Specifically, "applying optical excitation" involves adjusting the objective lens position using a precision three-dimensional displacement stage to accurately align the focused spot with the center of the grating coupler, with the laser power set to 2 milliwatts. The light propagates in the monitoring ridge waveguide as a guided mode, affected by sidewall scattering loss, material absorption loss, and passivation layer interface loss. When the guided mode reaches the grating coupler at the end of the monitoring ridge, the grating converts the guided mode into an upward-emitted beam. A photodetector is placed above the grating at the end of the ridge to receive the optical signal coupled from the grating. The "guided mode transmitted light intensity" refers to the output light intensity measured by the photodetector, measured in microwatts. Before measurement, the laser beam illuminated by the focusing objective lens is directly aimed at the photodetector, and the incident optical power onto the grating is measured. Considering that the coupling efficiency of the grating is approximately 30%, the effective incident optical power onto the waveguide is calculated as the reference incident light intensity. The objective lens was then aligned with the grating coupler at the beginning of the narrow-width monitoring ridge for coupling. The photodetector was moved above the grating at the end to measure the output light intensity, which was recorded as follows: The same measurement conditions were used to measure the width of the monitoring ridge, and the output light intensity was recorded as follows: For example, measured (This corresponds to an optical power of approximately 180 microwatts coupled into the waveguide) , The normalized transmittance is then obtained by dividing the output light intensity of each monitoring ridge by the reference incident light intensity. The "narrow-width monitoring ridge transmittance" is then described. The "wide-width monitoring ridge transmittance" .For example, =120 / 600=0.200, =135 / 600=0.225. The transmittance value is affected by both the grating coupling efficiency and the waveguide transmission loss; a higher value indicates lower waveguide transmission loss. Guided-mode transmission measurement reflects the optical performance of the waveguide through the actual optical transmission process. Transmittance is directly affected by the quality of the passivation layer. The better the passivation effect of the passivation layer on the sidewall surface states, the lower the non-radiative recombination loss, and the higher the transmittance of guided-mode transmission.

[0023] The transmittance ratio is then obtained by dividing the transmittance of the narrow-width monitoring ridge by the transmittance of the wide-width monitoring ridge. For example, when =0.200、 When =0.225, =0.200 / 0.225=0.889. The "transmittance ratio" is a parameter characterizing the uniformity of passivation layer coverage in waveguide regions of different widths. Since narrow-width and wide-width monitoring ridges have the same sidewall quality upon etching, the difference in loss before passivation mainly stems from the different responses of waveguide width to sidewall scattering. During passivation layer deposition, the growth of the passivation film is affected by the waveguide geometry. The smaller spacing between the sidewalls of the narrow-ridge waveguide forms a narrower trench structure. The diffusion resistance of the passivation layer precursor gas molecules entering the narrow trench is greater, and the concentration of deposited reactants at the bottom of the trench is lower than that of the wide trench. This results in a slower passivation layer deposition rate on the narrow-ridge sidewalls than on the wide-ridge sidewalls, ultimately leading to a thinner passivation layer thickness. This phenomenon is called the micro-loading effect in chemical vapor deposition. If the passivation process conditions are well optimized, the micro-loading effect is effectively suppressed, the passivation layer thickness difference between narrow and wide ridges is small, their passivation effects are similar, the relative difference in transmittance remains stable, and the transmittance ratio is close to the standard value when the process is stable. If the passivation process conditions deviate, the micro-loading effect intensifies, the passivation layer coverage of the narrow ridge is significantly thinner than that of the wide ridge, the passivation effect of the narrow ridge is weaker than that of the wide ridge, the transmittance of the narrow ridge decreases more, and the transmittance ratio deviates from the standard value. The transmittance ratio transforms the absolute transmittance information of the narrow and wide ridges into a relative index reflecting the uniformity of coverage, and this relative index has high sensitivity to changes in the uniformity of the passivation layer.

[0024] Then, multiple parameters are used to determine the transmittance ratio and absolute transmittance. First, the transmittance ratio is compared with a preset standard transmittance ratio. The "preset standard transmittance ratio" is the target transmittance ratio when the process is stable. During the process development phase, guided mode transmission measurements are performed on multiple batches of samples with stable passivation processes and qualified device performance. The distribution of transmittance ratios is statistically analyzed, and the preset standard transmittance ratio is determined to be 0.900. The transmittance ratio deviation is then calculated. = -0.900. For example, when When =0.889, =-0.011, a negative value, indicates that the transmittance of the narrow ridge is relatively lower than that of the wide ridge, and the passivation layer coverage on the narrow ridge is insufficient. Next, the transmittance of the narrow ridge and the wide ridge are compared with their respective preset standard transmittances. According to process development data, the preset standard transmittance for the narrow ridge is 0.210, and the preset standard transmittance for the wide ridge is 0.230. The transmittance deviation of the narrow ridge is calculated. = -0.210, transmittance deviation of wide ridge = -0.230. For example, when =0.200、 When =0.225, =-0.010, =-0.005. The "preset uniformity threshold" is the allowable deviation range for determining anomalies. The threshold for transmittance ratio deviation is set to 0.015, and the threshold for absolute transmittance deviation is set to 0.012. The determination logic is as follows: when the absolute value of the transmittance ratio deviation exceeds 0.015, or the absolute value of the narrow ridge transmittance deviation exceeds 0.012, or the absolute value of the wide ridge transmittance deviation exceeds 0.012, it is determined that the passivation layer coverage uniformity is abnormal. For example, when =-0.011、 =-0.010、 When the value is -0.005, the absolute values ​​of the three deviations are 0.011, 0.010, and 0.005, respectively, none of which exceed their respective thresholds, indicating that the passivation layer coverage uniformity is normal. When the value equals -0.020, the absolute value of 0.020 exceeds the threshold of 0.015 and is judged as abnormal. By using both the transmittance ratio and the absolute value of transmittance for judgment, uniformity problems and overall quality problems can be distinguished, thus improving the ability to identify different passivation defect modes.

[0025] When an abnormality in passivation layer coverage is detected, the deposition process parameters for the next batch are adjusted based on the sign and magnitude of the transmittance ratio deviation. The "deposition process parameters" refer to the process conditions of the plasma-enhanced chemical vapor deposition (PECVD) process for the passivation layer, including radio frequency (RF) power, process gas flow rate ratio, and process chamber pressure. RF power affects plasma density and reactivity, process gas flow rate ratio affects the chemical composition of the deposition reaction, and process chamber pressure affects the diffusion ability of gas molecules. When the transmittance ratio deviation is positive, the transmittance of the narrow ridge is relatively high, resulting in a relatively thick passivation layer on the narrow ridge, requiring a reduction in process parameters that promote uniform deposition. When the transmittance ratio deviation is negative, the transmittance of the narrow ridge is relatively low, resulting in insufficient passivation layer coverage on the narrow ridge, requiring an increase in process parameters that promote uniform deposition. "Reducing the value of deposition process parameters" specifically refers to reducing RF power or reducing process chamber pressure. Reducing RF power weakens plasma activity, reduces gaseous reaction products, and lowers the overall deposition rate; reducing pressure reduces gas molecule collisions, increases the mean free path of molecules, and improves gas diffusion in narrow trenches. The phrase "increasing the deposition process parameters" specifically refers to increasing the radio frequency (RF) power or the process chamber pressure. Increasing RF power enhances reaction activity and increases the concentration of reaction products; increasing pressure increases the frequency of molecular collisions, promoting the transport of reactants to the bottom of the narrow trench. The adjustment amount is determined based on the deviation magnitude. When the absolute value of the transmittance ratio deviation is between 0.015 and 0.025, the RF power is adjusted by 3% to 6% relative to the current value, or the pressure is adjusted by 8% to 15%. When the absolute value of the deviation exceeds 0.025, the RF power is adjusted by 6% to 10%, or the pressure is adjusted by 15% to 25%. For example, when the transmittance ratio deviation is -0.020, which is negative and an absolute value of 0.020 is considered a slight deviation, the current RF power of 80 watts is increased by 5% to 84 watts, or the pressure is increased from 150 millitonnes to 168 millitonnes by 12%.

[0026] When both the narrow ridge transmittance deviation and the wide ridge transmittance deviation are negative, the transmittance of both the narrow and wide ridges is lower than the standard value, indicating insufficient overall passivation effect of the passivation layer, requiring an increase in annealing process parameters. The "annealing process parameters" refer to the rapid thermal annealing process conditions after passivation layer deposition, including annealing temperature and annealing time. Annealing activates the passivation effect of the passivation layer on the dangling bonds of the sidewall surface, reducing the interface state density. "Increasing the annealing process parameter values" refers to increasing the annealing temperature or extending the annealing time. Increasing the annealing temperature enhances the thermal activation process of the passivation reaction, promoting the bonding of surface dangling bonds with hydrogen atoms in the passivation layer; extending the annealing time makes the passivation reaction more complete. When the absolute value of the negative deviation is less than 0.010, it is considered slightly insufficient; the annealing temperature should be increased by 2% to 4% relative to the current value, or the annealing time extended by 10% to 15%. When the absolute value exceeds 0.010, it is considered severely insufficient; the annealing temperature should be increased by 4% to 8%, or the annealing time extended by 15% to 25%. For example, when... =-0.008、 When the value is -0.009, all values ​​are negative and the absolute value is less than 0.010. Increase the current annealing temperature from 420 degrees Celsius by 3% to 433 degrees Celsius, or extend the annealing time from 25 minutes to 28 minutes by 12%.

[0027] The passivation layer quality is evaluated through guided mode transmission measurements. The transmittance ratio reflects coverage uniformity, while the absolute transmittance reflects the overall passivation effect, achieving multi-dimensional characterization. Deposition and annealing parameters are adjusted based on the type of deviation to avoid increased waveguide loss caused by passivation layer quality issues, thereby improving the consistency of the laser chip passivation process and device yield.

[0028] Furthermore, multiple sets of monitoring ridges are formed at different locations on the surface of the semiconductor substrate. For a circular semiconductor substrate with a diameter of 3 inches (approximately 75 mm), the "substrate center region" refers to the area with a radial distance of less than 10 mm from the geometric center of the substrate, and the "substrate edge region" refers to the area with a radial distance of less than 10 mm from the outer edge of the substrate. The first set of monitoring ridges is set at the center of the substrate; the second to fourth sets of monitoring ridges are evenly distributed along the circumference at a radial distance of 20 mm from the center; and the fifth to eighth sets of monitoring ridges are evenly distributed along the circumference at a radial distance of 8 mm from the outer edge, for a total of eight sets of monitoring ridges per substrate. Each set of monitoring ridges includes a narrow-width monitoring ridge and a wide-width monitoring ridge, spaced 5 micrometers apart along the radial direction of the substrate. For example, the first set is located at the center, and the fifth to eighth sets are located at the four edges (top, bottom, left, and right), respectively. Multiple monitoring ridges were set up in different locations because the etching and passivation processes exhibit radial non-uniformity on the substrate due to the influence of equipment structure. Plasma density is lower in the edge regions than in the center regions, and substrate temperature also deviates in the edge regions, resulting in relatively poor sidewall and passivation layer quality in the edge regions. Multi-location monitoring can identify this spatial distribution characteristic. Confocal laser scanning measurements and differential processing were performed on each of the eight monitoring ridges. The comprehensive sidewall roughness index of each group was calculated and compared with a preset roughness range to obtain the sidewall roughness judgment result (qualified or exceeding the standard). After passivation layer deposition, guided mode transmission measurements and transmittance comparison processing were performed on each of the eight monitoring ridges. The transmittance ratio and transmittance deviation of each group were calculated and compared with a preset uniformity threshold to obtain the passivation layer coverage uniformity judgment result (normal or abnormal) for each group. For example, the measurement results for a certain batch were: groups 1 to 3 were judged as qualified and normal, group 4 was judged as exceeding the standard and abnormal, and 3 out of the 4 groups in the edge region of groups 5 to 8 were judged as exceeding the standard and abnormal. The location coordinates and judgment results of each group were recorded for subsequent analysis. The spatial distribution pattern of abnormal monitoring ridges is then statistically analyzed. The "preset ratio" is set to 50%. When more than two out of the four monitoring ridges in the edge region are identified as abnormal, the abnormality ratio in the edge region is considered too high. Simultaneously, the abnormality ratio in the center region (100% for an abnormal ridge in the first monitoring ridge group, and 0% for normal) and the abnormality ratio in the edge region (number of abnormal groups divided by 4) are calculated. When the difference between the abnormality ratio in the edge region and the abnormality ratio in the center region is greater than 30%, the abnormality is considered concentrated in the edge region, and the process parameters affecting uniformity are adjusted. The "process parameters affecting uniformity" include the RF power of the etching equipment, the substrate heating temperature, and the process gas flow rate. Specifically, the RF power corresponding to the edge region is increased by 5% to 10% relative to the center region, the substrate heating temperature corresponding to the edge region is increased by 3 to 5 degrees Celsius, and the flow rate ratio of the gas spray nozzles in the edge region is increased.When the difference between the anomaly ratio in the edge region and the anomaly ratio in the center region is less than 30%, and anomalies are present in both the center and edge regions, the anomalies are determined to be randomly distributed, and process parameters affecting stability are adjusted. These "process parameters affecting stability" include the process chamber cleaning cycle, reactant gas purity, and equipment preheating time. Specifically, the chamber cleaning cycle is shortened from once every 20 wafers to once every 15 wafers; the reactant gas purity level is increased from 99.99% to 99.999%; and the equipment preheating time is extended from 10 minutes to 15 minutes. By analyzing the anomaly distribution pattern and specifically adjusting different types of process parameters, effective differentiation and precise control of process uniformity and stability issues are achieved.

[0029] In this embodiment of the invention, a main waveguide ridge and monitoring ridges of different widths are simultaneously formed on a semiconductor substrate by etching. The monitoring ridges include narrow-width and wide-width monitoring ridges. During etching, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and scattered light intensity signals are collected. The scattered light intensity signals are then differentially processed to determine the sidewall roughness. When the sidewall roughness deviates from a preset roughness range, the etching process parameters are adjusted. After passivation layer deposition, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and guided mode light intensity signals are collected. The guided mode light intensity signals are then compared to determine the passivation layer coverage uniformity. When the passivation layer coverage uniformity deviates from a preset uniformity threshold, the passivation process parameters are adjusted. This solution enables in-situ monitoring and adaptive adjustment of the sidewall quality and passivation layer coverage state of waveguide structures of different widths during etching and passivation processes, improving the uniformity and yield of the transmission performance of laser chip waveguides and increasing mass production yield.

[0030] Please see Figure 2 Another embodiment of the etching control method for low-loss waveguides of laser chips in this application includes: 201. A main waveguide ridge and at least two monitoring ridges of different widths are simultaneously formed on a semiconductor substrate by etching, wherein the at least two monitoring ridges include a narrow-width monitoring ridge and a wide-width monitoring ridge; In this embodiment, step 201 is similar to step 101 in the first embodiment, and will not be described again here.

[0031] 202. Simultaneously apply lateral incident light to the sidewall regions of the narrow-width monitoring ridge and the wide-width monitoring ridge at multiple sampling positions during the preset etching process, and collect scattered light intensity data within a preset angle range respectively; In this embodiment, after the etching process is completed, a confocal laser scanning microscope is used to scan and measure the sidewall of the monitoring ridge. The confocal laser scanning microscope includes a laser source with a wavelength of 633 nm, an objective lens with a numerical aperture of 0.9, a confocal pinhole, a photomultiplier tube detector, and a precision three-dimensional piezoelectric scanning platform. The objective lens focuses the laser onto the sidewall surface to form a spot with a diameter of approximately 400 nm. The "sampling position" refers to the measurement point selected along the length of the monitoring ridge. The confocal pinhole filters out defocused light signals, achieving a lateral spatial resolution of approximately 50 nm. A scanning area of ​​10 μm in length is selected for the monitoring ridge. The piezoelectric scanning platform controls the spot to move along the length of the sidewall, recording the light intensity signal every 50 nm, obtaining data from a total of 200 sampling positions. The "side-incident light" refers to the laser beam illuminating the sidewall from the side of the monitoring ridge at a 60-degree incident angle, rather than from a vertical incidence from the top. Side illumination allows the laser to act directly on the sidewall surface. At each sampling position, after the laser illuminates the sidewall, part of the light is reflected and scattered by the sidewall surface. The "synchronous application" refers to the piezoelectric platform first controlling the scanning area to align with the sidewall of the narrow-width monitoring ridge. After scanning 200 sampling positions, the platform moves laterally by approximately 3 micrometers to align the scanning area with the same axial position of the wide-width monitoring ridge. Measurements are performed using the same laser power of 0.5 milliwatts, an incident angle of 60 degrees, and a scanning step size of 50 nanometers, ensuring that the measurement conditions for the two monitoring ridges are completely consistent. For example, for a sampling position 1 micrometer from the starting end, the sidewalls of both the narrow and wide ridges at that position are respectively illuminated by laser and the light intensity signal is recorded. The 50-nanometer sampling interval meets the spatial resolution requirements for sampling roughness features on the order of 100 nanometers. The confocal system effectively suppresses background light interference from the top surface of the monitoring ridge and the material bulk, improving the correspondence between the measurement signal and the local features of the sidewall.

[0032] 203. The light intensity data of the narrow width monitoring ridge at each sampling location is differentially analyzed with the light intensity data of the wide width monitoring ridge to obtain the differential intensity sequence distributed along the monitoring ridge. In this embodiment, at each sampling location, a confocal photomultiplier tube detector collects the light signal passing through a pinhole. When the laser irradiates the sidewall surface, the interaction between the light and the surface generates reflected and scattered light. In areas where the sidewall is relatively smooth, most of the light energy propagates along the specular reflection direction; in areas where the sidewall has nanoscale roughness and undulations, the light is scattered in different directions by the undulating structure, and the light intensity in the specular reflection direction is weakened. The photomultiplier tube detector is placed in the specular reflection light path and receives the light signal from the focusing plane through the confocal pinhole. The "preset angle range" is an angle interval of ±30 degrees in the specular reflection direction, which covers the main reflected and forward scattered light energy. The photocurrent output by the detector is converted into a digital signal by an amplifier and an analog-to-digital converter. This signal represents the combined light intensity of the sidewall reflection and scattering at that sampling location, in microwatts. At the first sampling location, the light intensity signal of the narrow-width monitoring ridge sidewall is recorded as follows: The light intensity signal of the wide-width monitoring ridge is recorded as The scanning of 200 sampling positions was completed sequentially, and 200 light intensity data were obtained from the narrow-width monitoring ridge. to 200 data points were obtained from wide-width ridge monitoring. to When the roughness at a sampling location on the sidewall is high, scattering at that location is enhanced and specular reflection is weakened, resulting in a relative decrease in the light intensity received by the detector. Conversely, when the sidewall is relatively smooth, specular reflection is strong and scattering is weak, leading to a relative increase in the light intensity received by the detector. Therefore, the fluctuation of the light intensity signal along the monitoring ridge reflects the spatial distribution of the sidewall roughness. For example, the light intensity data for the first five sampling locations on the narrow-width monitoring ridge are 8.5, 7.2, 8.8, 6.5, and 7.9 microwatts, respectively. The fluctuation in these values ​​indicates differences in the sidewall roughness at these locations, with the lowest light intensity at the fourth location corresponding to the highest roughness at that point. The spatial filtering effect of the confocal pinhole ensures that the light intensity signal at each sampling location mainly originates from the local area of ​​the sidewall illuminated by the focused spot. The high-density scanning with a sampling interval of 50 nanometers provides sufficient spatially resolved data for subsequent extraction of the autocorrelation function characteristics of the roughness.

[0033] 204. Calculate the spatial autocorrelation function of the differential intensity sequence to obtain the differential autocorrelation function; In this embodiment, the light intensity data of the narrow-width monitoring ridge and the wide-width monitoring ridge at the same sampling position are subtracted point by point. The difference formula is as follows: , where i is the sampling position number (1 to 200), and ΔI(i) is the differential intensity at the i-th position. For example, the differential intensity at the first position ΔI(1) = 8.5 - 4.5 = 4.0 microwatts, and 200 differential intensity values ​​are calculated sequentially. The "differential intensity sequence" refers to the data sequence formed by arranging these 200 differential intensities in the order of sampling positions. Since the narrow-width monitoring ridge and the wide-width monitoring ridge are formed synchronously, their surface top roughness, material bulk defect density, and light absorption coefficient are completely consistent. In confocal measurement, systematic factors such as laser power fluctuation, objective lens transmittance, detector response characteristics, and ambient light interference have the same amplitude of influence on the light intensity measurement of the narrow-width ridge and the wide-width ridge. Differential operation eliminates these common-mode background signals. The signal retained by the differential intensity sequence comes from the scattering difference caused by the width difference between the narrow-width and wide-width ridges. According to waveguide scattering theory, the narrower the waveguide, the stronger the influence of sidewall roughness on light transmission. Under the same sidewall roughness conditions, light transmitted through a narrow ridge is more significantly affected by sidewall scattering. Therefore, when illuminated by an external laser, the scattering effect produced by the sidewall of a narrow ridge is stronger than that of a wide ridge, resulting in lower specular reflection intensity at the sidewall of a narrow ridge compared to a wide ridge. The differential intensity is positive, and its magnitude and spatial fluctuation characteristics reflect the amplitude and spatial distribution of the sidewall roughness, providing a disturbance-suppressed signal for subsequent autocorrelation function analysis.

[0034] 205. Perform a square root operation on the value of the differential autocorrelation function at the zero point to obtain the root mean square difference intensity; In this embodiment, the spatial autocorrelation function is calculated for the differential intensity sequence. The formula for calculating the autocorrelation function R(k) is: , Where k is the spatial lag step number (0 to N-1), N is the total number of sampling points (200), Σ represents the summation of i from 1 to Nk, and ΔI(i) and ΔI(i+k) are the differential intensities at the i-th and i+k-th positions. The actual spatial distance corresponding to a spatial lag of k steps is k × 50 nanometers. The calculation process is as follows: when k = 0, , We obtain R(0) = 38.6 microwatts²; when k=1 (spatial distance 50 nanometers), , R(1) = 36.2 μW²; when k=2 (spatial distance 100 nm), R(2) = 23.8 μW², the value decreases significantly. 200 autocorrelation function values ​​were calculated sequentially. The "differential autocorrelation function" refers to the curve of the autocorrelation function changing with spatial distance. R(0) has the largest value, representing the overall fluctuation amplitude of the differential intensity sequence; as the spatial interval increases, R(k) gradually decreases. When the spatial interval reaches 100 nm, R(k) decreases to about 62% of R(0), indicating that the correlation of sidewall roughness weakens at a 100 nm spatial scale. The decay rate of the autocorrelation function reflects the spatial periodic characteristics of the sidewall roughness; rapid decay indicates that the roughness changes rapidly over a short distance, while slow decay indicates that the roughness has a longer spatial period. Autocorrelation function analysis transforms discrete light intensity measurements into function curves describing the spatial statistical laws of roughness.

[0035] 206. Perform exponential decay fitting on the differential autocorrelation function and extract the characteristic decay length obtained from the fitting; In this embodiment, the square root operation is performed on the value R(0) of the difference autocorrelation function at the zero point, and the formula is as follows: ,in R(0) represents the root mean square difference intensity, measured in microwatts. R(0) is the mean square value of all data points in the difference intensity sequence, representing the overall fluctuation range of the difference intensity. For example, when R(0) = 38.6 microwatts², The "root mean square differential intensity" is a statistical measure of the overall fluctuation level of the differential intensity sequence. The magnitude of this parameter reflects the overall difference in sidewall light intensity signals between narrow-width and wide-width monitoring ridges. Since differential processing eliminates common-mode signals unrelated to the sidewalls, the root mean square differential intensity focuses on characterizing the differentiated optical response caused by sidewall roughness. Based on the relationship between sidewall scattering and roughness, a larger root mean square differential intensity corresponds to higher sidewall roughness. The zero-point value of the autocorrelation function is converted to a parameter with the same dimensions as light intensity through square root operation, facilitating comparison with the signal-to-noise ratio of the measurement system. When the root mean square differential intensity is significantly greater than the system noise level (approximately 0.1 microwatts), it indicates that the differential signal truly reflects the sidewall roughness characteristics. The root mean square differential intensity is the numerator in the subsequent calculation of the comprehensive roughness index.

[0036] 207. Calculate the ratio of the root mean square difference intensity to the characteristic attenuation length to obtain the roughness comprehensive index, and compare the roughness comprehensive index with the preset roughness range; In this embodiment, the differential autocorrelation function is fitted with exponential decay to extract the characteristic decay length. The fitting function is: , Where u is the spatial distance (nanometers). Let be the characteristic decay length (nanometers), and exp be the natural exponential function. For the k-th sampling point, u = k × 50 nanometers. The least squares method is used for fitting, and adjustments are made... The goal is to minimize the sum of squared errors between the fitted function value and 200 measured R(k) values. The "characteristic attenuation length" is mentioned here. This describes the spatial distance as the autocorrelation function decays to 1 / e (approximately 36.8%) of R(0), representing a characteristic scale characterizing the spatial correlation of sidewall roughness. For example, it can be obtained through fitting... = 108 nanometers. Verification: When u = 108 nanometers, It is approximately 36.8% of R(0). The characteristic decay length reflects the average spatial period of the sidewall roughness undulations, which is relatively small. (e.g., 70 nanometers) indicates dense and fluctuating roughness, with relatively large roughness. (e.g., 150 nm) indicates a gradual change in roughness. According to scattering theory, the shorter the correlation length, the greater the scattering loss. Exponential decay fitting extracts a single characteristic parameter from the complex autocorrelation function curve. This parameter corresponds to the spatial frequency characteristics of the sidewall roughness and is a key indicator for evaluating sidewall quality.

[0037] 208. When the comprehensive roughness index exceeds the preset roughness range, it is determined that the sidewall roughness exceeds the standard; In this embodiment, the ratio of the root mean square difference intensity to the characteristic attenuation length is calculated using the following formula: .For example, = 6.21 microwatts When the roughness is 108 nm, Q = 6.21 / 108 = 0.0575 μW / nm. The "comprehensive roughness index" integrates the amplitude characteristics (root mean square difference intensity) and spatial structure characteristics (characteristic attenuation length) of roughness. When the roughness amplitude increases or the correlation length decreases, the comprehensive index increases, corresponding to more severe scattering loss. Q is compared with a preset roughness range. The preset range is obtained through calibration: During the process development stage, standard samples with known sidewall roughness are prepared. The roughness is measured offline using a scanning electron microscope, ranging from 5 to 12 nm, the correlation length is from 80 to 150 nm, and the transmission loss is measured optically as 0.08 to 0.7 dB / cm. The standard samples are subjected to confocal scanning measurements according to this scheme, the comprehensive roughness index is calculated, and the correspondence between the comprehensive index and transmission loss is established. When the loss requirement is less than 0.5 dB / cm, the corresponding sidewall roughness must be less than 8 nm and the correlation length greater than 100 nm, determining the preset roughness range to be 0.045 to 0.070 μW / nm. The measured Q = 0.0575 microwatts per nanometer is within the acceptable range, indicating that the sidewall roughness is qualified. The comprehensive roughness index integrates two independent parameters into a single judgment criterion, simplifying the quality judgment process.

[0038] 209. Based on the degree of deviation of the comprehensive roughness index from the preset roughness range, calculate the adjustment amount of the etching process parameters, and adjust the etching process parameters of the next batch according to the adjustment amount. In this embodiment, when the overall roughness index exceeds the preset roughness range, it is determined that the sidewall roughness exceeds the standard. If Q is greater than the upper limit of 0.070 microwatts per nanometer, it is determined to exceed the standard. For example, when the measured Q = 0.085 microwatts per nanometer, it exceeds the upper limit. The deviation ΔQ is calculated as Q - 0.070 = 0.015 microwatts per nanometer. The etching process parameters are adjusted using a proportional control strategy. For RF power, the adjustment formula is: , Where α is the response coefficient, and the process experiment calibrated α = 0.12. This represents the current radio frequency power. The negative sign indicates that the power is reduced to weaken the plasma bombardment when the roughness exceeds the limit. For example, = 280 watt-hours, = -0.12×(0.015 / 0.070)×280 = -7.2 watts, the power for the next batch will be adjusted to 272.8 watts. For the process chamber pressure, the adjustment formula is: , Where β = 0.18, the positive sign indicates that increasing the pressure lowers the ion energy. For example, = 4.2 mTorr, = 0.18×(0.015 / 0.070)×4.2=0.16 mTorr, the pressure for the next batch is adjusted to 4.36 mTorr. The process parameters for the next batch are modified based on this adjustment, and the etching is repeated for verification, achieving closed-loop optimization. The quantitative adjustment strategy based on the deviation of the comprehensive roughness index avoids the subjectivity of manual experience-based adjustments, transforming process optimization from qualitative judgment to quantitative control. This shortens the process window convergence cycle, reduces iteration costs, and improves the stability of sidewall quality between batches and the mass production yield of laser chips.

[0039] 210. After depositing the passivation layer, photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the guided mode light intensity signal. The guided mode light intensity signal of the narrow-width monitoring ridge and the guided mode light intensity signal of the wide-width monitoring ridge are compared and processed to determine the uniformity of the passivation layer coverage. When the uniformity of the passivation layer coverage deviates from the preset uniformity threshold, the passivation process parameters are adjusted.

[0040] In this embodiment, step 210 is similar to step 103 in the first embodiment, and will not be described again here.

[0041] In this embodiment, a main waveguide ridge and monitoring ridges of different widths are simultaneously formed on a semiconductor substrate through etching. The monitoring ridges include narrow-width and wide-width monitoring ridges. During etching, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and scattered light intensity signals are collected. The scattered light intensity signals are differentially processed to determine the sidewall roughness. When the sidewall roughness deviates from a preset roughness range, the etching process parameters are adjusted. After passivation layer deposition, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and guided mode light intensity signals are collected. The guided mode light intensity signals are compared to determine the passivation layer coverage uniformity. When the passivation layer coverage uniformity deviates from a preset uniformity threshold, the passivation process parameters are adjusted. This solution enables in-situ monitoring and adaptive adjustment of the sidewall quality and passivation layer coverage state of waveguide structures of different widths during etching and passivation, improving the uniformity and yield of the laser chip waveguide transmission performance and increasing mass production yield.

[0042] The etching control method for the low-loss waveguide of the laser chip in the embodiments of the present invention has been described above. The etching control device for the low-loss waveguide of the laser chip in the embodiments of the present invention is described below. Please refer to [link to relevant documentation]. Figure 3 One embodiment of the etching control device for the low-loss waveguide of the laser chip in this invention includes: The structural building unit 301 is used to simultaneously form a main waveguide ridge and at least two monitoring ridges of different widths on a semiconductor substrate by etching, wherein the at least two monitoring ridges include a narrow-width monitoring ridge and a wide-width monitoring ridge. The roughness monitoring unit 302 is used to apply light excitation to the narrow width monitoring ridge and the wide width monitoring ridge, obtain the scattered light intensity signal during the etching process, and perform differential processing on the scattered light intensity signal of the narrow width monitoring ridge and the scattered light intensity signal of the wide width monitoring ridge to determine the sidewall roughness. When the sidewall roughness deviates from the preset roughness range, the etching process parameters are adjusted. The uniformity monitoring unit 303 is used to apply photoexcitation to the narrow-width monitoring ridge and the wide-width monitoring ridge after the passivation layer is deposited, obtain the guided mode light intensity signal, and compare the guided mode light intensity signal of the narrow-width monitoring ridge with the guided mode light intensity signal of the wide-width monitoring ridge to determine the uniformity of the passivation layer coverage. When the uniformity of the passivation layer coverage deviates from the preset uniformity threshold, the passivation process parameters are adjusted.

[0043] In this embodiment of the invention, a main waveguide ridge and monitoring ridges of different widths are simultaneously formed on a semiconductor substrate by etching. The monitoring ridges include narrow-width and wide-width monitoring ridges. During etching, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and scattered light intensity signals are collected. The scattered light intensity signals are then differentially processed to determine the sidewall roughness. When the sidewall roughness deviates from a preset roughness range, the etching process parameters are adjusted. After passivation layer deposition, photoexcitation is applied to both the narrow-width and wide-width monitoring ridges, and guided mode light intensity signals are collected. The guided mode light intensity signals are then compared to determine the passivation layer coverage uniformity. When the passivation layer coverage uniformity deviates from a preset uniformity threshold, the passivation process parameters are adjusted. This solution enables in-situ monitoring and adaptive adjustment of the sidewall quality and passivation layer coverage state of waveguide structures of different widths during etching and passivation processes, improving the uniformity and yield of the transmission performance of laser chip waveguides and increasing mass production yield.

[0044] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system or system / unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0045] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling the etching of a low-loss waveguide in a laser chip, characterized in that, The etching control method for the low-loss waveguide of the laser chip includes: A main waveguide ridge and at least two monitoring ridges of different widths are simultaneously formed on a semiconductor substrate by etching, wherein the at least two monitoring ridges include a narrow-width monitoring ridge and a wide-width monitoring ridge. Photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the scattered light intensity signal during the etching process. The scattered light intensity signal of the narrow-width monitoring ridge and the scattered light intensity signal of the wide-width monitoring ridge are differentially processed to determine the sidewall roughness. When the sidewall roughness deviates from the preset roughness range, the etching process parameters are adjusted. After depositing the passivation layer, photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the guided mode light intensity signal. The guided mode light intensity signal of the narrow-width monitoring ridge is compared with the guided mode light intensity signal of the wide-width monitoring ridge to determine the uniformity of the passivation layer coverage. When the uniformity of the passivation layer coverage deviates from the preset uniformity threshold, the passivation process parameters are adjusted.

2. The etching control method for low-loss waveguides of laser chips according to claim 1, characterized in that, The step of applying photoexcitation to the narrow-width monitoring ridge and the wide-width monitoring ridge, acquiring the scattered light intensity signal during the etching process, and performing differential processing on the scattered light intensity signal of the narrow-width monitoring ridge and the scattered light intensity signal of the wide-width monitoring ridge includes: Lateral incident light was simultaneously applied to the sidewall regions of the narrow-width monitoring ridge and the wide-width monitoring ridge at multiple sampling positions during the preset etching process, and scattered light intensity data within a preset angle range were collected respectively. The scattered light intensity data of the narrow-width monitoring ridge at each sampling location is differentially analyzed with the scattered light intensity data of the wide-width monitoring ridge to obtain the differential intensity sequence distributed along the monitoring ridge. The spatial autocorrelation function of the differential intensity sequence is calculated to obtain the differential autocorrelation function.

3. The etching control method for low-loss waveguides of laser chips according to claim 2, characterized in that, The determination of sidewall roughness includes: The root mean square difference intensity is obtained by taking the square root of the value of the differential autocorrelation function at the zero point. The differential autocorrelation function is subjected to exponential decay fitting, and the characteristic decay length obtained from the fitting is extracted; The ratio of the root mean square difference intensity to the characteristic attenuation length is calculated to obtain the roughness comprehensive index, and the roughness comprehensive index is compared with the preset roughness range.

4. The etching control method for low-loss waveguides of laser chips according to claim 3, characterized in that, When the sidewall roughness deviates from the preset roughness range, adjusting the etching process parameters includes: When the roughness comprehensive index exceeds the preset roughness range, it is determined that the sidewall roughness exceeds the standard; Based on the degree of deviation of the comprehensive roughness index from the preset roughness range, the adjustment amount of the etching process parameters is calculated, and the etching process parameters of the next batch are adjusted according to the adjustment amount.

5. The etching control method for low-loss waveguides of laser chips according to claim 1, characterized in that, After the passivation layer is deposited, photoexcitation is applied to the narrow-width monitoring ridge and the wide-width monitoring ridge to obtain the guided mode light intensity signal. The process of comparing the guided mode light intensity signal of the narrow-width monitoring ridge with the guided mode light intensity signal of the wide-width monitoring ridge includes: Photoexcitation was applied to the narrow-width monitoring ridge and the wide-width monitoring ridge respectively, and the corresponding guided mode transmitted light intensity was collected; The ratio of the transmitted light intensity of each guided mode to the corresponding incident light intensity is calculated to obtain the narrow-width monitoring ridge transmittance and the wide-width monitoring ridge transmittance, respectively. The transmittance ratio is obtained by calculating the ratio of the narrow ridge transmittance to the wide ridge transmittance.

6. The etching control method for low-loss waveguides of laser chips according to claim 5, characterized in that, The determination of passivation layer coverage uniformity includes: The transmittance ratio is compared with a preset standard transmittance ratio to calculate the transmittance ratio deviation; The narrow ridge transmittance is compared with the preset narrow ridge standard transmittance, and the wide ridge transmittance is compared with the preset wide ridge standard transmittance, and the narrow ridge transmittance deviation and wide ridge transmittance deviation are calculated respectively. Based on the transmittance ratio deviation, narrow ridge transmittance deviation, and wide ridge transmittance deviation, when the transmittance ratio deviation exceeds a preset uniformity threshold, or when either the narrow ridge transmittance deviation or the wide ridge transmittance deviation exceeds a preset uniformity threshold, the passivation layer coverage uniformity is determined to be abnormal.

7. The etching control method for low-loss waveguides of laser chips according to claim 6, characterized in that, The passivation process parameters include deposition process parameters and annealing process parameters. When the uniformity of the passivation layer coverage deviates from a preset uniformity threshold, adjusting the passivation process parameters includes: When the transmittance ratio deviation is positive, the deposition process parameter values ​​for the next batch are reduced; when the transmittance ratio deviation is negative, the deposition process parameter values ​​for the next batch are increased. When both the narrow ridge transmittance deviation and the wide ridge transmittance deviation are negative, the annealing process parameters for the next batch are increased.

8. An etching control device for a low-loss waveguide of a laser chip, characterized in that, The etching control device for the low-loss waveguide of the laser chip includes: A structural building block for simultaneously forming a main waveguide ridge and at least two monitoring ridges of different widths on a semiconductor substrate by etching, wherein the at least two monitoring ridges include a narrow-width monitoring ridge and a wide-width monitoring ridge; The roughness monitoring unit is used to apply light excitation to the narrow width monitoring ridge and the wide width monitoring ridge, obtain the scattered light intensity signal during the etching process, and perform differential processing on the scattered light intensity signal of the narrow width monitoring ridge and the scattered light intensity signal of the wide width monitoring ridge to determine the sidewall roughness. When the sidewall roughness deviates from the preset roughness range, the etching process parameters are adjusted. The uniformity monitoring unit is used to apply photoexcitation to the narrow-width monitoring ridge and the wide-width monitoring ridge after the passivation layer is deposited, obtain the guided mode light intensity signal, and compare the guided mode light intensity signal of the narrow-width monitoring ridge with the guided mode light intensity signal of the wide-width monitoring ridge to determine the uniformity of the passivation layer coverage. When the uniformity of the passivation layer coverage deviates from the preset uniformity threshold, the passivation process parameters are adjusted.

9. An etching control device for a low-loss waveguide of a laser chip, characterized in that, The etching control device for the low-loss waveguide of the laser chip includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the etching control device for the low-loss waveguide of the laser chip to perform the various steps of the etching control method for the low-loss waveguide of the laser chip as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the etching control method for the low-loss waveguide of the laser chip as described in any one of claims 1-7.

Images