WO2020048148A1

WO2020048148A1 - Surface defect measurement method based on spectral confocal sensor

Info

Publication number: WO2020048148A1
Application number: PCT/CN2019/086079
Authority: WO
Inventors: 居冰峰; 杜慧林; 张文浩; 孙安玉
Original assignee: 浙江大学
Priority date: 2018-09-03
Filing date: 2019-05-08
Publication date: 2020-03-12
Also published as: CN109283184A

Abstract

A surface defect measurement method based on a spectral confocal sensor, and an apparatus therefor. The measurement method comprises: mounting a plane mirror on an angular displacement stage, setting the tilt angle of the plane mirror (101), adjusting the distance between a spectral confocal sensor and the plane mirror (102), completing acquisition of spectral peak intensity data at the set tilt angle (103), and adjusting the angular displacement stage to obtain a different tilt angle, until acquisition of spectral peak intensity data at all specified tilt angles is completed (104); constructing a "steepness-distance-spectral peak intensity" characteristic curve (105); scanning and measuring a workpiece to be measured (106), and acquiring spectral peak intensity data of light reflected back from the surface of the workpiece to be measured (107), calculating the reflectivity of the measured point by the spectral peak intensity data in combination with the "steepness-distance-spectral peak intensity" characteristic curve (108); setting a reflectivity threshold, and identifying an area having an abrupt change in reflectivity by the set reflectivity threshold (110), to distinguish a defect area from a normal area, and completing defect locating and contour extraction (111). The measurement method provides a new method for the detection of surface defects of optical elements by establishing a mathematical model between the spectral peak intensity and the surface reflectivity and locating surface defects by means of the reflectivity.

Description

Surface defect measurement method based on spectral confocal sensor

Technical field

The invention relates to the field of surface defect measurement, in particular to a surface defect measurement method based on a spectral confocal sensor.

Background technique

Ultra-precision optical components are widely used in civil and military defense fields. Typical applications in the civilian field include lithography machine exposure mirrors, camera zoom lenses, optical instrument lenses, medical endoscopes and progressive lenses, etc .; military and defense applications include astronomical telescopes such as the Hubble Telescope, and laser weapons. , Infrared thermal imager, laser fusion, etc. The processing of ultra-precision optical components is inseparable from precision measurement technology. At the current stage, the processing of ultra-precision optical components requires not only higher-precision surface measurement technology, but also requirements for the detection of surface defects. Surface defects have become an important index for evaluating the quality of optical components.

When the light beam passes through the optical element with defects, local defects will cause beam scattering and energy loss, and may produce undesired diffraction, energy absorption, harmful flares, and film damage. In optical imaging applications, any Defects on the surface of an optical element will cause local brightness unevenness, which will affect the overall imaging quality. Therefore, the detection of surface defects is a necessary inspection process after high-end optical elements are processed. The distribution of surface defects on the surface scale of optical elements has great randomness. The microstructure of defects is different and has discrete structural features, so it is difficult to measure.

Traditional surface measurement equipment is difficult to measure surface defects. The reasons include the following:

1) The vertical size of the defect is small. The vertical size of some defects is smaller than the vertical resolution of the measurement sensor. In this case, it is difficult to distinguish the defects by ordinary topography measurement;

2) The horizontal size of the defect is small. When the lateral dimension of the defect is smaller than the lateral resolution of the sensor, the sensor cannot detect the presence of the defect. Taking the wavefront interferometer as an example, its longitudinal resolution is extremely high (up to λ / 100), but its lateral resolution is low. An interferometer equipped with a 1024 × 1024 pixel CCD is used to detect an optical element with a diameter of 1m. Its lateral resolution is only about 1mm, and it is difficult to detect defects at the sub-millimeter level.

3) Although some defects are large in scale, the traditional image processing methods are difficult to distinguish, and the manual detection workload is large, and it is easy to cause false detection and missed detection.

At present, the main methods of surface defect measurement include filter detection method, scattered light energy analysis method, laser spectrum method, and visual method. Most of the existing surface defect measurement methods require the workpiece to be transported to special equipment for measurement. The handling of high-end large-scale optical components is difficult, and the handling process has the risk of causing additional defects or damage. In addition, the handling process will cause secondary clamping errors, which is disadvantageous to the processing process. The visual method relies on manual screening with a magnifying glass or a microscope, which is less efficient and has the risk of missed and false detections.

Aiming at the above-mentioned defect detection needs and problems, the present invention establishes a mathematical model of "steepness-distance-spectral peak intensity" by using spectral peak intensity signals previously considered to have no engineering practical value in a spectral confocal sensor. Calculate the surface reflectance of the workpiece, and then complete the positioning and extraction of surface defects, which can achieve the parallel detection of the surface shape and defects based on the spectral confocal sensor.

Summary of the Invention

The present invention provides a surface defect measurement device and method based on a spectral confocal sensor based on the needs of optical element surface defect measurement. The invention groundbreakingly uses the spectral peak intensity signal of the spectral confocal sensor, experimentally establishes a mathematical model of the "steepness-distance-spectral peak intensity" of the spectral confocal probe, and realizes by calculating the reflectance of the surface of the workpiece Surface defect location and contour extraction.

The present invention uses the following technical solutions to achieve the above functions:

A method for measuring surface defects based on a spectral confocal sensor is specifically implemented as follows:

Step 1. Install the flat mirror (202) on the angular stage (203), and set the tilt angle of the flat mirror;

Step 2. Adjust the distance (102) between the spectral sensor and the workpiece to be measured, and complete the collection of spectral peak intensity data (103) at the set inclination angle;

Step 3. Adjust the angular stage to different inclination angles (101), and then repeat steps 1 and 2 until the acquisition of spectral peak intensity data at all specified inclination angles is completed;

Step 4. Use the collected spectral peak intensity data to construct a "steepness-distance-spectral peak intensity" characteristic curve (105).

Step 5. Scan and measure the workpiece to be tested (304), collect spectral peak intensity data (107) of the surface light of the workpiece, and use this data to calculate the reflectance matrix of the surface of the workpiece to be measured.

Step 6. Set the reflectance threshold, calculate the reflectance (108) of the measured point by combining the spectral peak intensity data (107) and the spectral peak intensity characteristic curve (105), and infer the position of the abnormal point from the abnormal point of reflectance Information to obtain the position set of abnormal reflectance points; by setting the reflectance threshold (110) area to identify the abrupt change of the reflectance area, thereby distinguishing the defective area from the normal area and completing the defect location and contour extraction (109).

The white light source (301) emits white light and enters the high-dispersion lens (303) through the optical fiber (209). After the light passes through the high-dispersion lens (303), light of different wavelengths will be focused at different positions on the optical axis. 304) The reflected light at the position of the measured point is returned to the spectrometer (208) to form a spectral peak, and the spectral peak intensity data is collected by a spectral peak intensity data acquisition device.

The spectral peak intensity data acquisition device includes a data acquisition card (206), a spectrometer (208), and a high-dispersion lens (303). The spectrometer (208) converts the returned light collected by the spectral confocal sensor (201) into a distance signal. And the light intensity signal, and then the distance signal and the light intensity signal are converted into two 0 to 10V analog voltage signals (207) and output. Then, the analog-to-digital conversion module of the data acquisition card (206) is used to convert the two analog voltage signals (207) into two digital signals for collection, and send them to the industrial computer (204) for data processing to obtain spectral peak intensity data information.

The "steepness-distance-spectral peak intensity" characteristic curve (205) of the spectral peak intensity data and the steepness and distance of the surface to be measured is established, and the surface reflectance of the measurement point is calculated from the spectral peak intensity signal using a mathematical experimental model. :

The "steepness-distance-spectrum peak intensity" characteristic curve and mathematical experimental model are set out from experiments, and a large number of data are used to establish the characteristic curve and mathematical experimental model. The experimental process is as follows: the plane mirror (202) is mounted on the angular stage (203), and the angular stage (203) is set to a minimum inclination range of 0 ° to a maximum γ _max with an interval δγ; for each inclination, let The spectral confocal sensor (201) moves up and down, records the relationship between the measured distance and the spectral peak intensity of each group, and draws a "steepness-distance-spectral peak intensity" characteristic curve (205).

The "steepness-distance-spectral peak intensity" characteristic curve and the mathematical experimental model, the spectral peak intensity i is expressed as a function of the local reflectance r, the local tilt angle γ, and the displacement d measured by the probe:

i = r · f (γ, d)

The "steepness-distance-spectral peak intensity" characteristic curve, whose spectral peak intensity increases as the distance measured by the spectral confocal sensor increases, and in the middle part of the sensor range, the spectral peak intensity, distance, and inclination angle are in the middle There is a monotonically decreasing relationship. An approximate expression of f (γ, d) is obtained by polynomial fitting.

The approximate expression of f (γ, d) in the relationship between the reflectivity and the inclination and distance of the measurement point is:

f (γ, d) = p ₀₀ + p ₁₀ · γ + p ₀₁ · d + p ₂₀ · γ ² + p ₁₁ · γ · d

Among them, p ₀₀ , p ₁₀ , p ₀₁ , p ₂₀ , and p ₁₁ are coefficients obtained by fitting experimental data;

γ is the inclination at the measurement point;

d is the measured distance.

The device for measuring the peak intensity signal of a surface point includes a spectral confocal sensor (201), a voice coil motor (405), an industrial computer (204), a motion controller (401), and a motor driver (402); a spectrum The confocal sensor (201) is fixed to the voice coil motor (405) moving in the vertical direction, and the workpiece is placed on the X-axis (403) and the Y-axis (404) of the linear linear translation stage in the horizontal direction. The industrial computer (204) sends a control instruction to the motion controller (401). The motion controller controls the motor driver (402) to drive the movement of the X axis (403) and the Y axis (404) to move the workpiece in the horizontal plane and drive the voice coil motor ( 405) Move up and down to ensure that the measurement point is within the range of the spectral confocal sensor to complete the scanning movement; collect two analog voltage signals (207) and three encoder signals (406) through the data acquisition card (206) and send them to The industrial computer (204) completes the acquisition of the spectral intensity signals of the corresponding points on the surface of the workpiece.

For the positioning and contour extraction of the surface defect, the reflectance threshold (110) is set to distinguish the abrupt change of the reflectance region, so as to distinguish the defective part from the normal part, and complete the positioning and contour extraction of the surface defect. Changing the reflectance threshold controls the sensitivity of surface defects.

The main significance and advantages of the present invention are:

The invention proposes the engineering significance of the spectral peak intensity of the spectral confocal sensor, and establishes a mathematical model between the spectral peak intensity and the surface reflectance experimentally, and uses the reflectance to locate the surface defect, which is an optical element surface defect. Detection provides a new method.

The invention can realize the parallel detection of the surface shape and the surface defect, at the same time avoid human error or secondary clamping error, can greatly improve the detection efficiency and accuracy of the optical element, and provide a strong guarantee for the quality control of the optical element.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a flow chart for measuring surface defects based on a spectral confocal sensor.

Figure 2 is a schematic diagram of the system used to establish the "steepness-distance-spectral peak intensity" characteristic curve.

Figure 3 is a schematic diagram of a system for measuring defects on a workpiece surface.

FIG. 4 is a schematic diagram of a three-degree-of-freedom mobile platform for collecting peak intensity data of a workpiece surface.

Fig. 5 is a graph showing the measurement results of surface defects of a rectangular mirror with a defective surface.

In the figure: spectral confocal sensor (201), plane mirror (202), angular stage (203), industrial computer (204), "steepness-distance-spectral peak intensity" characteristic curve (205), data acquisition card (206), pseudo-voltage signal (207), spectrometer (208), optical fiber (209), white light source (301), high-dispersion lens (303), workpiece (304), motion controller (401), motor driver (402) ), X-axis of linear electric stage (403), Y-axis of linear electric stage (404), voice coil motor (405), signal of modular encoder (406);

detailed description

The present invention is further described in detail below with reference to the accompanying drawings and embodiments.

The embodiment of the present invention relates to a method for measuring surface defects based on a spectral confocal sensor, which mainly includes the establishment of a "steepness-distance-spectral peak intensity" characteristic curve and the collection and processing of spectral peak intensity data of a workpiece to be measured.

A measurement process of a surface defect measurement method based on a spectral confocal sensor is shown in FIG. 1. First set the tilt angle of the plane mirror, adjust the distance between the spectral sensor and the workpiece to be measured, and collect the spectral peak intensity data. Repeat the above steps until the data acquisition at different inclination is completed. Then, the “steepness-distance-spectrum peak intensity” characteristic curve is constructed using the collected spectral peak intensity data. Then measure the peak intensity data of the workpiece to be measured, and use this data to calculate the reflectance matrix on the surface of the workpiece. Finally, the reflectance threshold is set, and the position information of the reflectivity abnormal point is inferred to obtain the position set of the reflectance abnormal point, so as to realize the surface defect positioning and contour extraction.

As shown in Figure 2, to establish the "steepness-distance-spectral peak intensity" characteristic curve, firstly, the plane mirror needs to be fixed on an angular stage, and the tilt angle γ of the plane mirror is adjusted to 0 °. The computer, controller, driver and spectrometer are turned on. , Adjust the voice coil motor so that the measurement point is at one end of the range of the spectral confocal sensor. Control the voice coil motor to drive the spectral confocal sensor to make continuous movement to the other end of the sensor range, and collect distance data d and spectral peak intensity data i. After completing the acquisition at 0 °, adjust the tilt angle of the plane mirror to increase δγ, and then adjust the workpiece so that the measurement point is at the end of the range. Also collect the distance and spectral peak intensity data, and repeat the above steps until the maximum required measurement tilt angle γ _{max is} reached. Each γ corresponds to a set of distance and spectral peak intensity data. The obtained inclination angle γ, spectral peak intensity i, and distance d are plotted as a "steep-distance-spectral peak intensity" characteristic curve and a second-order polynomial fitting is performed. The formula is as follows:

γ is the inclination at the measurement point;

d is the measured distance.

In this embodiment, the plane mirror produced using Mitutoyo standard flat crystal surface aluminized; taken δγ of 0.5 °, γ _max is 15 °; CL2 spectral sensor using the French company STIL confocal sensor. In this embodiment, the coefficients obtained by the polynomial fitting are as follows:

Table 1 Polynomial fitting coefficients

As shown in FIG. 3, after obtaining the characteristic curve, scan the workpiece to be measured to obtain the position matrix P, the distance matrix D and the spectral peak intensity matrix N of the surface point of the workpiece to be measured, where the position data in P and the distance in D And the peak intensity data in N correspond one-to-one. The steepness matrix S is calculated from the distance matrix D, and the steepness matrix S and the spectral peak intensity matrix N are substituted into the characteristic curve to calculate the reflectance matrix R of the workpiece surface point. Normalize the reflectance matrix and set the reflectance threshold to I _thres . A region with a reflectance higher than I _thres is determined as a normal region, and a region with a reflectance lower than I _thres is determined as a defective region. Since the position matrix P and the reflectance matrix R also have a one-to-one correspondence, the corresponding position data can be identified through the data with abnormal reflectance, and the position point set of the abnormal point can be found to realize the detection of surface defects and the contour extraction.

As shown in FIG. 4, a surface defect measurement device based on a spectral confocal sensor is specifically installed as follows: a spectral confocal probe (201) is fixed on a voice coil motor (405), and a workpiece (303) to be tested is placed on a linear motor drive The Y-axis is mounted on the X-axis driven by a linear motor. During measurement, the industrial computer (204) sets the X-axis measurement stroke X _max , the X-axis scanning step δX, and the Y-axis measurement stroke Y _max . The industrial computer sends instructions to the motion controller (401), and the motion controller sends motion instructions to the motor driver (402) to drive the X-axis, Y-axis, and voice coil motors to move. After the X-axis step δX, the Y-axis continuously moves Y _{max to} realize the scanning and data acquisition of a line segment. After the Y axis moves Y _max , the X axis steps δX, and the Y axis moves Y _{max in the} opposite direction to complete the scan acquisition. Repeat the above steps until the scanning motion of the entire workpiece is completed. During the measurement, the vertical movement of the voice coil motor (405) ensures that the measured point is always within the range of the spectral confocal probe. The data acquisition card (206) collects the distance voltage signal (207) and light intensity voltage signal output by the spectrometer in real time. (406). Simultaneously collect the three-axis grating encoder signals output by the X-axis, Y-axis, and the voice coil motor. After the signal is collected, it is sent to the industrial computer for signal processing and data storage in the industrial computer.

As shown in FIG. 5, the workpiece to be measured in this embodiment is a rectangular mirror in FIG. 5 (a), and a surface defect of the surface is slightly corroded by the alkali solution. Figure 5 (b) shows the actual surface shape measured by the grid scanning method. As shown in the figure, the surface defects of the rectangular workpiece cannot be distinguished only by the surface shape data. The reason is that the vertical dimension of the defect is small and cannot form abrupt changes . Figure 5 (c) shows the steepness matrix obtained from the distance data, and then substitutes the steepness matrix and the spectral peak intensity matrix into the "steepness-distance-spectral peak intensity" characteristic curve to calculate the surface reflectance. The data, as shown in the figure, can clearly distinguish the area of abnormal reflectivity. Figure 5 (d) is the reflectance data when the reflectance threshold is set to 0.8, as shown in the figure, accurate and effective surface defect location and contour extraction are realized.

The embodiment described above is only a preferred solution of the present invention, but it is not intended to limit the present invention. Those of ordinary skill in the relevant technical field may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, any technical solution obtained by adopting an equivalent replacement or equivalent transformation falls within the protection scope of the present invention.

Claims

A surface defect measurement method based on a spectral confocal sensor, which is characterized in that a plane reflector (202) is mounted on an angular stage (203), the distance (102) between the spectral sensor and the workpiece to be measured is adjusted, and the setting is completed Collection of spectral peak intensity data (103) under the inclination angle, adjust the angular stage to different inclination angles (101) until the acquisition of spectral peak intensity data at all specified inclination angles (103) is completed; use the collected spectral peak intensity data to construct the " "Slope-distance-spectral peak intensity" characteristic curve (105); scan the workpiece to be measured (304), collect spectral peak intensity data (107) of the surface light of the workpiece, and calculate the reflectance matrix of the surface of the workpiece to be measured; Set the reflectance threshold, calculate the reflectance (108) of the measured point by combining the spectral peak intensity data (107) and the spectral peak intensity characteristic curve (105), and infer the position information of the abnormal point from the reflectance abnormal point to obtain the reflection. The location set of the abnormality points of the rate; identify the abrupt change of the reflectivity by setting the reflectance threshold (110) area, so as to distinguish the defect area from the normal area and complete the defect location and contour extraction (109).
The method for measuring surface defects based on a spectral confocal sensor according to claim 1, characterized in that the white light source (301) emits white light and passes through the optical fiber (209) to the high-dispersion lens (303), and the light passes through the lens Light of different wavelengths will be focused at different positions on the optical axis, and the reflected light at the position of the measured point of the workpiece (304) will be returned to the spectrometer (208) to form a spectral peak, and the spectral peak intensity data will be collected by a spectral peak intensity data acquisition device.
The method for measuring surface defects based on a spectral confocal sensor according to claim 2, characterized in that the spectral peak intensity data acquisition device comprises a data acquisition card (206), a spectrometer (208), and a high-dispersion lens ( 303), the spectrometer (208) converts the returned light collected by the spectral confocal sensor (201) into a distance signal and a light intensity signal, and then converts the distance signal and the light intensity signal into two analog voltage signals of 0 to 10V (207) Output; then use the analog-to-digital conversion module of the data acquisition card (206) to convert the two analog voltage signals (207) into two digital signals for acquisition, and send to the industrial computer (204) for data processing to obtain spectral peak intensity data information.
The method for measuring surface defects based on a spectral confocal sensor according to claim 3, characterized in that a "steepness-distance-spectral peak intensity" characteristic curve of the spectral peak intensity data and the steepness and distance of the surface to be measured is established ( 205), using the mathematical experimental model of the characteristic curve, to calculate the surface reflectance of the measurement point through the spectral peak intensity signal calculation, specifically:

Starting from the experimental angle, fit the characteristic curve and establish a mathematical experimental model through the experimental data: Install the plane reflector (202) on the angular stage (203), and set the angular stage (203) to a minimum inclination range of 0 ° To the maximum γ max and the interval δγ; for each inclination, make the spectral confocal sensor (201) move up and down, record the relationship between the measured distance and the peak intensity of each group, and draw the "steepness-distance-spectrum peak intensity" Characteristic curve (205).
The method for measuring surface defects based on a spectral confocal sensor according to claim 4, characterized in that γ max is 15 °; and the interval δγ is 0.5 °.
The method for measuring surface defects based on a spectral confocal sensor according to claim 4 or 5, characterized in that the peak intensity i of the spectrum is expressed as the reflectance r at the measurement point, the inclination angle γ at the measurement point, and the distance d measured by the probe The function:

i = r · f (γ, d).
The method for measuring surface defects based on a spectral confocal sensor according to claim 6, characterized in that the intensity of the spectral peak increases as the distance measured by the spectral confocal sensor increases, and in the middle part of the sensor range, the spectral peak There is a monotonically decreasing relationship between intensity, distance, and inclination. The approximate expression of f (γ, d) in the relationship between reflectance and the inclination and distance of the measurement point is obtained by polynomial fitting:

f (γ, d) = p 00 + p 10 · γ + p 01 · d + p 20 · γ 2 + p 11 · γ · d;

Among them, p 00 , p 10 , p 01 , p 20 , and p 11 are coefficients obtained by fitting experimental data; γ is the inclination at the measurement point; and d is the distance measured by the probe.
The measuring device used in a surface defect measurement method based on a spectral confocal sensor according to claim 7, characterized in that

The spectral confocal probe (201) is fixed on the voice coil motor (405), the workpiece (303) to be measured is placed on the Y axis driven by the linear motor, and the Y axis is mounted on the X axis driven by the linear motor; (204) Set the X-axis measurement stroke X max , the X-axis scan step δX, and the Y-axis measurement stroke Y max ; the industrial computer sends instructions to the motion controller (401), and the motion controller sends motion instructions to the motor driver ( 402) Drive X axis, Y axis and voice coil motor movement; after X axis step δX, Y axis continuously moves Y max to realize scanning and data acquisition of one line segment; after Y axis motion Y max , X axis step δX, Y Axial and reverse movement Y max to complete the scanning acquisition; repeat the above steps until the entire workpiece is scanned and detected; during the measurement, the vertical movement of the voice coil motor (405) ensures that the measured point is always within the range of the spectral confocal probe. The data acquisition card (206) collects the distance voltage signal (207) and light intensity voltage signal (406) output by the spectrometer in real time, and simultaneously collects the three-way grating encoder signals output by the X-axis, Y-axis, and voice coil motor; the signals are collected After being conveyed to the industrial computer, complete the workpiece list Spectral intensity of the signal corresponding to the collection point.
The method for measuring surface defects based on a spectral confocal sensor according to claim 7, characterized in that the workpiece to be measured is scanned and measured to obtain a position matrix P, a distance matrix D, and a spectral peak intensity matrix N of the surface points of the workpiece to be measured , Where the position data in P corresponds to the distance in D and the spectral peak intensity data in N; the steepness matrix S is calculated from the distance matrix D, and the steepness matrix S and the spectral peak intensity matrix N are substituted into the characteristic curve Calculate the reflectivity matrix R of the workpiece surface points; normalize the reflectivity matrix and set the reflectivity threshold to I thres . The area with a reflectance higher than I thres is determined as a normal area, and the area with a reflectance lower than I thres is determined . It is determined as a defective area; since the position matrix P and the reflectance matrix R also have a one-to-one correspondence relationship, the corresponding position data can be identified through the data with abnormal reflectance, and the position point set of the abnormal point can be found to achieve surface defects. Detection and contour extraction.