High-precision single-side polishing device, method and system for ultrathin optical element

Abstract

The invention discloses a high-precision single-face polishing device, method and system for an ultrathin optical element, and belongs to the technical field of optical element polishing, and the technical scheme is characterized by comprising the steps that deformation quantity distribution data of the ultrathin optical element are obtained according to voltage data collected by a thin film sensor; according to the deformation quantity distribution data and a preset reduced-order model, obtaining predicted deformation quantity distribution data, the reduced-order model being obtained according to training; according to the predicted deformation quantity distribution data, control parameters are adjusted, and the control parameters comprise the negative pressure of the adsorption area. The optical element is fixed through the vacuum adsorption chuck comprising the adsorption area and the silica gel deformation layer, and the deformation quantity of the optical element is accurately monitored through the positive sensor and the negative sensor arranged in the silica gel deformation layer; the problem that optical elements are prone to deformation and breakage in a traditional disc feeding method is avoided, monitoring errors caused by the temperature drift effect are avoided, control parameters are adjusted finally based on deformation quantity, and polishing precision is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of optical element polishing, and more particularly to a high-precision single-side polishing device, method and system for ultra-thin optical elements. Background Art

[0002] Ultra-thin optical components are core components in high-end fields such as semiconductor chips, lidar, and aerospace remote sensing. The surface accuracy of these components directly determines the imaging quality and energy transmission efficiency of the optical system. However, existing clamping methods cannot achieve high-precision polishing requirements. For example, the wax bonding method uses thermosetting wax to bond the component to the carrier, but the thermal stress generated during the solidification and melting of the wax layer can cause irreversible changes in the back surface of the component's lower plate. Mechanical fixtures rely on rigid claws to fix the edges, which can easily cause the optical component to break. Therefore, the existing technology has shortcomings. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a high-precision single-sided polishing device, method and system for ultra-thin optical components, which avoids the problems of deformation and cracking of optical components by setting buffer stress in the adsorption area and silicone deformation layer in the vacuum adsorption chuck.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A high-precision single-side polishing device for ultra-thin optical elements, the device comprising a vacuum adsorption chuck, the vacuum adsorption chuck comprising:

[0006] An adsorption area, the adsorption area includes a main adsorption area and an edge buffer ring, the main adsorption area is located in the center of the chuck, the edge buffer ring is located on the periphery of the main adsorption area, and a micropore array is distributed on the adsorption area;

[0007] A silicone deformable layer, the silicone deformable layer is located on the surface of the adsorption area, the silicone deformable layer includes a thin film sensor, the thin film sensor includes a positive sensor and a negative sensor, the positive sensor and the negative sensor are arranged alternately, and the positive sensor and the negative sensor have opposite responses to deformation in the same direction.

[0008] The present invention provides a high-precision single-side polishing method for ultra-thin optical elements, which is applied to the above-mentioned high-precision single-side polishing device for ultra-thin optical elements, comprising:

[0009] Obtaining deformation distribution data of the ultra-thin optical element according to the voltage data collected by the thin film sensor;

[0010] Obtaining predicted shape variable distribution data based on the shape variable distribution data and a preset reduced-order model, wherein the reduced-order model is obtained through training;

[0011] According to the predicted deformation distribution data, control parameters are adjusted, wherein the control parameters include the negative pressure of the adsorption area.

[0012] As a further improvement of the present invention, the ultra-thin optical element includes a plurality of sub-regions, and the deformation distribution data of the ultra-thin optical element is obtained based on the voltage data collected by the thin film sensor, including:

[0013] Obtaining a voltage difference corresponding to each sub-region according to the voltage data, wherein each sub-region corresponds to a sensor pair, and each sensor pair includes a positive sensor and a negative sensor;

[0014] Obtaining a deformation amount corresponding to each sub-region according to a voltage difference corresponding to each sub-region;

[0015] The shape variable distribution data is obtained according to the shape variable corresponding to each sub-region and a triangulation algorithm.

[0016] As a further improvement of the present invention, obtaining the shape variable distribution data according to the shape variable corresponding to each sub-region and the triangulation algorithm includes:

[0017] For the deformation amount corresponding to each sub-region, calculating the corresponding curvature weighting factor;

[0018] Obtaining a weighted point set according to the deformation amount corresponding to each sub-region and the curvature weighting factor;

[0019] The deformation variable distribution data is obtained according to the weighted point set and the triangulation algorithm.

[0020] As a further improvement of the present invention, obtaining the shape variable distribution data according to the weighted point set and the triangulation algorithm includes:

[0021] For each coordinate point in the weighted point set, performing an iterative operation, the iterative operation comprising determining a circumscribed circle including the coordinate point, updating an edge list and a triangle list based on a distance between the coordinate point and a center of the circumscribed circle, until a predetermined termination condition is met, and outputting all triangles in the current triangle list;

[0022] The shape variable distribution data is obtained according to all triangles in the current triangle list.

[0023] As a further improvement of the present invention, obtaining predicted shape variable distribution data based on the shape variable distribution data and a preset reduced-order model includes:

[0024] Projecting the deformation variable distribution data onto a main mode to obtain a main mode coefficient vector, wherein the main mode is obtained according to the reduced-order model;

[0025] Obtaining predicted main modal coefficients according to current control parameters and the main modal coefficient vector;

[0026] The predicted deformation variable distribution data is obtained according to the predicted main modal coefficients.

[0027] As a further improvement of the present invention, the reduced-order model is obtained according to training, including:

[0028] Obtain historical deformation data and historical control parameter data;

[0029] Obtaining multiple main modes based on the historical deformation data and principal component analysis method;

[0030] The reduced-order model is obtained according to the historical control parameter data and the plurality of main modes.

[0031] As a further improvement of the present invention, adjusting the control parameters according to the predicted deformation variable distribution data includes:

[0032] Obtaining an error value based on the predicted shape variable distribution data and a preset threshold;

[0033] Obtaining an adjustment value of the control parameter according to the error value;

[0034] The adjustment value is updated according to a preset mapping table to obtain the control parameter.

[0035] As a further improvement of the present invention, the updating of the adjustment value according to a preset mapping table to obtain the control parameter includes:

[0036] determining a compensation value according to the adjustment value and the preset mapping table;

[0037] The adjustment value is updated according to the compensation value to obtain the control parameter.

[0038] The present invention provides a high-precision single-side polishing system for ultra-thin optical elements, which is applied to the above-mentioned high-precision single-side polishing method for ultra-thin optical elements, and includes a vacuum adsorption chuck and a server:

[0039] The vacuum adsorption chuck comprises:

[0040] An adsorption area, the adsorption area includes a main adsorption area and an edge buffer ring, the main adsorption area is located in the center of the chuck, the edge buffer ring is located on the periphery of the main adsorption area, and a micropore array is distributed on the adsorption area;

[0041] a silicone deformable layer, the silicone deformable layer being located on the surface of the adsorption area, the silicone deformable layer including a thin film sensor, the thin film sensor including a positive sensor and a negative sensor, the positive sensors and the negative sensors being arranged alternately, and the positive sensors and the negative sensors having opposite responses to deformation in the same direction;

[0042] The server includes:

[0043] an acquisition module, configured to obtain deformation distribution data of the ultra-thin optical element based on voltage data acquired by the thin film sensor;

[0044] a calculation module, configured to obtain predicted shape variable distribution data based on the shape variable distribution data and a preset reduced-order model, wherein the reduced-order model is obtained through training;

[0045] The control module is used to adjust control parameters according to the predicted deformation variable distribution data, and the control parameters include the negative pressure of the adsorption area.

[0046] Beneficial effect: The present invention utilizes the arrangement of an adsorption layer, a silicone deformation layer, and positive and negative sensors in a vacuum adsorption chuck to buffer stress to avoid deformation and cracking of optical elements. At the same time, the voltage difference between the positive and negative sensors effectively monitors the deformation of the optical elements, thereby avoiding the influence of temperature drift on monitoring accuracy. Ultimately, the control parameters are adjusted according to the deformation to ensure polishing accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Schematic diagram of the positional relationship between the vacuum adsorption chuck and the optical element in the present invention;

[0048] Figure 2 Schematic diagram of the steps of a high-precision single-side polishing method for ultra-thin optical elements in the present invention;

[0049] Figure 3 Schematic diagram of the position of the sensor pair;

[0050] Figure 4 Schematic diagram of part of the triangular mesh.

[0051] Figure numerals: 1. main adsorption area; 2. edge buffer ring; 3. micropore; 4. silicone deformation layer; 5. optical element; 6. positive sensor; 7. negative sensor. DETAILED DESCRIPTION

[0052] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations of the technical solution of the present invention.

[0053] Identical components are denoted by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom," "top," "inner," and "outer" refer to directions toward or away from the geometric center of a particular component, respectively.

[0054] The term "and / or" in the following text simply describes an association between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Additionally, the character " / " generally indicates an "or" relationship between the related objects.

[0055] The present invention provides a high-precision single-side polishing device for ultra-thin optical elements. The device includes a vacuum adsorption chuck, which includes:

[0056] Adsorption area, the adsorption area includes a main adsorption area and an edge buffer ring. The main adsorption area is located in the center of the chuck, and the edge buffer ring is located on the periphery of the main adsorption area. A micropore array is distributed on the adsorption area;

[0057] The silicone deformation layer is located on the surface of the adsorption area. The silicone deformation layer includes a thin film sensor. The thin film sensor includes a positive sensor and a negative sensor. The positive sensor and the negative sensor are arranged alternately. The positive sensor and the negative sensor have opposite responses to deformation in the same direction.

[0058] During polishing, the positional relationship between the chuck and the optical element is as follows: Figure 1 shown, but Figure 1 This is merely an example; the chuck and optical element should fit tightly together during polishing. Specifically, in this embodiment, the main adsorption area 1 is used to provide sufficient adsorption force, and its shape generally matches the shape of the optical element 5, covering the central main body area of ​​the optical element 5 to prevent the optical element from slipping or shifting. However, the edge strength of the optical element 5 is relatively low, and applying strong negative pressure to the main adsorption area 1 can easily lead to stress concentration at the edge of the optical element. Therefore, in this embodiment, an edge buffer ring 2 is provided around the periphery of the main adsorption area to ensure more uniform force on the edge and prevent edge cracking of the optical element.

[0059] The adsorption area is provided with a micropore array, which is composed of a large number of micropores 3 distributed on the surface of the adsorption area. The micropore array is connected to a vacuum system. When the vacuum system is activated, air is pumped out through the micropore array, thereby forming a negative pressure in the adsorption area. The vacuum system includes a vacuum pump. By adjusting parameters such as the speed and pumping power of the vacuum pump, the negative pressure in the adsorption area can be changed. In addition, since the main adsorption area and the edge buffer ring have different functions, the micropore array is also configured differently. The main adsorption area is used to provide sufficient adsorption force. Since the higher the density of the micropore array, the stronger the adsorption force, the micropore array in the main adsorption area is usually arranged with a higher density. That is, the micropores in the main adsorption area are usually arranged with a smaller pitch. Through the synergistic effect of a large number of micropores, a stable and strong adsorption force is generated. The edge buffer ring is used to reduce stress concentration at the edge of the optical element and prevent the edge from being broken due to excessive adsorption force. Therefore, the density of the micropore array in the edge buffer ring should be relatively small, and the density of the micropore array in the edge buffer ring should decrease gradually, that is, the density increases as the position is closer to the main adsorption area, so as to avoid edge breakage caused by sudden changes in adsorption force.

[0060] The silicone deformation layer 4 covers the surface of the adsorption area to buffer stress and prevent direct contact between the optical element and the adsorption area. At the same time, the silicone deformation layer includes a thin film sensor to monitor the deformation of the optical element. For example, the thin film sensor can use a PVDF (polyvinylidene fluoride) thin film sensor, but this embodiment does not limit this.

[0061] This embodiment utilizes an adsorption layer, a silicone deformation layer, and positive and negative sensors in a vacuum adsorption chuck to buffer stress to avoid deformation and cracking of the optical element. At the same time, the voltage difference between the positive and negative sensors effectively monitors the deformation of the optical element, so that the server adjusts the control parameters according to the monitored deformation to ensure polishing accuracy.

[0062] Further, such as Figure 2 As shown, an embodiment of the present application provides a high-precision single-side polishing method for ultra-thin optical elements, which is applied to the above-mentioned high-precision single-side polishing device for ultra-thin optical elements, comprising:

[0063] According to the voltage data collected by the thin film sensor, the deformation distribution data of the ultra-thin optical element is obtained;

[0064] According to the shape variable distribution data and the preset reduced-order model, the predicted shape variable distribution data is obtained, and the reduced-order model is obtained according to the training;

[0065] According to the predicted deformation distribution data, the control parameters are adjusted, and the control parameters include the negative pressure of the adsorption zone.

[0066] Among them, the reduced-order model is a simplified mathematical model. Its core function is to significantly shorten the calculation time of complex physical fields while ensuring accuracy. During the polishing process, the deformation of the optical component is affected by the coupling of multiple physical fields such as vacuum adsorption force, polishing heat and mechanical pressure. If the deformation is predicted by traditional finite element analysis, a large number of complex equations need to be solved, and the control parameters cannot be quickly adjusted according to the predicted deformation distribution. The principle of solving the problem through the reduced-order model is to extract key features from a large amount of historical data to construct a low-dimensional simplified model (reduced-order model), retaining only the core factors that affect the deformation, and then quickly predict the deformation.

[0067] This embodiment performs interpolation operations based on the voltage data collected by the thin film sensor to obtain the deformation variable distribution data of the ultra-thin optical element, and predicts the deformation variable distribution data of the optical element at the next moment based on the deformation variable distribution data and a preset reduced-order model. Then, the control parameters are adjusted in time according to the predicted deformation variable data to avoid excessive polishing caused by deformation variable accumulation.

[0068] Furthermore, an embodiment of the present application provides a step of obtaining deformation distribution data of an ultra-thin optical element based on voltage data collected by a thin film sensor, comprising:

[0069] Obtaining a voltage difference corresponding to each sub-region according to the voltage data, each sub-region corresponds to a sensor pair, and each sensor pair includes a positive sensor and a negative sensor;

[0070] According to the voltage difference corresponding to each sub-region, the deformation amount corresponding to each sub-region is obtained;

[0071] According to the shape variable corresponding to each sub-region and the triangulation algorithm, the shape variable distribution data is obtained.

[0072] The ultra-thin optical element includes multiple sub-regions, each of which corresponds to a sensor pair for monitoring the deformation of the corresponding sub-region. Each sensor pair includes a positive sensor and a negative sensor, and the positive sensor and the negative sensor have opposite responses to deformation in the same direction. The position of the sensor pair in the silicone deformation layer is as follows: Figure 3As shown, each grid in the figure corresponds to a sub-region. Specifically, when the optical element deforms during polishing or adsorption, its surface exerts pressure on the contacting silicone deformable layer. As a flexible intermediary, the silicone deformable layer deforms slightly in sync with the element's deformation. Because the sensor is located within the silicone deformable layer, the optical element's deformation is converted into mechanical pressure on the sensor. When the sensor is subjected to mechanical pressure, the charge distribution of its internal molecules shifts, generating a voltage between the two electrodes of the film. Through polarization processing, the voltage output by the sensor when subjected to pressure can be controlled. For positive sensor 6, when subjected to pressure, it outputs a positive voltage, i.e., the output voltage is positive, and the greater the deformation, the higher the positive voltage value. Negative sensor 7 uses the opposite polarization direction as the positive sensor, so that when subjected to the same pressure as the positive sensor, it generates a negative voltage, i.e., the output voltage is negative, and the greater the deformation, the smaller the negative voltage value. The sensor transmits the output voltage to the server for subsequent steps. This embodiment does not limit the division method or number of sub-regions.

[0073] During the polishing process, a temperature drift effect is likely to occur. That is, the temperature of the optical element surface will change during the polishing process, which will cause the piezoelectric constant of the thin film sensor to change, thereby affecting the voltage value output by the sensor, and ultimately leading to inaccurate detected deformation. That is, under the same deformation, temperature changes will cause the voltage value output by the same sensor to be different.

[0074] This embodiment solves the temperature drift effect by setting up positive and negative sensors. Specifically, when the temperature changes, for each sub-area, the voltage output by the corresponding positive sensor and negative sensor can be obtained as follows:

[0075] V 正 =k'×ε+ΔV

[0076] V 负 =-k'×ε+ΔV

[0077] Here, k' is the piezoelectric constant after the change. Since the distance between the positive and negative sensors in each sub-region in this embodiment is small, it can be approximately considered that the positive and negative sensors are in exactly the same temperature environment. Therefore, the piezoelectric constants of the positive and negative sensors after the change due to temperature are equal. However, since the polarization directions of the positive and negative sensors are different, the values ​​of the output voltages caused by the deformation should be opposite, k'×ε for the positive sensor and -k'×ε for the negative sensor, where ε represents the deformation corresponding to the sub-region.

[0078] ΔV represents the additional voltage caused by the temperature drift effect. The additional voltage is not affected by the polarization direction, so the additional voltage corresponding to the positive and negative electrodes is the same. Specifically, polarization refers to applying a strong electric field to the film, so that the originally randomly distributed molecular chains are oriented along the direction of the electric field. By controlling the polarization direction, the arrangement direction of the molecular chains can be controlled, so that the positive and negative sensors output voltages in different directions under the same deformation. The polyvinylidene fluoride film has pyroelectric properties, that is, when the temperature changes, the spontaneous polarization intensity of the material changes with the temperature, thereby generating charge accumulation on the surface of the material and forming an additional voltage. Since the additional voltage is caused by the microstructural changes of the material caused by temperature and the pyroelectric effect, it does not depend on the specific orientation of the molecular chain during polarization. Therefore, even if the polarization direction is different, when the temperature changes, the positive and negative sensors will generate additional voltages of the same size and direction. Moreover, based on this principle, when the temperature changes, even if the optical element does not deform, the sensor will output an additional voltage, resulting in inaccurate deformation distribution data and affecting the polishing accuracy.

[0079] For each sub-region, we get V 正 and V 负 Then, subtract the two and we can get the corresponding voltage difference:

[0080] V1=V 正 -V 负 =2k'×ε

[0081] When k' is known, the deformation variable ε corresponding to the sub-region can be obtained. Repeat the above steps for each sub-region to obtain the deformation variable corresponding to each sub-region. Among them, k' is determined according to the temperature measured by the temperature sensor. Specifically, the temperature sensor should also be located in the silicone deformation layer to accurately obtain the temperature change of the thin film sensor. After obtaining the temperature, the k' corresponding to the current temperature can be obtained according to the relationship between temperature and piezoelectric coefficient. The relationship between temperature and piezoelectric coefficient can be obtained by fitting multiple experimental data. Fitting the relationship based on experimental data is a technical means that can be implemented by those skilled in the art, and this embodiment will not be described in detail here. Among them, this embodiment does not limit the specific number and position of the temperature sensor, that is, the temperature detected by one temperature sensor can be used as the temperature corresponding to multiple sub-regions.

[0082] In this embodiment, a sensor pair is set for each sub-region, and the deformation variable corresponding to the sensor pair is used as the deformation variable corresponding to the sub-region. However, the deformation of the optical element is not a change of discrete points. When the range of the sub-region setting is large, the discrete deformation variable value corresponding to each sub-region cannot accurately represent the deformation variable distribution of the entire optical element. Therefore, this embodiment obtains the deformation variable distribution data based on the deformation variable corresponding to each sub-region and the triangulation algorithm.

[0083] This embodiment sets a sensor pair for each sub-area, and offsets the additional voltage caused by the temperature drift effect through the voltage difference to obtain accurate deformation data. In contrast, when only a single sensor is used to monitor the deformation data of the deformation monitoring sub-area, it is difficult to identify whether the output voltage is caused by deformation or only by temperature change. Moreover, when temperature change and deformation occur simultaneously, it is difficult to calculate the additional voltage through the data output by only a single sensor, which affects the accuracy of deformation monitoring.

[0084] Furthermore, the embodiment of the present application provides a step of obtaining shape variable distribution data based on the shape variable corresponding to each sub-region and a triangulation algorithm, including:

[0085] For the deformation variable corresponding to each sub-region, calculate the corresponding curvature weighting factor;

[0086] According to the deformation variable and curvature weighting factor corresponding to each sub-region, a weighted point set is obtained;

[0087] Based on the weighted point set and triangulation algorithm, the shape variable distribution data is obtained.

[0088] Specifically, for each sub-region, a point in the region is selected, and the coordinates of the point are used as the coordinates corresponding to the sub-region. This embodiment does not limit the specific selection method.

[0089] For each coordinate, calculate the corresponding curvature weighting factor as:

[0090]

[0091] Among them, P i represents the i-th coordinate, that is, the coordinate corresponding to the i-th sub-region, w i P i The curvature weighting factor, k i P i The corresponding curvature, k min is the minimum value of the curvature corresponding to all coordinates, k max is the maximum value of the curvature corresponding to all coordinates, α is a weight factor used to control the influence of the area on the density of the divided triangles, ∈ is a minimum value used to avoid the denominator being 0. This embodiment does not limit the specific values ​​of α and ∈, and those skilled in the art can set them according to actual needs.

[0092] Among them, the edge point is the point in the optical element that is greatly affected by the adsorption force and polishing pressure and has a sharp change in curvature. On the contrary, the internal point is a point with a relatively uniform adsorption force and a small change in curvature. The edge point requires a denser triangulation to capture complex deformations. This embodiment does not limit the specific division method of edge points and internal points. For example, the edge point is a point whose distance from the edge of the optical element is less than a preset distance, and the internal point is a point whose distance from the edge of the optical element is greater than or equal to the preset distance. Since the curvature of the edge point is large, a denser triangulation is required to capture complex deformations, that is, the curvature k of the edge point is 0. i When it is larger, the corresponding curvature weighting factor w i should also be large, so the curvature weighting factor w of the edge point i Should follow the curvature k i and by introducing k min and k max Normalization is performed to ensure that the curvature weighting factors of different edge points can be compared on the same scale. For internal points with gentle shape changes and small curvature changes, there is no need to adjust the density due to curvature. To simplify the calculation process, the corresponding curvature weighting factor is directly set to 1.

[0093] This embodiment does not limit the calculation method of the curvature of the edge point. For example, when calculating the curvature of an edge point, you can select the two edge points closest to it, draw a circle passing through these three edge points, and finally use the inverse of the circle radius as the curvature corresponding to the point.

[0094] After obtaining the curvature weighting factor corresponding to each coordinate point, the coordinate point is copied according to its corresponding curvature weighting factor, and the number of copies is times, among which Indicates rounding down. For example, for the coordinate point P i , assuming that the corresponding curvature weighting factor is 2, then copy it twice to get where σ i1 and σ i2 is a random perturbation to avoid the copied coordinate points from completely overlapping, and σ i1 ≠σ i2 , random perturbations can be randomly generated through Gaussian global distribution. After repeating the above copying steps for each coordinate point, the set of each original coordinate point and each coordinate point generated by copying is used as a weighted point set. Finally, based on the weighted point set and triangulation algorithm, the deformation variable distribution data is obtained.

[0095] This embodiment uses a triangulation algorithm based on a curvature weighting factor, so that points with large curvature correspond to larger curvature weighting factors, making the triangles smaller and denser after triangulation to accurately capture complex deformations, solving the problem of traditional uniform triangulation losing deformation data at edge points, and obtaining more accurate deformation distribution data.

[0096] Furthermore, an embodiment of the present application provides a step of obtaining shape variable distribution data based on a weighted point set and a triangulation algorithm, including:

[0097] For each coordinate point in the weighted point set, perform an iterative operation, the iterative operation including determining a circumscribed circle including the coordinate point, updating the edge list and triangle list based on the distance between the coordinate point and the center of the circumscribed circle, until a preset termination condition is reached, and outputting all triangles in the current triangle list;

[0098] Get the shape distribution data based on all triangles in the current triangle list.

[0099] The preset termination condition is to complete the traversal of each coordinate point in the weighted point set.

[0100] Specifically, to ensure that each coordinate point is within the decomposition range, an initial triangle is first constructed. The initial triangle includes each coordinate point in the weighted point set. The coordinates of each vertex of the initial triangle are then placed as a set in the triangle list to obtain the initial triangle list. The edge list is used to store the edges of the triangle to be processed. The initial edge list is empty. The circumcircle corresponding to each triangle in the triangle list is then calculated to obtain the circle and radius corresponding to each circumcircle. A coordinate point located inside any triangle in the triangle list is then selected from the weighted point set. The distance between the coordinate point and the center of the circumcircle of the arbitrary triangle is calculated. If the distance between the coordinate point and the center of the circumcircle is less than the radius of the circumcircle, the triangle corresponding to the circumcircle in the triangle list is added to the temporary list. Each edge corresponding to each triangle in the temporary list is then added to the edge list. Duplicate edges in the edge list are then deleted so that each edge appears only once in the edge list. A new triangle is then generated based on each edge in the edge list and the selected coordinate point, and placed in the triangle list. The triangle in the temporary list is then deleted from the triangle list, and the temporary list is cleared. Repeat the above steps for each coordinate point in the weighted point set, and finally output all triangles in the current triangle list. Each vertex in these triangles corresponds to a coordinate point in the weighted point set. These triangles can be used to connect each coordinate point in the weighted point set to obtain a triangular mesh. Based on the triangular mesh, the deformation corresponding to any point in the optical element can be calculated. Specifically, the minimum triangle to which the point belongs can be determined, and the weight coefficient is determined based on the distance from the point to each vertex of the triangle. Finally, the deformation of the point is obtained based on the deformation corresponding to each vertex and the weight coefficient.

[0101] Exemplarily, in order to further explain the above steps, a simple example is given below. Assume that the vertex coordinates of the initial triangle are A(-5,-5), B(10,-5) and A(-5,10), and the initial triangle includes each coordinate point in the weighted point set. Then, the coordinates of each vertex of the initial triangle are put into the triangle list as a set, and the initial triangle list T is obtained as T={T1(A,B,C)}, where T1 represents the initial triangle. Then, the circumscribed circle corresponding to each triangle in the triangle list is calculated to obtain the circle and radius corresponding to each circumscribed circle. At this time, the triangle list includes only one triangle, and the center coordinates are calculated to be (2.5,2.5) and the radius is about 7. Then, a coordinate point p1(2,3) located inside any triangle in the triangle list is selected from the weighted point set, and the distance between the coordinate point and the center of the circumscribed circle of the arbitrary triangle is calculated to be about 0.7, which is less than the radius of the circumscribed circle. Therefore, the triangle corresponding to the circumscribed circle in the triangle list is added to the temporary list to obtain the temporary list T. ′={T1(A,B,C)}, then add each edge corresponding to each triangle in the temporary list to the edge list, and get the edge list E={(A,B),(E,C),(B,C)}. All edges in the edge list appear only once, so there is no need to delete edges. Then, based on the edge list and p1, generate new triangles T2(p1,B,C), T3(A,p1,C) and T4(A,B,p1), and put them into the triangle list. Then delete the triangles in the temporary list from the triangle list, and get The updated triangle list is T = {T2(p1,B,C),T3(A,p1,C),T4(A,B,p1)}. Then, select a coordinate point from the weighted point set that is inside any triangle in the triangle list, such as the coordinate point p2(5,5) in T2(p1,B,C). Then calculate the distance between p2 and the center of the circumscribed circle of triangle T2(p1,B,C), and repeat the above steps until the preset termination condition is reached, and output all triangles in the current triangle list.

[0102] This embodiment uses a triangulation algorithm to obtain the deformation variables corresponding to other points on the optical element based on the deformation variable corresponding to each sub-region, and then obtain deformation variable distribution data, wherein the deformation variable distribution data includes the deformation variable corresponding to each sub-region and the deformation variables corresponding to other coordinate points calculated based on the triangular mesh. This embodiment does not limit the number and position of other coordinate points, as long as the deformation variable distribution data can relatively completely describe the deformation variable distribution of the optical element.

[0103] For example, Figure 4 As shown, assuming that one of the triangles in the triangular mesh is ΔABC, Figure 4 Not all triangles in the triangular mesh are drawn. When the deformation corresponding to point D needs to be determined, it needs to be calculated based on the distance from point D to points A, B, and C.

[0104] Moreover, the triangulation algorithm selected in this embodiment can construct a triangular mesh according to the characteristics of the edge points in the optical element, so that the triangular mesh can better reflect the characteristics of the edge area, and then the deformation variables corresponding to other coordinate points can be accurately calculated according to the triangular mesh, so as to facilitate the subsequent accurate adjustment of the control parameters according to the deformation variable distribution data, thereby ensuring the polishing accuracy.

[0105] Furthermore, the embodiment of the present application provides a step of obtaining a reduced-order model based on training, including:

[0106] Obtain historical deformation data and historical control parameter data;

[0107] According to historical deformation data and principal component analysis method, multiple main modes are obtained;

[0108] A reduced-order model is obtained based on historical control parameter data and multiple main modes.

[0109] The historical control parameter data includes multiple samples, each corresponding to a vector, and each element in the vector corresponds to a control parameter. This embodiment does not limit the number and type of control parameters. For example, the control parameters include the negative pressure of the main adsorption area, the negative pressure of the edge buffer ring, the polishing pressure, and the polishing head speed. To unify the dimensions, each element in the vector is normalized. The historical deformation variable data includes deformation variable distribution data corresponding to each sample in the historical control parameter data.

[0110] Specifically, assuming that the historical control parameter data includes N samples, the shape variable distribution data corresponding to each sample includes M shape variables, and each shape variable corresponds to a coordinate point, a high-dimensional data matrix U can be obtained. The number of rows in the matrix U is the number of samples N. The samples are arranged in the order of acquisition time. The acquisition interval between each sample is the same, and the number of columns is M. The high-dimensional data matrix is ​​then reduced to a certain order. First, the mean vector of all row vectors in the matrix U is calculated. Then subtract the mean vector from each row in the matrix U Get the matrix U after centralization C , then according to U C Calculate the covariance matrix Then, perform eigenvalue decomposition based on C: CΦ=ΦΛ to obtain the eigenvector matrix Φ and the eigenvalue matrix Λ. Performing eigenvalue decomposition on C based on principal component analysis to obtain the eigenvector matrix and the eigenvalue matrix is ​​a technical means that can be implemented by those skilled in the art, and this embodiment will not be described in detail here. Then select the largest k eigenvalues ​​from the eigenvalue matrix, and determine the eigenvectors corresponding to the k eigenvalues ​​from φ to obtain the main mode matrix in, is the eigenvector corresponding to the maximum value among the k eigenvalues, is the eigenvector corresponding to the second largest value among the k eigenvalues, and so on. This embodiment does not limit the specific value of k. The value of k can be obtained according to the deformation mode of the optical element. For example, through the analysis of historical deformation data, it is found that during polishing, the most likely deformation modes of the optical element include symmetrical bending of the main adsorption area and the edge buffer ring (symmetrical bending includes bending of the main adsorption area and the edge buffer ring in the same direction or in the opposite direction), local bulge of the optical element, and wavy warping of the edge. In this case, k=3 can be set, that is, the number of main modes is 3.

[0111] Then for the matrix U C For each row in the equation, calculate the corresponding main modal coefficient a j =u j Φ k, where j represents the number of rows, u j For U C In the jth row, a j is the main modal coefficient corresponding to the jth row, since the matrix U C Each row in corresponds to a sample, so a j It can be called the main modal coefficient corresponding to the jth sample, a j is a vector with k elements, so a j Each element in corresponds to a main mode. C Repeat the above steps for each row in to obtain N main modal coefficients, each of which contains k elements. Then, extract the first element of each main modal coefficient in turn to form the main modal coefficient corresponding to the first main mode. Then, extract the second element to the kth element of each main modal coefficient in turn to obtain the main modal coefficients corresponding to k main modes. The main modal coefficient corresponding to each main mode is a vector containing N elements.

[0112] Since each sample is arranged according to the acquisition time and the acquisition interval is the same, the N elements in the main modal coefficients corresponding to the k main modes are also arranged in time order. Therefore, based on the two adjacent elements, the change rate of the kth main mode at each moment can be obtained. For example, the main modal coefficient corresponding to the kth main mode is recorded as b k , b k The i-th element in is denoted as b k (i), each element corresponds to a sampling moment, so when the sampling interval is very small, b at the i-th moment k Rate of change It can be approximately expressed as where b k (i+1) is b k The i+1th element in . After obtaining the rate of change corresponding to each moment, the sample corresponding to each moment is obtained based on the historical control parameter data, and a fitting relationship between the rate of change and the sample is established. Since the number of main modes is k, a total of k relationship equations are obtained, and these k relationship equations are called reduced-order models. Among them, fitting the relationship equation according to the data is a technical means that can be implemented by those skilled in the art, such as performing linear fitting according to the least squares method. This embodiment does not limit the specific fitting method, and those skilled in the art can select an appropriate fitting method based on actual conditions.

[0113] Furthermore, an embodiment of the present application provides a step of obtaining predicted shape variable distribution data based on shape variable distribution data and a preset reduced-order model, including:

[0114] Project the deformation variable distribution data onto the main mode to obtain the main mode coefficient vector, and the main mode is obtained according to the reduced-order model;

[0115] According to the current control parameters and the main modal coefficient vector, the predicted main modal coefficient is obtained;

[0116] The predicted deformation variable distribution data is obtained based on the predicted main modal coefficients.

[0117] Specifically, first, the current shape variable distribution data can be obtained according to the above triangulation algorithm. The shape variable distribution data can be expressed as a column vector u. Each shape variable in the vector corresponds to a coordinate point and the number of elements should be M. Then, the current control parameters are obtained to obtain a vector including multiple control parameters.

[0118] Then project u into the main modal space to obtain the main modal coefficient vector Each element in vector a corresponds to a main mode. Then, the current control parameters are substituted into the above k relational expressions to obtain the transformation rate corresponding to each main mode. Then, for each main mode, the corresponding element in a is obtained. The corresponding transformation rate and element value of the main mode are added together to obtain the predicted main mode coefficient corresponding to each main mode. That is, a total of 3 predicted main mode coefficients are obtained. The 3 predicted main mode coefficients are arranged according to their corresponding main modes to obtain the row vector a1. Then, the predicted deformation variable distribution data is obtained based on vector a1. The predicted shape variable distribution data is the shape variable distribution data corresponding to the next moment.

[0119] This embodiment uses a reduced-order model to accurately extract the main mode from a large amount of data, clearly present the essential characteristics of the deformation variable, and obtain the relationship between the deformation variable and the control parameter. It can then accurately predict the deformation variable distribution data at the next moment, and then adjust the control parameters at the next moment to ensure polishing accuracy and avoid over-polishing.

[0120] Furthermore, the embodiment of the present application provides a step of adjusting control parameters based on the predicted deformation variable distribution data, including:

[0121] Obtain an error value based on the predicted shape variable distribution data and a preset threshold;

[0122] Obtain adjustment values ​​of control parameters according to the error values;

[0123] The adjustment value is updated according to the preset mapping table to obtain the control parameter.

[0124] Among them, the preset threshold is the maximum deformation acceptable during the production process, and the preset thresholds corresponding to the main adsorption area and the edge buffer ring may be different. This embodiment does not limit the value of the preset threshold, and technical personnel in this field can set it according to actual conditions.

[0125] After obtaining the predicted deformation variable distribution data u1, each element in u1 is compared with its corresponding preset threshold value, and the element exceeding the preset threshold value is determined. The element is subtracted from the preset threshold value to obtain the error value corresponding to the element, and the position of the element is determined, where the error value is the deformation variable that needs to be adjusted. For example, if an element value is 5μm and its corresponding preset threshold value is 4μm, then the error value is 1μm, which is the deformation variable that needs to be adjusted. The error value is then assigned to each control parameter to obtain the error value of each control parameter. The correlation between each control parameter and the deformation variable can be considered during the assignment. The correlation between the control parameter and the deformation variable can be obtained based on the historical deformation variable data and the historical control parameter data. The calculation of the correlation is a conventional technical means of those skilled in the art, and this embodiment does not limit this.

[0126] Each control parameter corresponds to a PID controller. After the error value corresponding to the control parameter is input into the corresponding PID controller, the adjustment value corresponding to the control parameter can be obtained. The adjustment value is the difference between the value before adjustment and the value after adjustment.

[0127] Furthermore, an embodiment of the present application provides a step of updating an adjustment value according to a preset mapping table to obtain a control parameter, including:

[0128] Determine the compensation value according to the adjustment value and a preset mapping table;

[0129] The adjustment value is updated according to the compensation value to obtain the control parameter.

[0130] Among them, the preset mapping table stores the relationship between position, adjustment value and error value. For example, when the control parameters include the negative pressure of the main adsorption area, the negative pressure of the edge buffer ring and the polishing pressure, one of the rows in the mapping table records that the negative pressure of the main adsorption area is reduced by 0.8MPa, the negative pressure of the edge buffer ring is reduced by 0.2MPa, and the polishing pressure is reduced by 5KPa. The deformation of the main adsorption area is reduced by an average of 0.8μm. The mapping table can be obtained based on historical deformation data and historical control parameter data. This is a technical means that can be implemented by technical personnel in this field, and this embodiment does not limit this.

[0131] Specifically, after obtaining the adjustment value of each control parameter, the mapping table is used to query whether the error value corresponding to the adjustment value in the table is equal to the error value calculated above. If they are equal, the current control parameter is updated according to the calculated adjustment value to obtain the control parameter at the next moment. If they are not equal, the proportional relationship between the error value in the calculation table and the calculated error value is used to obtain the compensation value corresponding to each control parameter through this proportional relationship. The compensation value is the value that needs to be adjusted after the control parameter is adjusted according to the adjustment value. Finally, the control parameter is updated according to the adjustment value and the compensation value to obtain the control parameter at the next moment. If the currently calculated adjustment value does not exist in the mapping table, the control parameter is updated according to the calculated adjustment value, and the next moment is checked to see whether the adjustment at the current moment is accurate, that is, whether the deformation at the next moment has a decreasing trend through the adjustment at the current moment. If so, a new row is added to the table. If not, the adjustment value is recalculated and the above steps are repeated.

[0132] This embodiment obtains accurate null parameters by simultaneously considering the output results of the mapping table and the PID controller, thereby avoiding the problem of being unable to respond to real-time changes in the adjustment amount when only the mapping table is used and avoiding the problem of calculation errors that are prone to occur when only the PID controller is used.

[0133] The present invention provides a high-precision single-side polishing system for ultra-thin optical elements, which is applied to the high-precision single-side polishing method for ultra-thin optical elements described above. The system includes a vacuum adsorption chuck and a server:

[0134] The chuck includes:

[0135] Adsorption area, the adsorption area includes a main adsorption area and an edge buffer ring. The main adsorption area is located in the center of the chuck, and the edge buffer ring is located on the periphery of the main adsorption area. A micropore array is distributed on the adsorption area;

[0136] A silicone deformation layer is located on the surface of the adsorption area. The silicone deformation layer includes a thin film sensor. The thin film sensor includes a positive sensor and a negative sensor. The positive sensor and the negative sensor are arranged alternately. The positive sensor and the negative sensor have opposite responses to deformation in the same direction.

[0137] The server includes:

[0138] An acquisition module, configured to obtain deformation distribution data of the ultra-thin optical element based on voltage data collected by the thin film sensor;

[0139] A calculation module is used to obtain predicted shape variable distribution data based on shape variable distribution data and a preset reduced-order model, where the reduced-order model is obtained based on training;

[0140] The control module is used to adjust the control parameters according to the predicted deformation variable distribution data, and the control parameters include the negative pressure of the adsorption area.

[0141] The embodiments of the present application provide a high-precision single-sided polishing device, method, and system for ultra-thin optical elements. First, the deformation of the optical element is accurately monitored by setting a sensor pair, and the overall deformation distribution data of the optical element is obtained in combination with a triangulation algorithm. Then, the deformation distribution data at the next moment is predicted based on the current deformation distribution data, and finally the control parameters are adjusted based on the predicted data. The method provided in this embodiment not only ensures the accuracy of deformation monitoring, but also reduces the amount of calculation through the setting of a reduced-order model, improves the response speed of the system, and can adjust the control parameters more quickly, thereby ensuring polishing accuracy.

[0142] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0143] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0144] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0145] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A high-precision single-side polishing device for ultra-thin optical components, characterized in that: The device includes a vacuum adsorption chuck, and the vacuum adsorption chuck includes: An adsorption area, the adsorption area includes a main adsorption area and an edge buffer ring, the main adsorption area is located in the center of the chuck, the edge buffer ring is located on the periphery of the main adsorption area, and a micropore array is distributed on the adsorption area; A silicone deformable layer, the silicone deformable layer is located on the surface of the adsorption area, the silicone deformable layer includes a thin film sensor, the thin film sensor includes a positive sensor and a negative sensor, the positive sensor and the negative sensor are arranged alternately, and the positive sensor and the negative sensor have opposite responses to deformation in the same direction.

2. A high-precision single-side polishing method for ultra-thin optical elements, applied to the high-precision single-side polishing device for ultra-thin optical elements as claimed in claim 1, characterized in that: include: Obtaining deformation distribution data of the ultra-thin optical element according to the voltage data collected by the thin film sensor; Obtaining predicted shape variable distribution data based on the shape variable distribution data and a preset reduced-order model, wherein the reduced-order model is obtained through training; According to the predicted deformation distribution data, control parameters are adjusted, wherein the control parameters include the negative pressure of the adsorption area.

3. The high-precision single-side polishing method for ultra-thin optical components according to claim 2, characterized in that: The ultra-thin optical element includes a plurality of sub-regions, and obtaining deformation distribution data of the ultra-thin optical element according to the voltage data collected by the thin film sensor includes: Obtaining a voltage difference corresponding to each sub-region according to the voltage data, wherein each sub-region corresponds to a sensor pair, and each sensor pair includes a positive sensor and a negative sensor; Obtaining a deformation amount corresponding to each sub-region according to a voltage difference corresponding to each sub-region; The shape variable distribution data is obtained according to the shape variable corresponding to each sub-region and a triangulation algorithm.

4. The high-precision single-side polishing method for ultra-thin optical elements according to claim 3, characterized in that: The step of obtaining the shape variable distribution data according to the shape variable corresponding to each sub-region and the triangulation algorithm includes: For the deformation amount corresponding to each sub-region, calculating the corresponding curvature weighting factor; Obtaining a weighted point set according to the deformation amount corresponding to each sub-region and the curvature weighting factor; The deformation variable distribution data is obtained according to the weighted point set and the triangulation algorithm.

5. The high-precision single-side polishing method for ultra-thin optical components according to claim 4, characterized in that: The step of obtaining the shape variable distribution data according to the weighted point set and the triangulation algorithm includes: For each coordinate point in the weighted point set, performing an iterative operation, the iterative operation comprising determining a circumscribed circle including the coordinate point, updating an edge list and a triangle list based on a distance between the coordinate point and a center of the circumscribed circle, until a predetermined termination condition is met, and outputting all triangles in the current triangle list; The shape variable distribution data is obtained according to all triangles in the current triangle list.

6. The high-precision single-side polishing method for ultra-thin optical components according to claim 2, characterized in that: The step of obtaining predicted shape variable distribution data based on the shape variable distribution data and a preset reduced-order model includes: Projecting the deformation variable distribution data onto a main mode to obtain a main mode coefficient vector, wherein the main mode is obtained according to the reduced-order model; Obtaining predicted main modal coefficients according to current control parameters and the main modal coefficient vector; The predicted deformation variable distribution data is obtained according to the predicted main modal coefficients.

7. The high-precision single-side polishing method for ultra-thin optical components according to claim 2, characterized in that: The reduced-order model is obtained according to training, and includes: Obtain historical deformation data and historical control parameter data; Obtaining multiple main modes based on the historical deformation data and principal component analysis method; The reduced-order model is obtained according to the historical control parameter data and the plurality of main modes.

8. The high-precision single-side polishing method for ultra-thin optical components according to claim 2, characterized in that: Adjusting control parameters according to the predicted deformation distribution data includes: Obtaining an error value based on the predicted shape variable distribution data and a preset threshold; Obtaining an adjustment value of the control parameter according to the error value; The adjustment value is updated according to a preset mapping table to obtain the control parameter.

9. The high-precision single-side polishing method for ultra-thin optical components according to claim 8, characterized in that: The updating of the adjustment value according to a preset mapping table to obtain the control parameter includes: determining a compensation value according to the adjustment value and the preset mapping table; The adjustment value is updated according to the compensation value to obtain the control parameter.

10. A high-precision single-side polishing system for ultra-thin optical components, applied to a high-precision single-side polishing method for ultra-thin optical components according to any one of claims 2 to 9, characterized in that: Includes vacuum chuck and server: The vacuum adsorption chuck comprises: An adsorption area, the adsorption area includes a main adsorption area and an edge buffer ring, the main adsorption area is located in the center of the chuck, the edge buffer ring is located on the periphery of the main adsorption area, and a micropore array is distributed on the adsorption area; a silicone deformable layer, the silicone deformable layer being located on the surface of the adsorption area, the silicone deformable layer including a thin film sensor, the thin film sensor including a positive sensor and a negative sensor, the positive sensors and the negative sensors being arranged alternately, and the positive sensors and the negative sensors having opposite responses to deformation in the same direction; The server includes: an acquisition module, configured to obtain deformation distribution data of the ultra-thin optical element based on voltage data acquired by the thin film sensor; a calculation module, configured to obtain predicted shape variable distribution data based on the shape variable distribution data and a preset reduced-order model, wherein the reduced-order model is obtained through training; The control module is used to adjust control parameters according to the predicted deformation variable distribution data, and the control parameters include the negative pressure of the adsorption area.

Images