TW569292B - Estimation method for machine processing parameters of semiconductor equipment and recording medium readable by computer - Google Patents

Abstract

The present invention is an estimation method for machine processing parameters of semiconductor equipment, which includes the following steps: first, calculating the first average of processing parameters for the first product manufacturing from multiple machines; then, configuring the first average as the processing parameters of the semiconductor equipments not processing the first product; then, calculating the second average of processing parameters during the manufacturing of multiple different products; using the expectation maximization algorithm to re-configure the processing parameters for producing the first products on the semiconductor equipments based on the first average and the second average.

Description

569292 V. Description of the invention (1) The present invention relates to a method for predicting process parameter values of a semiconductor device machine, and particularly to a process parameter value using a similar machine that has already manufactured a product, and predicts that the product has not yet been manufactured. A method for predicting a process parameter value of a semiconductor device machine when the semiconductor device machine is manufacturing this product. With the development trend of advanced process control (Apc) in the semiconductor industry, the standard and monitoring strategy for semiconductor integrated circuit systems in the future will continue to use the single machine, single module, single variable, and The implementation of post measurement quality variable process and equipment monitoring has been transformed into multi-machine, cross-module, multi-variable and wear-out process operation, state variables, and batch room (Run —10—Run, R2R) Immediate control and monitoring strategies. Most of the semi-conductors use single-variable monitoring, which is based on statistical or process yield and production. However, due to the occurrence of law, often a large number of flaws are discovered. , Also the number of manufacturing plants, local essential product value and quantity Process design, overall efficiency. Yu Zhizhang is the epidemic product of equipment, machine, conductor, and equipment. Inspection, goodness depends only on the manufacturer, monitoring and monitoring of semiconductor equipment, large manufacturing process The machine or section process machine performs operational measurement variables and statistical program control (Control, SPC) product quality control, and also incorporates routine maintenance between the machine to improve product square process operation, I state and equipment status. An abnormal machine with decay or no M and no warning type will cry according to the unsuitable process parameter value, and the abnormal situation of the unequipped machine usually requires testing to find out the problem. And ask the rule of thumb of the private and private teachers to propose temporary

569292 V. Description of the invention (2) = positive, and the crux of the problem cannot be effectively found and solved. In addition, it is necessary to rely on the process parameter values of the process engineer for the part of the non-ready machine σ ′ to make the product. This situation fails. In view of this, the main purpose of the present invention is to refer to a related process of a similar machine that has manufactured a product, and a method for predicting a parameter value of a semiconductor device that has not manufactured the product, and to dynamically update the planned parameter value at any time. In order to achieve the above purpose, it can be achieved by the method for predicting the process parameter value of the standby machine. According to the semiconductor design method of the embodiment of the present invention, first, a first average value of a plurality of machines for a value is calculated, and then, after the manufacturing process of the semiconductor equipment machine of the first product is set, the time when the semiconductor equipment machine is calculated is calculated. The second average value of the process parameter values and the (Expectation Maximization 5 EM) value and the second average value reset the process parameter values of the semiconductor device. According to the embodiment of the present invention, when a predetermined number of process parameters are established, the process parameters are terminated to predict. In addition, the rule of thumb of the invention that can be applied to semiconductors and cannot be manufactured can also be made to provide a value that can be manufactured and tested. The semiconducting effective pin is used to set the semiconductor device of the problem product to temporarily set a large number of products for trial planting. The use of the process equipment and device manufacturing system that has been manufactured to automatically predict when the product is still in use. The first product provided by the semiconductor device manufacturing system is average. Value parameter value. For multiple utilization expectation algorithms, the standby system makes numerical predictions. Bulk exposure machine parameter value prediction manufacturing process parameters have not yet processed the manufacturing of this different product. Increasing the number of times to make this first product based on the first average is greater than one. Semiconductor Research 0503-7587TWf; TSMC2001-1510; yianhou.ptd No. 6 Page 569292 V. Description of the invention (3) Semiconductor equipment such as grinding machine, semiconductor etching machine, and semiconductor ion implantation machine, and the corresponding process parameter values can be exposure energy, grinding time, etching time, With ion implantation dose and so on. Embodiment FIG. 1 is a flowchart showing a method for predicting a process parameter value of a semiconductor device machine according to an embodiment of the present invention. According to the method for predicting a process parameter value of a semiconductor device machine according to an embodiment of the present invention, first, in step S10, a first average value of process parameter values of a plurality of machines for a first product manufacturing is calculated. Next, in step S11, a process parameter value of a semiconductor device machine that has not yet processed the first product is set according to the first average value. Then, as step, S1 2, calculate the second average value of the process parameter values of the semiconductor device machine for manufacturing a plurality of different products, and as step S13, use Expectation Maximization (EM) algorithm, The first average value and the second average value reset the process parameter values of the semiconductor device machine to manufacture the first product. In addition, according to the embodiment of the present invention, when the number of times the process parameter value is predicted is greater than a predetermined number, the prediction of the process parameter value is terminated. It is worth noting that the invention can be applied to semiconductor equipment such as semiconductor exposure equipment, semiconductor polishing equipment, semiconductor etching equipment, and semiconductor ion implantation equipment, and the corresponding process parameter values can be exposure. Energy, grinding time, etching time, and ion implantation dose. Next, the semiconductor exposure machine and its exposure energy (process parameter values) for product manufacturing are described in detail as follows:

0503-7587TWf; TSMC2001-1510; yianhou.ptd page 7 569292

569292 V. Description of the invention (5); dti [k] ^ d ^ [k] / rij, where allow represents the number of times the process parameter value is predicted. Next, using the Expectation Maximization (EM) algorithm, the position where the exposure energy is newly filled (predicted) in the array D (that is, the part without the exposure energy data) is reset according to the row average value and the column average value. Its exposure energy data 'is as follows: ^ [ic + l] = + pc] ~ d ^ [A:]) ~ («X? 7] Xii [/:])) / ((«-1) (? 73 -1)) Among them, AW. Represents the process parameter value when the semiconductor device machine manufactured the first product the last time; 4 [* + 1] represents the process parameter when the semiconductor device machine manufactured the first product the next time Values; and the average of the column and row averages. Note that the reset operation can define conditions to stop it from predicting process parameter values. For example, it can be set to stop predicting the value of the process parameter when the number of predictions of the process parameter value (U value) is greater than a predetermined number; or to define a value, ~ = 1, that is, to check all predicted values before readjusting the settings Whether the gap between them reaches a predetermined threshold, such as judging whether it is greater than 1 (Γ-5, if μ > 1 (Γ-5 indicates that the fine-tuning difference has exceeded a certain distance, then terminate the prediction of the process parameter value, more A warning signal or e-mail can be sent to the process engineer to request further correction. Next, an example is used for illustration. Figure 2 shows an example of a process parameter value table. Assume that the process parameter value table includes four types of machines (DPPHE1, DPPHE1,

0503-7587TWf; TSMC2001-1510; yianhou.ptd page 9 569292 V. Description of the invention (6) DPPHE2, DPPHE5, DPPHE6) and three products (8205 1 20A, 8358120A, 8595120A) exposure energy data. Among them, the DPHHE1 machine has no exposure energy data for the product 8358 1 20A and the DPHHE2 machine has the product 8595 1 20A. That is, the DPHHE1 machine has not manufactured the product 8358120A and the 0 ^ 1 ^ 2 machine has not manufactured the product 8595 1 20 Eight. First, by using the formula. = ¾A / (W — L! Mountain = 1, ... ”, calculate the column average for each column: (Column 1) The column average is two (31.38 + 28.24 + 33.17 + 34.96 ) /4-31.94 (column 2) column average = (28.81 + 33.46 + 33.65) / 3 = 31.97 (column 3) column average = (29 · 11 + 31. 51 + 32. 72) / 3 = After 31.11, the portion of each column that has no exposure energy data is set to the average value of that column, as shown in Figure 3. Next, the first fine-tuning of the predicted value (fc = 1), using the formula respectively < _ [fc] ⑻ / WM == 1, · "and < thorn = the difference] /« j = i, ···, calculate the column average K.) and row average (\): (\): ( Column 1) column average = (31.38 + 28.24 + 33 · 17 + 34.96) / 4-31. 94 (column 2) 歹, J + sentence value: (31 · 97 + 28 · 81 + 33 · 46 + 33 · 65) /4=31.97 (column 3) 歹 11 Mean = (2 9. 1 1 +3 1. 1 1 +3 1. 5 1 + 32. 72) / 4 = 31.11 (row 1 ) Row average = (3 1. 38 + 3 1. 9 7 + 29 · 11) / 3 = 30. 82

569292 V. Description of the invention (7) (line 2) line average = (28 · 24 + 28 · 81 + 3111) / 3 = 29 · 39 (line 3) line average = (3317 + 3346 + 31.51) / 3 = 3271 (line 4) row average = (34 · 96 + 3365 + 32 · 72) / 3 = 3378 So 'can set a new prediction value: (4χ (3χ 0. 82-31. 97) + 3 χ (4 x 3 1.97-3 1.97) ~ (1 2 x 3 1.67-3 1.97)) / 6: 30.27 (4x (3x 29. 39-31. Ll) + 3x (4x 31.11- 31. 11) -(12x 31. 67-31. 11)) / 6-26. 82 After 'write the predicted value back to the table, as shown in Figure 4A. Now ^ 'To repeat the above-mentioned fine-tuning operation, the results after the second fine-tuning 0 = and the third-person fine-tuning (fc = 3) are shown in the "" and 4 (: Fig. 2). In addition, the invention also It can be coded into a computer program which can be read by the computer. In the recorded medium. The computer program is used to enable the prediction of the manufacturing parameter value of the semiconductor device machine, as described in the present invention. Therefore, 'provided by the present invention Method for measuring the process parameter of a semiconductor device machine 'can use the relevant ^ private parameter values of similar machines that have already manufactured products' to dynamically predict the semiconductor device machine that has not yet manufactured this product = the value of the process parameter when manufacturing this product In the second case, which does not require human intervention, the process parameter values in the semi-cold body δ and the machine can be automatically predicted and fine-tuned according to the manufacturing situation of the similar machine II, which greatly increases the overall production rate, improves the product manufacturing yield, and reduces Mistakes that may occur during artificial adjustments. Although the present invention has been disclosed above in the preferred embodiment, it is not intended to

569292 V. Description of the invention (8) The invention is limited. Anyone who is familiar with this technology can do some modifications and retouching without departing from the spirit and scope of the invention. Therefore, the scope of protection of the invention shall be regarded as the attached application. The patent scope shall prevail. 0503-7587TWf; TSMC2001-1510; yianhou.ptd page 12 569292

In order to make the above-mentioned objects, features, and advantages of the present invention clearer, the following specific embodiments are described in detail with the accompanying drawings. The following figure 1 shows a method for predicting a semiconductor programming parameter value according to an embodiment of the present invention. flow chart. Figure 2 shows an example of a process parameter value table. Of the embodiment

Fig. 3 shows the result of the process parameter value table of Fig. 2 after the process parameter value prediction of this issue. X

FIG. 4A shows the result of the process parameter value table of FIG. 3 after the process parameter value prediction of the embodiment of the present invention. K

FIG. 4B shows the result of the process parameter value table in FIG. 4A after fine-tuning the process parameter values according to the embodiment of the present invention. Fig. 4C shows the result of the process parameter value table in Fig. 4B after fine adjustment of the process parameter values according to the embodiment of the present invention. , Symbol description S 1 0, S11, ..., S1 3 ~ Operation steps.

Claims (1)

569292 6. Scope of patent application: 1. A method for predicting the process parameter value of a semiconductor equipment machine, including the following steps: _ Calculating the first average value of the process parameter value of a plurality of machines for the manufacture of a first product; and the basis The first average value sets a process parameter value of a semiconductor device machine of one of the first products. 2. The method for predicting the process parameter value of the semiconductor equipment machine as described in item 1 of the scope of the patent application, further comprising the following steps: Calculating a second average value of the process parameter value of the semiconductor equipment machine when manufacturing a plurality of different products; And H sets a process parameter value of the semiconductor device 'machine for manufacturing the first product according to the first average value and the second average value. · 3. The method for predicting the process parameter value of the semiconductor device machine as described in item 1 of the scope of the patent application, wherein the process parameter value of the semiconductor device machine that has not yet been processed for the first product is set according to the first average value. A process parameter value when the semiconductor device machine manufactures the first product is set to the first average value. 4. The method for predicting a process parameter value of a semiconductor device machine as described in item 2 of the scope of patent application, wherein the process parameter for the semiconductor device machine to manufacture the first product is set according to the first average value and the second average value. The value is specified according to the following equation: 4 [女 +1] = (wiX («Xd y [k] -άΗ [/ c]) + ηχ 1) (^-1)) where k represents the predicted value of the process parameter Number of times; m represents the 0503-7587TWf; TSMC2001-1510; yianhou.ptd Page 14 569292 VI. Scope of patent application The main blood & The number of two-conductor equipment machines; "Represents the honeycomb condition including the different sigma sockets and the first rape machine, indicating the previous semiconductor equipment ^ m Γ-process parameter value at the time of the product; 4 Bu 丨] Represents the next time = ¥ ° and the manufacturing process parameter value when the first product is manufactured by the machine; ^] wins the first average value; Qinet represents the second average value; and (㈦ represents the Guang-ping: The average value of the value and the second average value. The manufacturing method of the semiconductor equipment machine described in item 4 of the patent application No. 4 of the patent application. The "precision method" further includes the number of times U when the process parameter value is predicted. When the existing number is reached, the prediction of the process parameter value is terminated. 〇6. The method for predicting the private parameter value of the semiconductor equipment machine as described in the first item of the patent scope of the application, wherein the semiconductor equipment machine is a semiconductor exposure machine . 7. The method for predicting the private parameter value of the semiconductor device machine as described in item 6 of the scope of the patent application, wherein the parameter value is the exposure energy. 8 · The semiconductor device as described in the scope of the patent application Machine process parameter value prediction Measuring method, wherein the conductor equipment machine is a semiconductor polishing machine. 9. The method for predicting the process parameter value of the semiconductor equipment machine as described in item 8 of the scope of patent application, wherein the f-stable parameter value is a grinding time. 10. The method for predicting the process parameter value of a semiconductor device as described in item 1 of the scope of application for a patent, wherein the semiconductor device machine is a semiconductor money engraving machine. I!, A, ^ ^ ii of this conductor equipment machine II · Bull as claimed in item 10 of the patent scope ¥ 0503-7587TWf; TSMC2001 -1510: yianhou.ptd 15th tribute 569292 6. Scope of patent application: A method for predicting a process parameter value, wherein the process parameter value is an etching time. 1 2 · The method for predicting process parameter values of the semiconductor device machine as described in item 1 of the scope of the patent application, wherein the semiconductor device machine is a semiconductor device ^ 1 3 · The semiconductor device as described in the item 2 of the patent application scope Method for predicting process parameter values of equipment machines, wherein the process parameter value is an ion implantation dose of 0 1 4 · A method for predicting process parameter values of semiconductor equipment machines, including the following steps: Calculating a plurality of machines for a first product manufacturing The first average value of the process parameter values at that time; and the average value is the process parameter value of a semiconductor device machine that has not yet processed one of the first products; the process of manufacturing the semiconductor device machine for a plurality of different products is calculated A second average value of the parameter value; and an Expectation Maximization algorithm, which resets the process parameter values of the semiconductor device machine to manufacture the first product according to the first average value and the second average value . 15 · The method for predicting the process parameter value of the semiconductor equipment machine as described in Item 14 of the scope of the patent application, wherein the expected increase (EM) algorithm is used to re-calculate based on the first average value and the second average value. The process parameter values for setting the semiconductor device machine to manufacture the first product are specified according to the following equations: 0503-7587TWf; TSMC2001-1510; yianhou.ptd page 16 569292 VI. Patent application scope, λα-ir number; w stands for the machine that includes' which represents the prediction of process parameter values / including 誃 and other semiconductor equipment machines The number of "" generations of the product and the product of the first product; the previous one; the value of the process parameters when the conductor set f machine manufactures the first product; the manufacturer, the manufacturer of the semiconductor equipment machine The value of the process parameter for the first product represents the first average value; 0] represents the first; average value, and represents the average value of the first average value and the second average value. 1 6 · As described in Item 5 of the patent scope, the semi-conductor box is long, and the method of predicting the process parameter value further includes when the number of times of the process parameter value prediction is greater than a predetermined number, Terminate the prediction of the process parameter value 17 · The method for predicting the process parameter value of the semiconductor device and the machine as described in item 4 of the patent application scope, wherein the semiconductor equipment machine is a semiconductor exposure machine. 1 8 · The method for predicting the process parameter value of the semiconductor device and the machine as described in item 17 of the scope of patent application, wherein the process parameter value is the exposure energy. 19. The method for predicting process parameter values of the semiconductor equipment machine as described in item 14 of the scope of application for a patent, wherein the semiconductor equipment machine is a semiconductor grinding machine. 2 0. The method for predicting the process parameter value of the semiconductor equipment machine as described in item 19 of the scope of patent application, wherein the process parameter value is a grinding time. 2 1 · Semiconductor equipment as described in item 4 of the scope of patent application 0503-7587TWf; TSMC2001 -1510; yianhou.ptd page 17 569292 VI. Patent application scope Prediction method of process parameter values, in which the semiconductor equipment machine is a semiconductor etching machine. 2 2. The method for predicting a process parameter value of a semiconductor device machine as described in item 21 of the scope of patent application, wherein the process parameter value is an etching time. 2 3. The method for predicting the process parameter value of the semiconductor device machine as described in item 14 of the scope of patent application, wherein the semiconductor device machine is a semiconductor ion implantation machine. 2 4. The method for predicting a process parameter value of a semiconductor device machine as described in item 23 of the scope of patent application, wherein the process parameter value is an ion implantation amount. 2 5. A method for predicting the exposure energy of a semiconductor exposure machine, including. The following steps:-Calculate a first average value of the exposure energy of the plurality of exposure machines for the manufacture of a first product; set the first average value as the exposure energy of one of the semiconductor exposure machines of the first product; The second average value of the exposure energy of the semiconductor exposure machine for manufacturing a plurality of different products; and the redesignation of the exposure energy of the semiconductor exposure machine when manufacturing the first product according to the following equation: 0 4 [女 +1] = [ mx [nxd ^]-ά ^ [/ (:]) + «χ x ^ [fc] -iiv [ic])) / ((η -1) (^ 3-1)) where, c is the exposure energy The number of predictions; m represents the number of exposure machines and the semiconductor exposure machine; 0503-7587TWf; TSMC2001-1510; yianhou.ptd page 18 569292 6. The number of patent applications for the same product and the number of the first product; 4 Min represents the exposure energy when the semiconductor exposure machine manufactured the first product last time; 4k + i] represents the exposure energy at the next time the semiconductor exposure machine manufactures the first product; represents the first average value; Qi㈨ represents the second average value; and YM represents the first average value and the first The average of the two averages. 2 6 · The method for predicting exposure energy of a semiconductor exposure machine as described in item 25 of the scope of patent application, further comprising terminating the prediction of the exposure energy when the number (value) of the exposure energy prediction is greater than a predetermined number. 2 7 · —A computer-readable recording medium that encodes a computer program to enable the prediction of process parameter values for semiconductor equipment machines. The method for predicting the process parameter values of semiconductor equipment machines includes the following steps: A first average value of a process parameter value at the time of manufacturing a first product; setting the first average value to a value of a process parameter of a semiconductor device machine that has not yet processed the first product; calculating the semiconductor device machine for a plurality of different products A second average of process parameter values at the time of manufacture; and The Expectation Maximization (EM) algorithm is used to reset the process parameter values of the semiconductor device machine to manufacture the first product according to the first average value and the second average value. 2 8 · The computer-readable recording medium described in item 27 of the scope of patent application, wherein an expected increase (EM) algorithm is used to reset the semiconductor device according to the first average value and the second average value Machine made the first product 〇503-7587TWf *. TSMC2001 -1510 · yianhou.ptd Page 19 569292 6. The process parameter values of the patented products are specified according to the following equation: [mxinxd ^ [fc] ^ [k])-[nxrn ^ M-[^) / ^ __ ^ ^ where k represents the prediction of process parameter values W represents the number of the equipment and the semiconductor equipment; "represents the number of the different products and the first product; represents the previous one = when the semiconductor equipment machine was used to manufacture the first product Process parameter value; A [Ηι] represents the process parameter value when the semiconductor device machine manufactures the first product; ^ and represents The first average value is represented by the average value of the second average value and the second average value of the edge. 29. The computer-readable medium according to item 28 of the scope of the patent application, wherein the computer program is more capable of terminating the prediction of the process parameter value when the predicted value of the process parameter value is greater than a predetermined number. 30. The computer-readable recording medium as described in item 27 of the scope of patent application, wherein the Suan semiconductor equipment machine is a semiconductor exposure machine. 31. The computer-readable recording medium 'described in item 30 of the scope of patent application, wherein the process parameter value is exposure energy. 3 2 · The computer-readable recording medium as described in item 27 of the patent application scope, wherein the semiconductor equipment machine is a semiconductor polishing machine. 33. The computer-readable recording medium 'described in item 32 of the scope of patent application, wherein the process parameter value is a grinding time. 34. The computer-readable recording medium as described in item 27 of the patent application scope, in which 4 semiconductor equipment tools are semiconductor etching machines. 569292 6. Scope of patent application 35. The computer-readable recording medium described in item 34 of the scope of patent application, wherein the process parameter value is the etching time. 36. The computer-readable recording medium described in item 27 of the scope of the patent application, wherein the semiconductor device machine is a semiconductor ion implantation machine. 37. The computer-readable recording medium described in item 36 of the scope of patent application, wherein the process parameter value is the ion implantation dose. 3 8. —A computer-readable recording medium that encodes a computer program to enable the exposure energy prediction of a semiconductor exposure machine. The exposure energy prediction method of the semiconductor exposure machine includes the following steps: The first average value of the exposure energy during the manufacture of the first product; setting the first average value as the exposure energy of a semiconductor exposure machine that has not yet processed one of the first products; calculating the semiconductor exposure machine for the manufacture of a plurality of different products The second average value of the exposure energy; and the reassignment of the exposure energy when the semiconductor exposure machine manufactures the first product according to the following equation: Heart [Perspective +1] = (^ 3x (»xd [ic]) +« x 1 ^ 71M)-[nxm x ^ 1 [A:]-^^ [A:])) / ((? I- l) (w3-l)) where k is the number of times the exposure energy is predicted; m is the number containing The number of the exposure machines and the semiconductor exposure machine; η represents the number of the different products and the first product; represents the exposure energy when the semiconductor exposure machine manufactured the first product in the previous time; represents the next time 0503-7587TWf; TSMC2001-1510; yianhou.ptd page 21 569292 VI. Patent application scope Exposure energy of the semiconductor exposure machine when manufacturing the first product; 0] represents the first average value; represents the second average value ; And represents the average of the first average and the second average. 39. The computer-readable recording medium described in item 38 of the scope of patent application, wherein the computer program is more capable of terminating the exposure energy when the number of exposure energy prediction times (G value) is greater than a predetermined number. 0503-7587TWf; TSMC2001 -1510; yianhou.ptd page 22