Estimation devices, vacuum valves and vacuum pumps

Abstract

To provide an estimation device capable of accurately extracting and classifying state characteristic values of a vacuum pump.SOLUTION: An estimation device 6 estimates process gas conditions of a system 100 in which a vacuum chamber 1, in which a gas is introduced and subjected to a process treatment, is evacuated by a vacuum pump 3 mounted thereto via a vacuum valve 2. The estimation device estimates a first process gas condition including a type and a flow rate of the introduced gas, on the basis of correlation data between an opening degree of the vacuum valve 2 and an effective evacuation speed of the system 100 for a predetermined gas type, the opening degree of the vacuum valve 2, and a chamber pressure Pc of the vacuum chamber 1.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an estimation device for estimating process conditions in a vacuum device, and a vacuum valve and a vacuum pump having the estimation device. [Background technology]

[0002] Turbomolecular pumps are used as exhaust pumps in various semiconductor manufacturing equipment. However, during the etching process and other processes, reaction products accumulate inside the pump. In turbomolecular pumps, the rotor rotates at high speed with a gap between it and the stator. During etching, reaction products accumulate inside the pump, eventually filling the gap between the rotor and stator and causing the pump to become stuck, preventing rotation. Regarding this reaction product accumulation, methods have been proposed for evaluating the amount of accumulation based on the increase in motor current, as described in Patent Documents 1 and 2. The invention described in Patent Document 1 includes a special health check mode in which a known gas type and gas flow rate are set constant, and the amount of accumulation is evaluated based on the motor current measured in the health check mode.

[0003] The invention described in Patent Document 2 measures the motor current value without setting any special conditions like those in the health check mode described in Patent Document 1, and only uses the measured motor current values ​​that are equal to or greater than a set value.The invention then proposes a method of detecting product deposition by calculating the average value per unit time, arranging them in chronological order, determining the difference from the initial motor current value at the start of use, and detecting when this value exceeds a preset threshold.Furthermore, the motor current value is measured only during times when the motor current value in rated rotation mode is stable, and the above threshold judgment is performed. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2011 / 145444 [Patent Document 2] International Publication No. 2013 / 161399 Summary of the Invention [Problem to be solved by the invention]

[0005] As described above, when evaluating the amount of product deposition by comparing the increase in motor current value over time with a threshold value, if the pump exhaust conditions are not approximately the same when measuring the motor current value, changes in the motor current value due to factors other than product deposition (for example, changes in gas flow rate) will also be measured, resulting in the problem of not being able to perform an appropriate evaluation of the amount of deposition.

[0006] The invention described in Patent Document 1 does not have the above-mentioned problems because it measures the motor current value Im in a health check mode under the same conditions. However, it is necessary to periodically stop the manufacturing process and set up a health check mode period. Furthermore, the health check mode requires the installation of gas equipment and the addition or modification of operation modes in the equipment, which is a burden.

[0007] In the invention described in Patent Document 2, the motor current value is measured and calculated under the conditions of "only when the motor current value in rated rotation mode is stable" and "above a set value," and the calculation is performed under the conditions of "calculating the average value per unit time," "arranging them in chronological order," and "obtaining the difference from the initial motor current value at the start of use," thereby minimizing variation in the measurement conditions of the motor current even if the exhaust conditions such as gas type and gas flow rate are unknown. However, since processes such as etching and film formation are performed in the equipment using various process condition recipes, the obtained motor current value will include current values ​​measured under various conditions, and even if the calculation process described above is performed, it cannot be used as a practical selection extraction condition. [Means for solving the problem]

[0008] An estimation device according to a first aspect of the present invention is an estimation device that estimates process gas conditions in a system in which a vacuum chamber, into which gas is introduced and processing is performed, is evacuated by a vacuum pump attached via a vacuum valve, and estimates a first process gas condition including information on the gas type and gas flow rate of the introduced gas based on correlation data between the valve body opening of the vacuum valve for a specified gas type and the effective exhaust speed of the system, and the chamber pressure of the vacuum chamber. An estimation device according to a second aspect of the present invention estimates at least one of a first process gas condition, a second process gas condition, a third process gas condition, and a fourth process gas condition, and a fifth process gas condition, generates a plurality of data sets in time series, each data set including the estimated process gas conditions and a state characteristic value representing the operating state of a vacuum pump, extracts only a data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets, and sets a classification frame in which the process gas condition can be considered to be approximately the same as a predetermined state for one predetermined process gas condition or predetermined plurality of process gas conditions among the plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame. A vacuum valve according to a third aspect of the present invention is provided between a vacuum chamber in which a process is performed and a vacuum pump, and is equipped with the estimation device of the first aspect, and determines the product deposition state in the vacuum pump based on the product deposition amount estimated by the estimation device. A fourth aspect of the present invention is a vacuum valve into which at least one of first, second, third, and fourth process gas conditions, a fifth process gas condition, and a state characteristic value representing the operating state of a vacuum pump are input, and the vacuum valve generates a plurality of data sets in chronological order, each data set including at least one of the first to fourth process gas conditions, the fifth process gas condition, and the state characteristic value representing the operating state, extracts only the data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets, and sets a classification frame in which the process gas condition can be considered to be approximately the same as a predetermined state for one predetermined process gas condition or predetermined plurality of process gas conditions among the plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame. A vacuum pump according to a fifth aspect of the present invention comprises a pump rotor driven by a motor and the estimation device of the first aspect, and determines the product deposition state within the vacuum pump based on the product deposition amount estimated by the estimation device. A vacuum pump according to a sixth aspect of the present invention is a vacuum pump to which at least one of first, second, third, and fourth process gas conditions, and a fifth process gas condition, is input, and the vacuum pump generates a plurality of data sets in time series, each data set including at least one of the first to fourth process gas conditions, the fifth process gas condition, and a state characteristic value representing the operating state of the vacuum pump, extracts only the data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets, and sets a classification frame in which the process gas condition can be considered to be approximately the same as a predetermined state for one predetermined process gas condition or predetermined plurality of process gas conditions among the plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame. [Effects of the Invention]

[0009] The estimation device of the present invention can estimate process gas conditions including information on the gas type and gas flow rate of the introduced gas.The vacuum valve of the present invention can determine the product deposition state in the vacuum pump based on the product deposition amount estimated by the estimation device, and can extract only a data set representing an equilibrium state from multiple data sets including process gas conditions and state characteristic values, and classify the extracted data set into the classification frame.The vacuum pump of the present invention can determine the product deposition state in the vacuum pump based on the product deposition amount estimated by the estimation device, and can extract only a data set representing an equilibrium state from multiple data sets including process gas conditions and state characteristic values, and classify the extracted data set into the classification frame. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram showing a schematic configuration of a vacuum processing apparatus. [Figure 2] FIG. 2 is a schematic diagram showing the relationship between the process and the chamber pressure Pc. [Figure 3] FIG. 3 is a diagram illustrating the process pressure Ps. [Figure 4] FIG. 4 is a diagram showing the relationship between the valve element opening θ and the effective exhaust speed Se(θ). [Figure 5] FIG. 5 is a diagram showing the tendency of the effective exhaust speed Se(θ) in the low opening region and the high opening region. [Figure 6] FIG. 6 is a diagram for explaining a method for calculating (ao, Qino). [Figure 7] FIG. 7 is a diagram illustrating the relationship between the measurement timing of the motor current value and the process information data set. [Figure 8] FIG. 8 is a diagram showing a configuration in which the function of the estimation device is provided in another device, and shows the cases in which it is provided in a vacuum valve, a vacuum pump, and a main control device. [Figure 9] FIG. 9 is a graph showing an example of a current value function. [Figure 10] FIG. 10 is a flowchart showing a method for estimating the product deposition amount according to the third embodiment. [Figure 11] FIG. 11 is a diagram showing an example of a current value function calculated in time series. [Figure 12] FIG. 12 is a diagram showing an example of specific current values ​​calculated in time series. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. -First embodiment- 1 is a block diagram showing a schematic configuration of a vacuum processing apparatus 100. Examples of the vacuum processing apparatus 100 include an etching apparatus and a film forming apparatus. The vacuum processing apparatus 100 includes a vacuum chamber 1 in which a process is performed, a vacuum valve 2 attached to the vacuum chamber 1, a vacuum pump 3 attached to the vacuum chamber 1 via the vacuum valve 2, a flow rate controller 4 that controls the flow rate of gas introduced into the vacuum chamber 1, a vacuum gauge 5 that measures the pressure in the vacuum chamber 1, an estimation device 6, and a main control device 7 that controls the vacuum processing apparatus 100.

[0012] The vacuum valve 2 is an automatic pressure control valve that automatically adjusts the opening degree θ of the valve element 21 based on the target pressure Ps input from the main control device 7 and the chamber pressure Pc input from the vacuum gauge 5 so that the chamber pressure Pc becomes the target pressure Ps. The vacuum valve 2 outputs the chamber pressure Pc measured by the vacuum gauge 5 to the main control device 7. The vacuum pump 3 in FIG. 1 is a turbomolecular pump, and the pump rotor 32 is rotated at high speed by a motor. An auxiliary pump (not shown), such as a dry pump, is connected to the exhaust port 31. The vacuum valve 2 stores correlation data Se_reference(θ), which is the effective pumping speed for a reference gas. The chamber pressure Pc, the target pressure Ps, and the valve element opening degree θ are constantly input to the estimation device 6 from the vacuum valve 2, and the correlation data Se_reference(θ) is also input. The correlation data Se_reference(θ), which is the relationship between the effective pumping speed and opening degree of a known reference gas, only needs to be transmitted and input once. Furthermore, the estimation device 6 receives an input of the motor current value Im from the vacuum pump 3 as a state characteristic value that has a strong correlation with the amount of product deposition.

[0013] The estimation device 6 includes an estimation unit 61 and a storage unit 62. A data set generation unit 61A of the estimation unit 61 estimates the process conditions in the vacuum chamber 1 based on the chamber pressure Pc, target pressure Ps, valve element opening θ, and correlation data Se_reference(θ) input from the vacuum valve 2, and generates a process information data set consisting of a group of data related to the process gas conditions. In addition to generating the process information data set described above, the data set generation unit 61A also generates a deposition information data set by adding the motor current value Im input from the vacuum pump 3 to the process information data set. An deposition amount determination unit 61B of the estimation unit 61 extracts a deposition information data set to be used for deposition amount determination and determines the deposition amount based on the extracted deposition information data set. Deposition amount determination information is output from the estimation device 6, and in the example shown in FIG. 1, it is output to the main control device 7.

[0014] The estimation device 6 includes a processing unit such as a microcomputer having a CPU, memory (ROM, RAM), and peripheral circuits, and realizes the functions of the estimation unit 61 by a software program stored in the ROM. The storage unit 62 is configured by the memory of the microcomputer. Alternatively, it may be configured by a digital processor such as an FPGA (Field Programmable Gate Array) and its peripheral circuits.

[0015] As mentioned above, reaction products are deposited inside the vacuum pump 3 due to the exhaust of the process gas, and the motor current Im of the vacuum pump 3 is known as a characteristic value that is closely correlated with the amount of deposited products. Since the motor current Im depends on the process conditions in the vacuum chamber 1 evacuated by the vacuum pump 3, when using the motor current Im as a state characteristic value that indicates the amount of deposited products, it is necessary to also consider the process conditions at the time of current measurement.

[0016] FIG. 2 is a schematic diagram showing the relationship between process events (also called process events) and chamber pressure Pc. In the example shown in FIG. 2, three types of process events (e.g., etching and film deposition) labeled A, B, and C are repeatedly performed on wafers transferred into vacuum chamber 1. When an unprocessed wafer is loaded at timing W1in, processes A, B, and C are sequentially performed on the wafer. After the processed wafer is unloaded (W1out) and an unprocessed wafer is loaded (W2in), processes A, B, and C are sequentially performed on the loaded wafer. Process A, which begins after the wafer is loaded and process gas is introduced, is performed when the chamber pressure Pc stabilizes at target pressure Ps1 (equilibrium state) R1. Similarly, process B is performed when the chamber pressure Pc stabilizes at target pressure Ps2 (equilibrium state) R2. Furthermore, process C remains at the same target pressure Ps2 as process B, but because a different gas is introduced and replaced when process B switches to process C, the chamber pressure Pc temporarily deviates from the equilibrium state and then stabilizes at the target pressure Ps2 again (equilibrium state) R3. Normally, etching or film deposition processes are performed when the chamber pressure Pc reaches the target pressure Ps after passing through a transient response process of automatic pressure adjustment and remains at a constant pressure.

[0017] The timing of maintenance for deposit removal is determined by the increase in the motor current Im. Maintenance is determined when the increase in the motor current Im relative to the motor current Im when no reaction products are present exceeds a predetermined threshold. When reaction products accumulate in the pump, the load on the motor that rotates the pump rotor 32 increases. However, because the motor current Im also changes with changes in the gas load during repeated process events, the measured motor current Im includes both changes due to gas load fluctuations and changes due to an increase in the amount of deposition. Therefore, unless the motor current Im is acquired under the same gas load conditions, the amount of deposition cannot be accurately determined based on the increase in the current. For example, if the sampling data includes a mixture of motor current Im measured in state R1, state R2, and state R3 (see Figure 2), the amount of deposition cannot be accurately evaluated.

[0018] In this embodiment, by adding a process information data set consisting of a group of data related to process gas conditions to the motor current value Im, it is possible to determine the deposition amount based on the motor current value Im under the same process gas conditions (gas load conditions).

[0019] [1.Correlation Data Se_Standard(θ)] First, we will explain the correlation data Se_reference(θ) stored in the vacuum valve 2. As shown in FIG. 1, a process gas exhaust system for semiconductor processes (such as dry etching) comprises at least a vacuum chamber 1 where the process is performed, a gas introduction system (such as a flow rate controller 4), a vacuum gauge 5, a vacuum valve 2, and a vacuum pump 3 (including a backing pump). The effective pumping speed Se of the exhaust system is affected by the structure of the vacuum chamber 1, but is essentially determined by the conductance of the vacuum valve 2 and the pumping speed of the vacuum pump 3 downstream of the valve. Because the conductance of the vacuum valve 2 changes depending on the valve disc opening θ, the effective pumping speed Se is expressed as a function of the valve disc opening θ. Note that the effective pumping speed Se depends not only on the opening θ but also on the introduced gas flow rate Qin. However, because the influence of the introduced gas flow rate Qin is small, this embodiment will be described as a function of only the opening θ, such as Se(θ).

[0020] In this embodiment, correlation data Se_reference(θ) between the valve element opening θ of the vacuum valve 2 and the effective pumping speed Se for a known reference gas type (e.g., Ar gas) is stored in advance in the storage unit 62 of the estimation device 6. The correlation data Se_reference(θ) can be obtained, for example, by introducing a known reference gas type (e.g., Ar gas) into an exhaust system equipped with the vacuum valve 2 and vacuum pump 3 in a specified vacuum chamber and measuring the effective pumping speed Se at various valve element openings θ. The correlation data Se_reference(θ) is obtained before shipping to the user and stored in the storage unit 62 of the estimation device 6. Alternatively, a specified reference gas type and gas flow rate is introduced into the user's exhaust system, and the effective pumping speed Se is measured and stored in the storage unit 62 of the estimation device 6.

[0021] The correlation data Se_reference(θ) is also used in the automatic pressure regulation control of the vacuum valve 2, and is therefore normally stored in the vacuum valve 2. Therefore, the correlation data Se_reference(θ) stored in the vacuum valve 2 may be transmitted to the estimation device 6 via a communication function, as shown in Fig. 1. This data transmission only needs to be done once.

[0022] [2. Process gas conditions] Next, we will explain the process gas conditions. While other process conditions include the high-frequency power required for plasma excitation, we consider the following process gas conditions related to the introduced gas: reference gas equivalent flow rate Qin_a, gas species relative value a, gas flow rate Qin, process pressure Ps, and event time length Time_spn. However, the reference gas equivalent flow rate Qin_a depends on the gas species and gas flow rate. Depending on the situation, the reference gas equivalent flow rate Qin_a or the gas species relative value a and gas flow rate Qin may be used. Furthermore, even if the above process gas conditions can be estimated, there may be cases where the measurement timing is not desirable. Specifically, a chamber pressure Pc in an equilibrium state is desirable, but a transient state is undesirable. Therefore, in addition to the above process gas conditions, it is preferable to also use measurement recommendation timing information flag_OK as a process gas condition. The estimation device 6 generates a process information data set containing the above multiple process gas conditions: [a, Qin, Ps, Time_spn, flag_OK] or [Qin_a, Ps, Time_spn, flag_OK]. Here, five or four process gas conditions are used as a set, but the number can be selected depending on the situation. For example, only one process gas condition may be used, and even if only one process gas condition is included, it will be called a process information data set.

[0023] (1-1. Process pressure Ps) The process pressure Ps is set and input from the main controller 7 to the vacuum valve 2 as the target pressure for automatic pressure adjustment. Figure 3 shows the change in the target pressure Ps input to the estimation device 6 relative to the change in the chamber pressure Pc shown in Figure 2, indicated by a dashed line. The target pressure Ps is input, and the chamber pressure Pc is adjusted to the target pressure Ps by the automatic pressure adjustment of the vacuum valve 2. Typically, after a transient response process of automatic pressure adjustment, the target pressure Ps is reached and maintained at a constant pressure before etching or film formation processes (process events A, B, and C in Figure 3) are performed. Therefore, the process gas condition can be established when the chamber pressure Pc reaches a nearly equilibrium state and becomes the process pressure (target pressure) Ps. That is, in Figure 3, during the period when Pc = Ps1 (process event A), process pressure Ps1 is generated as the process gas condition. During the period when Pc = Ps2 (process events B and C), process pressure Ps2 is generated as the process gas condition. During other periods, the process gas condition (process pressure Ps) is not generated. Here, "other periods" refers to periods of high vacuum where no wafer loading or unloading processes are performed. For convenience, the dashed line of the target pressure Ps is also drawn for this period in Figure 3, but automatic pressure adjustment is not always necessary; the valve element may be driven to its full open position to forcibly exhaust any process gas remaining in the vacuum chamber.

[0024] (1-2. Event duration Time_spn) The event duration Time_spn is the time length from the start to the end of a process event. For example, the event duration Time_spn can be estimated based on whether the target pressure Ps changes or deviates from the equilibrium state. In the example shown in Figure 3, the target pressure Ps changes from high vacuum to Ps1 at t=t1 and from Ps1 to Ps2 at t=t2. t=t1 is considered the start of event A, and t=t2 is considered the end of process event A. Therefore, Time_spn = t2 - t1 is the event duration of process event A, with target pressure Ps1. Meanwhile, t=t2 is both the end of process event A and the start of the next process event B. However, at the end of process event B, the target pressure Ps remains unchanged at Ps2. Therefore, the end of process event B cannot be determined based on the change in target pressure Ps. Therefore, if there is no change in target pressure Ps, the end of process event B can be estimated based on the deviation of chamber pressure Pc from target pressure Ps2. In FIG. 3, the timing t=t13 can be considered as the end of process event B, so Time_spn=t13−t2 is the event time length of process event B.

[0025] t=t13 is the end of process event B and the start of the next process event, C. The end of process event C is either t=t3 or t=t14 in Figure 3. If we estimate the end based on the change in target pressure Ps from Ps2 to high vacuum, then the end is t=t3, and if we estimate it based on the deviation from the equilibrium state, then the end is t=t14. In either case, the end is approximately the same time, so if we consider t=t3 to be the end of process event C, then Time_spn=t3-t13 is the event duration of process event C, where the target pressure is Ps2.

[0026] As described above, the chamber pressure Pc just before the end of an event is typically in a constant equilibrium state at the target pressure Ps, which is the process pressure. Therefore, a process event is also the period from one equilibrium state to another. For example, if the high vacuum state just before process event A in FIG. 3 is also considered an equilibrium state, the period (t12-t11) from this equilibrium state (t=t11) to the equilibrium state at the end of process event A (t=t12) is estimated as the event duration Time_spn of process event A. Similarly, the period (t13-t12) from the equilibrium state at the end of process event A (t=t12) to the equilibrium state at the end of the next process event B (t=t13) is estimated as the event duration Time_spn of process event B. Furthermore, the period (t14-t13) from the equilibrium state at the end of process event B (t=t13) to the equilibrium state at the end of the next process event C (t=t14) is estimated as the event duration Time_spn of process event C.

[0027] In cases where a process event starts or ends while the target pressure Ps remains constant, as shown in FIG. 3, a change in the introduced gas usually causes a deviation from the equilibrium state. That is, the chamber pressure Pc deviates from the target pressure Ps, then converges back to the target pressure Ps and returns to equilibrium. Therefore, the event duration Time_spn is estimated by determining the start or end based on whether or not there is a deviation from the equilibrium state described above. However, if the change in the introduced gas is small, there may be almost no deviation from the equilibrium state. In such cases, the process gas conditions are determined from a change in the reference gas equivalent flow rate Qin_a, which will be described later.

[0028] The event duration Time_spn is undetermined from the time the start time can be estimated until the end time can be estimated. In other words, it is constantly updated by the time value obtained by subtracting the start time from the current time, and the event duration Time_spn value is determined by stopping the update at the end time.

[0029] (1-3. Gas type relative value a, gas flow rate Qin) As shown in Figure 4, the effective pumping speed Se(θ) monotonically increases with the valve disc opening θ. In Figure 4, lines L1, L2, and L3 represent the effective pumping speeds Se(θ) for helium gas, argon gas, and xenon gas, respectively. Furthermore, if the effect of the structure of the vacuum chamber 1 on the effective pumping speed Se(θ) can be ignored, the effective pumping speed Se(θ) is determined by the pumping speed Sp of the vacuum pump 3 and the conductance Cv(θ) of the vacuum valve 2, as shown in the following equation (1): Se(θ)=1 / (1 / Cv(θ)+1 / Sp) …(1) The relationship between the opening θ and the effective pumping speed Se(θ) generally shows the tendency shown in Figure 5. That is, in the effective pumping speed Se(θ) shown by line L10, in the region C where the opening θ is small (low opening), the conductance Cv (line L11) of the vacuum valve 2 becomes dominant, and in the region D where the opening θ is large (high opening), the pumping speed Sp (line L12) of the vacuum pump 3 becomes dominant.

[0030] Furthermore, under normal semiconductor process conditions, in the small valve element opening θ region, gas flow can be considered to be in the molecular flow region, and it is known that, if the molecular weight of the introduced gas species is M, then conductance Cv(θ) is inversely proportional to √M. Therefore, in the small valve element opening θ region, the relationship shown in the following equation (2) holds between the effective pumping speed Se of an arbitrary gas species (molecular weight M) and the effective pumping speed Se_reference of a reference gas species (molecular weight M0). Se(θ)=(√(M0 / M))×Se_standard(θ) …(2)

[0031] In equation (2), √(M0 / M) represents the ratio between the effective pumping speed Se(θ) and the effective pumping speed Se_reference(θ) of the reference gas species, and it can be seen that the gas species can be expressed using this ratio. Below, the gas species relative value a is defined using the following equation (3). Gas species relative value a = √(M0 / M) = Se(θ) / Se_reference(θ) ... (3) The gas species relative value a is greater than 1 for gases lighter than the reference gas species, and is smaller than 1 for gases heavier than the reference gas species. Note that, although the above describes the properties in the low opening region where the valve disc opening θ is small, the opening range used in normal processes generally overlaps with the above-mentioned low opening region, and therefore will not be specifically referred to as the low opening region in the following description.

[0032] To optimize automatic pressure regulation, the vacuum valve 2 may estimate the type and flow rate of the introduced gas. For example, if a control period is set in which the valve element position is fixed at a certain position during the initial stage of the transient response period of the chamber pressure Pc, when the previous process event ends and the current process event begins, the gas type and flow rate can be estimated during that period. Typically, the gas introduction and replacement is completed quickly by the flow controller 4 in a time significantly shorter than the fixed valve element position period. Therefore, if a period is set in which the valve element position θ is fixed, the gas type, flow rate, and position are constant except for that initial stage, so the effective pumping speed Se can be considered to be a constant value Se0. In this case, the pumping equations (4A) and (4B) hold true at any time t1 or t2 during the period in which the valve element position θ is fixed, excluding the initial stage. By eliminating Se0 from equations (4A) and (4B), the gas flow rate Qin can be estimated using the following equation (5). Qin=V×dPr1 / dt+Se0×Pr1 …(4A) Qin=V×dPr2 / dt+Se0×Pr2 …(4B) Qin=V×(Pr2×dPr1 / dt-Pr1×dPr2 / dt) / (Pr2-Pr1) …(5) Furthermore, from equations (4A) and (4B), the effective pumping speed Se0 can be expressed by the following equation (6). Se0=-V×(dPr2 / dt-dPr1 / dt) / (Pr2-Pr1) …(6)

[0033] In equations (4A), (4B), (5), and (6), Pr1 and Pr2 are the pressure measurement values ​​at t=t1 and t2, and dPr1 / dt and dPr2 / dt are the pressure change rates over time of the pressure measurement values ​​at t=t1 and t2. V is the chamber volume. Note that, in order to optimize automatic pressure regulation, the chamber volume value V is usually obtained in advance by measurement during a calibration process that is performed only once when the vacuum valve 2 is installed in the vacuum chamber.

[0034] The effective pumping speed Se0 calculated by equation (6) is the effective pumping speed at a fixed valve disc opening θ. This effective pumping speed Se0, for which the gas species is unknown, can be expressed as Se0(θ) = a × Se_reference(θ) using the effective pumping speed Se_reference(θ) of the reference gas species and the gas species relative value a. Using equation (6), the gas species relative value a can be expressed as in the following equation (7). a=-(V / Se_Reference(θ))×(dPr2 / dt-dPr1 / dt) / (Pr2-Pr1) …(7)

[0035] It is possible to use one estimated value of (a, Qin) calculated in a transient state during such pressure regulation control as (a, Qin) as is, or to use it as the reference gas equivalent flow rate Qin_a, but since it is estimated in a dynamic process called a transient state, there is generally a large error. Therefore, the following measures are applied to reduce the error.

[0036] When there are multiple (N) estimated values ​​of (a, Qin) obtained during the transient state of pressure adjustment, they are represented as (a_est1, Qin_est1), (a_est2, Qin_est2), (a_estk, Qin_estk), . . ., (a_estN, Qin_estN), where k = 1, 2, . . ., N. A process event that is valid as data always reaches an equilibrium state at the end, and the reference gas equivalent flow rate Qin_a can be calculated using Se_reference(θ) × Ps. Therefore, on the Qin-a plane shown in Figure 6, the equilibrium state is represented by a point on the line Qin = Qin_a × a, and the N estimated values ​​mentioned above are scattered around the line.

[0037] Therefore, the optimal point can be said to be the point (ao, Qino) on the line where the sum of squares (ΣLk^2) of the distances Lk between the line and each estimated point (k points) is smallest. To find this, apply Lagrange's method of undetermined multipliers with the constraint Qino = Qin_a×ao. By solving the simultaneous equations where the partial derivatives of F=(ΣLk^2)+λ(Qino-Qin_a×ao) with respect to ao, Qino, and the Lagrange multiplier λ are 0, we obtain the following equations (8A) and (8B). ao=Σ{Qink+(1 / Qin_a)×ak} / {N×(Qin_a+1 / Qin_a)} …(8A) Qino=Qin_a×ao …(8B) This (ao, Qino) can be applied as the gas species relative value and gas flow rate of the process event. Also, if there is a significant change in (ao, Qino) between the previous event and the current event, it can be determined that the current event has ended even if there is no change in Qin_a.

[0038] (1-4. Reference gas equivalent flow rate Qin_a) In a typical process, multiple gas species are typically introduced, with different flow rates for each gas species. The dissociation of gas molecules due to plasma excitation can also be considered a state in which multiple gas species are introduced. When a mixed gas consisting of multiple gas species is introduced, the pumping equation shown in Equation (9) below applies to each gas species. In other words, the pumping equation for a mixed gas is a superposition of the pumping equations for each gas species contained in the mixed gas. Equation (9) represents the pumping equation for N gas species (k = 1, 2, 3, . . . , N), where Q, P, and S(θ) are the gas flow rate, pressure (partial pressure), and effective pumping speed for gas species k, and V is the chamber volume. dP / dt represents the time derivative of pressure P. The effective pumping speed S(θ) for gas species k is expressed as Equation (10) below, using the relative value ak of gas species k relative to the reference gas. Qink=V×dPk / dt+Sek(θ)×Pk…(9) Sek(θ)=ak×Se_standard(θ) …(10)

[0039] Dividing both sides of the equation (9) for N exhausts for N types of gas (k=1, 2, 3, . . . , N) by a k, and then adding the left and right sides together, gives the following equation (11). Σ(Qink / ak)=V×Σ(dPk / dt / ak)+Se_standard(θ)×Σ(Pk) …(11) k=1,2,…, N Since (Pk) in equation (11) is the sum of partial pressures, it is the chamber pressure Pc measured by the vacuum gauge 5. Equation (11) is an equation related to a mixed gas, but if the mixed gas expressed by the relationship in equation (11) is considered to be an average single gas having a gas species relative value a_total, the equation corresponding to equation (11) can be expressed as the following equation (12). Qin_total / a_total=V×dPc / dt / a_total+Se_standard(θ)×Pc …(12)

[0040] In equation (12), Qin_total corresponds to the flow rate of the mixed gas, so Qin_total = ΣQink, and the left side = ΣQink / a_total. If the left side of equation (12) corresponds to the left side of equation (11) and the two left sides are considered equal, then Σ(Qink / ak) = ΣQink / a_total is obtained. In other words, the gas species relative value a_total of an average single gas corresponding to the mixed gas is expressed by the following equation (13). a_total=ΣQink / Σ(Qink / ak) …(13) In particular, in the pressure equilibrium state, dPk / dt=0 (k=1, 2, ... K), so in the pressure equilibrium state, equations (11) and (12) can be expressed as equations (14) and (15), respectively. Σ(Qink / ak)=Se_criteria(θ)× Σ(Pk) ( k=1,2,…, N) …(14) Qin_total / a_total=Se_standard(θ)×Pc …(15)

[0041] In both equations (14) and (15), the right-hand side is the product of the effective pumping speed and pressure of the reference gas, so Qin_total / a_total (=Σ(Qink / ak)) on the left-hand side can be interpreted as the gas flow rate equivalent to the reference gas. Hereinafter, Qin_total / a_total (=Σ(Qink / ak)) will be expressed as the reference gas equivalent flow rate Qin_a. The reference gas equivalent flow rate Qin_a is defined by the following equation (16). Qin_a=Qin_total / a_total(=Σ(Qink / ak)) …(16)

[0042] When the flow rate of gas species lighter than the reference gas is dominant in the process gas (gas mixture), Σ(Qink / ak) will be dominated by the Qink / ak of the lighter gas species. Therefore, Qin_a is roughly calculated by dividing Qin_total by the gas species relative value ak of the lighter gas species, and Qin_a tends to be smaller. Conversely, when heavier gas species are dominant, Qin_a tends to be larger. Therefore, in the following, the reference gas equivalent flow rate Qin_a expressed by equation (16) will be used as an index of gas characteristic values ​​(process gas conditions) including gas flow rate and gas species information, not only for mixed gases but also for single gas species. Note that in the following, Qin_total and a_total, expressed as Qin_total = ΣQink and a_total = ΣQnk / Σ(Qink / ak), will be expressed as Qin and a, respectively, for both mixed gases and single gas species. That is, equation (16) can be expressed as the following equation (17): Qin_a= Qin / a(=Σ(Qink / ak)) …(17)

[0043] Since the reference gas equivalent flow rate Qin_a is a gas characteristic value defined as Qin_total / a_total, it not only increases when the gas flow rate increases, but also when the average gas species becomes heavier and a_total decreases. In other words, an increase in gas flow rate and an increase in average molecular weight M appear as an increase in the reference gas equivalent flow rate Qin_a, and a decrease in gas flow rate and a decrease in average molecular weight M appear as a decrease in the reference gas equivalent flow rate Qin_a.

[0044] As described above in the explanation of the event time length Time_spn, in the case where the target pressure Ps remains constant even when the process event changes, and further, when the change in the introduced gas due to the change in the process event is small, there is little deviation from the equilibrium state, and it may be impossible to estimate the start or end of the process event. Even in such a case, it may be possible to determine the end of the event from the change in the reference gas equivalent flow rate Qin_a.

[0045] Even if the target pressure value remains unchanged, if the gas type and gas flow rate change (due to a change in process event), i.e., when the previous event ends and the current event begins, the vacuum valve 2 automatically adjusts the chamber pressure Pc to maintain a constant target pressure Ps. As a result, the valve travel θ generally changes from the previous event. This change in travel can be used to determine the start or end of a process event. From equations (15) and (16) above, Qin_a = Se_reference(θ) × Ps is established in the equilibrium state of each process event with a constant target pressure Ps. Therefore, a change in valve travel θ ultimately manifests as a change in Qin_a. The reference gas equivalent flow rate Qin_a, calculated based on the valve travel θ and the target pressure Ps, can be used as information on the gas type and gas flow rate of a process event. Furthermore, even if there is no deviation from equilibrium between process events, the end of an event can be determined from a change in the value of Qin_a.

[0046] (1-5. Measurement recommended timing information flag_OK) Consider a case in which the vacuum pump 3 (e.g., a turbomolecular pump) measures its motor current value Im itself and sequentially transmits the data to the estimation device 6. If there is a change in the introduced gas that significantly increases the gas load at the start of the current event period, the rotor rotation speed will temporarily drop from the rated rotation speed due to insufficient torque and then return to the rated rotation speed. At this time, the motor current value Im will suddenly increase and overshoot, before converging to a current value greater than before the change. Because the motor current changes during the event period in this way, it is preferable to obtain a measurement value when the target pressure Ps is in equilibrium as a representative value of the process event.

[0047] The recommended measurement timing information flag_OK is an indicator that indicates that the target pressure Ps has been roughly equilibrated. When the recommended measurement timing information flag_OK=1, that timing is the preferred measurement timing, indicating an equilibrium state. On the other hand, when the recommended measurement timing information flag_OK=0 (non-equilibrium state), the measurement timing is not preferred, and the motor current value Im measured at that timing is not used as the motor current value Im for determining the deposition amount. As an example of the recommended measurement timing information flag_OK, if the chamber pressure Pc is within Ps±0.02Ps, i.e., a ±2% range, for 5 seconds, it is determined to be in an equilibrium state, and the flag value flag_OK is set from 0 to 1.

[0048] FIG. 7 illustrates the relationship between the measurement timing of the motor current value Im and the process information data set. Similar to FIG. 3, it shows a dashed line indicating the target pressure Ps and a solid line indicating the chamber pressure Pc. Here, we consider the transition of the data set [Ps, Time_spn, flag_OK]. Δts is the time interval for the estimation calculation. While a short Δts is generally preferable, for simplicity, we set Δts = 3 seconds and use ts1 to ts8 as the times related to process event B. The start and end of the event duration Time_spn are defined as the timing of the change in the target pressure Ps or the deviation from the equilibrium state. That is, the start of process event B is defined as the timing of the change in the target pressure Ps from Ps1 to Ps2 (ts1 in FIG. 7), and the end is defined as the timing of the deviation from the equilibrium state (ts8 in FIG. 7).

[0049] The process pressure Ps is estimated to be Ps2 at least from t=ts1 to t=ts8. Here, the target pressure Ps first becomes equal to Ps2 at t=ts1, which is the start time. The chamber pressure Pc first deviates from the equilibrium state near Ps at t=ts8, which is the end time. Since the event duration Time_spn is calculated by subtracting the start time from the current time, as described above, a provisional value is estimated, such as Time_spn=ts3-ts1 at measurement timing ts3 and Time_spn=ts7-ts1 at measurement timing ts7. Then, when the end time ts8 is reached, the event duration Time_spn=ts8-ts1 for process event B is finalized, and all data related to the event duration Time_spn from measurement timing ts1 to ts8 is updated from the provisional value to the final value, Time_spn=ts8-ts1.

[0050] For the measurement recommendation timing information flag_OK, the chamber pressure Pc deviates from the process pressure Ps (deviation condition = within the vicinity of Ps2 for less than 5 seconds) at ts1 to ts5, so flag_OK = 0 is set, and at ts6 and ts7, the chamber pressure Pc is in an equilibrium state (equilibrium condition = within the vicinity of Ps2 for 5 seconds or more), so flag_OK = 1 is set. Also, at the end time ts8, the chamber pressure Pc deviates from Ps2 again (within the vicinity of Ps2 for less than 5 seconds), so flag_OK = 0 is set.

[0051] Therefore, the process information data set [Ps,Time_spn,flag_OK] for measurement timings ts1 to ts8 is, in order, [Ps2,ts8-ts1,0], [Ps2,ts8-ts1,0], [Ps2,ts8-ts1,0], [Ps2,ts8-ts1,0], [Ps2,ts8-ts1,0], [Ps2,ts8-ts1,0], [Ps2,ts8-ts1,1], [Ps2,ts8-ts1,1], [Ps2,ts8-ts1,0]. In this case, the motor current value Im measured between measurement timings ts6 and ts7, when the recommended measurement timing information flag_OK is 1, can be said to be data that can be used for deposition amount evaluation.

[0052] [3. Generation of sediment information dataset] As described above, the estimation device 6 generates a process information data set [a, Qin, Ps, Time_spn, flag_OK] or [Qin_a, Ps, Time_spn, flag_OK] that includes a set of multiple process gas conditions. The measurement recommendation timing information flag_OK is an index indicating whether or not an equilibrium state is reached. For example, if the equilibrium state is not reached and the process is terminated due to a change in the setting of the target pressure Ps, the flag_OK flag is not set, and the motor current value Im in that case is determined to be not a recommended value.

[0053] Furthermore, the estimation device 6 generates a deposition information dataset by adding a state characteristic value input from the vacuum pump 3, for example, the motor current value Im, which is a state characteristic value for evaluating the deposition amount, to the generated process information dataset. Since the measurement time Time_msr is constantly transmitted and input from the vacuum pump 3 together with the motor current value Im, the measurement time Time_msr is also added to the deposition information dataset, and the deposition information dataset is expressed as [a, Qin, Ps, Time_spn, flag_OK; Im, Time_msr] or [Qin_a, Ps, Time_spn, flag_OK; Im, Time_msr].

[0054] Note that the event start time or end time may be used instead of the measurement time Time_msr. Since the end timing of the event duration Time_spn is naturally undetermined and occurs suddenly, data sets for which the flag_OK flag is set are provisionally stored, for example, the past five data sets at one-second intervals. Then, when the end of the event is determined, only one set of the provisionally stored data sets is stored and the rest are discarded. The same process is repeated for the next process event. For example, in the example shown in FIG. 7, the interval is three seconds, so the past two data sets are provisionally stored. The event is determined to end at t=ts8, and one set of the past two data sets, including the measurement time ts7, which is provisionally stored as a data set with flag_OK=1, is stored. Since the accumulation amount evaluation only requires the change over time in the motor current value Im, multiple data sets for one process event are not necessarily required as accumulation data. As described above, it is sufficient to store one set of data sets with the flag_OK flag set.

[0055] At the end of an event, the data sets (process information data set and deposition information data set) for that process event are finalized, and this finalized data set (process information data set or deposition information data set) is transmitted and output to each peripheral component device as necessary. In the configuration shown in Figure 1, the estimation device 6 not only generates the deposition information data set but also evaluates and determines the product deposition, so there is no need to output it to other peripheral component devices.

[0056] (Dataset to be stored) The monitoring period for preventive maintenance ranges from several months to several years; here, we will use a one-year example. The event duration Time_spn for each process event for which a dataset is acquired varies from several seconds to several tens of minutes, and is longer for narrower and deeper holes in the etching process, and longer for thicker dense films in the film deposition process. For example, if the event duration Time_spn is assumed to be 60 seconds, the number of datasets will be approximately 500,000 over one year of continuous operation. If datasets are managed on a dedicated PC, all data can be stored, but if they are stored and saved in component equipment such as vacuum valve 2 and vacuum pump 3, storage capacity limitations are likely to be strict. In such cases, it is best to define classification frameworks into multiple classes in advance, perform classification and extraction processing each time, discard datasets that are not extracted, and store and save only the classified and extracted datasets.

[0057] [4. Classification and extraction of data sets] The definite data set for each process event is stored in a memory unit of the estimation device 6. The estimation device 6 classifies and extracts data sets that fit into a classification framework that can be considered to represent the same process condition from the definite data sets of a large number of accumulated process events from the present to the past.

[0058] (4-1. When determining the classification frame under a single process gas condition) When determining a classification framework that can be regarded as the same process conditions, it is preferable to use a classification framework that is strongly related to product deposition. The etching gas is usually a heavy gas, and it can be said that the correlation with product deposition is stronger when the gas flow rate is higher, that is, when the reference gas equivalent flow rate Qin_a is larger. Therefore, it is preferable to conduct the evaluation under conditions where the reference gas equivalent flow rate Qin_a is large. However, since the process conditions vary for each end user, if only the dataset with a large reference gas equivalent flow rate Qin_a is targeted, the number of data collected in the classification may extremely decrease. Therefore, it is preferable to use a plurality of evaluation target classes with a relatively large reference gas equivalent flow rate Qin_a as the classification framework.

[0059] For example, let one framework regarding the reference gas equivalent flow rate Qin_a be Qin_a_1, and generate its frame range with ±ratio×Qin_a_1. In the case of the example shown in Figure 3, among the reference gas equivalent flow rates Qin_a obtained in each process event, for example, if Qin_a_C regarding process C is within the frame range and Qin_a_A, Qin_a_B regarding processes A and B are outside the frame range, only the dataset of process C obtained in the equilibrium state will be classified and extracted.

[0060] When using the gas species relative value a and the gas flow rate Qin instead of the reference gas equivalent flow rate Qin_a, the frame range of (a, Qin) may be set, for example, as ±ratio×a, ±ratio×Qin. ratio is a value of 0 < ratio < 1, and usually about 0.1 to 0.2. The group of determined datasets classified into one framework are arranged in time series respectively. If the data group of the motor current value Im arranged in time series generally increases monotonically and exceeds a predetermined threshold value, it can be determined as the maintenance time period.

[0061] If, despite the classification framework being appropriate, there is a tendency for the data to not increase monotonically but to suddenly drop and then increase again, it can be evaluated and analyzed to determine that there is a high possibility that peripheral maintenance, such as replacement of the vacuum pump 3, has been carried out. In addition to directly evaluating the extracted data group as described above, it is also possible to generate an optimal trend line (or curve) from regression analysis, as in the prior art, and determine whether maintenance is necessary based on whether a predetermined threshold has been reached.

[0062] In a turbomolecular pump, the pump rotor 32, rotating at a speed close to the speed of sound, generates heat as it collides with gas molecules. This heat is then dissipated by thermal radiation to the opposing stator. As the amount of product deposition increases, the heat dissipation characteristics deteriorate due to the adhesion of product to the pump rotor 32 and stator surfaces, resulting in an increase in rotor temperature. In other words, rotor temperature can also be considered a state characteristic value that is strongly correlated with the amount of product deposition. Therefore, if the vacuum pump 3 is equipped with a rotor temperature sensor that measures the temperature of the pump rotor 32, rotor temperature can be used instead of the motor current Im as a state characteristic value for evaluating the amount of product deposition. In this case, too, by acquiring a rotor temperature data set and classifying it in the same way, it is possible to evaluate the temperature increase trend and determine the state by comparing it with a threshold value. Furthermore, by evaluating and judging using both the motor current Im and rotor temperature as state characteristic values, the accuracy of the product deposition evaluation can be further improved.

[0063] (Guideline for classification of gas type relative value a and gas flow rate Qin) The following describes the classification range of the gas species relative value a and the gas flow rate Qin. For example, in an etching process, the etching gas is introduced using Ar gas (M = 40) as the dilution gas. For etching gases such as SF6 (molecular weight M = 142) and C4F8 (molecular weight M = 200), with Ar gas as the reference gas, the gas species relative value a is calculated as a = 0.53 and a = 0.45, respectively, according to equation (3). Since the molecular weight of the etching gas is typically greater than that of Ar gas, the classification range of the gas species relative value a can be at least a < 1. However, when considering dissociation of gas molecules due to plasma excitation, it is preferable to set the value of the gas species relative value a in a wider range. For example, when using a fluorine (F)-based gas, the value can be set to a < 1.5 (= sqrt(40 / 19), where sqrt() is the square root of the number in parentheses). As for the gas flow rate Qin, the classification range is roughly several hundred sccm for a process chamber equipped with a 3000 L / s class turbomolecular pump.

[0064] (Impact of product deposition and classification framework) Let's consider the gas species relative value a or reference gas equivalent flow rate Qin_a, which is affected by the gas species. As mentioned above, in a typical process, equilibrium is reached in the low valve opening θ region of the vacuum valve 2, and the effective pumping speed at this point can be considered approximately the valve conductance value. For example, if the pump pumping speed Sp = 2000 L / s and the valve conductance Cv = 200 L / s, the effective pumping speed Se is 182 L / s (= 1 / (1 / Sp + 1 / Cv)), which is approximately 10% slower than when conductance alone is used. This range is considered the low opening region. Therefore, if we explicitly limit the region to the low opening region, we can add the valve opening θ to the event condition (dataset) and also add a classification frame that specifies whether the opening value is below a certain low opening value (e.g., 10% or less).

[0065] As the amount of product deposition increases, the gas flow inside the vacuum pump becomes clogged, and the motor current increases to maintain the rated rotation, but pumping performance still declines, causing the pumping speed Sp to tend to decrease. For example, if the pumping speed Sp decreases by about 10% to Sp = 1800 L / s, the effective pumping speed Se is 180 L / s, which, compared to the above 182 L / s, shows that this level has almost no effect on the effective pumping speed Se. Here, we will assume that product deposition has increased extremely, and consider what to do if the decrease becomes even greater.

[0066] The gas species relative value a is calculated as follows: gas species relative value a = Se(θ) / Se_reference(θ). Even if the effective pumping speed decreases over time, the stored correlation data Se_reference(θ) value for the reference gas remains fixed, so the estimated gas species relative value a also decreases as the effective pumping speed Se(θ) decreases. For example, if the reference gas is Ar, a = 1 is estimated when Ar gas is introduced in the initial state where there is no product deposition. However, if the effective pumping speed changes over time due to product deposition, the gas species relative value a is estimated to be a value smaller than 1. Furthermore, due to the relationship Qin_a = Qin / a, the estimated Qin_a value will increase as the effective pumping speed changes over time.

[0067] The classification frame setting (frame range ±ratio × a) must also be taken into consideration to account for changes in the gas type relative value a that accompany such changes in the effective pumping speed over time. In this case, the frame range for the gas type relative value a is set so that the negative deviation is greater than the positive deviation. Note that the frame range for the gas flow rate Qin can have the same negative and positive deviations. Furthermore, it is more appropriate to make the positive deviation greater than the negative deviation for the reference gas equivalent flow rate Qin_a.

[0068] (4-2. When determining the classification frame based on multiple process gas conditions) While the above describes a case in which classification frames are determined solely based on the simplest process gas conditions, such as gas type and gas flow rate, multiple process gas conditions may be combined. For example, classification frames may be established for each gas type, gas flow rate, process pressure, and event duration, and only data sets that satisfy all of the classification frames may be extracted as data sets with the same process conditions. Compared to a single condition, for example, if the same process pressure and event duration are used as conditions, the probability of identifying process events as belonging to the production of approximately the same product increases in the case of manufacturing equipment used for high-mix, low-volume production. As a result, the possibility of identifying unknown process environmental conditions that cannot be estimated increases, enabling more accurate evaluation and judgment.

[0069] (4-3. When class classification frames are determined by combining multiple event conditions) Furthermore, classification frames can be determined not only by the process conditions within a single event, but also by combining process events with different process conditions for each process event and determining the classification frame based on that combination. Typically, a single wafer undergoes multiple etching processes, with the exposure / development process, etching process, and film deposition process being performed in a cycle. Furthermore, even within a specific etching process, the etching process is typically performed under multiple different process conditions, rather than a single set of process conditions. For example, the process conditions can be changed as the holes or grooves to be machined become deeper, optimizing the process. Therefore, by identifying the combination of process conditions, it is possible to narrow down the data to a specific etching process.

[0070] That is, a data set block consisting of a plurality of data sets having a combination pattern of process conditions of interest is extracted and classified, and then only the target data sets for evaluation and judgment are extracted and classified from the data set block. Although the number of data sets stored and saved is reduced, they are carefully selected data sets, which further improves the accuracy of evaluation and judgment.

[0071] [Variation 1] In the above-described embodiment, the estimation device 6 is configured to generate a process information data set, generate a deposition information data set by adding the motor current value Im input from the vacuum pump 3 to the process information data set, extract a deposition information data set to be used for determining the deposition amount, and determine the deposition amount based on the extracted deposition information data set. However, for example, it is possible to determine the deposition amount by transmitting a definite data set for each process event (a data set including only process gas conditions) that does not include the motor current value Im from the estimation device 6 to the vacuum pump 3 or the vacuum valve 2, adding the motor current value Im and the measurement time Time_msr to the definite data set in the vacuum pump 3 or the vacuum valve 2, and classifying and extracting the data set and arranging it in chronological order.

[0072] As described above, the data set is finalized at the end of the event, so the vacuum pump 3 or vacuum valve 2 can add the motor current value Im and the measurement time Time_msr to the finalized data set by, for example, the following process. The estimation device 6 outputs a trigger signal to the vacuum pump 3 or vacuum valve 2 each time a data set with flag_OK=1 is estimated. The vacuum pump 3 or vacuum valve 2 updates and stores the motor current value Im and the measurement time each time a trigger signal is input. Once the end of the event is finalized and the finalized data set is input from the estimation device 6 to the vacuum pump 3 or vacuum valve 2, the vacuum pump 3 or vacuum valve 2 adds the stored motor current value Im and measurement time to the finalized data and stores it.

[0073] [Variation 2] 1, the estimating device 6 is connected in parallel to the vacuum pump 3 and vacuum valve 2 via communication lines and is configured to acquire information from the vacuum pump 3 and vacuum valve 2, but the vacuum pump 3, vacuum valve 2, and estimating device 6 may also be daisy-chained in order using RS485 communication or the like. The motor current value Im of the vacuum pump 3 is sent to the estimating device 6 via the vacuum valve 2.

[0074] [Variation 3] In the above-described embodiment, the estimation device 6 is provided independently, but as shown in Figure 8(a), the estimation function may be included in the vacuum valve 2. In this case, the vacuum valve 2 may evaluate the amount of product deposition in the vacuum pump 3 and transmit the determination result to the main control device 7 as shown in Figure 8(a), or may transmit the result to the vacuum pump 3 so that an alarm and warning are issued at the vacuum pump 3. Alternatively, the vacuum valve 2 may generate only a process information dataset excluding the state characteristic value to be evaluated and transmit it to the vacuum pump 3, and the vacuum pump 3 may generate a deposition information dataset from the received process information dataset and the motor current value Im and evaluate the amount of product deposition.

[0075] 8(b), the estimation function may be included in the vacuum pump 3. In this case, the chamber pressure Pc, target pressure Ps, valve element opening θ, and correlation data Se_reference(θ) are input to the vacuum pump 3 from the vacuum valve 2, and the vacuum pump 3 evaluates the amount of product deposition and sends the judgment result to the main control device 7. Alternatively, the estimation function may be included in the main control device 7 as shown in FIG. 8(c). The valve element opening θ, chamber pressure Pc, motor current value Im, and correlation data Se_reference(θ) are input to the main control device 7 from the vacuum valve 2.

[0076] Instead of inputting the correlation data Se_reference(θ) from the vacuum valve 2 to the main controller 7, the main controller 7 may set various values ​​for the valve element opening θ of the vacuum valve 2, measure the chamber pressure value Pc, and directly calculate and store the correlation data Se_reference(θ). Alternatively, the numerical value of the correlation data Se_reference(θ) described in the product specifications of the vacuum valve 2 may be directly input to the main controller 7 and stored therein. Furthermore, apart from Figures 8(a) to (c), all signals may be input once to the main controller 7, and then the signals may be output again from the main controller 7 to the vacuum valve 2 or the vacuum pump 3.

[0077] -Second embodiment- In the first embodiment described above, the case where the amount of product deposition in the vacuum pump 3 is evaluated based on the motor current value Im, which is a state characteristic value of the vacuum pump 3. In evaluating the amount of product deposition, the motor current value Im is monitored under the same process gas conditions, and it is determined whether the motor current value Im exceeds a predetermined threshold. It has also been explained that, in addition to the motor current value Im, rotor temperature can be used as a state characteristic value that has a strong correlation with the amount of product deposition. With regard to the maintenance of the vacuum pump 3, in addition to the amount of product deposition, there is creep strain of the pump rotor 32 of the vacuum pump 3. In the second embodiment, the application of this to the evaluation of creep strain will be described.

[0078] In a vacuum pump 3 in which the pump rotor 32 rotates at high speed, monitoring the amount of creep strain of the pump rotor 32 is important for safe operation of the vacuum pump 3. The amount of creep strain of the pump rotor 32 depends on the rotor temperature and stress due to centrifugal force, which in turn depends on the rotor rotation speed. Consider a rotor rotation speed within a predetermined rotation speed range (e.g., a predetermined rotation speed range including the rated rotation speed) where the progression of creep strain becomes a problem. The amount of creep strain can be said to be proportional to the time integral of the rotor temperature over a time range within the predetermined rotation speed range. When the time integral of the rotor temperature reaches a value equivalent to the allowable value for creep strain, the creep life can be determined (see, for example, JP 2018-3615 A).

[0079] (1. Rotor speed as a state characteristic value) As described above, when evaluating creep strain using the time integral of rotor temperature, it is necessary to calculate the time integral of rotor temperature only for the time when the rotor rotational speed, at which stress is high, is within the above-mentioned predetermined rotational speed range. Therefore, to evaluate creep strain using the time integral of rotor temperature, rotor temperature and rotor rotational speed are required as state characteristic values ​​of the vacuum pump 3. While turbomolecular pumps are typically equipped with a rotational speed sensor to detect rotor rotational speed, many do not have a rotor temperature sensor. Therefore, here, the reference gas equivalent flow rate Qin_a is used instead of rotor temperature. In other words, creep strain can be evaluated even without a rotor temperature sensor.

[0080] The reference gas equivalent flow rate Qin_a, one of the process gas conditions mentioned above, correlates closely with rotor temperature because the rotor temperature increases as the reference gas equivalent flow rate Qin_a increases. Similarly to the time integral of the rotor temperature, the rotor creep strain increases as the time integral of the reference gas equivalent flow rate Qin_a increases. Therefore, by inputting the rotor rotation speed into the estimation device 6 as a state characteristic value of the vacuum pump 3 and using the reference gas equivalent flow rate Qin_a estimated by the estimation device 6, it is possible to determine a time period in which the rotor rotation speed is within a predetermined range based on the magnitude of the reference gas equivalent flow rate Qin_a. Furthermore, during a time period in which the rotor rotation speed is within the predetermined range, the creep strain can be evaluated by calculating the time integral of the reference gas equivalent flow rate Qin_a instead of the time integral of the rotor temperature.

[0081] The state characteristic value is the rotor rotation speed, and the process gas condition for determining the recommended rotor rotation speed (the above-mentioned predetermined rotation speed range) is the reference gas equivalent flow rate Qin_a. When the value of the reference gas equivalent flow rate Qin_a is within a predetermined range, the rotor rotation speed is determined to be within the predetermined rotation speed range. The estimation device 6 calculates the time integral of the reference gas equivalent flow rate Qin_a during the period determined to be within the predetermined rotation speed range, and when the cumulative time integral of the reference gas equivalent flow rate Qin_a reaches the creep strain life determination value, it is determined to be creep strain life and issues an alarm and warning.

[0082] (2. Creep strain evaluation using process gas conditions only) When the rotor speed falls outside the predetermined range, including the rated speed, this occurs when the pump is stopped or when the pump is accelerating or decelerating during startup or shutdown. In either case, the value of the reference gas equivalent flow rate Qin_a can be considered zero. Therefore, if the time integral of the reference gas equivalent flow rate Qin_a is used instead of the time integral of the rotor temperature Tr and time integration is performed during a process in which the reference gas equivalent flow rate Qin_a is a large value, the time integral when the rotor speed falls outside the predetermined range, including the rated speed, is automatically excluded. Here, the measurement recommendation timing information flag_OK, which is an indicator of whether or not the process is in an equilibrium state, is used to determine whether or not the process is in an equilibrium state during a process in which the reference gas equivalent flow rate Qin_a is a large value.

[0083] As described above, a dataset whose measurement recommendation timing information flag_OK is set to 0 is considered an invalid dataset and is not included in the confirmed dataset for each process event. However, it is considered that the number of invalid datasets is small compared to the number of valid datasets among the process events in which the process is executed. For example, looking at process event B from ts1 to ts8 in Figure 7, the datasets at measurement timings ts1 to ts5 and ts8 where flag_OK=0 are invalid datasets, while the multiple datasets from measurement timings ts6 to ts7 where flag_OK=1 are valid datasets.

[0084] For process event B, the time integral of the reference gas equivalent flow rate Qin_a when using the valid data set is (Qin_a × Δts) × (Time_spn / Δts-2), which can be said to be approximately equal to (Qin_a × Δts) × Time_spn / Δts = Qin_a × Time_spn. Therefore, the time integral of the reference gas equivalent flow rate Qin_a for all process events = Σ(Qin_a × Time_spn) can be said to have a high correlation with the creep strain amount, and the time integral of the reference gas equivalent flow rate Qin_a can be used to evaluate and determine the creep strain amount. The estimation device 6 issues an alarm and warning when the time integral of the reference gas equivalent flow rate Qin_a reaches the time integral amount corresponding to the creep life.

[0085] While the above discussion focused on the case where the reference gas equivalent flow rate Qin_a is obtained, if the gas species relative value a is obtained based on Equation (8A), it is possible to narrow down whether the heavy gases that have a significant effect on creep strain increase are heavier than the reference gas Ar. In this case, accuracy can be further improved by accumulating the time integral of the reference gas equivalent flow rate Qin_a only for cases classified and extracted in a classification frame heavier than Ar gas, rather than just for cases where the reference gas equivalent flow rate Qin_a is large. On the other hand, when considering dissociation of gas molecules due to plasma excitation, it is preferable to narrow down the classification frame to include gases slightly lighter than Ar gas.

[0086] In the first and second embodiments, the target of preventive maintenance has been described as a TMP, but the product deposition can be similarly applied to a dry pump. Furthermore, products also deposit on the vacuum valve 2 and vacuum gauge 5, and the risk increases as the reference gas equivalent flow rate Qin_a increases and the time integral of the reference gas equivalent flow rate Qin_a increases, which is beneficial for evaluation and judgment.

[0087] -Third embodiment- In the first embodiment described above, the amount of product deposition in the vacuum pump 3 is evaluated by monitoring the motor current value Im when the process gas conditions are considered to be the same and determining whether the motor current value Im exceeds a predetermined threshold. When the motor current value Im is monitored under the condition that the process gas conditions are considered to be the same, the number of samples of the acquired motor current value Im may be too small, resulting in a decrease in the accuracy of estimating the amount of product deposition. Furthermore, the process gas conditions for acquiring the motor current value Im must be set for each user, which can be a significant setup burden.

[0088] For the above reasons, we have been studying ways to improve the accuracy of estimating the product deposition amount by relaxing the process gas conditions for obtaining the motor current value Im and increasing the number of samples of the motor current value Im. During this study, the inventors of the present invention discovered that the motor current value Im remains the same even if other process gas conditions are different, as long as the reference gas equivalent flow rate Qin_a remains the same. In other words, we discovered that the motor current value Im can be expressed as a function of only the reference gas equivalent flow rate Qin_a, as shown in Figure 9. Hereinafter, this function will be referred to as the current value function F(Qin_a). Furthermore, data fitting techniques such as the least squares method are used to calculate functions that represent the characteristics of a data set. When calculating a function using this technique, the more data points used in the data fitting technique and the greater the spread in values ​​of the data points, the more accurate the function can be calculated. Figure 9 is a graph of an example of a current value function.

[0089] Based on the above findings, in the third embodiment, as many deposition information data sets as possible are obtained in a time series, each including a reference gas equivalent flow rate Qin_a and a motor current value Im, and a current value function F(Qin_a) is calculated using these many deposition information data sets, and the amount of product deposition is estimated based on the change over time in the motor current value (called the specific current value Im0) calculated from the current value function F(Qin_a).

[0090] A method for estimating the product deposition amount according to the third embodiment will be described below with reference to FIG. 10. FIG. 10 is a flowchart showing the method for estimating the product deposition amount according to the third embodiment. First, the data set generation unit 61A of the estimation device 6 generates a plurality of data sets [Qin_a, Im, Time_msr] in chronological order for a predetermined period as deposition information data sets used to calculate the current value function F(Qin_a), and stores the data sets in the storage unit 62 (step S1). Whether the predetermined period for continuing data set generation has arrived can be determined, for example, from the difference between the current measurement time Time_msr and the measurement time Time_msr when the data set was first generated.

[0091] The predetermined period for continuing generation of deposition information data sets used to calculate the current value function F(Qin_a) is a period during which a large number of deposition information data sets can be generated to the extent that an accurate current value function F(Qin_a) can be calculated, and the influence of product deposition on the motor current value Im is negligible. Specifically, the predetermined period for continuing generation of deposition information data sets is, for example, one week. This period can be changed as appropriate depending on, for example, the type of gas used in the vacuum chamber 1, the flow rate of the gas, etc.

[0092] The deposition information data set is generated in step S1 when the inside of the vacuum chamber 1 is in an equilibrium state (ie, when the chamber pressure Pc is constant).

[0093] After the generation of the deposition information data set continues for a predetermined period, the deposition amount determination unit 61B of the estimation device 6 calculates a current value function F(Qin_a) that expresses the motor current value Im as a function of the reference gas equivalent flow rate using the reference gas equivalent flow rate Qin_a and the motor current value Im included in the deposition information data set generated in step S1 (step S2). Specifically, the current value function F(Qin_a) is calculated as a higher-order function of the reference gas equivalent flow rate Qin_a (for example, a1+a2*Qin_a+a3*(Qin_a) 2+a4*(Qin_a) 3 +···, a1, a2, a3, a4: constants), and this higher-order function is fitted to the data set [Qin_a,Im] using a data fitting method such as the least squares method to calculate the specific values ​​of the constants a1, a2, a3, and a4.

[0094] 9, when the reference gas equivalent flow rate Qin_a is a relatively large value (for example, when the reference gas equivalent flow rate Qin_a is 100 sccm or more), the motor current value Im increases linearly with an increase in the reference gas equivalent flow rate Qin_a. In other words, when the reference gas equivalent flow rate Qin_a is a relatively large value, the current value function F(Qin_a) can be approximated by a linear equation of a1' + a2' * Qin_a (a1', a2': constants). Therefore, in step S2, from the data sets generated in step S1, data sets in which the reference gas equivalent flow rate Qin_a is equal to or greater than a predetermined value (a range in which the current value function F(Qin_a) can be linearly approximated) may be extracted, and the extracted data sets may be data-fitted to the linearly approximated current value function F(Qin_a) = a1' + a2' * Qin_a to calculate specific values ​​of the constants a1' and a2'. This allows the current value function F(Qin_a) that represents the relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im to be calculated accurately and quickly.

[0095] The current value function F(Qin_a) calculated in step S2 corresponds to a function that represents the relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im at the midpoint of the predetermined period during which the deposition information data set was generated in step S1. For example, if the predetermined period during which the deposition information data set was generated is the period from time T1 to time T2, the current value function F(Qin_a) corresponds to a function that represents the relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im at time (T1+T2) / 2.

[0096] After calculating the current value function F(Qin_a), the accumulation amount determination unit 61B calculates the specific current value Im0 using the current value function F(Qin_a) calculated in step S2 (step S3). Specifically, the specific current value Im0 is calculated by substituting a specific value (specific value Qin_a0) of a predetermined reference gas equivalent flow rate Qin_a into the calculated current value function F(Qin_a).

[0097] In this way, the specific current value Im0 corresponds to the motor current value Im when the reference gas equivalent flow rate Qin_a is the specific value Qin_a0. Furthermore, as described above, the current value function F(Qin_a) corresponds to a function that represents the relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im at the midpoint of the predetermined period during which the deposition information data set was generated. Therefore, the specific current value Im0 corresponds to the motor current value Im when the reference gas equivalent flow rate Qin_a is the specific value Qin_a0 at the midpoint of the predetermined period during which the deposition information data set was generated. Specifically, for example, if the predetermined period during which the deposition information data set was generated is the period from time T1 to time T2, the specific current value Im0 corresponds to the motor current value Im when the reference gas equivalent flow rate Qin_a is the specific value Qin_a0 at time (T1+T2) / 2.

[0098] After calculating the specific current value Im0, the accumulation amount determination unit 61B determines whether the specific current value Im0 calculated in step S3 is equal to or greater than a predetermined threshold value (step S4). This predetermined threshold value can be set to an appropriate value as needed. If the specific current value Im0 is equal to or greater than the predetermined threshold value ("Yes" in step S4), the accumulation amount determination unit 61B determines that the accumulation amount of products is excessive (step S5). When it is determined that the accumulation amount of products is excessive, the accumulation amount determination unit 61B may, for example, output an alarm to the user that the accumulation amount of products is excessive. The accumulation amount determination unit 61B can notify the user that the accumulation amount of products is excessive by, for example, displaying a message on the display unit of the estimation device 6 that the accumulation amount of products is excessive, or by emitting a sound when the accumulation amount of products is excessive.

[0099] On the other hand, if the specific current value Im0 is less than the predetermined threshold value ("No" in step S4), the accumulation amount determination unit 61B determines that the amount of accumulation of products is small and appropriate (step S6).

[0100] After determining whether the amount of product deposition is excessive, it is determined whether to terminate the estimation of the amount of product deposition based on the specific current value Im0 (step S7). If the estimation of the amount of product deposition is to continue ("No" in step S7), steps S1 to S6 are executed again to calculate the specific current value Im0 for the next "predetermined period" (for example, the period from time T2 to time T3), and based on the specific current value Im0, it is determined whether the amount of product deposition is excessive in the next "predetermined period".

[0101] On the other hand, if the estimation of the product deposition amount is not to be continued ("Yes" in step S7), the estimation of the product deposition amount is ended.

[0102] By repeatedly executing the above steps S1 to S6, for example, it is possible to calculate a current value function F(Qin_a) in time series as shown in Fig. 11, and to calculate a specific current value Im0 in time series as shown in Fig. 12. Specifically, by repeatedly executing the above steps S1 to S6 n times, for example, from time T1 to time Tn, it is possible to calculate a current value function F(Qin_a) at time (T1+T2) / 2, a current value function F(Qin_a) at time (T2+T3) / 2, and a current value function F(Qin_a) at time (Tn+Tn+1) / 2 in time series as shown in Fig. 11. Furthermore, as shown in Fig. 12, a specific current value Im0(1) at time (T1+T2) / 2, a specific current value Im0(2) at time (T2+T3) / 2, and a specific current value Im0(n) at time (Tn+Tn+1) / 2 are calculated in time series. Fig. 11 is a diagram showing an example of a current value function calculated in time series. Fig. 12 is a diagram showing an example of specific current values ​​calculated in time series.

[0103] As described above, the current value function F(Qin_a) is calculated using a large number of data sets [Qin_a, Im] generated within a predetermined period, and therefore can accurately represent the relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im within the predetermined period. Therefore, the current value function F(Qin_a) calculated in time series can accurately represent the time-series changes in the motor current value Im that change in accordance with the accumulation of by-products.

[0104] Furthermore, the specific current value Im0 is calculated by substituting a specific value Qin_a0 of a predetermined reference gas equivalent flow rate Qin_a into the current value function F(Qin_a), so that it accurately reflects the time-series change in the motor current value Im due to the accumulation of by-products when the reference gas equivalent flow rate Qin_a is at the specific value Qin_a0. The amount of accumulation of by-products can be accurately estimated by estimating whether or not such specific current value Im0 is equal to or greater than a predetermined threshold.

[0105] [Variation 1] The relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im can be expressed as shown in FIG. 9 when the gas flow is in the molecular flow region. When the gas flow deviates from the molecular flow region, the relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im deviates from the relationship shown in FIG. 9. Therefore, a condition that the gas flow is in the molecular flow region may be set as a condition for generating a data set for calculating the current value function F(Qin_a). Specifically, the above data set may be generated when the valve element opening θ is less than a predetermined value (e.g., 10%).

[0106] [Variation 2] The relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im can be expressed as shown in Figure 9, the closer the molecular weight of the gas to be flowed in this system is to the molecular weight of the reference gas. If the molecular weight of the gas to be flowed deviates from the molecular weight of the reference gas, the relationship between the reference gas equivalent flow rate Qin_a and the motor current value Im deviates from the relationship shown in Figure 9. Therefore, as a condition for generating a data set for calculating the current value function F(Qin_a), a data set for calculating the current value function F(Qin_a) may be generated when the gas species relative value a is within a predetermined range (for example, 1.5 or less).

[0107] [Variation 3] The current value function F(Qin_a) can accurately represent the relationship between the reference-gas-equivalent flow rate Qin_a and the motor current value Im for values ​​of the reference-gas-equivalent flow rate Qin_a between the minimum and maximum values ​​of the reference-gas-equivalent flow rate Qin_a included in the generated data set. On the other hand, for values ​​of the reference-gas-equivalent flow rate Qin_a that are not between the minimum and maximum values ​​of the reference-gas-equivalent flow rate Qin_a included in the data set, the current value function F(Qin_a) may not accurately represent the relationship between the reference-gas-equivalent flow rate Qin_a and the motor current value Im. Therefore, when the specific value Qin_a0 of the reference-gas-equivalent flow rate Qin_a used to calculate the specific current value Im0 is not between the minimum and maximum values ​​of the reference-gas-equivalent flow rate Qin_a included in the data set, the specific current value Im0 may not be calculated.

[0108] In addition, when the current value function F(Qin_a) is calculated by linear approximation, the specific current value Im0 may be calculated even if the specific value Qin_a0 of the reference gas equivalent flow rate Qin_a is not between the minimum and maximum values ​​of the reference gas equivalent flow rate Qin_a included in the data set.

[0109] [Variation 4] After a certain number of specific current values ​​Im0 have been calculated in a time series, a function representing the relationship between the multiple specific current values ​​Im0 in the calculated time series and time T can be calculated using a data fitting method such as the least squares method, and this function can be used to estimate the time at which the specific current value Im0 will be greater than or equal to a predetermined threshold.

[0110] [Variation 5] It is not necessary to perform the generation of the data set, the calculation of the current value function F(Qin_a), and the calculation of the specific current value Im0 every predetermined period. For example, the generation of data sets may be continued for a long period of time, and a large number of data sets may be stored in the storage unit 62. After that, the large number of data sets may be divided into data sets for each predetermined period, and the current value function F(Qin_a) and the specific current value Im0 may be calculated.

[0111] [Variation 6] As shown in the above equations (15) and (16), the reference gas equivalent flow rate Qin_a is calculated by multiplying the correlation data Se_reference(θ) by the chamber pressure Pc in the equilibrium state. Therefore, in the above embodiment, the correlation data Se_reference(θ) has been presented on the premise that various opening values ​​and corresponding effective exhaust speed values ​​are stored as numerical data. On the other hand, if the correlation data Se_reference(θ) is also considered as a function of the valve disc opening θ, similar to the current value function F(Qin_a), the correlation data Se_reference(θ) can be calculated as a higher-order function of the valve disc opening θ (for example, b1+b2*θ+b3*θ). 2 +b4*θ 3 +···, b1, b2, b3, b4: known constants). In this case, only the constants (b1, b2, b3, b4) included in this higher-order function need to be stored as parameter data. Even if the correlation data Se_reference(θ) is defined as a higher-order function in this way, the current value function F(Qin_a) can be calculated in the same way as if it were calculated from the data set [Qin_a, Im]. Specifically, using the higher-order function of the correlation data Se_reference(θ), the current value function F(Qin_a) can be rewritten as a current value function F(θ, Pc), which is a function of the valve disc opening θ and the chamber pressure Pc. Then, by performing data fitting between this current value function F(θ, Pc) and the data set [θ, Pc, Im] using a data fitting method such as the least squares method, the current value function F(θ, Pc), which is a function of the valve disc opening θ and the chamber pressure Pc, can be specifically calculated.

[0112] [Variation 7] The current value function F(Qin_a) changes over time as product deposition progresses. However, the initial current value function (hereinafter referred to as the initial function F) when there is no product deposition or when the amount of product deposition is so small that the effect of product deposition can be ignored is ini (Qin_a) can be defined in advance as a known high-order function. That is, the initial function F ini (Qin_a) is a known high-order function of the reference gas equivalent flow rate Qin_a (e.g., a1''+a2''*Qin_a+a3''*(Qin_a) 2 +a4''*(Qin_a) 3 +···, a1'', a2'', a3'', a4'': known constants).

[0113] [Variation 8] In the above-described modification 6, the correlation data Se_reference(θ) is defined as a known high-order function, and the current value function F(θ, Pc), which is a function of the valve disc opening θ and the chamber pressure Pc, is calculated using this high-order function. Conversely, the correlation data Se_reference(θ) can be calculated by defining an unknown high-order function (e.g., b1'+b2'*θ+b3'*θ) 2 +b4'*θ 3 +···, b1', b2', b3', b4': unknown constants), and the unknown constants b1', b2', b3', b4' are defined as the initial function F ini (Qin_a), and the calculated constants b1', b2', b3', and b4' may be used as the known constants b1, b2, b3, and b4 shown in Modification Example 6. As will be described later, the constants b1', b2', b3', and b4' are calculated using a data set at the initial time, but in calculating correlation data Se_standard(θ) using a higher-order function, the constants b1', b2', b3', and b4' calculated using a data set at the initial time are valid not only at the initial time but also at other times, and can be used as they are as the known constants b1, b2, b3, and b4.

[0114] Specifically, given a known initial function F ini(Qin_a) is calculated by using the correlation data Se_criterion (θ) defined as an unknown higher-order function, and the (unknown) initial function F, which is a function of the valve disc opening θ and the chamber pressure Pc. ini (θ,Pc). The initial function F ini (θ,Pc) contains unknown constants b1', b2', b3', and b4'. Also, the initial function F ini (θ, Pc) is valid at the initial time when there is no or negligible product deposition. Therefore, this initial function F ini (θ, Pc) and the data set [θ, Pc, Im] at the initial time (for example, between times T1 and T2) are fitted using a data fitting method such as the least squares method to obtain the initial function F ini By calculating (θ, Pc), the unknown constants b1', b2', b3', and b4' can be calculated as specific values. The constants b1', b2', b3', and b4' calculated in this manner are then stored as the known constants b1, b2, b3, and b4 shown in Modification 6, and can be used when calculating the correlation data Se_standard(θ) thereafter.

[0115] It will be understood by those skilled in the art that the exemplary embodiments and variations described above are examples of the following aspects.

[0116] [1] In one embodiment, an estimation device estimates process gas conditions of a system in which a vacuum chamber into which gas is introduced and a process is performed is evacuated by a vacuum pump attached via a vacuum valve, and estimates a first process gas condition including information on the gas type and gas flow rate of the introduced gas based on correlation data between the valve body opening of the vacuum valve for a predetermined gas type and the effective exhaust speed of the system, and the chamber pressure of the vacuum chamber.

[0117] The estimation device 6 estimates a reference gas equivalent flow rate Qin_a, which includes information on the gas species relative value a of the introduced gas and the gas flow rate Qin, based on correlation data Se_reference(θ) between the valve disc opening θ and the effective pumping speed and the chamber pressure Pc. By classifying and extracting the motor current value using this estimated reference gas equivalent flow rate Qin_a as a process gas condition, it is possible to determine the change over time in the motor current value that depends on product deposition more accurately than before, and to more accurately determine the amount of product deposition.

[0118] [2] The estimation apparatus according to [1] above further comprises a state determination unit that determines whether the chamber pressure is in a state that can be regarded as an equilibrium state where it has converged to a process target pressure, and estimates the first process gas condition in the state that can be regarded as an equilibrium state. By determining that the chamber pressure Pc is in an equilibrium state in the estimation unit 61 that functions as the state determination unit, the reference gas equivalent flow rate Qin_a in the equilibrium state can be easily estimated from the target pressure Ps and the correlation data Se_reference(θ) at that time.

[0119] [3] In the estimation device described in [2] above, in addition to determining the state that can be considered as an equilibrium state, the state determination unit determines that the chamber pressure is in a transient state of pressure response when the chamber pressure is adjusted to the target pressure, estimates a plurality of second process gas conditions corresponding to the gas type and gas flow rate of the introduced gas in the transient state, and estimates a third process gas condition corresponding to the gas type and gas flow rate of the introduced gas in an equilibrium state where the chamber pressure has converged to the target pressure for the process based on the plurality of second process gas conditions and the first process gas condition in the state that can be considered as an equilibrium state. The transient state of the pressure response can be estimated from the change in the chamber pressure Pc after a change in Qin_a or a change in the valve disc opening θ. In this transient state, the third process gas conditions (gas species relative value ao, gas flow rate Qino) can be estimated from multiple second process gas conditions (gas species relative value a, gas flow rate Qin) and the reference gas equivalent flow rate Qin_a in an equilibrium state, thereby improving the estimation accuracy of the gas species relative value and gas flow rate.

[0120] [4] In the estimation device according to any one of the above items [1] to [3], an event time length corresponding to the required time of the process treatment and / or a process pressure are further estimated as a fourth process gas condition. Even if a plurality of process events are included, classification and extraction can be easily performed by using the event time length Time_spn corresponding to the time required for the process treatment and / or the process pressure Ps as the fourth process gas condition.

[0121] [5] In the estimation apparatus according to any one of the above items [1] to [4], information indicating whether the chamber pressure is in an equilibrium state is further estimated as a fifth process gas condition. The measurement recommendation timing information flag_OK is an index indicating whether the chamber pressure Pc is approximately in an equilibrium state, and by using the measurement recommendation timing information flag_OK as a process gas condition, the motor current value in the equilibrium state can be easily classified and extracted.

[0122] [6] An estimation device according to one embodiment estimates at least one of the first process gas condition [1] or [2], the second process gas condition [3], the third process gas condition [3], and the fourth process gas condition [4], and the fifth process gas condition [5], generates a plurality of data sets in time series, each including the estimated process gas conditions and a state characteristic value representing the operating state of a vacuum pump, extracts only the data sets in which the fifth process gas condition represents an equilibrium state from the plurality of data sets, and sets a classification frame in which the process gas condition can be considered to be approximately the same as a predetermined state for one predetermined process gas condition or predetermined plurality of process gas conditions among the plurality of process gas conditions included in the extracted data sets, and classifies the extracted data sets into the classification frame. By using at least one of the first to fourth process gas conditions and the fifth process gas condition (recommended timing information flag_OK), it is possible to accurately classify and extract state characteristic values ​​(motor current values) in an equilibrium state for the same process event.

[0123] [7] In the estimation device described in [6] above, the state characteristic value is a current value of a motor that rotates and drives a pump rotor, and multiple data sets classified into the same classification frame are arranged in chronological order, and the amount of product deposition in the vacuum pump is estimated based on the time-series changes in the state characteristic values ​​included in the data sets arranged in chronological order. Since the motor current values ​​in the equilibrium state for the same process event can be accurately classified and extracted, the amount of product deposition can be accurately evaluated.

[0124] [8] In the estimation device according to any one of the above items [1] to [5], the first process gas condition is a reference gas equivalent flow rate calculated based on the gas type of the introduced gas and the gas flow rate. The estimation device generates multiple data sets each including the reference gas equivalent flow rate and a current value of a motor that rotates and drives a pump rotor, calculates a current value function that expresses the relationship between the reference gas equivalent flow rate and the current value using the multiple data sets each including the reference gas equivalent flow rate and the current value, substitutes a predetermined specific value of the reference gas equivalent flow rate into the current value function to calculate a specific current value that is the current value at that specific value, and estimates the amount of product deposition in the vacuum pump based on the specific current value. The current value function is calculated using a large number of data sets generated within a predetermined period, and therefore can accurately represent the relationship between the reference gas equivalent flow rate and the motor current value within the predetermined period. Therefore, the current value function calculated in time series can accurately represent the time-series change in the motor current value that changes in accordance with the accumulation of by-products. Furthermore, the specific current value is calculated by substituting a predetermined specific value of the reference gas equivalent flow rate into the current value function, and accurately reflects the time-series change in the motor current value due to the accumulation of by-products when the reference gas equivalent flow rate is a specific value. The amount of by-product accumulation can be accurately estimated by estimating whether or not such a specific current value is equal to or greater than a predetermined threshold.

[0125] [9] In the estimation device described in [5] above, a plurality of data sets each including the first process gas condition, an event time length, and the fifth process gas condition are generated in a time series, only a data set in which the fifth process gas condition represents an equilibrium state is extracted from the plurality of data sets, a time integral of the first process gas condition is estimated based on the event time length and the fifth process gas condition included in the extracted data set, and creep strain of a pump rotor of a vacuum pump is evaluated based on the estimated time integral. Since creep strain is estimated based only on the process gas conditions, creep strain evaluation of a vacuum pump without a temperature sensor is easily possible.

[0126]

[10] In one embodiment, a vacuum valve is provided between a vacuum chamber in which a process is performed and a vacuum pump, and is equipped with the estimation device described in [7] or [8] above, and determines the product deposition state in the vacuum pump based on the product deposition amount estimated by the estimation device. By estimating the amount of product deposition using the estimation device [7] or [8] above, the amount of product deposition in the vacuum pump can be easily and accurately evaluated.

[0127]

[11] A vacuum valve according to one embodiment receives as input at least one of the first process gas condition [1] or [2], the second process gas condition [3], the third process gas condition [3], and the fourth process gas condition [4], the fifth process gas condition [5], and a state characteristic value representing the operating state of a vacuum pump, generates a plurality of data sets in chronological order, each data set including at least one of the first to fourth process gas conditions, the fifth process gas condition, and the state characteristic value representing the operating state, extracts only the data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets, and sets a classification frame in which the process gas condition can be considered to be approximately the same as a predetermined state for a predetermined one or a predetermined plurality of process gas conditions among the plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame. The process gas conditions are input from the estimation device 6, and the state characteristic values ​​(motor current values) are input from the vacuum pump 3 to the vacuum valve 2, and the vacuum valve 2 extracts and classifies the state characteristic values ​​based on the process gas conditions, thereby enabling more accurate extraction and classification of the state characteristic values.

[0128]

[12] A vacuum pump according to one embodiment includes a pump rotor that is driven to rotate by a motor and the estimation device described in [7] or [8] above, and determines the product deposition state within the vacuum pump based on the product deposition amount estimated by the estimation device.

[0129]

[13] A vacuum pump according to one embodiment receives as input at least one of the first process gas condition [1] or [2], the second process gas condition [3], the third process gas condition [3], and the fourth process gas condition [4], and a fifth process gas condition [5], generates a plurality of data sets in chronological order, each data set including at least one of the first to fourth process gas conditions, the fifth process gas condition, and a state characteristic value representing the operating state of the vacuum pump, extracts only the data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets, and sets a classification frame in which the process gas condition can be considered to be approximately the same as a predetermined state for a predetermined one or a predetermined plurality of process gas conditions among the plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame. Since the state characteristic values ​​are extracted and classified based on the process gas conditions input from the estimation device, the state characteristic values ​​can be extracted and classified with high accuracy.

[0130] Although various embodiments and modifications have been described above, the present invention is not limited to these. Other embodiments that are conceivable within the scope of the technical idea of ​​the present invention are also included within the scope of the present invention. [Explanation of symbols]

[0131] 1... vacuum chamber, 2... vacuum valve, 3... vacuum pump, 4... flow rate controller, 5... vacuum gauge, 6... estimating device, 7... main control device, 21... valve body, 32... pump rotor, 61... estimating unit, 62... memory unit

Claims

1. 1. An estimation device for estimating process gas conditions in a system in which a vacuum chamber into which a gas is introduced and a process is performed is evacuated by a vacuum pump attached via a vacuum valve, the device comprising: estimating first process gas conditions including information on the gas type and gas flow rate of an introduced gas based on correlation data between the valve element opening of the vacuum valve and the effective pumping speed of the system for a predetermined gas type that is one known reference gas, the valve element opening of the vacuum valve, and the chamber pressure of the vacuum chamber; estimating an amount of product deposition in the vacuum pump or an amount of creep strain of a pump rotor based on the first process gas condition and a state characteristic value input from the vacuum pump; estimating the first process gas condition in a state that can be regarded as an equilibrium state in which the chamber pressure has converged to a process target pressure; When the target pressure in the equilibrium state is Ps and the correlation data in the equilibrium state is Se_reference(θ), the first process gas condition includes a relational expression of reference gas equivalent flow rate Qin=Se_reference(θ)×Ps. Estimation device.

2. An estimation device for estimating process gas conditions in a system in which a vacuum chamber into which a gas is introduced and a process is performed is evacuated by a vacuum pump attached via a vacuum valve, comprising: estimating first process gas conditions including information on the gas type and gas flow rate of an introduced gas based on correlation data between the valve element opening of the vacuum valve and the effective pumping speed of the system for a predetermined gas type, the valve element opening of the vacuum valve, and the chamber pressure of the vacuum chamber; a state determination unit that determines whether the chamber pressure is in a state that can be regarded as an equilibrium state where the chamber pressure has converged to a process target pressure; estimating the first process gas condition at the assumed equilibrium state; the state determination unit determines whether the chamber pressure is in a transient state of pressure response when the chamber pressure is adjusted to the target pressure, in addition to determining whether the chamber pressure is in a state that can be considered to be an equilibrium state; estimating a plurality of second process gas conditions corresponding to the gas type and gas flow rate of the introduced gas in the transient state; estimating a third process gas condition corresponding to a type and a flow rate of the introduced gas in an equilibrium state where the chamber pressure has converged to a target pressure for the process, based on the plurality of second process gas conditions and the first process gas condition in the state considered to be an equilibrium state; Estimation device.

3. 3. The estimation device according to claim 1, An estimation device estimates an event time length corresponding to a required time for the process treatment and / or a process pressure as a fourth process gas condition.

4. 4. The estimation device according to claim 1, The estimation device estimates information indicating whether the chamber pressure is in an equilibrium state as a fifth process gas condition.

5. estimating at least one of the first process gas condition set forth in claim 1 or 2, the third process gas condition set forth in claim 2, and the fourth process gas condition set forth in claim 3, and the fifth process gas condition set forth in claim 4; generating a plurality of data sets in time series, each data set including the estimated process gas conditions and a state characteristic value representing an operating state of the vacuum pump; extracting only a data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets; An estimation device that sets a classification frame in which a process gas condition can be considered to be substantially the same as a predetermined state for one predetermined process gas condition or predetermined plurality of process gas conditions among a plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame.

6. 6. The estimation device according to claim 5, the state characteristic value is a current value of a motor that rotates and drives a pump rotor, An estimation device that arranges multiple data sets classified into the same classification frame in time series and estimates the amount of product deposition in a vacuum pump based on the time-series changes in the state characteristic values ​​included in the data sets arranged in time series.

7. The estimation device according to any one of claims 1 to 4, the first process gas condition is a reference gas equivalent flow rate calculated based on the gas type of the introduced gas and the gas flow rate; generating a plurality of data sets each including the reference gas equivalent flow rate and a current value of a motor that rotates and drives a pump rotor; calculating a current value function representing a relationship between the reference gas equivalent flow rate and the current value using a plurality of data sets including the reference gas equivalent flow rate and the current value; Substituting a predetermined specific value of a flow rate equivalent to a reference gas into the current value function to calculate a specific current value, which is the current value at the specific value; An estimation device that estimates the amount of product deposition in the vacuum pump based on the specific current value.

8. 5. The estimation device according to claim 4, generating a plurality of data sets in time series, each data set including the first process gas condition, an event time length, and the fifth process gas condition; extracting only a data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets; estimating a time integral of the first process gas condition based on the event duration and the fifth process gas condition included in the extracted data set; An estimation device that estimates the amount of creep strain of a pump rotor of a vacuum pump based on an estimated time integral amount.

9. A vacuum valve provided between a vacuum chamber where a process is performed and a vacuum pump, the vacuum valve including the estimation device according to claim 6 or 7, A vacuum valve that determines a product accumulation state in a vacuum pump based on the product accumulation amount estimated by the estimation device.

10. 1. A vacuum valve into which at least one of a first process gas condition according to claim 1 or 2, a third process gas condition according to claim 2, and a fourth process gas condition according to claim 3, a fifth process gas condition according to claim 4, and a state characteristic value representing an operating state of a vacuum pump are input, generating a plurality of data sets in time series, each data set including at least one of the first to fourth process gas conditions, the fifth process gas condition, and a state characteristic value representing the operating state; extracting only a data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets; A vacuum valve that sets a classification frame in which a process gas condition can be considered to be approximately the same as a predetermined state for one predetermined process gas condition or predetermined plurality of process gas conditions among a plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame.

11. a pump rotor that is rotationally driven by a motor; and the estimation device according to claim 6 or 7, A vacuum pump that determines a product accumulation state in the vacuum pump based on the product accumulation amount estimated by the estimation device.

12. 1. A vacuum pump to which at least one of a first process gas condition according to claim 1 or 2, a third process gas condition according to claim 2, and a fourth process gas condition according to claim 3, and a fifth process gas condition according to claim 5, is input, generating a plurality of data sets in time series, each data set including at least one of the first to fourth process gas conditions, the fifth process gas condition, and a state characteristic value representing an operating state of a vacuum pump; extracting only a data set in which the fifth process gas condition represents an equilibrium state from the plurality of data sets; A vacuum pump that sets a classification frame in which a process gas condition can be considered to be substantially the same as a predetermined state for one predetermined process gas condition or predetermined plurality of process gas conditions among a plurality of process gas conditions included in the extracted data set, and classifies the extracted data set into the classification frame.

Images