JP4737233B2 - Temperature controller and heat treatment equipment - Google Patents

Description

The present invention relates to a temperature controller and a heat treatment apparatus using a suitable model structure to model such measurement target and the control target, and more particularly, a suitable model structure such as a control or predictive control of a controlled object with interference using a temperature controller and to a heat treatment apparatus.

  A control target having multiple points of input / output interference, that is, a control having a plurality of operation amounts input to the control target and a control amount from the control target, and mutual interference exists between the operation amount and the control amount As a known technique for controlling the object to be non-interfering, there is a non-interacting PID control shown in FIG.

The control target 30 in this example is a control target with 2ch interference of 2 inputs (u ₁ , u ₂ ) 2 outputs (z ₁ , z ₂ ), and P ₁₁ , P ₂₁ , P ₁₂ , P ₂₂ are The transfer functions C ₁₁ and C ₂₂ output manipulated variables u ₁ ′ and u ₂ ′ based on deviations between the controlled variables z ₁ and z ₂ from the controlled object 30 and the target values r ₁ and r ₂ , respectively. C ₁₂ (s) and C ₂₁ (s) are cross controllers for non-interference.

In this conventional example, the relationship of interference of control objects is considered as a matrix, and the cross controllers C ₁₂ (s) and C ₂₁ (s) for non-interference in the adjusting unit 31 are set so as to cancel the interference. It is what decides.

If the cross controllers C ₁₂ (s) and C ₂₁ (s) are designed so that the controlled variable z ₁ is not affected by the manipulated variable u ₂ ′ and the controlled variable z ₂ is not affected by the manipulated variable u ₁ ′. , Non-interference can be achieved. There is also a method using an inverse matrix as means for eliminating such influence.
Nobuhide Suda et al. “PID Control” Asakura Shoten (Edition of System Control Information Society) March 10, 2000, p62

However, the presupposed relationship of interference of the controlled object is not a simple low-order matrix relationship. Therefore, ideal non-interference cannot be realized with the above-described first-order model of the conventional example. For example, when conventional non-interacting PID control is applied to a control target with thermal interference, the result is as shown in FIGS. In these drawings, a thick solid line indicates ch1 corresponding to u ₁ and z ₁ , and a thin solid line indicates ch2 corresponding to u ₂ and z ₂ .

  FIG. 33 shows the characteristics of the controlled object that has been made non-interfering with the primary model. This is the temperature (control amount) when a stepped heater output (operation amount) is input to ch1 to be controlled in 1000 seconds. In a steady state, the non-interference can be realized sufficiently, but it is a problem transiently. Contrary to the increase in the temperature of ch1, the temperature of ch2 is decreasing.

  FIG. 34 shows a target value response when the controlled object is controlled by the CHR adjustment rule. This is a step response in which the target value is increased by 10 ° C. for ch 1 from the state of 190 ° C. Since transient non-interference cannot be realized, it can be seen that the temperature of ch2 is also greatly changed even though only the target value of ch1 is changed.

  The reason why such transient non-interference cannot be realized is that the relationship of interference of the controlled object is not a simple and unilateral relationship from the manipulated variable u to the controlled variable z shown in FIG.

  The movement of the heat amount due to the interference is caused by the temperature difference. When the temperature difference between the points of the plurality of points to be controlled is large, the heat transfer due to interference is large, and when the temperature difference between the points is small, the heat transfer due to interference is small. This is because such a relationship is not taken into account, so that the error of the assumed control target model is large, and there is a limit to the factors that can be canceled by the inverse matrix of the non-interacting control.

  For this reason, the conventional non-interacting control has rarely been practically used.

The present invention was made in view of such circumstances, and an object thereof is to provide a temperature controller and a heat treatment apparatus using a suitable model structure such as a non-interference control or predictive control.

  In order to achieve the above object, the present invention is configured as follows.

The temperature controller of the present invention is based on a temperature control means for outputting an operation amount for the control object based on a detected temperature from the control object that is a mass having a heat capacity, and on the basis of the operation amount from the temperature control means. A temperature regulator for performing a dead time compensation control, comprising: a controlled object model; and a controlled object model excluding the dead time. The two controlled object models include a plurality of model elements, each having at least one input and at least one output, and for each of the two model elements of the plurality of model elements, from the output of one model element to the other And a model structure that feeds back the difference on the output side obtained by subtracting the output of the model element to the input side of the model element with different signs. Here, the lump having the heat capacity refers to a hot plate, a wafer, a support structure, and the like.

  According to the present invention, the controlled object model used for the dead time compensation control includes a plurality of model elements and includes a model structure that feeds back the difference on the output side of the model elements to the input side. Compared to a model including the model error, the model error is reduced, and overshoot and the like are suppressed.

The temperature controller according to the present invention is a temperature controller including a temperature control unit that outputs an operation amount for the control object based on a detected temperature from the control object, and includes at least one of the detected temperature and the operation amount. Based on a workpiece model for estimating the temperature of a workpiece to be processed which is a mass having a heat capacity to be processed by the control object, and the control means is a workpiece estimated using the workpiece model The manipulated variable is output based on the estimated temperature of the control object and the detected temperature from the control object, and the workpiece model includes a plurality of model elements, and two model elements of the plurality of model elements A model structure that feeds back the difference on the output side obtained by subtracting the output of the other model element from the output of one model element to the input side of the model element with different signs A.

According to the present invention, the object to be processed is estimated by using the object model having a model structure with a small model error, and the object to be controlled is controlled.
Processing can be performed with a desired temperature of the workpiece to be processed by the control target.

  In one embodiment of the present invention, the temperature control means includes a proportional differential control unit to which a deviation between the target temperature to be controlled and the detected temperature is input, the target temperature and the estimated temperature of the object to be processed. And an integration control unit to which the deviation is input.

  According to the present invention, the object to be controlled can be controlled so that the estimated temperature of the object to be processed matches the target temperature by the integration operation of the integration control unit.

  The heat treatment apparatus of the present invention includes a heat treatment means for heat-treating an object to be processed, and a temperature controller of the present invention for controlling the temperature of the heat treatment means based on the detected temperature of the heat treatment means detected by the temperature detection means, And operating means for operating the heat treatment means based on the output of the temperature controller.

  Here, adding an operation means heating and / or cooling the heat treatment means.

  According to the present invention, it is possible to provide a heat treatment apparatus that is temperature-controlled with high accuracy by the temperature controller of the present invention.

  As described above, according to the present invention, the model structure in which the difference on the output side corresponding to the difference in the physical state of the measurement target or the control target is fed back to the input side is compared with the model structure of the conventional example. As a result, the error of the model to be assumed is reduced, and the non-interacting control or the predictive control is performed using this model structure, so that highly accurate control or the like is possible.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram of the model structure of a controlled object model according to one embodiment of the present invention, and corresponds to the above-described conventional controlled object 30 of FIG.

The model structure 1 of this embodiment is a thermal model to be controlled of a thermal interference system with two inputs (u ₁ , u ₂ ) and two outputs (z ₁ , z ₂ ), and is a 2ch controlled model.

As the inputs u ₁ and u ₂ , for example, operation amounts that are two heater outputs for heating a control object such as a heat treatment board and a heat treatment furnace, respectively, and as outputs z ₁ and z ₂ , for example, A control amount that is a detected temperature from two temperature sensors that respectively detect temperatures can be assumed.

Model structure 1 of the controlled object model, the difference between the two outputs _z 1, _{z 2,} calculated by the subtraction unit 2, the two inputs _u 1, _{u 2} via a feedback element Pf, subtraction unit 3 and the adding section 4 Each is fed back with different signs.

P ₁₁ and P ₂₂ are transfer functions from the inputs u ₁ and u ₂ to the outputs z ₁ and z ₂ . In this example, the parts assigned to the two heaters to be controlled such as the heat treatment panel and the heat treatment furnace, that is, the control objects corresponding to each ch are respectively grasped as model elements having heat capacity, and each model The elements are shown as transfer functions P ₁₁ and P ₂₂ .
This model structure 1 is a thermal model of a heat interference system. When there is a temperature difference, the movement of heat occurs, and this movement of heat is equivalent to the meaning of Fourier's law that is proportional to the temperature difference. is there.

Fourier's law is, for example, “Heat Transfer Engineering”, Hideki Tasaka, p6 of Morikita Publishing Co., Ltd. The important factor that determines the amount of heat transfer is the spatial temperature gradient, and the distance between two points is expressed as Δx When the temperature difference between the two points is ΔT, the heat flux q (heat transfer amount per unit area) is λ as the thermal conductivity,
q = −λ (dT / dx)
It becomes.

  The feedback element Pf in FIG. 1 corresponds to the thermal conductivity λ of Fourier's law.

In this model structure 1, the difference between the two outputs z ₁ and z ₂ that are outputs of the above-described model elements, that is, the temperature difference is input to each model element via a feedback element Pf corresponding to the degree of interference or the like. A certain amount of input u ₁ , u ₂ , that is, an operation amount corresponding to the amount of heat is fed back with different signs, and the amount of heat is transferred from one ch to the other due to the temperature difference. This is a block diagram showing the phenomenon of thermal interference in which the amount of heat is deprived (negative) and the amount of heat is added to the other channel (positive).

  That is, in the model structure 1 of the controlled object model of this embodiment, the interference of the controlled object of the thermal system has two temperatures, and when there is a temperature difference, the amount of heat proportional to the temperature difference is transferred. Means the Fourier law that happens.

  The feedback element pf is a ratio of how much heat moves due to a temperature difference, and may be a coefficient value or a first-order lag element.

  Next, the difference in characteristics between the model structure 1 that is the controlled object model of this embodiment and the controlled object model of the conventional example of FIG. 32 will be described.

FIG. 2 shows a configuration in which temporary parameters are set in the model structure 1 of this embodiment. The steady-state characteristics of this model structure 1 are shown in FIG. Shows the control amount when the inputted two steps from 0 to a point ch1 only 1000 seconds at which the operation amount u ₁ input of the controlled object, the operation amount u ₂ of ch2 is always zero, thick A solid line indicates ch1 and a thin solid line indicates ch2.

  FIG. 4 shows the steady characteristics as a result of setting the parameters of the conventional controlled object model so as to match the steady characteristics of FIG. 3, and FIG. 5 shows the parameters of the controlled model of the conventional example.

  Even if the steady characteristics can be matched well, the transient characteristics cannot be matched.

  6 and 7 are enlarged views of the transient characteristics of the model structure 1 of this embodiment and the control target model of the conventional example. In the model structure 1 of this embodiment, as shown in FIG. 6, a temperature difference is generated and a delay occurs because heat transfer starts, whereas in the control target model of the conventional example, As shown in FIG. 7, the temperature of ch2 rises without delay. This difference appears as a difference in control performance.

If the interference factor of the control target model of the conventional example is made higher, it can be brought into a state close to the control target model 1 of this embodiment, but there is a disadvantage that the number of parameters increases and becomes complicated. To do. This will be further described later.
FIG. 8 is a configuration diagram of a temperature control system using the temperature controller of the present invention that performs non-interference control.

The temperature controller 5 of this embodiment performs non-interference using the model structure 1 that is the controlled object model of the present invention described above, and two detected temperatures z ₁ and z ₂ from the controlled object 6. And two target PID control means 7 ₁ , 7 ₂ for calculating and outputting manipulated variables u ₁ ′, u ₂ ′ based on the deviation between the target temperatures SP ₁ , SP _2, and both PID control means 7 ₁ , 7 includes a non-interacting unit 8 that processes the manipulated variables u ₁ ′ and u ₂ ′ from 7 ₂ so as to be decoupled using the model structure 1 and outputs the processed amount u ₁ to the controlled object 6.

Both the PID control means 7 ₁ , 7 ₂ and the non-interacting device 8 are constituted by a microcomputer.

  FIG. 9 is a block diagram of the decoupling device 8 and the model structure 1. The model structure 1 is obtained by modeling the controlled object 6 as described above.

The non-interfering unit 8 calculates a difference between the two outputs z ₁ and z ₂ of the model structure 1 of the controlled object 6, and outputs the output from the subtracter 9 to the model structure 1 of the controlled object 6. A compensation element Pf ′ corresponding to the feedback element Pf and an adder 10 and a subtractor 11 for adding or subtracting the output of the compensation element Pf ′ to the input manipulated variable u ₁ ′, u ₂ ′ are provided. The decoupling device 8 is configured to cancel the interference.

  Next, the difference in control characteristics between the non-interacting control of this embodiment and the conventional non-interacting control of FIG. 32 will be described.

A model structure 1 to be controlled is obtained by adding a dead time (1 second assuming a control period of pulse output) to the model structure 1 in FIG. FIG. 10 shows the state of the controlled object that has been made non-interfering. L ₁ and L ₂ are dead time elements.

  The method for evaluating the characteristics was performed by performing step input on the operation amount of ch1 and observing the change between ch1 and ch2 at that time.

  The configuration of the conventional non-interacting control is the configuration shown in FIG. 11, and the precompensator 12 for non-interacting uses the steady gain of the controlled object 6 as a matrix and its inverse matrix. The magnitude input to ch1 was set to such a value that the temperature as the output was about a steady value of about 600 ° C. before decoupling.

  Specifically, it is as follows.

  Next, FIG. 12 and FIG. 13 show control characteristics (step response) when the PID parameter is determined and controlled by the CHR adjustment rule. FIG. 12 shows a conventional example, and FIG. 13 shows this embodiment.

  Here, calculation of the PID parameter by the CHR adjustment rule will be described.

  There are the following conditions as conditions that do not go too far due to disturbance.

P = R · L / 0.95
I = 2.38 ・ L
D = 0.42 · L
In the case of the conventional non-interacting control, a step input is added to a control object that has been made non-interacting in the simulation, and measurement is performed as follows.

R = 0.0054
L = 1
Therefore:

P = R · L / 0.95 = 0.507
I = 2.38 · L = 2.38
D = 0.42 · L = 0.42
However, in order to hunting at this value, the proportional gain was weakened and the following values were adopted as the PID parameters.

P = R · L / 0.6 = 0.009
I = 2.38
D = 0.42
In the case of the non-interacting control of this embodiment, step input is added to the control object that has been made non-interfering on the simulation, and measurement is performed as follows.

R = 1.0
L = 1
Therefore, the following values were adopted as PID parameters.

P = R · L / 0.95 = 1.05
I = 2.38 · L = 2.38
D = 0.42 · L = 0.42
In the conventional non-interacting control, transient non-interacting cannot be realized as shown in FIG. Even when the PID control is performed, interference of the target value response is very large between the channels.

  On the other hand, in the non-interacting control using the model structure according to the present invention, as shown in FIG. 13, the transient characteristic of the control target after the de-interacting is also the target value response when the PID control is performed. The interference between them is very small. As described above, by using the model structure according to the present invention, a transient decoupling effect can be obtained.

In the above-described embodiment, the non-interacting device 8 performs the non-interacting based on the detected temperatures z ₁ and z ₂ of the controlled object 6, but as another embodiment of the present invention, the controlled object 6 The model structure 1 may be built in the non-interacting device 8 and made non-interfering without using the detected temperatures z ₁ and z ₂ of the controlled object 6.

  FIG. 14 is a configuration diagram of a temperature control system using the temperature controller of the present invention that performs gradient temperature control.

  As described above, the gradient temperature control is proposed by the present applicant as Japanese Patent Application No. 1999-215061, “Control Device, Temperature Controller and Heat Treatment Device” (Japanese Patent Laid-Open No. 2000-187514), and registered as Japanese Patent No. 3278807. It is.

  In this gradient temperature control, the temperature to be controlled is converted into a representative temperature, for example, an average temperature and a gradient temperature that is a temperature difference, and the average temperature and the gradient temperature are controlled as control amounts. is there.

The temperature controller 13 of this embodiment includes a first mode converter 14 that converts two detected temperatures z ₁ and z ₂ from the controlled object 6 into an average temperature and a gradient temperature according to a conversion matrix, and an average temperature. Two PID control means 7 ₁ and 7 ₂ for calculating and outputting manipulated variables u ₁ ′ and u ₂ ′ on the basis of the deviation between the temperature and the target average temperature or the deviation between the gradient temperature and the target gradient temperature, and both PIDs Distributing means 15 is provided for distributing the manipulated variables u ₁ ′, u ₂ ′ from the control means 7 ₁ , 7 ₂ so as not to interfere.

  In this embodiment, there is a feature in the configuration of the distribution unit 15, and the distribution unit 15 includes a second mode converter 16 that performs conversion according to an inverse matrix of the conversion matrix of the first mode converter 14, and And a non-interacting device 17 for making non-interfering using the model structure of the controlled object according to the present invention.

  FIG. 15 is a block diagram of the model structure 1 of the first and second mode converters 14 and 16, the non-interacting unit 17, and the controlled object 6.

The first mode converter 14 includes an adder 18 that adds two detected temperatures z ₁ and z ₂ from the model structure 1 that is the control target 6, and a multiplier that halves the output of the adder 18. 19 and a subtracter 20 that takes the difference between the two detected temperatures z ₁ and z ₂ from the model structure 1 that is the controlled object 6.

A second mode converter 16 'and the multiplier 21 to 1/2 the operation amount _{u 1} from the first PID controller _{7 1'} operation amount _{u 2} from the second PID controller _{7 2} from A subtracter 22 for subtracting the output of the multiplier 21 and an adder 23 for adding the output of the multiplier 21 to the operation amount u ₁ ′ from the _first PID control means 71 are provided.

  The non-interacting device 17 receives the compensation element Pf ′ to which the difference between the two detected temperatures of the model structure 1 of the controlled object 6 from the first mode converter 14 is given and the output of the compensation element Pf ′. An adder 10 and a subtractor 11 for adding or subtracting to the manipulated variable are provided, and the compensation element Pf ′ corresponds to the feedback element Pf of the model structure 1 of the controlled object 6. That is, the non-interacting unit 17 has a configuration in which the subtracting unit 9 is omitted and replaced with the subtracting unit 20 of the first mode converter 14 in the non-interacting unit 8 of FIG.

Note that the model structure 1 of the controlled object 6 includes dead time elements L ₁ and L ₂ as in FIG. 10 described above.

The first and second PID control means 7 ₁ and 7 ₂ , the first and second mode converters 14 and 16, the non-interacting device 17 and the like are constituted by a microcomputer.

  In each of the above-described embodiments, description has been made by applying to two controlled objects. However, the present invention can be similarly applied to three or more multi-point controlled objects.

For example, in the case of a control target of 3 inputs (u _{1 to} u ₃ ) and 3 outputs (z _{1 to} z ₃ ), for example, as shown in FIG. 16, the outputs z ₁ and z ₂ of ch 1 and ch ₂ The difference, the difference between the outputs z ₂ and z ₃ of ch2 and ch3, may be fed back to the corresponding two inputs u ₁ , u ₂ ; u ₂ , u ₃ via feedback elements Pf ₁ , Pf ₂ , respectively. In this case, the model structure to be controlled includes three model elements, and the model elements corresponding to each ch are indicated by transfer functions P ₁₁ , P ₂₂ , and P ₃₃ .

Further, four heaters for each of the ch1 to ch4 that heat the controlled object are linearly arranged as shown in FIG. 17 and control of four inputs (u _{1 to} u ₄ ) and four outputs (z _{1 to} z ₄ ). In the case of the target, for example, as shown in FIG. 18, the outputs z ₁ and z ₂ of ch ₁ and ch ₂ , the outputs z ₂ and z ₃ of ch ₂ and ch ₃ , the outputs z ₃ and z _{4 of} ch ₃ and ch ₄ , etc. The difference between the two outputs adjacent to each other may be fed back to the corresponding two inputs u ₁ , u ₂ ; u ₂ , u ₃ ; u ₃ , u ₄ . In this case, the model structure to be controlled includes four model elements, and the model elements corresponding to each channel are indicated by transfer functions P ₁₁ , P ₂₂ , P ₃₃ , and P ₄₄ , respectively. Become.

Further, as shown in FIG. 19, four heaters for each of the channels 1 to 4 that heat the control target are arranged in a radial manner around one heater of the channel 2 (u _{1 to} u ₄ ) 4 outputs. In the case of a control target (z _{1 to} z ₄ ), for example, as shown in FIG. 20, the output z _{2 of} ch2 corresponding to the central heater and the output z ₁ of the heaters of the other ch1, _{3, 4} , Z ₃ , z ₄ may be fed back to the corresponding two inputs u ₁ , u ₂ ; u ₂ , u ₃ ; u ₂ , u ₄ .

  As described above, if the interference element of the control target model of the conventional example is made higher order, it can be brought into a state close to the model structure 1 of this embodiment. It will be a lot. This will be described.

  From FIG. 6, to estimate the transfer function from the waveform of the interference relationship from ch1 to ch2, it is not a linear rise, but has a curvilinear motion in which the slope of the rise starts to be small and gradually increase. In order to express this waveform by a transfer function, it cannot be expressed unless it is at least second-order or higher. Even if it can be expressed in the second order, the following complexity arises. 2 rows and 2 columns have the following matrix orders.

Primary, secondary, secondary, primary This is converted into a parameter. To express the primary, 2 parameters of steady gain and time constant, and to express the secondary, 3 parameters are required, so a total of 10 A parameter (= 2 + 3 + 3 + 2) is required.

  On the other hand, the model structure according to the present invention is as follows.

Primary None None Primary + 1 parameters Similarly, when converted to parameters, the total is 5 parameters (= 2 + 2 + 1).

  Even in 2 rows and 2 columns, there are 5 parameters for 10 parameters, and there is an effect that can be greatly simplified. In addition, more effects are obtained when the number of matrices increases.

  The effect is shown in Table 1 below. It is assumed that 3 rows and 3 columns have the following matrix orders.

Primary, secondary, tertiary, secondary, primary, secondary, tertiary, secondary, primary

  From Table 1, it can be seen that when the number of parameters to be determined by tuning exceeds 7 lines, it becomes 10 times or more. The above is an assumption that the transfer function of adjacent interference can be approximated to the second order. If a higher order is required to reproduce a real machine waveform using a matrix, it becomes more complicated.

  Therefore, although the interference model in the matrix form can do the same as the model structure according to the present invention, it is very complicated and not practical.

  Here, an example of how to obtain the parameter of the feedback element Pf in the model structure to be controlled according to this embodiment will be described.

  For example, it can be obtained by using a search method (may be GA) such as Monte Carlo simulation.

  That is, since the thermal conductivity is a physically determined value, it can be easily assumed that there is a solution within a certain range of conditions. It is assumed that the range in which the parameter of the feedback element Pf will exist and the step response waveform when the step input is performed on the actual machine are obtained as data.

  The parameter of the feedback element Pf is randomly changed within a predetermined range. The output waveform is obtained by simulation using the parameters of the determined feedback element Pf and the step input. The waveform obtained by the simulation is compared with the actual machine waveform to calculate how much the error is (the square of the difference). While changing the parameter of the feedback element Pf at random, the parameter value of the feedback element Pf having the smallest error is obtained.

  The search is stopped when a certain number of times are repeated or the error becomes sufficiently small, and the parameter of the feedback element Pf at that time becomes the value of the parameter of the feedback element Pf as a tuning result.

  In addition, as a method for tuning the parameters of the model structure according to the present invention, all or part of the waveform of the step response, the impulse response, the temperature rise response, and the limit cycle response is stored. There is a method in which the parameter is varied and compared with the predicted waveform at that time, and the parameter with the smallest error is used as the model structure parameter.

  Next, still another embodiment of the present invention will be described.

  FIG. 21 is a configuration diagram of a temperature control system using a temperature controller according to another embodiment of the present invention.

  This embodiment is a temperature control system for a wafer heat treatment apparatus, which is a semiconductor manufacturing apparatus, and a temperature controller 35 controls the temperature of a hot plate 36 for heat-treating the wafer to a set temperature SP (target temperature). To do.

  In such a heat treatment apparatus, the wafer is heat-treated with a hot plate 36 set at a set temperature SP. This heat treatment is performed in a state where a cover called a chamber is put on the heat plate 36 on which the wafer is placed. Done.

  FIG. 22 is a diagram showing the temperature change of the wafer in a state where the wafer is placed on the hot plate 36 set at the set temperature SP and the chamber is covered. FIG. 22A shows the temperature immediately after the set temperature SP is set. FIG. 4B shows the case where the heat treatment is performed after several hours have elapsed after the set temperature SP is set.

  Conventionally, in the heat treatment of the wafer immediately after the set temperature SP is changed and the temperature of the hot plate 36 is set to the set temperature SP, as shown in FIG. Only the temperature indicated by the broken line lower than the set temperature SP is reached, and the heat treatment is performed at a temperature lower than the set temperature SP. On the other hand, in the heat treatment of the wafer after the temperature of the hot plate 36 is set to the set temperature SP and, for example, about several hours have passed, the temperature of the wafer is changed to the hot plate 36 as shown in FIG. The set temperature SP is reached and heat treatment is performed.

  As described above, although the temperature of the hot plate 36 is set to the set temperature SP, the difference in the reached temperature of the wafer to be heat-treated immediately after the settling and after several hours is caused by the hot plate 36. This is due to the influence of the above-described chamber, etc., which has a larger time constant than Until the temperature of the chamber or the like is stabilized, the temperature of the wafer is heat-treated without reaching the set temperature SP of the hot plate 36.

  As described above, conventionally, the temperature of the chamber or the like immediately after the temperature of the hot plate 36 is set to the set temperature SP is not stable, and the case where the temperature of the chamber or the like is stable after several hours has elapsed. Then, there is a problem that variation occurs in the heat treatment temperature of the wafer.

  Therefore, in this embodiment, the wafer temperature is estimated using the model structure according to the present invention, and control is performed so that the wafer temperature reaches the set temperature SP.

  Therefore, in this embodiment, as shown in FIG. 21, a wafer model 37 for estimating the wafer temperature is provided, and the deviation between the estimated temperature of the wafer predicted by the wafer model 37 and the set temperature SP is calculated. Are fed back to the integral (I) control unit 38.

  In this embodiment, the PID control means 39 includes a proportional differential (PD) control unit 40 and the integral control unit 38. The proportional differential control unit 40 includes a set temperature SP and a temperature of the heat plate 36. And the outputs of both the control units 38 and 40 are added to obtain an operation amount for the heater provided in the heat plate 36.

  As shown in FIG. 23, the wafer model 37 according to this embodiment includes a hot plate model element 41, a wafer model element 42, a chamber model element 43, a hot plate 36, and a wafer plate 37 corresponding to the thermal resistance of the wafer. It is a model structure comprising one feedback element 44 and a second feedback element 45 corresponding to the thermal resistance of the wafer and the chamber.

P _{1 to} P ₃ , P _f1 , and P _f2 indicate transfer functions.

  In this model structure, the difference in output between the hot plate model element 41 and the wafer model element 42 is fed back to the hot plate model element 41 and the wafer model element 42 through the first feedback element 44 with different signs. On the other hand, the difference in output between the wafer model element 42 and the chamber model element 43 is fed back to the wafer model element 42 and the chamber model element 43 through the second feedback element 45 with different signs. The output of the model element 42 becomes the estimated temperature of the wafer.

  This model structure takes in a hot plate and chamber in which heat is transferred to and from the wafer that is the target element of interest as a model element, and feeds back the temperature difference to the input side to estimate the wafer temperature. It is.

  Note that this model structure corresponds to the three-channel model structure of FIG. 16 described above in which one input and one output are provided.

  In this embodiment, although the hot plate model element 41 is included as the wafer model 37, the actual temperature of the hot plate 36 is set as shown in FIGS. 24 and 25 without using the hot plate model element 41. It may be used to estimate the wafer temperature with higher accuracy.

That is, in the model structure of FIGS. 24 and 25, the hot plate model element 41 is omitted, and the difference between the output of the wafer model element 44 and the output of the actual hot plate 36 is changed through the first feedback element 41 to the wafer. This is fed back to the input side of the model element 44. In the model structure of FIG. 25, each model element P ₂ , P ₃ is given a feedback input from the feedback elements Pf ₁ , Pf ₂ , but is not given any other original input. It has become.

  FIG. 26 is a diagram showing a change in wafer temperature (thin solid line) and conventional wafer temperature (thick broken line) according to the embodiment using the model structure of FIG.

  The temperature change of the wafer after changing the preset temperature SP of the hot plate 36 is shown.

  According to this embodiment, it can be seen that the wafer temperature reaches the set temperature SP earlier than in the conventional example.

  Therefore, according to this embodiment, it is possible to eliminate the difference in the wafer arrival temperature when the set temperature of the hot plate 36 is changed, and to reduce the variation in the heat treatment.

  In this embodiment, the chamber that affects the temperature of the wafer is incorporated as a model element. However, as another embodiment of the present invention, another model element, for example, a hot plate 36 is attached. A model structure incorporating a support structure or the like may be used.

  FIG. 27 is a block diagram of a temperature controller according to still another embodiment of the present invention.

  The temperature controller 46 of this embodiment controls the temperature of a control object 47 such as a hot plate to a set temperature SP, and provides a PID control means 48 that outputs a PID manipulated variable and a dead time compensation output. A dead time compensator 49 is provided, and the output of the dead time compensator 49 and the detected temperature from the control object 47 are added by an adder 50 and fed back to perform Smith compensation type dead time compensation control. It is.

  In the conventional Smith compensation type control, the model of the controlled object 47 is set to the dead time + 1st order delay, and the conventional dead time compensator 49 ′ includes the first order delay element 51 and the dead time as shown in FIG. A control target model including an element 52 and a model including a first-order lag element 51 obtained by removing the time delay element 52 from the control target model. The subtractor 53 outputs the control target model output from the model output with no time delay. The dead time compensation output is provided by subtracting at. In FIG. 28, temporary parameters are shown.

  In the conventional Smith compensation type control, as described above, the model to be controlled is set to the dead time + first order lag, whereas the actual characteristics of the controlled object are greatly different from the characteristics of the first order lag. For this reason, there is a problem that the effect of suppressing the overshoot of the target value response is often not sufficiently exhibited.

  For example, when applied to temperature control with a heater block as a control target, the heater block is not floating in the air but is fixed to a metal support structure having a large heat capacity. In this way, the heater block fixed to the support structure having a large heat capacity behaves completely different from the control target of the dead time + first order delay.

  Therefore, in this embodiment, Smith compensation control is performed using the model structure according to the present invention instead of the conventional dead time + first order lag model.

  That is, as shown in FIG. 29, the dead time compensator 49 of this embodiment replaces the conventional first-order lag element 51, for example, in the heater block temperature control, the model element of the heater block to be controlled 54, a model element 55 of the support structure that supports the heater block, and a feedback element 56 corresponding to the thermal resistance between the two, and the difference in output between the model elements 54 and 55, that is, the temperature A model structure is used in which the difference is fed back to the input of each of the model elements 54 and 55 through the feedback element 56 with different signs. In FIG. 29, temporary parameters are shown.

  FIG. 30 shows a temperature waveform (thin solid line) of the heater block with respect to a step-like manipulated variable (thick solid line) in an embodiment using such a model structure, and FIG. 31 shows a first-order lag instead of the model structure. FIG. 29 is a waveform diagram corresponding to the conventional example of FIG. 28 using elements.

  As can be seen from these figures, in the Smith compensation type control using the model structure according to the present invention, it is understood that overshoot is suppressed as compared with the conventional example.

  In this way, in the dead time + first order lag model, the overshoot suppression effect can be obtained by setting the model of the model structure according to the present invention to the dead time + the present invention with respect to the control target that cannot provide the overshoot suppression effect. .

In this way, it is possible to produce an overshoot suppression effect,
Quality in heat treatment can be improved, and the incidence of defective products can be reduced.

  Further, since overshoot can be suppressed, the settling time can be shortened and the tact time can be reduced.

(Other embodiments)
Although the above embodiment has been described by applying to PID control, the present invention is not limited to PID control but can be applied to other control methods such as on / off control, proportional control, and integral control.

  In the above-described embodiment, description has been made by applying to temperature control using a heater or the like. However, it is needless to say that the present invention may be applied to temperature control using a Peltier element or a cooler. You may apply to the temperature control which uses together.

  The model structure of the present invention can be suitably applied to internal model control (IMC), model reference adaptive control, or an observer.

  The present invention can be applied to various heat treatment apparatuses such as a thermal oxidation apparatus, a diffusion furnace, a CVD apparatus, a molding machine, and a packaging machine.

The present invention is useful as a temperature controller and a heat treatment apparatus.

It is a figure which shows the control object model which concerns on one embodiment of this invention. It is the block diagram which set the temporary parameter to the control object model of FIG. It is a figure which shows the steady state characteristic of the control object model of FIG. It is a figure which shows the steady state characteristic of the conventional control object model. It is the block diagram which set the parameter of the control object model of the prior art example. It is a figure which expands and shows the transient characteristic of the control object model of FIG. It is a figure which expands and shows the transient characteristic of the control object model of a prior art example. It is a schematic block diagram of the temperature control system using the temperature regulator of this invention. FIG. 9 is a block diagram of the decoupling device 8 and the control target model 1 of FIG. 8. FIG. 10 is a block diagram corresponding to FIG. 9 in which a dead time element is added to the controlled object model. It is a block diagram of the decoupling control of a prior art example. It is a figure which shows the control characteristic of a prior art example. It is a figure which shows the control characteristic of embodiment of FIG. It is a schematic block diagram of the temperature control system using the temperature regulator of this invention. FIG. 15 is a block diagram of first and second mode converters 14 and 16, a non-interacting unit 17, and a controlled object model 1 of FIG. 14. It is a figure which shows the control object model of other embodiment of this invention. It is a figure which shows arrangement | positioning of each ch. It is a figure which shows the control object model corresponding to FIG. It is a figure which shows arrangement | positioning of each ch. It is a figure which shows the control object model corresponding to FIG. It is a block diagram of the temperature control system which concerns on other embodiment of this invention. It is a figure which shows the temperature change of a wafer. It is a figure which shows the wafer model of FIG. It is a block diagram of the temperature control system of further another embodiment of this invention. It is a figure which shows the wafer model of FIG. It is a figure which shows the temperature change of embodiment of this invention and the wafer of a prior art example. It is a block diagram of the temperature control system of other embodiment of this invention. It is a figure which shows the structure of the dead time compensator of a prior art example. It is a block diagram of the dead time compensator of FIG. It is a figure which shows the temperature change of the heater block with respect to the step input of embodiment of this invention. It is a figure which shows the temperature change of the heater block with respect to the step input of a prior art example. It is a block diagram of the decoupling control of a prior art example. It is a characteristic figure of the control object of a prior art example. It is a figure which shows the target value response of a prior art example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Model structure 5, 13, 35, 46 Temperature controller 6, 30, 47 Control object 7 ₁ , 7 ₂ PID control means 8, 17 Decoupling device 37, 37 'Wafer model 14, 16 1st, 2nd Mode converter Pf Feedback element

Claims (4)

Temperature control means for outputting an operation amount for the control object based on a temperature detected from the control object that is a mass having a heat capacity, and a time delay compensation using a model based on the operation amount from the temperature control means A temperature controller for performing dead time compensation control, comprising a dead time compensation means for providing an output,
The model includes a control target model and a control target model excluding dead time, the both control target models include a plurality of model elements, and include at least one input and at least one output, Includes a model structure that feeds back the difference on the output side obtained by subtracting the output of the other model element from the output of one model element to the input side of the model element for each of the two model elements of the model element with different signs A temperature controller characterized by A temperature regulator comprising temperature control means for outputting an operation amount for the control object based on a detected temperature from the control object,
Based on at least one of the detected temperature and the manipulated variable, comprising a workpiece model for estimating a temperature of a workpiece to be a mass having a heat capacity to be processed by the control target,
The control means outputs the manipulated variable based on the estimated temperature of the workpiece estimated using the workpiece model and the detected temperature from the control target,
The workpiece model includes a plurality of model elements, and for each of the two model elements of the plurality of model elements, the difference on the output side obtained by subtracting the output of the other model element from the output of one model element is positive or negative The temperature controller is characterized in that it has a model structure that feeds back to the input side of the model element with different values. The temperature control means includes a proportional differential control unit to which a deviation between the target temperature to be controlled and the detected temperature is input, and an integration control unit to which a deviation between the target temperature and the estimated temperature of the workpiece is input. The temperature regulator of Claim 2 provided with these. The temperature controller according to any one of claims 1 to 3 , wherein a temperature of the heat treatment means is controlled based on a heat treatment means for heat treating the workpiece and a temperature detected by the temperature detection means. And an operation means for operating the heat treatment means based on the output of the temperature controller.

Images