JPS62204528A - Dry process processor - Google Patents

Abstract

PURPOSE:To improve the processing capacity of a dry process processor by calculating the gradient of pressure change in a processing chamber, and calculating to predict the brake amount of exhaust means to enhance the control responsiveness of dry process processing pressure. CONSTITUTION:A processing chamber 1 opens CV15 to exhaust gas, and a sample is conveyed thereto. An entire control microcomputer 14 feeds a flow rate set signal to MFC8-10, and inputs processing gas from valves 11, 12, 13, 3. A pressure control microcomputer 19 fully closes the CV15 by a command from the computer 14, and measures the pressure of the chamber 1 by a pressure sensor 21. After a predetermined time is elapsed, it calculates a gradient dp/dt, obtains a control target pressure from the computer 14 and predetermined exhausting velocity by dp/dt, obtains initial opening by the program of the computer 19, and calculates a predetermined time tv from full closure of the CV15 to the initial opening. The operation starting pressure Ps of the CV15 is predicted by using an equation (1), and when the pressure arrives at Ps, the CV15 starts moving to the predetermined opening. Thereafter, the pressure is controlled by a normal PLD control.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a dry process apparatus, and particularly to a dry process apparatus suitable for plasma processing samples such as semiconductor element substrates.

[Conventional device]

For example, as described in Japanese Patent Laid-Open No. 60-80225, a conventional device utilizes the fact that gas pressure is set as a processing condition, and calculates the relationship between this set pressure condition and the opening/closing ratio of a variable conductance valve. The program is stored in the memory, and the gas supply amount and exhaust speed are controlled according to the program values.

In this case, in order to shorten the time required to control the pressure in the processing chamber, the maximum flow rate that can be flowed by the mass flow controller is set at the time of gas introduction into the processing chamber, regardless of the set gas flow value determined by the process processing conditions.

The variable conductance valve is fully opened, that is, the pumping speed is set to zero, and then the pressure in the processing chamber is set to ±2 of the control target pressure.
When it reaches 0%, the gas flow rate is changed to the set value under the process conditions, and the opening degree of the variable conductance valve is changed to the programmed set value. In addition, in the second and subsequent processes, the contents of the memory are rewritten as the opening degree of the variable conductance valve to the opening degree when the gas pressure was stable in the previous process.

[Problem that the invention seeks to solve]

In the above-mentioned conventional technology, there is a problem that optimal control cannot be performed in the first treatment each time the process conditions are different, and the pressure control time becomes longer. In addition, in order to ensure a certain degree of controllability, it is necessary to preset the opening degree of the variable conductance valve that is estimated to be preferable each time the process processing conditions change (
There is a need for non-memory LSIs in preliminary experiments to establish process conditions and in semiconductor manufacturing processes.
When used in processes such as the production of products that frequently require different coating conditions, there is a problem in that it places a large burden on the operator.

Furthermore, in the above conventional technology, the pressure that is the starting point for changing the flow rate of the mass flow controller and the opening degree of the variable conductance valve to preset values is set in advance as a constant within a certain range (±20%) of the control target pressure. Therefore, if the set flow rate or control target pressure changes, the time required for the mass flow rate to stabilize from the maximum flow rate to the set flow rate, or the time required for the variable conductance valve to open from the fully closed state to the specified opening degree. As a result, the pressure significantly overshoots or underruns the set value, resulting in a problem that pressure control takes a long time.

An object of the present invention is to provide a dry process processing apparatus that can improve the control response of the dry process pressure and improve the processing capacity without putting a burden on the operator under any hot process conditions. be.

[Means for solving problems]

The above purpose is to measure the pressure in the processing chamber after introducing the strained process gas into the processing chamber with the variable conductance valve fully open, calculate the gradient of the pressure change in real time, and calculate the actual value from the gradient value and the volume of the processing chamber. A means for determining the flow rate of the processing gas flowing into the heat treatment chamber, and a required exhaust speed and opening degree of the variable conductance valve programmed in advance based on the required exhaust speed determined by the flow value and the control target pressure which is the set value. The process chamber calculates the opening degree of the valve that matches the control target pressure using this method, and also predicts the time required for the valve to open from the fully closed state to the opening degree determined from the above number. ill +,, means to back-calculate the pressure value that increases within the time, and to open the variable conductance valve from fully open to a predetermined opening degree when the processing chamber pressure reaches the pressure value obtained by subtracting the pressure value from the control target pressure. (This is achieved by providing a means for starting the operation.

[For production]

By measuring the gradient of pressure change when multiple process gases are flowing into a processing chamber with a volume of V with the variable conductance valve fully closed and the exhaust speed at zero,
The actual total fEtQ of gas am is determined by the following equation.

0 ・・・・・・・・・・・・・・・
(1) Q=V-T If the introduced gas flow rate does not fluctuate, the predetermined time lio'
If the actually measured pressure at 2 is P1eP2, the pressure gradient can be determined.

Substituting this into equation 11), if the control target pressure is pc, and the exhaust speed for keeping this pressure hot is S, then S can be found from the following equation.

If the evacuation speed of the vacuum pump is selected to be sufficiently high, for example, 2 to 5 times the above S', the pumping speed at the end of the processing chamber will depend on the conductance of the piping and the opening degree of the variable fundance valve. The pumping speed can be determined in advance by calculation or actual measurement.

Therefore, if the correlation between the opening degree of the variable conductance valve and the exhaust speed at the end of the processing chamber is stored in advance in the storage device, the opening degree of the variable conductance valve can be determined for the required exhaust speed shown by equation (4). can be easily determined.

On the other hand, for a variable conductance valve, once its drive system is determined, the opening/closing acceleration/deceleration and maximum speed are uniquely determined, so the time tv required to open from the fully closed state to the above opening degree can be easily determined. It will be done.

Figure 7 shows the rise curve of the pressure in the processing chamber as a function of time when gas is introduced into the processing chamber with the variable conductance valve fully open. , (excluding the initial unstable part, as is clear from equation (1), the pressure increases linearly with a constant gradient. However, if the variable conductance valve opens and the pumping speed becomes Sx, then the variable The pressure at the operation start point of the conductance valve is PS l.If the elapsed time after the start of operation is Δt, the pressure Pt after Δ has elapsed is expressed by the following equation.

P1= (P3-Q/3x) exp(-enX△t'
)+3-(51) In other words, the pressure increase curve deviates from a straight line, and the rate of pressure increase decreases in proportion to the reciprocal of the exponential function of the elapsed time.

Therefore, when the valve is opened from a pressure point A that is relatively low compared to the control target pressure pi, as shown by the two-dot chain line,
It takes time to reach Pt, and conversely, P
When the opening operation of the valve is started from a pressure close to t, an overshoot occurs as shown by the broken line, and feedback control is applied to lower the pressure, causing vibration, which also increases the time required for pressure control. As is clear from the above explanation, the pressure control time becomes the shortest depending on the gas flow rate and control target pressure. Appropriate opening operation start pressure P of the valve
3 exists. Therefore, it is necessary to predict an appropriate P based on various processing conditions set in advance and actual measurement data during processing, and to start opening the valve from the time when the pressure in the processing chamber being measured reaches Ps. Bye.

For example, if the pressure increases linearly with the valve fully open and the point where it reaches the control target pressure is point D, then the pressure P
The time required to reach a (i.e. from point B to point D) t
A becomes.

Therefore, during tAIIW, the variable conductance valve is expected to open from fully open to the required open position (Ps).

That is, by rearranging equation (7) using equation (13), Ps can be obtained.

In this way, by predicting the pressure Ps at which the valve should start operating based on the pressure Pl+P2 actually measured during processing, the set pressure (control target pressure), and the response time ty of the variable conductance valve, it is possible to adjust the pressure Ps at which the valve should start operating. , can complete hot water pressure control in a short time.

〔Example〕

An embodiment of the present invention will be described below with reference to FIGS. 1 to 7.

In FIG. 1, a processing chamber 1 for performing a dry process is provided with a supply pipe 2 for supplying processing gas, an on-off valve 3, and an exhaust pipe 4 for exhausting the inside of the processing chamber 1.

The gas supply unit includes a plurality of processing gas sources 5. 6. 7, a mass flow controller (hereinafter abbreviated as MFC) 8.9.10 that smoothly controls the flow rate of each gas set according to an external command, and a valve 11.10 that mixes these gases or stops supplying them. It consists of 12°13. MFC8, 9, 10 each have one internal D/A, A/
It is equipped with a D converter (not shown) and is connected to a microphone and computer 14 that control the entire processing device so as to be able to send and receive data. The gas rILffi is set by the overall control microcomputer 14, and each MF
From Cs~10, the measurement data of gas MtfA is sent back to the microcomputer 14.
4 monitors the operations of MF08 to MF10.

The exhaust section consists of a variable conductance valve (hereinafter abbreviated as Cv) and a vacuum exhaust device 16.

CV 15 is driven by drive motor 17, motor 1
7 is encoder 1 for detecting the opening degree of CVis.
8'4 are connected, and the signal is sent to the pressure control microcomputer 19.

The drive motor 17 is connected to the pressure control microcomputer 1.
910 controlled driver application.

A signal from a pressure sensor installed in the processing chamber l is sent to the microcomputer 19 via an A/D converter n.

The overall control microcomputer 14 sends pressure setting data to the pressure control microcomputer 19.

In the above configuration, the operation of each part during pressure control will be described below.

Processing chamber 1 has CV 15 in full UU state and is 10-'
It is evacuated to a low pressure of Torr to 10 Torr. Next, a sample (not shown) is carried into the processing chamber 14 by a transport device (not shown). When the loading is completed, the overall control microcomputer 14 sends a setting signal for the flow rate of each of MF08 to MF10, opens valves 11, 12, and 3, and introduces the processing gas into the processing chamber 1. .

At the same time, the pressure control microcomputer 19 starts the second
Start the pressure control shown in the figure.

That is, immediately open the CV 15 fully, and press the pressure sensor 21.
The pressure in the processing chamber 1 is sampled and measured. After a certain period of time has elapsed, the pressure gradient dp/dt is calculated using the sampling pressure value P1+''2 at time tit'2.Then, the pressure gradient dp/dt is calculated from the sampling pressure value P1+''2 at time tit'2. The required opening (that is, the initial opening) is determined by a program that calculates the required exhaust speed and calculates the correlation between the exhaust speed and the opening of the CV 15, which is stored in advance in the pressure control microcomputer 19. Required time t to reach the initial opening degree
Calculate y. From these calculation results, the operation start pressure Ps of the CVis is estimated using equation (8). When the sampling pressure by the pressure sensor 21 reaches P3, CV
15 starts moving to a predetermined opening degree. After this, pressure control is performed using normal PID control.

In this case, the volume of processing chamber 1 is V=1401! , the correlation between the opening degree of CMi5 and the exhaust speed at the exit of the processing chamber l is 10-3 Torr ~ l O-'' Torr using SF gas.
An experiment was conducted using the gas flow rate and control target pressure as parameters for the apparatus having the relationship shown in FIG. 3 in the pressure range of .

The drive motor 17 of the CV 15 uses a pulse motor with a step angle of 0.9 degrees, and the CV opening degree is 4 degrees from 0% to 100 degrees.
Moves with 000 pulses, moving speed is 1300P
Accelerate P8-1000PPS in 0.25 seconds, constant speed 10
00PPs, 1000PPS to 300PPs during deceleration
B'k The trapezoidal speed was set to decelerate in 0.25 seconds.

'JE 4 txt u, gas flow fi15 of SF6 gas
0 to 2008CCM, control target pressure Pt is 0
．． 01 to 0.09 Torr, and shows the relationship between Pt calculated by the microcomputer and the initial opening of CV15.The initial opening of Cv is selected between approximately 10 and 40 Torr depending on the process processing conditions. I understand.

FIG. 5 shows the relationship between the value obtained by dividing the pressure P at which cvis control should be started, which is preset by the pressure control microcombination 19, by the control target pressure pt, and Pt. The flow rate is expressed as a parameter.

As is clear from FIG. 5, the timing at which CVis control is started is not limited to a very certain range as shown in the above-mentioned prior art, and P@/Pt varies from 0.1 to 0.1 depending on the process conditions. It can be seen that it varies over a wide range of 0.98. This tendency becomes stronger as the control target pressure pt is lower.If the process conditions are limited to relatively high pressures where pt is 0.ITorr or higher, Ps/Pt is limited to a range of 0.85 to 0.98. It is also possible to do so.

The next summary in Figure 6 shows the process chamber 1 from the aforementioned predictive control method number.
The graph shows pt and the time required for the internal pressure to remain within ±3 degrees of Pt for 1.5 seconds or more after gas introduction (that is, pressure control time). In the figure, two straight lines passing through the origin are c v
The figure shows the time to reach the control target pressure pt theoretically at the earliest when is is fully closed (that is, the pumping speed is zero), and the upper side is the gas flow fl 50 SCCM.

The lower one shows the case where the gas flow rate is 11008 CC.

The difference between the inventor's experimental results and this straight line is the delay time caused by controlling c v ts, and corresponds to tB shown in FIG. 7. Although it is theoretically clear that this tB cannot be reduced to zero, the short tB provides good controllability and allows pressure control to be completed in a short time. In the conventional method, tB takes 10 to 30 seconds, but in this case,
As shown in FIG. 6, control is performed in an extremely short time of 1 to 3 seconds.

As described above, in this example, after introducing the processing gas, Cv
Measure the rising pressure gradient with the valve fully closed, and based on this, calculate the Cv opening to maintain the control target pressure and the delay time for opening to that opening, and calculate the optimal Cv operation υl. Since the Cv operation is performed with the starting pressure on the slave side, the pressure does not overshoot or underrun, and the pressure can be controlled in an extremely short time. Especially when the processing pressure is 0. I
For low-pressure process gases below Torr, this effect becomes remarkable, and the control delay time can be reduced from the conventional 115~'/'t.
o” Can be shortened.

In addition, in this embodiment, a pressure control method was described in which the pumping speed is controlled by controlling the opening degree of Cv, but basically, if the pumping method is capable of varying the pumping speed, the same method can be used. It is clear that it can be done.

For example, if the exhaust system is configured so that the pumping speed can be varied by controlling the rotational speed of a turbo molecular pump or mechanical booster pump, the correlation between the pumping speed and the rotational speed can be programmed into a microcomputer. Good.

In addition, in this example, the method of controlling the pressure before processing the sample was described, but in an actual dry process, after the pressure control is completed, energy such as RF, microwave, or magnetic field is applied from outside to the inside of the processing chamber. Generate plasma. For this reason, the thermal kinetic energy of the gas molecules increases, and as a result, as shown in FIG. 8, the pressure in the processing chamber increases after the start of discharge, so it is necessary to lower the pressure to the control target pressure by applying pressure control. Usually, the rate of pressure increase due to discharge is P
In a single-wafer dry process where the t reaches 10 to 20 inches and the processing time is short, such changes in processing conditions have a large effect on the characteristics of the sample (and must be avoided as much as possible). The present invention can be applied as a means to solve such problems.

That is, as shown in FIG. 9, the original control target pressure pt is lowered by the differential pressure corresponding to the pressure increase after the start of discharge.
The pressure at tt is programmed to be set as the control target pressure before the start of discharge, and the pressure control according to the present invention is performed on this pressure, and the control target pressure is set immediately after the start of discharge.
Program it to return from tt to the original PC. Such a method has the effect of preventing overshoot during discharge and quickly controlling the pressure.

In addition, in this embodiment, it is necessary to perform various calculations in a short time after the gas is introduced, and a microcomputer that controls the entire system including MFC control and a pressure control dedicated computer that detects pressure and controls the exhaust gas 21g1 are required. By installing a second microcomputer, we aim to speed up calculation processing.

Furthermore, in this example, a case was shown in which the correlation between the Cv opening degree and the exhaust speed was programmed in advance. If the inside or valve becomes dirty or clogged,
It is conceivable that the pumping speed will decrease.

In order to prevent the controllability from deteriorating due to this, it is necessary to check the correlation from time to time and change the program contents.

The present invention is equipped with a means for measuring and calculating the actual amount of gas flowing into the processing chamber, so if the gas is flowed while maintaining the Cv at a certain opening and the degree of vacuum reached at that time is measured, (
4) Using the formula, the pumping speed can be determined. If such a processing process is programmed into a microcomputer, it becomes possible to easily update the correlation between Cv opening and exhaust speed, thereby making it possible to self-diagnose the degree of deterioration of the exhaust system. There is also this effect.

〔Effect of the invention〕

According to the present invention, it is possible to predict and control the optimal pressure control conditions during the pressure increase after gas introduction under any process conditions, so that the parameters for pressure control can be changed every time the process conditions change. There is no need to set it each time (the burden on the operator can be reduced, and the response speed of pressure control is faster, so it has the effect of improving the throughput of the dry process.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram of a dry process processing apparatus showing an embodiment of the present invention, FIG. 2 is a pressure control flowchart, and FIG. 3 is a relationship diagram between Cv opening degree and pumping speed of the experimental equipment.
FIG. 4 is a relationship diagram between the control target pressure and the Cv initial opening degree obtained in the experiment, and FIG. 5 is a relationship diagram between the control target pressure and the Cv control start pressure ratio, '? Figure 1g6 is also the relationship diagram between the control target pressure and the pressure control time after the gas conductor, No. 7
The figure is a schematic diagram of the relationship between the elapsed time after gas introduction and the processing chamber pressure. Figure 9E8 is a schematic diagram showing an example of the pressure change in the processing chamber when electric discharge is performed. FIG. 3 is a schematic diagram of the elapsed time after gas introduction and the processing chamber pressure, showing an example of the present invention. l...Processing chamber, 5 to 7...Gas source, 8 to 10...MFC, 14...
Overall control microcomputer, 15...-CV
, 16... Vacuum exhaust device, 17... Drive motor, 18... Encoder, 19...
・・Pressure control microcomputer, ■・・・・・・
Driver, Figure 1 I~-11 processing! , 5-7---n source, 8-10
-----MFC/4----11JIFflfl
q, tya ko;j'i, -7,15---CV16
---! 2flF/lJ%f, /7---, #*t-
7,z8--- Knee/U-y”/'/---M1
1iII*MTf7ryv;t3-7.20----
F? 4ha'-21-1--IJtH-, 22--
-A10-18 Year 2 Figure 3 Prisoner Figure 4 Autopsy Qrasa Force (X10-'Torr) 4゛5 Figure PI Qing HaN-1) Pt (X10-'Tart)
Oro diagram #J @ eyes stiη P also (xlo-' 7arr) Off diagram "S 1L stock's shelf 18 Figure 19

Claims (1)

[Claims] 1. In a dry process processing apparatus having means for controlling the flow rate of processing gas introduced into the processing chamber and exhaust means capable of controlling the exhaust speed, after the processing gas is introduced into the processing chamber, means for measuring and calculating the gradient of pressure change to calculate the actual inflow amount of the processing gas into the processing chamber; and controlling the exhaust means for a required exhaust speed determined by the actual inflow amount and a set control target pressure. means for predicting and calculating the amount from the relationship between the pumping speed of the exhaust means and the control amount; and means for predicting the delay time required for changing from the control amount before the control of the exhaust means to the predicted and calculated control amount. A dry process processing apparatus comprising means for predicting the timing of starting control of the exhaust means in correlation with the delay time.