Patent Grant
7680563
(1) Field of the Invention
The present invention relates to pressure control techniques which control the pressure in a low pressure processing chamber such as a plasma processing chamber and more particularly to a pressure control technique which enables high speed control regardless of plasma dissociation or change in the effective flow rate.
(2) Description of the Related Art
There are various known methods of feeding back the reading of a pressure gauge for the valve opening degree and the most commonly used method is PID control. Usually in a typical PID control method, valve opening operation amount ΔVV is calculated for each control cycle in accordance with a PID control calculation formula (Formula 1 below) to control the opening degree of the valve.
[Formula 1]
ΔVV 32 VVn+1−VVn=Gi(Pn−P0)+Gp(Pn−Pn−1)+Gd(Pn−2Pn−1+Pn−2) (1)
Here,
In the above PID control method in which the valve opening operation amount ΔVV is calculated for each control cycle to control the valve opening degree, control is stable under a condition that the gain values in the PID control calculation formula are optimal. However, it may take long time to reach a target pressure. In addition, it may take extremely long time to reach the target pressure if there is a large difference from the optimal condition in terms of gas type, gas flow rate, gas dissociation state or target pressure level.
In other words, the above control method takes long control time and requires gain optimization for each condition. Besides, hunting often occurs with a butterfly throttle valve which shows a very nonlinear relation between valve opening and exhaust speed.
The present invention has been made in view of the above problem and provides a control technique which quickly adjusts the low pressure processing chamber to a desired pressure regardless of gas type, gas flow rate or target pressure simply by optimizing constants.
In order to address the problem, the present invention provides a pressure control device for a low pressure processing chamber which has the following constitution.
The device includes: a low pressure processing chamber; gas supply means which supplies processing gas to the low pressure processing chamber; plasma generating means which supplies electromagnetic energy to the processing gas supplied to the low pressure processing chamber and generates plasma; exhaust means which exhausts gas in the low pressure processing chamber; gas pressure measuring means which measures gas pressure in the low pressure processing chamber; exhaust speed adjusting means which adjusts exhaust speed of gas to be exhausted by the exhaust means; and an arithmetic and control unit which makes control calculation to calculate an exhaust speed to make the gas pressure measured by the pressure measuring means equal to a target value, and controls the exhaust speed adjusting means according to the calculation result.
Therefore, according to the present invention, the low pressure processing chamber can be quickly adjusted to a desired pressure regardless of gas type, gas flow rate or target pressure simply by optimizing constants.
The invention will be more particularly described with reference to the accompanying drawings, in which:
Next, the preferred embodiments of the present invention will be described in detail referring to the accompanying drawings. First, the present inventors have developed PID control calculation formulas in order to provide a control technique which quickly adjusts the low pressure processing chamber to a desired pressure regardless of gas type, gas flow rate or target pressure. The characteristics of the control method using the formulas are as follows.
The correlation between exhaust speed and valve opening can be determined by measurements using standard gas in advance. Therefore, it is not necessary to make measurements for each gas type and each gas flow rate in advance.
The integral gain Gi and proportional gain Gp which satisfy the above condition can be calculated in accordance with Formulas 2 and 3:
The opening degree of the exhaust valve, ΔVV, is determined in accordance with Formulas 4 and 5 using the above integral gain Gi and proportional gain Gp and a differential gain.
[Formula 4]
ΔS=Sn+1−Sn=Gi(Pn−P0)+Gp(Pn−Pn−1)+Gd(Pn−2Pn−1+Pn−2) (4)
[Formula 5]
ΔVV=F(Sn+1)−VVn (5)
Here,
Exhaust speed operation amount ΔS was calculated in accordance with the above formulas and valve opening operation amount ΔVV was calculated based on the calculated ΔS, and the valve was operated according to the calculated ΔVV. The result shows that the following effect is achieved by using the above formulas.
By setting appropriate values for the constants “an”, “bn”, “cn” in Formulas 2 and 3, the effective flow rate is calculated from the pressure and exhaust speed in each control cycle and it is automatically fed back and reflected in gain values so that control is optimized. Therefore, quick and stable control is done regardless of gas flow rate or target pressure. Furthermore, according to the result of calculation (ΔS) using control calculation formulas, the valve opening is not directly adjusted but the valve opening is adjusted so as to attain the above calculated exhaust speed operation amount (ΔS). Consequently, even if the function of the opening degree of a throttle valve and exhaust speed, F(S), is very nonlinear, control can be performed stably with less hunting.
Also, optimal control can be achieved regardless of throttle valve structure by optimizing the constants “an”, “bn”, “cn” in Formulas 2 and 3 for a specific gas type, a specific gas flow rate and a specific target pressure and pre-calculating the function of exhaust characteristics F(S). Besides, even when the gas type, gas flow rate, gas dissociation state or target pressure changes, optimal control can be achieved.
Next, the information about optimal control which has been thus obtained will be given in detail.
In this apparatus, plasma 8 is generated by introducing a microwave generated by a magnetron 5 through a wave guide 6 and a quartz window 7 into a low pressure processing chamber 1. Processing gas introduced through a gas inlet 9 is dissociated by the plasma 8 and a sample 11 placed on a sample holder 10 is processed using radicals generated by dissociation. The low pressure processing chamber 1 has a capacity of 59 liters.
The plasma processing apparatus includes a butterfly throttle valve 3 as an exhaust speed adjusting means between the low pressure processing chamber 1 and an exhaust device 2 so that the pressure in the low pressure processing chamber is automatically controlled by feeding back the difference between the reading of a pressure gauge 4 connected with the low pressure processing chamber 1 and the target pressure to let it reflected in the opening degree of the throttle valve 3 through an arithmetic and control unit 13. The operation speed of the throttle valve 3 is 25% per second (it takes four seconds for valve operation from the fully closed state to the fully open state).
The relation between the opening of the valve 3 and the processing chamber pressure was measured using O2 gas, as the processing gas, supplied at a flow rate of 150 sccm.
Next, consideration will be given to automatic pressure control in control cycles of 300 ms using the pressure control device as shown in
Next, the control method according to the present invention (in which PID control calculations are made in accordance with Formulas 2-5 to calculate the exhaust speed and the calculation result is fed back to the exhaust speed adjusting means) was carried out. The constants “an”, “bn”, “cn” used for the control calculations are shown in
As apparent from
This suggests that according to this embodiment, quicker control can be done regardless of target pressure by optimizing constants “an”, “bn”, and “cn”.
Then, the target pressure was increased from 0.5 Pa to 2.0 Pa under the condition that O2 gas as processing gas was supplied at a flow rate of 150 sccm and plasma discharge was generated. Tests were conducted on pressure response in the conventional control method and the control method according to this embodiment at different microwave power levels. In the conventional control method, gain values optimized under the condition of O2 gas supply at a flow rate of 150 sccm in the absence of plasma discharge were used. For constants “an”, “bn”, and “cn” in the control method according to this embodiment, the values shown in
As the microwave power is larger, dissociation progresses and one O2 gas molecule turns into two O radicals and thus the number of moles becomes larger and the effective flow rate increases. Therefore, in the conventional control method, if gain values obtained without plasma are used, it takes longer time to reach the target pressures and with 1000 W microwave power, target pressure 2 Pa is not reached in 15 seconds, as shown in
On the other hand, in the control method according to this embodiment, the effective flow rate is calculated for each control cycle and automatically fed back and reflected in gain values. Consequently, pressure 2 Pa is reached in about two seconds, whether the microwave power is 0, 500 or 1000 W, as shown in
In the conventional control method, as shown in
This demonstrates that the control method according to this embodiment permits quick control with optimized constants “an”, “bn”, and “cn”, regardless of plasma dissociation or effective flow rate.
Then, the target pressure was increased from 0.5 Pa to 2.0 Pa under the condition that SF6 gas as processing gas was supplied at a flow rate of 150 sccm and there was no plasma discharge. Tests were conducted on pressure response in the conventional control method and the control method according to this embodiment. In the conventional control method, gain values optimized under the condition of O2 gas supply at a flow rate of 150 sccm in the absence of plasma discharge were used. For constants “an”, “bn”, and “cn” in this embodiment, the values shown in
Since SF6 gas is harder to exhaust than O2 gas, the effective exhaust speed decreases about 60 percent. Consequently SF6 gas requires longer time to reach the target pressure than O2 gas, and as shown in
This demonstrates that the control method according to this embodiment permits quick control with optimized constants “an”, “bn”, and “cn”, regardless of gas type.
As explained above, the control method according to this embodiment permits robust quick pressure control which does not depend on gas pressure, gas flow rate, gas type and gas dissociation state. In the above examples of pressure control, the flow rate was constant; however, this embodiment achieves a similar effect even in pressure control at the time of gas change which involves a large change in the flow rate or in maintaining the pressure constant when the gas flow rate is changed.
In this embodiment, integral gain Gi and proportional gain Gp are calculated in accordance with Formulas 2 and 3 respectively and the valve opening operation amount is calculated in accordance with Formulas 4 and 5. However, a similar effect can be achieved irrespective of the above formulas if the following conditions are satisfied: (a) integral gain Gi is a function which has a positive correlation with exhaust speed Sn and also has a negative correlation with current pressure Pn and target pressure value P0; (b) proportional gain Gp is a function which has a negative correlation with current pressure Pn and target pressure P0; and (c) the valve opening is adjusted so that the value obtained by the PID control calculation formulas is the exhaust speed operation amount. Although differential gain Gd is 0 in this embodiment, a similar effect can be achieved using a differential gain value other than 0 as far as the value is appropriate.
One feature of the first embodiment is that the valve opening is not directly adjusted according to the value obtained from the PID control calculation formulas but the valve opening is adjusted so that the value obtained by the above calculation formulas is the exhaust speed operation amount. The advantage of this feature is discussed below.
First, a pressure control test was conducted using Formulas 2 and 3 which express integral gain Gi and proportional gain Gp and using PID control calculation formula, Formula 1, which directly expresses the valve opening operation amount.
In the test, O2 gas as processing gas was supplied at a flow rate of 150 sccm and the target pressure was increased from 0.5 Pa to 2.0 Pa in the presence of plasma discharge. Like the first embodiment, pressure response at different microwave power levels was investigated. For constants “an”, “bn”, and “cn”, the values optimized without plasma as shown in
As shown in
On the other hand, in the first embodiment, the value obtained from the PID control formulas using Formulas 4 and 5 is the exhaust speed operation amount. This may be the reason why control is done stably without hunting as shown in
Although the first embodiment uses the PID control calculation formula and Formulas 4 and 5 as the transform expressions from the PID control calculation formula for the valve opening operation amount, any PID control calculation formula may be used to achieve a similar effect if the valve opening can be adjusted so that the formula expresses the exhaust speed operation amount. Although the first embodiment uses a butterfly throttle valve which provides a nonlinear relation F(s) between valve opening and exhaust speed, a pendulum type throttle valve which demonstrates relatively linear exhaust characteristics may be used to achieve a similar effect.
Another feature of the first embodiment is that integral gain Gi and proportional gain Gp are functions which have a negative correlation with target pressure P0 in Formulas 2 and 3. The advantage of this feature is discussed below.
First, in order to eliminate the correlation of integral gain Gi and proportional gain Gp with target pressure P0, a pressure control performance test was carried out where b2 and c2 in Formulas 2 and 3 were 0.
In the test, O2 gas as processing gas was supplied at a flow rate of 150 sccm and the target pressure was increased from 0.5 Pa to 1.0, 2.0 and 3.0 Pa in the absence of plasma discharge and pressure response was investigated. For constants other than bc and c2, namely “an”, “bn”, and “cn”, the values optimized for target pressure increase from 0.5 pa to 2.0 Pa as shown in
Under the optimized condition for 2.0 Pa target pressure, the pressure rises almost as quickly as in the case shown in
Using the same constants, a test was conducted to investigate the pressure response in decreasing the target pressure from 1.0, 2.0 and 3.0 Pa to 0.5 Pa.
This suggests that quick control is possible in the first embodiment even under different pressure conditions because integral gain Gi and proportional gain Gp have a negative correlation with target pressure P0. Although the values of integral gain Gi and proportional gain Gp are given by Formulas 2 and 3 respectively in the first embodiment, as far as integral gain Gi and proportional gain Gp are functions which have a negative correlation with target pressure P0, they may be given in another way to achieve a similar effect.
Another feature of the first embodiment is that integral gain Gi is a function which has a positive correlation with exhaust speed Sn in Formula 2. The advantage of this feature is discussed below.
First, in order to eliminate the correlation between integral gain Gi and exhaust speed Sn, a pressure control performance test was carried out where a2 in Formula 2 was 0.
In the test, O2 gas as processing gas was supplied at a flow rate of 150 sccm and the target pressure was increased from 0.5 Pa to 1.0, 2.0 and 3.0 Pa in the absence of plasma discharge and pressure response was investigated. For constants other than a2, namely “an”, “bn”, and “cn”, the values optimized for target pressure increase from 0.5 Pa to 2.0 Pa as shown in
This suggests that quick control is possible in the first embodiment even under different pressure conditions because integral gain Gi has a positive correlation with exhaust speed Sn.
Although the value of integral gain Gi is given by Formula 2 in the first embodiment, as far as integral gain Gi is a function which has a positive correlation with exhaust speed Sn, any other function may be used to achieve a similar effect. In the test, the values optimized for pressure increase from 0.5 Pa to 2.0 Pa as shown in
A pressure control test was conducted on the pressure control device shown in
In the test, O2 gas as processing gas was supplied at a flow rate of 150 sccm and the target pressure was increased from 0.5 Pa to 1.0, 2.0 and 3.0 Pa in the absence of plasma discharge and pressure response was investigated. Formulas 2 and 3 were used for integral gain Gi and proportional gain Gp, and the values optimized for target pressure increase from 0.5 pa to 2.0 Pa as shown in
An attempt to further improve response in this situation would result in a further overshoot. Using the constants shown in
In order to solve this problem, the inventors have developed Formulas 6 and 7 in which pressure value Pn for each control cycle is added to the denominators of Formulas 2 and 3 which respectively express integral gain Gi and proportional gain Gp in the present invention.
Here, a3, b3, c3: positive constants (fixed values)
In the test, Formulas 6 and 7 were used to express integral gain Gi and proportional gain Gp respectively and O2 gas as processing gas was supplied at a flow rate of 150 sccm and the target pressure was increased from 0.5 Pa to 1.0, 2.0 and 3.0 Pa in the absence of plasma discharge and pressure response was investigated. The values optimized for target pressure increase from 0.5 Pa to 2.0 Pa as shown in
Using the constants shown in
This suggests that robustness against time lags is improved by using Formulas 6 and 7, in which pressure value Pn for each control cycle is added to the denominators of Formulas 2 and 3, in the calculation of integral gain Gi and proportional gain Gp.
Although the values of integral gain Gi and proportional gain Gp were given by Formulas 6 and 7 respectively in this embodiment, as far as integral gain Gi and proportional gain Gp are functions which have a negative correlation with both target pressure P0 and pressure value Pn for each control cycle, they may be given in another way to achieve a similar effect.
Using the pressure control device shown in
At the first step, the polysilicon 61 and the silicon oxide film 62 are etched. At the second step, the polysilicon 63 is etched until the silicon oxide film 64 is exposed. At this time, the polysilicon 63 is tapered by etching as illustrated in
By taking these three steps, the sample is expected to become a rectangular shape of the polysilicon from which said tapered portion is removed, as illustrated in
Regarding how a sample having the structure as shown in
When the control method according to the present invention was used, the silicon oxide film 64 remained almost intact across its thickness as illustrated in
Next, the reason for the partial loss of the silicon oxide film 64 at the pattern bottom was investigated.
As indicated in
Therefore, it may be considered that the silicon oxide film 64 was thin and part of the silicon oxide film 64 was etched in the period from the start of Step 3 until 2 Pa was reached.
Besides, at Step 2, the sudden decline in the flow rate could not be followed up by pressure control and the pressure remained as low as 0.3 Pa or less. Since the polysilicon etch rate was as low as 60 nm/min or less in the low pressure range below 0.3 Pa, etching of the polysilicon hardly progressed. It may be considered that polysilicon residue 66 was generated for this reason.
Consequently in the control method according to the present invention, even if the steps are continuously carried out, etching can be properly done and throughput can be improved.
As explained so far, according to the preferred embodiments of the present invention, an exhaust speed which makes the pressure in the processing chamber equal to the target pressure is calculated by PID control calculation; and feedback control of exhaust speed adjusting means (throttle valve) is performed to match the valve opening degree to the calculated exhaust speed. In the PID control calculation, the integral gain and proportional gain have a negative correlation with the target pressure value and the integral gain has a positive correlation with exhaust speed. Hence, irrespective of the throttle valve's exhaust characteristics (nonlinear relation between exhaust speed and valve opening), the pressure in the processing chamber can be quickly brought to the desired pressure level. Even when gas type, gas flow rate or target pressure is altered, optimization of the gains is not needed. Therefore, quick, flexible pressure control can be performed.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2007-021692 | Jan 2007 | JP | national |
| Number | Name | Date | Kind |
|---|---|---|---|
| 6142163 | McMillin et al. | Nov 2000 | A |
| 6942892 | Ishibashi | Sep 2005 | B1 |
| 20050167398 | Kitazawa | Aug 2005 | A1 |
| 20050269031 | Tyler et al. | Dec 2005 | A1 |
| 20080050538 | Hirata | Feb 2008 | A1 |
| Number | Date | Country |
|---|---|---|
| 10-011152 | Jan 1998 | JP |
| Number | Date | Country | |
|---|---|---|---|
| 20080183340 A1 | Jul 2008 | US |
