Method and apparatus for detecting optimal endpoints in plasma etch processes

• US 5,738,756 A
• Filed: 06/30/1995
• Issued: 04/14/1998
• Est. Priority Date: 06/30/1995
• Status: Expired due to Term

• Current Assignee: Lam Research Corporation
• Original Assignee: Lam Research Corporation
• Inventors: Liu, Alexander F.
• Primary Examiner(s): Breneman, R. Bruce
• Assistant Examiner(s): ADJODHA, MICHAEL
• Application Number:
  US08/497,461
• Time in Patent Office: 1,019 Days
• Field of Search: 156/626.1, 156/627.1, 156/345, 204/192.33
• US Class Current: 216/60
• CPC Class Codes: H01J 37/32935 Monitoring and controlling tubes by information coming from the object and/or discharge
• CPC Class Codes: H01J 37/32963 End-point detection
• CPC Class Codes: H01J 37/32972 Spectral analysis

Abstract

An improved method for specifying and reliably detecting endpoints in processes such as plasma etching, where the signal-to-noise ratio has been severely degraded due to factors such as "cloudy window" and low ratio of reactive surface area to non-reactive surface area. The improved method of the invention samples signals produced by photo sensitive equipment, digitally filters and cross-correlates the data, normalizes the data using an average normalization value, and provides further noise reduction through the use of three modes of endpoint specification and detection. The three modes of endpoint specification and detection require a pre-specified number of consecutive samples to exhibit a certain behavior before the endpoint is deemed detected and the process terminated as a result. The three modes of endpoint specification and detection also permit a very fine control of the etch time by permitting the user to adjust the specified endpoint by gradations of the sampling period. The improved method of the invention is capable of being implemented on known apparatus.

Claims

1. A method of detecting an endpoint for terminating a chemical process involving one or more chemical species, the process taking place in a chamber, said method comprising the steps of:

defining the endpoint in the form of a mode, a minimum value of a delay count and a threshold;

initiating the chemical process;

detecting intensity of light emitted during the chemical process by one or more of the one or more species in the form of an analog signal representing the intensity of each of the detected species;

sampling each analog signal at a rate and at the beginning of a current sampling period defined by the rate to produce a set of raw data samples for the current sampling period;

pre-filtering each of the set of raw data samples of the current sampling period;

normalizing each of the set of pre-filtered data samples;

auto-correlating the set of normalized data samples to produce a multichannel value for the current sampling period;

post-filtering the multichannel value;

incrementing the delay count when the threshold has been exceeded by the post-filtered multichannel value and the mode is met, otherwise setting the delay count equal to zero; and

repeating said pre-filtering, normalizing, auto-correlating, post-filtering, and incrementing steps for each set of raw data samples until the delay count reaches or exceeds the minimum value; and

terminating the process when the delay count reaches or exceeds the minimum value.

2. The method of claim 1 wherein the mode is met when the post-filtered multichannel value of an immediately preceding sampling period is less than the post-filtered multichannel value of the current sampling period.

3. The method of claim 1 wherein the mode is met when the post-filtered multichannel value of an immediately preceding sampling period is greater than the post-filtered multichannel value of the current sampling period.

4. The method of claim 1 wherein the mode is met when the post-filtered multichannel value of the current sampling period is less than the threshold and is less than the post-filtered multichannel value of an immediately preceding sampling period.

5. The method of claim 1 wherein said step of normalizing each of the set of pre-filtered data samples further comprises the steps of:

calculating an average normalization value corresponding to each of the sampled analog signals; and

dividing each of the set of pre-filtered data samples by the average normalization value corresponding to the analog signal from which each of the pre-filtered samples was taken.

6. The method of claim 5 wherein said step of normalizing is not performed for a first period of time with reference to a time at which the process is initiated.

7. The method of claim 6 wherein said step of dividing is not performed for a second period of time with reference to lapse of the first period of time.

8. The method of claim 7 wherein said step of calculating further comprises the steps of:

during the second period of time, summing together pre-filtered data samples taken from each of the sampled analog signals; and

dividing each of the summations by a total number of samples taken during the second predetermined period of time to produce the average normalization values corresponding to each of the sampled analog signals.

9. The method of claim 1 wherein said step of sampling each analog signal further comprises the steps of:

capturing a raw analog data sample of the amplitude of each analog signal; and

converting each of the raw analog samples to a raw digital data sample.

10. The method of claim 1 wherein said step of pre-filtering each of the raw data samples further comprises the step of calculating the pre-filtered data samples in accordance with the equation Df1(t)=(1-Q1)×

Df1(t-1)+Q1×

D1(t), wherein Df1(t) is the pre-filtered data sample being currently calculated, Q1 is a pre-filter constant, Df1(t-1)is the pre-filtered data sample calculated during a preceding sampling period.

11. The method of claim 1 wherein said step of auto-correlating further comprises the step of calculating the multichannel value in accordance with the equation MCA(t)= (A-B)+(C-D)!÷

(E), where A, B, C, D and E represent up to five of the normalized data samples, and MCA(t) is the multichannel value being calculated.

12. The method of claim 1 wherein said step of post-filtering the multichannel value further comprises the step of calculating the post-filtered multichannel value in accordance with the equation Df2(t)=(1-Q2)×

Df2(t-1)+Q2×

MCA(t), wherein Df2(t) is the post-filtered multichannel value of the current sampling period, Q2 is a post-filtering constant, Df2(t-1) is the post-filtered multichannel value of a preceding sampling period and MCA(t) is the multichannel value of the current sampling period.

13. The method of claim 1, 2, 3 or 4 wherein the derivative of the post-filtered multichannel value is used to determine whether the threshold has been exceeded and whether the mode has been met.

14. An apparatus for detecting an endpoint at which to terminate a chemical process involving one or more chemical species, the process taking place in a chamber, said apparatus comprising:

means for defining the endpoint in the form of a mode, a minimum value of a delay count and a threshold;

means for initiating the chemical process;

means for detecting intensity of light emitted during the chemical process by one or more of the one or more species in the form of an analog signal representing the intensity of each of the detected species;

means for sampling each analog signal at a rate and at the beginning of a current sampling period defined by the rate to produce a set of raw data samples for the current sampling period;

means for pre-filtering each of the set of raw data samples of the current sampling period;

means for normalizing each of the set of pre-filtered data samples;

means for auto-correlating the set of normalized data samples to produce a multichannel value for the current sampling period;

means for post-filtering the multichannel value;

means for incrementing the delay count when the threshold has been exceeded by the post-filtered multichannel value and the mode is met, otherwise setting the delay count equal to zero; and

means for terminating the process when the delay count reaches or exceeds the minimum value.

15. The apparatus of claim 14 wherein the mode is met when the post-filtered multichannel value of an immediately preceding sampling period is less than the post-filtered multichannel value of the current sampling period.

16. The apparatus of claim 14 wherein the mode is met when the post-filtered multichannel value of an immediately preceding sampling period is greater than the post-filtered multichannel value of the current sampling period.

17. The apparatus of claim 14 wherein the mode is met when the post-filtered multichannel value of the current sampling period is less than the threshold and is less than the post-filtered multichannel value of an immediately preceding sampling period.

18. The apparatus of claim 14 wherein said means for normalizing each of the set of pre-filtered data samples further comprises:

means for calculating an average normalization value corresponding to each of the sampled analog signals; and

means for dividing each of the set of pre-filtered data samples by the average normalization value corresponding to the analog signal from which each of the pre-filtered samples was taken.

19. The apparatus of claim 18 wherein said means for normalizing does not normalize the pre-filtered data samples for a first period of time with reference to a time at which the process is initiated.

20. The apparatus of claim 19 wherein said means for calculating calculates the average normalization values during a second period of time with reference to lapse of the first period of time.

21. The apparatus of claim 20 wherein said means for calculating further comprises:

means for summing together pre-filtered data samples taken from each of the sampled analog signals; and

means for dividing each of the summations by a total number of samples taken during the second period of time to produce the average normalization values corresponding to each of the sampled analog signals.

22. The apparatus of claim 14 wherein said means for sampling each analog signal further comprises:

means for capturing a raw analog data sample of the amplitude of each analog signal; and

means for converting each of the raw analog samples to a raw digital data sample.

23. The aparatus of claim 14 wherein said means for pre-filtering each of the raw data samples further comprises means for calculating the pre-filtered data samples in accordance with the equation Df1(t)=(1-Q1)×

Df1(t-1)+Q1×

D1(t), wherein Df1(t) is the pre-filtered data sample being currently calculated, Q1 is a pre-filter constant, Df1(t-1)is the pre-filtered data sample calculated during a preceding sampling period.

24. The apparatus of claim 14 wherein said means for auto-correlating further comprises means for calculating the multichannel value in accordance with the equation MCA(t)= (A-B)+(C-D)!÷

(E), where A, B, C, D and E represent up to five of the normalized data samples, and MCA(t) is the multichannel value being calculated.

25. The apparatus of claim 14 wherein said means for post-filtering the multichannel value further comprises means for calculating the post-filtered multichannel value in accordance with the equation Df2(t)=(1-Q2)×

Df2(t-1)+Q2×

MCA(t), wherein Df2(t) is the post-filtered multichannel value of the current sampling period, Q2 is a post-filtering constant, Df2(t-1) is the post-filtered multichannel value of a preceding sampling period and MCA(t) is the multichannel value of the current sampling period.

26. The apparatus of claim 14, 15, 16 or 17 wherein the derivative of the post-filtered multichannel value is used to determine whether the threshold has been exceeded and whether the mode has been met.