JPH03263829A - Method of end-point judgement - Google Patents

Abstract

PURPOSE:To judge the end point of an etching operation surely even when an etching rate is changed with an increase in the number of specimens to be treated by a method wherein the ratio of the end-point judgment time of a 1st specimen to the end-point judgment time of an nth specimen is computed and a constant corrected by multiplying the ratio is used to judge the end point of a treatment of an (n+1)th specimen. CONSTITUTION:In an end-point judgment method of a dry-etching operation by an emission spectroscopic method, the ratio tn/t1 of the end-point judgment time t1 of a 1st specimen to the end-point judgment time tn of an nth specimen is computed; a constant corrected by multiplying a constant in terms of time regarding the judgment of all end points of an end-point judgment sampling interval S, a dead time td, a difference interval and the like by said ratio tn/t1 is used to judge the end point of the treatment of an (n+1)th specimen. For example, when said ratio tn/t1 becomes a certain value or higher or lower, an alarm or a maintenance display is given by the instruction from a computer. Thereby, the inside of an etching treatment chamber is cleaned, and the maintenance of an apparatus is executed. The maintenance display at this time may be an alarm by using a lamp or the like or may be displayed on a CRT.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Minute Hand] The present invention relates to a method for determining the end point of dry etching using plasma.

[Conventional technology]

The conventional endpoint determination method is to change the gain offset of the amplifier that amplifies the plasma emission amount converted from one short signal as the number of processed sheets increases, and to adjust the endpoint determination dead time, sampling time, etc. The elements were fixed. This is because the change in the amount of plasma light emitted over time is caused by the window that determines the amount of light emitted being clouded by the debossing film in the etching chamber.
This is because it was thought that the cause was changes in the optical system over time.

However, in reality, there is a phenomenon in which the etch/rate decreases due to changes over time due to deposition and the like, and in such cases, the conventional method may not be able to accurately determine the final A.

Regarding this type of information, for example, Japanese Patent Application Laid-open No. 6
3-254732 is mentioned.

[Problem to be solved by the invention]

As mentioned above, the conventional technology has eight problems in that it cannot deal with changes in the amount of light in the five emission spectra in the time axis direction, which is caused by the phenomenon that the etch/rate decreases due to an increase in the number of processed sheets.

For example, in an apparatus shown in Figure 2, which performs etching by generating plasma in a quartz chamber through the action of a microwave electric field and a magnetic field, if a process with many deposits is used, deposits may adhere to the quartz chamber. However, as the number of processed sheets increases, the amount of light emission decreases. Incidentally, if this is all that is needed, it can be corrected using the auto-gain function, but in reality, the passage of microwaves becomes poor due to the deboss, and the etching rate also changes, or in this case, decreases. Therefore, as shown in FIGS. 4(a) to 5(1), the rate of change of the waveform near the end point also changes gradually as the rate decreases. In addition, Figure 4 shows the processing of the first sheet. @5 Figure shows the processing of the n-th sheet. With auto gain, during the first change (Vl) and during the nth change (Vl)
Vn) are adjusted so that they are equal, but the time required for change changes (T1-Tn), so d2V/dt2(
(secondary difference) becomes smaller. Therefore, the only way to correct this is to reduce the decision position or increase the difference interval.

Furthermore, recent wafers are complex, and multilayer films are often etched by -degree etching. The emission intensity waveform in such a case is, for example, as shown in FIG. 6(a). In order to detect just etching of the second layer with such a waveform, etching of the 2M1st layer is started,
It is necessary to provide dead time where the waveform is stable.

However, if the etch rate decreases and the dead time is always constant, the dead time will occur during the etching of the first layer, as shown in Figure 6(b), and the dead time will occur at the boundary between the first and second layers. There is a risk of detecting the end point.

SUMMARY OF THE INVENTION An object of the present invention is to provide an end point determination method that can reliably determine the end point of etching even when the etching rate changes as the number of processing branches of a sample increases.

[Means for solving 111m] In order to achieve the above purpose, the ratio of the end point judgment time for the first sample and the end point judgment time for the nth sample is calculated, and the end point judgment sampling interval, 7-' rad time, is calculated. .

A constant corrected by multiplying the time constants related to all end point determinations such as differential intervals by the ratio is used for end point determination of the processing of the (n+1)th image.

In addition, in order to achieve the above purpose, the interval between the peak interval of the second difference value of the luminescence amount when processing the first sample and the peak interval of the second difference value of the luminescence amount when processing the nth sample is This method calculates the ratio by comparing the times, corrects all the time constants related to end point determination such as the end point determination sampling interval, dead time, and difference interval using the ratio, and performs the end point determination for processing the n+1th sheet. This is what I did.

[For production]

Now, the just etch time for the first wafer is tl, and the just etch time for the nth wafer is t.

Assuming that the plasma emission spectrum waveforms of the 1&month and nth sheets have similar shapes expanded and contracted in the time axis direction, the ratio between t1 and tfl can be considered to be the ratio of etch rates.

In the present invention, based on this idea, the above t] and 1
Each time constant for end point determination is corrected using the ratio of n, 1n/fl. The time constant is the sappling interval for end point determination,
Refers to dead time, differential interval, etc.

In determining the end point of the n+11th sheet, the above F! of the first sheet!
For the constant between #, use something like Lx/in t" 14-↓.

By doing this, it is possible to perform correction in the rIll axis direction when determining the end point, and even when the etching rate decreases as the number of processed sheets increases, it is possible to accurately determine the etching end point.

〔Example〕

An embodiment of the present invention will be described below with reference to FIGS. 1 and 2.

FIG. 2 shows the end point determination system. In this case, etching gas is supplied to the inside of the etching processing chamber 1, which is evacuated to a predetermined pressure, and the electric field of the microwave and M
Plasma is generated by interaction with the i-field. The emission spectrum of the plasma generated during etching of the wafer 2 in the etching processing chamber 1 passes through the spectrometer 3 and is guided to the photoelectric conversion element 4 only at specific wavelengths. The amount of emission spectrum converted into an electrical signal by the photoelectric conversion element 4 is amplified by the amplifier 5 and input to the combinator 6. Hereinafter, all processing is performed by software within the computer 6.

FIG. 1 shows a diagram of the principle of the present invention. Figure 1 (a).

(b) and (C) are the 1st and nth processed sheets, respectively.

n + is the amount of emission spectrum of the first sheet.

Now, as shown in Figure (a), the first wafer is sampled at the sampling interval S1. It is assumed that the end point is determined at dead time ta1 and the end point is detected at tl. Also, as shown in Figure (b), for the n-th wafer, 1. Suppose that the end point is detected at . This h and t. It is considered that the ratio of etching rates for the first and nth sheets is the same.

Next, when etching the n+1th sheet, the above ratio is applied to all time constants such as the sampling interval, dead time, and although not shown in the figure, the differential interval.
Use the product multiplied by n/lx.

In other words, as shown in Figure (C), the sampling interval is s
n+1 := t(1/ tl X 81, dead time is 'anal=tn/h X tBH.% However, since there is no data to correct for the second sheet, it is assumed to be the same as the first sheet.

With the above method, by correcting constants related to various times for end point determination, fluctuations in the emission spectrum in the time axis direction can be coped with.

In addition, in this embodiment, the method of correction when determining the etching end point was described, but the above ratio [n/'h
When the value exceeds or falls below a certain value, an alarm or maintenance display may be issued according to an instruction from the computer, so that the interior of the etching processing chamber 1 can be cleaned and maintenance of the apparatus can be performed. In addition, the maintenance display at this time may be a warning such as an error message, or may be displayed on a CRT.

Further, as a method of notifying the maintenance time, the above ratio may be displayed in a graph on 0)'LT to inform the user of the trend and perform maintenance.

In addition, a dummy wafer is installed in the etching equipment.
When the computer detects that it is time for maintenance, the signal is sent to the host computer, which manages the wafer processing and automatically starts processing the wafer at the right time, such as when the loft changes. It is also possible to cancel the etching process, carry the dummy wafer into the etching process chamber, perform plasma cleaning, and then return to the normal 0-process etching process. When the etching gas is restarted, the time constant for reaching the etching end point is returned to the constant for the first sheet.

In addition, when correcting the time constant, it is possible to keep the reference time constant and correct it by increasing the counting time, or as shown in Figure 3, the end point judgment The clock cycle of the computer that performs this may be changed.

In this case, 6 is a microcomputer that performs end point determination, and 7 is a clock 8 that serves as a reference for end point determination.
9 is a clock control circuit that adjusts the period of the fire, and 9 is a real time counter that performs time management.

Furthermore, in this embodiment, the ratio is determined by comparing the time until just etch, but as shown in FIG. 4(b).
The peak f! of the secondary difference value of the amount of lightning during processing of the first wafer shown in ]Vi3t]' and the peak li5 interval tn' of the second-order difference value of the luminescence amount when processing the n-th image shown in FIG. 5(b).
Even if the ratio is determined by comparing the interval time with
...Photoelectric conversion element, 5...Amplifier, and 2 can be made.

6...Computer [Effects of the Invention] According to the present invention, it is possible to correct fluctuations in the emission spectrum in the temporal direction, and it is possible to correct the fluctuations in the emission spectrum in the temporal direction, and to cope with the processing due to the increase in the number of processed sheets, the reduction in the processing rate, etc. It becomes possible to determine the end point.

[Brief explanation of drawings]

Claims (1)

[Claims] 1. In a method for determining the end point of dry etching using emission spectroscopy, the ratio between the end point determination time for the first sample and the end point determination time for the nth sample is calculated, and the end point determination sampling interval and the dead point determination time are calculated. The constant corrected by multiplying the time constants related to end point determination such as time and difference interval by the above ratio is n+1
An end point determination method characterized in that it is used to determine the end point of processing for a second sheet. 2. The end point determination method according to claim 1, wherein an alarm or maintenance display is displayed when the ratio becomes above or below a certain value. 3. The end point determining method according to claim 1, wherein the ratio is displayed in a graph on a CRT to inform the user of the trend and to inform the user of maintenance timing. 4. The endpoint determination method according to claim 1, wherein the time constant is determined by changing a clock cycle of a microcomputer that performs endpoint determination. 5. In the method of determining the end point of dry etching using emission spectroscopy, the peak interval of the secondary difference value of the luminescence amount when processing the first sample and the peak of the secondary difference value of the luminescence amount when processing the nth sample The ratio is calculated by comparing the interval time with the interval, and the end point judgment sampling interval, dead time,
An end point determination method characterized by correcting time constants related to all end point determinations, such as difference intervals, and determining the end point of processing the (n+1)th image.