JPH08148474A - Dry etching end point detection method and apparatus - Google Patents

Abstract

(57) [Summary] [Object] To provide a dry etching end point detection method capable of automatically setting a desired etching end point to perform highly reliable etching processing and automating the etching processing. [Structure] In an end point detection method using a single-wafer dry etching apparatus for observing the emission intensity of plasma, the data of the emission intensity change of the first etching process and the real-time emission intensity change are output to an image output device. The end set time Ts of the second and subsequent etching processes is calculated based on the delay time up to (steps S1 to S5),
Based on this end set time Ts, the second and subsequent etching processes are ended or the etching conditions are changed (steps S6 to S8).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting an etching end point of a single wafer type dry etching apparatus used in a semiconductor manufacturing process.

[0002]

2. Description of the Related Art As a method for detecting the end point of etching in a dry etching apparatus utilizing plasma,
A method of observing a specified emission wavelength from plasma and utilizing the intensity change of the emission wavelength for end point detection is widely used.

Besides the method of observing the plasma emission wavelength, a method of observing the change of the film thickness to be etched by utilizing the interference of laser light has been partially put into practical use. However, the spot diameter of laser light is too large. At present, it is impossible to keep up with the miniaturization of the processing size of the LSI, and the restrictions on the material to be etched are large.

On the other hand, in the etching end point detecting method for observing the intensity change of the plasma emission wavelength, the emission wavelength to be observed is appropriately selected depending on the etching gas and the material to be etched, regardless of the type of the etching gas or the material to be etched. It has a wide range of features. Also, as a method for determining the etching end point, many methods such as a method using a second derivative method based on a change in light emission intensity have been proposed.

In any of the methods currently proposed in the method of utilizing the intensity change of the emission wavelength of the plasma for detecting the etching end point, the control after the intensity change of the emission wavelength occurs is a premise. .

[0006]

However, after the intensity change of the emission wavelength has occurred, the controllability is insufficient in the recent etching of VLSI, which has progressed in the miniaturization and thinning and the large diameter of the wafer substrate. Is becoming.

For example, the etching rate of polysilicon by a single-wafer etching apparatus for sequentially processing semiconductor substrates (wafers) one by one is about 300 nm per minute, and the etching rate of the underlying gate oxide film at this time is every minute. 30n
m. Therefore, in this case, the thickness of the polysilicon film should be 5 depending on the etching condition that the etching selectivity of the oxide film against polysilicon during etching is 10 and the uniformity of the etching rate is ± 5%.
If 0 nm of polysilicon is etched to stop the etching on the underlying oxide film having a thickness of 10 nm, the underlying oxide film starts to be exposed in only 9.5 seconds from the start of etching.

However, the emission wavelength of the reaction product SiClx hardly changes at this point. The reason is that the polysilicon film of the material to be etched is not etched on the wafer substrate at this point, and most of it remains, and the change in the emission wavelength is processed by a computer to perform an image output device (CRT) operation. ) It takes time to output. Therefore, the change of the emission wavelength output to the CRT occurs several seconds after the underlying oxide film is exposed on the surface, and the etching is surely progressed in the underlying oxide film within these several seconds. .

As a practical solution for solving the above problems, the etching time is set in advance, and the etching process is executed based on this set time.
The current situation is that almost no automation is performed. This tendency is expected to become a serious problem in the near future, when the etching speed is further increased due to the increase in the diameter and the number of single wafers, and the miniaturization and thinning of the VLSI are further advanced.

The present invention has been made in view of the above-mentioned drawbacks of the prior art, and a dry etching capable of automatically setting a desired etching end point to perform a highly reliable etching process and automating the etching process. The purpose of the present invention is to provide an endpoint detection method.

[0011]

In order to achieve the above object, the present invention provides an end point detection method using a single-wafer dry etching apparatus for observing the emission intensity of plasma,
Data of the light emission intensity change of the first etching process,
The end set time of the second and subsequent etching processes is calculated based on the delay time from the real-time change of the emission intensity to the output to the image output device, and the second and subsequent etching processes are performed based on this end set time. Provided is a method for detecting the end point of dry etching, which is characterized by changing the termination or etching conditions.

In the preferred embodiment, from the data of the change in the emission intensity of the first etching process, the time T1 of the decay start point of the emission intensity, the decay end time T2, and the etching end point of the portion where the etching rate is average are obtained. Of time T3 is calculated, and based on these times T1, T2, T3 and the delay time Tx, the end set time Ts of the second and subsequent etching processes is Ts = T3-{(T3 × U / 100) + Tx } However, the feature is that U is calculated as the uniformity of the etching rate.

In another preferred embodiment, the time of the emission start point of the emission intensity in the emission intensity change data of the first etching process is T1, and the delay time is T.
It is characterized in that the end set time Ts of the second and subsequent etching processes is calculated as Ts = T1-Tx as x.

In the present invention, further, the plasma emission spectrum detecting means, the emission spectrum data processing means by the detecting means, the image output device for displaying the intensity change of the emission spectrum by the data processing means, and the data processing means. In the dry etching end point detecting device having an etching control means for performing an etching process based on the calculation processing result of, the data processing means includes the data of the intensity change of the emission spectrum and the real-time emission intensity change to obtain the image. Provided is a dry etching end point detection device, which is configured to calculate an end set time of an etching process based on a delay time until output to an output device.

[0015]

The first etching process is executed to the end to obtain the data of the change in the emission spectrum, and the output attenuation start time point, the attenuation end time point and the average intermediate time point are calculated based on this data, and this data output In consideration of the delay time and the uniformity of the etching process, the end set time for stopping the etching process (or changing the conditions) is calculated. For the second and subsequent etching processes, the end set time obtained by this calculation is applied, and the etching is stopped (or the condition is changed) when the end set time is reached from the etching start time.

In other words, to explain the mode of operation of the present invention, an approximate etching rate is calculated from the data of the emission intensity change of the first material to be etched (eg, semiconductor wafer), and the etching rate In consideration of uniformity, the shortest time from the etching start time to the time when the base is first exposed is calculated. Next, the difference between the shortest time until the base is exposed and the delay time until the emission intensity change is displayed on the image output device (for example, CRT) is calculated, and this time is used as the etching end time or the etching condition change time as the second time. It is applied to the subsequent wafer processing. By incorporating the above calculation function and wafer etching control function into the control program of the etching apparatus,
The series of etching processes can be fully automated.

If the etching is finished at the time when the delay time is subtracted from the shortest time of exposing the underlayer by the above-mentioned etching, that is, the time until the underlayer is exposed in a portion where the etching is fast, the etching is completed. Before the exposure of, that is, the etching is finished in the under-etch state where the material to be etched is completely left,
It is possible to prevent etching of the base.

In actual etching, a just etching step of mainly etching the material to be etched and an overetching step of etching the remaining portion of the material to be etched while ensuring a high underlayer selectivity are performed. Therefore, one embodiment of the present invention There is no problem even if the just etching step is finished in the under-etched state in which the material to be etched remains.

Further, by incorporating the arithmetic function into the control program of the etching apparatus, the etching time of the just etching process, which has been manually set in the past, is obtained by arithmetic operation based on the processing result of the first wafer, and this is calculated. The etching process can be automatically performed by applying it to the second and subsequent wafers.

[0020]

1 is a partial cross-sectional view of a semiconductor wafer to which an end point detecting method according to a first embodiment of the present invention is applied. In the present embodiment, in processing the gate electrode made of polysilicon, etching is performed with a mixed gas of Cl ₂ / O ₂ , and the emission wavelength 391 of SiClx is generated during the etching.
nm was observed with an emission spectrometer. As shown in FIG. 1, the structure of the etching sample is such that a polysilicon film 3 having a thickness of about 100 nm is laminated on a Si (silicon) substrate 1 with a gate oxide film 2 having a thickness of about 10 nm, and further on top of that. The wafer is formed of a resist mask 4 having a predetermined pattern and a thickness of about 1.2 μm.
The gate oxide film 2 was formed by thermally oxidizing the silicon substrate 1 using a batch thermal oxidation furnace. Further, the polysilicon film 3 is formed of SiH by LPCVD (Low Pressure CVD) method.
₄ and PH ₃ were used to form a film. The resist mask 4 was formed as a fine pattern mask using a positive type chemically amplified resist material and a KrF excimer laser stepper.

Next, the above wafer was set in a single-wafer type RF bias application type magnetic field microwave plasma etching apparatus, and was etched by a mixed gas of Cl ₂ / O ₂ .

Simultaneously with the start of etching, a reaction product Si generated during the etching was measured by an emission spectrum meter.
The intensity change of the emission spectrum 391 nm of Clx was observed. FIG. 4 is an observation graph of the intensity change of this emission spectrum, in which the horizontal axis represents time T and the vertical axis represents emission intensity I. FIG.
As shown in, the emission intensity of SiClx changes from approximately α1 until the etching starts (T = 0) and immediately before the underlying gate oxide film 2 is exposed.

FIG. 2 shows a wafer cross section at a time point (time T = T1) when the spectral intensity of the emission intensity α1 starts to decrease. At this time T = T1, the etching of the polysilicon film 3 is completed at the portion where the etching rate of the polysilicon 3 on the wafer is high, and the exposed surface 5 of the underlying gate oxide film 2 appears (FIG. 2 (B)). On the other hand, at this time T = T1, as shown in FIG. 2A, the polysilicon film 3 still remains in the portion where etching is slow.

When the etching is further continued, the emission spectrum intensity continues to decrease, and when the polysilicon film 3 is completely etched and removed, the attenuation ends and the emission intensity becomes almost constant α2. The time at this point is T = T2 (FIG. 4). A cross section of the wafer at this time T2 is shown in FIG.

As shown in FIG. 3, when the polysilicon film 3 on the wafer is completely etched (that is, FIG. 3A).
When the polysilicon film 3 is completely removed and the underlying gate oxide film 2 is exposed in the slow etching portion, the reaction product SiClx emission intensity decreases to α2, and the etching of the polysilicon film 3 is completed. The change becomes very small after the time T2. In this case, from time T1 to T2
In the part where the etching rate is fast,
As shown in, the gate oxide film 2 is continuously etched,
A so-called over-etched state occurs, and an etching loss 6 of the oxide film 2 occurs.

However, in actual etching, T
Even if the etching is completed at the time point of 2, the etching progresses faster than the display of the change in the emission intensity, the etching cannot be stopped by the underlying gate oxide film 2, and it often reaches the silicon substrate 1. It was It is considered that this is because the time required for signal processing and calculation for capturing the emission spectrum with a detector and displaying it on an output device such as a CRT is included in the etching time as a delay time.

That is, if the change in the emission intensity and the etching amount match in real time, the change in the emission intensity can be used as the etching end point detecting means, but as described above, the delay time is included in the change in the emission intensity. This change in emission intensity cannot be used as it is as a method for detecting an end point. In this embodiment, this point is improved as follows.

FIG. 9 is a graph showing the intensity change of the emission spectrum displayed on the output device such as a CRT, and FIG. 10 is an explanatory diagram showing the intensity change incorporating the delay time. FIG. 11 is a flow chart showing the procedure of the embodiment of the method of the present invention.

First, as described above, etching of the first wafer is started (step S1). The etching process of the first wafer is completely performed until the end, and the emission spectrum of the reaction product SiClx during the etching process is detected and recorded as shown in FIG. 9 (step S
2). This intensity change of the emission spectrum may be recorded as a graph display and stored in a memory means such as a RAM as data. As shown in FIG. 3, the polysilicon film 3 as the material to be etched is completely etched and removed so that the gate oxide film 2 is exposed in the entire etching region. Step S3). Next, the data processing of the intensity change of the emission spectrum of the first sheet is performed as follows.

First, in step S4, the emission intensity at time T = T1 when dI / dT <0 from the change in emission intensity of the wafer, that is, the emission intensity I of the emission spectrum starts to decay from a constant intensity after etching is started. α1,
Each data is obtained with the emission intensity α2 at the time when dI / dT = 0, that is, the time T = T2 when the etching is completely finished and there is no change in emission intensity. The time T1 represents the time when the base is first exposed in the portion where the etching is the fastest (FIG. 2B), and the time T2 is the time when the base is exposed in the portion where the etching is the slowest (FIG. 3).
(A)). Further, the time T3 when the etching of the polysilicon film 3 is completed and the underlying gate oxide film 2 is exposed at the portion where the etching rate is average is determined. For this time T3, the emission intensity at the end of etching in the portion where the etching rate is average is calculated as α3 = (α1−α2) / 2, and the time point of this emission intensity α3 is obtained from the data.

Next, at step S5, the time T3
And etching rate uniformity U%, and the aforementioned delay time T
Based on x, the set time Ts for exposing the underlying gate oxide film is calculated as Ts = T3-{(T3 * U / 100) + Tx}. Here, the uniformity U% uses data measured and calculated in advance by a step gauge or a film thickness measuring device. The delay time Tx may be set empirically as, for example, 1 to 2 seconds, or the etching is finished at an appropriate time point (for example, T2) displayed, and the overetched film thickness is measured here. By doing so, the delay time can be calculated based on the etching amount which is known in advance.

Next, when etching the second and subsequent wafers, the set time Ts obtained by the above calculation is used.
That is, in step S6, the nth etching (n is an integer of 2 or more) is started, and after the start, the set time Ts is set.
Etching is completed when the temperature reaches (step S
7). As a result, the etching control of the polysilicon film 3 on the underlying gate oxide film 2 is performed with a desired high accuracy in consideration of the delay time. When the etching is finished at time Ts, the etching may be stopped completely, or the etching conditions such as the selection ratio with respect to the underlying layer may be changed. Further, by appropriately selecting the delay time Tx, the etching end time can be arbitrarily set to any of a high etching rate portion, a low etching rate portion and an average portion.

The end set time Ts is applied to all the second and subsequent wafers, and the etching process is repeated until the end of all the wafers (step S8).

In order to easily calculate the set time Ts for the end of etching (or change of etching conditions), even if Ts = T1−Tx, high accuracy is obtained when the etching uniformity is ± 5% or less. Etching control is achieved.

FIG. 5 is a partial cross-sectional view of a semiconductor wafer to which the end point detecting method according to the second embodiment of the present invention is applied.
In this example, contact hole processing for etching the oxide film was performed with a mixed gas of CHF3 / CF4 / Ar, and an emission spectrum of the reaction product CO during etching, 483.5 nm, was observed.

The structure of the etching sample has a thickness of about 500 nm on a Si (silicon) substrate 7, as shown in FIG.
Oxide film 8 is formed by a CVD apparatus, and a resist mask 9 having a predetermined pattern and a thickness of about 1.2 μm is formed thereon.
It is composed of a wafer on which is formed. The resist mask 9 was formed as a fine pattern mask using a negative chemically amplified resist material and a KrF excimer laser stepper.

Next, the above wafer was set in a single-wafer type magnetron etching apparatus, and was etched with a mixed gas of CHF3 / CF4 / Ar.

At the same time when the etching is started, the reaction product CO generated during the etching is measured by the emission spectrum meter.
The intensity change of the emission spectrum of 483.5 nm was observed. FIG. 8 is an observation graph of the intensity change of this emission spectrum, in which the horizontal axis represents time t and the vertical axis represents emission intensity i. FIG.
As shown in, the emission intensity of CO changes from about β1 from the etching start time (t = 0) to immediately before the underlying silicon substrate 7 is exposed.

FIG. 6 shows a wafer cross section at a time point (time t = t1) when the spectral intensity of the emission intensity β1 starts to decrease. At this time t = t1, the etching of the oxide film 8 is completed at the portion where the etching rate of the oxide film 8 on the wafer is high, and the exposed surface 10 of the underlying silicon substrate 7 appears (FIG. 6).
(B)). On the other hand, at the time of this time t = t1, FIG.
As shown in (A), the oxide film 8 still remains in the portion where etching is slow.

When the etching is further continued, the emission spectrum intensity continues to decrease, and when the oxide film 8 is completely etched and removed, the attenuation ends and the emission intensity becomes substantially constant β2. The time at this point is t = t2 (FIG. 8). A cross section of the wafer at this time t2 is shown in FIG.

As shown in FIG. 7, when the oxide film 8 on the wafer is completely etched (that is, as shown in FIG. 7A), the oxide film 8 is completely removed at the portion where the etching is slow and the base layer is removed. When the exposed surface 10 of the silicon substrate 7 appears, the emission intensity of the reaction product CO decreases to β2, and the change becomes very small when the etching end time t2 of the oxide film 8 is the boundary. In this case, in the portion where the etching rate is high from time t1 to t2, as shown in FIG. 7 (B), the silicon substrate 7 is continuously etched and becomes a so-called over-etched state, and an etching loss 11 of the underlying silicon occurs. .

However, in the actual etching, t
Even if the etching is completed at the time point of 2, the etching progresses faster than the display of the change in the emission intensity, and the etching cannot be stopped at the shallow portion of the underlying silicon substrate 7, and the silicon substrate 7 is so affected as to affect the quality and characteristics. It often happened to reach deep inside. This is because the emission spectrum is captured by the detector and the CRT is used as described above.
It is considered that this is because the time required for signal processing and calculation displayed on the output device such as is taken into the etching time as a delay time.

That is, if the change in the emission intensity and the etching amount match in real time, the change in the emission intensity can be used as the etching end point detecting means, but as described above, the delay time is included in the change in the emission intensity. This change in emission intensity cannot be used as it is as a method for detecting an end point. The present embodiment improves on this point as in the case of the first embodiment described above.

That is, when the first wafer is etched, the reaction product C is removed without stopping the etching.
The emission intensity of O is recorded as shown in FIG. From the change in the emission intensity of the wafer, the time for di / dt <0 is t
The emission intensity at each time point is defined as β1 and β2, where t2 is the time for which 1 and di / dt = 0. Next, the emission intensity β3 at the end of the etching of the film to be etched (oxide film 8) in the portion where the etching rate is average is β3 =
It is calculated as (β1-β2) / 2, and the time point is defined as t3. Next, this t3, uniformity U%, and delay time tx
Therefore, similarly to the first embodiment, the time ts for exposing the underlying silicon substrate 7 is set as ts = t3-{(t3 * U / 100) + tx}. The method of obtaining the uniformity U and the delay time ts is the same as in the case of the first embodiment. By etching the second and subsequent wafers using such set time ts, the etching of the oxide film 8 can be controlled with high accuracy as required without deeply etching the underlying silicon substrate 7. . As in the first embodiment, ts is used as a simple calculation method of the set time ts.
= T1-tx may be used for the calculation.

[0045]

As described above, in the present invention, since the end point setting time of the second and subsequent wafers is calculated based on the measured data of the first wafer, the emission spectrum intensity change is output. The etching can be stopped or the conditions can be changed at a desired point in consideration of the delay time until it is displayed on the equipment, and highly accurate and reliable etching control can be performed, and the second and subsequent wafers can be automatically controlled. The etching process becomes possible.

Further, if a program for executing the calculation of the end point setting time according to a series of sequences is incorporated in the etching apparatus, the automation of the etching process can be achieved completely.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a wafer according to a first embodiment of the method of the present invention.

FIG. 2 is an explanatory diagram of an etching state at the time when the emission spectrum intensity starts to attenuate in the embodiment of FIG.

FIG. 3 is an explanatory diagram of an etching state at the time point when the emission spectrum intensity is attenuated and becomes substantially constant in the embodiment of FIG.

FIG. 4 is a graph showing changes in emission spectrum intensity in the example of FIG.

FIG. 5 is a configuration diagram of a wafer according to a second embodiment of the method of the present invention.

6 is an explanatory diagram of an etching state at the time when the emission spectrum intensity starts to attenuate in the embodiment of FIG.

FIG. 7 is an explanatory diagram of an etching state at the time point when the emission spectrum intensity is attenuated and becomes substantially constant in the embodiment of FIG.

FIG. 8 is a graph showing changes in emission spectrum intensity in the example of FIG.

FIG. 9 is a detection display graph of intensity change of emission spectrum.

FIG. 10 is a graph for comparing and explaining an actual change in emission intensity and a change in emission intensity on a display.

FIG. 11 is a flowchart of an etching process according to the present invention.

[Explanation of symbols]

1: Silicon substrate, 2: Gate oxide film, 3: Polysilicon film, 4: Resist mask, 5: Exposed surface of gate oxide film, 6: Etching loss of gate oxide film, 7: Silicon substrate, 8: Oxide film, 9: resist mask, 10: exposed surface of silicon substrate, 11: etching loss of silicon substrate.

Claims (4)

[Claims] 1. In an end point detection method using a single-wafer dry etching apparatus for observing the emission intensity of plasma, data of emission intensity change of the first etching process and real-time emission intensity change are used as an image output device. The end setting time of the second and subsequent etching processes is calculated based on the delay time until output, and the second and subsequent etching processes are ended or the etching conditions are changed based on this end setting time. And a method for detecting the end point of dry etching. 2. From the data of the emission intensity change of the first etching process, the time T1 of the decay start point of the emission intensity, the decay end time T2, and the time T3 of the etching end point of an average etching rate are calculated. Calculated, and based on these times T1, T2, T3 and the delay time Tx, the end set time Ts of the second and subsequent etching processes.
Ts = T3-{(T3 * U / 100) + Tx}, where U is the uniformity of the etching rate, and the arithmetic processing is performed. 3. The time of the emission start point of the emission intensity in the emission intensity change data of the first etching process is T1.
2. The method for detecting an end point of dry etching according to claim 1, wherein the delay time is set to Tx, and an end set time Ts of the second and subsequent etching processes is set to Ts = T1-Tx. 4. A means for detecting a plasma emission spectrum,
Data processing means of the emission spectrum by the detecting means, an image output device for displaying the intensity change of the emission spectrum by the data processing means, and etching control means for performing etching processing based on the calculation processing result of the data processing means. In the dry etching end point detection device having the above, the data processing means performs the etching process based on the data of the intensity change of the emission spectrum and the delay time from the real-time emission intensity change to the output to the image output device. An end point detection device for dry etching, which is configured to calculate a set end time.