JPH01152286A - Method for monitoring final point trace of plasma etching reactor - Google Patents

Abstract

PURPOSE: To enable a method for detecting the abnormality during a production process operation and ameliorates the same by analyzing the characteristic of the candidate region of an actual end point trace over the entire part of the characteristic of the region of a reference end point trace to be matched.

CONSTITUTION: The reference end point trace(EPT) is first regulated. This reference EPT is divided to plural regions and the region of the actual EPT is matched. The characteristic of the desired matching region of the actual EPT and the characteristic of the region of the reference EPT to be matched are compared. As a result, whether the abnormality occurs at what time during the etching operation may be verified by this system. The regions are analyzed in details and the process specification rule is applied, by which the estimation of the case to the verified abnormality is made possible by this system.

COPYRIGHT: (C)1989,JPO

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of the Invention] The present invention relates to automated quality control and process operation techniques, and more particularly to an improved method and apparatus for detecting anomalies during manufacturing process operations. It is.

[Prior art and its problems]

In the manufacture of semiconductor devices, a large number of steps are combined in the process of each individual semiconductor device. During the manufacture of such semiconductor devices, various attempts have been made to establish quality control inspections in various processes. Typically, these systems for quality control checks include methods for physically inspecting or testing selected specimens of these semiconductor devices at various stages during the manufacturing process. . However, such systems are cumbersome and relatively expensive. Moreover, in many cases these systems are not able to guarantee the quality of each manufactured device because they are applied only to devices that are arbitrarily selected from the manufacturing process. Furthermore, previously developed systems operate to detect problems in the manufacturing process at a point after an error has already occurred. As a result, the same error may repeatedly occur during a particular process operation for a large number of semiconductor devices before the error is detected in a subsequent quality control inspection. As is clear from an economical point of view, there is a need for a manufacturing method that detects abnormalities during the manufacturing process as soon as possible after these abnormalities occur. Furthermore, it is desirable that these abnormalities can be detected in real time. In other words, it is desirable to be able to detect these abnormalities when they occur. By utilizing such instantaneous detection, it is possible to avoid repeating the same process abnormality on semiconductor devices during manufacturing.

A common process operation in the manufacture of semiconductor devices is plasma etching. In this plasma etching, a semiconductor device in the form of a "slice" (commonly called "slice") is placed in an etching chamber by supplying RF (radio frequency) power under a specific gas and a predetermined pressure and temperature. let During the etching process, certain materials on the surface of the semiconductor slice react with the gas in the chamber and then become evaporated from the slice surface. A typical etching process for a slice is one that lasts up to a minute or up to several minutes or more.

A typical conventional etching reactor is equipped with equipment for monitoring certain aspects of the etching process. These reactors are equipped with equipment to perform on-line hardware monitoring, such as equipment to monitor reactor temperature and pressure, RF power supply power, and feed gas flow rates.

Conventional etching apparatus also include devices designed to detect the end point of this etching operation. Solid state technology (5solid 5tate Te) disclosed herein by reference.
chnology), April 1981, No. 115-1
“Method for Detecting the End Point of Plasma Etching” (Paul J, Marcus and Pang, p. 22).
Several methods have been proposed for such devices, according to Don Foo (author). These proposed methods include emission spectroscopy, optical reflectance, mass spectrometry, impedance monitoring, Langmuir probe monitoring, and pressure monitoring methods.

No matter which monitoring method is used, the end point monitoring device
It serves to detect the completion of the desired etching reaction, so that the etcher can signal that the etching cycle is complete and that the unprocessed slice is ready for etching.

A frequently used method for endpoint detection utilizes endpoint tracing (abbreviated as EPT). This EPT is realized by an emission spectroscopy means and is used to measure the concentration of gas in the plasma over the surface of the slice to be etched. This EPT is designed to monitor the reactants or products of the etching reaction. For example, EPT
By adjusting , the monitoring device can detect when these etching products are no longer released into the plasma, thereby signaling the end of the desired etching reaction. In general, endpoint detectors locate abrupt changes in the concentration of the species being monitored to the etching process;
It is designed to detect the approximate time at which this end point is expected. This device is beneficial in the following respects. That is, a signal is provided to the etching apparatus to terminate the etching cycle, remove the etched semiconductor slice, and insert a new (unprocessed) semiconductor slice into the etching chamber.

However, there are several limitations to these endpoint detectors, whether applied or merely theoretical. The basic limitation is that these detectors only function to detect the end of the etching operation. Therefore,
These detectors have the disadvantage that they do not give any information as to whether the etching process is continuing in the best possible manner or whether anomalies have occurred during this process.

A need has arisen for a process or apparatus that can determine whether an etching process is optimally underway or complete. An etching monitoring system that can automatically check the etching status of all slices would solve the above-mentioned problems in modern semiconductor manufacturing quality control systems.

Additionally, any system capable of monitoring the etching process in a manner that provides information to determine whether the layers formed on the semiconductor device have been formed and processed in a predetermined manner prior to the etching operation. There are further advantages.

Conventional equipment and processes have not been able to achieve the desired benefits described above.

A co-pending patent application (U.S. Patent Application Ser. A method and apparatus for detecting anomalies is disclosed. The particular invention in this application relates to plasma etching processes in the manufacture of semiconductor devices. In accordance with an embodiment of the invention, the actual end point trace (EPT) for each semiconductor slice etch is compared in detail to a predetermined reference EPT to determine whether the etch is progressing as desired. It not only determines, but also determines whether a predetermined operation has been properly performed on the slice before etching.

[Purpose of the invention]

The present invention aims to improve the methods and apparatus disclosed in the related inventions mentioned above.

[Summary of the invention]

According to the present invention, an improved method and apparatus for detecting anomalies in cyclical or repeated process operations can be provided.

A particular application of the invention is plasma etching processes in the manufacture of semiconductor devices. Further, in accordance with one embodiment of the present invention applied to endpoint trace data, actual endpoint trace data for etching a semiconductor slice and a predetermined reference endpoint trace for the particular etching process being performed. This results in improved analysis between In accordance with the present invention, it has been found that in many cases, analysis of this endpoint trace data allows the substantial cause of a detected anomaly in the etching reaction to be deduced. Additionally, in accordance with the present invention, analysis of this endpoint data can provide valuable information related to the layers of material formed on the semiconductor device prior to the etching process.

According to one embodiment of the present invention in conjunction with plasma etching, the etch progresses as desired by closely comparing the actual end point trace for each slice etch with a predetermined reference EPT. In addition to determining whether a given operation has been properly performed on a slice before etching. For example, according to the present invention, it is possible to analyze the conditions where the deposition is too thick or thin, where the photolithography step is omitted or performed incorrectly and the predetermined features are still surrounded. Can be done.

According to the embodiment described above, a reference EPT is first defined. This reference EPT is divided into a plurality of regions, each of which has significance with respect to a corresponding connection point or step during the etching process. The actual EPT is obtained during the etching of each semiconductor slice. This actual EPT is divided into a plurality of regions based on the change in the slope point of this EPT. A series of practical functions are applied to align the area of the actual endpoint trace with the area of the reference EPT. This is the actual E
This is done by comparing the characteristics of the desired matching region of the PT with the characteristics of the region of the reference EPT to be matched. Reference EP for each of the areas of actual EPT
When matching the regions of T, the predetermined characteristics of each region are compared with the characteristics of the corresponding region of the reference EPT. Such a comparison is performed by executing a set of rules adapted to the particular process to be monitored. These rules are called "process specific" rules. These process-specific rules invoke a set of rules that compares the general set of characteristics of any two domains. These latter rules are referred to as process independent rules. By comparing the area of the actual EPT and the area of the reference EPT, the system can prove when an anomaly occurred during the etching operation. In addition to analyzing the area in detail,
By applying process-specific rules, the system can infer causes for proven anomalies.

The present invention has the advantage of providing a system that has the ability to compare and analyze continuously varying data curves with continuously varying data curves.

According to a particular advantage of the present invention, changes in the shape or duration of particular regions of a particular data curve can be tolerated, and predefined regions of the actual data curve can be demonstrated; can be compared with the corresponding area of the reference curve. This is a particular advantage for various reasons. That is, the duration of one or more regions of an actual data curve will vary widely from curve to curve.

In order to accurately analyze each of the actual data curves, it is extremely important to first accurately establish the starting and ending points, as well as accurately determining the duration of each region of the curve.

Further, according to an advantage of the present invention, it is possible to define and compare the regions of the curves that can vary continuously, and to determine which regions of the actual data curves are inconspicuous and whose duration can be "extended" over time. It can also “float”. For example, if the initial region of the data curve is of excessively long duration,
The present invention has the advantage of being able to tolerate such extended durations while still proving the boundaries of continuous regions in the data curve.

Still further, it is an advantage of the present invention that the system of the present invention is standardized so that it can be easily and quickly applied to modify etching or manufacturing operations.

[Embodiments of the invention]

Embodiments of the present invention will be described in detail below with reference to the drawings.

As previously mentioned, the present invention provides an improved method and apparatus for detecting anomalies in repetitive type process operations.

The present invention is applied to such repetitive process operations, which can be monitored by a device that generates continuously variable electrical signals that are correlated to the progress of the process to be monitored. signal or E
PT is obtained. One type of such continuously variable signals utilized in tree inventions are time-based. That is, this signal changes over time. Also, this signal is one of time varying intensity. Such signals can often be classified into multiple areas, which areas correspond to steps or parts of the process to be performed. It is also important to be able to define this signal as a preferred or reference signal. Here, the reference signal represents the signal obtained when the process is proceeding under favorable conditions.

Therefore, the actual signal received during process operation can be matched to the characteristics of the reference signal.

In the present invention, the reference curve is defined as follows. That is, this curve corresponds to the continuously variable signal that would be expected if the sensing device were monitoring the process. During process operation, an actual curve is obtained which represents the output of the sensing device. During or after a process cycle, the area of the actual curve is matched against the corresponding area of the reference curve, so that the actual curve determines whether the process is well underway or has progressed. or an abnormality occurred during the process. The characteristics of each region of the actual curve compared to each region of the reference curve vary from region to region, and are determined by process and data considerations for the particular process step to which these regions correspond. It is something that will be done.

The results of the comparison of the actual and reference curves can be used in various ways. For example, sending a signal to processing equipment to mVt a process operation, sending a signal to processing equipment to inhibit this operation, sending a signal to the operator of this equipment, specifying a specific process cycle for later use. storing information about, sending signals to processing equipment to change the process during or during subsequent process cycles, or pointing out specific anomalies or analyzing the causes of detected anomalies. There is a way.

The aforementioned pending patent application No. 046,497 discloses a method and apparatus for detecting anomalies in cyclically repeated process operations.

This patent application describes a system for defining a reference curve and its area for a particular process. Additionally, actual curves representing data from this process are obtained during the process and compared to a reference curve to determine whether an anomaly is occurring or has occurred during the process to be monitored. Also,
This application describes the use of this system in conjunction with plasma etching during process operation.

According to this system, while analyzing the actual curve, the area of this curve is defined as a function of time from the start of the process operation.

According to the present invention, an improved process and apparatus for comparing a reference continuously variable curve and an actual continuously variable curve can be provided. The present invention includes an improved method for defining regions in actual curves, so that these regions can be more accurately defined and the analysis thereof more accurate. The invention also provides an improved method that utilizes a practical set of functions to define and match the area of an actual curve to the area of a reference curve. This provides an accurate definition and analysis of the area of the actual curve. Additionally, the present invention can be modularized, and furthermore, a set of process-independent rules and a process-specific
A set of files can be used for the analysis. As a result, the system of the invention can be particularly well and easily adapted to different and varying etching processes or other manufacturing processes.

An embodiment of the present invention has been found to be suitable for application in conjunction with semiconductor manufacturing processes, particularly plasma etching processes. Hereinafter, an example in which an embodiment of the present invention is applied to a plasma etching operation will be described with reference to FIGS. 1 to 9.

First, FIG. 1 is a conceptual diagram illustrating a typical plasma etch reactor used in conjunction with one embodiment of the present invention. This reactor includes a plasma etching reaction vessel 10.
A shower head electrode 1 is provided inside the shower head electrode 1.
2 is installed.

This showerhead electrode is electrostatically coupled to an RF power source 14. Typically 13.5MHz by this power supply 14
of high-frequency power. This RF power supply 14 includes a first
An RF tuning network, not shown in the figure, is provided. A hole 16 is formed in this electrode 12 through which gases are released and supplied to the electrode 12 via a gas line 18. The flow rate of this gas is controlled by a mass controller not shown in FIG. The semiconductor slice 20 to be etched is placed on a substrate 22 of the reactor, and the substrate 22 is grounded as indicated at 24. The reactor is also provided with a pressure regulator 26 and a gas outlet line 28, which line 28 is operatively connected to a vacuum tube not shown in FIG.

In operation of this plasma etching reactor, a mixture of gases is supplied through the electrode 12 and power is supplied to the RF power source 1.
4 through this electrode.

A low pressure glow discharge is established and reactive species are generated in the plasma 30. These reactive species selectively react with the thin film of material on the slice 20, resulting in products that evaporate from the slice 20 and, ultimately, are pumped through the gas outlet line 28. Served. Such etching operations, if the etching is successful, continue until a predetermined amount or depth of the desired material has been etched from the surface of slice 20.

EPT is obtained by emission spectroscopy from the etching reactor used in the embodiment of the invention shown in FIG. This EPT detection device allows the detection of light emitted by electronically excited species in the plasma.

Although the light emitted by atoms, molecules or free radicals in an unbalanced plasma is not strictly proportional to the density of these species, for the purposes of this EPT it is assumed that the proportionality property is preserved. It can be assumed. Install a detection device and
The concentration (density) of the etching reaction products, ie, the reactive species in the plasma, can be monitored.

Once it is determined which etching product or species is to be detected, the system can be directed to monitor the wavelength of light that corresponds to the light emitted by the species to be monitored. According to this example, the system is designed to detect light emitted by a particular etching product.

Returning to Figure 1 again, the EPT for etcher processing
Also shown is an apparatus for obtaining . The plasma etching reaction vessel 10 is provided with a window 32 through which light emitted from the plasma passes. Light from this window passes through an optical filter 34. This filter 34
to block light outside a preselected emission band.

According to this example, since this device is set to monitor carbon monoxide, the filter 34 is used to detect carbon monoxide.
) passes light in the optical band. Light emitted from plasma 30 passes through window 32 and, if its wavelength is suitable, passes through optical filter 34 and reaches photodiode 35 .

An electrical signal corresponding to the intensity of the light impinging on the photodiode 36 is generated from the photodiode 36 and is also supplied to the EPT recorder 38 .

In a typical conventional plasma etching reactor,
The EPTI-lace recorder 38 provides a chart paper record of the EPTI-lace of the etching operations performed in the reactor.

One of the desired reactions in this example is the etching of silicon dioxide with CF: SiO□+CF4===>SiF4+CO.

This production of CO, and therefore its spectral intensity in the plasma, decreases when all of the SiO□ is etched from the slice and the above-mentioned reactions stop producing more CO. Therefore, the decrease in 00947 intensity;
That is, the point in time at which this end point is predictable is an indication that the layer has been etched (ie, the end point has been reached).

A digital processor or computer 40 is illustrated in FIG. 1 and is operatively connected to photodiode 36 to receive the output signal therefrom. According to this example, the computer 40 stores programs and subroutines for the system, stores the reference EPT and other data, analyzes the actual EPT, and compares the EPT with the reference EPT. . Additionally, information on the results of these EPT comparisons can be stored and predetermined signals or controls can be provided to the plasma etcher or apparatus operator to provide other desired functions. The computer 40 can be programmed to perform EPT comparisons, as well as perform other functions described below, and can also be provided with a separate computer and installed as part of a computer control system for the plasma etch reactor. You can also. In this case, such a computer control system needs to have sufficient ability to perform the various functions described below. According to one embodiment of the invention, computer 40 comprises a Texas Instruments Explorer computer with software written in LISP. This software is designed to be able to analyze any form of EPT and compare it to a reference EPT for the etching process.

FIG. 2 shows a simulation of a reference EPT for a particular notching process for a layer of oxide material on a semiconductor slice in one embodiment of the present invention. In Figure 2, intensity is plotted as a function of time. This intensity corresponds to the intensity of the signal generated by photodiode 36 in FIG. As previously mentioned, the EPT device is preset to monitor carbon monoxide enoting reaction products vaporized into the plasma. The EPT of FIG. 2 simulates the signal from the photodiode 36 from a point before the etching reaction starts, up to the end of the plasma etching process, during several steps encountered during the plasma etching process. It is. Therefore, in FIG. 2, prior to T1 in region RO, the signal from photodiode 36 is
It is below a threshold value of 50 representing the absence of a plasma of excited species. This indicates that the plasma etching reaction has not yet started. At time T1, the EPT intensity exceeds this threshold level 50 and a plasma etch reaction is initiated. The intensity of this EPT increases sharply from time T1 to T2 in the region designated R1, thus representing the presence of a plasma of the excited species as well as a sharp rise in the concentration of the monitored species in the plasma. be able to. This R1 corresponds to a particular phase of the etching operation, which is similar to other regions R2-R of the EPT.
5. For example, R2 represents the transition point from the starting phase R1 to the relatively stable phase R3, during which a predetermined amount of oxide material is etched from the surface of the slice and is now in the plasma. The concentration (density) of the species being monitored remains approximately constant.

Region R4 represents a drop in the density of carbon monoxide species being monitored in the plasma and is similar to region R3.
This is an indication that the oxide material or layer being etched into it has been depleted. The flat portion of region R5 corresponds to the output signal from the plasma itself with little or no carbon monoxide present in the plasma.

In this R5 flat region, the intensity remains relatively constant, indicating that little or no species are being produced. The etching operation continues for a predetermined period of time after the end point of the oxide etch, as indicated by region R4, thereby removing oxide material from the etched area of the slice. Can be completely washed. After a predetermined time, the etching operation is complete and plasma is no longer generated. This is represented by a steep drop in intensity in the latter half of region R5. In region T6, the signal falls below the threshold value. The reason is that there is no plasma of excited species.

The etching operation is designed to proceed through two or more layers of different materials. In such a case, the reference EPT would, of course, be varied from the EPT shown in FIG. 2 and would include areas corresponding to the various layers and materials to be etched. An example of such an EFT is shown in Figure 7, where etching is performed through four different materials consisting of polysilicon, mitox, silicon nitride and silicon oxide. represents the trace of

3-6 are configuration charts displaying the functional architecture of a software program for controlling computer 40, by which one embodiment of the present invention is implemented.

FIG. 3 is a general block diagram illustrating the functionality of one embodiment of the present invention designed to monitor plasma etch operations. The embodiment shown in FIG. 3 is referred to as the "Plasma Etching Diagnostic Expert System" or PEDX.

Module 6 represents the PEDX system.

The overall block diagram of FIG. 3 displays a system illustrated in three general functional modules.

The first general functionality module is illustrated at 62 and includes the functionality that defines a normal trace, or EPT. A block diagram detailing various specific functions of general module 62 is illustrated in FIG. 4 below.

The second general function of this PEDX system 60 is the signal-to-symbol conversion function represented by module 64. The various functions of module 64 are illustrated and described more fully with respect to FIG. 5 below. A third general function of PEDX system 60 is to compare the actual endpoint trace and the reference endpoint trace as indicated by module 66. The functionality of this module 66 will also be described in connection with FIG. 6 below.

FIG. 4 is a functional block diagram of module 62 of FIG. Module 62 represents the general functionality for defining the reference trace in this example. As shown in FIG. 4, the general function for defining a reference trace includes multiple steps for defining regions in the trace. The functionality of defining regions is shown in module 70. Arrow 68 indicates that function 70 is executed several times, for example for each region of the trace, and this function is analogous to a loop in a computer program chart. Function 7
0 acts to define the target for describing each region of the reference EPT.

The functions of defining each region of trace 70 include obtaining a start and end point encompassing each time of region 72;
obtaining or verifying the material to be etched during step 4; obtaining the shape of region 76; computing the average slope of region 78 and finding the maximum and minimum points; and programming file 80. Contains functionality to print information about the area.

The function of FIG. 4 for defining regions is performed for each region of the reference trace. The characteristics of FIG. 4 for each region are defined by the importance of these regions and the operator's knowledge of the characteristics of the region of the reference trace.

FIG. 5 is a block diagram of the operation of a system for performing the general functions related to signal-to-symbol conversion shown in FIG. 3. There are two primary steps in function 64. 1st
This step is for obtaining the actual EPT as shown at 82. The second step, as indicated by module 84, is to match the actual EPT to the reference EPT. There are two primary components to this function 84 that matches the actual EPT to the reference EPT. The first is represented by 86 and proves a point in the actual EPT, where there is a large slope change. These points are called critical points. Adjacent points between these critical points have the same slope, so they are grouped together. A region is defined to be the portion of the trace between two consecutive critical points. The functions of the program for specifying these critical points are illustrated by function blocks starting with function block 86. This program first sets up a starting point in list file 88. Then move on to the next point,
Calculate slope 90. Points are added to list 92 or new starting points are set in list 94 by the program in relation to the slope calculated relative to the slope of the starting point of this list.

As discussed above in connection with FIG. 4, an object is defined;
Each area in the EPT will be explained.

Attributes of this object include average slope, area maximum, minimum, critical point intensity, slope and time, and other intensities as desired. 1 of the object relative to the reference EPT
Although (cents) - once determined, never changes again, the set of objects representing the actual EPT can also be corrected by the matching function.

This is the second component of the signal-to-symbol converter and is designated as functional block 84 in FIG. EP based on the actual EPT area by matching components
Combine in the area of T. If these areas correspond to the same state of the process, they form a set. The matching operation is difficult. The reason is that the actual and reference EPT critical points are fundamentally different;
This is particularly because there are problems with etching.

Overall, the matching function matches the area of the actual EPT to the area of the reference EPT from left to right. For each region of the reference EPT, the actual EPT
Consider the critical point in the inconsistency part. The order of comparison is used to select two critical points of the actual EPT that define the limits of the matching region. This system is practical as follows. That is, attributes are compared for their importance until a critical point is found in the actual EPT that best matches the critical point in the reference EPT.

In this system, we consider the slope at the critical point to be the most important attribute. As a result, the matching component can merge several specified regions of the actual EPT, split these regions into one, or select regions without making any actual changes.

Referring again to FIG. 5, the second component of function block 84 that aligns the actual EPT to the reference EPT includes matching the actual EPT region to the reference EPT region, as indicated by function block 96. Means for matching are included. The series of functions subordinated from module 96 are repeated for each region of the EPT, as indicated by arrow 98. The first step in this function is to find the starting point of region 100.

In this step, an analysis is performed to establish whether the point under consideration is the first point in EPT 102 or the last point in the final region of EPT 104.

At 106, the program finds the end point of region 106. In determining the end points of regions and matching these points to the regions of the reference EPT, this program employs a practical set of functions. The first function is to obtain candidate regions along with alignment slopes 108. If several candidates exist, the program obtains more identification features 110 and employs additional features. These additional features include obtaining candidates with the correct height 112, correct duration 114;
as well as obtaining a candidate with the correct time 116. On the other hand, 106
If there are no candidates in , the program obtains fewer discriminative features 112 and selects candidates with the correct time but the wrong slope 120;
Furthermore, obtaining a candidate with the correct height but the wrong slope 122 is included.

A third function of module 96 is fine tuning of results 124. In performing this tuning, the program collects adjacent regions of both the reference and actual EPT 126 and determines whether these adjacent reference regions are oriented from top to bottom or from bottom to top. The system then checks the association of the region of the actual EPT with adjacent regions, determines the state of the region relative to the left side of the region of the actual EPT 130, and changes the state of this region relative to the right side 132. and then adjust the changes at 134.

This includes shifting the critical point of the region of the actual EPT to the left 136, shifting this critical point to the right 136, or continuing in the region until a change occurs in the direction of slope 14. .

FIG. 6 shows the functionality shown in module 66 of FIG.
FIG. 2 is a block diagram showing the work of a program for execution in comparison with PT. As shown in FIG. 6, this function involves obtaining a reference region 142 for each region and
4 and then executing a set of process specific rules for region 146. As previously mentioned, process specific rules are executed to determine the actual
Each independent region of PT will be analyzed in detail.

Additionally, the rules include different sets of comparisons and further include analysis of one area against the other. Initiating execution of the set of process-specific rules for region 146 includes initiating process-specific rules 148 and initiating process-independent rules 150.

With reference to FIGS. 7 and 8, an example of the operation of the system of the present invention in the area of EPT matching is illustrated.

FIG. 7 shows an example of a reference end point trace and intensity plotted as a function of time. In Figure 8,
An example of an actual endpoint trace to be matched by the system relative to the reference endpoint trace of FIG. 7 is illustrated. In matching the area of the actual end point trace in FIG. 8 with the reference end point trace in FIG.
Assume that it matches point A in the figure. This corresponds to the functional module 100 in FIG. The program then selects a point from the actual endpoint trace that corresponds to point B of the reference endpoint trace in FIG. 7 (see, eg, functional module 106 in FIG. 5). Critical points B' and D' can be confirmed as candidates for matching critical point B. The reason is that all these points lie at the end of a region with a slope comparable to that of point B above. The next attribute of the system comparison is the intensity comparison of the critical points B', D' with respect to the intensity at point B. In this case, critical point B' is not close to the intensity of point B, but D' is close to B, and then D' is selected and matched with B, and points A' and D' are The regions in between are fused to form a single region terminating at point D'. In this example, the remaining areas of the actual EPT are matched in a direct manner.

Another example of a program's ability to match between regions is shown in FIG. Figure 9 shows the actual EP
This shows an example of T, and is compared here with the reference end point trace in FIG. For comparison under this system, point A' in FIG. 9 is matched with point A in FIG. 7. Also, point B' is matched with point B, and point A'
-Match point B' (represented by region 2') with region B-C (represented by region 2) in FIG. However,
Region 2' in FIG. 9 is of considerably longer duration, compared to region 2 (corresponding) in FIG. 7. With this system, the extended duration of region 2' of the EPT in FIG. 9 can be easily handled. This is because the system does not define a critical point from point B' until there is a change in slope of a predetermined magnitude. In FIG. 9, a change in slope occurs at point C'. This system thus allows region 2' of FIG. 9 to be matched to the shorter duration region 2 of FIG. 7, although there is an extended duration. According to this system, point C' is area 2
A suitable critical point at the end of ′ can be confirmed by checking that the other data for the region, e.g. Become. Furthermore, the present system allows regions adjacent to region 2' to be verified for masociation with respect to the corresponding expected regions of the reference EPT of FIG. This illustrates two possibilities of the invention: even if the area of the actual endpoint trace extends considerably beyond the expected duration with respect to the reference endpoint trace, the actual endpoint trace The area of the point trace can be matched to the area of the reference end point trace. Furthermore, this illustrates the potential of the present invention to allow regions to "float" in time and be accurately matched and compared. The regions dependent on point C' (e.g. regions C'-D' and D'-E') are
It is "floating" and is much slower than you would normally expect from these points. Additionally, the present invention has the ability to accurately confirm and compare these areas.

The system of the present invention also allows matching of regions of the actual endpoint trace, the duration of which is shorter than that of the corresponding region of the reference endpoint trace. .

The examples in Figures 8 and 9 illustrate the potential of the system, using a set of practical features to The area of the actual end point trace can be matched to the area of the reference end point trace. In accordance with the present invention, the output of the signal-to-symbol converter can be used to detect and diagnose problems associated with etching, for example, using rules for operating module 64 of FIG. This system consists of two sets of rules: process-independent rules and process-specific rules.

According to the process-independent rule, the area of the actual trace and the area of the reference trace are compared and anomalies are searched for. Apply these rules to any plasma etching process by comparing the intensity, slope, and time of the critical point in the reference and actual endpoint traces. The conclusions of these process-independent rules are symbolic, e.g.
The result of a rule that compares the times of two critical points can be expressed as "too early,""toolate," or "no problem." This process-independent rule also allows searching for anomalies and irregularities in the region, such as positive or negative peaks and multiple peaks.

On the other hand, process specific rules allow checking the cause of the problem. By dividing these specifying rules into groups, each group can represent knowledge about a different area. Process specific rules are
Developed by process engineers, who have experience working with specific processes and
Familiar with various issues and corresponding endpoint trace shapes. Rules are written by process engineers to address symptoms by identifying appropriate process-independent rules and combining one or more causes for each problem.

In addition, a set of these rules (
set) and compare or possibly analyze areas of the trace. Therefore, these process-specific rules continue using what can be found from the process-independent rules.

In analyzing the actual end point trace of FIG. 8, the system of the present invention compares each region of FIG. 8 with the region of FIG. In this comparison, these rules ignore cases where regions are nearly identical, and detect the presence of large differences between regions. Points A' to D in Figure 8
According to the process-independent rule that compares the times of critical points in the case of the region up to It can show that it is too slow.

According to the conclusion of the process-specific rule, which called for a process-independent rule that checks the time of the critical point, it indicates that the last critical point of the region occurred too late, and
This area indicates that it was intended to represent the etching of the polysilicon layer. Process specific rules allow the actual endpoint trace to suggest a conclusion indicating that thin material was present on the polysilicon area to be etched during the etch.゛This thin material on the polysilicon allows point A'
An abnormality is induced during EPT from to D'.

These rules allow detection of anomalies regardless of the size of the difference outside the tolerance range. Tolerances depend on empirical knowledge of the process in the manufacturing environment. Although failure ranges are given, process engineers can emphasize values when writing these process specific rules. If a problem is detected, the system of the present invention allows the plasma etcher to be stopped and the diagnostic results to be reported to a technician for corrective action.

According to the present invention, it is possible to detect and diagnose etching deterioration due to errors occurring during the aforementioned process. One of the major causes of these errors is due to the deposition process. Either too much or too little material would be deposited here. Such-
An example is illustrated in the actual EPT of FIG. Area 2'
has an unusually long duration, which corresponds to criterion E in Figure 7.
It is longer than region 2 of P'r'. Area 2
The process specification rule applied in step 2 informs us that this region has been given an unusually long etch time, and informs us that the material being etched in this region is polysilicon. An abnormally long duration may indicate that the particular polysilicon layer to be etched has been deposited too thinly on the semiconductor device. Examples of other process steps that may be inadvertently performed before monitoring the etch process include photolithography steps, etching steps, or other processes.

According to the present invention, endpoint traces can be interpreted by combining signal-to-symbol transformation as well as rule-based reasoning. These signal-to-symbol conversion and process-independent rules are applicable to any plasma etching process or other process that provides a suitable continuously variable data curve. If a problem arises with the etching process, it can be detected by a small set of process-independent rules. - Diagnose the problem by combining the cause with one or more symptoms through a set of process-specific rules.

Also, as is clear, since the system of this embodiment of the present invention is modular, knowledge regarding the process to be monitored can be isolated at the process specific rule level. The other main components of the system, the signal-to-symbol converter and the set of process-independent rules, depend on the particular process (or etching process) to be monitored.
It is applicable to any process (or etching process) that provides a suitable continuously variable data curve. Therefore, embodiments of the present invention can be easily, effectively and
Can be applied quickly.

There are also particular benefits when the present invention is used in conjunction with plasma etching reactors. Many of these benefits are due to the operational complexity of plasma etch reactors. This complexity results from the many factors that affect whether the etching operation proceeds optimally. Since there are many factors that give shape to the actual etching reaction, the present invention, which directly observes the characteristics of the reaction itself rather than the operating parameters of the reaction vessel, has a much better effect than previously proposed methods of monitoring the etching process. It is highly reliable and can provide a direct means. In summary, embodiments of the present invention monitor and monitor the process itself, as opposed to process control parameters. Examples of process control parameters that can be varied in a plasma etching reactor include power, pressure, checking the relative concentration and content of gases supplied to the system, total gas flow to the reactor, and changing its pressure and temperature. This includes things that can be done. Some of these process control parameters are monitored by the hardware controls of the etching reactor.

However, many additional factors beyond these control processes affect the processes within the etch reactor. Some of these parameters include what is referred to as the chamber sheathing or character of the material on the interior surfaces of the reaction vessel. Recombination of surface materials on the interior surfaces of the chamber frequently occurs within the etching reactor, thereby affecting the progress of the etching reaction within the chamber.

Another factor is the distance of the electrode from the slice surface. This electrode itself is also slightly etched during reactor operation. After multiple etchings have been carried out, the distance between the slice surface and the electrode becomes increased due to the etching process of the electrode and the resulting etching reaction affected. These factors beyond the control parameters are not easily influenced by monitoring actions.

Slight variations in one or more of the above-mentioned factors can have a significant impact on whether the etching reaction can proceed optimally or acceptably. However, due to the complexity in the interrelationship of species factors, it is difficult to predict the effect of variations in one or more factors on the actual etching reaction itself. Because of this complexity, special advantages are obtained according to the invention. Conventional hardware allows the etcher monitor to control only a few of the factors that affect the response, and as mentioned earlier, many of these factors do not easily influence this monitoring behavior. .

Furthermore, this hardware controls only the monitor factors that affect the reaction, not the reaction itself.

Therefore, in many cases, some factors will change and have a detrimental effect on the etching reaction, but if the hardware monitor displays the exact set of process control parameters that exist (e.g., correct pressure,
This hardware monitor could not detect that the etching reaction was proceeding incorrectly (temperature, gas flow rate). However, according to the present invention, by monitoring the process in an observable manner (for example, the production rate of reaction product species), it is possible to monitor the actual etching reaction more accurately, and to detect errors in the process when they occur. ,
It can detect more intensively than conventional systems.

Furthermore, in accordance with the present invention, it is possible to provide a system capable of defining and matching the region of an actual continuously variable curve, such as an actual EPT curve. According to this system, the above-mentioned operation is realized even when these regions deviate significantly from the regions of the continuously variable curve of a specific reference that have been defined. For example, the invention allows defining areas of the actual curve that can float or extend in time. This provides a significant benefit to curve comparison systems that compare regions of curves whose regions are defined invariably as a function of time. Furthermore, our practical approach to matching critical points and regions allows the regions of these curves to be accurately defined, matched, and compared.

In accordance with one embodiment of the present invention, process-related problems that occur during typical plasma etch processes can be advantageously detected by inspecting and analyzing the end point (EPT) of the plasma etch process. Early and automatic detection of such problems can greatly improve the slice generation rate while preventing erroneous processing of subsequently processed slices. Furthermore, according to the present invention, problems can be recognized because predetermined causes have been detected, and signals representative of these causes can be generated.

Additionally, the invention can be used in real-time applications. That is, the etching process can be performed in real-time, in collaboration with other systems, such as expert systems, or in real-time, so as to change or correct the etching parameters during etching when anomalies are detected, while actually proceeding independently. It can be monitored by

Further, according to an additional benefit of the present invention, it is not necessary to install a strip chart recorder in the clean room, and a large number of endpoint traces can be edited by saving only endpoint traces exhibiting anomalous behavior. effective.

Additionally, the system of the present invention functions as a process monitoring tool, allowing the microprocessor to interpolate based on hardware monitoring capabilities. Along with setting alarms by hardware monitor,
According to embodiments of the present invention, by inhibiting processing based on hardware problems (e.g., no RF mains power, no gas, inadequate pressure, or other similar problems). , the operator is alerted and can eventually stop the etcher of any reaction-related problems.

According to the present invention, it is possible to provide a system that allows the etching reaction to be observed and detects errors during the etching operation that cannot be detected by monitoring the equipment; That is, it merely monitors gas flow, temperature, or etcher RF power settings. Furthermore, it is possible to provide a system of the present invention having a bank amplifier monitoring system for a matching reactor hardware monitoring device. Therefore, the hardware monitor sensor fails and
When a hardware function becomes below performance, the present invention allows this problem to be detected as soon as the out-of-performance hardware function changes the etching response from the reference response.

Further, the present invention can be effectively utilized under various situations. For example, it can be applied to various etching operations, each etching operation having its own reference EPT. Each of these various EPTs has a different number of regions, and each of these regions has a different shape. Furthermore, it has the ability to define various properties of each region and to detect or measure such properties by considering what is unique to the region of etching or to a particular etch. Furthermore, the present invention can be used for processes other than plasma etching reactions.

Further, in the embodiment shown in FIG. 1, analog signals from the photodiode signal generator are received, but not only analog signals but also digital signals can be used.

As detailed above, the present invention is not limited to the above-described embodiments, and it is understood that those skilled in the art can easily make various changes without departing from the technical idea of the present invention. be.

In connection with the above description, the following items are disclosed.

1. Establishing a reference endpoint trace for a predetermined etching process, dividing this reference endpoint trace into predetermined regions, and applying the predetermined etching process to a semiconductor device by four people; Collecting actual end point traces related to etching of semiconductor devices, dividing the actual end point traces into candidate regions according to predetermined criteria, and matching characteristics of the candidate regions of the actual end point traces. matching the area of the actual end point trace to the corresponding area of the reference end point trace by analyzing the entire characteristics of the area of the reference end point trace; A method of monitoring an endpoint trace of a plasma etch reactor, the method comprising the step of comparing characteristics of a region of the trace with corresponding characteristics of a matched region of said reference endpoint trace. .

2. The method further comprises the step of generating a signal representing a variable of the characteristic of the area of the actual endpoint trace exceeding a predetermined limit value from the corresponding characteristic of the matching region of the reference endpoint trace. The method according to claim 1 (hereinafter simply referred to as "Claim 1"), characterized in that:

3. The method of claim 2, wherein the predefined criteria for dividing the actual endpoint trace into candidate regions includes changes in the slope of points along the actual endpoint trace. -4. In the step of comparing the characteristics of the N area of the actual end point trace with the corresponding characteristics of H-hli matched with the reference end point trace, - sets of rules are applied to the actual end point trace. 4. A method according to claim 3, further comprising the step of applying to at least one of the regions.

5. The method according to claim 4, wherein the set of rules includes -a process-independent rule and a set of process-specific rules.

66 The step of applying the set of process-independent rules comprises: between the region of the actual endpoint trace and the matched region of the reference endpoint trace; 6. The method of claim 5, further comprising the step of comparing at least one of the values of the points, the slope, and the time.

7. Determine whether the deviation between the characteristic of said actual endpoint trace and the corresponding characteristic of said reference endpoint trace exceeds a predefined value, and convert such deviation into at least one predetermined value. 4. The method of claim 3, further comprising the step of: attributing the deviation to a predefined cause and generating a signal representative of the at least one predefined cause to which the deviation has been attributed.

8. said step of dividing the actual end point trace into candidate regions according to said predefined criteria, determining the slope of this actual end point trace at a point along said actual end point trace; The method further comprises the step of determining a critical point of the actual end point exceeding a predefined value of Rnn, and further dividing the actual end point trace into candidate regions between these determined Rnn points. The method recited in claim 2.

9. By analyzing the characteristics of the candidate region of the actual end point trace with respect to the characteristics of the region of the reference end point trace to be matched, the correspondence between the candidate region of the actual end point trace and the reference end point is determined. 3. The method according to claim 2, wherein the step of matching the area of the trace includes the step of applying the set of practical functions to the area of the actual end point trace.

10. In the practical function; identifying a region of the actual endpoint trace whose average slope varies by a value less than a predefined value from the average slope of the region of the reference endpoint trace to be matched; Identify areas of these traces in which the intensity value of the final point of the actual endpoint trace changes by a value less than a predetermined intensity change from the intensity value of the final point of the region of the reference endpoint trace to be matched. identifying a region of the actual endpoint trace whose duration changes by a value less than a predetermined duration change value from the duration of the region of the reference endpoint trace; and Claim 9 characterized in that at least one operation of checking the area of the actual end point trace occurring within the range is provided.
Method described.

11. In monitoring the operation of a process device that executes a predetermined process, this process is first monitored by a detector to obtain a continuously variable signal curve corresponding to the progress of the process, and a reference continuously variable signal is defining a reference continuously variable signal curve for a predetermined process in such a way that the curve corresponds to a predetermined acceptable behavior of the predetermined process; defining characteristics for at least one region of a signal curve, introducing said predetermined process, and collecting from said detector an actual continuously variable signal curve corresponding to the progression of said introduced process; Divide the actual continuous variable signal curve into candidate regions based on the characteristics thereof, and apply the characteristics of the candidate region of the actual continuous variable signal curve to the entire characteristics of the region of the reference continuous variable signal to be matched. matching the candidate region of the actual continuous variable signal curve with the region of the reference continuous variable signal curve; The reference continuously variable signal is compared with the characteristic of the matched region in the continuously variable signal curve, and furthermore, the signal representing the variable in one characteristic of the region of the actual continuously variable signal curve that exceeds a predefined limit value is A method of monitoring comprising the step of generating from a characteristic of correspondence of matched regions of curves.

12. The method according to claim 11, characterized in that, on the basis of which the actual continuously variable signal curve is divided into candidate regions, the characteristics thereof do not in principle include the time component of this curve.

13. The method according to claim 11, characterized in that by identifying critical points in the actual continuous variable signal curve, this signal curve is divided into candidate regions, and these candidate regions are made to exist between the critical points. Method.

14. The method of claim 13, wherein the critical point is determined based on a change in slope of the continuously variable signal curve.

15. Claim 11, characterized in that the step of matching the actual continuously variable signal candidate region to the reference continuously variable signal region includes a step of applying a practical set of functions to the candidate region. the method of.

16. The practical function includes: an area of the actual continuously variable signal curve having an average slope that is less than a predefined average slope value from the average slope of the area of the reference continuously variable signal curve to be matched; Check that the intensity value at the end point of this actual signal curve varies by less than a predetermined intensity variation value from the intensity value at the end point of the region of the reference signal curve to be matched. identifying an area of the signal curve; identifying an area of this actual signal curve having a duration that varies by less than a predefined duration change value from the duration of the reference continuously variable signal curve; 16. The method of claim 15, further comprising at least one step of: ascertaining the region of the actual continuously variable signal curve that occurs within a predetermined time range.

17. The method according to claim 13, wherein the step of matching the candidate region of the actual continuously variable signal with the region of the reference continuously variable signal includes a step of applying a practical set of functions to the candidate region. Method.

18. The practical function includes: an area of the actual continuously variable signal curve having an average slope that varies by a value less than a predefined average slope value from the average slope of the area of the reference continuously variable signal curve to be matched; Check that the intensity value at the end point of this actual signal curve varies by less than a predetermined intensity variation value from the intensity value at the end point of the region of the reference signal curve to be matched. identifying an area of the signal curve; identifying an area of this actual signal curve having a duration that varies by less than a predefined duration change value from the duration of the reference continuously variable signal curve; The method of claim F further comprising at least one step of: ascertaining the area of the actual continuously variable signal curve that occurs within a predetermined time range.

19. The step of comparing the area of the actual variable signal curve with the area of a reference variable signal curve comprises the step of applying a set of rules to at least one of the areas of the actual variable signal curve. The method according to claim 11, characterized in that:

20. The method according to claim 19, wherein the -set of rules includes a -set of process-independent rules and a set of process-specific rules.

21. The step of applying the set of process-independent rules includes determining at least one predefined point value between the region of the actual signal curve and the matched region of the reference signal curve in each of these regions; , slope and time.

22. determining whether a deviation between a characteristic of said actual signal curve and a corresponding characteristic of said reference signal curve exceeds a predefined value; and a predefined cause for such deviation; 12. The method of claim 11, further comprising the steps of attributing at least one of: and generating a signal confirming said at least one predetermined cause.

23. said step of dividing said actual signal curve into candidate regions, determining the slope of said curve at points along said actual signal curve, said region changing by more than a first predefined value; 12. The method of claim 11, further comprising the steps of: determining critical points on said actual signal curve, and dividing said actual signal curve into candidate regions between these determined critical points.

24. The method according to claim 11, characterized in that plasma etching is provided in the @ligand-determined process.

25. The method according to claim 18, characterized in that said predetermined process includes plasma etching.

26. The method of claim 21, wherein said predetermined process includes plasma etching.

27. The method of claim 8, wherein the duration of the candidate region is varied to improve matching of the candidate actual endpoint trace to the region of the reference endpoint trace.

28. In monitoring the endpoint trace of a plasma etch reactor, establish a reference endpoint trace for a predetermined etching process, identify critical points during this reference endpoint trace, and identify these points. Identify the area of the reference endpoint trace between the critical points, introduce this predetermined etching process into a semiconductor device, collect the actual endpoint trace for the etch of this semiconductor device, and Based on the change in the slope of the endpoint trace, identify critical point candidates in this trace, divide this actual endpoint trace into candidate regions between the critical point candidates, and divide this actual endpoint trace into candidate regions between the critical point candidates. matching a candidate region to a region of the reference endpoint trace; comparing at least one characteristic of at least one region of the actual trace to a corresponding characteristic of a corresponding region of the reference endpoint trace; , representing whether at least one said property of at least one region of said actual endpoint trace varies by more than a predefined limit value from a corresponding property of a corresponding region of said reference endpoint trace. A method for monitoring an end point trace, comprising the step of generating a display.

29. The step of matching the candidate region of the actual end point trace with the region of the reference trace includes the step of comparing the characteristics of the candidate region of the actual trace with the characteristics of the region of the reference trace. The method according to claim 28, characterized in that:

30. The step of matching the candidate region of the actual trace to the corresponding region of the reference trace further provides a practical paper function to compare the characteristics of the candidate region with the characteristics of the reference end point trace. The method according to claim 29, characterized in that a step is provided.

31. According to the set of practical functions: the slope at the end of the candidate region; the average slope of the region; the intensity of the last point of the region; the duration of this region; and the end point from the beginning of the trace. A method characterized in that at least one of the characteristics consisting of a region and a time is compared.

32. The step of matching the candidate area of the actual end point trace with the area of the reference end point trace includes at least one area adjacent to the candidate area of the actual trace.
32. The method of claim 31, further comprising the step of comparing the characteristics of one region with the characteristics of a corresponding region adjacent to the region of the reference endpoint trace to be matched.

33. Adjusting the limit value of the candidate region to be matched beyond the identified critical point candidate based on the change in slope, whereby the candidate region of the actual trace is adjusted to the region of the reference trace. 33. The method according to claim 32, characterized in that the method further precisely matches.

34. a data storage device for receiving reference data representing a reference continuously variable signal corresponding to a predefined process; and an actual data storage representing an actual continuously variable signal curve corresponding to the actual operation of said predefined process; a processor for dividing the actual data into candidate regions based on characteristics of the actual data; and a processor for dividing the actual data into candidate regions based on characteristics of the actual data; a matcher that matches at least one candidate region of the actual data to a region of reference data by analyzing the characteristics of the at least one region of the actual data; a comparator for comparing the characteristics of the area of the reference data, which has been determined, with the area of the actual data, and a reference for matching a signal representing a characteristic variable of the area of the actual data exceeding a predetermined limit value with the area of the actual data; A process monitoring device comprising: a generator that generates data based on characteristics of correspondence between areas of data.

35. Claim 34, wherein the processor is operative to divide the actual data into candidate regions based on changes in the slope of the actual data.
The device described.

36. The apparatus according to claim 35, wherein the matcher applies a practical set of functions to the candidate region to match it to the region of the reference data.

37. The apparatus of claim 36, wherein said predetermined process includes a plasma etching operation.

38. The apparatus of claim 37, wherein the actual continuously variable signal curve includes an endpoint trace signal from a plasma eschatter.

39. In monitoring the endpoint trace of a plasma etch reactor, establishing a reference endpoint trace for a predetermined etching process, dividing the reference endpoint trace into predefined regions, implementing the predetermined etching process on a semiconductor device; collecting an actual endpoint trace for etching the semiconductor device; and dividing the actual endpoint trace into candidate regions according to predefined criteria. Furthermore, by analyzing the characteristics of the candidate region of this actual end point trace over the characteristics of the region of the reference end point trace to be matched, the candidate region of this actual end point trace is determined as described above. A method of monitoring an end point trace, comprising the step of matching a reference end point trace to a corresponding region.

40. a data store for receiving reference data representing a reference continuously variable signal curve corresponding to a plasma etching process; and collecting actual data representing an actual continuously variable signal curve corresponding to the actual operation of the plasma etching process; a processor for dividing the actual data into candidate regions based on characteristics of the actual data; and a processor for analyzing the characteristics of the candidate regions over the characteristics of the at least one region of the reference data. A process monitoring device characterized by comprising a matching device for matching one candidate region of the actual data (at least one candidate region) to the region of the reference data.

41, an improved apparatus and process is provided for detecting anomalies during manufacturing process operations. According to one embodiment, the operation of the plasma etch reactor is monitored to detect anomalies during the etch operation. A baseline end point trace (EPT) is defined for this etch process by the plasma etch diagnostic expert system (60). Regions are defined within the endpoint trace and characteristics of this criterion, and tolerances are defined for each region. The etcher operates and the actual EPT is obtained from this operation. This actual EPT is analyzed to confirm the area of this EPT, and then the area of this EPT is matched to the area of the reference EPT. The system utilizes a series of practical functions in matching the actual EPT area to the reference EPT area. The characteristics of the matched region of the actual endpoint trace are compared to the characteristics of the corresponding region of the reference endpoint trace to determine whether an anomaly has occurred during the etching process. According to the present invention, it is possible to compare an actual end point trace with a reference end point trace and achieve good matching.

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram illustrating a typical plasma etch reactor having an endpoint trace monitoring device and computer means according to an embodiment of the present invention; FIG. 2 is a reference endpoint diagram according to an embodiment of the present invention; A graph representing a trace and a specific etching process; FIG. 3 is a chart of the overall software configuration; FIG. 4 is a diagram of a configuration showing general functions defining a reference trace according to the embodiment of FIG. 3; FIG. 5 is a diagram of a configuration representing the general function of a signal-to-symbol converter; FIG. 6 is a diagram of a configuration representing a function of comparing end point traces; FIG. Graph illustrating an example of a reference endpoint trace with intensity plotted as a function; FIG. 8 is a graph illustrating an example of a first actual endpoint trace to be matched to the reference endpoint trace of FIG. , and FIG. 9 are graphs showing an example of a second actual end point trace to be matched against the reference end point trace of FIG. 10... Plasma etching reaction vessel 12...
... Electrode 20 ... Semiconductor slice 30 ... Plasma 32 ... Window 34 ... Optical filter 36 ... Photodiode 38 ... ...EPTI-Race Recorder 40...
...Computer Phrases Lesson 5 Iku Invitation

Claims (1)

Claims: Establishing a reference endpoint trace for a predetermined etching process; dividing the reference endpoint trace into predefined regions; and introducing the predetermined etching process into a semiconductor device. collecting actual end point traces for etching of the semiconductor device, dividing the actual end point traces into candidate regions according to predetermined criteria, and matching characteristics of the candidate regions of the actual end point traces. By analyzing over the characteristics of the region of the reference endpoint trace to be matched, the region of the actual endpoint trace is matched to the corresponding region of the reference endpoint trace, and further this actual endpoint trace is matched to the corresponding region of the reference endpoint trace. A method of monitoring an endpoint trace of a plasma etch reactor comprising the step of comparing characteristics of a region of the endpoint trace with corresponding characteristics of a matched region of the reference endpoint trace.