U.S. patent application number 13/067542 was filed with the patent office on 2011-10-06 for method and system for managing semiconductor manufacturing device.
This patent application is currently assigned to Kabushiki Kaisha Toshiba. Invention is credited to Masafumi Asano, Hiroshi Matsushita, Junji Sugamoto.
Application Number | 20110245956 13/067542 |
Document ID | / |
Family ID | 39528506 |
Filed Date | 2011-10-06 |
United States Patent
Application |
20110245956 |
Kind Code |
A1 |
Matsushita; Hiroshi ; et
al. |
October 6, 2011 |
Method and system for managing semiconductor manufacturing
device
Abstract
A management system includes a variable-period setting unit that
sets a variable period in which quality-control values vary. Then,
a retrieving unit retrieves events sandwiching the variable period.
The events can be a maintenance of the semiconductor manufacturing
device and/or a change of a correction value. An analysis-period
setting unit sets an analysis period for analyzing a cause of
variation of the quality-control values between the events
retrieved by the retrieving unit.
Inventors: |
Matsushita; Hiroshi;
(Kanagawa, JP) ; Sugamoto; Junji; (Oita, JP)
; Asano; Masafumi; (Kanagawa, JP) |
Assignee: |
Kabushiki Kaisha Toshiba
Tokyo
JP
|
Family ID: |
39528506 |
Appl. No.: |
13/067542 |
Filed: |
June 8, 2011 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11959968 |
Dec 19, 2007 |
7979154 |
|
|
13067542 |
|
|
|
|
Current U.S.
Class: |
700/110 |
Current CPC
Class: |
Y02P 90/02 20151101;
G05B 2219/32221 20130101; G05B 2219/45031 20130101; G05B 19/41875
20130101; G05B 2219/32179 20130101; Y02P 90/22 20151101 |
Class at
Publication: |
700/110 |
International
Class: |
G06F 19/00 20110101
G06F019/00 |
Foreign Application Data
Date |
Code |
Application Number |
Dec 19, 2006 |
JP |
2006-340937 |
Oct 30, 2007 |
JP |
2007-282277 |
Claims
1. (canceled)
2. (canceled)
3. A management system that manages a semiconductor manufacturing
device, the management system comprising: a first storage unit that
stores therein a plurality of quality-control values, the
quality-control values being values obtained by measuring a
dimension of wafers at different times during processing of the
wafers by the semiconductor manufacturing device; a second storage
unit that stores therein a plurality of equipment parameters, the
equipment parameters being values obtained by monitoring a state of
the semiconductor manufacturing device at different times during
processing of the wafers by the semiconductor manufacturing device;
a third storage unit that stores therein a maintenance log of the
semiconductor manufacturing device; a first setting unit that sets
a correction value for correcting the equipment parameters to
control the quality-control values based on the quality-control
values in the first storage unit and the equipment parameters in
the second storage unit; a fourth storage unit that stores therein
the correction value set by the first setting unit; a detecting
unit that detects an abnormality in the state of the semiconductor
manufacturing device based on monitoring of the equipment
parameters and an error detection rule, the error detection rule is
for determining an abnormality in the state of the semiconductor
manufacturing device; an acquiring unit that acquires events from
the third storage unit and the fourth storage unit, the events
including a maintenance of the semiconductor manufacturing device
stored in the third storage unit and a change of the correction
value stored in the fourth storage unit; a determining unit that
determines whether the events acquired by the second acquiring unit
are associated with the equipment parameters monitored by the
detecting unit; and an instructing unit that outputs an instruction
indicative of a necessity to update the error detection rule when
it is determined by the determining unit that the events are
associated with the equipment parameters.
4. The management system according to claim 3, wherein when it is
determined by the determining unit that the events are associated
with the parameters, the instructing unit outputs an instruction to
the detecting unit to automatically update the error detection
rule.
5. The management system according to claim 3, wherein when the
semiconductor manufacturing device is an exposure device, the
equipment parameters monitored by the detecting unit include one or
more of following: (1) synchronization accuracy between a wafer
stage and a reticle stage (2) orthogonality of a wafer (3) relative
magnification of a reticle image (4) delay in a processing between
a developing process and an exposure process, and the detecting
unit outputs an instruction indicative of a necessity to perform a
maintenance of the exposure device when detecting an error in a
resist width based on an error detection rule that is used for
detecting error in equipment parameters associated with a resist
width.
6. The management system according to claim 4, wherein when the
semiconductor manufacturing device is an exposure device, the
equipment parameters monitored by the detecting unit include one or
more of following: (1) synchronization accuracy between a wafer
stage and a reticle stage (2) orthogonality of a wafer (3) relative
magnification of a reticle image (4) delay in a processing between
a developing process and an exposure process, and the detecting
unit outputs an instruction indicative of a necessity to perform a
maintenance of the exposure device when detecting an error in a
resist width based on an error detection rule that is used for
detecting error in equipment parameters associated with a resist
width.
7. A method of managing a semiconductor manufacturing device, the
method comprising: acquiring events, the events including a
maintenance of the semiconductor manufacturing device and a change
of a correction value, the correction value being a value for
correcting equipment parameters to control quality-control values
based on the quality-control values, the equipment parameters being
values obtained by monitoring a state of the semiconductor
manufacturing device at different times during processing of wafers
by the semiconductor manufacturing device, the quality-control
values being values obtained by measuring a dimension of the wafers
at different times during processing of the wafers by the
semiconductor manufacturing device; determining whether the events
acquired at the acquiring are associated with the equipment
parameters being monitored by a detecting unit that detects an
abnormality in the state of the semiconductor manufacturing device
based on monitoring of the equipment parameters and an error
detection rule, the error detection rule is for determining an
abnormality in the state of the semiconductor manufacturing device;
and outputting an instruction indicative of a necessity to update
the error detection rule when it is determined at the determining
that the events acquired at the acquiring are associated with the
equipment parameters monitored by the detecting unit.
8. The method according to claim 7, wherein the outputting includes
outputting an instruction to the detecting unit to automatically
update the error detection rule when it is determined at the
determining that the events acquired at the acquiring are
associated with the equipment parameters monitored by the detecting
unit.
9. The method according to claim 7, wherein when the semiconductor
manufacturing device is an exposure device, the equipment
parameters monitored by the detecting unit include one or more of
following: (1) synchronization accuracy between a wafer stage and a
reticle stage (2) orthogonality of a wafer (3) relative
magnification of a reticle image (4) delay in a processing between
a developing process and an exposure process, and the detecting
unit outputs an instruction indicative of a necessity to perform a
maintenance of the exposure device when detecting an error in a
resist width based on an error detection rule that is used for
detecting error in equipment parameters associated with a resist
width.
10. The method according to claim 8, wherein when the semiconductor
manufacturing device is an exposure device, the equipment
parameters monitored by the detecting unit include one or more of
following: (1) synchronization accuracy between a wafer stage and a
reticle stage (2) orthogonality of a wafer (3) relative
magnification of a reticle image (4) delay in a processing between
a developing process and an exposure process, and the detecting
unit outputs an instruction indicative of a necessity to perform a
maintenance of the exposure device when detecting an error in a
resist width based on an error detection rule that is used for
detecting error in equipment parameters associated with a resist
width.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is based upon and claims the benefit of
priority from the prior Japanese Patent Application No.
2006-340937, filed on Dec. 19, 2006 and Japanese Patent Application
No. 2007-282277, filed on Oct. 30, 2007; the entire contents of
both of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention relates to a technology for managing a
semiconductor manufacturing device.
[0004] 2. Description of the Related Art
[0005] Better results can be obtained in a semiconductor
manufacturing process if target parameters of a process are
maintained at respective predetermined target values. For example,
in a process of laminating a film, it is necessary to laminate a
film with a desired width previously designed in the production
recipe. Furthermore, in an exposure process, it is necessary that
the dimensions after the exposure process are within an allowable
range of the design values.
[0006] However, in semiconductor manufacturing processes, one or
more target parameters may deviate from the target values due to
various external causes. For example, in the exposure process,
light exposure may change depending on a state of a lighting
optical system or a state of a reticle that transfers a circuit
pattern. Such a change of light exposure can cause dimensional
change of a semiconductor. If the target parameters vary, elements
that make up a semiconductor integrated circuit (IC) do not work in
a desirable manner. Such a semiconductor IC is considered as a
defective product and cannot be sold in the marketplace, resulting
in decreasing a production yield.
[0007] One approach is to monitor a target parameter, i.e., a
quality control (QC) value, for each process. Such monitoring
includes monitoring a physical quantity, i.e., a QC value, while a
process is being executed. Examples of QC values include resist
width in the exposure process and finished dimensions of an element
manufactured through a processing process.
[0008] It is common to monitor an internal state, so-called an
equipment engineering system (EES) data, of the semiconductor
manufacturing device by using various sensors. For example, in an
exposure device used in the exposure process, several hundreds of
EES parameters, such as light exposure, focus value, and
temperature of a developer, are acquired.
[0009] JP-A 2005-197323 (KOKAI), discloses a conventional
technology for identifying a cause of a variation of a QC value in
a semiconductor manufacturing device from EES parameters by
performing a correlation analysis on the QC value and the EES
parameters.
[0010] Recently, an advanced process control (APC) has been
developed. In the APC, a state of a manufacturing device is
controlled based on a value of the target parameter to maintain a
target parameter to a predetermined value. For example, in the
exposure process, light exposure of the exposure device is
controlled so that a resist width is maintained to a predetermined
value based on the values of the resist width measured as the QC
values. A liner relation can be seen between the light exposure and
the resist width. Therefore, if it is detected a tendency from the
QC values that the resist width become excessively wide, light
exposure is controlled so that the resist width reduces to a
desired value.
[0011] A fault detection and classification (FDC) is performed
based on the EES parameters. FDC is a method of monitoring a
parameter, such as an EES parameter, to check occurrence of a
defect, and classify the defect when a defect has occurred. In the
FDC, it is determined that a known defect has occurred when certain
EES parameters have values in a predetermined range. In other
words, in the FDC, it is necessary to prepare an FDC model for
detecting defects based on EES parameters.
[0012] Maintenance operations are often performed on the
semiconductor manufacturing device to maintain the device in a
normal state. The maintenance operation includes, for example,
cleaning of a vacuum chamber, or adjustment of various units.
Although the semiconductor manufacturing device is maintained in a
normal state by such maintenance operation, some of the EES
parameters may vary unexpectedly due to the maintenance
operation.
[0013] In the technique of identifying a cause of the change in QC
values by using EES parameters, if the state of the device changes
because an APC control or a maintenance operation is performed
during an analysis period, the true cause of the change can not be
extracted. Moreover, if the state of the device changes because an
APC control or a maintenance operation is performed, the FDC model
needs to be updated to suit the current state of the device. In
other words, if one FDC model is used before and after an APC
control or a maintenance operation, there is a possibility that a
defect is erroneously detected or even overlooked.
SUMMARY OF THE INVENTION
[0014] According to an aspect of the present invention, there is
provided a management system that manages a semiconductor
manufacturing device. The management system includes a first
storage unit that stores therein a plurality of quality-control
values, the quality-control values being values obtained by
measuring a dimension of wafers at different times during
processing of the wafers by the semiconductor manufacturing device;
a second storage unit that stores therein a plurality of equipment
parameters, the equipment parameters being values obtained by
monitoring a state of the semiconductor manufacturing device at
different times during processing of the wafers by the
semiconductor manufacturing device; a third storage unit that
stores therein a maintenance log of the semiconductor manufacturing
device; a first setting unit that sets a correction value for
correcting the equipment parameters to control the quality-control
values based on the quality-control values in the first storage
unit and the equipment parameters in the second storage unit; a
fourth storage unit that stores therein the correction value set by
the first setting unit; a second setting unit that sets a variable
period in which the quality-control values vary; a retrieving unit
that retrieves events sandwiching the variable period, the events
including a maintenance of the semiconductor manufacturing device
from the maintenance log stored in the third storage unit and a
change of the correction value in the fourth storage unit; a third
setting unit that sets an analysis period for analyzing a cause of
variation of the quality-control values between the events
retrieved by the retrieving unit; and an extracting unit that
performs statistical analysis to quantitatively calculate a
correlation between quality-control values and equipment parameters
within the analysis period, and extracts a cause of variation of
the quality-control values based on calculated correlation.
[0015] According to another aspect of the present invention, there
is provided a method of managing a semiconductor manufacturing
device. The method includes setting a variable period in which
quality-control values vary, the quality-control values being
values obtained by measuring a dimension of wafers at different
times during processing of the wafers by the semiconductor
manufacturing device; retrieving events sandwiching the variable
period, the events including a maintenance of the semiconductor
manufacturing device and a change of a correction value, the
correction value being a value for correcting equipment parameters
to control the quality-control values based on the quality-control
values, the equipment parameters being values obtained by
monitoring a state of the semiconductor manufacturing device at
different times during processing of the wafers by the
semiconductor manufacturing device; setting an analysis period for
analyzing a cause of variation of the quality-control values
between the events retrieved at the retrieving; performing
statistical analysis to quantitatively calculate a correlation
between quality-control values and equipment parameters within the
analysis period; and extracting a cause of variation of the
quality-control values based on the correlation calculated at the
performing.
[0016] According to still another aspect of the present invention,
there is provided a management system that manages a semiconductor
manufacturing device. The management system includes a first
storage unit that stores therein a plurality of quality-control
values, the quality-control values being values obtained by
measuring a dimension of wafers at different times during
processing of the wafers by the semiconductor manufacturing device;
a second storage unit that stores therein a plurality of equipment
parameters, the equipment parameters being values obtained by
monitoring a state of the semiconductor manufacturing device at
different times during processing of the wafers by the
semiconductor manufacturing device; a third storage unit that
stores therein a maintenance log of the semiconductor manufacturing
device; a first setting unit that sets a correction value for
correcting the equipment parameters to control the quality-control
values based on the quality-control values in the first storage
unit and the equipment parameters in the second storage unit; a
fourth storage unit that stores therein the correction value set by
the first setting unit; a detecting unit that detects an
abnormality in the state of the semiconductor manufacturing device
based on monitoring of the equipment parameters and an error
detection rule, the error detection rule is for determining an
abnormality in the state of the semiconductor manufacturing device;
an acquiring unit that acquires events from the third storage unit
and the fourth storage unit, the events including a maintenance of
the semiconductor manufacturing device stored in the third storage
unit and a change of the correction value stored in the fourth
storage unit; a determining unit that determines whether the events
acquired by the second acquiring unit are associated with the
equipment parameters monitored by the detecting unit; and an
instructing unit that outputs an instruction indicative of a
necessity to update the error detection rule when it is determined
by the determining unit that the events are associated with the
equipment parameters.
[0017] According to still another aspect of the present invention,
there is provided a method of managing a semiconductor
manufacturing device. The method including acquiring events, the
events including a maintenance of the semiconductor manufacturing
device and a change of a correction value, the correction value
being a value for correcting equipment parameters to control
quality-control values based on the quality-control values, the
equipment parameters being values obtained by monitoring a state of
the semiconductor manufacturing device at different times during
processing of wafers by the semiconductor manufacturing device, the
quality-control values being values obtained by measuring a
dimension of the wafers at different times during processing of the
wafers by the semiconductor manufacturing device; determining
whether the events acquired at the acquiring are associated with
the equipment parameters being monitored by a detecting unit that
detects an abnormality in the state of the semiconductor
manufacturing device based on monitoring of the equipment
parameters and an error detection rule, the error detection rule is
for determining an abnormality in the state of the semiconductor
manufacturing device; and outputting an instruction indicative of a
necessity to update the error detection rule when it is determined
at the determining that the events acquired at the acquiring are
associated with the equipment parameters monitored by the detecting
unit.
BRIEF DESCRIPTION OF THE DRAWINGS
[0018] FIG. 1 is a block diagram of a management system for
managing a semiconductor manufacturing device according to a first
embodiment of the present invention;
[0019] FIG. 2 is a flowchart of a process of managing the
semiconductor manufacturing device performed by the management
system shown in FIG. 1;
[0020] FIG. 3A is a graph of a relation between exposing time and
resist width as a QC value;
[0021] FIG. 3B is a graph of a relation between exposing time and
light exposure as an APC set value;
[0022] FIG. 3C is a graph of a relation between exposing time and
maintenance log;
[0023] FIGS. 4 to 6 are examples of display of EES parameters,
correlation coefficient, and associated event data on a display
unit of a user interface (I/F) shown in FIG. 1;
[0024] FIG. 7 is a block diagram of a management system for
managing a semiconductor manufacturing device according to a second
embodiment of the present invention;
[0025] FIG. 8 is a flowchart of a process of managing the
semiconductor manufacturing device performed by the management
system shown in FIG. 7;
[0026] FIG. 9 is an example of display of an error detection state
for each FDC model;
[0027] FIG. 10 is an example of a table containing event data
associated with an FDC model;
[0028] FIG. 11 is a scatter plot of a relation between mean of
standard deviation of Y-component of synchronization accuracy and
amount of dimensional change (absolute value);
[0029] FIG. 12 is a scatter plot of a relation between amount of
dimensional change (absolute value) and orthogonality of a
wafer;
[0030] FIG. 13 is a graph for explaining a relation among delay in
a processing after a developing process, shot magnification, and
amount of dimensional change (absolute value); and
[0031] FIG. 14 depicts examples of FDC models according to a third
embodiment of the present invention.
DETAILED DESCRIPTION OF. THE INVENTION
[0032] Exemplary embodiments of the present invention are explained
in detail below with reference to the accompanying drawings.
[0033] The following embodiments include acquiring data about APC
and maintenance as event data in a system in which target
parameters for each process are controlled so as to be constant by
performing APC during a semiconductor manufacturing process. Then,
extraction of a cause of change in the state of the semiconductor
manufacturing device, or updating of an FDC model during an
interval between events is performed by using the event data as a
trigger.
[0034] In the embodiments explained below, the present invention is
applied to an exposure process as a manufacturing process of a
semiconductor IC. However, the present invention can be applied to
other processes in the manufacturing process.
[0035] The first embodiment, data on an APC control and maintenance
are acquired as event data from a semiconductor manufacturing
device that is controlled by APC, and a period for analyzing a
cause of variation of QC values is set based on event occurrence
time. Moreover, an FDC model is updated based on occurrence of
events as a trigger.
[0036] An exposure process for exposing a gate of a transistor is
described in the first embodiment.
[0037] FIG. 1 is a block diagram of a management system 100
according to the first embodiment for managing a semiconductor
manufacturing device 2.
[0038] The semiconductor manufacturing device (e.g., an exposure
device) 2 and a QC-value measuring device 3 are accommodated in a
clean room 1 and the atmosphere in the control room 1 is controlled
accurately.
[0039] The semiconductor manufacturing device 2 receives a wafer
that has been subjected to a previous process, and it further
processes that wafer. A resist coating process is an example of the
previous process, and an exposure process is an example of the
process performed by the semiconductor manufacturing device 2.
[0040] The QC-value measuring device 3 measures QC values of the
wafer, which has been processed by the semiconductor manufacturing
device 2, and passes that wafer to a next device for a next
process. Resist width of a gate portion of a transistor is an
example of the QC value. Etching process is an example of the next
process. The QC-value measuring device 3 is, for example, a
critical dimension scanning electron microscope (SEM).
[0041] The semiconductor manufacturing device 2 is controlled by
APC. The management system 100 manages the semiconductor
manufacturing device 2. The management system 100 includes various
databases such as a production-management-information database 4, a
QC-value database 5, an EES parameter database 6, and a maintenance
information database 7.
[0042] The production-management-information database 4 contains
production management information for identifying each of the
wafers processed by the semiconductor manufacturing device 2. The
QC-value database 5 serves as a first storage unit and contains QC
values obtained by measuring dimension of a processed area of each
of the wafers processed by the semiconductor manufacturing device
2. The EES parameter database 6 serves as a second storage unit and
contains EES parameters obtained by monitoring the state of the
semiconductor manufacturing device 2. The maintenance information
database 7 serves as a third storage unit and contains maintenance
log of the semiconductor manufacturing device 2.
[0043] Data is input in each of the databases 5 to 7 from a data
collection server (not shown).
[0044] The production management information is for identifying
which lot (wafer) is being processed by the semiconductor
manufacturing device 2, and contains lot (wafer) number, brand
name, process name, processing date/time, and the like.
[0045] An example of the QC values includes a resist width
(dimension of a processed area) of a gate of a transistor exposed
in the exposure process.
[0046] The EES parameters are various data acquired by sensors (not
shown) provided in each unit of the semiconductor manufacturing
device 2, and they are the keys to know the internal state of the
semiconductor manufacturing device 2. In a typical semiconductor
manufacturing device, it is possible to collect about 200 types of
the EES parameters such as actual light exposure, follow focus
capability, synchronization accuracy, chamber temperature, chamber
pressure, and inclination of axis.
[0047] Maintenance information is a log data indicating when and
what type of maintenance operation was performed on the
semiconductor manufacturing device 2. The maintenance information
contains data on maintenance target device, maintenance time,
particulars of maintenance, and the like.
[0048] The management system 100 also includes a user interface
(I/F) 8 and an APC setting device 9. The user I/F 8 displays
various data on a display unit (not shown) and outputs various
control signals in response to operations by an operator of the
management system 100. The APC setting device 9 serves as a first
setting unit and generates an APC set value (a correction value)
that is used for correcting an EES parameter to control the QC
values based on the production management information, the
distribution of the QC values, and the EES parameters.
[0049] The QC values stored in the QC-value database 5 are
displayed in the form of a time-series graph on the display unit of
the user I/F 8.
[0050] The APC setting device 9 acquires an EES parameter, such as
actual light exposure in the exposure process, and predicts a QC
value for a target lot based on QC values of past five lots. The
APC setting device 9 then calculates an APC set value corresponding
to the acquired EES parameter (light exposure) based on the
predicted QC value to achieve a desired resist width, and outputs
the calculated APC set value to the semiconductor manufacturing
device 2. The semiconductor manufacturing device 2 uses the APC set
value received from the APC setting device 9 in the exposure
process for the target lot. The APC set value is calculated by
using a table containing items of previously measured actual EES
parameter (light exposure) in association with a QC value (resist
width).
[0051] The management system 100 also includes an APC-set-value
database 10 and a computer (central processing unit (CPU)) 11. The
APC-set-value database 10 serves as a fourth storage unit and
contains an APC set value generated by the APC setting device 9.
The CPU 11 outputs information in response to a control signal
received from the user I/F 8. The databases 4 to 7 and 10 can be
realized by using a magnetic disk or other computer-readable
recording media.
[0052] The CPU 11 includes a QC-value variable-period setting unit
11a, an event-data retrieving unit lib, an analysis-period setting
unit 11c, and a cause extracting unit lid. The QC-value
variable-period setting unit 11a serves as a second setting unit
and sets a QC-value variable period in which a QC values varies
based on the QC values and a control signal received from the user
I/F 8. The event-data retrieving unit 11b searches the maintenance
information database 7 and the APC-set-value database 10, and
retrieves events, such as maintenance of the semiconductor
manufacturing device 2, or change in the APC set value, that took
place just before/after the QC-value variable period.
[0053] Examples of the event data include time at which the APC set
value was changed, or time at which a maintenance operation was
conducted.
[0054] The analysis-period setting unit 11c serves as a third
setting unit and sets a period between the retrieved events that
sandwich the QC-value variable period as an analysis period. The
cause extracting unit 11d performs correlation analysis between the
QC values and the EES parameters during the analysis period, and
outputs the calculated correlation coefficient in association with
the EES parameter to which the correlation analysis has been
conducted to the user I/F 8. The user I/F 8 displays the
correlation coefficient in association with the EES parameter on
the display unit.
[0055] FIG. 2 is a flowchart of a process of managing performed by
the management system 100 when managing the semiconductor
manufacturing device 2. To begin with, the QC-value variable-period
setting unit 11a sets a QC-value variable period based on the QC
values present in the QC value database 5 and a control signal
received from the user I/F 8 (step S1).
[0056] The event-data retrieving unit 11b retrieves event data
sandwiching the QC-value variable period (step S2). The event data
is information on events such as a maintenance or a change of an
APC set value. Information on whether a maintenance was performed
can be obtained by searching the maintenance information database
7, and information on whether an APC set value was changed can be
obtained by searching the APC-set-value database 10. The events
sandwiching the QC-value variable period are a pre-even that
occurred immediately before a start of the QC-value variable period
and a post-even that occurred immediately after an end of the
QC-value variable period.
[0057] The analysis-period setting unit 11c sets an analysis period
(step S3). The analysis period starts from a time point at which
the pre-event occurred and ends at a time point at which the
post-event occurred. In other words, the analysis period includes
the QC-value variable period.
[0058] The cause extracting unit 11d performs a correlation
analysis between the QC values and the EES parameters during the
analysis period, and identifies an EES parameter having a
correlation coefficient larger than a predetermined target value as
a cause of variation of the QC-values (step S4).
[0059] The cause extracting unit 11d outputs the calculated
correlation coefficient associated with the EES parameter to the
user I/F 8, and the user I/F 8 displays them on the display unit
(step S5). Thus, the operator can understand the cause of variation
of the QC values by looking at the information displayed on the
display unit.
[0060] The event-data retrieving unit 11b determines whether event
data related to the EES parameter is present during the analysis
period (step S6). Whether event data during the analysis period is
present can be decided by searching the maintenance information
database 7 and the APC-set-value database 10. When such an event
data is present, the CPU 11 outputs the event data to the user I/F
8, so that the user I/F 8 displays that event data in associated
manner with the cause of variation of the QC values (step S7). On
the other hand, when there is no such event data, the process
control ends.
[0061] With the operation described above, it is possible to
retrieve a true cause of variation of the QC values excluding a
cause of variation due to events.
[0062] FIG. 3A is a graph of a relation between an exposing time
and a resist width as a QC value; FIG. 3B is a graph of a relation
between an exposing time and light exposure as an APC set value;
and FIG. 3C is a graph of a relation between an exposing time and a
maintenance log according to the first embodiment.
[0063] The graph shown in FIG. 3A is displayed on the display unit
of the user I/F 8 in accordance with output of data from the
QC-value database 5. The resist width shown in FIG. 3A is a
measured value of a resist width of a transistor gate processed in
the exposure process. For example, five points are selected per
wafer, their widths are measured, and a mean of measured width is
plotted as the resist width for each wafer. Each resist width
measured at each exposure process time is sequentially plotted as a
point along the horizontal axis (exposing time), and those points
are connected by a line.
[0064] In the example shown in FIG. 3A, a desired resist width is
100 nanometers, and an allowable range of the resist width is from
95 nanometers to 105 nanometers. It can be seen from FIG. 3A that
the resist width increased after time t1, and exceeded the
allowable range.
[0065] The APC setting device 9 predicts that the resist width of
the subsequent lots is likely to increase based on the fact that
the resist width (QC value) is increasing at time t2. Accordingly,
the APC setting device 9 makes a correction to increase the light
exposure (one of the EES parameters) to control the resist width to
be the desired value (i.e., the APC set value is set) (see FIG.
3B). As a result, the resist width is controlled to the desired
value, i.e., within the allowable range.
[0066] At time t3, an engineer conducts a maintenance operation on
the semiconductor manufacturing device 2. For example, the engineer
adjusts a focus system of the semiconductor manufacturing device 2
(see FIG. 3C). As a result, focus value changes, making the resist
width thinner than the desired value.
[0067] At time t4 until which a thin resist width has been
continued for five lots or more, the APC setting device 9 decreases
the light exposure (see FIG. 3B) to control the resist width to be
the desired value. As a result, the resist width increases and it
is near the desired value.
[0068] Because dimension of a transistor gate largely affects the
characteristics of the transistor, it is preferable to maintain the
resist width at the desired value.
[0069] It is examined below why dimensional change occurs from time
t1 to time t2 as shown in FIG. 3A.
[0070] An operation for extracting EES parameters that could be the
cause for the dimensional change is performed. A defected portion
of the exposure device that is the cause of the dimensional change
is identified from the extracted EES parameters.
[0071] Then, a correlation analysis is performed for extracting a
cause of variation of the dimensional change. A correlation
coefficient R between the resist width (QC value) and each of the
EES parameters can be obtained from Equation (1):
R = 1 n i ( x i - .mu. x ) ( y i - .mu. y ) ( i x i 2 n - .mu. x 2
) ( i y i 2 n - .mu. y 2 ) ( 1 ) ##EQU00001##
[0072] A mean of measured five points per wafer is used. In
Equation (1), x.sub.i is a value of the EES parameter of i-th
wafer, y.sub.i is the QC value of the i-th wafer, .mu..sub.x is a
mean of x.sub.i, .mu..sub.y is a mean of y.sub.i, and n is the
total number of wafers. Specifically, x.sub.i and y.sub.i are means
for each wafer, while .mu..sub.X and .mu..sub.y are means of the n
numbers of wafers.
[0073] An EES parameter having a correlation coefficient R obtained
from Equation (1) equal to or larger than 0.6 is extracted as an
EES parameter related to the change of the resist width (QC
value).
[0074] In such a correlation analysis, where the analysis period is
taken affects the result.
[0075] Specifically, if the correlation analysis is performed for
the entire period, it is difficult to extract the cause of
variation; because, an EES parameter having strong correlation with
the dimensional value may be different for each period.
[0076] For example, at time t2 and time t4 shown in FIG. 3A, the
light exposure controlled by the APC setting device 9 will be
extracted as the EES parameter as such having strong correlation
with the dimensional change as shown in FIG. 3B. On the other hand,
at time t3, the focus value that has been changed due to a
maintenance operation will be extracted as the most related EES
parameter as shown in FIG. 3C.
[0077] As can be seen from FIGS. 3A to 3C, there is no common EES
parameter having correlation with the dimensional value for the
entire period. Therefore, if the correlation analysis is performed
for the entire period, no EES parameter will be extracted.
[0078] According to the first embodiment, an operator specifies
(step S1 in FIG. 2) a variation for which the cause is to be
extracted. For example, the operator can point the variation for
which the cause is to be extracted on the graph FIG. 3A that is
displayed on the display unit of the user I/F 8.
[0079] For example, if the operator specifies a portion (period)
between times t1 and t2, where the resist width is increasing, in a
graph shown in FIG. 3A, the event-data retrieving unit 11b
retrieves event data sandwiching the specified period (step S2 in
FIG. 2). As described above, the event data contains time at which
the APC setting device 9 changes an APC set value and time when the
maintenance operation is performed.
[0080] In the examples shown FIGS. 3A to 3C, the event-data
retrieving unit lib retrieves that the APC setting device 9 has
changed the APC set value at times t0 and t2 (see FIG. 3B). The
analysis-period setting unit 11c then sets the analysis period
between times t0 and t2 (step S3 in FIG. 2). The analysis period is
set between the events sandwiching the period in which the target
dimensional change occurs. Furthermore, it is preferable to set the
analysis period as long as possible unless the analysis period
includes extracted events from a view of correlation analysis.
[0081] The cause extracting unit 11d then acquires the QC values
and the EES parameters from each database for the set analysis
period, and calculates a correlation coefficient from Equation (1)
(step S4 in FIG. 2).
[0082] FIGS. 4 to 6 are examples of EES parameters, correlation
coefficients, and associated event data that are displayed on the
display unit of the user I/F 8.
[0083] As shown in FIG. 4, a result of extraction of a cause of
variation of the QC values is displayed on the display unit,
indicating that Y-component of synchronization accuracy has the
largest correlation coefficient of 0.85. As a result, the operator
can decide that the Y-component of the synchronization accuracy is
the cause of variation for the analysis period (step S5 in FIG.
2).
[0084] As described above, the operator can find a true cause of
variation excluding a cause of variation due to events; because,
such analysis period is set in such a manner that events
sandwiching the analysis period are automatically retrieved. Thus,
it is possible to better manage the semiconductor manufacturing
device 2 based on the true cause.
[0085] When the operator specifies time t2 or time t3 shown in FIG.
3A, events sandwiching a period where the QC values vary are
retrieved, and a correlation analysis is performed in the same
manner.
[0086] For example, assume now that the operator specifies a period
including time t2 but not including time t1 and time t3. As shown
in FIG. 3B, facts that the APC setting is changed at time t0 and
time t2 and a maintenance operation is performed at time t3 are
extracted as events, and the analysis period is set between time t0
and time t3.
[0087] As shown in FIG. 5, a result of extraction of a cause of
variation of the QC values, and event data at time t2 as event data
present during the analysis period are displayed on the display
unit of the user I/F 8. Assume now that a table (not shown)
containing event data in association with the EES parameter is
provided, e.g., the EES parameter name associated with light
exposure is also associated with an event indicative of change in
the APC set value. At this point, if there is association between
the event data and the target EES parameter in the table (not
shown), the cause extracting unit 11d displays associated event
data on the display unit of the user I/F 8 (steps S6 and S7 in FIG.
2).
[0088] When the operator specifies time t3 shown in FIG. 3A,
maintenance data of a focus system is displayed as the event data
at time t3 in the similar manner as shown in FIG. 6.
[0089] As described above, if the event data is displayed in
addition to the EES parameter as shown in FIGS. 5 and 6, the
operator can find out that the dimensional change is caused by
extracted known events.
[0090] Instead of the operator setting the QC-value variable
period, the QC-value variable period can be set automatically. For
example, it is possible to previously set a threshold, and
automatically extract a period where the level of a change of the
QC values exceeds the threshold as the QC-value variable period. In
this case, the QC-value variable-period setting unit 11a can
automatically set the QC-value variable period without the need for
any instruction from the operator via the user I/F 8.
[0091] As described above, according to the first embodiment, when
analyzing the cause of variation of the QC values from the EES
parameter, event data sandwiching the QC-value variable period is
automatically retrieved, and the analysis period is set between the
retrieved events. As a result, it is possible to extract a true
cause of variation of the QC values excluding a cause of variation
due to events. Furthermore, an event occurs during the QC-value
variable period, it is possible to identify that the event has
caused a change of the QC values by examining association between
the event and the EES parameter. Thus, it is possible to detect an
error depending on events, such as the APC control or the
maintenance operation, during a semiconductor manufacturing
process.
[0092] The cause of variation of the QC values is identified based
on a correlation coefficient calculated from Equation (1) between
the QC values and the EES parameters. However, if it is possible to
quantitatively calculate a correlation between change of the QC
values and the EES parameter, a calculation method is not limited
to the correlation analysis and other statistical analysis or
statistical method can be used. For example, a partial least square
(PLS) analysis, or a principal component analysis can be used to
calculate the correlation. Furthermore, an EES parameter as a cause
of variation is extracted by using a univariate correlation
analysis. Specifically, it is assumed that a change of the QC
values occurs due to a single EES parameter. However, there is a
possibility that the QC values vary due to a plurality of EES
parameters. In this situation, it is possible to use a multivariate
analysis instead of the univariate analysis.
[0093] A second embodiment of the present invention is described
below. Particularly, a semiconductor manufacturing device according
to the second embodiment is controlled by APC, data on an APC
control and maintenance are acquired as event data, and an FDC
model is updated based on occurrence of events as a trigger.
[0094] FIG. 7 is a block diagram of a management system 200
according to the second embodiment for managing the semiconductor
manufacturing device 2. Components assigned with the same reference
numbers as those shown in FIG. 1 are configured similarly to those
described in the first embodiment.
[0095] The semiconductor manufacturing device 2 is controlled by
APC. The management system 200 manages the semiconductor
manufacturing device 2. The management system 200 includes various
databases such as the production-management-information database 4,
the QC-value database 5, the EES parameter database 6, and the
maintenance information database 7.
[0096] The production-management-information database 4 contains
production management information for identifying each of the
wafers processed by the semiconductor manufacturing device 2. The
QC-value database 5 serves as a first storage unit and contains QC
values obtained by measuring dimension of a processed area of each
of the wafers processed by the semiconductor manufacturing device
2. The EES parameter database 6 serves as a second storage unit and
contains EES parameters obtained by monitoring the state of the
semiconductor manufacturing device 2. The maintenance information
database 7 serves as a third storage unit and contains maintenance
log of the semiconductor manufacturing device 2.
[0097] Data is input in each of the databases 5 to 7 from a data
collection server (not shown).
[0098] The management system 200 also includes the user I/F 8 and
the APC setting device 9. The user I/F 8 displays various data on a
display unit (not shown) and outputs various control signals in
response to operations by an operator of the management system 100.
The APC setting device 9 serves as a first setting unit and
generates an APC set value (a correction value) that is used for
correcting an EES parameter to control the QC values based on the
production management information, the distribution of the QC
values, and the EES parameters.
[0099] The management system 200 also includes the APC-set-value
database 10, a first FDC-model unit 12, and a second FDC-model unit
13. The APC-set-value database 10 serves as the fourth storage unit
and contains an APC set value generated by the APC setting device
9. The first FDC-model unit 12 serves as a first device-error
detecting unit, while the second FDC-model unit 13 serves as a
second device-error detecting unit.
[0100] The first FDC-model unit 12 monitors a value of a first EES
parameter stored in the EES parameter database 6, detects an
abnormality of the semiconductor manufacturing device 2 based on a
first FDC model as a first error-detection rule for determining an
error of the semiconductor manufacturing device 2. When an
abnormality of the semiconductor manufacturing device 2 is
detected, the first FDC-model unit 12 outputs information on the
detected abnormality to the user I/F 8 so that the display unit of
the user I/F 8 displays that information.
[0101] The second FDC-model unit 13 monitors a value of a second
EES parameter stored in the EES parameter database 6, detects an
abnormality of the semiconductor manufacturing device 2 based on a
second FDC model as a second error-detection rule for determining
an error of the semiconductor manufacturing device 2. When an
abnormality of the semiconductor manufacturing device 2 is
detected, the second FDC-model unit 13 outputs information on the
detected abnormality to the user I/F 8 so that the display unit of
the user I/F 8 displays that information.
[0102] Specifically, the first FDC-model unit 12 and the second
FDC-model unit 13 issue commands indicating that a defective event
previously set as an error is occurring in the semiconductor
manufacturing device 2 when an input EES parameter is a certain
value or changes in a certain way by using the production
management data, the QC-value data, and the EES parameter.
[0103] Assuming that the semiconductor manufacturing device 2 is an
etching device, and that an abnormality that produces dust occurs
in a chamber of the semiconductor manufacturing device 2 if
pressure in the chamber during an etching process changes suddenly.
In this example, the first FDC-model unit 12 continuously monitors
change of pressure in the chamber during the etching process, and,
when the amount of change exceeds a threshold level, outputs a
warning indicating that an abnormality that causes generation of
dust has occurred to the user I/F 8. The user I/F 8 causes the
display unit to display the warning.
[0104] The management system 200 includes a table 14 and a computer
(CPU) 211. The table 14 contains, in an associated manner, data on
the first FDC-model unit 12 and events monitored by the first
FDC-model unit 12 that are associated with (cause effect on) the
first EES parameter, and data on the second FDC-model unit 13 and
events monitored by the second FDC-model unit 13 that are
associated with (cause effect on) the second EES parameter. The CPU
211 outputs data to the user I/F 8.
[0105] The CPU 211 includes an event-data acquiring unit 211a, an
event-data determining unit 211b, and an update instructing unit
211c. Specifically, the management system 200 includes the
event-data acquiring unit 211a, the event-data determining unit
211b, and the update instructing unit 211c that are realized by
executing the CPU 211.
[0106] The event-data acquiring unit 211a acquires, as event data,
maintenance information of the semiconductor manufacturing device 2
from the maintenance information database 7 and acquires
information on a change of the APC set value from the APC-set-value
database 10.
[0107] The event-data determining unit 211b determines whether the
event data acquired by the event-data acquiring unit 211a is
associated with one of the first FDC-model unit 12 (or the first
EES parameter monitored by the first FDC-model unit 12), the second
FDC-model unit 13 (or the second EES parameter monitored by the
second FDC-model unit 13), and others based on data in the table
14.
[0108] The update instructing unit 211c outputs to the user I/F 8
an instruction for updating the first FDC model as the first
error-detection rule when it is determined that the event data
acquired by the event-data acquiring unit 211a is associated with
the first FDC-model unit 12 (or the first EES parameter monitored
by the first FDC-model unit 12). The display unit of the user I/F 8
displays that instruction for the operator.
[0109] On the other hand, the update instructing unit 211c outputs
to the user I/F 8 an instruction for updating the second FDC model
as the second error-detection rule when it is determined that the
event data acquired by the event-data acquiring unit 211a is
associated with the second FDC-model unit 13 (or the second EES
parameter monitored by the second FDC-model unit 13). The display
unit of the user I/F 8 displays that instruction for the
operator.
[0110] Furthermore, when it is determined that the event data
acquired by the event-data acquiring unit 211a is associated with
the first FDC-model unit 12 (or the first EES parameter monitored
by the first FDC-model unit 12), the update instructing unit 211c
instructs the first FDC-model unit 12 to automatically update the
first FDC model as the first error-detection rule.
[0111] Moreover, when it is determined that the event data acquired
by the event-data acquiring unit 211a is associated with the second
FDC-model unit 13 (or the second EES parameter monitored by the
second FDC-model unit 13), the update instructing unit 211c
instructs the second FDC-model unit 13 to automatically update the
second FDC model as the second error-detection rule.
[0112] It is possible to set whether the update instructing unit
211c outputs an instruction for updating the first FDC model to the
user I/F 8 for displaying the instruction on the display unit, or
the update instructing unit 211c instructs the first FDC-model unit
12 to automatically update the first FDC model, depending on a type
of the first EES parameter monitored by the first FDC model.
Similarly, it is possible to set whether the update instructing
unit 211c outputs an instruction for updating the second FDC model
to the user I/F 8 for displaying the instruction on the display
unit, or the update instructing unit 211c instructs the second
FDC-model unit 13 to automatically update the second FDC model,
depending on a type of the second EES parameter monitored by the
second FDC model.
[0113] FIG. 8 is a flowchart of a process of managing performed by
the management system 200 when managing the semiconductor
manufacturing device 2. To begin with, the event-data acquiring
unit 211a acquires, as event data, maintenance information of the
semiconductor manufacturing device 2 and information on change of
the APC set value from the maintenance information database 7 and
the APC-set-value database 10 (step S21).
[0114] The event-data determining unit 211b determines whether the
event data acquired by the event-data acquiring unit 211a is
associated with the first FDC-model unit 12 (or the first EES
parameter monitored by the first FDC-model unit 12), or the second
FDC-model unit 13 (or the second EES parameter monitored by the
second FDC-model unit 13) based on the data in the table 14 (step
S22).
[0115] When it is determined that the event data acquired at step
S21 is associated with the first FDC-model unit 12 (or the first
EES parameter monitored by the first FDC-model unit 12) (Yes at
step S23), the update instructing unit 211c determines whether to
automatically update the first FDC model based on an associated FDC
model determined at step S23 (step S24). Similarly, when it is
determined that the event data acquired at step S21 is associated
with the second FDC-model unit 13 (or the second EES parameter
monitored by the second FDC-model unit 13) (Yes at step S23), the
update instructing unit 211c determines whether to automatically
update the second FDC model based on an associated FDC model
determined at step S23 (step S24).
[0116] When it is determined to perform an automatic update (Yes at
step S24), process control proceeds to step S25, so that the update
instructing unit 211c automatically updates the first or the second
FDC model to a corresponding FDC model unit, and outputs to the
user I/F 8 data indicating that automatic update has been
instructed. Then, process control ends. The display unit of the
user I/F 8 displays that data for the operator.
[0117] On the other hand, when it is determined not to perform an
automatic update (No at step S24), process control proceeds to step
S26, so that the update instructing unit 211c outputs to the user
I/F 8 data instructing a manual update of the first or the second
FDC model. The display unit of the user I/F 8 displays that
instruction for the operator. Upon viewing seeing the instruction
displayed on the display unit, the operator recognizes a need for
manually updating the FDC model, and changes the FDC model as
appropriate.
[0118] When the update instructing unit 211c determines at step S23
that the acquired event is not associated with either one of the
first FDC-model unit 12 (or the first EES parameter monitored by
the first FDC-model unit 12) and the second FDC-model unit 13 (or
the second EES parameter monitored by the second FDC-model unit
13), process control ends.
[0119] As described above, according to the second embodiment, when
an event that requires update of the error-detection rule (FDC
model) occurs, it is possible to appropriately update a
corresponding FDC model and prevent misinformation and overlook of
an error.
[0120] An example of applying the management method according to
the second embodiment is described below. Similar to the first
embodiment, the management method is applied to the exposure
process of a transistor gate.
[0121] FIG. 9 is an example of display of an error detection state
for each FDC model on the display unit of the user I/F 8; and FIG.
10 is an example of contents of the table 14 containing event data
in association with an FDC model.
[0122] As assumed in the first embodiment in connection with FIG.
4, Y-component of the synchronization accuracy is identified as a
cause of a resist width error. In other words, dimensional error
occurs when Y-component of the synchronization accuracy exceeds a
predetermined value.
[0123] Assume now that an algorithm is installed in the first
FDC-model unit 12 to continuously monitor Y-component of the
synchronization accuracy as an FDC model in the exposure process,
and when the value of the Y-component of the synchronization
accuracy exceeds a previously set threshold, an warning indicative
of occurrence of dimensional error is issued.
[0124] Similarly, it is assumed that an algorithm is installed in
the second FDC-model unit 13 to detect occurrence of a dimensional
error when a value of a focus tracking parameter drops below a
predetermined value.
[0125] An error detection state for each of the first FDC-model
unit 12 or the second FDC-model unit 13 is displayed on the display
unit of the user I/F 8 in the manner shown in FIG. 9). The level of
an error is classified into one of three levels of good (i.e., no
attention is required because the corresponding value is within an
allowable range), caution needed, and bad (i.e., immediate
attention is required because the corresponding value is out of the
allowable range), depending on the level of excess from the
threshold. A solid black circle is displayed for the appropriate
error level. A middle circle between good and bad corresponds to
caution needed in the example shown in FIG. 9.
[0126] Upon seeing a warning displayed on the display unit of the
user I/F 8, the operator determines whether to suspend a production
line to perform examination and maintenance.
[0127] Alternatively, the operator just suspends the production
line, i.e., without making a determination as to whether to suspend
the production line.
[0128] It should be noted that the error-detection rule (FDC model)
needs to be reconsidered when a state of the semiconductor
manufacturing device 2 changes due to occurrence of an event such
as a maintenance operation.
[0129] For example, the error-detection rule for the
synchronization accuracy needs to be reconsidered and changed when
adjustment of a synchronization mechanism between a wafer stage and
a reticle stage is performed. Similarly, the error-detection rule
for focus tracking needs to be reconsidered and changed when
adjustment of a focus system is performed.
[0130] If the error-detection rule, i.e., the FDC model, is
continuously used without reconsidering, it is possible to cause
erroneous detection of an error or overlook of an error.
[0131] The event-data acquiring unit 211a acquires event data
associated with the semiconductor manufacturing device 2 (step S21
in FIG. 8). The event data contains, for the exposure device,
information on change of the APC set value in relation to the light
exposure and maintenance information of the exposure device.
[0132] The event-data determining unit 211b refers to the event
data acquired by the event-data acquiring unit 211a and the table
14, and determines whether there is occurrence of an event
associated with a currently working FDC model unit (or the EES
parameter monitored by the FDC model or a currently working
error-detection rule) (steps S22 and S23 in FIG. 8).
[0133] Types of event data associated with each FDC model unit (the
EES parameter or the error-detection rule) is stored in the table
14 in the form of the table shown in FIG. 10. It can be seen from
FIG. 10 that the first FDC-model unit 12 is associated with the
synchronization accuracy, and the second FDC-model unit 13 is
associated with the event in relation to a focus.
[0134] For example, the event-data acquiring unit 211a detects that
an event of a maintenance operation has occurred at time t3 shown
in FIG. 3A, and recognizes that the event is associated with a
focus due to the fact that the maintenance operation is an
adjustment of a focus system. The event-data determining unit 211b
determines that the second FDC-model unit 13 that is detecting the
dimensional error by using the follow focus parameter (EES
parameter) corresponds to the event on a focus from the table shown
in FIG. 10.
[0135] Upon receiving a determination result from the event-data
determining unit 211b, the update instructing unit 211c outputs an
update instruction. For example, the update instructing unit 211c
sends the update instruction indicative of a fact that an update of
the error-detection rule (FDC model) is required to the user I/F 8,
and the display unit of the user I/F 8 displays the update
instruction for the operator in the manner shown in FIG. 9. At this
point, if the update type is specified as a manual update in the
table shown in FIG. 10, the update instructing unit 211c refers to
that table and displays that "suspended, need manual update" on the
display unit of the user I/F 8. At the same time, the update
instructing unit 211c instructs the target FDC model unit (in the
example shown in FIG. 10, the second FDC-model unit 13) to suspend
operation of the target FDC model (step S26 in FIG. 8). There is no
indication on any one of good, caution needed, and bad for the
second FDC-model unit 13 as shown in FIG. 9; because, the second
FDC-model unit 13 is suspended.
[0136] Upon receiving an instruction to suspend, the FDC model unit
ends an error determination process. The operator recognizes an
instruction displayed on the display unit as shown in FIG. 9, and
manually updates the target error-detection rule (FDC model). For
the event at time t3 shown in FIG. 3A, the operator checks change
of the follow focus parameter after maintenance of the focus
system, and sets a determination condition suitable for detecting a
dimensional error as a new error-detection rule (FDC model) to the
second FDC-model unit 13.
[0137] On the other hand, when the type of an update is set as an
automatic update in the table shown in FIG. 10, the update
instructing unit 211c displays "update needed, automatically
updating" instead of "suspended, need manual update" on the screen
shown in FIG. 9, and issues instruction indicative of an automatic
update to the target FDC model unit (steps S24 and S25 in FIG. 8).
The FDC model unit that received the instruction updates the
error-detection rule in accordance with the previously installed
algorithm.
[0138] For example, upon receiving the instruction for updating,
the first FDC-model unit 12 suspends the error detection process,
and automatically sets a threshold suitable for detecting the
dimensional error based on a relation between Y-component of the
synchronization accuracy and the resist width by using data for ten
lots (i.e., automatically updates the error-detection rule). After
completion of the setting, the first FDC-model unit 12 restarts the
error detection process.
[0139] The update instruction unit 211c deletes a corresponding
displayed data indicating an update state from displayed on the
display unit (see FIG. 8) after the manual or the automatic update
is completed.
[0140] As described above, according to the second embodiment,
event data is acquired, and a corresponding error-detection rule
(FDC model) can be appropriately updated when it is needed in
accordance with a change in the state of the device due to
occurrence of an event. Thus, it is possible to prevent erroneous
information, an erroneous detection, and overlook of an error.
[0141] Specifically, it is possible to effectively use the
error-detection rule (FDC model) in accordance with a change in
state of the device due to occurrence of an event. Therefore, an
error can be detected in accordance with the APC control and the
maintenance. As similar to the first embodiment, the
error-detection rule (FDC model) is updated in accordance with a
change of a state of the device due to occurrence of an event.
Thus, such an update is performed between events.
[0142] Although it is explained that the two FDC model units are
arranged in the management system 200 (i.e., the first FDC-model
unit 12 and the second FDC-model unit 13), it is possible to
arrange three or more of the FDC model units depending on the
number of EES parameters to be monitored.
[0143] A third embodiment of the present invention is described
with reference to FIGS. 11 to 14. As the third embodiment, an
exposure-device parameter that changes the resist width is taken as
an example, and an FDC model as the error-detection rule is
configured based on those parameters. The exposure-device
parameters are the same as those described in the first and the
second embodiments.
[0144] The following examination was performed as assumption of the
third embodiment: the management system having the same
configuration as that in the first embodiment was operated for one
year; a plurality of the exposure parameters that change a value of
resist width were extracted; and those extracted parameters were
further examined for their relation to the resist width.
[0145] Examples of the exposure-device parameters include
synchronization accuracy, orthogonality of a wafer, shot
magnification, delay in processing after a developing process as
will be described in detail. Other extracted exposure-device
parameters are just listed in addition to the above parameters.
Correlation between the exposure-device parameter and the resist
width has not been recognized in the conventional techniques, while
such a correlation was recognized with the management system
100.
[0146] The synchronization accuracy is explained below. As a result
of operation of the management system 100, it can be seen that
there is a correlation between a mean of standard deviation of
Y-component of the synchronization accuracy and a distribution
width of a first distribution layer. The synchronization accuracy
is associated with a follow (synchronization) accuracy between a
wafer stage and a reticle stage, i.e., associated with Y component
of the synchronization accuracy from X component and Y component
set on the surface of the wafer stage. The synchronization accuracy
is a mean of standard deviation of Y-component of the
synchronization accuracy. The mean of the standard deviation of
Y-component of the synchronization accuracy is obtained as follows:
there is a fact that the exposure process to a single wafer is
separated into a plurality of shot exposure; a predetermined number
of the shot exposure is put into a group; standard deviation of
Y-component of the synchronization accuracy for each group is
calculated; and a mean of obtained standard deviation is calculated
for the wafer.
[0147] Although Y-component of the synchronization accuracy is
described below, X-component is similar to the Y-component, so that
a mean of the standard deviation of the X-component of the
synchronization accuracy can be used as the exposure-device
parameter.
[0148] In the actually-examined exposure process, a target value of
a distribution width of a first distribution layer is 150
nanometers, while an allowable range in design is from 140
nanometers to 160 nanometers. FIG. 11 is a scatter plot of a
relation between a mean of standard deviation of Y-component of the
synchronization accuracy and an amount of dimensional change
(absolute value). The amount of dimensional change (absolute value)
means an absolute value of amount of change from the target value
of the distribution width. Each point is plotted for each wafer on
the scatter plot. It can be seen from FIG. 11 that as a mean of the
standard deviation of Y-component of the synchronization accuracy
increases, a difference between the amount of dimensional change
(absolute amount) and the target value increases. When the mean of
the standard deviation of Y-component of the synchronization
accuracy exceeds 5 nanometers, the amount of dimensional change
(absolute amount) exceeds 10 nanometers, exceeding the allowable
range in design.
[0149] A management system according to the third embodiment is
provided with a third FDC model and a third FDC model unit in
addition to the configuration of the management system 200. The
third FDC model issues a warning when the mean of the standard
deviation of Y-component of the synchronization accuracy exceeds 5
nanometers. The third FDC model unit is such that in which the
third FDC model is installed. Upon receiving the mean of the
standard deviation of Y-component of the synchronization accuracy,
the third FDC model detects an error that causes the distribution
width of the first distribution layer to be changed equal to or
more than 10 nanometers. At the same time, an instruction is issued
to the operator to perform maintenance of the wafer stage and the
reticle stage based on the parameter of the exposure device used
for error detection (i.e., the mean of the standard deviation of
Y-component of the synchronization accuracy). Specifically, a
maintenance instruction is displayed on the display unit of the
user I/F 8, and the operator conducts maintenance in accordance
with the displayed instruction. The synchronization accuracy is
registered as event data associated with the third FDC model.
[0150] As shown in FIG. 10, the first FDC model is explained as an
FDC model in relation to the synchronization accuracy according to
the second embodiment. However, for detailed explanation of such an
FDC model in distinction from the first FDC model, the third FDC
model is employed in the third embodiment.
[0151] As another example of the extracted exposure-device
parameter, the orthogonality of a wafer is described below. As a
result of an operation of the management system 100, it can be seen
that there is a correlation between the orthogonality of a wafer
and a gate dimension. The orthogonality of a wafer means a mean of
orthogonality measured per shot of exposure to a wafer.
[0152] In the actually-operated exposure process, the target value
of the gate dimension is 100 nanometers, while allowable range in
design is from 95 nanometers to 105 nanometers. FIG. 12 is a
scatter plot of a relation between an amount of dimensional change
(absolute value) and orthogonality of a wafer. Each point in the
scatter plot corresponds to a separate wafer. As the orthogonality
of a wafer increases/decreases from zero, amount of dimensional
change increases. When the absolute value of the orthogonality of a
wafer exceeds 0.1 microradian (.mu.rad), the amount of dimensional
change (absolute value) exceeds 5 nanometers, so that the gate
dimension exceeds the allowable range.
[0153] A fourth FDC model that issues a warning when the absolute
value of the orthogonality of a wafer exceeds 0.1 microradian is
prepared. Moreover, a fourth FDC model unit in which the fourth FDC
model is installed is arranged in the configuration of the
management system according to the third embodiment. Upon receiving
the orthogonality of a wafer, the fourth FDC model detects an error
that causes the gate dimension to be changed equal to or more than
5 nanometers from the target value. At the same time, an
instruction is issued to an operator to perform maintenance of the
wafer stage based on the parameter of the exposure device used for
detection (i.e., the orthogonality of a wafer). Specifically, a
maintenance instruction is displayed on the display unit of the
user I/F 8, and the operator conducts the maintenance in accordance
with the displayed instruction. An "adjustment" is registered as
event data associated with the fourth FDC model.
[0154] As a still another example of the extracted exposure-device
parameter, shot magnification and delay in processing after a
developing process are explained below. As a result of operation of
the management system 100, it can be seen that there is a
correlation between the shot magnification, the delay in processing
after a developing process, and the gate dimension. The shot
magnification is a relative magnification of a reticle image in the
exposure process. The delay in processing after a developing
process means delay from a time when a predetermined developing
process is finished to a time when a wafer is actually discharged
from a developing unit, in a resist developing process to the wafer
after the exposure process. The resist developing process and the
exposure process are integrally explained as the exposure process;
because the resist developing process is performed right after the
exposure process and each processing device is integrally arranged
with each other.
[0155] The correlation between the shot magnification and the delay
in processing after a developing process and the dimensional change
is not clear. However, as shown in FIG. 13, if X-coordinate defines
the delay in processing after a developing process, Y-coordinate
defines the shot magnification, and contour is depicted in
accordance with an absolute value of an amount of change from the
target value of the gate dimension, a certain correlation can be
seen. The absolute value of the amount of change from the target
value of the gate dimension represents dimensional change,
representing each of curved lines with dimensional change of 1
nanometer, 3 nanometers, and 5 nanometers. It can be seen from FIG.
13 that, when the delay in processing after a developing process
exceeds 10 seconds or more, and the shot magnification becomes
equal to or larger than 0.1 parts per million, the gate dimension
exceeds the allowable range in design.
[0156] A two-variable function is provided that presumes an
absolute value of the amount of change from the target value of the
gate dimension based on the delay in processing after a developing
process and the shot magnification. A fifth FDC model using the
two-variable function is prepared to issue a warning when detecting
the dimensional change of equal to or more than 5 nanometers from
the allowable range of the gate dimension. A fifth FDC model unit
in which the fifth FDC model is installed is also arranged in the
configuration of the management system according to the third
embodiment. As described above, upon receiving the shot
magnification and the delay in processing after a developing
process, the fifth FDC model detects an error that causes the gate
dimension to be changed equal to or more than 5 nanometers from the
target value. At the same time, an instruction is issued to an
operator so that maintenance of an optical stage and a developing
unit is conducted based on the parameter of the exposure device
used for an error detection (i.e., the shot magnification and the
delay in processing after a developing process). Specifically, a
maintenance instruction is displayed on the display unit of the
user I/F 8, and the operator conducts the maintenance in accordance
with a displayed instruction. An "adjustment" and a clean track
serving as a unit for resist coating, baking, and developing
process are registered as event data associated with the fifth FDC
model.
[0157] As the two-variable function used in the fifth FDC model,
multivariate function such as Mahalanobis distance can be used. As
a method of extracting the exposure-device parameter associated
with the dimensional change, a univariate correlation analysis is
used in the first embodiment, while the two-variable function using
the shot magnification and the delay in processing after a
developing process is used in the third embodiment. However, it is
possible to use the PLS analysis or the principal component
analysis instead of the correlation analysis using a correlation
coefficient.
[0158] FIG. 14 depicts examples of the third to the fifth FDC
models. As a result of operation in the exposure process performed
by the management system 100, other exposure parameters causing
change of the resist width are extracted. The exposure-device
parameters associated with the third to the fifth FDC models and
the other exposure-device parameters are the following:
[0159] (1) Parameter indicative of synchronization accuracy between
a wafer stage and a reticle stage (synchronization accuracy, mean,
standard deviation)
[0160] (2) Parameter indicative of difference between a target
value and an actually-measured value of a focus position (follow
focus, mean, standard deviation)
[0161] (3) Parameter indicative of difference between a target
value and an actually-measured value of a tilt amount (tilt to
Z-axis, mean, standard deviation)
[0162] (4) Parameter associated with alignment (parallel movement,
rotation, magnification, orthogonality)
[0163] (5) Parameter associated with resist coating, baking, and
developing (temperature, flow rate, processing time)
[0164] In the above example, the third FDC model is classified into
(1), the orthogonality of a wafer is (4), the shot magnification is
(4), and the delay in processing after a developing process is (5).
The description "synchronization accuracy, mean, standard
deviation" added to the parameter (1) indicates that a mean,
standard deviation, and a mean of the standard deviation for the
parameter are also used as parameters in addition to the
synchronization accuracy. This is the same for the parameters (2)
and (3). The parameter (2) indicates follow property of a focus
position in a projection optical system. The parameter (3)
indicates the level of tilt to Z-axis orthogonal to the surface of
the wafer stage; for example, tilt amount of an optical axis. The
parameter (4) is associated with alignment of a reticle and a
wafer, indicating parallel movement, rotation, magnification,
orthogonality. The parameter (5) is associated with the exposure
process, such as resist coating, baking, and temperature, flow
rate, and processing time in a developing process.
[0165] As described in the third embodiment, an FDC model is
prepared by examining relation between the exposure-device
parameter and a value of target parameter, and obtaining a
detection rule for detecting when the target parameter exceeds the
allowable range in design.
[0166] The event data is also registered in association with the
arranged FDC model. By installing such FDC models in the management
system according to the second embodiment, it is possible to
automatically issue an update instruction for the FDC model when an
event requiring an update of the FDC model occurs.
[0167] As described above, according to the third embodiment, the
exposure-device parameter as the cause of change of the target
parameter is extracted, and an FDC model is arranged in which a
detection rule for detecting a situation where the target parameter
exceeds the allowable range in design. Therefore, an error in
target parameter can be detected and maintenance instruction can be
automatically issued depending on the cause of the error. Examples
of the detection rule include a management value of each of the
exposure-device parameters with which the target parameter is
within the predetermine range, and a univariate/multivariate
detection function.
[0168] It is effective to stabilize a target parameter in a
predetermined range during the exposure process for improving
productivity of the semiconductor devices. In the conventional
technique, dimensional change is adjusted by using a relatively
easy parameter such as adjustment of light exposure. Thus, true
cause of variation in a device is left as it is without taking a
countermeasure, resulting in causing the same error. However, cause
of variation can be extracted as described in the first embodiment,
and target parameter can be stably set in a predetermined range by
installing an FDC model in relation to an extracted exposure-device
parameter as described in the second embodiment.
[0169] As set forth hereinabove, according to an aspect of the
present invention, it is possible to extract a cause of variation
of the QC values excluding cause of variation due to events such as
an APC control or a maintenance. Thus, an error such as a defective
state of the semiconductor manufacturing device can be detected in
the semiconductor-device manufacturing process.
[0170] Furthermore, it is possible to use an error-detection rule
in accordance with a change of a state of a device due to events
such as the APC control or a maintenance operation.
[0171] Additional advantages and modifications will readily occur
to those skilled in the art. Therefore, the invention in its
broader aspects is not limited to the specific details and
representative embodiments shown and described herein. Accordingly,
various modifications may be made without departing from the spirit
or scope of the general inventive concept as defined by the
appended claims and their equivalents.
* * * * *