U.S. patent application number 12/719207 was filed with the patent office on 2010-06-24 for exposure data generation method and device, exposure data verification method and device and storage medium.
This patent application is currently assigned to FUJITSU MICROELECTRONICS LIMITED. Invention is credited to Kozo OGINO.
Application Number | 20100162198 12/719207 |
Document ID | / |
Family ID | 38370238 |
Filed Date | 2010-06-24 |
United States Patent
Application |
20100162198 |
Kind Code |
A1 |
OGINO; Kozo |
June 24, 2010 |
Exposure data generation method and device, exposure data
verification method and device and storage medium
Abstract
Exposure verification is applied to exposure data indicating a
pattern to be exposed by a charged particle beam. If an error point
is extracted from the exposure data by the exposure verification,
the values of coefficients are modified and exposure data is
regenerated taking into consideration the coefficients whose values
have been modified. Thus, exposure data is re-generated by changing
each of the coefficient values within its appropriate range.
Inventors: |
OGINO; Kozo; (Kawasaki-shi,
JP) |
Correspondence
Address: |
Fujitsu Patent Center;C/O CPA Global
P.O. Box 52050
Minneapolis
MN
55402
US
|
Assignee: |
FUJITSU MICROELECTRONICS
LIMITED
Yokohama-shi
JP
|
Family ID: |
38370238 |
Appl. No.: |
12/719207 |
Filed: |
March 8, 2010 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11510556 |
Aug 28, 2006 |
7707540 |
|
|
12719207 |
|
|
|
|
Current U.S.
Class: |
716/54 |
Current CPC
Class: |
H01L 21/0277 20130101;
G03F 1/36 20130101 |
Class at
Publication: |
716/21 ;
716/19 |
International
Class: |
G06F 17/50 20060101
G06F017/50 |
Foreign Application Data
Date |
Code |
Application Number |
Feb 14, 2006 |
JP |
2006-037006 |
Claims
1-6. (canceled)
7. An exposure data generation method for generating exposure data
for exposing a resist film formed on a multi-layered semiconductor
substrate by a charged particle beam, comprising: calculating a
plurality of amounts of exposure obtained on the resist film taking
into consideration an error caused in dimensions of a pattern
formed on a layer constituting the semiconductor substrate; and
extracting a point to be inappropriate from the exposure data,
based on the plurality of calculated amounts of exposure.
8-9. (canceled)
10. An exposure data verification device for verifying exposure
data for exposing a resist film formed on a multi-layered
semiconductor substrate by a charged particle beam, comprising:
calculation unit for calculating a plurality of amounts of exposure
obtained on the resist film taking into consideration an error
caused in dimensions of a pattern formed on a layer constituting
the semiconductor substrate; and exposure verification unit for
performing exposure verification for extracting a point to be
inappropriate from the exposure data, based on the plurality of
calculated amounts of exposure by the calculation unit.
11-12. (canceled)
13. A computer-readable storage medium on which is recorded a
program for enabling a computer to perform functions, the functions
comprising: calculation function for calculating a plurality of
amounts of exposure obtained on the resist film taking into
consideration an error caused in dimensions of a pattern formed on
a layer constituting the semiconductor substrate; and exposure
verification function for performing exposure verification for
extracting a point to be inappropriate from the exposure data,
based on the plurality of calculated amounts of exposure by the
calculation function.
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. 2006-037006
filed on Feb. 14, 2006, the entire contents 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 generating
exposure data to expose a resist film formed on a multi-layered
semiconductor substrate by a charged particle beam.
[0004] 2. Description of the Related Art
[0005] Lately, in the manufacture of semiconductor devices, such as
a large-scale integrated circuit (LSI) and the like, it is desired
to form a very fine pattern. Thus, currently a charged particle
beam is usually used for pattern generation exposure. It is common
to use an electron for a charged particle.
[0006] By charged particle beam exposure, a part of charged
particles inputted to a resist film is forward-scattered, apart of
the particles that have transmitted the resist film are
backward-scattered and it is inputted the resist film again. Thus,
even when a charged particle beam is inputted to one point on the
resist film, an area exposed by the charged particle is not only
one point, but it also covers its neighborhood (proximity effect).
Therefore, exposure data indicating the exposure pattern of a
charged particle beam is usually formed by applying proximity
effect correction to layout data indicating a pattern to be formed
on the resist film in order to optimize the amount of exposure or
dimensions of an exposure pattern (Japanese Patent Application Nos.
2005-101501, 2003-149784 and H11-8187).
[0007] With the recent fine semiconductor devices, the form of an
exposure pattern for expose a semiconductor substrate has become
fine and also its multi-layer structure has become complex. The
backward scatter intensity of exposure to such a semiconductor
substrate can be calculated with high accuracy by simulation based
on a physical model (for example, simulation by a Monte Carlo
method). However, actually it takes a very long time to calculate
the intensity. Thus, it is desired to calculate the intensity in a
shorter time while realizing higher accuracy.
[0008] As publicly known, the scatter of a charged particle varies
depending on its material. In the prior art disclosed by Japanese
Patent Application No. 2005-101501 (hereinafter "patent reference
1"), scatter distribution depending on a distance is prepared as a
coefficient a and the backward scatter intensity of each area is
calculated by an area density method. If intensity in an area (i,
j) is expressed Fb.sub.i, j, the Fb.sub.i, j is finally calculated
as follows.
Fb i , j = l m E n - 1 ( i + l , j + m ; i , j ) .alpha. i + l , j
+ m Q i + l , j + m ( 1 ) ##EQU00001##
[0009] In the above equation, .alpha..sub.i+1, j+m, Q.sub.i+1, j+m
and E.sub.n-1 (i+1, j+1; i, j) represent pattern density in an area
(i+1, j+m), the amount of exposure applied to an area (i+1, j+m)
and a charged particle intensity coefficient indicating the degree
of influence on an area (i+1, j+m) of the amount of exposure
applied to an area (i, j), respectively.
[0010] The charged particle intensity coefficient E.sub.n-1
corresponds to the coefficient a. The coefficient a can be
calculated using a reflection coefficient R, which is a ratio
indicating the reflection of a charged particle on a layer, and a
transmission coefficient T indicating its ratio of transmitting
through the layer prepared by each material. Thus, the backward
scatter intensity Fb.sub.i, j taking the material of each layer
into consideration can be calculated to realize high accuracy. This
exposure data can also be appropriately generated in high
accuracy.
[0011] In the manufacture of semiconductor devices, a factor of
accuracy degradation due to a multi-layer structure, such as
unevenness in thickness of lower layers, due to non-uniformity of
chemical machine polish (CMP) or the accuracy error in dimensions
of the pattern of a lower layer, sometimes occur. Stored energy
distribution to the resist film varies depending on such a factor.
Each value of the coefficients R and T varies depending on the
occurrence and the degree of such a factor. Thus, an error occurs
in the backward scatter intensity Fb.sub.i, j which is calculated
according to equation (1). The influence of the error has a
tendency to increase due to fine semiconductor devices. Therefore,
in even wholly appropriate exposure data, an inappropriate point
(poor resolution point) is easily detected by exposure
verification. Thus, in the generation of exposure data, including
exposure verification, it is also important to take such a factor
of accuracy degradation into consideration.
[0012] In the prior art disclosed by Japanese Patent Application
Nos. 2003-149784 and H11-8187, the amount of calculation is reduced
by limiting a point for calculating stored energy including
backward scatter intensity as an evaluation point.
SUMMARY OF THE INVENTION
[0013] It is an object of the present invention to provide a
technology for generating exposure data taking into consideration a
factor of accuracy degradation due to a multi-layer structure.
[0014] The exposure data generation method of the present invention
generates exposure data for exposing a resist film formed on a
multi-layered semiconductor substrate by a charge particle beam.
The exposure data generation method comprises obtaining exposure
data indicating a pattern to be exposed by the charge particle
beam, which is generated from layout data indicating a pattern to
be formed on the resist film, performing exposure verification the
exposure data, using at least one changeable coefficient, modifying
the value of a coefficient when an error point is extracted from
the exposure data by exposure verification and re-generating
exposure data taking into consideration the coefficient whose value
is modified.
[0015] It is preferable to perform the exposure verification taking
into consideration the backward scatter of the charged particle
beam by a layer located below an exposure target layer on which a
resist layer is formed. It is preferable to perform the exposure
verification in two steps; the first exposure verification using a
coefficient and the second exposure verification performed based on
the result of the first exposure verification. The first exposure
verification can also be performed by calculating the degree of
risk for approximating the size of the backward scatter intensity.
It is preferable to perform at least one of film thickness margin
verification taking into consideration the error of the film
thickness of a layer constituting the semiconductor substrate and
area density margin verification taking into consideration the
error of the dimensions of a pattern formed on the layer.
[0016] The exposure data verification methods in the first and
second aspects of the present invention both verify exposure data
for exposing a resist film formed on a multi-layered semiconductor
substrate by a charged particle beam and each of them performs
exposure verification as follows.
[0017] The exposure data verification method in the first aspect of
the present invention calculates a plurality of the amount of
exposure obtained on the resist film taking into consideration the
error in film thickness of a layer constituting a semiconductor
substrate and extracts a point to be considered inappropriate from
exposure data.
[0018] The exposure data verification method in the second aspect
of the present invention calculates a plurality of the amount of
exposure obtained on the resist film taking into consideration the
error in dimensions of a pattern formed on a layer constituting a
semiconductor substrate and extracts a point to be considered
inappropriate from exposure data.
[0019] In the present invention, exposure verification is applied
to exposure data indicating a pattern to be exposed by a charge
particle beam, using at least one changeable coefficient. If the
exposure verification extracts an error point from the exposure
data, the value of the coefficient and exposure data is regenerated
taking into consideration the coefficient whose value is
modified.
[0020] Some coefficient has an appropriate range taking into the
factor of accuracy degradation due to a multi-layer structure. In
exposure verification using such a coefficient, there is a
possibility that the error point varies depending on a value
adopted for the exposure verification. By such a possibility, a
point that is not actually erroneous sometimes regarded as an error
point. However, if a coefficient value is changed within the
appropriate range and exposure data is re-generated, a point which
should not be regarded as an error point can be prevented or
suppressed from being regarded as an error point. Thus, an error
point to be coped with can be more easily coped with. As a result,
exposure data can also be more easily generated taking into
consideration the factor of accuracy degradation due to the
multi-layer structure.
[0021] In the present invention, a plurality of the amount of
exposure obtained on a resist film can be calculated taking into
consideration the error in film thickness of a layer constituting a
semiconductor substrate, and a point which should be regarded
inappropriate can be extracted from exposure data, based on the
plurality of the calculated amount of exposure. Therefore, a part
which is made erroneous by a film thickness error can be surely
extracted.
[0022] In the present invention, a plurality of the amount of
exposure obtained on a resist film can be calculated taking into
consideration the error in dimensions of a pattern formed on a
layer constituting a semiconductor substrate, and a point which
should be regarded inappropriate can be extracted from exposure
data, based on the plurality of the calculated amount of exposure.
Therefore, a part which is made erroneous by a dimensional error of
the formed pattern can be surely extracted.
[0023] Either of them facilitates coping with the factor of
accuracy degradation due to a multi-layer structure.
BRIEF DESCRIPTION OF THE DRAWINGS
[0024] FIG. 1 shows the configuration of the exposure data
generation device of the present invention;
[0025] FIG. 2 shows how to detect an error point in exposure
verification;
[0026] FIG. 3 shows a correction parameter extraction data
management table;
[0027] FIG. 4 shows a verification error management table;
[0028] FIG. 5A shows how to display an error when the error is
detected by a pattern edge resolution position error verification
method;
[0029] FIG. 5B shows how to detect an error by the pattern edge
resolution position error verification method;
[0030] FIG. 6A shows how to display an error when the error is
detected by an exposure intensity contrast verification method;
[0031] FIG. 6B shows how to detect an error by the exposure
intensity contrast verification method;
[0032] FIG. 7A shows how to display an error when the error is
detected by an exposure amount margin verification method;
[0033] FIG. 7B shows how to detect an error by the exposure amount
margin verification method;
[0034] FIG. 8A shows how to display an error when the error is
detected by a lower layer film thickness margin verification
method;
[0035] FIG. 8B shows how to detect an error by the lower layer film
thickness margin verification method;
[0036] FIG. 9A shows how to display an error when the error is
detected by a lower layer area density margin verification
method;
[0037] FIG. 9B shows how to detect an error by the lower layer area
density margin verification method;
[0038] FIG. 10 is the flowchart of the exposure data generation
process; and
[0039] FIG. 11 shows an example of the hardware configuration of a
computer capable of realizing the exposure data generation device
of the present invention.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0040] The preferred embodiments of the present invention are
described in detail below with reference to the drawings.
[0041] FIG. 1 shows the configuration of the exposure data
generation device of the present invention.
[0042] The exposure data generation device generates exposure data
by inputting the layout data of each layer, and if a point which
should be poor (error point) is detected by exposure verification,
the generation device re-generates exposure data in such a way as
to correct the point.
[0043] An input unit 11 is used to externally input layout data and
exposure experiment data. In this preferred embodiment, exposure
verification is performed using the method disclosed by Patent
reference 1. The exposure experiment data is used to extract a
coefficient needed for the exposure verification. Since the value
of the coefficient is modified (updated) taking into consideration
the factor of accuracy degradation due to a multi-layer structure,
the coefficient is hereinafter called a "parameter". In this case,
a plurality of parameters exists.
[0044] A storage unit 12 stores various types of data inputted by
the input unit 11. A parameter extraction unit 13 extracts or
modifies a parameter value from the exposure experiment data. A
proximity effect correction unit 14 performs proximity effect
correction using the parameter value inputted from the parameter
extraction unit 13 to generate exposure data.
[0045] An exposure verification unit 15 applies exposure
verification to the generated exposure data to extract an error
point. The exposure verification unit 15 comprises a simple
exposure verification unit 15a for performing (simple) exposure
verification using the method disclosed by Patent reference 1 and a
detailed exposure verification unit 15b for performing exposure
verification by Monte Carlo method.
[0046] Simple exposure verification is applied to the entire
exposure data while detailed exposure verification is applied to an
error point detected by the simple exposure verification. Thus,
exposure verification which takes much calculation time can be
minimized, and exposure data can be generated more rapidly while
realizing high accuracy.
[0047] If the error point extracted by simple exposure verification
is confirmed to be erroneous by detailed exposure verification, the
exposure verification unit 15 stores the error point in the storage
unit 12 and instructs the parameter extraction unit 13 and the
proximity effect correction unit 14 to perform the process again.
Thus, the parameter extraction unit 13 updates the parameter value
within an appropriate range determined taking into consideration
the factor of accuracy degradation due to a multi-layer structure
in such a way as to correct the error point. The proximity effect
correction unit 14 re-generates exposure data after performing
proximity effect correction using the updated parameter value.
[0048] The exposure verification unit 15 applies exposure
verification to the regenerated exposure data again. If an error
point is extracted again by the exposure verification, the exposure
verification unit 15 instructs the parameter extraction unit 13 and
the proximity effect correction unit 14 to perform the process
again. Thus, exposure data which includes no error point when the
factor of accuracy degradation due to a multi-layer structure is
taken into consideration is generated. Such exposure data is
outputted from an output unit 16.
[0049] FIG. 11 shows an example of the hardware configuration of a
computer capable of realizing the exposure data generation device.
Prior to the detailed description of FIG. 1, the configuration of a
computer capable of realizing the exposure data generation device
is described in detail. For convenience' sake, the following
description is made hereinafter presuming that the exposure data
generation device is realized by one computer with the
configuration shown in FIG. 11.
[0050] The computer shown in FIG. 11 comprises a CPU 61, memory 62,
an input device 63, an output device 64, an external storage device
65, a storage medium driving device 66 and a network connection
device 67, which all are connected to each other by a bus 68. The
configuration shown in FIG. 11 is only one example and is not
limited to this.
[0051] The CPU 61 is a central processing unit for controlling the
entire computer.
[0052] The memory 62 is RAM or the like for temporarily storing a
program or data stored in the external storage device 65 (or
portable storage medium MD) when updating data or so on. The CPU 61
controls the entire computer by reading out the program into the
memory 62 and executing it.
[0053] The input device 63 is connected to an input device, such as
a keyboard, a mouse or the like or has it. The input device 63
detects the operator operation of such an input device and notifies
the CPU 61 of the detection result.
[0054] The output device 64 is connected to a display or the like
or has it. The output device 64 outputs data transmitted under the
control of the CPU 61 on the display.
[0055] The network connection device 67 is used to communicate with
another device via a network, such as an intranet, the Internet or
the like. The external storage device 65 is a hard disk device or
the like. The external storage device 65 mainly stores various
types of data and programs.
[0056] The storage medium driving device 66 is used to access a
portable storage medium MD, such as an optical disk, a
magneto-optical disk or the like.
[0057] The layout data and the exposure experiment data are
obtained via the network connection device 67 or the storage medium
driving device 66 and are stored in the external storage device 65
or the like. Thus, the input unit 11 can be realized by the CPU 61,
the memory 62, the external storage device 65, the network
connection device 67, the storage medium driving device 66, and the
bus 68. The storage unit 12 can be realized by the external storage
device 65 or the memory 62. The generated exposure data is
outputted via the network connection device 67 or the storage
medium driving device 66. Therefore, the output unit 16 can be
realized by the CPU 61, the memory 62, the external storage device
65, the network connection device 67, the storage medium driving
device 66, and the bus 68. The others can be realized by the CPU
61, the memory 62, the external storage device 65, and the bus
68.
[0058] The exposure data generation device (exposure data
verification device) of this preferred embodiment can be realized
by the CPU 61 executing the program mounting functions necessary
for it. The program can be recorded in the storage medium MD and be
distributed. Alternatively, it can be obtained by the network
connection device 67.
[0059] FIG. 2 shows how to detect an error point in exposure
verification. A "target graphic", an "evaluation point" and a
"resolution position" correspond to a pattern represented by layout
data, a point to be verified, on the edge of the target graphic and
an actually formed evaluation point on the pattern, respectively.
In this preferred embodiment, basically an error point is extracted
focused on the deviation in a position between the resolution
position and the evaluation point.
[0060] FIG. 3 shows a correction parameter extraction data
management table. The correction parameter extraction data
management table is stored in the storage unit 12. The verification
error management table shown in FIG. 4 is also stored in the
storage unit 12. Hereinafter, the correction parameter extraction
data management table and the verification error management table
are called a "data management table" and an "error management
table", respectively.
[0061] The data management table is prepared to manage/store
parameter extraction data. The data management table stores each
coordinates of the evaluation point and resolution position for
each data type. A "layout data identification name" in FIG. 3
indicates layout data to which proximity effect correction is
applied, that is, is an exposure data identification name.
[0062] The parameter extraction unit 13 refers to data stored in
the data management table to determine the value of the parameter.
For example, firstly, a parameter value such that the evaluation
point and the resolution position is matched is extracted and
determined. After that, a parameter value is extracted from newly
stored data and an optimal one is determined taking into
consideration each extracted parameter. The optimal parameter value
is determined by least squares method, for example, using the first
determined value as an initial value. As the parameter extraction
method, the technology disclosed by Japanese Patent Application No.
2005-211042, whose patent the applicant has applied for Jul. 21,
2005 can be used.
[0063] Each parameter has an appropriate range. A parameter value
must be determined taking the range into consideration. Thus,
actually the parameter value is determined as follows.
[0064] In the backward scatter intensity calculation method
disclosed by Patent reference 1, each of the transmission
coefficient T, reflection coefficient R and diffusion length a
indicating the 1/e radius of Gaussian distribution for each
material of each layer has a physical range. The possible range of
each parameter value of the film thickness of 0.about..infin. of a
semiconductor substrate each layer of which is made of one kind of
material is as follows.
0.apprxeq.T.ltoreq.1 (2)
0.ltoreq.R.ltoreq..eta. (backward scatter coefficient) (3)
0.ltoreq..sigma..ltoreq..beta.b (backward diffusion length) (4)
[0065] Since the range of each parameter value is for fixed film
thickness, the range can be narrowed to some extent by extracting
each parameter in the film thickness by simulation or the like in
advance.
[0066] If simulation by Monte Carlo method is used, usually scatter
in the case where a film made of a target material is placed at the
vacuum and a lot of charged particles (hereinafter called
"electron") is inputted to one point on its surface is simulated.
In that case, the ratio of the total energy of electrons outputted
from the surface of the film to the total energy of inputted
electrons is the reflection coefficient R and a stretch (1/e
radius) obtained when the energy distribution of electrons
outputted from the surface of the film approximates Gaussian
distribution corresponds to the diffusion length .sigma.. The ratio
of the total energy of electrons outputted from the back surface of
the film to the total energy of inputted electrons corresponds to
the transmission coefficient T. Strictly speaking, since the amount
of energy of electrons and the stored amount of energy in the
resist is not matched, some allowance must be given to each of the
calculated parameters T.sub.o, R.sub.o and .sigma..sub.o. Thus, a
constant q indicating its allowance is externally specified, and
the range of each of the parameter values T, R and .sigma. is set
according to the following expressions.
(1-q)T.sub.o.ltoreq.T.ltoreq.q+(1-q)T.sub.o (5)
(1-q)R.sub.o.ltoreq.R.ltoreq.q.eta.+(1-q)R.sub.o (6)
(1-q).sigma..sub.o.ltoreq..sigma..ltoreq.q.beta.b+(1-q).sigma..sub.o
(7)
[0067] If q=1, expressions (5).about.(7) coincide with expressions
(2).about.(4). If q=0, they coincide with the values calculated by
the simulation. Hereinafter, the lower and upper limits of the
parameters are expressed as T.sub.min, R.sub.min and
.sigma..sub.min, and T.sub.man, R.sub.man and .sigma..sub.man,
respectively.
[0068] Firstly, each parameter value with an appropriate range
indicated by expressions (5).about.(7) is determined. The second
time and after, the previously obtained value is corrected taking
into consideration data newly stored in the data management table.
Thus, an optimal value in the neighborhood of the previously
obtained value can be determined for the parameter.
[0069] In this case, the range of each parameter value is
restricted to the range indicated by expressions (5).about.(7).
However, if the number of the data newly stored in the data
management table is small, the change of each parameter value is
expected to be small. Therefore, the previously applied range can
also be further narrowed around the initial value according to the
ratio of the newly added data to the entire data. In that case, if
the initial values are T.sub.k-1, R.sub.k-1 and .sigma..sub.k-1 and
the ratio of newly added data to the entire data is p, the possible
ranges of parameter values this time T.sub.k, R.sub.k and
.sigma..sub.k can also be restricted as follows.
pT.sub.min,k-1+(1-p)T.sub.k-1.ltoreq.T.sub.k.ltoreq.pT.sub.max,k-1+(1-p)-
T.sub.k-1 (8)
pR.sub.min,k-1+(1-p)R.sub.k-1.ltoreq.R.sub.k.ltoreq.pR.sub.max,k-1+(1-p)-
R.sub.k-1 (9)
p.sigma..sub.min,k-1+(1-p).sigma..sub.k-1.ltoreq..sigma..sub.k.ltoreq.p.-
sigma..sub.max,k-1+(1-p).sigma..sub.k-1 (10)
[0070] If each parameter exists out of a predetermined range when
least squares method is used as a method for correcting the
parameter value within the range, sometimes no square-sum x.sup.2
becomes a minimum by existence of the parameter that a value is out
of a range. Least squares method corresponds to calculating a set
of parameter values in which square-sum x.sup.2 becomes a minimum,
using square-sum x.sup.2 as the function of the set of parameter
values. In this case, if square-sum x.sup.2(.alpha.) is calculated
according to an expression which monotonously increases out of the
boundary as follows, assuming that only one parameter value .alpha.
is used and its range is .alpha.1.ltoreq..alpha..ltoreq..alpha.2,
for convenience' sake, the range can be made to include a minimum
value without fail when calculate square-sum x.sup.2(.alpha.) in an
expression to increase monotonous outside from a border from the
border.
x.sup.2(.alpha.)(.alpha.1.ltoreq..alpha..ltoreq..alpha.2) (11)
x.sup.2(.alpha.)+k.alpha..times.(.alpha.1-.alpha.)(.alpha.<.alpha.1)
(12)
x.sup.2(.alpha.)+k.alpha..times.(.alpha.-.alpha.2)(.alpha.2<.alpha.)
(13)
[0071] In the above expressions, k.alpha. is a constant (>0) for
speeding up the return of .alpha. from its deviation in the search
of a minimum value by increasing the difference of .alpha. to some
extent. The value can also be 1.
[0072] The above method can also extend similarly even if a
plurality of parameter values exists. Therefore, by adopting this
method, each parameter value can be corrected (determined) within
each predetermined range. By such correction, only an actual error
point or a point with such a high possibility remains in the
exposure data. As a result, the developer can cope with a point to
be coped with, thereby improving work efficiency and
serviceability.
[0073] FIG. 4 shows a verification error management table.
[0074] This error management table is prepared to manage/store
error points detected by exposure verification. The error
management table stores the coordinates of an evaluation point, an
outward direction evaluation vector, a resolution position error
and its error contents for each error point.
[0075] An evaluation point is provided on an edge of a target
graphic. For this reason, the outward direction evaluation vector
(hereinafter called an "outward vector") indicates a vector which
pass through the evaluation point from inside the target graphic
and goes outward in unit vectors. Thus, for example, in FIG. 2, if
there are the evaluation point and the resolution position on the
X-axis and the resolution position is located in smaller
coordinates than the revaluation point on the X-axis, the outward
vector becomes (1, 0). If they are located reversely on the X-axis,
it becomes (-1, 0).
[0076] The resolution position error indicates the respective
errors of the evaluation point and the resolution position in the
outward vector direction. Therefore, for example, similarly, if
there are the evaluation point and the resolution position on the
X-axis and the resolution position is located in smaller
coordinates than the revaluation point on the X-axis, the
resolution position error becomes negative. Thus, if an error
occurs in the thick direction of a target graphic, the resolution
position error becomes positive. If an error occurs in the reverse
direction, it becomes negative.
[0077] In this preferred embodiment, a plurality of types of
exposure verification is attempted. Each of E1.about.E3 shown in
FIG. 4 indicates the type of exposure verification which detects an
error. The exposure verification performed in this preferred
embodiment is specifically described below with reference to FIGS.
5A.about.9B.
[0078] FIGS. 5A and 5B shows a pastern edge resolution position
error verification method. FIG. 5A shows how to display an error
when the error is detected by the pattern edge resolution position
error verification method. FIG. 5B shows how to detect an error by
the pattern edge resolution position error verification method.
[0079] In FIG. 5A, the resolution position .DELTA. is located on
the left of the evaluation point. Hereinafter, in FIGS. 6A.about.9A
too, it is assumed for convenience' sake that the direction where
the resolution position .DELTA. and the evaluation point are
arrayed is the X-axis. The direction from the resolution position
.DELTA. toward the evaluation point is the ascending direction of a
position on the X-axis. A direction where layers are piled is
called a vertical direction.
[0080] The vertical and horizontal axes of the graph shown in FIG.
5B indicate exposure intensity (amount of exposure) and a position
on the X-axis, respectively. E.sub.th represents the threshold of
the exposure intensity with which a pattern is formed.
.DELTA..sub.max represents the maximum value in the allowance in
the outward vector direction (the X-axis direction here) using the
evaluation point as the reference. Thus, the resolution position
.DELTA. range of
-.DELTA..sub.max.ltoreq..DELTA..ltoreq..DELTA..sub.max is specified
as its error allowance and no error is detected in the range.
[0081] In FIG. 5A, the resolution position .DELTA. is located far
away from -.DELTA..sub.max. The error graphic shown in FIG. 5A
notifies that an error occurs due to it, and can be displayed as a
rectangle having a width on an edge (the Y-axis here) covered by
the evaluation point and a width (error) on the X-axis, of the
evaluation point and the resolution position .DELTA.. Data for
displaying the error graphic can be outputted in the same file
format as the layout data, exposure data or the like.
[0082] FIGS. 6A and 6B shows an exposure intensity contrast
verification method. Like FIGS. 5A and 5B, FIG. 6A shows how to
display an error when the error is detected by the exposure
intensity contrast verification method. FIG. 6B shows how to detect
an error by the exposure intensity contrast verification
method.
[0083] In this exposure intensity contrast verification method, a
contrast value C(=(E.sub.max-E.sub.min)/(E.sub.max+E.sub.min)) is
calculated using the maximum exposure intensity E.sub.max in a
target graphic and the minimum exposure intensity E.sub.min in the
neighborhood of its outside. Whether or not an error occurs is
checked by whether the calculated value C is equal to or more than
the threshold C.sub.min which is predetermined as the allowable
minimum value. The width on the X-axis of the error graphic in the
case where the error occurs is a width obtained by multiplying a
width with the target graphic adjacent on the X-axis, which has a
corresponding evaluation point, by the value of 100-C.
[0084] FIGS. 7A and 7B shows an exposure amount margin verification
method. Like FIGS. 5A and 5B, FIG. 7A shows how to display an error
when the error is detected by the exposure amount margin
verification method. FIG. 7B shows how to detect an error by the
exposure amount margin verification method.
[0085] Exposure intensity needed to form a pattern depends on the
material of an adopted resist film. This means that an actual edge
position varies depending on the material of a resist film.
Therefore, in this exposure amount margin verification method, as
shown in FIG. 7B, the maximum change ratio k.sub.max is prepared to
set the allowable range of exposure intensity (amount of exposure).
Then, two resolution positions .DELTA..sub.1 and .DELTA..sub.2,
which are 1/(1-k.sub.max) and 1/(1+k.sub.max) respectively of an
exposure intensity threshold E.sub.th are calculated, and it is
checked whether these resolution positions .DELTA..sub.1 and
.DELTA..sub.2 both are within the error allowance. The width on the
X-axis of the error graphic are one between those resolution
positions .DELTA..sub.j and .DELTA..sub.2.
[0086] FIGS. 8A and 8B shows a lower layer film thickness margin
verification method. Like FIGS. 5A and 5B, FIG. 8A shows how to
display an error when the error is detected by the lower layer film
thickness margin verification method. FIG. 8B shows how to detect
an error by the lower layer film thickness margin verification
method.
[0087] Backward scatter intensity varies depending on not only a
material but also layer film thickness. If the parameter values are
T, R and .sigma. and their respective values in the case where the
film thickness becomes n times are T.sub.n, R.sub.n and
.sigma..sub.n, the following relationship exists between those
values. Therefore, exposure intensity varies depending on film
thickness.
T.sub.n=T.sup.n (14)
R.sub.n=R(1-T.sub.n.sup.2)/(1-T.sup.2)=R(1-T.sup.2n)/(1-T.sup.2)
(15)
.sigma..sub.n=n.sup.1/2.sigma. (16)
[0088] The film thickness varies depending on the non-uniformity of
chemical machine polish (CMP). It also varies depending on its
position on the semiconductor substrate, its manufacturing process
and the like. For this reason, in this lower layer film thickness
margin verification method, as shown in FIG. 8B, the maximum
possible amount of change d of the film thickness of a lower layer
is taken into consideration. In each of the case where the film is
thick by the maximum amount of change d and the case where the film
is thin by the maximum amount of change d, resolution positions
.DELTA..sub.1 and .DELTA..sub.2 whose exposure intensity is the
threshold E.sub.th are calculated and it is checked whether those
resolution positions .DELTA..sub.1 and .DELTA..sub.2 both are
within its error allowance. Since it can be considered that an
actual resolution position exists between those resolution
positions .DELTA..sub.1 and .DELTA..sub.2, an error due to the
error of film thickness can be surely detected. In this case, the
width on the X-axis of an error graphic is one between the
evaluation point and the resolution position .DELTA..sub.2.
[0089] FIGS. 9A and 9B shows a lower layer area density margin
verification method. Like FIGS. 5A and 5B, FIG. 9A shows how to
display an error when the error is detected by the lower layer area
density margin verification method. FIG. 9B shows how to detect an
error by the lower layer area density margin verification
method.
[0090] Backward scatter intensity varies depending on a material.
The area density of the lower layer area density margin
verification method varies depending on an error in pattern
generation. For example, as shown in FIG. 9B, if an object
indicated by a rectangle in the top view of a layer is a contact
hole, it can be considered that the maximum amount of change d
exists in the width (dimensions) of the contact hole. If such an
amount of change d exists, exposure intensity varies between the
case where the dimensions is thick by the maximum amount of change
d and the case where the dimensions is thin by the maximum amount
of change d. For this reason, in this lower layer area density
margin verification method, as shown in FIG. 9B, the maximum
possible amount of change d of the dimensions of a pattern formed
on a lower layer is taken into consideration. Then, two resolution
positions .DELTA..sub.1 and .DELTA..sub.2, which are the case where
the dimensions is thick by the maximum amount of change d and the
case where the dimensions is thin by the maximum amount of change
d, respectively of an exposure intensity threshold E.sub.th are
calculated, and it is checked whether these resolution positions
.DELTA..sub.1 and .DELTA..sub.2 both are within the error
allowance. Since it can be considered that an actual resolution
position exists between those resolution positions .DELTA..sub.1
and .DELTA..sub.2, an error due to an error in the dimensions of a
formed pattern can be surely detected. In this case, the width on
the X-axis of an error graphic is one between the evaluation point
and the resolution position .DELTA..sub.2.
[0091] E1-3 shown in FIG. 4 indicate errors detected by the pattern
edge resolution position error verification method, the exposure
intensity contrast verification method and the exposure amount
margin verification method, respectively. In their parentheses of
symbols "E1", "E2" and "E3", an error from the evaluation point of
the resolution position .DELTA., contrast value C and errors from
the evaluation point of two resolution positions .DELTA..sub.1 and
.DELTA..sub.2, respectively are shown.
[0092] In the exposure verification unit 15, each of the simple
exposure verification unit 15a and the detailed exposure
verification unit 15b perform exposure verification using the
various types of verification methods described above. Thus,
exposure data is generated by modifying parameter values while
avoiding the actual occurrence of an error with high accuracy.
[0093] FIG. 10 is the flowchart of the exposure data generation
process. Next, the process of generating exposure data and its flow
are described in detail with reference to FIG. 10. The generation
process can be realized, for example, by the CPU 61 shown in FIG.
11 executing a program stored in the external storage device 65 or
a storage medium MD. Thus, the exposure data generation device of
the present invention can be realized by the computer whose
configuration is shown in FIG. 11 executing the program. Here, the
following description assumes that the external storage device 65
(storage unit 12) stores layout data D1 and exposure experiment
data.
[0094] Firstly, in step S1, the layout data D1 is read from the
external storage device 65 and layout data of an exposure target
layer and its lower layer is extracted. In step S2, the correction
parameter extraction data management table (FIG. 3) is stored in an
area secured in RAM 62 or the like. Then, the exposure experiment
data 2 is read from the external storage device 65 and is
registered in the table.
[0095] In step S3, parameter values are extracted from the data
stored in the data management table. In step S4, proximity effect
correction is applied to the layout data of the exposure target
layer using the extracted parameter values to generate exposure
data. In step S5, a window process is applied to the entire
exposure verification area (exposure data) to extract the
importance point of design.
[0096] In this window process, its design rules are checked and a
point requiring high accuracy is extracted as the importance point
of design.
[0097] In step S6, simple exposure verification is performed to
extract an error point. Then, the extracted error point is
registered in the verification error management table (FIG. 4). In
step S7, detailed exposure verification is applied to the error
point registered in the error management table, and a point that is
not determined to be erroneous by the verification is deleted from
the error management table.
[0098] In step S8, it is determined whether the process termination
criterion is met. If no error point is registered in the error
management table, if no change exists in the registered error or if
a parameter value cannot be updated within a predetermined range,
it is determined that the process termination criterion is met and
its determination is yes. Then, lastly, in step S9, the exposure
data D3 that is generated when performing step S4 is outputted and
the series of processes terminate. Otherwise, its determination is
no and the flow proceeds to step S10. In step S10, the error point
registered in the error management table is registered in the data
management table and the flow returns to step S3.
[0099] In step S3, the parameter values are updated (corrected)
taking into consideration the data (error point) added to the data
management table. In step S4, exposure data is re-generated. In
step S5, the window process is applied only to the error point
registered in the error management table and then the error
management table is cleared (initial setting). In this way, the
second time and after, in steps S3.about.S5, processes whose
contents are different from those of the first time are performed.
In step S6, since simple exposure verification can be performed in
high speed, the simple exposure verification is applied to the
entire exposure verification area. However, the simple exposure
verification can also be applied only to the point extracted by the
window process.
[0100] In the multi-layered semiconductor device, the above
described exposure data generation process is performed for each
layer. Thus, appropriate exposure data can be obtained for each
layer.
[0101] Although in this preferred embodiment, the method disclosed
by Patent reference 1 is adopted as the simple exposure
verification method (for the calculation of the backward scatter
intensity), another method can also be adopted. Since, for example,
in a point greatly influenced by backward scatter, the exposure
intensity is high in a non-exposure part, the amount of exposure
must be reduced in an exposure part. As a result, resolution
(contrast) tends to degrade. For that reason, such a point is
sometimes called the degree of risk as a simple evaluation index.
The degree of risk corresponds to the size of backward scatter
intensity, that is, approximates the size. The degree of risk (i,
j) of an area (i, j) is calculated as follows. In the following
equation, ".alpha..sub.k(i, j)" and "(1-.alpha..sub.k(i, j))" are
the pattern area density of a heavy material in a layer k and that
of a light material, respectively.
Degree of risk ( i , j ) = k = 1 N [ C k .alpha. k ( i , j ) + D k
{ 1 - .alpha. k ( i , j ) } ] ( 17 ) ##EQU00002##
[0102] When the surface of the semiconductor substrate is the same
in a sufficiently wide range, the size of the backward scatter
intensity has the following tendency.
(1) The higher the ratio of a heavy material (corresponding to a
pattern area density) is, the larger the backward scatter intensity
is. (2) The thicker the thickness of a heavy material is, the
larger the backward scatter intensity is. There is a relationship
of C.sub.k>D.sub.k between coefficients C.sub.k and D.sub.k
which are multiplied to .alpha..sub.k(i, j) and (1-.alpha..sub.k(i,
j)), respectively, due to the tendency (1). Therefore, the larger
the number of layers containing a heavy material, the higher the
degree of risk becomes. Thus, the tendency (2) is also taken into
consideration. For example, if in the case of two-layered
semiconductor substrate, 100% of a heavy material and 100% of a
light material exist in the first and second layers, the degree of
risk becomes C.sub.1.times.D.sub.2. If 100% of a heavy material
also exists in the second layer, the degree of risk becomes
C.sub.1.times.C.sub.2. Since C.sub.2>D.sub.2, in this example,
the degree of risk of the latter is higher. Thus, in the calculated
degree of risk, the tendencies (1) and (2) both are taken into
consideration. Therefore, it can be used to specify an error point
or a point with such a possibility.
[0103] The appropriate value of each of the coefficients C.sub.k
and D.sub.k varies depending on film thickness and area density
(pattern dimensions, etc.) like the parameters T, R and .sigma. and
has an appropriate range. For this reason, even when the degree of
risk is used for exposure verification, as a whole, each process
can be performed according to the flow shown in FIG. 10.
* * * * *