U.S. patent application number 16/659034 was filed with the patent office on 2020-02-13 for method and apparatus for semiconductor manufacturing.
The applicant listed for this patent is IMEC VZW. Invention is credited to Christopher Ausschnitt, Vincent Truffert.
Application Number | 20200050112 16/659034 |
Document ID | / |
Family ID | 59014655 |
Filed Date | 2020-02-13 |
![](/patent/app/20200050112/US20200050112A1-20200213-D00000.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00001.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00002.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00003.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00004.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00005.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00006.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00007.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00008.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00009.png)
![](/patent/app/20200050112/US20200050112A1-20200213-D00010.png)
View All Diagrams
United States Patent
Application |
20200050112 |
Kind Code |
A1 |
Ausschnitt; Christopher ; et
al. |
February 13, 2020 |
Method and Apparatus for Semiconductor Manufacturing
Abstract
Examples herein are related to a method and apparatus for
determining dimensions of features in a patterned layer of a chip
produced on a semiconductor production wafer. The production of the
patterned layer includes a lithography step and an etching step,
where the lithographic mask applied for producing the patterned
layer is provided with one or more asymmetric marks. The position
of printed and etched mark features is sensitive to lithographic
and etch parameters. Changes in these positions are measured by
overlay measurements, i.e. the measurement of the change in
position of one mark relative to another. The obtained `pseudo`
overlay data are fitted to a parametric model, while characteristic
feature dimensions are measured on a test wafer. The inverted model
allows determination of feature dimensions on a production wafer.
Application of the method on two different layers allows
determination of edge placement errors between features of the two
layers.
Inventors: |
Ausschnitt; Christopher;
(Heverlee, BE) ; Truffert; Vincent; (Woluwe
Saint-Pierre, BE) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
IMEC VZW |
Leuven |
|
BE |
|
|
Family ID: |
59014655 |
Appl. No.: |
16/659034 |
Filed: |
October 21, 2019 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
16308363 |
Dec 7, 2018 |
10481504 |
|
|
PCT/EP2017/064145 |
Jun 9, 2017 |
|
|
|
16659034 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G03F 7/70625 20130101;
G03F 7/70616 20130101; G03F 7/70633 20130101; G03F 7/705
20130101 |
International
Class: |
G03F 7/20 20060101
G03F007/20 |
Foreign Application Data
Date |
Code |
Application Number |
Jun 10, 2016 |
EP |
16174048.5 |
Mar 31, 2017 |
EP |
17164311.7 |
Claims
1. An apparatus for lithography and etching of features on a
semiconductor wafer, wherein the apparatus comprises: a metrology
tool configured to determine the position of a printed mark pattern
and an etched mark pattern obtained from an asymmetric metrology
mark included in a lithography mask applied in the apparatus, a
computer-implemented verification unit comprising a memory provided
with a computer program for executing at least the following steps,
when run on the verification unit: acquiring through the metrology
tool a plurality of position-representative values associated
respectively to a plurality of asymmetric marks, calculating values
of lithographic and/or etch parameters based on a first parametric
model that links the position-representative values to the
parameters, calculating characteristic dimensions of features on
the wafer based on a second parametric model that links the
lithographic and/or etch parameters to the characteristic
dimensions, evaluating the characteristic dimensions in comparison
with a tolerance.
2. The apparatus according to claim 1, wherein the verification
unit is furthermore configured to update one or more of the
lithographic and/or etch parameters, as required by the evaluation
result.
3. The apparatus according to claim 1, wherein the metrology tool
is an IBO tool or a DBO tool.
4. A verification unit as described in claim 1, comprising a memory
provided with a computer program for executing the steps, when the
program is run on the verification unit.
5. A computer program product applicable in the verification unit
of claim 4, and configured to execute the steps, when the program
is run on the verification unit.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] The present application is a division of U.S. patent
application Ser. No. 16/308,363 filed Dec. 7, 2018, which is a U.S.
National Phase Application pursuant to 35 U.S.C. .sctn. 371 of
International Application No. PCT/EP2017/064145 filed Jun. 9, 2017,
which claims priority to European Patent Application No. 16174048.5
filed Jun. 10, 2016 and also claims priority to European Patent
Application No. 17164311.7 filed Mar. 31, 2017. The entire
disclosure contents of these applications are herewith incorporated
by reference into the present application.
FIELD OF THE DISCLOSURE
[0002] The present disclosure is related to semiconductor
processing, in particular to a metrology method for monitoring
lithographic and/or etch processes during the fabrication of a
semiconductor chip.
BACKGROUND
[0003] Semiconductor processing can include many lithography and
etch steps, where a pattern defined in a lithographic mask is
exposed to a light source, resulting in the printing of the pattern
on a photoresist film deposited on a layer stack that is built up
layer by layer on a semiconductor wafer. After development of the
resist, etching reproduces the pattern in a layer of the stack, for
example for realizing a metal conductor pattern in a level of the
back-end-of-line stack of an integrated circuit chip. Feature
dimensions of the printed and etched patterns in present day
processing technology are on the order of nanometers, and the
monitoring of the printed/etched features can require specific
metrology tools.
[0004] The dependence of the feature dimensions on a plurality of
lithographic parameters has been the subject of intensive study.
Methods for monitoring the litho-process can include establishing a
parametric model that correlates dimensions of specific control
patterns to the set values of dose and focus applied in the
lithography tool. The measurement of dimensions of the control
patterns can take place via Scanning Electron Microscopy (SEM),
Atomic Force Microscopy (AFM), optical microscopy, or other means.
The determination of feature dimensions by SEM or AFM is however
time-consuming and often destructive. Also, the control patterns
applied in this way are typically not design rule compatible. As a
result, the influence of etch parameters on the manufactured
pattern is generally not taken into account. There is a need
therefore for alternative methods that allow a more reliable and
more non-intrusive way of verifying a pattern after lithography and
etching.
SUMMARY
[0005] This disclosure is related to a method for determining
dimensions of features in a patterned layer of a chip produced on a
semiconductor production wafer, i.e., a wafer used in a
manufacturing process for producing multiple chips on the wafer.
Within the present context, the term "production wafer" is
typically distinct from a "test wafer" used in the calibration
steps included in this disclosure. The production of the patterned
layer includes a lithography step and an etching step, where the
lithographic mask applied for producing the patterned layer is
provided with one or more asymmetric marks which are design rule
compatible, i.e. which are reproduced in the resist layer including
fine elements of the marks. The position of printed and etched mark
features is sensitive to lithographic and etch parameters.
According to some embodiments, changes in these positions are
measured by overlay measurements, i.e. the measurement of the
change in position of one mark relative to another. In particular,
optical overlay measurement is typically fast, non-destructive and,
since overlay is intrinsically a differential, isolation from
variation in the layers underlying the patterns of interest can be
achieved. Current overlay measurement capability routinely exhibits
single-Angstrom level precision. The obtained "pseudo" overlay data
obtained on a Focus Exposure Matrix (FEM) wafer, possibly also
including modulation of etch parameters, are fitted to a parametric
model, while characteristic feature dimensions are measured on the
FEM. The inverted model allows determination of feature dimensions
on a production wafer. Application of the method on two different
layers allows determination of edge placement errors between
features of the two layers.
[0006] According to the disclosure, a method is provided for
determining dimensions of features in a patterned layer of a chip
produced on a semiconductor wafer, the production of the layer
including a lithography step and an etching step, the lithography
step including: applying a lithographic mask which defines a
pattern of features and exposing a resist layer to a light source
through the mask, thereby obtaining printed pattern features, the
method comprising the steps of: including in the mask one or more
asymmetric marks which result in printed mark features (i.e. in a
printed mark pattern for each asymmetric mark) in the resist layer
after the lithography step, and in etched mark features after the
etching step, wherein: the position of the printed mark features
depends on the set values of one or more lithographic parameters
applied in the lithography step, the position of the etched mark
features depends on the set values of one or more etch parameters
applied in the etching step, determining on one or more test
wafers, the position or a value representative thereof, of the
printed and/or etched mark features, for a plurality of different
"set" values of the lithographic and/or etch parameters,
determining dimensions of one or more pattern features of the
patterned layer for the same plurality of set values of the
lithographic and/or etch parameters, determining one or more first
parametric models correlating the position (i.e. a value
representative of the position) of the printed and/or etched
asymmetric mark features to the plurality of different set values
of the lithographic and/or etch parameters, determining one or more
second parametric models correlating the measured feature
dimensions to the plurality of different values of the lithographic
and/or etch parameters, inverting the one or more first parametric
models.
[0007] The preceding steps are the calibration steps of the method.
The following steps may be referred to as the "production steps" of
the method. The calibration steps are performed once and result in
the first and second parametric model and the inverted first model.
The production steps may be applied at a plurality of locations on
a plurality of chips produced on one or more production wafers:
performing the lithography and etch steps on the production wafer
and determining the position of the same printed and/or etched
asymmetric mark features, calculating from the inverted first one
or more models a set of "get" values of the lithographic and/or
etch parameters, determining the dimensions of one or more
characteristic features on the production wafer by applying the
get" values to the one or more second parametric models.
[0008] The position of a printed or etched asymmetric mark feature
or a value representative thereof is determined by determining the
position of the centroid of the printed or etched mark pattern in
the direction of asymmetry, or a position that is representative of
the centroid position. A value representative of the centroid
position is for example a distance of the centroid to an offset
position, which may be the position of the centroid of a symmetric
or asymmetric offset mark pattern equally included in the litho
mask. A position representative of the centroid position for an
asymmetric printed or etched mark pattern may be determined based
on an intensity profile of an image of the mark pattern, in the
direction of asymmetry. The centroid-representative position may be
obtained by integrating the intensity profile with respect to a
reference intensity value and determining the centroid of the area
between the reference intensity and the profile.
[0009] According to an embodiment, the marks comprise at least one
pair of a symmetric mark and an asymmetric mark or at least one
pair of oppositely oriented asymmetric marks, and wherein in the
method step "determining on one or more test wafers, the position
or a value representative thereof, of the printed and/or etched
mark features, for a plurality of different `set` values of the
lithographic and/or etch parameters," a value is determined that is
representative of the position of the resulting pair of printed or
etched mark features, the value being a function of the distance
between the pair of printed or etched features, obtained by an
overlay measurement, the value being termed a "pseudo overlay
response." Typically, the distance is determined as the distance
between the centroids of the pair of oppositely oriented asymmetric
marks or of the pair of a symmetric mark and an asymmetric mark.
The pseudo-overlay response may be a shift of the distance with
respect to a reference value, or it may be distance itself.
[0010] According to an embodiment, the overlay measurement is an
image based overlay (IBO) measurement. The overlay measurement may
be based on the determination of the centroid of an image of the
asymmetric mark, for example obtained by determining the edges of
an intensity profile across the mark in the direction of
asymmetry.
[0011] According to an embodiment, the marks are arranged in one or
more IBO targets, located within or in the vicinity of the
patterned layer, where one IBO target comprises a plurality of
different asymmetric marks and/or a plurality of reproductions of
the same asymmetric mark in different orientations, allowing one to
obtain multiple pseudo-overlay responses as a function of the
lithographic and/or etch parameters. According to an embodiment,
one or more of the IBO targets comprises one or more cross-shaped
or wind-mill shaped clusters of reproductions of the same mark
arranged around a central point.
[0012] According to an embodiment, the overlay measurement is a
diffraction based overlay (DBO) measurement. According to an
embodiment, the marks are arranged in one or more DBO targets,
located within or in the vicinity of the patterned layer, where one
DBO target comprises a pair of interleaved gratings, each grating
formed of two marks (A,B) arranged adjacently and in a repeated
manner, at least one of the marks (A,B) being an asymmetric mark,
and wherein the distance between two adjacent marks A and B is
different in the first grating compared to the second grating, the
difference between the distances being pre-defined. According to a
further embodiment, a DBO target comprises two of the pairs of
interleaved gratings, the first pair being oriented perpendicularly
to the second pair.
[0013] According to an embodiment, additional feature dimensions in
the patterned layer of the production wafer are determined by
interpolation of a plurality of the characteristic feature
dimensions.
[0014] According to an embodiment, at least some of the asymmetric
marks comprise mark features, which are dimensioned and/or arranged
in a manner to improve the type of response of the mark pattern's
position or position-representative value to a change in one or
more lithographic and/or etch parameters, so that at least one
response exhibits a monotonic change as a function of one of the
lithographic and/or etch parameters
[0015] According to an embodiment, the chip is produced multiple
times on one or more semiconductor production wafers, and the
determination of the characteristic feature dimensions is done on a
chip produced on the one or more production wafers, the
characteristic feature dimensions are verified on the basis of a
manufacturing tolerance, the verification results are used to
calculate updated values of the lithographic and/or etch
parameters, the updated values being configured to maintain the
characteristic feature dimensions within the tolerance, the updated
values are applied in the production of one or more subsequent
chips.
[0016] The disclosure is furthermore related to a method for
determining an edge placement error (EPE) between two features of
two respective patterned layers of a semiconductor chip, comprising
the steps of: determining the dimensions of the first and second
feature by the above-described method, determining the overlay
error between the first and second layer, determining the edge
placement error on the basis of the overlay error, taking into
account the dimensions of the first and second feature as
determined in the first step.
[0017] According to an embodiment of the method for determining an
EPE, the lithographic masks for producing the two layers comprise
respective parts of a hybrid target, the first part comprising
marks provided in the first mask, the second part comprising marks
provided in the second mask, and wherein the measurement of the
overlay error between the first and second layer is obtained from
an overlay value measured between printed and/or etched mark
features resulting from the first and second parts. The overlay
error can be determined using either the relative layer-to-layer
placement of symmetric marks on each layer or the relative
layer-to-layer placement among the sum of the opposing asymmetric
marks on each layer.
[0018] According to some embodiments, the chip is produced multiple
times on one or more semiconductor production wafers, and the
determination of the EPE is done on a chip produced on the one or
more production wafers, based on the first and second parametric
models determined by the calibration steps of the method, and
verified on the basis of a manufacturing tolerance, the
verification results are used to calculate updated values of the
lithographic and/or etch parameters, the updated values being
configured to maintain the EPE within the tolerance, the updated
values are applied in the production of one or more subsequent
chips.
[0019] According to some embodiments of any of the above-described
methods, the lithographic parameters are the dose and defocus
values and wherein the etch parameter is the etch bias.
[0020] According to embodiments, the lithographic mask comprises
one or more IBO targets as described in any of FIGS. 4 to 10 and
16a, and any combination thereof, as well as any obvious variations
thereof in terms of the number of marks, the number of mark types,
the relative location of the marks with respect to each other,
etc.
[0021] The disclosure is equally related to any IBO targets
described in this specification. The disclosure is thus related to
an IBO target comprising one or more cross-shaped or wind-mill
shaped clusters of reproductions of the same mark arranged around a
central point. The disclosure is furthermore related to IBO targets
as described in any of FIGS. 4 to 10 and 16a, and any combination
thereof, as well as any obvious variations thereof in terms of the
number of marks, the number of mark types, the relative location of
the marks with respect to each other, etc.
[0022] According to an embodiment, at least one of the IBO targets
comprises a first group of asymmetric marks of which the uniform
and periodic portions are arranged in a first direction, and a
second group of the same asymmetric marks, of which the uniform and
periodic portion are arranged in a second direction perpendicular
to the first.
[0023] The disclosure is equally related to the DBO targets
described in this specification. The disclosure is thus related to
DBO targets as described in FIGS. 11b, 11c and FIGS. 12a and 12b,
as well as any obvious variations thereof in terms of the number of
marks, the number of mark types, the relative location of the marks
with respect to each other, etc. According to an embodiment, a DBO
target according to the disclosure comprises a pair of interleaved
gratings, each grating formed of two marks (A,B) arranged
adjacently and in a repeated manner, at least one of the marks
(A,B) being an asymmetric mark, and wherein the distance between
two adjacent marks A and B is different in the first grating
compared to the second grating, the difference between the
distances being pre-defined. According to a further embodiment, a
DBO target according to the disclosure comprises two of the pairs
of interleaved gratings, the first pair being oriented
perpendicularly to the second pair.
[0024] The disclosure is also related to an apparatus for
lithography and etching of features on a semiconductor wafer,
wherein the apparatus comprises: a metrology tool configured to
determine the position of a printed mark pattern and an etched mark
pattern obtained from an asymmetric metrology mark included in a
lithography mask applied in the apparatus, a computer-implemented
verification unit comprising a memory provided with a computer
program for executing at least the following steps, when run on the
verification unit: acquiring through the metrology tool a plurality
of position-representative values associated respectively to a
plurality of asymmetric marks, calculating values of lithographic
and/or etch parameters based on a first parametric model that links
the position-representative values to the parameters, calculating
characteristic dimensions of features on the wafer based on a
second parametric model that links the lithographic and/or etch
parameters to the characteristic dimensions, and evaluating the
characteristic dimensions in comparison with a tolerance.
[0025] The verification unit may furthermore be configured to
update one or more of the lithographic and/or etch parameters, as
appropriate for the evaluation result. The metrology tool may be an
IBO tool or a DBO tool. The verification unit may comprise a memory
provided with a computer program for executing the steps, when the
program is run on the verification unit.
[0026] The disclosure is also related to a computer program product
applicable in the verification unit of the disclosure, and
configured to execute the steps, when the program is run on the
verification unit.
BRIEF DESCRIPTION OF THE FIGURES
[0027] FIG. 1a shows marks, according to an example embodiment.
[0028] FIG. 1b shows marks, according to an example embodiment.
[0029] FIG. 1c shows marks, according to an example embodiment.
[0030] FIG. 1d shows marks, according to an example embodiment.
[0031] FIG. 1e shows marks, according to an example embodiment.
[0032] FIG. 1f shows marks, according to an example embodiment.
[0033] FIG. 1g shows marks, according to an example embodiment.
[0034] FIG. 1h shows marks, according to an example embodiment.
[0035] FIG. 2a illustrates how marks can be used to measure a
pseudo-overlay value, according to an example embodiment.
[0036] FIG. 2b illustrates how marks can be used to measure a
pseudo-overlay value, according to an example embodiment.
[0037] FIG. 3a illustrates how the centroid of a mark may be
determined by an image based overlay (IBO) tool, according to an
example embodiment.
[0038] FIG. 3b illustrates how the centroid of a mark may be
determined by an image based overlay (IBO) tool, according to an
example embodiment.
[0039] FIG. 4a shows IBO targets, according to an example
embodiment.
[0040] FIG. 4b shows IBO targets, according to an example
embodiment.
[0041] FIG. 5a shows IBO targets, according to an example
embodiment.
[0042] FIG. 5b shows IBO targets, according to an example
embodiment.
[0043] FIG. 6a shows IBO targets, according to an example
embodiment.
[0044] FIG. 6b shows IBO targets, according to an example
embodiment.
[0045] FIG. 7a shows IBO targets, according to an example
embodiment.
[0046] FIG. 7b shows IBO targets, according to an example
embodiment.
[0047] FIG. 8a shows IBO targets, according to an example
embodiment.
[0048] FIG. 8b shows IBO targets, according to an example
embodiment.
[0049] FIG. 9 shows IBO targets, according to an example
embodiment.
[0050] FIG. 10 shows IBO targets, according to an example
embodiment.
[0051] FIG. 11a shows a DBO target for the measurement of overlay
errors, according to an example embodiment.
[0052] FIG. 11b shows the structure of gratings of a diffraction
based overlay (DBO) target, according to an example embodiment.
[0053] FIG. 11c shows the structure of gratings of a diffraction
based overlay (DBO) target, according to an example embodiment.
[0054] FIG. 12a shows the layout of a DBO target adapted to
determine a pseudo-overlay response, according to an example
embodiment.
[0055] FIG. 12b shows an example of a DBO target designed for
obtaining a response in 2 orthogonal directions, according to an
example embodiment.
[0056] FIG. 13 shows pseudo-overlay data points as a function of
dose and defocus for two different asymmetric marks, as well as the
best fit of a parametric model based on the data points, according
to an example embodiment.
[0057] FIG. 14 illustrates the duty cycle parameter on a comb-type
metrology mark, according to an example embodiment.
[0058] FIG. 15 illustrates a method that is applied in two layers
and the overlay between the layers is taken into account to
determine an edge placement error between features in the two
layers, according to an example embodiment.
[0059] FIG. 16a shows an IBO target having marks on two different
layers, according to an example embodiment.
[0060] FIG. 16b shows a DBO target having marks on two different
layers, according to an example embodiment.
DETAILED DESCRIPTION
[0061] Example embodiments will now be described more fully
hereinafter with reference to the accompanying drawings. That which
is encompassed by the claims may, however, be embodied in many
different forms and should not be construed as limited to the
embodiments set forth herein; rather, these embodiments are
provided by way of example. Furthermore, like numbers refer to the
same or similar elements or components throughout.
[0062] According to embodiments of the disclosure, the lithographic
mask used for the production of a patterned layer is provided with
one or more metrology targets, each target comprising at least one
asymmetric mark. In some embodiments, the target comprises at least
one pair of a symmetric mark and an asymmetric mark, and/or one or
more pairs of oppositely oriented, possibly mirrored asymmetric
marks. Within the present context, a "mark" can be defined as a
predefined pattern designed for metrology purposes. A target
comprises one or more marks and is included inside or in the
vicinity of the area of the mask that contains the actual pattern
that is to be printed. Typical examples of symmetric and asymmetric
marks are shown in FIGS. 1a-h, which illustrate marks of the comb
type, both symmetric (FIG. 1a) and asymmetric (FIGS. 1b-1d) and of
the rail type, equally symmetric (FIG. 1e) and asymmetric (FIGS.
1f-1g). The asymmetric marks of this type are known as such, as
well as the fact that features printed by lithography according to
these patterns are sensitive to modulations of various lithography
parameters. The present disclosure applies this characteristic in a
method that is applicable for monitoring lithographically printed
features in a semiconductor chip production line, and also for
monitoring features obtained after etching. For this reason, the
symmetric and asymmetric marks applied in this disclosure are
generally to be design rule compatible, meaning that the marks are
dimensioned and designed so that fine-pitched features of the
patterns, such as the fine-pitched teeth of the comb-type marks,
can be transferred to a resist layer, so that the fine-pitched
pattern can further be transferred to an underlying layer, by an
etch process. Note that the asymmetric marks can be further
segmented in the direction transverse to the comb or rail to
maintain design rule compatibility on layers comprised of
two-dimensional patterns; e.g., contact or block layers.
[0063] The two lithographic parameters that are used in this
disclosure are the dose and defocus values, de-focus being the
deviation from a pre-defined zero-defocus setting on the exposure
tool, expressed for example in nm. Dose is defined as the energy
applied through the mask during exposure, expressed for example in
mJ/cm.sup.2. Both defocus and dose are values that can be set on
the lithography tool. These dose and defocus set values are
hereafter referred to as "Eset" and "Fset" respectively, or simply
"E" and "F." The etch parameter used in the present description is
the etch bias, which refers to the difference in the lateral
dimension between the etched image and the lithographically printed
image. The determination of the etch bias E generally involves a
well-defined lithographically printed feature. A "set" value of the
etch bias is referred to in the present description as "Bset" or
"B," expressed for example in nm. The etch bias can be set by
applying specific settings to control parameters during the etch
process, like etch time, etch rate, voltage, chamber pressure etc.
It is pointed out that other parameters may be used instead of or
in addition to dose, defocus and etch bias.
[0064] First, a number of configurations of marks will be
explained. According to embodiments of the disclosure, the position
of printed or etched mark features obtained from an asymmetric mark
is to be understood as the position of the centroid of the mark
pattern obtained after litho or etch respectively, in the direction
of asymmetry (i.e. the direction perpendicular to the longitudinal
direction of the mark in the case of elongated marks as shown in
the drawings). The centroid is the center line of the integrated
area of a mark pattern (either symmetric or asymmetric), i.e. the
area covered by the printed or etched mark pattern is the same to
the left and to the right of the centroid. In embodiments, the
position of a printed and/or etched asymmetric mark pattern is
expressed in terms of the distance from the centroid of the mark
pattern to an offset position that is either independent from the
litho and etch parameters or that itself changes in a similar way
as the mark pattern, as a function of the parameters. According to
an embodiment, the offset position is itself defined by a symmetric
or an asymmetric mark included in the litho-mask. The offset
asymmetric mark may be the mirror image of the mark of which the
position is being determined. The target then comprises one or more
mark pairs, designed for metrology purposes.
[0065] FIG. 2a shows a target comprising a pair of a symmetric mark
and an asymmetric comb-type mark. FIG. 2b shows a target comprising
a pair of two mirrored asymmetric comb-type marks. In each of the
FIGS. 2a and 2b, a view is shown at the top of the figures, of the
as-designed shape of the marks. The distance S.sub.ref between the
centroids of the two marks is the distance obtained when the marks
are reproduced in an essentially ideal way by litho or etch: the
width w of the mark, the pitch p, and the shape and dimensions of
the comb features is exactly according to the design of the mark.
These ideal process conditions are hereinafter also referred to as
the nominal process conditions. After the lithography step,
illustrated at the bottom of the figures, these marks result in
printed reproductions of the marks in a resist layer, i.e. printed
mark features. As the marks are design rule compatible, they are
equally reproduced after etching, for example in the form of
trenches in a layer underneath the resist ("etched mark features").
The printed and etched reproductions depend on the process
parameters such as E, F, and B. When these parameters differ from
the ideal set of parameters, the dimensions of the reproduced mark
patterns differ from the as-designed values, as is illustrated in
FIG. 2b. The position of the centroid of a printed or etched
symmetric mark pattern is nevertheless insensitive to changes in E,
F and B, as a consequence of the symmetry. Due to its asymmetric
structure, the position of the centroid of a printed or etched
asymmetric mark pattern shifts in the direction perpendicular to
the longitudinal direction of the mark, as a function of changes in
E, F and B. The shift is indicated as a function .delta.(E,F,B) in
FIGS. 2a and 2b. When two mirrored asymmetric marks are used, the
shift between the two equals 2.delta.. The distance S between the
centroids is equally indicated in the drawings. .delta. and
2.delta. are obtained as:
[0066] .delta.=S-S.sub.ref in the case of FIG. 2a
[0067] 2.delta.=S-S.sub.ref in the case of FIG. 2b
[0068] In embodiments, the position of asymmetric marks is
determined by detecting and measuring the shift .delta. or 2.delta.
by metrology tools known as such in the art for measuring overlay
errors, i.e. measurements for determining the shift between two
different layers produced by different lithographic masks and
caused by an overlay error between the masks applied for printing
the two layers. Because the shift is now measured in a single
layer, the measured shift is referred to in the present context as
a "pseudo-overlay value" or "pseudo-overlay response," as the value
responds to (is sensitive to) the litho and etch parameters used
for printing and etching the mark features. The pseudo-overlay
response will hereafter be expressed by the letter O. Instead of
the shift .delta. or 2.delta., the distance S itself can be
measured and used as a pseudo-overlay response. The pseudo-overlay
response can take the form of a value representative of the
position of a printed and/or etched asymmetric mark pattern. The
pseudo overlay response can be measured by image based overlay
(IBO) tools or diffraction based overlay (DBO) tools. These tools
are known for the measurement of overlay errors. The layout of the
marks will be different in each case as will be explained
hereafter.
[0069] An IBO target can consist of a plurality of marks of the
types as shown in FIGS. 2a and 2b, distributed across the field of
view (FOV) of the IBO tool. The printed features have an overall
width "w" that is resolvable by the IBO tool. For example, "w" may
be of the order of 500 nm. The fine comb structure has a pitch "p"
that is not resolvable by the IBO tool, e.g., "p" is on the order
of 100 nm. The IBO tool is thus able to determine the relative
distance between the centroids of the target features, as
illustrated in FIGS. 3a and 3b which illustrate the image as seen
by the IBO tool of a symmetric mark pattern and an asymmetric
comb-type mark pattern, as well as the intensity profile as a
function of the position in the direction perpendicular to the mark
pattern. As the fine pitch p is not resolvable by the IBO tool, the
tool sees an asymmetric intensity profile across the mark in the
direction of asymmetry. It can approximate the centroid position
X.sub.C by measuring the relative position of the edges of the
intensity profile at a given intensity setting I.sub.set, i.e.,
X.sub.C=(X.sub.1+X.sub.2)/2. A typically more accurate way of
determining the centroid position is by integrating the intensity
profile with respect to a reference value I.sub.ref, i.e.
calculating the area between I.sub.ref and the intensity profiles,
and determining the centroid of the profiles, i.e. the location of
the centroid line so that the calculated areas to the left and
right of the line are the same. Because the centroid position is
based on the image, it may differ from the mathematical centroid
position of the printed or etched mark pattern as defined above in
relation to FIGS. 2a and 2b. The measured shift .delta. or 2.delta.
of the image centroid will however be proportional to the
mathematical centroid shift when measured by the IBO tool.
Metrology targets comprising at least one pair of oppositely
oriented asymmetric marks or a pair of a symmetric mark and an
asymmetric mark, as shown in FIGS. 2a and 2b, are hereafter
referred to as IBO targets.
[0070] FIGS. 4a and 4b show an IBO target that takes up a minimal
portion of the printed or etched area. The area of the printed
target may be about 2 square microns. The target comprises two
marks 1 and 1' of the same asymmetric design and rotated
180.degree. one with respect to the other. The marks may be any of
the asymmetric types shown in FIG. 1. The asymmetry of the marks is
illustrated in a symbolic way by a distinct hatching type of
portions 2 and 3 to indicate a pattern density gradient across the
two regions. For example, for a comb type mark, portion 2 is a full
rectangle and portion 3 is an array of comb features. The marks are
positioned side by side along their longitudinal direction (i.e.
the direction of the dividing line between the differing pattern
densities of portions 2 and 3). FIG. 4a shows the as-designed
printed or etched marks (obtained at nominal process conditions):
the distance between the centroids is S.sub.ref. When the process
conditions are perturbed and thereby deviate from the nominal
conditions, a shift 2.delta. between the centroids can be detected
by an optical IBO tool (see FIG. 4b).
[0071] Other types of IBO targets are illustrated in FIGS. 5a, 5b,
6a, and 6b. They each comprise two pairs of mutually oppositely
oriented asymmetric marks 4/4' and 5/5', arranged in a windmill
pattern (FIGS. 5a-b) or a cross pattern (FIGS. 6a-b). The four
marks are all reproductions of the same mark design. The centroid
shift 2.delta. can thereby be measured in two orthogonal directions
x and y. It is seen that the four centroids of the four marks form
a rectangle with a fixed center O.sub.C, regardless of whether the
nominal process conditions (FIGS. 5a and 6a) or perturbed process
conditions (FIGS. 5b and 6b) are applied. This characteristic,
namely the fact that the point O.sub.C is fixed in a particular
layer, can be used to detect an overlay error between two layers
applied in the process. When the IBO target of FIG. 5a, 5b, 6a, or
6b is included at the same location in two masks applied
consecutively in the process, the IBO tool can detect the shift of
the point O.sub.C between the two layers, and thereby detect an
overlay error in two orthogonal directions. It is to be noticed
that the midpoint (line) between the centroids in the target of
FIGS. 4a and 4b is a fixed line, and can thus be used to determine
overlay errors in one direction. The area of printed/etched surface
taken up by the targets of FIGS. 5a, 5b, 6a, and 6b may be about 4
square microns.
[0072] FIG. 7a shows yet another IBO target formed by a mark
cluster as printed under nominal conditions. Again the design is a
windmill cluster of asymmetric marks of the same design, but this
time each of the two pairs of marks 6/6' and 7/7' extending in x
and y direction respectively are formed of 2 marks (6 and 6' for
one pair and 7 and 7' for the second pair) which have the same
orientation. As a consequence, when the process is perturbed (FIG.
7b), the center O.sub.C of the rectangle formed by the centroids
shifts to a new position O.sub.C'. In order to measure the shift
2.delta., a 180.degree. oriented version of the windmill cluster of
FIG. 7a can be used.
[0073] FIG. 8a shows a target design comprising the cluster 10 of
FIG. 7a repeated in the upper right and lower left quadrant of the
target layout, and a 180.degree. rotated version 10' of the cluster
in the two remaining quadrants (upper left and lower right). This
target allows measurement of the centroid shifts in two directions
by measuring the relative movement of the center points O.sub.C1
with respect to the center points O.sub.C2. The center point of the
four clusters O.sub.D however remains fixed regardless of
perturbations from the nominal process, and can thus be used for
actual overlay measurements between printed/etched layers in the
manner described above.
[0074] FIG. 9 finally shows a similar target design as in FIGS. 8a
and 8b, but wherein two different mark types M1 and M2 are applied,
for example a comb type and a rail type, thereby allowing to
distinguish two different E,F,B responses in x and y directions
(e.g., four pseudo-overlay responses in total). The center point
O.sub.D is again usable for overlay measurements.
[0075] FIG. 10 shows a further example of an IBO target. The marks
included in the IBO target shown are arranged in a matrix of 24 x-
and y-directed crosses that fit within the field-of-view (FOV) of
the IBO tool, where the FOV center is denoted by a unique pattern
recognition mark 15. Asymmetric mark-crosses A.sub.1, A.sub.2,
A.sub.3 and A.sub.4 are identified in FIG. 10, as well as symmetric
mark-crosses S.sub.1 and S.sub.2. Each of the asymmetric
mark-crosses A.sub.1-A.sub.4 comprises 4 identical marks (i.e.
reproductions of the same mark design) oriented so that the center
point of the cross shifts as a function of the process parameters
(as in FIGS. 7a and 7b). But the marks in the four crosses
A.sub.1-A.sub.4 are all distinct. It is seen that the asymmetry of
the asymmetric marks A.sub.1-A.sub.4 in the upper right (UR) and
lower left (LL) target quadrants is opposite that of the marks in
the upper left (UL) and lower right (LR) target quadrants. A set of
4 crosses consisting of marks of a particular asymmetric design
(for example the 4 A.sub.1 crosses in the four quadrants) then
defines a specific (E, F, B) response in both x and y directions. A
potential advantage of the IBO target is that it enables the
simultaneous capture of multiple (E, F, B) responses within one
target. In the example of FIG. 10, the two sets of symmetric marks
S.sub.1 and S.sub.2 are usable as a reference. This leaves the 4
sets of 4 asymmetric marks A.sub.1-A.sub.4, usable to differentiate
4 different (E, F, B) responses in both x and y directions (i.e. 8
pseudo-overlay responses).
[0076] As an alternative to IBO targets, diffraction based overlay
(DBO) targets can be applied. FIG. 11a shows the layout of a DBO
overlay target as known in the art, formed of an arrangement of
optically distinct elements (A,B) in an interlaced diffraction
grating configuration with a pitch "P," typically on the order of
500-1000 nm. In other words, a first grating consisting of elements
A is interlaced with a second grating consisting of elements B.
Within each period "P" are the repeating elements (A,B). The
grating elements A and B must be sufficiently different to cause a
measurable difference between the plus and minus first order
diffracted intensity .DELTA.I=I.sub.+1-I.sub.-1. This intensity
difference is measured in a DBO overlay tool. It is a measurement
for the overlay value OVL which allows determining overlay errors
when the elements A and B are printed through different lithography
masks.
[0077] A DBO target can include a similar interlaced grating
configuration, but with elements A and B included in the same
lithographic mask. In the terminology of the present disclosure,
elements A and B are referred to hereafter as marks. In addition,
at least one of the marks A or B is asymmetric in the sense
described above. FIGS. 11b and 11c show two examples of the layout
of such a DBO target. In the first example (FIG. 11b), mark A is
symmetric and mark B is asymmetric: it is a comb mark comprising a
fine structure arranged at a maximum pitch p<<P that
precludes diffraction induced by the fine structure itself, which
could confound or detract from the diffraction from the grating
elements at the coarse pitch P. FIG. 11c shows an example where
both A and B are asymmetric. A and B cannot be mirror images
however: they must have dimensions sufficiently different to cause
a measurable difference in the plus and minus first order
diffracted intensity. The marks applied in at least one grating
must be asymmetric in order for patterning process conditions (E,
F, B) to induce a relative movement of the centroids C of the
asymmetric marks B of one printed and/or etched grating with
respect to the centroids C' of the marks A of the symmetric grating
(FIG. 11b) or with respect to the centroids C' of the marks A of
the opposing asymmetric grating (FIG. 11c). Such relative movements
cause a proportional change in the relative intensity of the plus
and minus first order diffraction from the gratings. Thus, a DBO
tool can measure an (E, F, B) response by measuring that intensity
difference on a printed or etched version of the DBO target, formed
of the gratings comprising the printed or etched marks A and B.
[0078] In order to determine the shift of the distance S between
two adjacent marks A and B within an interleaved grating, a DBO
target is provided (see FIG. 12a) that comprises two gratings 30
and 31 formed by the same marks A and B but with a well-defined
difference in the distance between two adjacent marks A and B. The
period P is the same for both gratings 30 and 31. In grating 30,
the "as designed" distance between the centroids of two adjacent
marks A and B is 0.5P+.DELTA.D. In grating 31, this distance is
0.5P-.DELTA.D. The shift .DELTA.D is a well-defined value that is
significantly smaller than 0.5P so that gratings 30 and 31
represent slightly shifted versions of the grating with .DELTA.D=0.
When the target is printed or etched, the centroid position of the
asymmetric mark B will be shifted over a distance .delta., as
explained with reference to FIG. 2. This means that: the distance
between the printed or etched mark patterns A and B in grating 30
is 0.5P+.DELTA.D+8, and the distance between the printed or etched
mark patterns A and B in grating 31 is 0.5P-.DELTA.D+.delta..
[0079] The intensity differences .DELTA.I.sub.1 and .DELTA.I.sub.2
are measured on the printed or etched versions of the two gratings,
using a DBO tool. These parameters are proportional to the values
.DELTA.D+.delta. and -.DELTA.D+.delta. via a diffraction
coefficient K:
.DELTA.I.sub.1=.kappa.(.DELTA.D+.delta.)
.DELTA.I.sub.2=.kappa.(-.DELTA.D+.delta.)
[0080] Solving .kappa. and .delta. from these equations leads
to:
.kappa.=(.DELTA.I.sub.1-.DELTA.I.sub.2)/2.DELTA.D
.delta.=.DELTA.D(.DELTA.I.sub.1+.DELTA.I.sub.2)/(.DELTA.I.sub.1-.DELTA.I-
.sub.2)
[0081] The latter equation allows one to determine the
pseudo-overlay response .delta. on the basis of a measurement of
.DELTA.I.sub.1 and .DELTA.I.sub.2. A typical DBO target layout is
shown in FIG. 12b. The target comprises two pairs of gratings. The
first pair 35,36 has marks A and B with the direction of asymmetry
of the asymmetric mark B in the x-direction and with the
shifts+.DELTA.D and -.DELTA.D applied between the positions of
marks A and B (i.e. identical to the gratings 30 and 31 of FIG.
12a). The second pair 37,38 are the same as 35,36 but rotated over
90.degree.. The set of 4 gratings is therefore designed to measure
one (E,F,B) response in the x and y directions (i.e. two
pseudo-overlay responses .delta., one in x and one in y direction).
DBO targets therefore have the potential disadvantage compared to
IBO targets that they can require a larger area of the mask if
multiple responses in multiple orientations are required.
[0082] So according to some embodiments of the disclosure, a
lithographic mask for producing a given pattern is provided with
one or more IBO and/or DBO targets as described above, each target
providing one or more (E,F,B) responses in the form of
pseudo-overlay values O.sub.1 to O.sub.n. In order to establish the
relation between O.sub.1 . . . O.sub.n and the set values of E, F
and B, a calibration procedure is performed. The procedure will be
described in detail for the calibration with regard to dose and
defocus only, but the method is analogous when the etch bias is
included.
[0083] As an example, an IBO target may be provided that permits
the measurement of a total of n pseudo overlay values O.sub.1 . . .
O.sub.n (for example 4 in x direction and 4 in y direction in the
case of the target of FIG. 10), which are a function of dose and
defocus settings. In order to determine the functions, a
calibration procedure is performed on one or more test wafers,
known as Focus Exposure Matrix (FEM) wafers. On such a test wafer,
consecutive prints are made in a resist layer, each print being
made with a different value of the set dose value E.sub.set and the
set de-focus value F.sub.set. For example, the de-focus value may
be incremented in constant steps for a number of constant dose
settings, the latter equally being incremented in constant steps.
The ranges for the dose and defocus settings may be based on
preliminary simulations for example. In each print, the values of
O.sub.1 . . . O.sub.n are measured in an overlay measurement tool
as described above. This results in a data set wherein O.sub.1 . .
. O.sub.n can be set out as data points in a graph, as a function
of E and F.
[0084] In addition to this, the dimensions of a number of features
within the printed pattern are equally measured in each print on
the FEM wafer. These features are selected within the pattern,
e.g., in the vicinity of the target or targets. Features are
selected which determine the functionality of the chip, i.e. when
these features are not produced according to a predefined
manufacturing specification, the chip will or is very likely to
malfunction. In the present disclosure, these selected features are
referred to as characteristic features. Alternatively or in
addition to the measurement of characteristic feature dimensions on
the printed pattern, the dimensions may be measured after etching
the pattern at a given value of the etch bias. These dimensions may
be measured by any suitable technique, such as SEM, AFM or the
like. A total of m of these measurements yields a set of data
CD.sub.1 . . . CD.sub.m, as a function of E and F.
[0085] A first parametric model is then fitted to the pseudo
overlay data O.sub.1 . . . O.sub.n. In other words, mathematical
functions O.sub.1=f.sub.1(E,F) . . . O.sub.n=f.sub.n(E,F) are
determined which represent the best approximation of the measured
overlay data.
[0086] A second parametric model is fitted to the characteristic
feature dimension data i.e. mathematical functions
CD.sub.1=d.sub.1(E,F) . . . CD.sub.m=d.sub.m(E,F) are determined
which represent the best approximation of the measured feature
dimension data.
[0087] The first parametric model is then inverted. This is a
mathematical operation, yielding functions E.sub.get=F(O.sub.1, . .
. , O.sub.n) and F.sub.get=G(O.sub.1, . . . , O.sub.n). The dose
and defocus values are labeled E.sub.get and F.sub.get in order to
distinguish them from the set values applied initially. The "get"
values are the dose and focus values derived from overlay
measurements on a particular target. E.sub.get and F.sub.get may
differ from the set values due to the model approximation and due
to other factors such as measurement noise and/or process variation
on the FEM wafer that maps onto effective dose/focus or errors in
the dose/focus settings.
[0088] According to the disclosure, the inverted first model and
the second model are subsequently used for determining dimensions
of the characteristic features on a production wafer, i.e. during
the production of a chip comprising a layer that is printed and
etched using the same mask, including the same targets used in the
calibration. When the layer is printed and etched on the production
chip, the n overlay values O.sub.1 . . . O.sub.n are measured by an
overlay tool. These values are used for calculating E.sub.get and
F.sub.get using the inverted first parametric model, which are
subsequently entered in the second parametric model to calculate
the characteristic feature dimensions CD.sub.1(E.sub.get,F.sub.get)
. . . CD.sub.m(E.sub.get,F.sub.get) on the production chip. The
feature dimensions are thereby determined without requiring
potentially destructive measurements such as SEM and AFM on the
production chip.
[0089] When the etch bias B is taken into account as a parameter in
the model, the above-described methodology is the same. The
calibration data as a function of B will generally be gathered on
the same or further FEM wafers, onto which the previously produced
resist patterns are transferred by etching into an underlying
layer. On consecutive dies of the FEM wafer, the settings of the
etch tools are configured so that each die or each group of dies is
etched with a different etch bias B. Measurement of the
pseudo-overlay of the FEM wafer(s) after etching now yields a data
set of O.sub.1 . . . O.sub.n as a function E,F and B. A parametric
model can be fitted to this data in an analogue manner as described
above. Possibly pseudo-overlay data could also be gathered after
litho and before etching. A first parametric model can then be
built as a function of dose and defocus based on the pre-etch
pseudo overlay and another first parametric model can be built
based on the post-etch data, as a function of dose, defocus, and
etch bias. Another embodiment would entail basing the calibration
on the etch bias B only, and not on E and F. As indicated already,
the characteristic feature dimensions CD.sub.1 . . . CD.sub.m may
be measured on the printed pattern or on the etched pattern or on
both. When measured on both, the second parametric model will in
fact be determined both for the printed features and for the etched
features. Instead of or in addition to E, F and B, any number of
other lithography and etch parameters can be used in the method.
For example, systematic conditions that affect imaging in resist
can take the form of aberration of lens systems and scan stability
in the litho tool, resist thickness and bake plate temperature in
the resist processing tool, as well as systematic conditions that
affect the transfer of the resist image by etch, such as the
pressure, voltage, gas mix of the etch reactor and pattern loading
effects on the wafer.
[0090] The characteristic feature dimensions obtained on the
production chip by the method of the disclosure may be fed back to
a control unit incorporated in or connected to the lithographic
and/or etching tools applied in the production process, where they
are verified on the basis of a given manufacturing tolerance with
respect to theoretical values. The verification allows the
calculation of updated values of the processing parameters used in
the first and second parametric models (E,F,B). These updated
values help ensure that the characteristic dimensions on subsequent
production chips stay within the manufacturing tolerance. Any known
control algorithm and control hardware may be applied for realizing
the above-described control strategy. The control strategy is
typically applied by performing the feedback and update of the
processing parameters on the basis of sample measurements rather
than on measurements taken on each chip on the production wafer,
i.e. the characteristic dimensions are determined by the method of
the disclosure on a limited number of chips, each determination
allowing updating the processing parameters applied in producing
one or more subsequent chips.
[0091] By way of example, FIG. 13 shows a data set obtained from an
IBO target having two different asymmetric marks, a comb pattern as
shown in FIG. 1b and a dagger pattern as shown in FIG. 1c. So
overlay measurements O.sub.C and O.sub.D are obtained by measuring
the pseudo overlay between mirrored comb patterns and mirrored
dagger patterns respectively. The data in FIG. 13 is shown as a
function of normalized "set" dose and defocus values E and F. The
dots show the overlay data measured on an FEM wafer, the upper data
sets being obtained from the comb-type marks and the lower data
sets being obtained from the dagger-type marks. The curves show the
best fit for a parametric model of the following form:
O.sub.C=a.sub.00+a.sub.10E+a.sub.01F
O.sub.D=b.sub.00+b.sub.10E+b.sub.02(F-F.sub.0).sup.2
[0092] The best fit to the data was obtained for the following set
of parameters:
TABLE-US-00001 TABLE 1 O.sub.C a.sub.00 -0.89 a.sub.10 0.68
a.sub.01 -0.86 O.sub.D b.sub.00 -2.17 b.sub.10 0.42 b.sub.02 -0.35
F.sub.0 -1.99
[0093] The inverse model was then calculated as follows:
E get = - B .+-. B 2 - 4 AC 2 A ##EQU00001## F get = O C - a 00 - a
10 E get a 01 ##EQU00001.2##
[0094] wherein:
A = b 02 ( a 10 a 01 ) 2 ##EQU00002## B = b 10 - 2 b 02 ( a 10 a 01
) ( O C - a 00 a 01 - F 0 ) ##EQU00002.2## C = b 00 - O D + b 02 (
O C - a 00 a 01 - F 0 ) 2 ##EQU00002.3##
[0095] The calibration procedure is typically done for a plurality
of targets distributed across the area of a printed pattern, i.e.
each target allows one to determine a set of overlay measurements
O.sub.1 . . . O.sub.n which are fitted to the set E/F and/or B
values to yield a parametric model. Characteristic feature
measurements CD.sub.1 . . . CD.sub.m are made in the vicinity of
each target. Provided that the model parameters (the parameters
a.sub.mn, b.sub.mn and F.sub.0 in the above example) between
targets differ insignificantly or linearly, the characteristic
feature dimensions derived from overlay measurements on the
production wafer can then be used to determine feature dimensions
of any number of pattern features by linear interpolation between
feature dimensions determined by the method of the disclosure.
[0096] The disclosure generally involves the measured shifts over
the lithography energy-focus process window being sufficiently
distinct and continuous (low-order) to enable unique inversion
using analytical or numerical techniques; i.e., the responses of
two or more marks are transformable to physically meaningful dose
and focus values. Meeting this condition can require the
identification of an appropriate set of marks during process setup.
Marks typically exhibit a monotonic, substantially linear response
to dose, but a more complex and variable focus response. Thus, mark
selection usually devolves to the evaluation and identification of
distinguishable focus response.
[0097] The inventors have studied the influence of the mark shape
and dimensions on the quality of the parametric model that links
the pseudo-overlay values to dose and focus (E and F). As seen in
FIG. 13, the pseudo-overlay response of the dagger-type mark in the
X-direction is a parabolic function. The peak of the parabola
indicates an optimum focus for the dagger mark. The comb-type mark
exhibits a monotonic response to focus, however the data points
show deviations from the linear model. Also, the response of both
the dagger and comb marks as a function of dose is monotonic and
substantially linear.
[0098] Further analysis has now led to the finding that, for a
given lithography exposure condition, the type of response to focus
of X and Y-shifts derived from comb- or dagger-type marks is
sensitive to the duty cycle and pitch of the comb or dagger array.
These parameters are illustrated in FIG. 14. The duty cycle is the
ratio of the width of the periodic features to their pitch. In a
DUV exposure case (193 nm illumination, 1.35NA), it was found that
a substantially linear monotonic response to focus is obtained when
the duty cycle of the comb mark is 64 nm/96 nm or 0.67. The
behavior changes to a non-monotonic response at lower duty
cycles.
[0099] A monotonic response is not required for all responses, but
it can make the inversion task more straightforward and ensure
uniqueness. What can be required are two distinct realizable
responses that enable unique inversion. This benchmark is satisfied
by a monotonic dose response and a parabolic focus response paired
with a monotonic dose response and a monotonic focus response (as
illustrated above). It could also be satisfied by a monotonic dose
response and a parabolic focus response having one vertex focus
paired with a monotonic dose response and a parabolic focus
response having a substantially different vertex focus.
[0100] Therefore, the set of marks needed to realize the inventive
method can be appropriate for the lithography exposure conditions.
For DUV exposure, an embodiment applies at least one target
comprised of comb and dagger marks that cover a range of duty cycle
and pitch tailored to the lithographic exposure condition. IBO
targets are well suited to this method because multiple marks can
be evaluated simultaneously in a single target.
[0101] Generally, according to embodiments of the disclosure, at
least some of the asymmetric marks comprise mark features (for
example legs of the comb), which are dimensioned and/or arranged in
a manner to adjust the type of response of the mark pattern's
position to a change in one or more process parameters, so that at
least one response exhibits a monotonic change as a function of one
of the process parameters.
[0102] The disclosure is equally related to a method for estimating
edge placement errors between pattern features of two different
layers in a layered structure of a semiconductor chip, for example
between consecutive layers applied in the back end of line process.
FIG. 15 shows a detail of two superimposed patterns, each having a
line feature 16 and 16' respectively. The line features are shown
with respect to the as-designed features 17 and 17'. Each of these
features 16/16' is produced using a separate litho and etch step,
i.e. with two different masks and with a potential overlay error
between the two. According to the disclosure, the dimensions
CD.sub.1 and CD.sub.2 are determined in each layer by the
above-described method, i.e. after lithography and/or after
etching. In other words, in each of the two masks, a target
comprising one or more asymmetric marks is provided allowing the
measurement of O.sub.1 . . . O.sub.n after the litho and/or etching
steps, the calculation of E.sub.get/F.sub.get/B.sub.get, and the
subsequent calculation of CD.sub.1 and CD.sub.2 as a function of
the respective E.sub.get/F.sub.get/B.sub.get values. Then a "real"
overlay measurement is done between a symmetric mark in the first
layer and a symmetric mark in the second layer. A pair or a cluster
of mirrored asymmetric marks can thereby be used as a symmetric
mark, as mentioned in relation to FIGS. 4-9 (the measurement of the
position of O.sub.C or O.sub.D in the two layers yields the overlay
value). This results in the overlay error OVL indicated in the
figure. The edge placement error (EPE), equally indicated in FIG.
15, between the features 16 and 16' in the final device is then
calculated on the basis of this real overlay value, but taking into
account the CD values obtained by the method in each layer. This
allows an accurate determination of the EPE.
[0103] In order to facilitate this multi-layer approach, the
disclosure provides for the use of hybrid targets, as illustrated
in FIGS. 16a and 16b. FIG. 16a shows a hybrid IBO target arranged
in a matrix of 48 x- and y-directed crosses that fits within the
field-of-view (FOV) of the IBO tool. Twenty four crosses are in the
central area 18 and twenty four are in the surrounding area 19. The
central area 18 may for example have the layout shown in FIG. 10.
Generally the same marks are used in the two areas. The crosses in
the central area are included in the litho-mask for producing a
first layer, and the crosses in the surrounding area are included
in the litho-mask for producing a second layer. Each mask thus has
4 sets of asymmetric marks, allowing the determination of O.sub.1 .
. . O.sub.4 in each layer and in each direction x and y. For
determining the overlay OVL between the layers, the shift can be
measured for one symmetric mark in the first layer with respect to
the same symmetric mark in the second layer, or the shift can be
measured for one pair of mirrored symmetric marks in the first
layer to the same pair of mirrored asymmetric marks in the second
layer.
[0104] FIG. 16b shows a DBO target distributed over two layers,
i.e. distributed over two litho-masks for producing the two layers.
The targets 20/21 on the left yield two different E/F/B responses
in layer 1 in both x and y directions. These targets are each
formed of four gratings, in the same way as shown in FIG. 12b, i.e.
two pairs of gratings formed of the same marks A and B, one pair
oriented in the x direction and the other pair oriented in the y
direction, each pair comprising two gratings with the same marks A
and B but with the distance between adjacent marks A and B shifted
respectively over +.DELTA.D and -.DELTA.D. The marks A and B used
in target 20 are different than the ones used in target 21. For
example, target 20 is formed of symmetric marks A and asymmetric
comb marks B, while target 21 is formed of symmetric marks A and
asymmetric rail marks B. So in total, the marks 20 and 21 allow one
to determine four pseudo-overlay responses O.sub.1 . . . O.sub.4 in
the form of .delta.-values as defined with respect to FIG. 12a: 2
overlay values in the x direction (respectively the .delta.-values
of the x-oriented gratings in marks 20 and 21) and 2 in the y
direction (respectively the .delta.-values of the y-oriented
gratings in marks 20 and 21). In the same way, the targets 22/23 on
the right yield two different E/F/B responses in layer 2 in both x
and y, i.e. four pseudo overlay responses O.sub.1 . . . O.sub.4.
Targets 22 and 23 are typically identical to the targets 20 and 21.
The central target 24 comprises gratings with elements A and B
which are symmetric but which are partially in the first layer and
partially in the second layer (as in FIG. 11a). An overlay
measurement between the two layers based on this target 24 thereby
allows one to measure the overlay error OVL between the layers.
[0105] The edge placement error determined in the above-described
way may equally be used in a control strategy for determining
updated values of the litho and/or etch parameters based on a
manufacturing tolerance, configured to maintain the edge placement
error within the tolerance. The EPE determination according to the
disclosure is typically done on a sample number of chips produced
on a production wafer, allowing an update of processing parameters
in the process for producing one or more subsequent chips.
[0106] While some of the focus in the above description has been on
the use of the pseudo-overlay response as a position representative
parameter, it should also be noted that the disclosure is not
limited to the use of the pseudo-overlay values as such a
parameter. Any parameter that is sensitive to the position of a
printed or etched asymmetric mark can be used, as long as it allows
one to determine invertible first and second parametric models in
the manner described above.
[0107] The disclosure is equally related to an apparatus for
lithography and etching of features on a semiconductor wafer,
wherein the apparatus comprises a metrology tool configured to
determine the position of a printed mark pattern and an etched mark
pattern obtained from an asymmetric metrology mark included in a
lithography mask applied in the apparatus. The metrology tool may
comprise an IBO or DBO based overlay tool as described above,
configured to measure a pseudo-overlay value or a value equivalent
thereto, that is representative of the position of the asymmetric
mark pattern. The apparatus furthermore comprises a
computer-implemented verification unit comprising a memory provided
with a computer program for executing at least the following steps,
when run on the verification unit: acquiring through the metrology
tool a plurality of position-representative values associated
respectively to a plurality of asymmetric marks, calculating values
of lithographic and/or etch parameters based on a first parametric
model that links the position-representative values to the
parameters, calculating characteristic dimensions of features on
the wafer based on a second parametric model that links the
lithographic and/or etch parameters to the characteristic
dimensions, and evaluating the characteristic dimensions in
comparison with a tolerance.
[0108] Typically, the verification unit is furthermore configured
to update one or more of the lithographic and/or etch parameters,
as appropriate for the evaluation result.
[0109] While some embodiments have been illustrated and described
in detail in the appended drawings and the foregoing description,
such illustration and description are to be considered illustrative
and not restrictive. Other variations to the disclosed embodiments
can be understood and effected in practicing the claims, from a
study of the drawings, the disclosure, and the appended claims. The
mere fact that certain measures or features are recited in mutually
different dependent claims does not indicate that a combination of
these measures or features cannot be used. Any reference signs in
the claims should not be construed as limiting the scope.
* * * * *