Umted States Patent [19]
`Sogard
`
`[54] METHOD AND APPARATUS FOR AERIAL
`IMAGE ANALYZER
`
`[75] Inventor: Michael R. SOgardQ Menlo Park. Calif
`
`_
`. .
`_
`[73] Assrgnee: Nikon Precision, Inc.. Belmont, Calif.
`
`[21] Appl. N0.: 209,026
`[22] Filed:
`Mar. 9, 1994
`
`6
`
`[51] Int. Cl. ............................. .. G01J 1/42, G03B 27/42
`[52] U:S. Cl. .............................. .. 356/121; 355/53; 355/68
`[58] Fleld of Search ................................... .. 356/121, 122,
`356/123, 124, 218, 223, 225, 226, 399-401;
`250/306, 307, 309, 310. 311; 355/53. 551%
`
`.
`
`[56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`US00563173 1A
`[11] Patent Number:
`[45] Date of Patent:
`
`5,631,731
`May 20, 1997
`
`4,929,083
`5,001,737
`5,105,305
`5,123,734
`
`5/1990 Brunner.
`3/1991 Lewis et a1. .
`4/1992 Betzig et a1. .
`6/1992 Spence et a1. ........................ .. 356/121
`
`FOREIGN PATENT DOCUMENTS
`
`0147925 6/1990 Japan ................................... .. 356/121
`OTHER PUBLICATIONS
`Dieter W. Pohl, et a1. “Optical 'Ihnnelling through and
`Adjustable Liquid Metal Gap”. Nato Advance Research
`Workshop on Near Field Optics, v_ 242. Presented Oct
`26_28’ 1992, Published Aug. 1993, pp. 51__58_
`H_ E Maya, et a1_ “A new step-by-step aligner for very
`large scale integration (VLSI) production.” Semiconductor
`Microlithography V 9-18, SPIE vol. 221 (1980).
`Primary Examiner—Hoa O. Pham
`flittomeydAgeéztbror FlilrlirlzgrDavid G. Beck; Townsend and
`
`ownsen an
`[57]
`
`ew
`
`ABSTRACT
`
`Re. 32,795 12/1988 Matsuura et a1. ..................... .. 356/121
`3,938,894
`2/1976 Nanba .
`4,357,100 11/1982 Mayer et a1. .
`4,443,096
`4/1984 Joh?lmsmeief el al- -
`4,456,368
`6/1984 Isaka et al- ~
`4,498,767
`2/1985 McGovern 6t
`4,540,277
`9/1985 Mayer et al. .
`4 585 342 4,1986 Lin et aL _
`4,660,981
`4/1987 Stridsberg ............................. .. 356/398
`4:662:747
`5/1987 Isaacson et al _
`4,684,206
`8/1987 Bednorz et a1. .
`4,725,727
`2/1988 Harder et a1. .
`4,917,462
`4/1990 Lewis et al. .
`
`.................. .. 356/121
`
`A method and apparatus to analyze the aerial image of an
`optical system using a subwavelength slit. A slit con?gura
`tion yields a higher signal-to-noise ratio than that achievable
`with a round aperture. The slit also allows the polarization
`of the aerial ilnage to be analyzed In an alternative emb0di_
`u 1i .
`d Th '
`u
`1i
`.
`ment a tunne ng s t1s use .
`e tunne ng s tcompnses an
`Optically transPment ridge-11kB Strum“ mwntcd to a
`subsatlrzgleI~n the combined structure covered by a thin, planar
`met
`1
`
`38 Claims, 22 Drawing Sheets
`
`N2
`//I‘/' \\\
`
`/
`
`Nikon Exhibit 1008 Page 1
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 1 0f 22
`
`5,631,731
`
`N
`/ ,/ \\
`
`" 'I U‘
`
`I
`
`10
`
`/
`
`Nikon Exhibit 1008 Page 2
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 2 of 22
`
`5,631,731
`
`/
`I
`
`REPLACE IIASK
`WITH SPECIAL PATTERN
`
`REPLACE NAFER IYITH
`SUBNAVELENCTII SLIT
`
`SCAN SLIT
`ACROSS INACE
`
`/-32
`
`//34
`
`NONITOR AND RECORD
`NTENSITY PROFILE OF TII
`NEAR FIELD TRANSNISSID
`
`I
`I
`/
`
`CONPARE NONITORED
`INTENSITY PROFILE TO
`THEORETICAL ‘PERFECT’
`INTENSITY PROFILE
`
`DETERMINE PERFORIIANC
`CHARACTERISTICS OF
`STEPPER SYSTEM
`
`7.“
`5/
`
`F/(i l5.
`
`Nikon Exhibit 1008 Page 3
`
`

`

`U.S. Patent
`
`May 20, 1997
`
`Sheet 3 of 22
`
`5,631,731
`
`F/GI
`
`2.
`
`Nikon Exhibit 1008 Page 4
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 4 0f 22
`
`5,631,731
`
`4|
`
`l
`
`42
`42
`45 —%¢//7////"/////‘/////'////
`4 6
`
`FIG 3A.
`
`-— SCAN
`
`| Ir
`
`11F
`
`lg
`
`FIG 4.
`
`Nikon Exhibit 1008 Page 5
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 5 of 22
`
`5,631,731
`
`Slit
`
`—-°— Mo I00 nm
`
`—'=*— M0 200 nm - -X- - Mo 50 nm
`IN AIR
`
`‘1
`
`PERF. CONDCTR.
`50 nm
`
`0
`
`PERE
`CONDCTR
`I00 nm
`
`A‘ PERE
`CONDCTR.
`200 nm
`
`musmmuct o 2
`
`0.001
`
`0.000!
`0
`
`0.05
`
`0.!
`
`0.25
`
`0.3
`
`0.35
`
`0.2
`015
`METAL THICKNESS (um)
`
`FIG: 5.
`
`Nikon Exhibit 1008 Page 6
`
`

`

`U.S. Patent
`
`May 20, 1997
`
`Sheet 6 of 22
`
`5,631,731
`
`ii
`
`I
`
`/7|
`
`I
`
`FIG 6/].
`
`77‘
`
`/////////'/////4//
`‘In
`,QM
`/////////,i///////\//
`
`P76: 65.
`
`>77
`
`P15
`
`i
`
`/75
`
`12
`
`72
`
`//////////l’//////t///
`
`FIG: 6C
`
`Nikon Exhibit 1008 Page 7
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 7 0f 22
`
`5,631,731
`
`+ Cr50nm + Cr|00nm + Cr200nm
`
`TRANSNlSSION
`
`p 2
`
`0.00!
`
`0
`
`0.05
`
`0.l
`
`0.2
`M5
`METAL THICKNESS (um)
`
`0.25
`
`0.3
`
`FIG. Z
`
`Nikon Exhibit 1008 Page 8
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 8 0f 22
`
`5,631,731
`
`+ Cr50nm + Cr I00 nm + Cr 200 nm
`
`TRANSMISSION
`
`000i
`0
`
`0.05
`
`0|
`
`0.2
`0. I5
`METAL THICKNESS (um)
`
`0.25
`
`0.3
`
`0.35
`
`FIG: 8!
`
`Nikon Exhibit 1008 Page 9
`
`

`

`U.S. Patent
`
`May 20, 1997
`
`Sheet 9 of 22
`
`5,631,731
`
`4- Cr 50 nm
`
`+ Cr I00 nm
`
`+ Or 200 on
`
`-°- Mo 50 nm
`
`-<>— No [00 am
`
`+ No 200 nm
`
`l0000
`
`I000
`
`0
`
`10
`
`0.02
`
`0.04
`
`0.:
`0.08
`0.06
`METAL mom-:55 (um)
`
`F I61 9.
`
`Nikon Exhibit 1008 Page 10
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 10 0f 22
`
`5,631,731
`
`+ Cr50nm
`
`-°- M050 nm + CrIOO nm
`
`—<>— Mo I00 mn + Or 200 nm + H0 200 nm
`
`2:315
`
`METAL THICKNESS (um)
`
`F761 /0.
`
`Nikon Exhibit 1008 Page 11
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 11 of 22
`
`5,631,731
`
`+ Cr 50 nm + Cr I00 nm + Or 200 nm
`
`0.4
`
`k
`
`?
`E 03
`
`o N
`
`2;
`
`/
`
`I / M4
`
`0
`
`0.05
`
`DJ
`
`0.2
`015
`METAL THICKNESS (um)
`
`0.25
`
`0.3
`
`0.35
`
`F/GI ll
`
`Nikon Exhibit 1008 Page 12
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 12 0f 22
`
`5,631,731
`
`-<>— METAL THICKNESS=
`I20 nm
`
`° METAL THICKNESS=
`H5 nm
`
`A METAL TH|CKNESS=
`I25 nm
`
`0.7
`
`0.6 L
`
`
`0.5 if 04
`2%
`
`' i
`'- o2.
`
`0.2
`
`0]
`
`\d,
`
`‘
`
`’
`
`0
`40
`
`45
`
`50
`
`T0
`
`75
`
`B0
`
`65
`60
`55
`SLIT WIDTH (nm)
`
`FIG /2.
`
`Nikon Exhibit 1008 Page 13
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 13 0f 22
`
`5,631,731
`
`— -O—— TE MODE + TM MODE
`50 0m
`50 nm
`- —A-- TE MODE
`—*— TM MODE
`200 nm
`200 nm
`
`‘ '°" TE MODE + TM MODE
`I00 nm
`I00 “m
`
`I
`|
`coueasm- NA= 0.6
`//
`f
`
`\
`\\\\\
`\ I‘ ‘ \
`\X:\ \
`\ \\
`\\
`
`\
`'\
`\
`\\
`
`‘
`
`2
`
`,
`
`mconénsm-wus
`
`A
`
`'
`
`\ \
`
`03
`
`0.0
`
`0.7
`
`06
`
`0.4
`
`0.5
`
`0.2 0.1
`
`0
`0
`
`0.5
`
`I
`
`2.5
`2
`1.5
`SPATIAL FREQUENCY (l/um)
`
`F l6: l3.
`
`a
`
`.
`3.5
`
`U
`4
`
`Nikon Exhibit 1008 Page 14
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 14 of 22
`
`5,631,731
`
`INTENSITYHTF
`
`0
`
`0.5
`
`I
`
`I5
`
`2
`
`2.5
`
`3
`
`3.5
`
`4
`
`SPATIAL FREQUENCY (I/ um)
`
`FIG:
`
`/4.
`
`Nikon
`
`Exhibit1008
`
`Page15
`
`Nikon Exhibit 1008 Page 15
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 15 0f 22
`
`5,631,731
`
`LOOE-OB
`
`MOE-09
`
`
`
`TRANSIITTED POWER (I)
`
`LOOE-ID
`
`LOOE-ll
`
`LOOE-IZ
`0
`
`0.05
`
`0.!
`
`0 2
`015
`FILM THICKNESS (um)
`
`0.25
`
`0.5
`
`+ Cr 50 nm
`
`-°- Cr I00 nm + Cr 200 nm
`
`4-H!) 50 nm
`
`-°-Mo I00 nm
`
`-¢- M0 200 nm
`
`FIG: /5.
`
`Nikon Exhibit 1008 Page 16
`
`

`

`U.S. Patent
`
`May 20, 1997
`
`Sheet 16 of 22
`
`5,631,731
`
`
`
`I'll‘ III‘I
`
`.9 wt
`
`“2 2 m2 2 ma . _
`
`E: on Q2 lol E: cow .6 lil Ea Q9 .5
`
`523525 E:
`
`a 23
`
`.2 ca 2.
`
`.6. =2 8:: LT
`
`E. cm a IT
`
`2-23
`
`:22
`
`#22
`
`Q22
`
`2.23 :55
`
`
`
`2.32 5%“?
`
`lllllll
`
`s-§._
`
`2-82
`
`Nikon Exhibit 1008 Page 17
`
`

`

`US. Patent ‘
`
`May 20, 1997
`
`Sheet 17 of 22
`
`5,631,731
`
`190?}
`194%
`
`19s»
`
`19|
`
`/ / / / / A192
`FIG. l7?
`
`0.2
`
`»
`
`-
`
`015 -
`
`n1.) M 0.: -
`
`g;
`
`'
`
`‘
`
`‘I
`
`5 0.05 ~'
`
`.200
`
`-me
`
`200
`
`I00
`0
`1 (ma)
`F/GI 20A
`10
`I
`\"Eb _ X
`\ 1U OPTICALLY TRANSPARENT
`TI MATERIAL
`/ / / / ‘/ x0/ /
`suasmm
`F/GI 20B.
`
`METAL
`
`Nikon Exhibit 1008 Page 18
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 18 0f 22
`
`5,631,731
`
`COAT SUBSTRATE
`WITH PHOTURESIST
`U
`EXPUSE PHOTORESIST USING
`A 'LINE' (0R 'DOT') MASK
`L
`DEVELOP PNOTORESIST
`
`DEPOSIT TRANSPARENT
`NRTERIAL
`
`STRIP RESIST
`
`r205
`
`rZIO
`
`/2|5
`
`/220
`
`/ 225
`
`[Z50
`
`ROUND OFF
`PROTRUDING STRUCTURETS)
`
`[240
`
`/250
`
`DEPOSIT PLANAR
`METAL FILM
`
`THIN PLANAR
`NETM. FILN
`
`F I61 I84.
`
`Nikon Exhibit 1008 Page 19
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 19 of 22
`
`5,631,731
`
`OPTICAL umocmm? APPROX 500 nm
`
`OPTICAL 0R FLUORESCENT
`gamma
`DEPOSIT OPTICAL MATERIAL
`
`\ A
`
`/
`
`CHEMICAL ETCH 0R me
`
`L
`
`DEPOSIT METAL W @//
`
`M
`PLANARIZE (POLISH) W
`LW____J
`
`FIG. /8B.
`
`Nikon Exhibit 1008 Page 20
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 20 0f 22
`
`5,631,731
`
`OPTICAL LITHOGRAPHY
`
`
`RESIST
`
`
`
`
`APPROX. 500 nm
`
` SUBSTRATE
`
`
`
`Egg—:7
`
`RIDGE 0N SUBSTRATE
`
`Q/SPUTTER THIN METAL FILM
`
`
`
`IREMOVE FORM
`
`MASTER
`
`STAMP
`
`\
`
`k
`
`
`
`/
`,’
`MAyERIAO/
`/
`
`
`
`,
`/
`
`/
`
`OPTICAL
`/
`/I
`
`
`DEPOSIT METAL / PLANARIZE
`\
`(POLISH) \
`’ /"'
`7 / ‘7',
`
`$4,224..
`
`————-—
`
`{11%
`
`FIG I86:
`
`Nikon
`
`Exhibit1008
`
`Page 21
`
`Nikon Exhibit 1008 Page 21
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 21 of 22
`
`5,631,731
`
`0.8
`
`0.6
`
`24x).
`IOOO
`
`5-H!) 0.4
`
`0.2
`
`-200
`
`-100
`
`0
`
`1
`(nm)
`
`IOO
`
`200
`
`FIG.
`
`[9.4.
`
`
`Z(x )
`OPTICALLY TRANSPARENT
`METAL
`
`MATERIAL
`
`
`SUBSTRATE
`
`FIG [.98.
`
`Nikon
`
`Exhibit 1008
`
`Page 22
`
`Nikon Exhibit 1008 Page 22
`
`

`

`US. Patent
`
`May 20, 1997
`
`Sheet 22 of 22
`
`5,631,731
`
`-+-INVERTED
`Vac-TE
`
`+INVERTED
`Vac-TM
`
`_+-PARABOLA
`-TE
`
`+ PARABOLA
`~Tl
`
` 0
`
`0.5
`
`I
`
`l.5
`
`2
`
`2.5
`
`3
`
`3.5
`
`4
`
`SPATIAL FREQUENCY (l/ um)
`
`FIG: 21
`
`'
`
`Nikon
`
`Exhibit1008
`
`Page23
`
`Nikon Exhibit 1008 Page 23
`
`

`

`1
`METHOD AND APPARATUS FOR AERIAL
`IMAGE ANALYZER
`
`BACKGROUND OF THE INVENTION
`
`The present invention relates generally to self metrology
`of optical projection lithography systems, and more
`specifically, to a method and apparatus for analyzing the
`aerial image of such a system using a subwavelength slit.
`As the feature sizes on integrated circuit devices have
`grown ever smaller, the demands on the optical train in the
`optical lithography system have grown ever greater. This has
`led to an increased need for techniques for accurate and
`real-time monitoring of the optical system’s performance.
`In a typical prior art technique for monitoring system
`performance. a reticle pattern is imaged onto a resist coated
`substrate. The substrate is then developed and used to
`analyze the optical system’s performance. The analysis is
`based on the pattern being a convolution of a perfect image
`of the reticle and the performance characteristics of the
`stepper’s optical system Unfortunately. the nonlinear quali-
`ties of the photoresist technique adds a third variable making
`it diflicult to accurately reduce the data and derive the
`performance of the optical system. A further problem with
`this method of analysis is that it is slow and time consuming.
`SUMMARY OF THE INVENTION
`
`The invention provides a method and apparatus for ana-
`lyzing the aerial image of an optical system using a sub-
`wavelength slit (a slit narrower than the wavelength of the
`light illuminating it) or its functional equivalent. The aerial
`image refers to the intensity distribution of an image in or
`near the image focal plane of an optical system. Analysis of
`this image, especially in systems operating near the diffrac-
`tion limits of the optical system, provides valuable real-time
`information on the performance of the optical system.
`A slit is used instead of a round aperture in order to
`increase the amount of energy transmitted beyond the slit.
`Subwavelength round apertures in metal films of finite
`thickness transmit essentially only evanescent light. Eva-
`nescent light refers to electromagnetic fields produced in the
`immediate vicinity of the aperture edges. These fields don’t
`propagate like normal electromagnetic waves and therefore
`cannot be detected remotely. However.
`if a dielectric
`medium is present, some fraction of the evanescent light is
`converted to propagating waves which can be detected.
`Although a slit also produces evanescent light, for incident
`light with the magnetic vector parallel to the slit edge a
`propagating mode always exists. which produces a substan-
`tial increase in the slit’s transmission.
`
`In the preferred embodiment. a slit plate is included on the
`wafer holder stage of an optical projection lithography
`system. In order to examine the aerial image of the optical
`system prior to exposing the wafer, the stage is repositioned
`such that the slit plate will lie proximate to the image plane
`of the optical system. After replacing the reticle with a
`special test pattern. the slit is scanned across the image plane
`while the transmission of the slit is monitored. This provides
`the user with a high resolution intensity profile of the
`generated image. In the preferred embodiment. a fluorescent
`material mounted in the near field of the slit converts
`
`transmitted light. including evanescent light. to longer wave-
`lengths which are subsequently detected by a photodetector.
`From the intensity profile of the aerial image. errors in the
`optical system can be determined. Thus this system offers a
`real time. high fidelity method of monitoring the perfor-
`mance of the optical system.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`4o
`
`45
`
`50
`
`55
`
`60
`
`65
`
`5,631,731
`
`2
`
`The aerial image most readily analyzed by the slit is a line
`image. oriented parallel to the slit long axis. and longer than
`the slit length. The slit plate preferably includes slits with
`several orientations. so that line patterns with different
`orientations can be analyzed. The slits and their correspond-
`ing patterns are positioned so that only the slit, or slits, of a
`given orientation are illuminated at a given time.
`In an alternative embodiment, a tunneling slit is used
`instead of the subwavelength slit. The tunneling slit, a
`functional equivalent to the subwavelength slit. has no
`actual slit or interruption in its surface. It is designed and
`fabricated to have transmission characteristics similar to
`
`those of a subwavelength slit. The tunneling slit is poten-
`tially easier to fabricate then the subwavelength slit and yet
`provides comparable performance.
`In an alternative embodiment, the aerial image consisting
`of a parallel series of images is analyzed using a series of
`parallel slits with the same spacing as the image periodicity.
`In this embodiment the slits can either be subwavelength
`slits or tunneling slits. Due to the multiple slit arrangement,
`the intensity is increased as is the signal-to-noise ratio.
`Studies show a 150 nm wide slit of molybdenum (Mo) or
`silicon (Si) with vertical walls and thickness of about 120
`nm can represent close to an ideal aerial image monitor. This
`slit reproduces the width of the aerial image to within about
`5% for both transverse electric mode (TEM) and transverse
`magnetic mode (TMM) polarization states. Furthermore the
`transmissions for TEM and TMM agree to within about 5%.
`The image contrast determined by the slit is in good agree-
`ment with that of the aerial image. From signal to noise
`considerations it should be possible to measure an image in
`less than 1 sec using a single slit 10 mn long.
`A subwavelength slit for use with the present invention
`may be fabricated with vertical sidewalls (sometimes
`referred to as a standard slit) or sloping sidewalls
`(sometimes referred to as a vee slit). Vertical walled slits can
`be fabricated using electron-beam lithography. However,
`vee-shaped slits and tunneling slits can be fabricated from
`silicon (Si) using optical lithographic techniques.
`Reference to the remaining portions of the specification
`and the drawings will provide further understanding of the
`nature and advantages of the invention.
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`FIG. 1A shows an optical projection lithography stepper
`system utilizing the aerial image analyzer of the present
`invention;
`FIG. 1B shows the use of one embodiment of the present
`invention as a means of measuring the optical performance
`of the stepper;
`FIG. 2 shows one example of a special test pattern;
`FIG. 3A shows the simultaneous examination of several
`parallel line images;
`FIG. 3B shows an alternative embodiment utilizing a light
`pipe and a photomultiplier;
`FIG. 4 shows the analytical waveguide model;
`FIG. 5 shows a series of plots of transmission versus film
`thickness for slits in which the slit walls are parallel and
`vertical;
`FIGS. 6A—6C show other slit geometries;
`FIG. 7 shows a series of plots of transmission versus film
`thickness for vee geometry slits with 50. 100, and 200 um
`slit widths;
`FIG. 8 shows a series of plots of transmission versus film
`thickness for inverted vee geometry slits with 50. 100. and
`200 nm slit widths;
`
`Nikon
`
`Exhibit 1008
`
`Page 24
`
`Nikon Exhibit 1008 Page 24
`
`

`

`5,631,731
`
`3
`
`FIG. 9 shows the signal-to—background ratio versus film
`thickness for 50. 100, and 200 um slit widths;
`
`FIG. 10 shows the fraction of TM polarization in the
`transmitted signal for the vertical side wall case;
`FIG. 11 shows the degree of TM polarization versus film
`thickness for the vee slit geometry;
`FIG. 12 shows the TM polarization versus slit width for
`the vee slit geometry;
`FIG. 13 shows the modulation transfer function (MTF)
`for a vee geometry slit with a Mo film thickness of 120 nm
`and slit widths of 50. 100, and 200 nm;
`FIG. 14 shows conditions in which the TB and TM MTF’s
`are almost the same;
`
`FIG. 15 shows the transmitted power from a slit. assum-
`ing a slit length of 10 run and an illumination intensity of 100
`mW per square centimeter;
`FIG. 16 shows the calculated signal-to-noise ratio (S/N);
`FIG. 17 shows the tunneling slit;
`FIG. 18A. 18B. and 18C show the preferred fabrication
`techniques for the tunneling slit.
`FIG. 19A shows the calculated transmission through a
`parabolic tunneling slit;
`FIG. 19B shows the dimensional variables used in FIG.
`19A;
`FIG. 20A shows the calculated transmission through a
`triangular (or “inverted vee”) tunneling slit;
`FIG. 20B shows the dimensional variables used in FIG.
`20A; and
`FIG. 21 shows the MI‘F’s for tunneling slits.
`
`DESCRIPTION OF SPECIFIC EMBODIMENTS
`
`System Overview
`FIG. 1A shows an optical stepper lithography system 10.
`In a typical chip processing configuration. a light source 12
`and optics 14 image a reticle 15 onto a wafer 17. Wafer 17
`is mounted to an x-y—z positioning stage 18. Stage 18 allows
`step and repeat patterns to be applied to wafer 17. Also
`mounted to stage 18 is a slit plate 11 which,
`in this
`embodiment. contains several slits 13. each oriented in a
`different direction. A rectilinear coordinate system 20 is
`denoted schematically.
`FIG. 1B shows the use of one embodiment of the present
`invention as a means of measuring the optical performance
`of modified stepper 10. As in the prior art technique, reticle
`15 is replaced with a special test pattern (step 30). Stage 18
`is translated until wafer 17 is replaced with subwavelength
`slit 13 or its functional equivalent (step 32). In this position
`the slit is lying in. or within a few wavelengths of. the image
`plane. The slit is then moved or scanned across the image
`plane. the long axis of the slit being held perpendicular to the
`scan direction (step 34). During the scanning operation. the
`intensity of the slit’s transmission is monitored and recorded
`(step 35). The measured intensity profile is compared to a
`previously calculated profile expected from a “perfect”
`stepper (step 36). From this comparison the performance
`characteristics of the optical system are determined (step
`38).
`FIG. 2 shows one example of a test pattern. Several sets
`of line patterns. at diiferent orientations to the stepper’s x—y
`coordinate system are positioned in different regions of the
`image field of the stepper optics. Such a pattern should allow
`the image quality to be well characterized over the entire
`image field and allow the separate determination of a
`number of basic lens aberrations. such as spherical aberra-
`
`4
`tion. coma. and astigmatism. Additional sets of line patterns
`of different line width can provide additional information.
`which can be used to optimize exposure conditions. The
`present invention therefore offers the advantage of being
`able to perform real time. high fidelity analysis of the
`performance of the optical system
`FIG. 3A shows the simultaneous examination of several
`
`parallel aerial line images 41. Instead of a single slit being
`used. multiple slits 42 are used, thereby increasing the signal
`level. The period of the slits should be either equal to or a
`multiple of the period of the aerial image to be analyzed in
`order to achieve the desired higher signal levels. The long
`axes of all of the slits 42 are held perpendicular to the scan
`direction. In this embodiment. a fluorescent material 45 is
`mounted within the near field of slits 42. Material 45
`converts the light to longer wavelengths where it is detected
`by a photodetector 46. The use of fluorescence has the
`advantage that it is easier to detect light in the longer
`wavelengths. FIG. 3B shows an alternative embodiment in
`which fluorescent material 45 and detector 46 are replaced
`with a light pipe 47 which transmits the photons to a
`photomultiplier 48.
`Slit Characteristics
`
`FIG. 4 shows an example of the analytic waveguide
`model used to analyze the behavior of various slit designs.
`The slit is treated like a waveguide in the z direction (normal
`to the plane of the film). If the slit dimensions vary with 2,
`several waveguide sections or ‘slabs’ 51, one above the other
`and of varying sizes to match the z dependence. are used.
`For each slab 51 the eigenfunctions are calculated. Polar-
`ization effects and evanescent light are included. For an
`incident plane wave. boundary conditions at the interfaces
`between the slabs. the substrate. and the air are matched to
`obtain eigenvalues for all slab eigenfunctions. For conduct-
`ing slabs the eigenvalues are complex. Next the S—matrix
`which couples the incident plane wave to the transmitted
`wave is calculated. The S-matrix completely defines the
`transmission properties of the slit structure. For an aerial
`image, the image is Fourier transformed into plane wave
`components. Then the S—matrix is applied to transmit these
`components through the slit where they are recombined to
`give the transmitted amplitude and intensity. Because the
`above analysis uses a grating equation formalism in its
`calculations. a set or array of slits is assumed in all calcu-
`lations. A repeat distance 52 of 2 pm is used.
`FIG. 5 shows a series of plots of transmission versus film
`thickness for slits in which the slit walls are parallel and
`vertical. The definition of transmission here requires some
`explanation. Normally transmission would be defined as the
`fraction of light incident on the slit which is transmitted. The
`transmitted power/th length is then proportional to the slit
`width and the transmission coefficient. However. transmis—
`sion is defined here as the fraction of light incident over one
`repeat distance which is transmitted. to allow the effects of
`the partial transparency of very thin films to be included.
`In the preferred embodiment. the slit is fabricated using
`standard micro~circuit processing techniques which are well
`known in the art. The slit film sits on top of a dielectric
`substrate representing the fluorescent converter which, in the
`preferred embodiment,
`is made of CaFZ. The index of
`refraction is taken as 1.468. The presence of the substrate
`has some effect on the transmission. Slit films of both
`chrome (Cr) and Mo are analyzed for slit widths of 50. 100.
`and 200 nm. each. Mo and Si have almost identical optical
`constants and therefore can be expected to perform similarly.
`Slits fabricated from Si. however, should be much easier to
`make than those of Mo.
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`45
`
`50
`
`55
`
`65
`
`Nikon
`
`Exhibit 1008
`
`Page 25
`
`Nikon Exhibit 1008 Page 25
`
`

`

`5,631,731
`
`5
`As shown in FIG. 5, in most cases for small metal film
`thickness the transmission through the slits in the real metal
`films exceeds that for the perfectly conducting case. This is
`because of the tunneling of light through the film which is
`significant for thin films. The effect is less for the Mo films,
`because the extinction factor of M0 is about twice that of Cr.
`
`FIGS. 6A—C show other slit geometries. A slit 71 is a
`‘standard’ slit in which the slit walls 72 are vertical and
`parallel to the optic axis 73. A slit 74 has a ‘vee’ geometry
`while slit 75 has an ‘inverted vee’ geometry. The angles
`between the slit walls and optic axis 73 for both slits 74 and
`75 are 45 degrees. but other angles (say 30—60 degrees) can
`be used. Arrow 76 indicates the direction of the light
`incident on the slits. A material 77 indicates the slit sub-
`
`strates. Creating either the vee slit 74 or the inverted vee slit
`75 in metal would probably be diflicult, but they can be
`made in a straightforward way using single crystal Si in the
`<100> orientation. and anisotropic etching. However. the
`angle between the slit wall and the optic axis would then be
`35.26°. These structures are modeled by approximating the
`sloping edges by a series of rectangles/slabs whose horizon-
`tal dimensions change in a regular manner.
`FIG. 7 ShOWS a series of plots of transmission versus film
`thickness for vee geometry slits with 50. 100. and 200 nm
`slit widths. The transmission dependence on metal thickness
`is more complicated than the case with vertical walls. Also
`there is less decrease in transmission with increasing metal
`thickness than for the vertical wall case. To some extent the
`higher transmission is probably due to the thinner walls near
`the vee. so that this slit behaves like a vertical wall slit with
`a larger opening. There may be a collection horn effect as
`well. with the sloping walls allowing a better impedance
`match between free space and the interior of the slit.
`FIG. 8 shows a series of plots of transmission versus film
`thickness for inverted vee geometry slits with 50. 100. and
`200 nm slit widths. Like the vee geometry there is less
`decrease with increasing metal thickness than for the vertical
`wall case. However the variations in transmission with metal
`thickness are smaller than the vee case.
`FIG. 9 shows the signal-to-background ratio (s/b) for a slit
`with vertical walls scanning a line pattern which is assumed
`to be as long as the slit. and with uniform illumination
`extending over a distance of 0.5 pm normal to the slit edge.
`In this case. the signal is the light going through the slit
`while the background is the light tunneling through the
`metalfilm. An acceptable s/b value is one in which the
`background has no significant effect on the measurements
`even when the slit is set near the edge of the image where
`the illumination is reduced. Furthermore there may be
`several line images or patterns illuminating the metal film
`for a single slit. so the results in FIG. 9 may represent an
`underestimate of the background by as much as a factor of
`10. Therefore a reasonable value for s/b may be in excess of
`104. The required minimum thicknesses of the films are then
`about 60 to 70 nm for Mo and about 120 to 130 nm for Cr.
`The incident polarization can always be decomposed into
`components of Transverse Magnetic mode (TM—magnetic
`vector parallel to the sides of the slit) and Transverse Electric
`mode (TE-electric vector parallel to the sides of the slit).
`Theoretically. at least one TM mode can always propagate
`through even a very narrow slit.
`i.e..
`it doesn’t decay
`exponentially with the thickness of the slit material. This is
`an important advantage of the slit over a round aperture.
`Polarization effects are becoming important in lithography
`systems with high numerical aperture (NA) lenses, therefore
`an aerial image sensor which can measure polarization may
`be desirable. Furthermore. because the aerial image is mea-
`
`6
`sured in air rather than resist, the polarization efiects are
`larger since the NA is reduced in the resist by its index of
`refraction.
`
`FIG. 10 shows the fraction of TM polarization in the
`transmitted signal for the vertical side wall slit case. For
`smaller width slits the polarization can be 100 percent TM
`for film thicknesses consistent with the S/B requirements
`shown in FIG. 9. Note that if a non-polarizing sensor were
`desired. a slit width between 100 and 200 nm would be
`required.
`FIGS. 11 and 12 show the result of calculations for the vee
`slit geometry. FIG. 11 shows the degree of TM polarization
`versus film thickness while FIG. 12 shows the TM polar-
`ization versus slit width for a Mo slit with the vee slit
`
`geometry. Very high degrees of TM polarization appear
`possible for this geometry. It also appears possible to get a
`TM polarization of 0.5 (corresponding to no polarization)
`for a slit width of about 60 nm and a metal thickness of 120
`nm. The above analysis assumes normally incident waves.
`However, the results are little changed if light converging
`from a finite NA is used instead.
`FIG. 13 shows the modulation transfer function (MTF)
`for a vee geometry slit with a Mo film thickness of 120 nm
`and various slit widths. The MI‘F definition for a slit is
`completely analogous to the normal definition of the MTF
`for a lens system in optics. The analysis is based on a
`normally incident plane wave and two other plane waves
`which are incident at angles (i6). The amplitude (to within
`a phase factor) at the surface of the film is 1+cos(21rf1x)
`where stme/x is the transverse spatial frequency. The
`transmitted signal from this illuminated sinusoidal grating is
`calculated for values of 2117fo=0 and 2717fo=1:, at which the
`signal amplitudes are maximum (2) and minimum (0). This
`is equivalent to scanning the slit across the sinusoidal
`pattern. If the transmitted signal is 1(21thx) the contrast C
`can be defined as C=(I(0)—I(1r))/(I(0)+I(7r).
`The value of C depends on f1. For small values of fT. i.e.,
`very long transverse periods, the slit will follow the intensity
`variation very closely and the contrast will be ~1. However,
`as fT increases, the contrast will drop. The maximum value
`of fT is ll?» Where 9» is the wavelength of the light. The slit
`must be at least less than A. in order to pass all the spatial
`frequency components of the incident wave. For a coherent
`image produced by a lens with numerical aperture NA, the
`maximum spatial frequency is NA/k. In addition the contrast
`for the components must be high enough that noise doesn’t
`degrade the signal.
`Since the MI‘F will in general depend on slit width, film
`thickness, and film composition. there is a considerable
`range of properties to be explored. Shown for comparison in
`the figure are the MI‘F’s for an aberration free lens with an
`NA=0.6, illuminated by purely coherent or incoherent light.
`The slit MTF’s extend to higher spatial frequencies than in
`the coherent case, and exceed the modulation level of the
`incoherent case. A line/space pattern illuminated with par-
`tially coherent light will
`lie between the coherent and
`incoherent lines. Therefore it is likely the slit will pass all the
`spatial frequencies of significance in the aerial image. How-
`ever the higher frequency features will be reduced, because
`the slit MI'F is less than 1.0. Therefore a correction process
`may be needed. For purely coherent or incoherent illumi—
`nation the correction process can consist of Fourier trans-
`forming the scanned image, dividing the frequency spectrum
`by the slit MTF, and Fourier transforming back. For spatial
`frequencies where the projection lens MTF is small, this
`correction process does not have to be very accurate.
`FIG. 13 also shows that the MTF is different for TB and
`TM modes for this particular case. This is related to differ-
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`Nikon
`
`Exhibit 1008
`
`Page 26
`
`Nikon Exhibit 1008 Page 26
`
`

`

`5,631,731
`
`7
`ences in transmission of the two modes described above.
`FIG. 14 shows a situation where the TE and TM M'I‘F’s are
`almost the same. For this example. a vee slit with a 65 nm
`slit width in M0 with a metal thickness of 120 nm is used
`(see FIG. 12). As will be seen below, a 150 nm vertical wall
`slit will have the MTF’s approximately the same for TE and
`TM modes. and thus create no polarization effects.
`The fact that the slit behavior is sensitive to polarization
`for narrower slit widths suggests that there is a trade-off
`between resolution and polarization insensitivity. Polariza-
`tion sensitivity is no problem, however, if its effects can be
`measured. This can be done by measuring the aerial image
`sequentially with two slits with different polarization depen-
`dencies. The TE and TM images can then be determined
`These slits may not have the same MTF. but as long as it is
`possible to correct for the effects of an MTF which is less
`than 1. there should not be a problem
`First consider the case Where the polarization dependence
`is approximately independent of the spatial frequency com-
`position of the image. Then the image is scanned with slits
`1 and 2, the scanned intensities Il(x)EI1 and I2(x)EI2 can be
`written in terms of the TM and TE polarization components
`of the aerial image as
`
`11:6 171917151“E lmlm
`
`[fezrrlrr‘fezmlm
`
`where elm. elm. 62m, and 63”, are the polarization
`dependent transmissions for slits 1 and 2. The intensities for
`the two polarization states