Note: Descriptions are shown in the official language in which they were submitted.
2~3
-- 1 --
DEEP-W LITHO~RAPHY
This is a division of copending Canadian Patent
Application serial number 484,~58 which was ~iled on June
20, 1985.
Backgr_und__f_ he_ I_v _ti__
This invention relates to optical lithography
and, more particularly, to apparatus and methods for
achieving short-wavelength optical lithography adapted for
~abricating high-quality fine-line semiconductor devices.
It is known that the resolution limit (Lmin)
for equal lines and spaces in an optical imaging system
can be expressed as
L i ~ K~/NA (1)
where R is a constant whose value is typically between 0.
and 1.0 deperlding on processing and illumination
conditions and resist characteristics, ~ is the wavelength
of the exposing radiation and ~A is the numerical aperture
of the projection optics.
It is apparent Erom (1) that the minimum
printable feature can be reduced by decreasing ~ or by
increasing NA. But, since the depth of focus of the
system varies inversely as tNA)2, it is usually
preferable in a practical high-resolution system to
achieve the desired Lmin by reducing ~ rather than
;ncreasing ~A.
The present invention is directed to a novel
short-wavelength lithograph system.
Summary of the Invention
______ _________________
In accordance with an aspect of the invention
there is provided apparatus for optical lithography,
comprising stationary equipment including a laser source,
equipment including a stepping table physically separated
from said stationary equipment, and means for directing
signals from said laser source in said stationary equipment
to said second-mentioned equipment to maintain a prescribed
alignment between the direction of propagation of said
~d~
;2ZZ~
-- 2 --
signals and said second-mentioned equipment even during
movement of said second-mentioned eq~i~ment due to
stepping of said table.
In accordance with another aspect of the invention
there is provided a method of maintaining the output beam
of a laser aligned relative to movable equipment at which
the beam is directed, said method comprising the steps of
directing said beam at a photodetector array mounted on
said movable equipment, said array being adapted to
generate electrical signals representative of variations
of said beam relative to a prescribed alignment with
respect to said array, and applying said electrical
signals via a feedback loop to a deflection assembly that
maintains said beam aligned with respect to said array.
~ lithograph system according to the invention
incl~de~ a narrow-bandwidth tunable laser operating at a
wavelength in tlle deep- W range. A monochromatic all-
fused-silica lens assembly is utilized to direct the
output of the laser to successive portions of the surface
of a resist-coated wafer mounted on a movable support,
e.g., a known "stepping table".
The combination of a narrow-bandwidth laser and
a monochromatic lens assembly in accordance with the
invention makes it possible to quickly and easily
accomplish focus tracking in the system. This is done
by directing the output of a laser onto the surface of a
workpiece via a projection lens and controlling the
wavelength of the laser to maintain the focal length of
the lens equal to the lens-to-workpiece surface spacing.
Brief Descri_tion of the Drawin~
_ __________ __________________
The present invention taken in conjunction with
the invention disclosed in copending Canadian Patent
~pplication serial number 48~,658, which was filed on June
20, 1985 will be described in detail hereinbelow with the
~5;22'~8
aid of the accompanying drawings, in which:
FIGS. 1 and 2 taken together schematically depict
an apparatus in accordance with the invention for
achieving short-wavelength optical lithography; and
FIG. 3 shows in more detail a particular
implementation of a portion of FIG. 1.
Detailed Description
_______ _______ ___
The invention is described in connection ~Jith a
known type of "step-and-repeat" lithographic system.
In accordance with one feature of the invention,
a laser illumination source is physically separated from
the movable table portion of the system. Thus, for
example, the equipment 10 including laser 12 shown in FIG.
1 is advantageously located at a site removed from the
equipment 1~ including stepping table 16 shown in FIG. 2.
In this arrangement, the laser heam provided by the
equipment 10 i~c~ propagated through air or controlled space
to the equipment 14, as inciated by arrow 18 in FIG. 1.
(Actually, as will be clear later below, two coaxially
disposed laser beams are transmitted from the equipment 10
in the direction of the arrow 18. One beam constitutes
the exposing radiation. The other beam is utilized only
for alignment control purposes.)
There are several reasons for physically
separating the aforementioned equipments 10 and 14. In
some cases, for example, the laser 12 shown in FIG. 1
includes a toxic constituent which ~or safety
considerations should he located at a site remote from
operating personnel.
In operation, the stepping table moves successive
portions of a resist coated semiconductor wafer 40 into the
2~1
path of the illuminating radiation. In the past~ after
each indexing of the table, a delay was necessary to allow
damping out of vibrations caused by the table movements
before the illuminating beam was turned on. This is time
consuming.
In accordance with a feature of the invention,
instrumentalities are provided in the equipments 10 and 14
for instantaneously moving the laser beams in conformity
with the vibrations in the table 16 for maintaining the
l~ser beams t8 aligned relative to the table. Thus, after
each indexing of the table, the beam can be more quickly
turned on. Illustratively, these instrumentalities include
standard driven galvanometer mirrors 20 and 22 in the
equipment 10 and a conventional quadrant photodetector or
position-sensitive photodetector 24 mounted on the table
16. Electrical signals provided by the array 24 are
applied to differential amplifiers 26 and 28 in feedback
loops that respectively control galvanometer motors 30 and
32. The motor 30 is mechanically coupled to the mirror 20
via a Y-axis parallel shaft 34, and the motor 32 is coupled
to the mirror 22 via a Z-axis-parallel shaft 36. By
selective rotation of the mirrors 20 and 22, the
orientation of the laser beams 18 emanating from the
equipment 10 is varied in a controlled manner to compensate
for vibratory movement of the table 16.
Under quiescent conditions, the desired
orientation of the laser beans 18 relative to the
table 16 is established by steady-state signals applied to
the galvanometer motors 30 and 32 from a control
computer 38. The additional variable signals supplied by
the differential amplifiers 26 and 28 to the motors 30 and
32 are superimposed on the steady-state signals supplied by
the computer. This proces~s is clescribed furtl1er
hereinafter.
Each of the lenses includeci in the equipment 14
of llG. 2 is made only of fused silica. ~`used silica is a
highly sta~le material that is highly transr?arer1t to sl!ort-
~25;2~
wavelength light. Moreover, fused silica can be ~abricated
with good precision to form specified lens designsO
Despite these apparent advantages, applicant is the first
to have proposed the use of a single optical material
S (fused silica) to make a high-quality lens assembly for
short-wavelength (for example, deep-U~) optical lithography
based on laser illumination. Hereto~ore, it has been
customary to fabricate lenses utilizing multiple materials
to correct for chromatic aberrations.
Once he designed an all-fused-silica lens
assembly, applicant recognized that a laser source to be
combined with the assembly must as a practical matter have
an extremely narrow bandwidth if chromatic ~berrations in
the single-optical-m~terial assembly were to be avoided~
Since chromatic aberrations are unavoidable with a sinyle-
optical-material design, if the laser source bandwidth is
not suitably narrow, the projected image on the laser-beam-
illuminated wafer 40 (FIG. 2) would be unacceptably
blurred.
But applican~ found that all suitable short-
wavelength laser sources of adequate power were pulsed
lasers that inherently exhibit excessively broad
~andwidths. At that point, the obvious thing to have
done, as other workers in the art have, would have been to
redesign the lens assembly to be free of chromatic
aberrations with the available source bandwidths. But this
would have entailed employing optical materials other than
only fused silica. Instead, applicant embarked on the
unobvious course of retaining an all-fused-silica lens
design and redesigning the laser source to exhibit an
appropriately narrow bandwidth. This unique approach
allows the realization of a superior len.s aesi(3n and,
moreover, is t"e basis for achievin~ electronic focus
tracking as well as for achieving electronic tuning of the
laser source. Such tuning allows the source to be matched
to the operating characteristics o~ the lens assembly, as
will be speciLIed in detail later below.
~L~S;22~
Illustratively, the laser 12 included în the
equipment 10 of FIG. 1 comprises an excimer laser. This
category of lasers is capable of W emission at wavelengths
from, for example, below 4000 R to below
2000 A.
Excimer lasers and their applicatiOn to
lithography are described in a number of publications.
These include: "Laser Projection Printing" by
G. M. Dubroeucq et al, Proceedings of Microcircuit
Engineerin~ Conference, Aachen, Germany, September 1979,
pages 328-33-/; "Applications of Excimer Lasers in
Microelectronics" by T. McGrath, Solid State Techno ~ ,
Deccmber 1983, pages 165-169; "Deep UV Exposure of
Ag2Se/GeSe2 Utilizing an Excimer Laser" by
K~ ~. Polasko et al, IEEE Electron Device Letters,
Vol. EDL-5, ~o. 1, January 1984, pages 24-26; and "Excimer
Laser Projection Lithography" by K. Jain et al, Applied
Optics, Vol. 23, No. 5, March 1, 1984, pages 648-650.
By way of example, the laser 12 of FIG. 1
comprises a pulsed KrF gas excimer laser designed to
operate at a nominal center wavelength of 2484 ~O
(The fluorine constituent in KrF is highly toxic.)
Illustratively, the pulse repetition rate of the laser 12
is selected to be approximately 1000 pulses per second.
Inherently, the KrF excimer laser 12 (FIG. 1) has
a spectral bandwidth at the half-power point of
approximately l0 ~. But, recognizing that an all-
fused-silica lens assembly for hi~h-resolution lithography
requires a source bandwidth of less than about
0.1 ~ to be free of chromatic aberrations,
applicant combined a bandwidth-narrowing assembly with the
laser 12 to achieve an output at 2484 ~~
characteri~.ed by a half-power-point bandwidth of only
approximately 0.05 ~. At a repetition rate o~ 1~00
pulses per second, the power oE each such pulse is about
5 millijoules, which characteristics provide an adequate
basis ~or uniform high-resolution hiqh-throughput
3LZS2'~
lithography.
Various techniques are avai:Lable for narrowing
the inherent bandwidth of the laser 12 One suitable
assembly for doing this is shown in FIG. 3 (the figure also
showing portions 42 and 44 of the laser). Beam 46
emanating from the laser propagates through a standard low-
finesse etalon 48 and impinges upon a conventional grazinq-
incidence grating 50 which is spaced apart from a facing
high-reflectivity mirror 52. Illustratively, the
grating 50 has 3000-to-4000 grooves per millimeter. The
elements 48, 50 and 52 constitute both a tuning and a
bandwidth narrowing assembly. This assembly is shown in
FIG. 1 and identified therein by reference numeral 54
("bandwidth narrowing" is al50 reEerred to in tt~e art as
"line-narrowing").
Thle desired narrow-bandwidth output emanates from
assembly 54 and propagates in the direction indicated by
arrow 55 in FIG. 3. The arrow 55 is also shown in FIG. 1
wherein it is oriented parallel to the X axis.
The assembly 54 also provides a means for
establishing the center wavelength of the laser output 55
at a predetermined value and thereafter precisely
maintaining the wavelength at that value (or purposely
moving the wavelength off that value to accomplish
electronic focus trac~ing). This is accomplished by
rotating any one or combination of the elements 48, 50 and
52 about an axis perpendicular to the plane of the paper on
which FIG. 3 is drawn. Yor coarse tuning, rotating the
mirror 52 and/or the grating 50 is satisfactory. For fine
tuning, rotating only the etalon 48 is effective. In
practice, it is usually advantaqeous to initially establish
the predetermined center wavelength by rotating one or both
oE the elements 50 and 52. Thereafter, the laser can be
maintained at that wavelength or Eine-tuned therefroln by
selectively controlling the orientation of only the
etalon 48.
As schematically indicated in FIG. 3, a
.. ,, ., ,., ,. ", ,, .~ .
~52~2~3
micropositioner 56 is connected via a mechanical coupler 58
to the etalon 48. In response to signals applied to the
micropositioner 56 on line 60; the orientation of the
etalon is thereby controlled to maintain the wavelength of
the laser beam 55 at a 2redetermined value or to move the
wavelength off that value by a specified amount to
accomplish electronic focus tracking. The manner in which
the micropositioner 56 is so controlled will be described
in detail later below.
Various instrumentalities are ~nown in the art
for tuning and line-narrowing the output of a short-
wavelength laser as is done by assembly 54. In this
connection, see, for example: "Injection-Locked, Narrow-
Band KrF Discharge Laser Using an Unstable Resonator
Cavity" by J. Goldhar et al, ~tlcs Letters, Vol. 1 t No. 6,
December 197J, pp. 199-201; "Operating and Beam
Characteristics, Including Spectral Narrowing, of a TEA
Rare-Gas Hal:ide Excimer Laser" by T~ J. McKee et al, IEEE
Journal of Quantum Electronics, Vol. QE-15, No. 5,
May 1979, pp. 332-334; "Grazin~ Angle Tuner for CW Lasers"
by K. R. German, APplied Optics, Vol. 20, No. 18,
September 15, 1981, pp. 3168-3171; and "A Simple Tunable
KrF Laser System with Narrow Bandwidth and Diffraction-
Limited Divergence" by R. G. Caro et al, Journal Physics D:
Applied Physics, 15, 1982, pp~ 767-773.
The aforedescribed feedback loops for controlling
the galvanometer motors 30 and 32 (FIG. 1) require
continuous electrical input signals. But the pulsed
laser 12 is not capable of providing such signals via the
detector 24 tFIG. 2). Hence, a continuous-wave (CW)
laser 62 (for example, a standard helium-neon laser
operating at 6328 R) is also included in the
~equipment 10. The laser 62 is desi,3ned to provide a
continuous reference beam that is coaxial with the
a~orespeci~ied beam at 2484 A. In turn, the beam
at 6328 A is converted by the detector 24 into
continuolls electrical siqnals that are ap~lied to the
~s~
feedback loops that respectively control the galvanometer
motors 30 and 32.
Whenever the detector 24 senses that the
reference beam at 6328 R is off-center relative to
its prescribed alignment with the det,ector 24, correction
signals are applied to the motor 30 and/or to the motor 32
to re-establish the prescribed alignment. And, since the
beams at 6328 ~ and 2484 ~ are propagated
coaxially in the equipment 10, these correction signals are
effective to re-estabiish the prescribed alig~ment of the
exposing beam at 2484 ~.
The output of the CW laser 62 (FIG. 1) is
directed via a high-reflectivit~ mirror 64 to a dichroic
mirror 66. ~he mirror 66 is designed to reflect the
incident beam at 63~8 R to the right along an X-
direction path indicated by arrow 68. The mirror 66 is
also designed to transmit most of the incident beam at
2484 ~ emanating from the assembly 54 along the
same X-direction path. Thus, the arrow 68 indicates the
path along which the Cw beam at 6328 ~ and the
pulsed beam at 2484 ~ propagate coaxially to
successively impinge upon the galvanometer mirrors 20 and
22. After reflection from the mirrors 20 and 22, the
coaxial beams are propagated from the equipment 10. These
coaxial beams constitute the previously specified
beams 18.
A small portion (for example, about one percent)
of the beam at 2484 A emanating from the
assembly 54 of FIG. 1 is reflected upward by the mirror 66
along a Y-direction path indicated by arrow 69 to a
wavemeter 70. In response thereto, the wavemeter 70
provides an electrical signal to one input of a
differential amplifier 72. The other input to the
amplifier 72 is ~supplied by the control computer 38. In
3S that way, the output of the amplifier 72 applied to the
lead 60 is effective to maintain the output of the
assembly 54 at a predetermined center wavelenath specified
~5~2~3
-- 10 --
by the computer 38. Or, as will be described in more
detail later below, the signal applied to the amplifier 72
by the computer 38 can also be utilized to purposely change
the center wavelength to achieve electronic focus
tracking.
Further, the aforespecified ability of the
computer 38 (FI~. 1) to easily control or adjust the center
wavelength of the beam emanating from the assembly 54
simplifies the overall design and fabrication of the
e~uipment described herein. This is so because in practice
one can rarely make an all-fused-silica lens assembly
precisely to a specified prescription. Normally, one would
have to install the ~abricated lens assembly in the
equipment, test the assembly at a prespecified c:enter
frequenc~, and then ~ake the assembly apart to make
adjustments therein by further machining, polishing, etc~
Subsequently, re-installation of the lens
assembly and further testing in the equipment would be
done, and so forth in an interactive fashion until a near-
optimal match between the lens assembly and the centerwavelength was obtained. Such mechanical tuning or match-
ing of the lens assembly to a fixed prespecified center
wavelength is obviously time-consuming and expensive.
By contrast, .in applicant's unique design, it is
often possible, after initially installing the all-fused-
silica lens asseMbly in the herein-described equipment, to
achieve a near-ideal adjustment of the equipment withol~t
removing the lens assembly. This is done by leaving the
installed assembly intact and adjusting the center
wavelength of the exposinq beam, under control of the
computer 38, to obtain a near-optimal match between the
operating characteristics of the lens assembly as initially
fabricated and the operatinq wavelength of the equipment.
.Such electronic, rather than mechanical, adjustment of the
equipment is manifestly advantageous.
The laser beams at 2484 A an(i ~32~ A
~5~8
-- 11 --
that emanate from the equipment 10 in the direction of
arrow 18 are directed at the equipment: 14 shown in FIG. 2.
In particular, these beams are directed at a mirror 74 in
the equipment 14. The mirror 74 is designed to be highly
reflective at 6328 A and highly transmissive at
2484 A. As a result, most of the 6328 A
-beam is directed to the right in the clirection of arro-~ 76
and most of the 2484 R beam is directed downward in
the direction of arrow 78.
The 6328 A beam propagated in the
direction of the arrow 76 in FIG. 2 passes through a
filter 80. The filter 80 is designed to pass the
6328 ~ bea~ but to block any portion of the
2484 R beam that was reflected by the mirror 74 in
the direction o~ the arrow 76. Accordingly, only the CW
beam at 6328 ~ impinges upon the detector 24. In
turn, as described earlier above, the detector 24 provides
continuous electrical signals in the feedback loops shown
in FIG. 1 to control the operation of the galvanometer
motors 30 and 32. In that way, the orientation of the
6328 ~ beam 76 relative to the table 16, and thus
also the orientation of the coaxially disposed 2484
exposing beam 78, are maintained fixed even durinq
vibratory motion of the table 14.
The 2484 A beam propagated downward in
FIG. 2 in the direction of the arrow 78 is directed at an
adjustable field stop or aperture 82. Illustratively, the
diameter of the beam at the stop 82 is designed to be
larger than the diameter of the opening in the stop.
In practice, the equal-intensity contour lines of
the 2484 R beam transmitted through the stop 82 of
FIG. 2 are not svmmetrically disposed with respect to the
opening in the stop. Moreover, this asymmetry tends to
vary from pulse to pulse. If not compensated ~or~ these
factors can result in unsatisfactory illumination
uniformity and consequent poor linewidth control at the
surface o~ the wafer 40.
~Z5~
- 12 -
In accordance with a featur~e of the invention,
the slightly oversize beam directed at the stop 82 is
dithered or moved systematically by small amounts AX and
~Z. For an e~posure that cornprises, for example, several
hundred successive laser pulses, such movement is effective
to accomplish area avera~ing of the pulses transmitted
through the stop 82. In turn, this results in better
illumination uniformity at the surface of the wafer 40.
Movement of the 2484 ~ beam 78 directed
at the stop ~2 (YIG. 2~ to carry out area averaging is
controlled by the computer 38 (FIG. 1). Signals applied by
the computer 38 via leads 8O and 87 to the galvanometer
motors 30 and 32, respectively, are e~fective to implement
the aforementioned AZ and ~X movements of the beam across
the aperture in the stop 82.
The 2484 A beam 78 that passes through
the stop 82 impinges upon a mirror 88 shown in FIG. 2.
This n~irror is designed to reflect a relatively small
amount (for example about one percent) of the incident
beam. In turn, the reflected portion is directed through a
filter 90 to a photodiode 92. The filter 90 is designed to
pass light at 2484 ~ but to block li~ht at
6328 ~. In that way, any 6328 A component
in the bea~ transmitted by the stop 82 is prevented from
impinging on the photodiode 92.
The photodiode 92 of FIG. 2 constitutes part of a
dose control and laser trigger arrangement. The
photodiode 92 samples a portion of each 24~4 2
pulse and, in response thereto, generates an electrical
signal that is applied to a light integrator 94 (FIG. 1).
The control computer 38 s~pplies a second input signal (a
dose control signal) to the light integrator 94. In turn,
the output of the integrator 94 is applied to the laser 12
as a trigger signal there~or
The operation o~ the dose control and laser
trigger arrangement is as ~ollows. Under computer control,
~;Z5~
the stepping table 16 is moved by a micropositioner 95 to
bring a chip site on the resist-coated wafer 40 into
position for exposure to the pattern contained on the
reticle 84. (Illustratively, the reticle is assumed to
contain a single chip pattern thereon.) The computer 38
then activates the integrator ~4 to tricger the laser 12 to
start emitting pulses at 2484 ~. A portion of each
pulse is sampled by the photodiode 92 and a signal
representative~ thereof is applied to the integrator 94.
When the integrator 94 detects that the prescribed dose set
by the computer 38 has been attained, the laser 12 is
signaled to cease emitting pulses. Subse~uently, the
table 16 is moved to position ~nother chip site on the
wafer 40 in po~sition for the next exposure.
Exposure is accom~lished by projection of
illumination through a condenser lens 110, through a
reticle 84 containing the pattern to be imaged on the wafer
40, and through a projection lens 108. The provision of a
virtual source o illurnination for the lens 110 is now
described.
As indicated in FIG. 2, the 2484 ~ beam
78 that is transmitted through the mirror 88 is directed
toward a collimating lens 96. The lens 96 serves to focus
the beam to a small spot and to direct it at a mirror or
prism element 98. The small spot comprises a first virtual
source of illumination. ~he element 98 is designed to
deflect the beam into the edge of the field of a scanning
lens assembly 101 represented here by field lens 100 and
additional lens elements 102 and 104. In turn, the beam is
reflected by a scanning mirror assembly 106 back into the
scanninq lens assembly 101.
To achieve ade~1uate illumination of the entrar1ce
pupil of the projection lens 108 and hence pro~er ima~ing
characteristics on the wafer 40, it is advanta~eous to
illuminate about 50-to-75 percel1t of the diameter of the
entrance pupil of the lens 108. IllustL-atively, the
er1trance pu~il of the lens 108, as defined by drl apeL-tllre
~%5;;~Z8
- 14 -
stop 109 (shown schematically in FIG. 2), is approximately
tO0 millimeters in diameter. It is apparent, therefore,
that simply relaying the small first virtual source to the
lens 108 will not in practice provide adequate illumination
of the entrance pupil of the lens 108.
In accordance with one aspect of the invention,
the effective size of the virtual source that is actually
relayed to the projection lens 108 is substantially
increased in size over that of the first virtual source
produced by the lens 96. This is done by means of the
scanning lens assernbly 101 and the scanning mirror
assembly 106. By means of these assemblies, the operation
of which is described hereinafter, both the si~e and shape
oE the effective virtual source can be selectively varl~d
under control of signals ap~lied to the assembly 106 from
the computer 38. Significantly, in the co~Jrse of making
such variations, no laser light at 2484 g is wasted
in the depicted equipment. Hence, all available exposing
light is delivered to the surface of the wafer 40 even as
changes are made in the size and shape of the illumination
relayed to the entrance pupil of the projection lens 108.
The ability to change both the size and shape of
the illuminated portion of the entrance pupil of the
projection lens 108 can be significant in that it allows,
for example, tailoring the illumination to optimize
resolution of certain critical features on the wafer 40 and
to take maximum advantage of any nonlinearities in the
characteristics of the resist layer on the wafer 40.
The scanning mirror assembly 106 schematically
depicted in FIG. 2 comprises a mirror 112 and two
independently rotatable shafts 114 and 116. The shaft 114
- is oriented parallel to the X axis, whereas the shaft 116
is oriented perpendicular to the X axis. Rotation back and
forth of the shaft 116 causes the mirror 112 to rock back
and forth, as indicated by double-headed arrow 118. At the
same time that rocking occurs, the shaft 114 is turning, as
indicated ~y arrow 120. In other words, as the mirror
1;~5;~228
rocks in response to rotation of the shaft 116, the rocking
mirror is independently rotated by the shaft 114. As a
consequence, a relatively large area A in a Y-Z plane
immediately to the left of the lens 100 can be
substantially filled with the successive pulses supplied by
the laser 12 of FIG. 1 to form a large-area virtual source.
Illustratively, several hundred successive pulses are
supplied by the laser 12 during each interval in which a
chip site on the wafer 40 is being exposed.
The size of the aforespecified large-area virtual
source can be changed by varying the extent to which the
mirror 112 is rocked by the shaft 116. Additionally, the
shape of the virtual source can be changed by varyin~ the
speed o~ rot~tion of the shat 11~ while the mirror 112 is
being rocked
Liclht emanating from the relatively large-area
virtual source formed to the left of the lens 100 (~IG. 2)
is directed by the relay lens 110 to illuminate the pattern
contained on the reticle 84. In turn, light propagated
through the reticle 84 is imaged by the lens 109 onto a
chip site on the surface of the wafer 40. Illustratively,
the lens 109 forms a reduced (for example 5-to-1 reduced)
version of the reticle pattern on the wafer surface.
In a standard exposure system, conventional focus
tracking is accomplished by mechanically changing the
projection lens-to-wafer distance. Typically, this is done
either by moving the optical column of the system or by
moving the wafer. In either case, the adjustment is time-
consuming and, moreover~ may cause undesirable mechanica~
3n resonances in the system.
In accordance with one aspect of the present
invention, focus tracking is carried out quickly in a
nonmechanical manner. The ability to do so stems from the
use of lenses made from a single optical material. Such
lenses, unlike those corrected for chromatic aberrations,
exhibit an appro~imately linear relationship between
wavelenatl1 and tocal distance. I~ence, by electronically
- 16 -
changing the wavelength of the laser 12 (FIG. 1), the focal
plane of the projection lens tO8 (FIG. 2) is also changed.
The operation of the focus tracking arrangement
is as follows. First, a standard foc~ls sensor 111 detects
S whether or not the distance between the projection lens 108
and the surface of the wafer 40 has changed from a
prespecified value. Assume, for example, that that
distance has changed (decreased) by one micron, due~ for
example, to warpage in the wafer 40. A signal
representative of the change is then sent to the
computer 38 via lead 122. In response thereto, the
computer applies a corresponding correction signal to the
differential ampliier 72 (FIG. 1) included in the
aforedescribed frequency control loop. In turn, a signal
lS is applied by the amplifier 72 to the tuning and line-
narrowing asslembly 54 to increase the center wavelength of
the pulses emanating from the assembly 54. In one
illustrative case, the center wavelength was increased by
0.1 ~. This was sufficient to decrease the focal
distance of the lens 108 by one micron, thereby to
compensate exactly for the assumed one-micron decrease in
the lens-to-wafer spacing.
