U.S. patent application number 13/303327 was filed with the patent office on 2012-04-05 for system and method for laser processing at non-constant velocities.
This patent application is currently assigned to GSI GROUP CORPORATION. Invention is credited to Bo Gu, Shepard D. Johnson.
Application Number | 20120083049 13/303327 |
Document ID | / |
Family ID | 38805776 |
Filed Date | 2012-04-05 |
United States Patent
Application |
20120083049 |
Kind Code |
A1 |
Johnson; Shepard D. ; et
al. |
April 5, 2012 |
SYSTEM AND METHOD FOR LASER PROCESSING AT NON-CONSTANT
VELOCITIES
Abstract
A method is disclosed for on-the-fly processing at least one
structure of a group of structures with a pulsed laser output, The
method includes the steps of relatively positioning the group of
structures and the pulsed laser output axis with non-constant
velocity, and applying the pulsed laser output to the at least one
structure of the group of structures during the step of relatively
positioning the group of structures and the pulsed laser output
axis with non-constant velocity.
Inventors: |
Johnson; Shepard D.;
(Andover, MA) ; Gu; Bo; (North Andover,
MA) |
Assignee: |
GSI GROUP CORPORATION
Bedford
MA
|
Family ID: |
38805776 |
Appl. No.: |
13/303327 |
Filed: |
November 23, 2011 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11532160 |
Sep 15, 2006 |
8084706 |
|
|
13303327 |
|
|
|
|
60832082 |
Jul 20, 2006 |
|
|
|
Current U.S.
Class: |
438/4 ;
219/121.62; 257/E21.002 |
Current CPC
Class: |
H01L 2924/0002 20130101;
B23K 26/0853 20130101; B23K 26/0624 20151001; H01L 2924/0002
20130101; H01L 23/5258 20130101; H01L 2924/00 20130101; B23K
2101/40 20180801 |
Class at
Publication: |
438/4 ;
219/121.62; 257/E21.002 |
International
Class: |
H01L 21/02 20060101
H01L021/02; B23K 26/00 20060101 B23K026/00 |
Claims
1.-32. (canceled)
33. A method of on-the-fly laser processing first and second
structures of respective first and second groups of structures on a
moving stage with a pulsed laser output along a pulsed laser output
axis, each structure having a respective optimal processing time,
said method comprising the steps of: relatively positioning the
first and second structures and the pulsed laser output axis with
respective first and second velocities, wherein the first and
second velocities are different velocities; and applying the pulsed
laser output to the first and second structures during the step of
relatively positioning the first and second structures and the
pulsed laser output axis with the respective first and second
velocities by selecting one or more laser pulses from a laser that
is pulsed at a constant repetition rate that is not synchronized
with relatively positioning at least one of the first or second
structures and the pulsed laser output axis, such that laser
processing of said at least one structure is performed during at
least one time interval that comprises the at least one respective
optimal processing time and a pulse-to-pulse time between
sequential constant repetition rate laser pulses, wherein a
distance that is characterized by a product of the at least one
time interval and a relative positioning velocity at the optimal
processing time is within an acceptable position enor between the
pulsed laser output axis and the at least one structure of the
first and second structures.
34. The method as claimed in claim 33, wherein the pulsed laser
output is provided by selecting pulses of a mode locked laser
during the at least one time interval, each time interval
corresponding to a position of said at least one structure of the
first and second structures.
35. A method of laser processing selected structures in an array of
structures carried on a stage comprising: providing a pulsed laser
output by pulsing a laser with a substantially constant pulse to
pulse period. moving the structures at a velocity relative to the
pulsed laser output, wherein the distance between adjacent
structures divided by the velocity comprises a period that is not
synchronized with the pulse to pulse period; selecting pulses from
the pulsed laser output corresponding to respective selected
structures, and applying the selected pulses to the selected
structures with acceptable position errors between the selected
pulses and respective selected structures.
36. The method as claimed in claim 35, wherein the pulse to pulse
period times the velocity comprises a distance less than an
acceptable laser pulse to structure positioning error
37. The method as claimed in claim 35, wherein selecting comprises
down counting from a higher pulse repetition rate.
38. The method as claimed in claim 35, wherein the pulsed laser
output comprises an amplified seed laser output.
39. The method as claimed in claim 35, wherein moving the
structures comprises moving according to an acceleration profile,
the acceleration profile comprising at least first and second
velocities.
40. The method as claimed in claim 39, wherein one of the first and
second velocities is a reduced velocity near the end of travel of a
stage.
41. The method as claimed in claim 35, wherein the velocity is
selected from a continuous range of velocities.
42. The method as claimed in claim 35, wherein applying further
comprises deflecting the laser output axis with a beam
deflector.
43. The method as claimed in claim 35, further comprising planning
a trajectory with at least one commanded stage velocity set within
a continuous range of velocities.
44. The method as claimed in claim 35, wherein selecting pulses
comprises selecting a group of pulses applied to one structure,
wherein for each pulse a distance that is characterized by a
product of the at least one time interval and a relative
positioning velocity at the optimal processing time is within an
acceptable position error between the pulsed laser output axis and
the at least one structure of the group of structures.
45. A laser-based semiconductor processing system comprising: a
laser configured to pulse at a substantially constant rate, the
rate characterized by a pulse to pulse period to provide a pulsed
laser output, means for positioning structures relative to the
pulsed laser output in a plane at velocities within a continuous
range of velocities, means for selecting one or more pulses from
the pulsed laser output to irradiate respective structures, and a
system controller configured to receive target structure
coordinates, to initiate movement of the structures at one or more
velocities in a continuous range of velocities along a trajectory
for processing target structures, and to send pulse selecting
signals to the pulse selecting means when positional error between
each selected pulse of the pulsed laser output and a target
structure is below a predetermined error limit, wherein the laser
pulse to pulse period is unsynchronized with respect to the at
least one velocity and the positions of the moving structures.
46. The system as claimed in claim 45, wherein means for
positioning farther comprises beam deflecting means for combined
relative positioning.
47. The system as claimed in claim 45, where the structures are
uniformly spaced apart structures.
48. The system as claimed in claim 45, wherein the laser comprises
a mode-locked laser.
49. The system as claimed in claim 45, wherein the velocities
comprise non-continuous velocities.
50. The system as claimed in claim 45, wherein the continuous range
of velocities is 50-200 mm/s.
51. The system as claimed in claim 45, wherein means for
positioning further comprises a multi-spot beam positioning
system.
52. The system as claimed in claim 45, wherein means for
positioning further comprises a multi-path beam positioning system.
Description
PRIORITY
[0001] The present application claims priority to U.S. Provisional
Patent Application Ser. No. 60/832,082 filed Jul. 20, 2006.
BACKGROUND
[0002] The present invention relates generally to laser processing
of integrated circuits on semiconductor wafers, and relates in
particular to severing conductive links (e.g., link blowing) on
memory integrated circuits. For example, "LIA Handbook of Laser
Materials Processing," by J. F. Ready, Laser Institute of America,
2001, discloses general information regarding link blowing at
Chapter 19.
[0003] When manufactured, memory die typically include some number
of defective memory cells due to limitations in semiconductor
fabrication technologies. To make memory die with defective memory
cells useable, memory die are typically manufactured containing
extra memory cells that may be used in place of defective cells.
The defective memory cells must then be isolated. Integrated
circuit memory repair systems employ a focused laser beam to open
(or blast) fusible links on integrated circuit memory die in order
to provide that only properly functioning memory cells are coupled
to the circuit memory.
[0004] The processing speed of conventional memory repair systems,
however, may be limited by the pulse repetition rate of the laser
system. As the demands of laser processing systems increase, there
is a need for memory repair systems to be faster and more
efficient. Efforts to change the pulse rate however, for example by
increasing the Q switch rate, result in changes in pulse shape and
energy, which may adversely affect memory system repair.
[0005] There is a need, therefore, for faster and more efficient
memory repair systems.
SUMMARY
[0006] The invention provides a method for on-the-fly processing at
least one structure of a group of structures with a pulsed laser
output. In accordance with an embodiment, the method includes the
steps of relatively positioning the group of structures and the
pulsed laser output axis with non-constant velocity, and applying
the pulsed laser output to the at least one structure during the
step of relatively positioning the group of structures and the
pulsed laser output axis with non-constant velocity. In accordance
with an embodiment, a mode locked laser is used at pulse rate
frequencies from about 10 MHz to about 200 MHz, and in accordance
with further embodiments, each link to be blasted may receive a
plurality of pulses. In accordance with an embodiment of the
invention therefore, it is no longer required to synchronize the
substrate/stage velocity with the laser pulse rate to provide that
a link to be blasted is positioned at the location of the focused
laser beam at the time of a laser pulse.
BRIEF DESCRIPTION OF THE ILLUSTRATED EMBODIMENTS
[0007] The following description may be further understood with
reference to the accompanying drawings in which:
[0008] FIG. 1 shows an illustrative diagrammatic view of a wafer
that may be processed in accordance with an embodiment of the
invention that includes a plurality of circuits, each of which has
groups of memory links;
[0009] FIG. 2 shows an illustrative diagrammatic enlarged view of
selected circuits of FIG. 1 showing groups of memory links having
identified links that require blasting in accordance with an
embodiment of the invention;
[0010] FIGS. 3A, 3B and 3C show illustrative diagrammatic views of
a) a group of links, b) a linear ramped acceleration across the
group of links of FIG. 3A with a central peak, and c) a linear
ramped acceleration across the group of links of FIG. 3A with an
extended constant acceleration peak in accordance with an
embodiment of the invention;
[0011] FIGS. 4A, 4B and 4C show illustrative diagrammatic views of
a) a group of links, b) a non-linear ramped acceleration across the
group of links of FIG. 4A with a central peak, and c) a non-linear
ramped acceleration across the group of links of FIG. 4A with an
extended constant acceleration peak in accordance with an
embodiment of the invention;
[0012] FIG. 5 shows an illustrative diagrammatic view of a heat
affected zone of a link being blasted with a single pulse in
accordance with an embodiment of the invention;
[0013] FIG. 6 shows an illustrative diagrammatic view of a
plurality of heat affected zones of a link being blasted with
multiple pulses in accordance with an embodiment of the
invention;
[0014] FIG. 7 shows an illustrative diagrammatic view of a laser
processing system in accordance with an embodiment of the
invention; and
[0015] FIGS. 8A-8C show illustrative graphical and diagrammatic
views of link blasting time charts in a constant velocity system
and a non-constant velocity system in accordance with an embodiment
of the present invention.
[0016] The drawings are shown for illustrative purposes only.
DETAILED DESCRIPTION
[0017] The memory cells on a memory die are typically arranged in a
matrix of rows and columns of memory cells. Extra memory cells are
included on the memory die by increasing the number of rows and/or
columns of the memory matrix to include excess rows and columns of
memory cells. Defective memory cells in the memory matrix are
avoided (not used) by modifying memory matrix addressing to provide
matrix rows and columns that are defect free. Fusible links are
used to modify memory matrix addressing, and a laser is used to
open (or blast) the selected fusible links in a laser memory repair
system. Memory die are processed therefore to select only defect
free memory cells before the wafer is diced. Typically memory
wafers are 200 mm or 300 mm in diameter.
[0018] The fusible links on memory die are typically arranged in
groups of links where each group consists of a row or column of
links. Within each row or column the links are generally spaced at
equal increments. Link size and spacing vary significantly
dependent on the manufacturer and the memory design. Link
dimensions for a typical memory design may be for example, 0.4
.mu.m wide and 4 .mu.m long, with 3 .mu.m space between links.
[0019] Further designs may include at least some links having
widths of about 0.1-0.2 .mu.m, spaced apart at about 1-1.5 um. Link
processing at the finer scale using a 50 MHz mode locked laser as
disclosed, for example, in "Laser Processing of Ultra Fine Pitch
Fuse Structures in 65 .mu.m Node Technology", Society for Equipment
and Materials International, SEMICON West 2004 by Joohan Lee, James
Cordingley and Joseph J. Griffiths shows in FIG. 6 thereof a burst
of (amplified) mode locked pulses, at least a portion which was
applied to a link has also been achieved in accordance with the
invention.
[0020] The memory repair system is provided a map of link locations
on the memory die and a file listing the links on each die on the
wafer that need to be opened (blasted). Typically relatively few
links on a circuit are blasted and the locations of the links
blasted on each circuit are typically different. The memory repair
system opens the links using a pulsed laser beam focused to a
desired beam waist.
[0021] The links to be blasted are positioned in the XY plane
relative to the focused laser beam axis. For example, the wafer may
be carried on a precision XY stage during processing. Other
relative positioning devices and systems can be used and combined
to structures for processing. These may include split stage
positioning systems, multi-rate positioning systems, galvanometer
scanners, acousto-optic deflectors, fast steering mirrors,
electro-optic deflectors and piezo driven positioners. It is to be
understood that although for convenience positioning may be
referred in terms of stage motion, the scope of the invention
includes other positioning schemes. Laser focus is maintained
during link blasting by adjusting laser focus position (beam waist)
in the Z axis (vertical axis).
[0022] Prior to link blasting, the position and orientation of the
die sites carried on the wafer must be precisely registered with
the laser focus position. Alignment may include camera imaging and
laser scanning of alignment targets or of other target features.
Typically, laser edge scanning is used for fine alignment and the
scanning overhead contributes to the overall link processing
time.
[0023] When blasting a group of links a laser is fired (pulsed) at
approximately a constant repetition rate. Firing the laser at a
constant repetition rate helps maintain a precise and constant
amount of energy in each laser pulse thereby providing consistent
laser energies to each link blasted. The constant rate may result
from a laser trigger signal, or may be an inherent property of the
laser cavity.
[0024] Conventionally, the stage is moved at a constant velocity
during link blasting in order to position successive equally spaced
links of a link group at the location of the focused laser beam at
the time of the next laser pulse such that each laser pulse
corresponds to a single link in a group of links. The constant
velocity move used during link blasting is referred to as a CV
(constant velocity) move. In accordance with an embodiment of the
invention however, it is no longer required to synchronize the
substrate/stage velocity with the laser pulse rate to provide that
a link to be blasted is positioned at the location of the focused
laser beam at the time of a laser pulse.
[0025] Not all links in a group are typically blasted, therefore
not all fired laser pulses are used to blast links. A pulse
selector (typically an acoustic optic modulator) is used to route
pulses either through the focusing lens to the link if the link is
to be blasted or to a beam dump if the link is not to be left
blasted. An acoustic optic modulator is typically also used to
reduce the laser pulse energy to the desired energy for blasting a
link.
[0026] Various motion profiles, aspects of path planning, and
trajectory generation are disclosed, for example in U.S. Pat. Nos.
6,144,118; 6,483,071 and 6,662,063, the disclosures of which are
hereby incorporated by reference. The teachings of these references
may be applied to trajectory planning in systems in accordance with
various embodiments of the invention utilizing the split stage
positioning systems, multi-rate positioning systems, fast steering
mirrors, sold state deflectors, and piezo-electric driven
positioners.
[0027] After processing a group of links the stage is then moved to
the next link group to be blasted. The stage move trajectory for
moves between link groups is computed to position the stage at the
beginning of the next group with the appropriate velocity for
blasting the next group. These non-constant velocity moves between
link groups are referred to as PVT (position velocity time) moves.
Conventionally, the end point requirements for the move are a
position and a velocity at a specified time. The specified time is
required in order to coordinate the stage X direction move with the
stage Y direction move so that at the end of the move both axis
meet the end point requirements at the same time.
[0028] A system controller coordinates all activities during
processing. These activities include laser firing, stage/substrate
motion, and pulse selection. Typically stage motions are commanded
to provide move segments such that the links to be blasted are
positioned at the location of the focused laser beam and then laser
firing is synchronized to the stage motion. With conventional
lasers, laser firing is controlled by a laser trigger signal sent
from the system controller to the laser. When the laser receives a
trigger signal a laser pulse is generated. The laser pulse
generated occurs a small delay time after the trigger signal active
edge. The delay time typically varies slightly for each pulse
resulting in a small jitter in the laser firing time. The laser
typically generates a pulse at the time of the trigger signal by
either changing the state of a Q switch or by pulsing a seed
laser.
[0029] The laser repetition rate selected is determined by the
capabilities of the laser and the characteristics of the memory
repair system. Typically, higher laser repetition rates result in
greater system throughput. The laser is typically operated at a
single repetition rate for all link groups on a wafer. Operating at
a single repetition rate simplifies laser pulse energy control as
pulse energy control only needs to be calibrated for the single
repetition rate.
[0030] Laser pulse energy must be carefully controlled and pulse
energies should be matched pulse to pulse for example, to better
than 2% RMS. Firing the laser at a constant repetition helps to
achieve this matching of pulse energies. This need to carefully
control and match laser pulse energies, pulse to pulse, has been a
primary reason for moving the stage at a constant velocity when
blasting links.
[0031] In accordance with an embodiment of the invention, a very
high repetition rate laser, typically in the MHz, is used such as
may be provided by a mode locked solid state laser. The laser is
not triggered to generate pulses at desired times; rather the laser
is free running generating pulses that are not synchronized to
stage/link positions. The velocity of the substrate/stage (and
therefore the motion of the links) is not synchronized to the laser
repetition rate. A predetermined pulse time interval is used to
select a laser pulse, or plurality of pulses, from the free running
laser to blast a link. This pulse time interval occurs
substantially at the optimal time for positioning at the desired
link. This interval starts before a pulse is selected, and ends
after the last pulse is selected. No pulses are selected outside of
the pulse time interval. Note that the laser is not triggered to
generate a pulse at a specific time but a laser pulse, or plurality
of pulses, is selected that occur at a time close to the optimal
laser pulse time. Because the laser repetition rate is high as
compared to the velocity of the substrate/stage, the error in link
position at the time of the laser pulse is small and
acceptable.
[0032] The difference in time of the actual laser pulse to the
optimal time for the laser pulse is referred to as pulse time
jitter. The maximum amount of pulse time jitter is approximately
equal to the time between laser pulses of the free running laser.
Because the substrate/stage is moving at a velocity, this pulse
time jitter corresponds to position error in blasting. If the
motions of the stage are acceptably small during the pulse jitter
time then the resulting position errors are acceptable. For
example, if the substrate/stage is moving at 150 mm/s and the laser
free running repetition rate is 50 MHz, then the time between laser
pulses is 1/(50 MHz)=20 ns. The substrate/stage motion during 2Ons
when moving at 150 mm/s is (20 ns)*(150 mm/s)=3 nm. If a link is
0.4 .mu.m wide then 3 nm corresponds to (3 nm/0.4
.mu.m)*(100)=0.75% of the width of the link, which in some cases
may be an acceptable motion/position error during link
blasting.
[0033] As shown in FIG. 1, a semiconductor wafer 10 may include on
a surface 12 thereof a plurality of circuits 14, each of which
includes a plurality of fusible link groups 16. Each group 16
includes one or more links, and each group 16 of each circuit 14
may include one or more links 18 that have been determined to need
to be blasted or opened.
[0034] During conventional link blasting, small timing corrections
(phase corrections) are made in laser firing time to correct for
small stage positioning errors. In accordance with the invention,
substrate/stage position errors from the commanded position are
compensated for by changing the start of the laser pulse time
interval. Laser pulse time however, is not changed by changing the
time a laser pulse is generated. Rather, laser pulse time is
changed by selecting a different pulse from the free running laser
within the laser pulse time interval.
[0035] Because it is not required to use substrate/stage velocity
to synchronize the laser pulse rate with the stage/substrate
motion, it is not required that the substrate/stage move at a
substantially constant velocity during blasting. This provides
significant advantages in the movement of the stage and efficiency
of the system. In accordance with various embodiments of this
invention, the substrate/stage may be accelerating, decelerating,
or moving at a substantially constant velocity during blasting.
This allows for accelerating and decelerating during blasting of a
single link group. As shown in FIG. 2 for example, the substrate or
stage may be accelerating during processing of a single group, and
then decelerate near the end of the group and either reverse
direction along one axis (e.g., the X direction) as shown at A, or
may change direction from one axis to another (e.g., from Y to X)
as shown at B in FIG. 2. In either case, processing speed is
increased by permitting the substrate or stage to be moved at a
non-constant velocity during processing of a single group prior to
the change in direction as indicated at A or B. Moreover, the
ability to process at non-constant velocity also permits groups of
circuits that are at the edge of the processing field (stage travel
limits) to be processed more slowly at the end of the group near
the edge, and then more quickly at the center and other end of the
group.
[0036] As shown for example in FIG. 3A, a group 30 of memory links
32 may be provided along a single linear direction. As shown in
FIG. 3B, the acceleration of the substrate or stage may be
increasing linearly until the mid-point of the travel along the
group 30, and then decreasing linearly as shown at 34. As shown in
FIG. 3C, the acceleration of the substrate or stage may be
increasing linearly at a rapid linear rate until it reaches a
plateau. The acceleration may then remain constant for an extended
period time, and then decrease linearly as the substrate or stage
reaches the end of the group shown at 36. More generally, the
substrate or stage may be accelerating or decelerating during the
blasting of a link group.
[0037] As shown in FIG. 4A, a group 40 of memory links 42 may be
provided along a single linear direction, and as shown in FIG. 4B,
the acceleration of the substrate or stage may be increasing
non-linearly until the mid-point of the travel along the group 40,
and then decreasing non-linearly as shown at 44. As shown in FIG.
4C, the acceleration of the substrate or stage may be increasing
non-linearly at a rapid rate until it reaches a plateau. The
acceleration may then remain relatively constant for an extended
period time, and then decrease non-linearly as the substrate or
stage reaches the end of the group shown at 46. Again, more
generally, the substrate or stage may be accelerating or
decelerating during the blasting of a link group.
[0038] Acceleration may be limited by thermal loading, power,
mechanical or other system parameters. For example, high
acceleration may produce high motor forces that may introduce
unacceptable mechanical vibrations into the system and affect
positioning accuracy. Acceleration may be reduced in combination
with accelerating during blasting to improve system performance.
For example, reduced acceleration may reduce mechanical vibration
or other adverse conditions that generally limit usable
acceleration and thereby increase system accuracy while maintaining
or increasing system throughput. Reduced acceleration may be part
of a structured acceleration profile and may be used in conjunction
with increased periods of acceleration and with velocities that
exceed conventional pulse rate limited synchronization speeds.
[0039] The laser pulse rate may be unsynchronized relative to the
stage/substrate motion, and so the commanded stage velocity can be
set within a continuous range of velocities. This continuous range
of velocities is in contradistinction to conventional systems,
where the commanded stage velocity is limited to discrete sets of
velocities that are synchronized during blasting based on the laser
pulse rate and the regularly spaced link pitch such that the
substrate/stage is at a desired location at the time of a laser
pulse. This available continuous range of velocities provides added
degrees of freedom in trajectory planning that may be used to
optimize system throughput.
[0040] Two directions that are generally referred to in connection
with stage/substrate trajectories during link blasting are on-axis
and cross-axis. The term on-axis refers to the direction along a
particular group of links; and this is the direction of
stage/substrate motion when blasting the particular link group. The
term cross-axis refers to the direction in the XY plane orthogonal
to the on-axis direction, which is the direction across the
particular link group. The commanded velocity in the cross-axis
direction when blasting link groups is zero.
[0041] Systems and methods of various embodiments of the present
invention may permit increased system throughput during the
blasting of a single link group, as well as processing of each
circuit. For a single link group the velocity at the beginning of
the link group may be less than the maximum blast velocity, a
velocity that allows for a quick transition from a previous link
group to the present link group. Then during the blasting of the
link group the velocity may be increased (acceleration) to reduce
the blast time and then decreased (deceleration) to bring the
velocity down to a velocity that allows for a quick transition from
the current link group to the next link group. Examples of such
transitions are shown at A and B in FIG. 2.
[0042] Systems and methods of certain embodiments of the invention
also permit increased throughput by reducing the stage/substrate
travel time between link groups that are to be blasted in
accordance with certain embodiments. This reduction in time is
possible because of a reduction in requirements on the generated
trajectory between two link groups. In conventional systems, there
are requirements on the end conditions for the trajectory between
two link groups and there are constraints on the actual trajectory
generated for the trajectory between two link groups. In accordance
with various embodiments of this invention, the number of these
requirements (constraints) is reduced.
[0043] In particular, in typical conventional systems, there are
nine requirements on the end conditions of a trajectory between two
link groups. These requirements are the on-axis and cross-axis
positions and velocities at the beginning and end of the trajectory
(total of 8 requirements) and a time requirement that the end point
conditions for both the on-axis and cross-axis move occur at the
same time.
[0044] For the methods of certain embodiments of this invention,
there are only seven requirements on the end conditions of a
trajectory between two link groups. These requirements are the
cross-axis positions and velocities at the beginning and end of the
trajectory (total of 4 requirements), the on-axis positions at the
beginning and end of the trajectory (total of 2 requirements) and a
time requirement that the end point conditions for both the on-axis
and cross-axis more occur at the same time. There are no longer
fixed requirements on the on-axis velocities at the beginning and
end of the trajectory, rather there are constraints only on the
range of values for on-axis velocities at the beginning and end of
the trajectory. Effectively, for the on-axis move between two blast
groups, the on-axis trajectory may begin before the blasting
trajectory is complete and end after blasting has begun for the
next link group. There are of course also a number of constraints
for the move between two groups of links. These constraints include
maximum/minimum velocity, maximum/minimum acceleration, stage
travel range limits, and constraints on profile shape.
[0045] In general, a throughput increase may result for moves
between blast groups using the methods of this invention when the
time required to position the substrate to the correct off-axis
position at zero velocity is less than the time required to
position the substrate to the correct on-axis position and desired
on-axis blast velocity. For these cases it is often possible to
modify the velocities at the end of the previous link group and/or
at the beginning of the next link group such that it is possible to
make the delay time between link groups equal to the time required
for the off-axis move or at least less that the move time of
conventional systems.
[0046] Increased throughput may also be provided in certain
embodiments when blasting near the stage limits of travel. When
blasting a long link group where the on-axis link group end is
close to the limits of travel of the stage, it is not typically
possible to blast the end of the link group at maximum on-axis
velocity using conventional systems and methods. It is not possible
because there is not enough distance after the end of the link
group before the stage end of travel limit to reduce the stage
velocity in the on-axis direction to zero. In order to blast the
link group the blast velocity of the entire link group is reduced
to a velocity that may be brought to zero velocity in the distance
after the end of the link group before the stage end of travel
limit. Alternatively, conventional systems may divide a long
trajectory near the edge of stage movement of one long link group
into two link groups: a long trajectory that is blasted at maximum
velocity for the majority of the link group and a short trajectory
near the stage travel limits that is blasted at a reduced
velocity.
[0047] Using the methods of various embodiments of this invention,
the entire link group does not have to be processed at a reduced
velocity or broken into two link groups. The blast velocity of the
link group may be reduced as the stage travel limit is approached
resulting in an overall throughput improvement.
[0048] Increased throughput may also be provided in accordance with
various embodiments of the invention by blasting at optimal
velocity. In certain conventional systems, only a limited number of
synchronized velocities may be used to blast links. The velocities
available may not be optimal, for example, if the maximum stage
blast velocity is 200 mm/s and the maximum blast velocity available
to blast a particular link group is 150 mm/s. In this case, a
sub-optimal velocity must be used for long blast groups. Using the
methods of certain embodiments of this invention, the link group
may be blasted at the maximum velocity resulting in increased
throughput.
[0049] A link blasting system, for example, may have the following
constraints on stage motion: Acceleration must be in the range -15
m/s.sup.2 to +15 m/s.sup.2; Blast velocity must be in the range
-0.2 m/s to +0.2 m/s, with typical speeds of 50-200 mm/sec.;
Acceleration pulse shapes are raised cosine shaped; and Minimum
width acceleration pulse is 0.5 ms. The raised cosine acceleration
pulse shape is generally shown in FIG. 4B and described by the
following equation
a ( l ) = 1 2 A max ( 1 - cos ( 2 .pi. t Tp ) ) ##EQU00001##
[0050] The acceleration pulse end point equations for raised cosine
acceleration pulses are not unique to raised cosine acceleration
pulses profiles. The same acceleration pulse end point equations
may result from a number of other acceleration pulse shapes. Two
other acceleration pulse profile shapes that result in the same
acceleration pulse end point equations are shown in FIGS. 3B and
3C.
[0051] A laser processing system for use in accordance with an
embodiment of the invention may include a mode locked laser
operating above 20 MHz. The laser may generate pulses at a fixed
pulse repetition rate with a fixed phase. As shown at 50 in FIG. 5,
each link to be blown 52 may receive one pulse having a heat
affected zone 54. Selecting pulses from a free running pulse train
results in increased uncertainty in the actual time that the pulse
will be delivered to the target material, which increases pulse
time jitter. This increase in pulse time jitter translates into a
small increase in positional error when blasting a link if the
stage is moving. The corresponding increased pulse time jitter is
approximately equal to the time between laser pulses of the free
running laser. For example, if a substrate supporting the target
material and positioning stage are moving at 150 mm/s and the free
running repetition rate is 50 MHz, then the time between laser
pulses is 1/50 MHz=20 ns. The substrate/stage motion during 20 ns
when moving at 150 mm/s is 3 nm. If a link is 0.4 .mu.m wide, then
3 nm corresponds to 0.75% of the width of the link, an acceptable
positional error during link processing. In accordance with further
embodiments, and as shown at 60 in FIG. 6, each link to be blown 62
may receive a plurality of pulses providing a plurality of heat
affected zones 64 across the link 62.
[0052] In accordance with various embodiments, the laser processing
system may be as disclosed, for example, in U.S. Published Patent
Application No. 2006-0191884 filed Jan. 18, 2006, the disclosure of
which is hereby incorporated by reference. For example and with
reference to FIG. 7, a system may include a mode locked laser
system 70 (e.g., operating at 25 MHz), a pulse picker 72, an
optional optical amplifier 74, relay optics 76, a pulse selection
unit 78 (e.g., an acousto-optic modulator), a zero order beam stop
80, relay optics and a beam expander 82 a fold mirror 84, zoom
optics 86 and an objective lens 88. The laser pulses are focused at
the focal plane onto a substrate 90 having links that is supported
by an XY stage 92. The position of the stage 92 is controlled by a
controller 94, and the controller 94 also controls the selection of
pulses by the pulse picker 72 and/or the pulse selection unit 78.
The selection of pulses may be done either before or after the
optical amplifier, and either or both of the pulse picker 72 and
pulse selection unit 78 may be used. The system may also include an
optional wavelength shifter to change the frequency of the
illumination from, for example, 1064 .mu.m to 532 .mu.m.
[0053] The throughput advantages of systems and methods of the
invention may be shown by the following examples. In a first
example, a link group 30 mm in length is processed where links are
3 .mu.m apart. As a comparison, the links are first processed at a
constant velocity with a laser repetition rate of 50 kHz. The
substrate/stage velocity during blasting is (3 .mu.m)(50 kHz)=150
mm/s. The time to blast the link group is (30 mm)/(150 mm/s)=200
ms. In accordance with an embodiment of the invention, however, it
is possible to begin blasting the link group at a velocity of 150
mm/s, to accelerate up to 200 mm/s during the blasting, and then to
decelerate back down to 150mm/s such that the end velocity is 150
mm/s. This trajectory requires a total time of 151.66 ms resulting
in a time savings of 48.333 ms or about 24.2%.
[0054] In another example, two parallel link groups 9 mm in length
are separated by 0.1875 mm in the cross-axis and where the on-axis
start and end coordinates are the same. The links are blasted in
opposite on-axis directions at 150 mm/s and at -150 mm/s; one link
group is being blasted and then the substrate/stage is reversing
direction, moving 0.1875 mm in the cross-axis and blasting the
second link group. For a constant velocity trajectory, the total
time is calculated is 3 parts, the time to blast the first link
group, the time to reverse on-axis direction and step 0.1875 mm in
the cross-axis, and the time to blast the second link group. The
time to blast each link group is (9 mm)/(150 mm/s)=60 ms. The time
to reverse on-axis direction is 40 ms, the time to step in the
cross-axis by 0.1875 mm is 10 ms. Since the time to reverse on-axis
direction is greater than the time to step in the cross-axis this
is effectively the time required for the turn around. Therefore the
total time to process the two link groups is about (60 ms)+(40
ms)+(60 ms)=160 ms.
[0055] Using the methods of certain embodiments of this invention
it is possible to begin the turn around in the on-axis direction
before the blasting is complete. In this example, the deceleration
is begun 15 ms before the end of the link group blasting. Then the
total time to process the two link groups is (60 ms-15 ms)+(40
ms)+(60 ms-15 ms)=130 ms resulting in a time savings of 30 ms or
about 19%. This trajectory also requires less total stage travel,
therefore, this example also shows how total stage travel can be
reduced near the stage travel limits.
[0056] In a third example, the laser system in a memory repair
system is used to measure the location of a substrate on a stage.
This operation is referred to as alignment. During an alignment
operation the substrate is moved at a low constant velocity such
that a feature on the substrate crosses the focused laser beam
spot. I accordance with an embodiment, the laser is fired at a
constant repetition rate, the pulses are optically attenuated by
the AOM, and directed to the substrate along the laser processing
output axis. The pulses are attenuated such that the feature on the
substrate is not damaged by the laser pulse. During alignment
reflected light from the feature is collected by a detector and a
signal is generated. From this collected light signal it is
possible to determine the location of the feature on the substrate.
A similar method is used to determine best focus location. During
these operations the velocity of the stage is typically about 10
mm/s. The length of a scan is typically about 50 .mu.m. Often a
number of these scans are done serially crossing the same feature
on the substrate numerous times in both directions with the stage
stopping between operations. The time for one of these scan
operations using constant velocity methods is the time to
accelerate from stopped to 10 mm/s, the time for the actual scan,
and the time to decelerate to stopped. Typically the acceleration
and deceleration times are equal to the minimum acceleration pulse
time, assumed here to be 5 ms each. The time to scan is equal to
(50 .mu.m)/(10 mm/s)=5 ms. The total scan time, therefore, is 15
ms. Using the methods of certain embodiments of the invention, the
constant velocity segment is not required. The stage could be
accelerated up to 10 mm/s in 5 ms time (the minimum acceleration
pulse time) and then decelerated back to stopped in another 5 ms
time (the minimum acceleration pulse time). The acceleration could
be set such that the total move distance is 50 .mu.m, the scan
distance. For this example an acceleration of 4 m/s.sup.2 is
required. The resulting total scan time is 10 ms resulting in a
time savings of 5 ms or about 33%.
[0057] In accordance with another embodiment, the alignment
operation may be a carried our during acceleration, with a low
power laser, high speed detector, and data converter. The low power
laser output axis may be aligned to the laser processing axis.
[0058] Typically the substrate is located on the XY stage with some
finite rotation of the link group axis relative to the ideal stage
X and Y axes. When blasting link groups therefore, there is
typically a non-zero stage velocity in both the X & Y stage
axis. During link blasting in conventional constant velocity
systems, the stage motion in both stage axis, X and Y, are constant
velocity. Because of the small rotation of the stage ideal axis
relative to the substrate on-axis and cross-axis directions the
stage velocity in the stage ideal coordinate system in one axis
will be nearly equal to the on-axis velocity and the stage velocity
in the stage ideal coordinate system in the other axis will be
close to but not equal to zero. Therefore, when planning stage
trajectories using the methods of this invention, it is necessary
to consider this small angle between the stage ideal coordinate
system and the substrate on-axis/cross-axis coordinate system. If
stage trajectories are planned in the substrate on-axis/cross-axis
coordinate system then the trajectory can be planned such that the
cross-axis velocity is zero at the beginning of a link group to be
blasted. If stage trajectories are planned in the stage ideal
coordinate system then more care may be required in trajectory
planning. If stage trajectories are planned in the stage ideal
coordinate system and the stage axis close to the cross-axis is
planned to have a constant velocity motion at the beginning of the
link group to be blasted then, if the stage is moving at a
non-constant velocity (accelerating or decelerating) in the stage
axis close to the on-axis direction a small cross-axis error in
blast position may occur. There are a number of methods to handle
this possible cross-axis error. Some of the methods include: 1)
Plan the trajectories in the substrate on-axis/cross-axis
coordinate system. 2) Plan the trajectories in the stage ideal
coordinate system and accept a small cross-axis error. Larger
cross-axis errors are acceptable in link blasting because links are
typically long in the cross-axis direction. 3) Put a small rotation
into the XY stage to make the XY stage ideal coordinate system be
the same as the substrate on-axis/cross-axis coordinate system. 4)
Plan a non-constant velocity during blasting in the stage ideal
coordinate axis that is close to the substrate cross-axis
direction.
[0059] In accordance with further embodiments, the present
invention does not require that each of the links to be blasted in
a group be blasted along a single on-axis line across the group of
links. In other words, some off-axis movement across a group of
links may be permitted in certain embodiments.
[0060] In accordance with further embodiments, a system of the
present invention does not required that the links be evenly
spaced. In further embodiments, the system may provide increased
maximum positioning velocity. In further embodiments, the system
may provide increased laser output pulse rates and burst rates. In
further embodiments, the system may reduce acceleration forces to
reduce attendant mechanical perturbations thereby improving system
accuracy. In further embodiments, the system may be adapted for use
with multi-path and multi-spot beam positioning systems.
[0061] Pulsed lasers utilized in various embodiments may include
mode locked lasers, high speed laser diodes, or various
combinations thereof. For example, a GHz repetition rate/rate mode
locked semiconductor lasers is disclosed in "Passively Mode-Locked
Diode-Pumped Surface-Emitting Semiconductor Laser, IEEE Photonics
Technology Letters, vol. 12, No. 9, September 2000 by S. Hoogland
et al., and includes a semiconductor saturable absorber mirror
(SESAM) in an output cavity coupled to a semiconductor laser diode.
In accordance with other embodiments, mode locked lasers of
Coherent, Inc. of Santa Clara, Calif., or of Lumera Laser GmbH of
Germany may also be used. In accordance with further embodiments,
mode locked laser as disclosed, for example, in U.S. Pat. Nos.
4,914,663 and 6,210,401 may also be used.
[0062] As previously noted, link pitch and dimensions may continue
to shrink, approaching 1 um pitch with 0.1-0.4 um link widths. Link
processing systems will produce decreasing spot size to match the
trend. For instance, as reported in "Laser Processing of Ultra Fine
Pitch Fuse Structures in 65 .mu.m Node Technology", Society for
Equipment and Materials International, SEMICON West 2004 by Joohan
Lee, James Cordingley and Joseph J. Griffiths discloses a green
laser that produced a spot size as fine as about 0.7 .mu.m to
process fine pitch links. The decreasing spot dimensions provide
for a potential decrease in the energy of a pulse. For example, if
the spot diameter is decreased from about 1.5 .mu.m to about 1.0
.mu.m, the total energy in a pulse may be decreased two-fold to
achieve a given energy density at the target structure. Further, in
some embodiments it may be advantageous to increase the number of
pulses applied to a link, and to proportionately decrease the pulse
energy, for example. In such embodiments the output of the pulsed
laser may be sufficient to process links with minimal gain in the
optional amplifier 74, or to possibly operate without the amplifier
74. In some embodiments an amplifier may be advantageous, or
required. For instance, such a situation may arise if a few pulses
are to process the target material.
[0063] An optional optical amplifier may be used to amplify the
output of the pulsed laser. For example, a mode locked oscillator,
high speed diode, or fiber laser oscillator may produce picosecond
or femtosecond pulses with about 100 picojoules to a few nanojoules
of energy per pulse. The amplifier will generally be used to
increase the pulse energy to levels suitable for the material
processing task, for instance with gain of 6 dB-50 dB. The
amplifier may be a fiber laser amplifier, waveguide amplifier, or a
diode pumped solid state amplifier. The amplifier may be CW or
pulse pumped.
[0064] The amplifier performance will generally be limited based on
a compromise of several parameters: gain, output repetition rate,
and number of pulses (duty cycle) as a result of at least the
achievable average power limitations.
[0065] In certain embodiments at least one of the shape and energy
distribution of a group of output pulses may be altered if the
amplifier is pulse pumped or otherwise operated at varied
repetition rates. For example, if a pulse picker is used prior to
the amplifier the inputs to the amplifier will be spaced at
non-uniform temporal intervals, corresponding to a group of pulses
are selected for processing. The effective repetition rate will
vary, and the resulting effects will be noticeable if the amplifier
is operated within a frequency range wherein the output pulse
energy varies with the rate.
[0066] In some embodiments a high speed modulator, for instance the
output modulator 78, may be used to control the pulse shape and
energy distribution based on pre-determined spacings. For example,
U.S. Pat. No. 5,128,601, the entire disclosure of which
incorporated by reference herein, teaches the use of a modulator to
provide for constant output pulse energy characteristics, as well
as other methods for pulse control. Other systems, such as
disclosed for example in U.S. Pat. Nos. 5,812,569 and 5,226,051
involve output stabilization, and the principles have been employed
for use in q-switched laser systems.
[0067] In some preferred embodiments of the present invention the
pulsed laser is to be operated at a constant rate with pulses
selected with the output modulator 78. In such cases amplification
is preferably achieved with the use of a multistage amplifier
design, wherein each stage is operated so as to amplify the input
pulse train while avoiding distortion, and also with gain low
enough such that the amplifier operates well within the average
power specification. This embodiment is particularly suited for
operating at short wavelengths, for instance, green or UV
wavelengths, wherein a non-linear crystal performs wavelength
shifting. One stage of amplification may include a pre-amplifier.
Cascaded amplifiers 74 may be fiber amplifiers. PCT Publication
WO98042050, entitled "Pulsed Semiconductor High Power Amplifier and
Exemplary Applications", describes many aspects of high power fiber
amplifier systems, including multistage amplifiers, and exemplary
applications in laser material processing.
[0068] FIGS. 8A-8C compare blasting links using conventional
methods and using the methods of this invention. FIG. 8A shows the
distance verses time curves when blasting the same link group using
conventional methods 100 and using the methods of an embodiment of
the present invention 102. When blasting using conventional methods
the velocity is the same (constant) for the entire link group
resulting in a linear constant slope distance verses time curve
100. When blasting using the methods of this invention the velocity
is not constant during the blasting time. In the example shown in
FIG. 8A, at the beginning and end of curve 102 the blasting
velocity is equal to the blasting velocity of curve 100, in the
center of curve 102, however, the blasting velocity has increased
resulting is a steeper slope to the distance verses time curve and
resulting in an overall decrease in the total blast time.
[0069] In this example the links to be blasted have a constant
pitch; the links are equally spaced in distance. When using
conventional methods blasting at a constant velocity the blasts for
the equally spaced links result in blasts that are equally spaced
in time as is shown at 104 in FIG. 8B, represent the times for each
blast. When using the methods of this invention blasting at
changing velocity the blasts for the equally spaced links (constant
pitch) result in blasts that are unequally spaced in time as is
shown in FIG. 8C. In particular, the lines shown at 106 represent
the times for each blast at the beginning, the lines shown at 108
represent the times for each blast in the middle, and the lines
shown at 110 represent the times for each blast at the end of a
link group. For this example the blasts at the beginning of the
trajectory 106 and the blasts at the end of the blast trajectory
110 are spaced in time at the same spacing as the blasts using
conventional methods because the velocity is the same at these
times. The blasts at the center of the trajectory 108 are more
closely spaced in time but are equally spaced in distance with
respect to the blasts when using conventional methods because the
velocity at this time is greater. Since the same number of blasts
are required when using conventional methods as when using the
methods of this invention the same number of blasts are shown in
FIGS. 8B and 8C.
[0070] Those skilled in the art will appreciate that numerous
modifications and variations may be made to the above disclosed
embodiments without departing from the spirit and scope of the
invention.
* * * * *