Semiconductor device manufacturing: process – Continuous processing
Reexamination Certificate
2002-03-06
2003-01-14
Talbott, David L. (Department: 2827)
Semiconductor device manufacturing: process
Continuous processing
C438S908000, C438S716000, C029S025010
Reexamination Certificate
active
06506693
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to semiconductor processing systems, and more particularly, to semiconductor processing systems having removable cassettes for holding processed and unprocessed workpieces.
In order to decrease contamination and to enhance throughput, semiconductor processing systems often utilize one or more robots to transfer semiconductor wafers, substrates and other workpieces between a number of different vacuum chambers which perform a variety of tasks. An article entitled “Dry Etching Systems: Gearing Up for Larger Wafers”, in the October, 1985 issue of Semiconductor International magazine, pages 48-60, describes a four-chamber dry etching system in which a robot housed in a pentagonal-shaped mainframe serves four plasma etching chambers and a loadlock chamber mounted on the robot housing.
FIG. 1
of the present application illustrates a typical loadlock chamber
10
having a cassette
12
for holding unprocessed wafers
13
to be unloaded by a robot
14
and transferred to various processing chambers (not shown) attached to a mainframe
16
.
The loadlock chamber
10
is a pressure-tight enclosure which is coupled to the periphery of the mainframe
16
by interlocking seals
18
which permit the loadlock chamber to be removed and reattached to the mainframe as needed. The cassette
12
is loaded into the loadlock chamber
10
through a rear door (not shown) which is closed in a pressure-tight seal. The wafers are transferred between the mainframe
16
and the loadlock chamber
10
through a passageway
20
which may be closed by a slit valve
22
.
Although only eight wafers are illustrated for purposes of clarity, a typical cassette
12
may be initially loaded with as many as 25 or more unprocessed wafers
13
or other workpieces before the cassette is loaded into the loadlock chamber
10
. After the loadlock access door is closed and sealed, the loadlock chamber is then pumped by a pump system (not shown) down to the vacuum level of the mainframe
16
before the slit valve
22
is opened. The robot
14
which is mounted in the mainframe
16
then unloads the wafers from the cassette one at a time, transferring each wafer in turn to the first processing chamber. As best seen in
FIG. 2
, the robot
14
includes a robot hand or blade
24
which is moved underneath the wafer
13
to be unloaded. The robot
14
then “lifts” the wafer
13
from the wafer supports
26
supporting the wafers
13
in the cassette
12
. By “lifting,” it is meant that either the robot blade
24
is elevated or the cassette
12
is lowered by a suitable lifting mechanism
30
such that the wafer
13
is lifted off the cassette wafer supports
26
. The wafer may then be withdrawn from the cassette
12
through the passageway
20
and transferred to the first processing chamber.
Once a wafer has completed its processing in the first processing chamber, that wafer is transferred to the next processing chamber and the robot
14
unloads another wafer from the cassette
12
and transfers it to the first processing chamber. When a wafer has completed all the processing steps of the wafer processing system, the robot
14
typically returns the processed wafer back to the cassette
12
from which it came. Once all the wafers have been processed and returned to the cassette
12
, the cassette in the loadlock chamber is removed and another full cassette of unprocessed wafers is reloaded.
One problem associated with systems having single loadlock chamber and cassette is that the robot will run out of unprocessed wafers to unload before all the wafers are processed and returned to the cassette. Processing chambers will become idle as the last wafer works its way through the system. The time necessary for each wafer to pass through each of the processing chambers and to return to the cassette depends upon the number of processing chambers in the system and the time required for each processing step. Some systems have as many as 10 processing chambers in which each processing step can take 2 minutes or more. Thus, in some systems it can take 20 minutes or longer for the last wafer to return to the cassette, during which time the processing of additional wafers is halted.
Furthermore, once the cassette has been filled with the processed wafers, it is necessary to vent the loadlock chamber, remove the old cassette and insert a cassette of unprocessed wafers into the loadlock and then pump down the loadlock chamber to the vacuum level of the robot chamber. Although the increased vacuum isolation provided by such systems can improve product quality, it has been difficult to achieve commercially acceptable throughput for high vacuum processes, such as, for example, physical vapor processes such as sputtering. The time required to pump down the loadlock chambers to the base level, following loading of a cassette of wafers into the loadlock chamber, can be excessive.
In order to increase throughput, it has been proposed to utilize two loadlock chambers as described in U.S. Pat. No. 5,186,718, which is incorporated in its entirety by reference. In such a two loadlock system, both loadlock chambers are loaded with full cassettes of unprocessed wafers.
FIG. 3
of the present application illustrates in time line form several processing cycles for such a two loadlock system
100
(
FIG. 4
) having a mainframe
116
to which a plurality of processing chamber
140
are coupled. As shown in
FIG. 3
, the robot
14
starts unloading unprocessed wafers from the first loadlock chamber A (“L.L.A.”). Following a certain delay, the first wafer will have completed processing by the system
100
. At that time, it is believed that the robot typically loads the first processed wafer back into the cassette of loadlock chamber A, the same cassette from which it was unloaded. This process continues until the robot completes unloading the unprocessed wafers from the cassette of loadlock chamber A. However, because the system has two loadlock chambers, the processing chambers are not idled as the last wafer from loadlock chamber A passes through the system. Instead, once the supply of unprocessed wafers from loadlock chamber A is exhausted, the robot begins to unload unprocessed wafers from loadlock chamber B to start another unload-load cycle (cycle
2
) as shown in FIG.
3
. As the processing of individual wafers is completed, the robot continues to return processed wafers to the cassette of loadlock chamber A until the loadlock chamber A cassette is filled with processed wafers (completing unload-load cycle
1
). At that time, the next processed wafer (the first wafer unloaded from loadlock chamber B) will be returned to the cassette of loadlock B from which it was unloaded.
As the robot unloads unprocessed wafers from the cassette of loadlock chamber B and returns processed wafers to the cassette of loadlock chamber B, the slit valve to loadlock chamber A may be closed to permit the loadlock chamber A to be vented and the cassette removed and replaced with a cassette of unprocessed wafers. Once loadlock chamber A has been pumped back down to the base vacuum level and the robot unloads the last unprocessed wafer from loadlock chamber B, the robot can resume unloading unprocessed wafers from the cassette of loadlock chamber A, initiating a third unload-load cycle (cycle
3
). Thus, if the loadlock chamber A with the new cassette of unprocessed wafers can be pumped down before the robot finishes unloading the cassette of loadlock chamber B, processing of wafers can be continued uninterrupted.
When the robot has finished loading processed wafers into the cassette of loadlock chamber B ending unload-load cycle
2
, the cassette in loadlock chamber B can be removed and replaced with a full cassette of unprocessed wafers. In the meantime the robot continues to unload unprocessed wafers from and load processed wafers into the cassette of loadlock chamber A. In this manner, wafer processing can be maintained more continuously, significantly increasing throughput over the single loadlock systems.
Howe
Brezocsky Thomas B.
Davenport Robert E.
Heyder Roger V.
Applied Materials Inc.
Konrad Raynes & Victor & Mann LLP
Talbott David L.
Thai Luan
LandOfFree
Multiple loadlock system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Multiple loadlock system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Multiple loadlock system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3059747