STRASBAUGH CONFIDENTIAL INFORMATION
2
Altitude Above Sea Level
If a 7AF is located above sea level, it has been getting incorrect values for its vacuum readings. For example, at
5000 feet, an idle 7AF, which should read zero inches of Mercury, actually reads (-3.7). This causes problems
when a test is made for sufficient vacuum. If you experience this problem, set machine variable 39, Altitude
Vacuum Adjust, to zero. Then read the work chuck vacuum from the screen while the 7AF is idle. Finally,
return to machine variable 39 and set it to the number you just read.
Rough Wafers
Recently it has been attempted to grind poly-silicon wafers having small bumps. If a bump happens to lie on the
circular path taken by the probe as it travels around the wafer, the probe is smacked out of the way. It can be hit
with enough force to make it lose track of its current position. The coarse grind cycle has been modified to
recover from this loss of information.
Machine Variable 37 is called Probe Bounce Threshold. If you place a value into this variable, software will
monitor the variation in the probe reading. This monitoring takes place over the number of revolutions of the
wafer specified in machine variable 96, Grind Probe Average Revolutions. While the variation remains greater
than the Probe Bounce Threshold, the probe is said to be bouncing. During probe bouncing, the probe value
may drift farther and farther from reality, but its actual value will be ignored. Only the variation in value will be
monitored. When bouncing falls below the threshold given by this variable, the bumps will have been removed
by the grind wheel. At this point the probe will be raised and lowered to reset itself, and then the coarse grind
cycle will run to completion. If you set this variable to zero, grinding behaves normally, without examining
probe bouncing.
A second item of data is also needed in this process. Without reliable probe readings, how close can the grind
wheel approach the wafer at high speed before it risks a collision? Feed Wafer Thickness taken from the
grinding recipe will be used for this. Feed Wafer Thickness is already required to be as large or larger than the
thickest wafer. It is recommended that you add a bit more, perhaps fifty microns, as an extra cushion.
Warning. Your rough wafers may differ from the rough wafers already tested. It is strongly recommended that
you test with dummy wafers before grinding any valuable wafers.
Warning. The obstacles on some of the rough wafers tested at Strasbaugh were strong enough to bend the
probe in addition to smacking it out of the way. If your wafers damage the probe, the 7AF should not be used to
grind them until some other tool has removed the bumps.
Warning. Because of the lag in examining the variation over a large number of revolutions of the wafer, a large
value for Grind Probe Average Revolutions in variable 96 can increase the chances of overshooting your grind
target.
Extra Rinsing of the Wafer
The D.I. cutting water continues to spray and the chuck and grind wheel continue to spin during the Lift step of
the grind. At the end of the Lift, all three normally stop. However, a non-zero value can be placed into machine
variable 46, Chuck Wafer Clean secs, to keep the water spraying and the chuck and wheel spinning for this
many seconds after the Lift ends. When the time has expired, the water will turn off and the grind wheel will
stop, but the chuck will keep spinning as the bridge moves. This provides an extra rinse, intended to prevent
grind debris from transferring from the wafer to the robot end effector.
To Next Page
Loading...
