IBM Spin Coating Process. Greg Burtt Greg Hewitt Dave Valente Contact Engineer: Kevin Remillard Faculty Mentor: Jeff Marshall. Photo Resist. Process Setup:. Chuck. Silicon Wafer. Nozzle. Vacuum. 200. Problem Statement.
Contact Engineer: Kevin Remillard
Faculty Mentor: Jeff Marshall
Resist Buildup Target
Fluid Exit Hole
Clockwise from top right:
Solvent droplet defect
Ball-bearing defect (photoresist)
XMO (excursion monitor wafer)
XMO scan results (the dot is the location of the found defect)
The photographs on top are from an excursion monitor ran at IBM during testing. .75µm lines are etched on a wafer and it is then developed and scanned for defects.
Vertical lines are spaced .75µm apart
Eventually discarded due to complex manufacturing and the cost associated.
Preliminary Airflow Calibration: Between 650 and 700 linear feet per minute
Backside Rinse Volumetric Flow Calibration: 100mL/minute
This corresponded to only .1mm in fluid height, a number that was questioned until initial tests confirmed the calculations were correct.
When the primary concept was to add additional holes aiming upward, the velocity out the existing holes and one additional hole was plotted as a function of the diameter of the new hole.Fluid Exit Velocity Calculations
The flow out of the slots on the modified disk is more of a spray than the stream that is observed to come from the original holes
Photoresist was applied to the top bowl
After the cleaning process was run, the top bowl was inspected.
The prototype cleans above the knee, where the original disk fails to reach.
A particle detector was used during the last round of testing at IBM to measure the amount of solvent dispersed into the air inside the photolithography machine. Three minute long cycles were used, sampling the air at a rate of .9cfm.
IBM had never done this test with their existing cleaning process, but it was a major concern that if the new cleaning disk put more solvent particles into the air, it could cause more defects.
To everyone’s surprise, the cleaning process created no particles between .1 and 1µm. The prototype cleaning disk created minimal amounts of particles on the first three runs, and then dropped to zero. Presumably, the first particles were left over from machining and transporting the prototype.
Particle detection results for the first cleaning disk test.
Six wafers tested. Only nine defects were found. This is well below the threshold value.
1.) Run cleaning disk
2.) Process 3 wafers
3.) Inspect wafers for defects