Comparison of before/after blanking amplifier replacement

1/19/11

Process Flow

• substrate:• silicon piece

• spin coat:• 6% HSQ• 2500 rpm, 1250 rpm/sec, 60 sec• hot plate bake 80 C, 4 min• thickness = 137 nm

• expose• 100 kV, 2 nA, shot pitch = 2 nm• dose = 1500, 3000, 5000, 7000, 10000 uC/cm2

• develop• 25% TMAH, 30 sec immersion• 1 min 30 sec, DI water rinse

conversion 1hsqcircle500-001.v30, JBXFILER, output step = 1000 (1 nm grid)

width = 250 nm

outer diameter = 5 um

Dose = 1500 uC/cm2

before after

Dose = 3000 uC/cm2

before aftergap

gap

before after

Dose = 5000 uC/cm2

rough edge / fracture shape positioning error

gap

before after

Dose = 7000 uC/cm2

rough edge

rough edge

Dose = 10,000 uC/cm2

before after

rough edgerough edge

Conclusion

• new blanking amplifier confirmed to have working waveform, but the pattern result of the ring is the same

conversion 5single3.v30, created by Ming Lu, Brookhaven National Laboutput step = 1000 (1 nm grid)

Ming’s pattern has primitives exposed in a clockwise progressive order. (JBXFILER and Layout BEAMER primitives are exposed in random order.) The first primitive is at 3 o’clock.

first primitiveexposure order

Sequential ordering of primitive / fracture shape exposure

compare conversion method 1 to conversion method 5(both with new amplifier)

Dose = 1500 uC/cm2

conversion method 1 conversion method 5

Dose = 3000 uC/cm2

conversion method 1 conversion method 5

beginning / end

Dose = 5000 uC/cm2

conversion method 1 conversion method 5

Dose = 7000 uC/cm2

conversion method 1 conversion method 5

Dose = 10,000 uC/cm2

conversion method 1 conversion method 5

beginning / end

Conclusion

• Ming’s pattern looks much smoother at all doses• for conversion method 1, primitives are exposed in random order, so larger position changes are being made by PDEF• for conversion method 5, each primitive is exposed in a clockwise order, so there are smaller position changes being made by PDEF• therefore, the ring is suffering from a positioning related error• the rings expose very quickly (<< 1 sec) so the positioning error is not likely due to beam drift. • it seems the error is either due to PDEF itself, or some noise that is coupling into PDEF