The imprint resolution, process cycle time, and process life are all critically dependent on material,
mold and mold release agents.
The published resolution records for directly imprinted features include;
• 6nm half pitch, edge lithography generated by Chou (Chou 1996)
• 9 nm half pitch, e beam generated by AMO (Vratsov 2004)
• 2 nm nanotube replicated by Willson
• 1 nm line discontinuity replicated by Motorola (Mancini 2003)
Line width control is as important as minimum feature size.
• < 1 nm line edge roughness demonstrated by (Kurashima 2003)
• < 2 nm field to field repeatability reported by Motorola (Mancini 2003).
In pattern transfer applications, the etch step contributes to resolution and line width control because
of the thickness and uniformity of the residual layer. The residual layer thickness is a based on
conservation of volume and depends on the starting thickness, and the duty cycle and height of the
Residual layer variation is a function of pattern density simply caused by material not being able to
redistribute. Can be minimized by long fill times, in practice people use short fill time and deal with
the variation. Post imprint planarization processes such as SFIL-R can reduce the line width variation.
Drop dispense systems can simple change the volume of material depending on the local pattern
duty cycle. MII have reported using a automated program to analyze pattern data files and change the
material dispense appropriately.
All the results above were for hard templates, soft PDMS templates tend to distort during imprint. A
three layer mold with a 4 x harder organo-silicon top surface has shown 50 nm resolution at close to 1
atmosphere (Plachetka 2005).
The cycle times have been improving significantly over the last few year. Depends on many factors
including materials pressure and pattern type. Some typical data that has been reported includes;
• 5-15 mins for thermal imprint depending on material and molecular weight (Roos 2001),
• 3 mins for thermal imprint using the latest materials and “hero “ process reported by Microresist
• < 60 secs for UV spin on at Nanonex (Li 2004)
• 7 secs for UV drop dispense at MII (Shumaker 2005)
• 1 - 5 secs for thermal embossing on a thick film (Ohta 2005)
• 50 um for whole wafer thermal imprint
• 1.3 um 3 sigma for whole wafer UV imprint by Nanonex (Tan 2004) and (Islan 2002)
• 250 nm 3 sigma for S&R, room temp, drop dispense, UV cure Imprio 100 system by MII
• 32 nm 3 sigma relative to an optically stepped reference pattern for a S&R, room temp, drop
dispense, UV cure Imprio 250 system with magnification correction by MII (Choi 2005)
Process life and defect density are intertwined. In most cases some form of release treatment on the
mold is required to get the mold to release. Based on the onset of gross defectivity
• 50 imprints as a typical number in a review by Torres.
• 1000 imprints for improved reactive surface treatment release systems reported by EVG (Vratsov
• 3000 imprints for 30 nm features inspected by SEM using proprietary release system with drop
dispense materials reported by MII (Schumaker 2005).
Depends critically on the overall facility and process cleanliness.
Defect density experiments are particularly challenging. The latest results for immersion lithography
are 0.04 random defects per squ cm > 90 nm at 40 wafer an hour (Ronse 2005). By comparison,
imprint has achieved;
• < 0.1 random defects per squ cm > 250 nm, and 12 fixed defects per squ cm maintained for 12
wafers, 30 fields per wafer. The fixed defect count was achieved on the third mold that had been
fabricated, the defect density had dropped 100 x over the 3 cycles of learning. Reported by MII using
drop dispense and a proprietary release strategy (Shumaker 2005)
Also, depends critically on the overall facility and process cleanliness.
For more information Contact Us
Copyright © 2005 Impattern Solutions. All rights reserved.