Archive for September, 2010

Sep 19 2010

D2S and eBeam Initiative – available for std machines

Published by under Uncategorized

At the recent Bacus/SPIE Litho conference in Sept, the eBeam Initiative and its lead D2S made a couple of major announcements that expand the applicability of eBeam technology.  This years’ progress featured two major advancements in mask creation and writing, and a full day at the Bacus event focused on direct write for wafers.
Prior advancements by the group included, character based projection through custom eBeam stencils and also curved shaped beams.  Both of these improved pattern accuracy while dramatically reducing write time for both masks and wafers.  However, these techniques requied new eBeam machines for the imaging floor, and were not retrofittable to existing installations.
At the event in Monterey, they announced similar results to the curve based beams and shots, by using a VSB (variable shaped beam) shot pattern.  These patterns are Manhattan based and can be overlapped to create the resulting device patterns.  The results in larger shot patterns with less variability and also higher coverage of the intended pattern as shown in the following figure.  This overlapping shots have a significantly shorter write time for greater throughput.  The shots can be created with currently installed eBeam machines, and the new patterns are created using a new fracture/writing routine.
The second major announcement and improvement in throughput and writing time is the proof for results for the alternating shot patterns.  This was designed primarily for SRAF (sub-resolution assist features) but also works for regular devices such as fringe caps, inductors and most multi-legged structures.  The method works by having a fracture/writing pattern that utilizes a minimization of the blanking time, as all the patterns are written at full beam power.  The technique is shown in the following figure.
For the direct write of wafers, there was discussion of trends for the 3Xnm and the 2Xnm technologies.  As EUV is being pushed out beyond the 22nm node, direct write is becoming a viable options.  There were several presentations, including a keynote by D2S about the applicability of the direct write method for ASICs and SOCs being created with these tools.  In order to address the throughput issues, these flows are targeting the use of either the Jeol Multi-beam or the Avantest Multi-column ebeam machines.  For these designs, the direct write is being used primarily for “hole” type patterns such as contacts and vias.  This helps push the 22nm issue in manufacturing from “not perfect masks” with variability to “not perfect wafers” with variability.  With the challenges that are present at the 22nm node, direct write helps bring higher predictability to the end wafer by creating systematic patterning in high variability layers.
PC

At the recent Bacus/SPIE Litho conference in Sept, the eBeam Initiative and its lead D2S made a couple of major announcements that expand the applicability of eBeam technology.  This years’ progress featured two major advancements in mask creation and writing, and a full day at the Bacus event focused on direct write for wafers.

Prior advancements by the group included, character based projection through custom eBeam stencils and also curved shaped beams.  Both of these improved pattern accuracy while dramatically reducing write time for both masks and wafers.  However, these techniques requied new eBeam machines for the imaging floor, and were not retrofittable to existing installations.

At the event in Monterey, they announced similar results to the curve based beams and shots, by using a VSB (variable shaped beam) shot pattern.  These patterns are Manhattan based and can be overlapped to create the resulting device patterns.  The results in larger shot patterns with less variability and also higher coverage of the intended pattern as shown in the following figure.  This overlapping shots have a significantly shorter write time for greater throughput.  The shots can be created with currently installed eBeam machines, and the new patterns are created using a new fracture/writing routine.

overlapping VSB shots

Figure - Overlapping VSB shots

The second major announcement and improvement in throughput and writing time is the proof for results for the alternating shot patterns.  This was designed primarily for SRAF (sub-resolution assist features) but also works for regular devices such as fringe caps, inductors and most multi-legged structures.  The method works by having a fracture/writing pattern that utilizes a minimization of the blanking time, as all the patterns are written at full beam power.  The technique is shown in the following figure.

alternating shot eBeam pattern

Figure - Alternating shot eBeam pattern

For the direct write of wafers, there was discussion of trends for the 3Xnm and the 2Xnm technologies.  As EUV is being pushed out beyond the 22nm node, direct write is becoming a viable options.  There were several presentations, including a keynote by D2S about the applicability of the direct write method for ASICs and SOCs being created with these tools.  In order to address the throughput issues, these flows are targeting the use of either the Jeol Multi-beam or the Avantest Multi-column ebeam machines.  For these designs, the direct write is being used primarily for “hole” type patterns such as contacts and vias.  This helps push the 22nm issue in manufacturing from “not perfect masks” with variability to “not perfect wafers” with variability.  With the challenges that are present at the 22nm node, direct write helps bring higher predictability to the end wafer by creating systematic patterning in high variability layers.

PC

No responses yet