Imaging across the world. Hiroshi Matsumoto, Munehiro Ogasawara and Kiyoshi Hattori April 18 th, 2013

Transcription

1 Imaging across the world PMJ 213 Panel Discussion Challenges for future EB mask writers Hiroshi Matsumoto, Munehiro Ogasawara and Kiyoshi Hattori April 18 th, 213

2 vs. pixelated gray beam Two shaping apertures used to form a triangular or rectangular beam. Pixelated gray beams A number of square shaped beams, of fixed size, created by array of shaping apertures. Electron gun 1 st shaping aperture Shaping deflectors Condenser lenses Projector lenses 2 nd shaping aperture Sub deflectors Main deflectors Objective lens Reticle Yoshitake et al., Proc. of SPIE Vol D-8, 211 Platzgummer et al., Proc. of SPIE Vol , 211 Slide 2

3 vs. pixelated gray beam Two shaping apertures used to form a triangular or rectangular beam. Pixelated gray beams A number of square shaped beams, of fixed size, created by array of shaping apertures. Electron gun 1 st shaping aperture Amplitude: 1.875V (x1 xffff) 1LSB:.229mV (±15 V/ 17bit DAC resolution) Condenser lenses (IMS poc tool) Shaping deflectors Projector lenses 2 nd shaping aperture Sub deflectors Main deflectors +/- 3mV Objective lens 32 ns, or shorter, settling time was demonstrated on test bench with EBM-8 Sub DAC Amp Reticle Yoshitake et al., Proc. of SPIE Vol D-8, x512 blanker array in 2 mm sq. chip Platzgummer et al., Proc. of SPIE Vol , 211 Slide 3

4 Motivation for multibeam technology Throughput independent of pattern size In systems smaller shot size results in smaller exposure current and larger shot count, to increase total exposure time and total settling time. Curvelinear features can be written more easily systems use rectangular or triangular figures. Slide 4

5 vs. pixelated gray beam Shaped beam (triangle, rectangle) exposed with uniform dose. Pixelated gray beam Square beam exposed with modulated dose a : pixel size Dose profile can be reproduced with gray beam, if beam size is sufficiently small D=D D=D D=.5D Slide 5

6 vs. pixelated gray beam Shaped beam (triangle, rectangle) exposed with uniform dose. Pixelated gray beam Square beam exposed with modulated dose dose shaped beam design pattern size (3 nm) irradiated dose deposited dose threshed dose for resist process dose pixilated beam (2 nm) design pattern size (3 nm) irradiated dose deposited dose deposited dose () position position pixilated beam (5 nm) irradiated dose dose deposited dose design pattern size (3 nm) deposited dose () position Slide 6

7 vs. pixelated gray beam Shaped beam (triangle, rectangle) is exposed with uniform dose. Pixelated gray beam Square beam is exposed with modulated dose deposited dose with, 2nm pixel, 5 nm pixel a : pixel size dose threshed dose for resist process design pattern size Dose (3 profile nm) can be reproduced with gray beam, if beam size is sufficiently position small D=D D=D D=.5D Slide 7

8 Writing experiment (1) writing and pixelated gray beam writing were compared in writing experiments using the EBM-8 (single writer) and FUJIFILM PRL-9 Shot sizes of 1, 2, 5 and 1 nm, with 5% dose for edge pixels. Edge pixels were written in different write pass. Several chips were written with different dose D. (a = 1 nm) (a = 5 nm) (a = 2 nm) (a = 1 nm) pixelated beam 8

9 Writing experiment (2) Shot size : 1 nm pixelated beam 1 nm 1 nm 9

10 Writing experiment (3) Shot size : 2 nm pixelated beam 2 nm 2 nm 1

11 Writing experiment (4) CD [nm] 5 a [nm] dose [AU] CD [nm] 5 pixelated gray beam average of a [nm] dose [AU] Writing accuracy of pixelated beam improves as beam size decreases. Beam size of 1nm and 2 nm brings the same CD accuracy, with a discernible slope difference to writing. 11

12 Challenges for multi-beam writers Beam size needs to shrink, as beam blur reduces. beam blur (FWHM) [nm] assumed from ITRS forecast for direct write year pixel size [nm] year Introduction of multi-pass exposure with grid offset can improve gray beam write accuracy, but this is not addressed in this discussion. How can pixel size shrink? Increased demagnification Performance of high demag. optics is questionable. Reduction of aperture size, accompanied by either of : Increased # of beams with reduced beam pitch Increased beam current density Slide 12

13 pixel size [nm] [ 1 14 ] 8 data volume [Byte] Challenges for MBMW year Inevitable with 2x/node shot count increase MB year Which is the practical option? Low sensitivity resist may multiply write passes and data volume demagnification [AU] J [AU] beam pitch [AU] beam pitch 6 4 # of beam year year year Slide 13 # of beam [AU]

14 Summary Pixelated gray beam can have writing accuracy equivalent to, with sufficiently small beam size Error budget is needed to estimate feasible accuracy, as actual, beamlets have error in position, size and exposure current.. Challenges for multi-beam mask writers Smaller beam size for smaller beam blur Multi-pass writing with grid-offset is necessary. Shrinkage of beam pitch required with increase of # of beams. Otherwise, J or optical demag. should be increased. Integrity of explosive data volume Roadmap for 1-year evolution Challenges for mask writers Smaller shot size for smaller patterns Further increase of J and reduction of settling time is required. Shift to multi-column strategy Slide 14

15 END Slide 15