UV Nanoimprint Tool and Process Technology. S.V. Sreenivasan December 13 th, 2007

Transcription

1 UV Nanoimprint Tool and Process Technology S.V. Sreenivasan December 13 th, 2007

2 Agenda Introduction Need tool and process technology that can address: Patterning and CD control Alignment and Overlay Defect Control Throughput Conclusions

3 Step & Flash Imprint Lithography (S-FIL) High resolution fused silica template, coated with release layer Template Step 1: Dispense drops Planarization layer Substrate Imprint fluid dispenser Low viscosity fluid (Si-containing for S-FIL, Organic for S-FIL/R) Step 2: Lower template and fill pattern Step 3: Polymerize imprint fluid with UV exposure Template very low imprint pressure < 1/20 atmosphere at room temp Planarization layer Substrate Same Process Used for Step & Repeat and Whole Substrate Patterning Planarization layer Substrate Step 4: Separate template from substrate Template Planarization layer Substrate Step & Repeat

4 Stepper Technology for Non-Volatile Memory Non-Volatile Memory is the fastest growing segment in the semiconductor industry Production insertion targeted at 22nm Imprint on the International Technology Roadmap for Semiconductors (ITRS) Toshiba using tool for device prototyping 18nm feature size <1nm CD uniformity <2nm LER, 3σ Sub-15nm overlay <10 defects/cm 2 IMPRIO 250 Fully automated 300mm Tight overlay capability Sub 20nm resolution IBM has used stepper to make a critical level in a 7 layer storage class memory device 30nm FinFET lithography demonstrated though etch with excellent CD control and line edge roughness (LER 2nm 3σ)

5 Whole Substrate Technology for HDD Hard Disk Drives Moving To Patterned Media 1000 Technology Driver: Greater Areal Density Areal Density (Gigabits/in 2 ) Longitudinal Perpendicular recording recording Patterned media Fully automated whole disk imprinter Sub 20nm resolution High throughput Multiple commercial tools installed HGST is using our tool

6 Patterning and CD Control Patterning Requirements Need to handle pattern density variations Need to achieve thin and uniform residual layer Typical aspect ratio is 2:1 to 2.5:1 to avoid feature collapse issues Typical mean residual layer is <1/4 th of the feature height More drops for denser patterns on disk Feature Height of 75nm for 30nm features RLT = 10nm mean

7 Uniform Residual Layer: Pattern Density Variations 15 nm residual layer, independent of pattern density

8 Transfer and Residual Layer Thickness Measurements 1 nm adhesion layer thickness measurement Ultra-thin adhesion layer Residual layer mean <20nm and thickness variation to < 6 nm TIR MII Metrosol Mean: 1.09nm Sigma: 0.05nm Max: 1.22nm Min: 0.94nm x, y (mm) Residual Layer Thickness (nm) Average Std Dev Min Max Range Residual layer thickness fully populated wafer Position # nm HP

9 Alignment and Overlay Alignment is achieved using matched Moiré fringe patterns on both template and substrate based on a technique originally developed for X-ray proximity printing at MIT Imprint fluid lubricates movement and dampens vibration Align mechanics Template Substrate <100nm Magnification mechanics Substrate Template d(phase) Target Acquisition Resolution Test dx(nm)-template Alignment resolution is better than 2nm

10 Imprint to Photo Overlay Budget Components J Other Process Distortions (CMP, Film Depositions, Etc.) G Template/Mask Pattern Generation Distortion E Thermal In-Plane Template Distortions A X- Y- Alignment Noise (Machine Noise) D B Full Mix-and-Match Process Overlay (FM&MPO) Mix-and-Match, Multi- Template/Mask Overlay (M&MMTO) Single Machine Overlay (SMO) Field Alignment Accuracy MagX, MagY, Ortho Noise (Mag Actuator Noise) H Distortion Due to Tool to Tool Template/Mask Chuck Shape Difference F Thermal In-Plane Wafer Distortions C i-mat Moiré Alignment Metrology Noise I Distortion Due to Imprio to Photo Tool Wafer Chuck Shape Difference Temperature Control Machine Precision K Photo Tool Lens & Scan Speed Matching Distortion

11 Representative Alignment Data (From Four Corners) Four corner alignment data over the wafer represents basic machine precision in X, Y, Theta, MagX, MagY and Ortho directions: 5nm 3sigma in X 6nm 3sigma in Y

12 Multi-Machine Mix-and-Match Overlay Results Mix and match overlay performance with two different 193nm scanners has been demonstrated Overlay metrology performed using an industry standard KT overlay tool. 32 fields per wafer, 81 positions per field Achieved sub 20nm, 3σ results Challenges to move to sub- 10nm overlay appear to be engineering related.

13 Progress in S-FIL Defect Reduction at MII Defect Density by Date(KT-2132) Defect Density cm Non-commercial templates Commercial templates Improved adhesion layer Improved wafer cleanliness and template dicing process 02/ / / / / /2005 Date 02/ / / /2007 How can we achieve <1 cm -2?

14 S-FIL Defectivity 1. Template Fab Defects Template 4. Bubbles 2. Material Contaminants 3. Front Side Particles Planarization layer Substrate 5. Back Side Particles Template Planarization layer Substrate Planarization layer Substrate 6. Improper Release Template Planarization layer Substrate 7. Post-Imprint Fall-On Particles

15 Imprint Defectivity Imprio 250 Template was not inspected during Fabrication Template has 3 defects defect density = 3.1 cm -2 Imprinted wafers 89 fields Inspected al fields Inspected area per field ~1 cm 2 Pareto at right shows total defect densities for random and repeating defects Defect sizes > 200 nm (KT- 2132) Total wafer defect density = 3.4 cm -2 Imprint defectivity = 0.2 cm Template Ion Contm. Defect Density by Type Fall on Particle Plug Random 0.04 Prior Particle 0.04

16 Template Defect Inspections (counts) Defect Total Template Defectivity Through Fabrication After Cr and quartz etch After Mesa etch After Cr strip 0 defects Inspections were performed with a KT-576 tool by mask vendor 90 nm pixel Reflected light mode Maximum sensitivity Inspection performed on 6025 plates only Template # Defects detected on the template were not always seen during wafer inspection No attempt was made to repair these defects

17 Comparison of KT es32 and KT-2132 inspections Edge-roughness defects: Found only on the bottom edge of horizontal lines False defects Particles defects did not contribute to repeating defectivity Three template pattern defects were found repeating imprint defects One contaminated contact was detected repeating plug Type Edge roughness Particle Template Plug KT es32 Density cm KT-2132 Density cm -2 Repeating Template Defects Found only by es No new imprint related defectivity was identified

18 Template Life Template damage limited by S-FIL process Very low uniform pressure, low viscosity materials No impact: In S-FIL, the drops (1-2um high) prevent high-speed impact of particles with template In S-FIL Imprinting, particles are shielded by droplets 1 μm tall droplet Particle Template damage is rare, small percentage of particles cause damage We have the potential for a reliable approach to detect particle events in the tool based on low resolution fullfield image capture and analysis Small Z height causes large XY hot-spot

19 Particles that Do Not Damage Template Imprint um Particle that did not cause template feature contamination, no change in repeating defect density Particle that caused limited repeating defect in 4 fields

20 Example of Rare Template Damage Even a 10 um particle leads to small damage LARGE Front-side particle (>10um) Next imprint has small defect Small Damage!

21 Multiple Wafer Run Defectivity: I250-3 I250 tool with improved ECU and template handling. Internally coated wafers -older generation tools Manually cleaned templates 3 Template defects Inspected 21 fields per wafer Defect Density cm Wafer Defect Density Total Defect Density with Template Defects Removed Pilot Total Defect Density Repeating Defect Density Imprint # Wafer #

22 Particles that May Cause Template Damage Organic (soft) particles and particles that are embedded underneath polymer films do not appear to cause damage Inorganic and metal particles that show up on the wafer just before the imprint step are most relevant to template damage We are undertaking a particle classification study to understand where these hard particles may be coming from

23 Throughput Risk: Fast Fluid Fill For HVM, need fluid fill of < 1 second/field for > fields per wafer Keys to Fast Fluid Fill Low viscosity imprint resist (monomer) Small drop volume: Pico liter sized drops Template contact geometry control GDS based volume targeting Inclined template geometry creates fluid wave-front to avoid air trapping between liquid drops 6pl drop Contact geometry control Fluid fill direction

24 GDS Based Volume and Contact Geometry Control No Contact Geometry Control Contact Geometry Control Drop Pattern Uniform Grid GDS Based Uniform Grid GDS Based MagAlign >40 secs 25 secs 30 secs 4 secs (High Pattern Density Variation) Contacts through Pitch 30 secs 15 secs 10 secs 3 secs (Low Pattern Density Variation) GDS based volume compensation and contact geometry control are both needed This data collected for 6 pl minimum drop volume and for 25nm mean residual layer

25 Fill Time Set to < What is Needed (2.5 rather than 4sec) Dominant non-fill defective feature (last features to fill) X and Y moiré Cross and L Highly repeatable Filling Defects Cross and L Y moiré UL Verniers BnB LR Verniers X moiré CD Bars

26 Throughput Summary & Next Steps Further improvement in filling speed in being sought using Smaller drops of liquid Better understanding of drop placement optimization Improved control over contact geometry

27 Conclusions Template Throughput Align Defects Photolithography Progression Multiple levels in volume manufacturing Mix-and-match for critical layers in memory Thin film heads Device prototyping, mix & match, small volume devices Defect/align tolerant apps (storage, optical devices, etc.) Litho Performance (Resolution, LER, CD Control) S-FIL Progression

28 Acknowledgements