Scale-Up Activities in Atomic Layer Deposition at Argonne Jeffrey Elam, Anil Mane, Joe Libera December 9, 2011 Large Area Picosecond Photodetector Collaboration Meeting December 9-10, 2011, Argonne National Laboratory
Outline Introduction to ALD (1 slide) ALD on 33mm plates Challenges with scale-up to larger substrates Conclusions 2
Al CH 3 OH Al CH 3 A) B) OH Al(CH 3 ) 3 OH Trimethyl Aluminum (TMA) CH 4 Al CH 3 Al CH 3 H2OH2O Al CH 3 OH Al CH 3 Al CH 3 Al CH 3 OH Al CH 3 H2OH2O H2OH2O OH CH 4 OH Binary Reaction Sequence for Al 2 O 3 ALD 1 ALD Cycle of TMA/H 2 O Deposits 1 Al 2 O 3 “Monolayer” 3
4 Multiple 33mm MCPs in Tubular ALD Reactor Thickness uniformity on monitor Si(100) <2% The resistive layer thickness ~800A Similar thickness trend observed on second batch of 5 MCPs Excellent batch-to-batch reproducibility Chem-2 Resistive Coating
5 Resistance Comparison for 9 MCPs (air vs. vacuum) Average resistance (2 batches of 5 plates) 115 ±12 M ~10% resistance variation Chem-2 Resistive Coating
Challenges to Coating Larger Areas Need a bigger reactor Non-ideal ALD surface reactions High aspect ratios High surface areas 6
Large Substrate Reactor From 1x1-in plates to 12-in x 18-in plates 7 2” tube 12x18” box
Coating 8” MCPs in Beneq mm chamber
Non-Ideal ALD Surface Reactions 9 InCp + O 3 →In 2 O 3 In 2 O 3 O 3 O* + O 2 Indium oxide catalyzes ozone destruction In 2 O 3 films were thinner downstream in reactor In 2 O 3 ALD in Large Substrate Reactor
Scale-up of ALD In 2 O 3 to 12”x18” Substrates InCp + H 2 O + O 2 InCp + O 3 Thickness deviation: 2.5% Resistivity deviation: 6% Thickness deviation: 45% 10 Non-self limiting ALD reactions can lead to non-uniform films in larger ALD systems
15 μm Conformal coating on all exposed surfaces Aspect ratio ~ nm pores, 70 microns long 11 Coating High Aspect Ratios: ALD W in Anodic Alumina
Large Substrate Surface Area: Silica Gel Powder 100 micron particles, 30 nm pores (aspect ratio ~ 2000) Surface area = 100 m 2 /g Powder bed fixture for ~1 g support 12
Self-Limiting Al 2 O 3 ALD Self-limiting growth on planar and porous surfaces Exposures increased by x100 Silica GelPlanar Surface 13
Surface Areas of Glass Capillary Arrays: 14 MCP Typed (µm)γ = l/dαA Total (cm 2 ) 33mm diam , ,317 8in square , ,712 (thanks Jason M.) Empty tube reactor: ~2000cm 2 Empty LSR: ~4000 cm 2 Empty Beneq 300mm: ~1600 cm 2 MCP aspect ratio = 60 (we’ve done 10^5) 8” MCP surface area = 6.5 m 2 (we’ve done 10^3) Piece of cake…
Work Plan 15 1) Qualify Beneq for Chem2 coatings using coupons and 33mm MCPs in new 300mm chamber 2) Coat single 8”x8” MCPs in 300 mm chamber 200 mm chamber 3) Coat multiple (1-4) 8”x8” MCPs in 300 mm chamber X) Coat multiple (20) 8”x8” MCPs in 3D chamber 300 mm chamber
<1% non-uniformity in 100 nm coating 16 Al 2 O 3 ALD in 300 mm Reactor Works great!
Chem-2 ALD in 300 mm Reactor 17 Test Metrics: Optimized chemistry -2 baseline process on 300mm chamber: Deposit 300mm wafer at same condition as MCP deposition. Evaluated the resistivity uniformity across the large area Precursor inlet direction 300mm
Resistivity of Chem2 coating at different locations on 300mm wafer 18 Works great for Chem-2!
Chem-2 thickness NU on 300mm wafer with 8”x8” MCP on top:2D thickness map 19 ItemsRun-3 Average (nm) STDV sigma0.105 %sigma Max (nm) Min(nm) % diff We may need square chamber? 300mm 1-sigma thickness: 10% August 22, 2011
8” 40µm MCP Pair Gain Map Y gain slice X gain slice Y gain slice FIRST Gain map – Looks awful - multifiber visible, lots of gain non-uniformity This is from the thickness non-uniformity (thanks Ossy, Jason, SSL) 20
mm wafer 8” MCP reactor wall Troubleshooting in Beneq Top View Side View ALD conductive coating with MCP substrate installed over Si(100) 300 mm wafer Measure film thickness on Si wafer using 4-point probe conductivity (easy, quantitative) assumed flow distribution
Process testing on 8” MCP on Beneq 22 ItemsThickness data from Resistance Average (nm) STDV sigma0.045 %sigma4.539 Max(nm) Min(nm) % diff Precursor inlet Baseline process: 50 cycles conductive coating (No MCP) 1-sigma = 4%
Thickness values form resistance and 2D map: 23 ItemsBaselineRun-1Run-2Run-3 # of cycles50cycles 30cycles 25cycles Expected thickness (nm) Dose time (s) Average (nm) G (nm/cycle) STDV sigma %sigma Max (nm) Min(nm) % diff Tested many ideas, hardware, software, reactor breaks, etc. (3 months) 1-sigma = 4%57%52%10%24%
24 Troubleshooting in Beneq: Al 2 O 3 Side View ALD Al 2 O 3 coating with MCP substrate installed over Si(100) 300 mm wafer Visually assess coating uniformity Measure film thickness on Si using ellipsometry
Troubleshooting in Beneq: Al 2 O 3 Anil: “We have a flow problem… we need a square chamber” 50 nm 100 nm 150 nm Precursor inlet 25
Si MCP Troubleshooting in Tubular Reactor 26 2cm x 30 cm Si(100) substrate, rest 2x20cm MCP 2 mm above Si 500 cycles TMA/H 2 O for Al 2 O 3 Narrow gap under MCP High surface area 2 inches 26
Troubleshooting in Tubular Reactor , 0.05 Torr doses , 0.2 Torr doses , 1 Torr doses Some non-uniformity, not as bad as Beneq
Si MCP Si Troubleshooting in Tubular Reactor Narrow gap under MCP Narrow gap over MCP High surface area 28
Troubleshooting in Tubular Reactor , 0.05 Torr doses Same bad non-uniformity as in the Beneq 29
30 Si Troubleshooting in Tubular Reactor , 0.05 Torr doses Nearly perfect uniformity Narrow gap under MCP” Narrow gap over MCP No high surface area
300 mm wafer 8” MCP reactor wall What is going wrong in Beneq? Top View Side View actual flow distribution Narrow gap under MCP Narrow gap over MCP High surface area No flow in the gap Slow outgassing/diffusion from MCP Precursors mix, CVD 31
The problem: flow is bypassing the MCP Solution: we need to confine the flow so that it is forced to pass in the gap between the plate(s) and the reactor wall. We have a flow problem… we need a square reactor. What is going wrong in Beneq? 50 nm 100 nm 150 nm Precursor inlet 32
300 mm wafer 8” MCP reactor wall Potential Solution, Conclusions Top View Side View Equal gap Convert circle into square CFD modeling would be helpful Listen to Anil 33