Outline• Introduction• **Flare** evaluation methods and early results– Donut pattern method• What are donuts ?• Point Spread Function data from 2 resists– LS method• How to determine flare **distance** ?• Results from 3 resists• How to determine flare magnitude, η ?• Results from 3 resists• Summary2008 International Symposium on Extreme Ultraviolet Lithography 2

Introduction• The cause of EUV flare is mirrorsurface roughness.• **Flare** specification for the ADT is16%; the actual value is ~12%.• We were successful in our 45nmdevice demonstration using rulebased-OPC derived from mirrorpower spectral density (PSD) data.– Residuals thru pitch were lessthan 5 nm.• Do the following patterns measure“actual” flare ?– Donut patterns• to determine flare **distance**– LS pattern• traditional e-beam proximity effectcorrectionToday, I I will will show early results of of Donut and and LS LS evaluation method.2008 International Symposium on Extreme Ultraviolet Lithography 3

Outline• Introduction• **Flare** evaluation methods and early results– Donut pattern method:• What are donuts ?• Point Spread Function data from 2 resists– LS method:• How to determine flare **distance** ?• Results from 3 resists• How to determine flare magnitude, η ?• Results from 3 resists• Summary2008 International Symposium on Extreme Ultraviolet Lithography 4

What are donuts ?Small flare200nm post0.25/ 198.8 0.5/ 198.6 0.75/ 196.8 1/ 194.7Large flareouter radius (μm)/ inner CD (nm)2.5/ 190.2 5/ 191.2 10/ 190.3 25/ 187.5 50/ 183.4-- CD CD of of inner inner post post is is affected by by outer outer radius radius of ofclear clear annulus.-- CD CD measurements of of resist resist images of of donut donutpatterns yield yield a value value for for flare flare **distance**.2008 International Symposium on Extreme Ultraviolet Lithography 5

How to determine **Flare** PSF from Donuts data• Process:– Determine CD sensitivity to dose– Going from smallest to largest donutouter rings:• Determine change in CD due to eachadditional ringdeltaCD(n) = waferCD(n) – waferCD(n-1)• Determine dose required to cause thisamount of CD changedeltaDose(n) = deltaCD(n) / DoseSlope• Assuming spillover intensity to center isconstant for any given ring, determine themagnitude of the spillover to cause therequired change in dose at the centerspillover(n) = deltaDose(n) / (2π(r outer –r inner ))• Plot this spillover intensity vs. donutradius. This should be a crudeapproximation to the **Flare** Intensity PSFwaferCD(n)waferCD(n-1)deltaCD(n)2008 International Symposium on Extreme Ultraviolet Lithography 6

**Flare** evaluation by donut patternsSPILLOVER COEFFICIENT (a.u.)2.01.81.61.41.21.00.80.60.40.20.0-0.2Resist A: longer diffusion length (DL)Line: best fit double Gaussian curveResist B: shorter diffusion length0.1 1 10 100OUTER RADIUS (um)There There is is a huge huge spillover area, area, especially at at < 2 μm μm outer outer radius. radius.Both Both Resist Resist A and and B show show the the same same trends trends even even though they they have havedifferent diffusion lengths.2008 International Symposium on Extreme Ultraviolet Lithography 7

Outline• Introduction• **Flare** evaluation methods and early results– Donut pattern method:• What are donuts ?• PSF data from 2 resists– LS method:• How to determine flare **distance** ?• Results from 3 resists• How to determine flare magnitude, η ?• Results from 3 resists• Summary2008 International Symposium on Extreme Ultraviolet Lithography 8

How to determine flare **distance** ? (50nm LS pattern)GDS dataexamplemeasurement100um50nmLS area50nm : Effective area is 50%45nm : **Flare** effective area is 40%40nm : **Flare** effective area (density) is 25%SPACE CD**Flare****Flare****distance****distance**Space CD should be affected bypattern density **distance** from dark regionDistance from dark region (μm)2008 International Symposium on Extreme Ultraviolet Lithography 9

**Flare** **distance** (50nm LS w/ 100 µm-sq)6056Resist ASPACE CD (nm)52484440363228?0 10 20 30 40 50DISTANCE FROM DARK REGION (μm)Resist BResist CMASKResists A, A, B and and C show show few few mm, mm, ~16 ~16 μm, μm, and and 23 23 μm μm flare flare **distance**, respectively.Fall Fall off off of of Space Space CD CD near near dark dark region region is is not not due due to to mask mask error. error.2008 International Symposium on Extreme Ultraviolet Lithography 10

How to determine flare magnitude, η ?• Find optimum dose at the center of a largearea that contains LS patterns at variouspattern densities:• Calculate η by fitting the dose data to thefollowing equation:DxD50Dose=1+η1+2ηxη FittingDose vs Pattern Density2.01.81.61.41.21.00.80.60.40.20.00 0.2 0.4 0.6 0.8 1pattern densityDx: Optimized dose for various densityD50: Optimized dose at 50% densityx : pattern densityη0.20.40.60.8Pattern density (%)Line width (nm)50 40 32110203040506070809099NOVACD pattern for evaluating η(each square is 100 μm-sq area)2008 International Symposium on Extreme Ultraviolet Lithography 11

DxHow was = derived ?D501+η1+2ηx• Assumption (1): pattern area > flare **distance** > LS pitchEDxEx = (0.5 + ηx) DxEx….….0.5 DxηxDx (=**Flare**)Ex: Optimized energyX : pattern densityDx: Optimized dose(We can think of η as ratio of flare in BF.)• Assumption (2): optimum energy for target CDs shouldbe at the same Ex, even if its pattern densities aredifferent. (in this case, base-line density is 50%.)Ex = (0.5+ ηx) Dx = (0.5+ η 0.5) D50 = E50Dx =1+ η1+ 2ηxD502008 International Symposium on Extreme Ultraviolet Lithography 12

Parameter η (flare magnitude)1.20Optimum Dose η1.151.101.051.00η= 0.1η= 0.05η= 0.2Resist A: long DL (η = 0.184)Resist B: short DL (η = 0.0654)Resist C: long DL (η = 0.146)Lines : best fit curves0.950 10 20 30 40 50 60PATTERN DENSITY (%)Optimum η values values are are 18.4, 18.4, 6.45, 6.45, & 14.6% 14.6% for for 3 Resists A, A, B & C, C, respectively.Long Long DL DL resists resists show show higher higher η values. η strongly depends on on resist resist material.2008 International Symposium on Extreme Ultraviolet Lithography 13

Summary• Early results of 2 unique flare evaluation methods are shown.– Donut patterns: huge spillover area at < 2 μm outer radiusfor Resist A and B. (No data yet for Resist C.)– LS pattern: resists with long diffusion lengths show higher ηvalues. η strongly depends on resist material.MagnitudeResist Along DLResist Bshort DLResist Clong DLη 18.4 % 6.45 % 14.6%PSF Same trend No dataDistance LS edge Few μm ~16 μm 23 μm• These results will be utilized for our next device demo.2008 International Symposium on Extreme Ultraviolet Lithography 14

AcknowledgementsThis work was performed by the Research Alliance Teamsat various IBM Research and Development Facilities• ASML– Kevin Cummings– Bill Pierson– Sang-In Han– Rick Zachgo• College of Nanoscale Science andEngineering– Sudhar Raghunathan2008 International Symposium on Extreme Ultraviolet Lithography 15