02/24/ th FIB/SEM User Group Meeting, Washington DC Valery Ray Developing FIB GAE Recipes: Practical Application of “Unfinished Theory” PBS&T

02/24/ th FIB/SEM User Group Meeting, Washington DC 2 Outline  Yield Enhancement and Milling Rate criteria for characterizing beam GAE processes  Choosing beam raster parameters to maximize Milling Rate  Numerical and image examples  Conclusion

02/24/ th FIB/SEM User Group Meeting, Washington DC 3 GAE Recipe Development: Dose Enhancement vs. Milling Rate  Established GAE theory (K. Edinger, JVST B 18(6) 2000 and Microelectron. Eng. 57–58, 2001, also I. Utke et. al. JVST B 26(4) 2008) is developed with emphasis on yield enhancement criteria  Miling rate is better suitable for practical applications: » Recipes with maximized milling rates are required for etching High Aspect Ratio vias » Recipes with minimized milling rates are required for uniform deprocessing of Cu with minimal dielectric over-etch » Recipes with highest ratio of chemical milling rate to physical sputtering rate are required for high material selectivity

02/24/ th FIB/SEM User Group Meeting, Washington DC 4 GAE Recipe Development: Yield Equation Yield = = AR + AS Removed Atoms Incident Ions Jt D AR (Atoms Reacted) – FAST, parameter-sensitive, not limited by aspect ratio. AS (Atoms Sputtered) –SLOW, limited by aspect ratio J - Ion Beam Current Density t D – Time of beam dwell within the pixel

02/24/ th FIB/SEM User Group Meeting, Washington DC 5 GAE Recipe Development: Reactive Yield vs. Mill Parameters Effect on AR (Reactive Yield) Pixel Refresh 1~ 10mSec Pixel Overlap ~ 0 PixelDwell 50 nSec Parameter and Limit

02/24/ th FIB/SEM User Group Meeting, Washington DC 6 GAE Recipe Development: Phases of GAE Within Dwell Point t D = t AR + t AS t D → t AR, and t AS → 0 High-Rate GAE within dwell point requires shortest practically possible dwell time Chemical reactions occur on pico-second scale FIB dwell times are 10s or 100s of nanoseconds

02/24/ th FIB/SEM User Group Meeting, Washington DC 7 GAE Recipe Development: View from the Dwell Point  GAE process is happening within dwell point  Replenishment of gas begins when primary particle beam moves away from the dwell point  Therefore refresh time of each dwell point (not the “raster”) is critical for gas replenishment

02/24/ th FIB/SEM User Group Meeting, Washington DC 8 GAE Recipe Development: Optimal Raster Time t Raster = t Refresh = Σt Di i=0 n Raster time equivalent to optimal refresh time provides most efficient GAE

02/24/ th FIB/SEM User Group Meeting, Washington DC 9 GAE Recipe Development: Refresh Time and Number of Dwell Points N DP = t Refresh t D(Min.) Practical Practical t D(Min.) for modern FIBs is in the range of 200nSec to 50nSec

02/24/ th FIB/SEM User Group Meeting, Washington DC 10 GAE Recipe Development: Via Size Defines Dwell Point Distance dX = dY = L (Sqrt (N) - 1) Dwell points are desirable on the edges of the via

02/24/ th FIB/SEM User Group Meeting, Washington DC 11 GAE Recipe Development: Dwell Point Distance = Beam Size  Beam diameter equivalent to pixel distance ensures high reactive yield  Corresponding beam current value is controlled by ion optics; diffused beam is desirable D Beam = dX = dY For uniform orthogonal raster:

02/24/ th FIB/SEM User Group Meeting, Washington DC 12 GAE Recipe Development: Numerical Example of HAR Recipe 2μm x 2μm via in Si milled with Cl 2 (t Refresh = 1 mSec) on system with minimal dwell 0.2 μSec: N = 1000μSec / 0.2μSec = 5000 pixels / raster Beam Diamter = dX = dY = 2μm / (Sqrt(5000) – 1) = ~ 30 nm Corresponding beam current depends on FIB system, but typically will be around 20pA to 10pA or even lower Beam current can be increased for low (less then 5:1) aspect ratio work (surface micromachining)

02/24/ th FIB/SEM User Group Meeting, Washington DC 13 GAE Recipe Development: 10:1 HAR Via Etching Example 6.2μm 0.68μm 0.67μm Needle gas injector Trifluoroacetic Acid Precursor SiO2 substrate ~10min. etching time ~0.6um/min etching rate for 10:1 aspect ratio

02/24/ th FIB/SEM User Group Meeting, Washington DC 14 Dose nC Time min. Contact, μm All vias are 5μm deep XeF2 precursor on SiO2 substrate “Proof of Concept” test, milling small HAR vias with Beehive concentrator on FEI Vectra 986+ system GAE Recipe Development: 10:1 to 25:1 HAR Via Etching Example Flat ~0.7um/min milling rate with gas concentrator for aspect ratios 10:1 to 25:1

02/24/ th FIB/SEM User Group Meeting, Washington DC 15 GAE Recipe Development: Cu deprocessing Examples 100nm Cu line cut over 100nm dielectric Flat deprocessing through four layers of “Dummy” Cu

02/24/ th FIB/SEM User Group Meeting, Washington DC 16 Conclusions  Milling rate criteria is better suitable for characterizing practical GAE processes  Reviewing GAE process from the perspective of beam dwell point allows converting yield-based theory to rate- based practical applications  Rate-optimized GAE process establishes direct relationship between size of repair and beam diameter

02/24/ th FIB/SEM User Group Meeting, Washington DC