Ratcheting deformation in thin film structures Z. SUO Princeton University Work with MIN HUANG, Rui Huang, Jim Liang, Jean Prevost Princeton University Q. MA, H. Fujimoto, J. He Intel Corporation
Interconnect Structures Complex architecture Small feature sizes Diverse materials Intel: 130nm technology ibm.com
Thermal Cycling: Qualification Test ~1000 cycles Temperature -55 C 15 C 150 C 15 C -40 C Time Make-and-break Time consuming, a bottleneck Multiple failure modes Little understanding of mechanisms New materials
Flip-chip structure Package substrate (1.1mm thick, 30-40mm size) ~1000 cycles Underfil/fillet (150um thick) Solder bumps (100 um diameter, 150 um spacing) Silicon die (0.7mm thick, 10-15mm size) Underfill SiN (0.45um thick) Al-Cu um thick Polyimide (4 um thick) -55 C 15 C Lower level interconnects (10-15 um thick) Silicon SiN film cracking Metal pad crawling
Plan view 500 cycles between -55 C and 15 C Edge view Courtesy of Dr. J.B. Han, Agilent Technologies Singapore
Empirical Observations of Cracking Organic substrate Silicon die At die corners After temperature cycles In SiN film over Al pads, but not in SiN film over silica More likely when Al pad is wide
Flip-chip structure Questions Package substrate (1.1mm thick, 30-40mm size) Underfil/fillet (150um thick) Solder bumps (100 um diameter, 150 um spacing) Silicon die (0.7mm thick, 10-15mm size) Why cycling? SiN does not fatigue. SiN (0.45um thick) Al-Cu Underfill um thick Polyimide (4 um thick) Lower level interconnects (10-15 um thick) Silicon Why cracking? *Shear stress is limited by the yield strength of polymer, say 100 MPa. *Breaking strength of SiN is 1000 MPa.
Ratcheting Plastic Deformation Temperature 150 C 15 C -55 C Time 0.5 µm SiN Huang, Suo, Ma, Fujimoto, J. Mater. Res., 15, 139 (000) µm 30~40 mm Organic substrate Silicon die Biased Shear Stress polymer Aluminum pad 10~100 µm Silicon 0.7 mm Silica and low level interconnects (10~15µm thick) 1.1 mm 150 µm thick polymer layer Biased shear stress, from organic substrate Metal yields, from CTE mismatch with SiO
Stress builds up in SiN. STEADY STATE First cycle SiN passivation film Metal pad Many cycles, STEADY STATE SiN passivation film Metal pad t L t = τ L 0 τ 0L = t
Ratchet and Pawl Cyclic motion One-directional motion
Metal Pad Crawling Al pad Polyimide Silica Al pad SiN After ~1000 thermal cycles Polyimide SiN Al pad Al pad Silica Large plastic deformation over many thermal cycles
Finite Element Simulation 1 cycle 10 cycles 50 cycles 100 cycles
Shear stress in metal relaxes as temperature cycles Shear Stress in Al pad decreased as thermal cycle 6.00E+07 τ m Shear stress in Al pad (Pa) 5.00E+07 4.00E+07 3.00E+07.00E+07 1.00E+07 0.00E+00-1.00E+07 1 10 50 100 0 5 10 15 0 5 30 35 40 45 50 -.00E+07-3.00E+07-4.00E+07 Position (µ) SiN passivation film Metal pad τm
Membrane stress in SiN builds up as temperature cycles Stress built up in SiN film as thermal cycle.5e+09.0e+09 100 & Liquid Model Normal stress in SiN film (Pa) 1.5E+09 1.0E+09 5.0E+08 0.0E+00-5.0E+08-1.0E+09-1.5E+09 50 10 1 0 10 0 30 40 50 -.0E+09 -.5E+09 Position (µ)
Metal pad geometry. Slots. SiO Al Temperature 150 C 15 C -40 C Direct finite element calculations for 3D, many cycles take VERY long time. Time
3D Structure xx xy + y yx yx y τ mx τ my yx dy x yy + y + x xy Steady-state: yy Shear stress in metal vanishes Elastic plane stress problem xy yy dx dy xx + x xx dx z y y x τ my u τ mx v passivation film metal substrate x
Slot!
Further Issues Why cycle, again? Why not just plastic collapse? Number of cycles to approach the steady state.
ε p 3 4 5 1 Blanket film dε p = α ( m α s )dt Aluminum stress Y Silicon E strain +Y 1 ( ) d = E α m α s 1 ν dt Y T L 3 5 4 T H T ( ) α f α s ET H T L ( 1 ν)y ( ) >, cyclic plasticity <, shakedown
Why Crawling? Temperature Organic Substrate Silicon die 150 C 15 C -55 C Time Metal film Silicon stress Strain increment film Y E strain Huang, Suo, Ma, Acta Materialia, 49, 3039-3049 (001)
Elastic-Plastic Model stress, ε p Strain increment Y E strain Matching strains of film & substrate, ε p film, ε p, γ p, ε p dε p = α ( f α s )dt + 3 = Y Yield condition J flow theory dε ij p = s ij dλ dε p /3 = dγ p Shear strain increment dγ p = 6 ( α f α s)dt
E ( α α )( T T ) f s H ( 1 ν )Y L Four-color map Cyclic Plasticity Crawling Plastic Collapse Shakedown 0 1 3 Y
a) b) Y Stress and strain change with temperature Y 3τ 3τ t 0 i 0 E A T L A E T L B B F F Strain per cycle D D 1 E C T H T H C 0 γ p = T T ( α α ) f 1 ν s C(C ) 1 ( 1 ν ) E c) d) ε p γ p 0 E A A T L E T L B B F F 1 1 D 6τ 0 Y ( α ) f 3τ E( α f α s )T ( H T L ) ( 1 ν) Y 3τ 0 f α s s 0 Al Si α Al > α Si C(C ) T H C D C ( α α ) T H p γ T
Ratcheting-Creep Analogy Normalized shear stress, τm/y Y 1 3 3 0 τ E m E m ( α m α s )( TH TL ) ( 1 ν ) Y ( α m α s )( TH TL ) ( 1 ν ) Y linear approximation γ p = m m > 1 ( 1 ν ) E < Strain per cycle Normalized ratcheting strain rate, E m ( α m α s )( TH TL ) ( 1 ν ) Y γ p p γ Y / m ( 1 ν m ) Em = E( α f α s )T ( H T L ) ( 1 ν) Y 3τ 0 dγ dn dγ dt
substrate Elastic passivation film Equilibrium Elasticity N 0 ~ L D ~ L Hh Number of cycles to reach steady-state h x = τ = E p u x u N = D u x H η E m E p u E m α T Y L L 1 metal film passivation film Ratcheting metal film τ = η H η = E m 1 D = EHh η u N H E m α T Y h 1 Huang,Suo, Ma, J Mech.Phys. Solids. 50, 1079 (00)
Concomitant crack extension and underlayer ratcheting Tensile Film da dn = 0.6 Hh ηg c Ratcheting Layer Rigid Substrate η = E m 1(1 v m ) E m α T (1 v m )Y 1 J. Liang, R. Huang, J.H. Prevost, Z. Suo, submitted
Conclusions Thermal cycling test is a bottleneck. Ratcheting: cyclic temperature, aided by biased shear stress, causes uni-directional deformation. Stress relaxes in metal, and builds up in SiN. Ductile, low k dielectrics may ratchet. Needthin film data on large plastic deformation over many cycles, and long time. www.princeton.edu/~suo