Electromigration reliability: Question bank — Hinglish deep dive

6.4.8 · D5 · Hardware › Power, Thermal & Reliability › Electromigration reliability

Shuru karne se pehle, ek shared picture words mein: ek wire ek corridor hai metal atoms ka; ek strong current ek electron ki hawa hai jo un atoms ko dhakelta hai. Kuch break nahi hota jab tak atoms kisi jagah pile up ya drain away na karein — woh "jagah" hamesha woh spot hoti hai jahan atom flow badalta hai (ek divergence). Woh sentence dhyan mein rakho; neeche ke zyaadatar traps bas us sentence ka hi disguise hain.

Definition Is page ke symbols (ek baar define karein, neeche freely use karein)

Kisi bhi trap ke mention karne se pehle, is page ka har symbol plain words aur units mein:

$J$ = current density, current jo har unit cross-sectional area se squeeze hoti hai (units $A/cm^{2}$ ). Iska sign/direction conventional current wala hai — woh direction jisme ek positive charge flow karta. Electrons ulti direction mein flow karte hain, isliye electron wind atoms ko $J$ ke opposite direction mein dhakelta hai.
$L$ = wire length (units $cm$ ), woh distance jo current interconnect ke saath travel karti hai. Yeh $J$ ke saath milkar Blech product $J \cdot L$ banata hai jo short-line immortality decide karta hai.
$E$ = electric field metal ke andar (units $V/cm$ ). Yeh $J$ ke saath (same direction mein) point karta hai, kyunki ek resistive wire ke andar field hi hai jo current ko aage drive karta hai. Yeh koi bahar se apply kiya field nahi hai — yeh internal field hai jise Ohm's law current se relate karta hai.
$ρ$ = resistivity metal ki (units $Ω \cdot cm$ ): material current ko kitna resist karta hai. Local form mein Ohm's law kehta hai $E = ρ J$ — ek zyada resistive metal mein same current push karne ke liye zyada bada field chahiye.
$q$ = elementary charge ( $1.6 \times 1 0^{- 19}$ coulombs), ek unit electric charge ka size; yeh "field" ko "charged ion pe force" mein convert karta hai.
$Z^{*}$ = effective charge, ek unit-less lumped number jo direct field push aur (dominant) electron-wind push ko bundle karta hai. Iska sign usually wind term se set hota hai, isliye yeh atoms ko anode ki taraf point karta hai.
$k_{B}$ = Boltzmann's constant ( $8.617 \times 1 0^{- 5}$ eV/K), temperature aur energy ke beech ka bridge — yeh "kitna hot" ( $T$ kelvin mein) ko "atoms mein kitni thermal jiggle energy" mein convert karta hai.
$E_{a}$ = activation energy (eV mein): woh energy hill jo ek atom ko naye site pe jump karne ke liye climb karni padti hai.
$D$ = diffusivity (atoms kitni easily bhatak sakte hain, units $cm^{2} / s$ ); $N$ = atomic density (atoms per unit volume). Zyada mobile, zyada crowded atoms → bada atom flux.
$MTTF$ = Mean Time To Failure (time ki units, jaise years) — wire ke EM se fail hone se pehle ki average lifetime. Yahi woh quantity hai jo Black's equation predict karta hai; dekho MTTF and FIT rates.
$n$ = current exponent Black's equation mein (ek unit-less power); $A$ = ek process/geometry constant; $T$ = absolute temperature kelvin mein.

Intuition Force se atom flux tak (kyun Einstein relation appear hota hai)

Atoms pe force tabhi drift cause karta hai jab atoms actually move kar sakein — aur woh kitni easily move karte hain woh unki mobility hai. Einstein relation kehta hai ki mobility aur diffusivity ek hi wandering ke do views hain: $mobility = D / (k_{B} T)$ . Iska matlab hai ki woh thermal jiggle jo atoms ko diffuse hone deta hai ( $D$ ) exactly wahi hai jo ek force ko unhe drag karne deta hai ( $mobility$ ), thermal energy $k_{B} T$ se scale karke. To atomic flux (atoms crossing per area per time) hai: $J_{a t o m} = mobility \times N \frac{N D}{k _{B} T} F$ Takeaway formula nahi balki uski shape hai: flux atom density $N$ ke saath, diffusivity $D$ ke saath (hotter $\Rightarrow$ bada), aur force $F$ ke saath badhta hai — aur $J_{a t o m}$ mein ek divergence (ek jagah jahan woh badalta hai) hi actually mass ko voids aur hillocks mein accumulate karti hai.

Neeche ke teen pictures traps ko ground karte hain: s01 dikhata hai flux divergence, s02 dikhata hai void vs. hillock, aur s03 dikhata hai Blech back-stress. Har figure top corner mein apna label carry karta hai taaki in-text references ("figure s01", etc.) unambiguously point karein.

True or false — justify

Metal atoms cathode ki taraf drift karte hain kyunki electric field positive ions ko wahan kheenchti hai.

False. Good conductors mein electron-wind force dominate karta hai, isliye net atom motion anode ki taraf hoti hai — wohi direction jisme electrons flow karte hain, conventional current

J

ke opposite.

Ek perfectly uniform wire jisme uniform current density ho woh bhi apne middle mein eventually EM se fail ho jaayegi.

Idealized case mein False. Uniform flux ka matlab hai mass-in equals mass-out har jagah (figure s01 ke balanced arrows dekho), isliye kuch bhi accumulate nahi hota — failure ke liye ek flux divergence chahiye (ek via, grain triple-point, width step, ya barrier), jo real wires mein hamesha hoti hai.

Chip ko cool karna hamesha electromigration lifetime extend karta hai.

Hamesha nahi. Arrhenius reliability model term improve hoti hai, lekin Joule heating & self-heating local

T

high rakh sakti hai, aur low temperature Stress migration aur thermomechanical stress ko worsen kar sakta hai jo nayi divergences create karte hain.

Copper interconnects electromigration ke liye effectively immune hain.

False. Copper zyada resistant hai (higher

E_{a}

), immune nahi; yeh surfaces aur interfaces ke saath migrate karta hai (Cu–cap, Cu–barrier), isliye CoWP jaisi cap layers important hain — dekho Copper damascene process.

Black's equation mein exponent

n = 2

ek universal physical constant hai.

False.

n \approx 2

sirf tab hota hai jab void growth life ko limit kare;

n \approx 1

tab hota hai jab void nucleation limit kare. Galat

n

choose karna lifetime ko orders of magnitude misjudge kara sakta hai.

Ek chhhoti enough wire huge current density forever carry kar sakti hai bina EM failure ke.

True — yahi Blech effect hai (figure s03). Jab

J \cdot L

(current density times wire length

L

) ek critical product se neeche hota hai, piled-up atoms ka back-stress electron wind ko cancel kar deta hai void grow hone se pehle.

Black's equation mein exponential

e^{+ E_{a} / k_{B} T}

hai, aur wahan positive sign ka matlab hai ki zyada temperature zyada lambi life deta hai.

False. Yahan

k_{B}

(Boltzmann's constant)

T

ko energy se link karta hai; ek positive exponent jo $T$ badhne par chhhota hota hai ka matlab hai ki Mean Time To Failure (MTTF) temperature ke saath drop karta hai — exactly wahi jo hum expect karte hain, kyunki hotter atoms zyada fast diffuse karte hain (dekho MTTF and FIT rates).

Voids aur hillocks ek hi defect ke do naam hain.

False. Void atom ka depletion hai → ek open circuit (cathode/upstream side); hillock atom ka pile-up hai → ek short neighbour ke saath (anode side) — figure s02 ke do ends.

Spot the error

"Wire ki lifetime double karne ke liye, bas uski width double karo taaki current density half ho jaaye — kyunki MTTF

1/ J

ke proportional hai."

Proportionality

J^{- n}

hai,

J^{- 1}

nahi.

n \approx 2

ke saath,

J

half karna life ko

2^{2} = 4

se multiply karta hai,

2

se nahi — us insaan ne quietly

n = 1

assume kar liya.

"EM current limits aur RC-delay limits unrelated design rules hain, isliye unhe separately optimize karo."

Woh same wire geometry share karte hain. Narrowing a wire

J

raise karta hai (worse EM) aur resistance raise karta hai (worse Interconnect RC delay); ek width choice dono ko ek saath trade karti hai — dekho Design rules & current density limits.

"Kyunki

E = ρ J

, driving force

F = Z^{*} q ρ J

dikhata hai ki force resistivity se aati hai, isliye ek lower-resistivity metal mein hamesha kam EM force hoti hai."

Ion per force

ρ

ke saath drop karta hai, lekin lifetime puri flux law pe depend karti hai jisme diffusivity

D

, atomic density

N

E_{a}

, aur failure geometry shaamil hain. Low

ρ

field term mein help karta hai, phir bhi interface diffusion dominate kar sakta hai (jaise copper mein).

"Kyunki atoms anode ki taraf flow karte hain, hamen pehle hillocks (pile-ups) ki chinta karni chahiye — wahan wire snap open hoti hai."

Ek snap-open ek void hai, jo cathode/upstream side pe banti hai jahan atoms drain away ho jaate hain. Hillocks shorts cause karte hain, ek alag failure mode.

"Effective charge

Z^{*}

literally metal atom ka ionic charge hai."

Nahi —

Z^{*}

ek lumped effective factor hai jo direct field force aur (dominant) electron-wind momentum transfer ko bundle karta hai. Yeh often bada hota hai aur iska sign wind term se set hota hai, actual ionic charge se nahi.

"Blech immortality ka matlab hai ek chhhoti wire mein bilkul bhi EM force nahi hoti."

Force abhi bhi hai; yeh growing back-stress se balance hoti hai void nucleate hone se pehle. Current hatao aur dono relax ho jaate hain — immortality ek equilibrium hai, force ki absence nahi.

Why questions

Har chhhote technology node pe electromigration manage karna kyun mushkil hota jaata hai?

Wires narrow hoti hain jabki current similar rehti hai ya badhti hai, isliye current density $J$ climb karta hai, aur kyunki MTTF

\propto J^{- n}

, lifetime steeply collapse karti hai.

Atomic flux mein ek divergence asli villain kyun hai, flux khud nahi?

Uniform flux har region mein same mass in aur out move karta hai, isliye kuch bhi accumulate nahi hota; sirf jahan flux badalta hai wahan atoms build up (hillock) ya deplete (void) hote hain — figure s01.

Resistivity

ρ

EM driving force mein aata hi kyun hai?

Kyunki designers current density

J

set karte hain, internal field

E

nahi; Ohm's law

E = ρ J

ek ko doosre mein translate karta hai, isliye

ρ

woh conversion factor hai jo

F = Z^{*} q ρ J

J

ke terms mein likhne deta hai.

Black's equation ek Arrhenius temperature term ko

J

ki power se multiply kyun karta hai instead of sirf temperature use karne ke?

Electron wind (current-driven force) set karta hai ki atoms ko kitna zor se push kiya jaata hai, jabki temperature (

k_{B} T

ke zariye) set karta hai ki woh kitni easily diffuse karte hain — EM ko dono chahiye, isliye law ek

J^{- n}

mechanical factor ko

e^{E_{a} / k_{B} T}

thermal factor ke saath combine karta hai.

Vias, grain-boundary triple points, aur diffusion barriers EM hot-spots ke roop mein kyun show up karte hain?

Har ek woh jagah hai jahan atom flux bina change ke continue nahi kar sakta — flow ek rate pe enter karta hai aur doosri rate pe leave karta hai — jo hillocks aur voids ko seed karne wala accumulation ya depletion produce karta hai.

Hum EM guarantee karne ke liye current simply forever kyun nahi lower kar sakte?

Current lower karna performance bhi lower karta hai aur option nahi bhi ho sakta; instead designers current caps, short-line (Blech) routing, temperature control, aur material choices combine karte hain — koi single knob free nahi hai.

Edge cases

Jab current density

J

zero ki taraf approach kare to EM risk ka kya hota hai?

Driving force

F = Z^{*} q ρ J \to 0

, isliye electron wind vanish hoti hai aur EM effectively ruk jaati hai — halaanki residual Stress migration thermal stress se atoms ko bina kisi current ke bhi move kar sakta hai.

Exactly critical Blech product

(J \cdot L)_{cr i t}

pe, wire immortal hai ya fail ho rahi hai?

Yeh marginal boundary pe baith jaata hai — back-stress exactly wind ko balance karta hai (figure s03), isliye theory mein koi net void growth nahi; practice mein designers safely $(J L)_{cr i t}$ se neeche rehte hain kyunki real geometry aur temperature vary karti hai.

Ek jumper itna chhhota hai ki Blech-immortal ho, lekin ek long line se connect karta hai — kya puri path safe hai?

Nahi. Immortality per-segment hai; long line abhi bhi apni divergences pe fail ho sakti hai. Ek chhhota piece sirf apni raksha karta hai, apne neighbours ki nahi.

Agar do identical wires sirf is mein differ karein ki ek

2 0^{\circ} C

zyada hot chalti hai, to kaunsi pehle fail hoti hai aur roughly kitni jaldi?

Hotter wire, kyunki Arrhenius term

e^{E_{a} / k_{B} T}

diffusion ko faster banata hai;

350

K ke aaspaas

E_{a} = 0.7

eV ke saath ratio roughly teen-guna chhhoti life hai.

Jab void nucleation dominate kare (

n \approx 1

), to void-growth regime (

n \approx 2

) ki tulna mein current double karna zyada hurt karta hai ya kam?

Kam —

n = 1

ke saath

J

double karna life half kar deta hai, jabki

n = 2

ke saath yeh life quarter kar deta hai. Chhhota exponent ka matlab hai ki current density nucleation-limited case mein ek weaker lever hai.

Recall Jaane se pehle ek one-line self-test

Is page ke sabse zyaadatar traps ko kaunsa single sentence explain karta hai? ::: "Kuch bhi fail nahi hota jab tak atom flux kahi change na kare, aur electron wind — field nahi — anode ki taraf direction set karta hai." Har void/hillock, Blech, aur copper trap us line ki corollary hai.