Transcript
Page 1: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 1

Transient Thermal Response

• Transient Models– Lumped: Tenbroek (1997), Rinaldi

(2001), Lin (2004)– Introduce CTH usually with approximate

Green’s functions; heated volume is a function of time (Joy, 1970)

– Finite-Element methods

( , ) erfc2 2Si

P rT r tk r t

Temperature evolution of a step-heated point source into silicon half-plane (Mautry 1990)

Simplest (~ bulk Si FET)

Instantaneous T rise

E P tTC cV

Due to very sharp heating pulse t ‹‹ V2/3/

2

3/ 2 3/ 20

1 ( ')( , ) exp ' '8 ( ) ( ') 4 ( ')

t P t rT r t dV dtcV t t t t

More general

Temperature evolution anywhere (r,t) due to arbitrary heating function P(0<t’<t) inside volume V (dV’ V) (Joy 1970)

Page 2: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 2

Instantaneous Temperature Rise

• Neglect convection & radiation• Assuming lumped body• Biot = hL/k << 1, internal

resistance and T variation neglected, T(x) = T = const.

Instantaneous T rise

E P tTC cV

Due to very sharp heating pulse t ‹‹ V2/3/

LW d

2

02

''' ( )T P hA c TT Tx k kV k t

1/

Page 3: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 3

Lumped Temperature Decay

• After power input switched off• Assuming lumped body

• RTH = 1/hA

• CTH = cV

• Time constant ~ RTHCTH

T decay

( / )0( ) e hA cV t

HT T T

LW d

T(t=0) = TH

2

02

''' ( )T P hA c TT Tx k kV k t

Page 4: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 4

Electrical and Mechanical Analogy

• Thermal capacitance (C = ρcV) normally spread over the volume of the body

• When Biot << 1 we can lump capacitance into a single “circuit element” (electrical or mechanical analogy)

There are no physical elements analogousto mass or inductance in thermal systems

Page 5: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 5

Transient Edge (Face) Heating

Also http://www.uh.edu/engines/epi1384.htmLienhard book, http://web.mit.edu/lienhard/www/ahtt.html

When is only the surface of a body heated?I.e. when is the depth dimension “infinite”?

2

2

1T Tx t

Note: Only heated surface B.C. is available

erf2

fire

i fire

T T xT T t

Page 6: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 6

• If body is “semi-infinite” there is no length scale on which to build the Biot number

• Replace Biot (αt)1/2

2erf exp erfc2 2

fire

i fire

T TT T

xt

h tk

Note this reduces to previous slide’s simplerexpression (erf only) when h=0!

Transient Heating with Convective B.C.

Page 7: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 7

Transient Lumped Spreading Resistance

• Point source of heat in material with k, c and α = k/c

• Or spherical heat source, outside sphere

• This is OK if we want to roughly approximate transistor as a sphere embedded in material with k, c

Source: Timo Veijola, http://www.aplac.hut.fi/publications/bec-1996-01/bec/bec.html

2

2

1 ( ) 1rT P Tr r k t

( , ) erfc2 2Si

P rT r tk r t

Temperature evolution of a step-heated point source into silicon half-plane (Mautry 1990)

~ Bulk Si FET transient

Characteristic diffusion length LD = (αt)1/2

Page 8: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 8

Transient of a Step-Heated Transistor

In general:

Carslaw and Jaeger (2e, 1986)

“Instantaneously” means short pulse timevs. Si diffusion time (t < LD

2/α) orshort depth vs. Si diffusion length (L < (αt)1/2)

Page 9: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 9

Device Thermal Transients (3D)

Page 10: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 10

Temperature of Pulsed Diode

Holway, TED 27,433 (1980)

Page 11: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 11

Interconnect Reliability

Page 12: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 12

Transient of a Step-Heated Interconnect

When to use “adiabaticapproximation” andwhen to worry aboutheat dissipation intosurrounding oxide

Page 13: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 13

Transient Thermal Failure

Page 14: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 14

Understanding the sqrt(t) Dependence

• Physical = think of the heated volume as it expands ~ (αt)1/2

• Mathematical = erf approximation

Page 15: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 15

Time Scales of Thermal Device Failure

• Three time scales:– “Small” failure times: all heat dissipated within defect, little

heat lost to surrounding ~ adiabatic (ΔT ~ Pt)– Intermediate time: heating up surrounding layer of (αt)1/2 – “Long” failure time ~ steady-state, thermal equilibrium

established: ΔT ~ P*const. = PRTH

Page 16: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 16

Ex: Failure of SiGe HBT and Cu IC

Wunsch-Bell curve of HBT

Page 17: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 17

Ex: Failure of Al/Cu Interconnects

• Fracture due to the expansion of critical volume of molten Al/Cu. (@ 1000 0C)

Metal 4

~ 12 mm Failed Interconnect

AlCu on polymer Intel Corporation

Ju & Goodson, Elec. Dev. Lett. 18, 512 (1997)

00

t = 200 ns

150 ns

100 ns

T along Interconnect

x

T

200 K

Banerjee et al., IRPS 2000

Page 18: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 18

Temperature Rise in ViasVia and interconnect dimensions are not consistent from a heat generation / thermal resistance perspective, leading to hotspots. New model accounts for via conduction and Joule heating and recommends dimensions considering temperature and EM lifetime.

Based on ITRS global lines of a 100 nm technology node(Left: ANSYS simulation. Right: Closed-Form Modeling)

S. Im, K. Banerjee, and K. E. Goodson, IRPS 2002

Metal Line

ILD

Via Hotspot

0

5

10

15

1 2 3 4 5 6 7 8T n

-T0

[o C]

Metal Level (n)

no vias

isothermalvias

vias withJoule heating

Page 19: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 19

Time Scales of Electrothermal Processes

10-12 10-11 10-10 10-9 10-8 10-7

seconds

MechanisticTimescales

StressTimescales

electron & phononrelaxation times

conduction processes within transistor and gate oxide diffusion processes within

interconnects, vias, passivation

clockperiod

gateswitching ESD / EOS Phenomena

-1

123

0 20 40 60 80 100 120ns

I ESD

(A)

CDM

HBM

MM

456

Governs peak transistor temperature & mobility reduction, leakage augmentation

Governs total power consumption, package-level cooling

Governs relative importance of failure modes for ESD

Source: K. Goodson

Page 20: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 20

ESD: Electrostatic Discharge

• High-field damage• High-current damage• Thermal runaway

J. Vinson & J. Liou, Proc. IEEE 86, 2 (1998)

Page 21: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 21

Common ESD Models

Combined, transient,electro-thermal device models

Gate

DrainSourceJ. Vinson & J. Liou, Proc. IEEE 86, 2 (1998)

Lumped: Human-Body Model (HBM)

Lumped: Machine Model (MM)

Page 22: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 22

Reliability

• The Arrhenius Equation: MTF=A*exp(Ea/kBT)

• MTF: mean time to failure at T• A: empirical constant• Ea: activation energy

• kB: Boltzmann’s constant• T: absolute temperature

• Failure mechanisms:• Die metalization (Corrosion, Electromigration, Contact spiking)• Oxide (charge trapping, gate oxide breakdown, hot electrons)• Device (ionic contamination, second breakdown, surface-charge)• Die attach (fracture, thermal breakdown, adhesion fatigue)• Interconnect (wirebond failure, flip-chip joint failure)• Package (cracking, whisker and dendritic growth, lid seal failure)

• Most of the above increase with T (Arrhenius)

• Notable exception: hot electrons are worse at low temperatures

Source: M. Stan

Ea = 1.1 eV

Ea = 0.7 eV

Page 23: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 23

Improved Reliability AnalysisM. Stan (2007), Van der Bosch, IEDM (2006)

( )

0

1( )

failure at EkT t

the dt constkT t

life consumption rate

• There is NO “one size fits all” reliability estimate approach

• Typical reliability lifetime estimates done at worst-case temperature (e.g. 125 oC) which is an OVERDESIGN

• Apply in a “lumped” fashion at the granularity of microarchitecture units

Page 24: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 24

Combined Package Model

Steady-state:

Tj – junction temperature

Tc – case temperature

Ts – heat sink temperature

Ta – ambient temperature

Rtotal(ITRS 2003)

RchipRchip

Rhe

atsi

nk

0

0.1

0.2

0.3 oC/W

2000 2005 2010 2015 2020

3DIC3DIC

MulticoreMulticore

Page 25: Transient  Thermal  Response

© 2010 Eric Pop, UIUC ECE 598EP: Hot Chips 25

Thermal Design Summary

• Temperature affects performance, power, and reliability

• Architecture-level: conduction only– Very crude approximation of convection as equivalent resistance– Convection, in general: too complicated, need CFD!

• Use compact models for package• Power density is key• Temporal, spatial variation are key• Hot spots drive thermal design


Top Related