3D Chip Stacks Novel Thermal Interface Materials - IEEEewh.ieee.org/soc/cpmt/presentations/cpmt1203a.pdf · 3D Chip Stacks Novel Thermal Interface Materials ... Prof. Kaustav Banerjee

IEEE SF Bay Area Chapter, Components, Packaging and Manufacturing Technology Chapter

March 14, 2012

www.cpmt.org/scv/

[email protected]://www.nanoheat.stanford.edu

3D Chip StacksNovel Thermal Interface Materials

Srilakshmi [email protected]

PI: Prof. Ken GoodsonDepartment of Mechanical Engineering

IEEE CPMT SocietyMarch 14, 2012

Outline

• Stanford Nano/Micro Heat Lab

– Overview of Metrology and Materials

• Materials for Thermal Management

– Aligned CNT nanotape

– High density aligned CNT composites

– Mechanical Characterization

– 3D chip attachments and conductive underfills

IEEE Components Packaging and Manufacturing Technology Society 2


March 14, 2012

www.cpmt.org/scv/

Nano Heat Transfer Lab


Alumni

Prof. Dan Fletcher UC Berkeley Dr. Jeremy Rowlette Daylight Solns

Prof. Evelyn Wang MIT Dr. Patricia Gharagozloo Sandia LabsProf. Katsuo Kurabayashi U. Michigan Dr. Per Sverdrup IntelProf. Sungtaek Ju UCLA Dr. Chen Fang Exxon-MobileProf. Mehdi Asheghi Stanford Dr. Milnes David IBMProf. Bill King UIUC Dr. Max Touzelbaev AMDProf. Eric Pop UIUC Dr. Roger Flynn IntelProf. Sanjiv Sinha UIUC Dr. Julie Steinbrenner Xerox ParcProf. Xeujiao Hu Wuhan Univ. Dr. John Reifenberg Alphabet EnergyProf. Carlos Hidrovo UT Austin Dr. David Fogg CreareProf. Kaustav Banerjee UCSB Dr. Matthew Panzer KLA-TencorProf. Sarah Parikh Foothill CollegeProf. Ankur Jain UT Arlington

Current Group

Josef Miler Elah Bozorg-Grayeli Woosung ParkYuan Gao Amy MarconnetJaeho Lee Shilpi Roy (EE)Sri Lingamneni Michael Barako Prof. Mehdi AsheghiSaniya Leblanc Lewis Hom Dr. Yoonjin WonJungwan Cho Zijian Li Dr. Takashi Kodama

Thermal and Mechanical Characterization


Cross-sectional IR Microscopy

Heater

growth substrate

CNT Film

0 200 400 60015

20

25

30

35

40

Distance (um)

Tem

per

atu

re (

C)

CNT

kdx

dT 1

BRT

dx

dTk

1

500 μm

Distance (μm)

Pico/Nanosecond Thermoreflectance

Time

Re

fle

cte

d S

ign

al

TR SignalHeating Pulse

CNT Film

Metal

Metal CNT Catalyst

TransparentSubstrate

Mechanical Characterization

CNT Film

Photodetector Laser Beam

Microresonators (In-Plane) Nanoindentation (Out-of-Plane)

Laser Doppler Vibrometer

(LDV)

Electrothermal Characterization


March 14, 2012

www.cpmt.org/scv/

Metrology


Sample Complexity

Rig

Co

mp

lexi

ty SAMPLE FILM

Kaeding, Skurk, Goodson, Applied Physics Letters (1993)

Pump-probe optics

Hybrid Optical-Electrical Methods

Multi-property Measurements

Nanodevices and Materials


Data Storage

Thermoelectrics

C. Xi, C-M HsuCui group, Stanford

Tilke et al., LMU Munich

Nanotransistors

Numonyx/Intel

Solar Electrodes

R. Noriega and S. PhadkeSalleo group, Stanford

Vuckovic et al., Stanford

CMOS Lasers

Optical Nanocrystals

R. KekatpureBrongersma group, Stanford

Thermal Interfaces

Goodson group, Stanford


March 14, 2012

www.cpmt.org/scv/

Nanothermal Impact


…on Performance

…o

n R

elia

bili

ty

nano-transistors

high-densitynonvolatile memory

RF electronics

3D chip integration

opticalinterconnects

rapid PCR &blood analysis

thermoelectricenergy harvesting

Phase Change Memory


SiO2

PCMW W

Current (A)V GND

Thermal Characterization

Electrothermal/Crystallization Modeling

John ReifenbergZijian Lee

ThermoElectric Effect (Seebeck) Threshold Switching Phenomena

Novel Geometries, Synthesis, & Multibit

Groups of H.S. Philip Wong (EE) and Kenneth E. Goodson (ME)

Intel (D. Kau, K-W. Chang, Ilya V Karpov, G. Spandini)

NXP (F. Hurckx), Micron (John Smythe), IBM (Raoux, Krebs, et al.)

National Science Foundation (NSF), Semiconductor Research Corporation (SRC)

Sponsors & Collaborators:

Metal heater

PCInsulator

Jaeho LeeZijian LiElah Bozorg-GrayeliSangBum KimJohn Reifenberg

MicroThermal Stage (MTS)

SangBum KimRakesh Gnana,

John Reifenberg, Jaeho Lee,

Zijian Li

Jaeho Lee, Rakesh Gnana,

Zijian Li

Sample film (CNT Array)

Heat sink (silicon or metal substrate)

Metal coating

Nanosecond Heating from Nd:YAG

ToDetector

CW Radiation for

Thermometry

GST

Crystalline Amorphous

Yuan ZhangRakesh GnanaSangbum KimJohn Reifenberg


March 14, 2012

www.cpmt.org/scv/

Key Challenges for TEs in Combustion Systems


…Low resistance interfaces that are stable under thermal cycling.

…High-temperature TE materials that are stable and promise low-cost scaleup.

…Characterization methods that include interfaces and correlate better with system performance.

hot side / heat exchanger

insulation / e.g. ceramic plate

conductor/ e.g. copper


cold side / heat exchanger

conductor/ e.g. copper conductor/ e.g. copper

n p

thermalthermal

electrical& thermal

electrical& thermalthermalthermal

diffusion barrier

joining technology

contact resistanceHeat

Cooling

hot side / heat exchanger


conductor/ e.g. copper


cold side / heat exchanger

conductor/ e.g. copper conductor/ e.g. copper

n p

thermalthermal

electrical& thermal

electrical& thermalthermalthermal

diffusion barrier

joining technology

contact resistanceHeat

Cooling

600 °C

90 °C

Improvements in the intrinsic ZT of TE materials are proving to be very difficult to translate into efficient, reliable power recovery systems.

Major needs include…

Automotive Waste Heat RecoveryThermoelectric Modules and Electro-Thermo Interfaces


TE Material

Hot Side Cold Side

IR Camera

250 W

Michael Barako, Lewis Hom, Saniya Leblanc, Yuan Gao, Woosunng Park, Amir Aminfar, Amy Marconnet, Dr. Mehdi Asheghi

Solder Bonding High Temperature IR Imaging

Thermal Cycling of TE Modules

50 μm

Fracture

After 45,000 Cycles

1010

010

110

210

310

410

5

0.5

0.55

0.6

Cycles

ZT

Figure of Merit

~ 400 oC drop across TE module


March 14, 2012

www.cpmt.org/scv/

Thermal Management Challenges for Microprocessors


Importance of Hotspot Thermal Management• Peak temperatures limit the reliability of

interconnects• Thermo-mechanical strains due to temperature non-

uniformities can degrade packaging and interfaces

Challenges Ahead for Thermal Management• Transistor scaling• Increasing number of cores• 3D integration• Constraints of mobile applications

Adapted from: http://en.wikipedia.org/wiki/Heat_sink

Images of Electromigration FailureRalph Group, Cornell University

Solder joint failure due to thermomechanical stressRidgetop Group Inc.,

Flow = 2 ml/min, Mass Flux = 103 kg/s-m2

0

5

10

15

20

25

0 10 20 30 40 50 60 70

Heat Flux [W/cm2]

P

tot /

P

tot,l

o

Control Vent

Heat Flux [W/cm2]

∆P

tot/∆

Pto

t,liq

Hot Spot Detection and Thermal Management


(b)

Distributed Temperature

Sensor Network

(c)

Reproduced power map

(a)

Power map

Rapid Hotspot Prediction & Power Distribution

Non uniform heat generating chip

Base

IR transparent window

Imaging lens

To detector

IR radiation

IR transparent fluid

(a)

Through-Wafer Hotspot Imaging

Milnes David, Joe Miler, Lewis Hom, Dr. Mehdi Asheghi

To resolve transient chip hotspots with increased accuracy and cool them with high-heat flux cooling solutions

Vapor-Venting Microfluidic Heat Exchangers

David et al.,


March 14, 2012

www.cpmt.org/scv/

Outline

• Stanford Nano/Micro Heat Lab

– Overview of Metrology and Materials

• Materials for Thermal Management

– Aligned CNT nanotape

– High density aligned CNT composites

– Mechanical Characterization – Resonator

– 3D chip attachments and conductive underfills


3D chips: Material Requirements


DRAM DRAM

lnterposer

Logic

Chip Carrier

LogicMemoryMemory

Chip Carrier

TIM 1 &2 (metal alloys, particle filled organics, aligned CNT films)- High thermal conductivity- Mechanical compliance

Flowable Underfill- Electrically insulating- Mechanical stiffness- Viscosity and capillary forces- High thermal conductivity

3D Chip Attachment (Adhesives, Thermal compression bonding)- High thermal conductivity- Electrically insulating- Thermal cycling stability

Encapsulation- High thermal conductivity- Electrically insulating (on the side facing the chip)- Mechanical compliance

Goal: Discover and characterize advanced materials containing nanoscale inclusions (particles, platelets, tubes), targeting the unique property needs of packaging applications including TSV, interposer, and 3D


March 14, 2012

www.cpmt.org/scv/

Challenges Posed by Material Requirements


Problem: Heat removal from 3D stacked circuits is severely limited by low conductivities of the underfill/BGA and the thermal interface.

Solution: Nanostructured organics promise high thermal conductivity with the mechanical and viscoelastic properties required for manufacturing and lifetime reliability.

Application:

• Passivation layer during the BEP of the wafer

• Non-conductive adhesive as a replacement to under-fill

Challenges:

• Maintaining nanostructure during dispensing and CMP

• Minimizing negative impact from contacting the nanostructure to

the adjacent materials

• Impact of thermal pressure bonding during assembly process

• Thermo-mechanical stresses between the polymer and the micro-bump/Cu pillar

• Effects of reflow process

LogicMemoryMemory

Chip Carrier

IBM and 3M collaboration


High thermal conductivity underfill adhesives for building silicon skyscrapers


March 14, 2012

www.cpmt.org/scv/

Thermal Interface Materials


Courtesy of Dr. Boris Kozinsky (Bosch)

Thermo-mechanical Stresses in TEs

Goal: Engineer materials with high thermal conductivityand low elastic modulus

GOAL

Thermal Resistivity (m K / W)1.00.1

Elas

tic

Mo

du

lus

(MP

a)

Greases & Gels

Phase Change Materials

Indium/ Solders Adhesives

0.01

Nano-gels

104

103

102

101

Latest Stanford CNT Data1

CNT Die Attachment


Carbon 2012

LogicMemoryMemory

Chip Carrier

Primary ThermalInterface

2010


March 14, 2012

www.cpmt.org/scv/

Nanotape to Replace Solder Pads


SRC Patent: Hu, Jiang, Goodson, US Patent 7,504,453, issued 2009SRC Patent: Panzer, Goodson, et al., 2009/0068387 (pending)Panzer, Maruyama, Goodson et al., Nanoletters (2010) Hu, Fisher, Goodson et al., J. Heat Transfer (2006)

Removable mechanical backer

Nanofibers

Low melting temperature binder (e.g. alloys of Ga, In, Sn)

Adhesion layer

Adhesion layer wets nanotubes and promotes adhesion of binder (Pd, Pt, or Ti).

~100 nm is the typical variation in CNT height.

Upon heating, the low melting binder conforms to CNT and substrate topography.

Mechanical Properties of CNT Films New Technique: Mechanical Resonators


Resonator length and shape variation

• LDV (laser Doppler velocimetry) experimental setup : resonant frequency of various thickness films. • Resonant frequency shift : mechanical modulus• Ring-down and fitting measurements : quality factors

Thermal and Mechanical Characterization

CNT on a Cantilever

Experimental Setup


March 14, 2012

www.cpmt.org/scv/

Experimental Method and Data Interpretation


Transformed section method

wn

wn,o

ESiISi EcntIcnt

SiASi cnt Acnt

SiASi

ESiISi,o

1

ECNT Esi

ICNT

1 CNTACNT

SiASi

1

wn

wn, Si, 0

2

ISi, 0 ISi

i

ii AA i

ii IEEI

Euler-Bernoulli differential equation for multi-layer beam

04

4

2

2

x

wEI

dt

wA

Polysilicon deposition

Resonator outline etching

Catalyst deposition

Carbon Nanotube Growth

Resonator etching

Oxide layer removal

Won et al, Carbon (2011)

Beam thickness, tSi

Carbon nanotube thickness, tcnt

Beam Width, b (25-100 um)

Beam Length, L (200-1000 um)

Si-CNT Cantilever

Mechanical Behavior of CNT Films


MWCNTs (Monano)

SWCNTs (U. of Tokyo)

Thickness: 0.5 - 220 µmModulus : 1 - 370 MPaDensity : 25 - 140 kg/m3

MWCNTs (MIT)


March 14, 2012

www.cpmt.org/scv/

Mechanical Behavior of Nonhomogeneous CNT Films


Three-layer Analysis

• Interweaving of a thin layer of entangled and randomly oriented nanotubes• Vertical-aligned growth• Density decay

Growth Stages

crust

middle

DimensionlessEffective Modulus

SWCNTs (U. of Tokyo)

MWCNTs (Monano)

MWCNTs (MIT)

Two-layer analysis Three-layer analysis

Stage 1 Stage 2

Self-organization Vertical-growth

crust middle

SWCNT films

Theoretical curve for MWCNT films

wn

wn,Si,0

ESiISi,1 EMiddleIMiddle ETopITop

SiASi Middle AMiddle Top ATop

SiASi

ESiISi,0

1

: Moduli are scaled by reference sample and volume fraction

Aligned CNT Nanocomposites


Remove 1 vol.% CNT film from growth substrate

Biaxial compression to increase volume fraction

Infiltrate with epoxy (RTM-6)

Axi

al

Transverse

1 m

1 vol.% CNT film

2 m

1 vol.% CNT Composite

Marconnet et al, ACS Nano (2011)


March 14, 2012

www.cpmt.org/scv/

Thermal Conductivity of Aligned CNT Nanocomposites


0.01.0

5.0

10.0

15.0

20.0

Th

erm

al C

on

du

ctiv

ity

En

han

cem

ent

[k/k

epo

xy]

0% 5% 10% 15% 20%0.0

1.0

2.0

3.0

4.0

5.0

Th

ermal C

on

du

ctivity [W/m

/K]

Volume Fraction

Composites - AxialComposites - TransverseUnfilled Arrays - Axial

(Surfaces polished to ~3nm roughness and coated with 200 nm of Pt to ensure nearly identical contact conditions for all samples and improves contact between composite and reference layers)

•Non-linear increase at higher volume fraction suggests that CNT-CNT and CNT-polymer interactions are important to the thermal transport•CNT’s contribute at a rate of about 10 W/m/K per CNT at low volume fractions, much lower than expected.

Effective medium approachPower law

Ax

ial

Transverse

Aligned CNT Nanocomposites


Heat Sink

Heat Source

Variation in CNT Heights

CNT-CNT Contacts

Defects

Voids

CNT-Epoxy Boundary Resistance

Composite Interface Resistance

Individual CNT Thermal Conductance

Spatially Varying Alignment

He

at S

ink

He

at S

ou

rce

CNT-CNT Contacts

CNT-Epoxy Boundary Resistance

Composite Interface Resistance

Spatially Varying Alignment

Heat SinkIndividual CNT Thermal Conductance

Transverse Conduction

Axial Conduction


March 14, 2012

www.cpmt.org/scv/

Nanoparticle Based Composite Packaging


Cu - Eastman et al, 2001; Carbon Nanofibers - Sui et al, 2008; CuO – Karthikeyan et al, 2008; Fe3O4 - Philip et al, 2008; SiC – Zhou et al, 2008; xGNP –Fukushima et al, 2006 and Yu et al, 2007; hBN – Hsuan et al, 2006; hBN+cBN – Yung et al, 2007; hBN, Daimond, Silica – Lee et al, 2005; BN+MWCNT –

Teng et al, 2011; Al2O3 – Fu et al, 2011 and Lee et al, 2005; 3D AlN – Shi et al, 2009; AlN Particles + Whiskers – Xu et al, 2001.

Aligned CNTs

40% Ni + 2% MWCNT

30% Ni +2.3% MWCNT

CNTs

30% BN + 1% MWCNT

AlN Particles

+ Whiskers

hBN

hBN +cBN

Al2O3

SiC 3D AlN

Whiskers

Al2O3

Silica

hBN

xGNP

Fe3O4

xGNP

Carbon

Nanofibers

Ni

Ni

CuO

Cu

xGNP

Diamond

1

10

100

0 10 20 30 40 50 60 70

Ther

ma

l Co

nd

uct

ivit

y E

nh

an

cem

en

t [k/

k mat

rix]

Filler Concentration (% Vol)

CNTs

Ceramics

OTHERS

Goodson group data

Key Notes from Literature – Percolation Pathways


Fe3O4 - Philip et al, 2008; Ni - Hu et al, 2004; BN - Yung et al, 2007; AlN - Xu et al, 2001.

•Application of magnetic field in the direction parallel to the temperature gradient

•CNT inclusions to form conductive pathways

•Multimodal particle size mixing

•Combination of different crystalline forms

•Distribution in particle shapes to form better networks


March 14, 2012

www.cpmt.org/scv/

Key Notes from Literature – Chemical Treatment


SiC – Zhou et al, 2008; hBN - Yung et al, 2007.

Silane surface treatment of particles to form particle-resin interface structures

Candidate Filler Micro/Nano Particles


2 μm

(a) AlN – 10 µm

2 μm

(b) BN – 1 µm (c) SiC – 100 nm

2 μm

1 μm

(d) xGNPs – 25 µm; 5-15 nm

500 nm

(e) MWCNTs – 6-13 nm OD; 2.5-20 µm LMaterial

Thermal Conductivity

(W/mK)

Electrical Resistivity

(Ω-cm)

Aluminum Nitride

285 (single)70–210 (poly)

> 1014

h-Boron Nitride

600; 30 > 1014

Silicon Carbide

120 102–106

xGNP 3000; 6 10-5 ; 1

MWCNTs 100-3000


March 14, 2012

www.cpmt.org/scv/

Preliminary Thermal Conductivity Data


Samples: particles dispersed in silicone oilTechnique : IR imaging

•1% xGNP performed better than 1% MWCNT

• 1% xGNP and 1% MWCNT performed better that 10% AlN

• 1% xGNP addition to 10% AlN showed promising enhancement

Particle Composition

(Vol %)

Thermal Conductivity

Enhancement (k/kmatrix)

1 % xGNP 2.1

1% MWCNT 1.5

10% AlN 1.3

10% AlN + 1% xGNP

2.8

Ongoing Work and Future Directions

• Ongoing work– Cured polymer nanocomposites– Various surface treatments of filler particles– Electrical characterization of polymer nanocomposites– Compare thermal and electrical conductivity data against

existing effective medium theories

• Future directions– Bonding of composite materials to Si– CTE measurement of composite materials bonded to

mechanical resonator– Real-time evolution of interfacial adhesion, fatigue,

debonding



March 14, 2012

www.cpmt.org/scv/

Conclusions - Materials for Thermal Management

– Development of novel thermal interface materials is crucial for 3D circuits performance

– Nano tape is a promising replacement to solder pads

– Measurements of aligned CNT films and composites showed thermal conductivity and elastic modulus comparable to or better than commercial TIMs

– Preliminary thermal conductivity data of nanosuspensions in silicone oil showed promising trends, and this work is being extended to nanocomposites


Documents

3D Chip Stacks Novel Thermal Interface Materials - IEEEewh.ieee.org/soc/cpmt/presentations/cpmt1203a.pdf · 3D Chip Stacks Novel Thermal Interface Materials ... Prof. Kaustav Banerjee