Electronics Packaging Society

Chris BaileyPresident

1

Reliability Challenges for the Aerospace Sector and the Use ofCommercial Off‐The‐Shelf Components (COTS)

Chris Bailey

University of Greenwich, London, UK

Computational Mechanics and Reliability Group

• Established in 2004• Staff: 3 Profs, 5 Post Doc’s, 7 PhD’s• Research Mission

To develop and apply CAE technologies to predict the physical behaviour, performance, reliability, and maintainability of complex engineering components/systems.

12

Content

• COTS

• Modelling <-> Metrology

• Refinishing Processes

▪ Hot Solder Dip

▪ Reballing

• Board Assembly & Reliability

▪ QFN’s

▪ uBGA’s

• Conclusions

Commercial off-the-shelf Components (COTS)

• Avionics Systems

• Components

▪ Leaded components

▪ BGA’s

▪ QFN’s

▪ etc

• Use of commercial components for high reliability application

• Lead-free solders still an issue

Reliability Challenges

• Ruggedising components

• Refinishing components

▪ Hot solder dipping

▪ Deballing/Reballing

• Impact of assembly materials

▪ Solders

▪ Conformal coatings

▪ Underfills

▪ Compliant PCB’s

▪ ….

Solder Dipping Deballing/Reballing

Conformal Coatings

Content

• COTS

• Modelling <-> Metrology

• Refinishing Processes

▪ Hot Solder Dip

▪ Reballing

• Board Assembly & Reliability

▪ QFN’s

▪ uBGA’s

• Conclusions

Reliability Predictions:Physics of Failure

17

Modelling

Metrology →Modelling• COTS need to be chacterised (usually limited data available)

• SEM/EDX, CSAM, SEM/EDX, Tomography, etc…

• Data to feed into models – Geometry, Materials Data

Computational Intelligence and Machine Learning for Predictive Modelling

• Similarity-based reliability qualification approach• Optimisation of product qualification processes• Technologies: self-organising maps (SOM), support vector

machines, neural network, state-space models, etc

S. Stoyanov, C. Bailey and T. Tourloukis (2016), Similarity approach for reducing qualification tests of

electronic components, Microelectronics Reliability, Vol. 67, pp. 111-119

S. Stoyanov, C. Bailey, et al (2019) Predictive analytics methodology for smart qualification testing

of electronic components, Journal of Intelligent Manufacturing, Vol. 30 (3), pp. 1497–1514

New PART

Part Characterisation (geometry, materials,

integrity)

Sufficiently similar part?

Part Qualified or Failed

Experimental Qualification Tests

Yes

No

Database

previously

qualified parts

Cluster coordinate X

Clu

ster

co

ord

inat

e Y

Part No.Normalised Attribute Value

a1 a2 a3 a4 a5

Part No. 1 0.427 0.318 1.000 1.000 0.240

Part No. 2 0.284 0.000 0.068 0.615 0.200

………..............................................................................

Part No. 35 0.873 0.373 0.731 0.065 0.832

Part No. 36 0.211 0.263 0.353 0.458 0.524

Similarity Approach to Qualification

Normalised attribute #1

No

rmal

ised

att

ribu

te #

2

Normalised attribute #1

No

rmal

ised

att

ribu

te #

2

Cluster coordinate X

Clu

ster

co

ord

inate

Y

Normalised attribute #1

New Part B

No

rmali

sed

att

ribu

te #

2

Similarity based predictions

Electronic Parts and Data

ProcessSOM Similarity Model Development

Content

• COTS

• Modelling <-> Metrology

• Refinishing Processes

▪ Hot Solder Dip

▪ Reballing

• Board Assembly & Reliability

▪ QFN’s

▪ uBGA’s

• Conclusions

Refinishing (Hot Solder Dip) Process

• Undertaken by Micross Components Ltd

• Double dip hot solder refinishing process

• Robot arm automated

Refinishing components for High-Rel Apps• Remove lead-free finishes (Tin Whiskers)

• Standard: ANSI/GEIA-STD-0006

• Impact of process on reliability

3D-CT Scans

Thermal

Stress/Damage

COTS Package

Temperature (⁰C)

Heat Transfer Process from

Model

Model Validation on Optimised Process

Refinishing Step

Temperature at Die Centre

( ⁰C )

Experimental

Thermo-couple

Readings

Model

Prediction

End of First pre-heat 105.6 125.0

End of First Solder Dip 127.7 128.0

Prior to second pre-heating 121.1 108.3

End of Second pre-heat 118.9 119.4

End of Second Solder Dip 119.6 120.6

End of forced air cooling 87.6 86.7

End of water rinse 65.8 66.1 HSD Normalised Time0 1

S. Stoyanov et al., Modelling methodology for thermal analysis of hot solder dip process, Microelectronics Reliability, 53, 2013, 1055-1067

1

-20

-13.333-6.667

06.667

13.33320

26.66733.333

40

SEP 14 2011

11:26:05

Example of Modelling Stress Response1

-6

7.66721.333

3548.667

62.33376

89.667103.333

117

SEP 2 2011

17:58:32

NODAL SOLUTION

STEP=31

SUB =1

TIME=64

S1 (AVG)

DMX =.039098

SMN =-17.73

SMX =277.488

1

-6.52

7.20120.923

34.64548.367

62.08975.811

89.533103.255

116.977

SEP 2 2011

17:53:24

NODAL SOLUTION

STEP=31

SUB =1

TIME=64

S1 (AVG)

DMX =.009946

SMN =-6.52

SMX =116.977

1

-20

-13.333-6.667

06.667

13.33320

26.66733.333

40

SEP 14 2011

11:27:13

Die Stress

Response

Peel stress at

wire bond

interface

Can we predict delamination?

Model

CSAM

Model

CSAM

Deballing Process and Risk of Damage

Hot Nitrogen Deballing

• Heat dissipates through the top most metal layers of the BGA substrate

• Thermal effects are localised. Very little propagation of the temperature front from the pad towards the die and first-level solder joints

Stoyanov, Dabek and Bailey, Proc. International Spring Seminar on Electronics Technology, Dresden, Germany, 2014, pp. 1-6

Reballing Process and Risk of Damage

Tungsten

Nozzle

BGA

substrate

φ = ID nozzle

Solder

mask

Copper

padBT

core

N2

pressure

Silicon dieBased on PacTechjetting equipment

0

50

100

150

200

250

300

0.1 1 10 100 1000

Tem

pera

ture

(C)

Logscale time of laser re-balling for a single 2nd level joint

(ms)

P1 P2

P3 P4

Laser Re-balling : Local/Global BGA Level

Temperature (C) results from local model

Solidification time window

• Temperature front from laser re-balling does not propagate towards the 1st level solder interconnects (even in the case of direct signal path)

Stoyanov, Dabek and Bailey, Proc. Electronics System-Integration Technology Conference, Helsinki, Finland, 2014, pp. 1-6

Temperature (C) results from global model over 60 ms time interval

Solidification Process during Re-balling

TEMPERATURE

MIN

MAX

TIME

LIQUIDFRACTION

SOLID FRACTION

t = 6.5 ms t = 7.0 ms t = 7.5 ms t = 8.0 ms t = 8.5 ms t = 9.0 ms t = 9.5 ms t = 10.0 ms

t = 13.3 ms t = 16.6 ms t = 20.0 ms t = 23.3 ms t = 26.6 ms t = 30.0 ms t = 33.3 ms t = 36.6 ms

• Solidification of the solder material is extremely fast process

• Solder material of the deposited droplet solidifies within 36 ms

Stoyanov and Bailey, Microelectronics Reliability, Vol. 55, Issue 9-10, 2015, pp. 1271-1279

Tungsten

Nozzle

BGA

substrate

φ = ID nozzle

Solder

mask

Copper

padBT

core

N2

pressure

Silicon die

Content

• COTS

• Modelling <-> Metrology

• Refinishing Processes

▪ Hot Solder Dip

▪ Reballing

• Board Assembly & Reliability

▪ QFN’s

▪ uBGA’s

• Conclusions

Conformal Coatings• Protection to PCB’s

▪ contamination, salt spray, moisture, fungus, dust, and corrosion

▪ Mitigation to tin whiskers

• Materials

▪ Silicone, Urethane, Acrylic

P1 P2 P3 P4 C1

Yin, Stoyanov, Bailey, Stewart, Thermo-mechanical Analysis of Conformally Coated QFNs for High Reliability

Applications, IEEE Transactions CPMT, 9 (11), 2210-2218, 2019

Finite Element Model

Material Stress Free Temperature

Solder 180

Coating A 90

Coating B 22

PCB 170

All others 145

Development of Lifetime Model

Yin, C, Stoyanov, S, Bailey, C, Stewart, P, and McCallum, S, Reliability Assessment of QFN Components for Aerospace Applications,

IEEE 66th Electronic Components and Technology Conference, 1996-2002, 2016

Impact of Conformal Coatings

• Coating A increases damage for all cases

• Mixed situation for Coating B

• Damage reduced in smaller packages (P1, P2, P3, C1),

• Increased in the larger package (P4)

• Complex interaction• Package size, coating penetration & properties

Impact of Edgebond And Conformal Coating

Yin, Stoyanov, Bailey, Stewart, “Modelling the impact of

conformal coating penetration on QFN reliability”, IEEE ICEPT, 1021-1026, 2017

Damage Mechanism

◼ When conformal coating is not used

Shear stress/strain in solder joint due to global CTE mismatch between the PCB and component.

◼ When conformal coating is used, coating presence

◼Reduces CTE miss-match induced shear stress/strain in solder joint (positive role).

◼Constrains the out-of-plane movement in solder joint (negative).

Content

• COTS

• Modelling <-> Metrology

• Refinishing Processes

▪ Hot Solder Dip

▪ Reballing

• Board Assembly & Reliability

▪ QFN’s

▪ uBGA’s

• Conclusions

Micro-BGA Assembly and Scope of Study• mBGA COTS packages (High Rel. Apps)

▪ Small size package architecture

▪ A range of IC functions

▪ Increasingly available at lower cost

• Assembly variants assessment (temperature cycling)

▪ PCB’s: Rigid and Compliant

▪ Resins: Underfill and Edgebond

• All combinations of PCB types and resin optionswere tested and modelled

Critical balls with

redundancy

1

2

34

5

Assembly with underfill

Dual connection

redundancy

Micro-BGAEdgebond

PCB

Assembly Variants• Two multi-layer PCB variants

1. Rigid PCB: an all-rigid stack-up, with the top-most layer (#1) being a “rigid” dielectric material layer

2. Compliant PCB: same PCB stack-up but with a compliant material (lower elastic modulus) top layer

• Three resin variants3. No resin (Edgebond or Underfill)

4. Edgebond - typically penetrating to a line half-way through the second row of joints in the grid array)

5. Underfill - a complete fill of the gap between the package and the PCB

Conformal coating applied after resins applied

Reliability Test Programme: Setup and Results

• Daisy chains

▪ Continuously monitored

▪ Cross-section failure analysis

• Temperature cycle

▪ −25˚C to 100˚C

▪ Ramp times of 10 minutes

▪ 30-minute dwells

• Failure data statistically analysed

Tests Outcome

• Substantial improvement with resins,particularly with underfill

• Compliant PCB advantageous whenresin is used

Finite Element Modelling

• Finite element moles developed for all assembly variants

• Ansys APDL, element type SOLID 185, transient analysis

• Quarter symmetry assumption

FE Models for 6

Assemblies

S. Stoyanov, C. Bailey, P. Stewart, G. Morrison, Reliability Impact of Assembly Materials for Micro-BGA Components

in High Reliability Applications, Proc. IEEE ESTC, 2020, 1-7

Discussions

• Test data and the model results for solder joints damage in good agreement

• The linear regression line represents a powerlaw relation between characteristic life andmodel predictions for solder damage value 1,220 cycles

948 cycles

3,991cycles

8,855 cycles

>10,000 cycles

Life cycles

from Test

Solder Damage from Model

S. Stoyanov, C. Bailey, P. Stewart, G. Morrison, Reliability

Impact of Assembly Materials for Micro-BGA Components in

High Reliability Applications, Proc. IEEE ESTC, 2020, 1-7

Findings: PCB Impact on mBGA Reliability

• The rigid PCB topmost layer (CTE<15 ppm/C) gives better match to the package CTE (mould resin CTE ~7 ppm/C and Si die CTE ~4 ppm/C) compared with the compliant (CTE>20 ppm/C) layerThe better CTE compliance of the rigid material delivers higher reliability than the lower modulus but less CTE-matching compliant material

• But, the compliant PCB provides an improved reliability for assemblies with resinsThe compliant layer has both lower modulus and better CTE match to the CTEs of the resin and the solder compared with the rigid layer

PCB type Impact

Findings: Resin Impact on mBGA Reliability and Solder Joint Damage Mechanisms

• Higher damage in solder joint at the package side interface

• Solder damage (plastic work) due to global, in-plane CTE miss-match between PCB and mBGA. Clear shear deformation and notable movement of solder joints under thermal cycling

• No solder damage drivers at local joint level

• Higher damage in solder joint at the PCB pad side interface

• More uniform plastic work distribution across the joint

• Solder damage (plastic work) due to global, in-plane CTE miss-match between PCB and mBGA greatly reduced. Minimum shear deformation of joints, movement restricted

• Main contribution to solder damage from local level CTE miss-match: PCB resist - resin - solder

• Edgebond achieves similar effect as the underfill but to lesser extent

Plastic work

Deformed shape factor 10 / displacement contours

Plastic work

No Resin With Resin

Solder

Copper

Moulding

Resin

Solder Resist

PCB

Solder

Copper

Moulding

No

ResinSolder Resist

PCB

No Resin

Resin (Edgebond or Underfill)

Conclusions• Ruggedised COTS

• Refinishing Processes

• Assembly and Reliability

▪ Conformal coatings

▪ Edgebond or Underfills

▪ Ridgid or Compliant PCB’s

• Worst option

▪ PCB without additional resins

• Best options:

▪ Compliant PCB and underfill is superior

▪ Compliant PCB with Reworkable Edgebond

Conclusions• Design for Reliability

▪ Modelling <-> Metrology

• AI/Machine will also play significant role in future

https://eps.ieee.org/technology/heterogeneous-integration-roadmap.html

PublicationsS. Stoyanov, C. Bailey, P. Stewart, G. Morrison, Reliability Impact of Assembly Materials for Micro-BGA Components in High Reliability Applications, Proc. IEEE ESTC, 2020, 1-7

S. Stoyanov, C. Bailey, P. Stewart, M. Parker, J.F. Roulston, Experimental and modelling study on delamination risks for refinished electronic packages under hot solder dip loads, IEEE Transactions on CPMT, 10 (3), 2020, 502-515

C. Yin, S. Stoyanov, C. Bailey, P. Stewart, Thermomechanical Analysis of Conformally Coated QFNs for High-Reliability Applications, IEEE Transactions on CPMT, 9 (11), 2019, 2210-2218

S. Stoyanov, M. Ahsan, C. Bailey, T. Wotherspoon, C. Hunt (2019) “Predictive analytics methodology for smart qualification testing of electronic components”, Journal of Intelligent Manufacturing, 30, 1497-1514

S. Stoyanov, C. Bailey and G. Tourloukis, Similarity approach for reducing qualification tests of electronic components, Microelectronics Reliability, 67, 2016, 111–119

C. Yin, S. Stoyanov, C. Bailey, P. Stewart and S. McCallum, Reliability Assessment of QFN Components for Aerospace Applications, Proc. IEEE ECTC, 2016, 1996-2002.

S. Stoyanov, P. Stewart and C. Bailey, Vulnerability Study of Hot Solder Dipped COTS Components, Proc. IEEE ISSE, 2016, 193-198

S. Stoyanov and C. Bailey, Modelling the impact of refinishing processes on COTS components for use in aerospace applications, Microelectronics Reliability, 55, (9-10), 2015, 1271-1279

C. Yin, C. Best, C. Bailey, S. Stoyanov, Statistical Analysis of the Impact of Refinishing Process on Leaded Components, Microelectronics Reliability, 55 (2), 2015, 424-431

S. Stoyanov, G. Tourloukis and C. Bailey, Similarity Based Reliability Qualification of Electronic Components, Proc. IEEE ISSE, 2015, 202-207

S. Stoyanov, A. Dabek and C. Bailey, Thermo-mechanical Impact of Laser-induced Solder Ball Attach Process on Ball Grid Arrays”, Proc. IEEE ESTC, 2014, 1-6

S. Stoyanov, A. Dabek and C. Bailey, Hot Nitrogen Deballing of Ball Grid Arrays, Proc. IEEE ISSE, 2014, 1-6

S. Stoyanov et al., Modelling methodology for thermal analysis of hot solder dip process, Microelectronics Reliability, 53, 2013, 1055-1067

C. Bailey et al., Assessment of Refinishing Processes for Electronic Components in High Reliability Applications, Proc. IEEE EPTC, 2013, 156-161

S. Stoyanov, A. Dabek and C. Bailey, Thermo-mechanical Sub-modelling of BGA Components in PCB Reflow, Proc. IEEE ISSE, 2013, 253-258

S. Stoyanov, C. Best, C. Yin, M. O. Alam, C. Bailey, P. Tollafield, Experimental and Modelling Study on the Effects of Refinishing Lead-Free Microelectronic Components, Proc. IEEE ESTC, 2012, 1-6

C.Y. Yin, C. Best, C. Bailey, S. Stoyanov, M.O. Alam, Statistical analysis of the impacts of refinishing process on the reliability of microelectronics components, Proc. ICEPT-HDP, 2012, 1377-1381

S. Stoyanov, C. Best, M. O. Alam, C. Bailey, P. Tollafield, M. Parker and J. Scott, Modelling and Testing the Impact of Hot Solder Dip Process on Leaded Components, Proc. IEEE ISSE, 2012, 303-308

S. Stoyanov, C. Bailey, P. Tollafield, R. Crawford, M. Parker, J. Scott and J. Roulston, Thermal Modelling and Optimisation of Hot Solder Dip Process, Proc. IEEE EuroSime, 2012, 1-8

Thank youc.bailey@greenwich.ac.uk