3D LOCALIZATION BY LOCK-IN THERMOGRAPHY EUFANET – Toulouse – November 2011 Shermin Danaie - Master1 EEATS at Joseph Fourier University Grenoble Quentin Saulnier – STMicroelectronics Grenoble – Failure Analysis Arnaud Loubaresse – STEricsson Grenoble – Failure Analysis

Purpose Increasing functionalities in mobile phone

System In Package (SIP)

NEW CHALLENGES: ∙ Defect localization in 3D (X,Y,Z) mainly when dices are stacked ∙ Non destructive methods

ONE SOLUTION: ∙ Use lock-in thermography for Z location of a defect

11/25/2011

CONFIDENTIAL

2

Outline • Lock-in thermography principle • Thermography equipment • Trial outline and analysis cases • Modeling for simple cases

• Z measurements with Hamamatsu software • Conclusion & outlook

11/25/2011

CONFIDENTIAL

3

LOCK-IN THERMOGRAPHY PRINCIPLE

Lock-in thermography principle

•Two images (0° & -90°) are converted as below: Amplitude: Phase shift: ∙ Phase shift is about propagation time of the wave between defect and sample surface ∙ Z localization can be theorically found

11/25/2011

CONFIDENTIAL

5

THERMOGRAPHY EQUIPMENT

Thermography equipment ∙ Themos 1000 from Hamamatsu is used: ∙ InSb detector, max efficiency between 3.7µm to 5.2µm

∙ Examples of images for the 3 infrared lenses (Germanium): ∙ CSP product, silicon thickness around 350µm

0.8X, FOV 12mm*9mm

4X, FOV 2.4mm*1.8mm 11/25/2011

15X, FOV 0.64mm*0.48mm CONFIDENTIAL

7

TRIAL OUTLINE & ANALYSIS CASES

Trial outline

•REF unit reverse biased  Diode connected to GND is polarized in direct.

•Reference value for phase shift measurement. •Comparison of phase shift values between REF reverse biased and FAIL units.

11/25/2011

CONFIDENTIAL

9

Analysis case: back to back device Problem description and aim of analysis: bumps

Digital die wire bonding

Analog die

∙ Case 1: Short-circuit on digital signal ∙ Case 2: Short-circuit on analog signal

TRIAL OUTLINE: ∙ Z localization of defect by Lock-in thermography ∙ Physical characterization of defect ∙ Effect of each layers on phase shift 11/25/2011

CONFIDENTIAL

10

Case 1 REF reverse biased, digital die: Frequency (Hz) Phase ( )

10 180

5 124

Failing unit:

1 60

Frequency (Hz) Phase ( )

10 188

5 131

1 68

Hypothesis: Defect is located at digital die level Themos Localization

X-ray Image

Package Layout

Defect is on wire bonding but Z is very close to die 11/25/2011

CONFIDENTIAL

11

Case 2 (1) REF reverse biased, analog die: Frequency (Hz)

10

5

Phase ( )

229

172

Failing unit:

1

Frequency (Hz)

10

5

1

88

Phase ( )

240

180

100

Hypothesis: Defect is located at analog die level Effect of each material on phase shift (F lock-in = 10Hz) 100° (thickness: around 140µm) 50° (Process ~10µm) 0° (thickness: silicon 320µm, glue)

Conclusion:

• Process and resin have a great influence on phase shift • Silicon and glue have a low influence on phase shift 11/25/2011

CONFIDENTIAL

12

Case 2 (2) Checking the hypothesis: Thermography of analog die after polishing

Optical microscope

OBIRCH image

∙ Analog die is broken ∙ Remark: • Phase shift of thermal signal at defect level is not null, it’s around 35° • This is called Initial Phase Shift 11/25/2011

CONFIDENTIAL

13

MODELING FOR SIMPLE CASES

Thermal wave propagation inside a solid • For isotropic and homogeneous material, Phase shift of thermal wave between defect (z=0) and surface (z=l)

z=l

Where z=0

µ (in mm) is called thermal diffusion length, and is depending on: • Thermal conductivity (λ en W/m*K) • Thermal capacity (cp en J/g*K) • Density (ρ en g/cm³) • Lock-in frequency (f en Hz) 11/25/2011

CONFIDENTIAL

15

Phase shift vs Z localization

Phase shift (°)

φ(f lock-in) for resin and silicon:

Silicium Silicon Résine

Resin

Frequency (Hz)

• As seen during trials: • Resin has a great influence on phase shift • Silicon has a low influence on phase shift

11/25/2011

CONFIDENTIAL

16

Checking modeling for Case 1 • Localization of defect is close to resin thickness: Z=140µm • Resin is the only material on top of defect λ (W/m*K)

cp (J/g*K)

ρ (g/cm³)

μ (mm)

0,98

1,43

2.03

0.32

Experimental Theorical shift phase shift (°) phase (°) at f=1Hz at f=1Hz 25

68

But initial phase shift has to be take into consideration, it’s around 35°. Theorical: Φ=25°  Z=140µm Trial: Φ=68-35=33°  Z=184µm

11/25/2011

CONFIDENTIAL

17

Z MEASUREMENTS WITH HAMAMATSU SOFTWARE

Z measurements with Hamamatsu software • Need also to know characteristics below: • Thermal conductivity (λ en W/m*K), Thermal capacity (cp en J/g*K) & Density (ρ en g/cm³) Wire bonded device with 450µm of resin

• Works fine for simple cases • Trials performed on 3D packages but accuracy is not good enough

Measured: 448µm 11/25/2011

CONFIDENTIAL

19

CONCLUSION & OUTLOOK

Conclusion • Knowledge on: • Lock-in frequency influence on measurements

• Low frequencies are suitable for Z localization while high frequencies are better for X/Y localization • Material influence on phase shift • Z localization improvement: • Relation between Z and phase shift is proven • Experimental outline is in place • Modeling and Z measurements are working for simple cases

11/25/2011

CONFIDENTIAL

21

Outlook • To developed: • What about Z localization on others 3D packages like face to face packages? • Need to work on package description, means to know very

well Thermal conductivity, Thermal capacity & Density (ρ en g/cm³) • Create a database with material characteristic? • Keep strong collaboration with Hamamatsu 11/25/2011

CONFIDENTIAL

22

THANK YOU