Upload
truonglien
View
216
Download
0
Embed Size (px)
Citation preview
François KERISIT
2014 October,16 Rouen, France
Confidential
Tools dedicated to chip access
1
DC in summary • President : Véronique POULAIN • Founder : Michael OBEIN
• Specialization : IC Decapsulation (20+ years)
• R&D : Worldwide
• Manufacturing : FabLess since 2002
2
FA FLOW
Fig. 9
423
Fig. 9
423
Decapsulation is the beginning
of the Flow just after electrical
and visual inspection !!!
Without a good decapsulation :
No analysis.
From Peter JACOB - EMPA
…
3
Why open a component?
Make the chip visible
Front side
Back side
Make bondings accessible
See all connections
Find contaminants
…
4 usual techniques are possible in this day but
depend on purpose Picture property : LaMIP NXP CAEN / DIGIT CONCEPT
4
4 MAIN WAYS CHIP ACCESS
4 MAIN WAYS
MECHANICAL CAN OPENER
MICROABRASIV
E BLASTING
SsameSiPrep
DIP OPENER
LASER
SesameLASER
SesamePLASMA
and PLASER
PLASMA
ACIDS SesameACID
5
4 MAIN WAYS : 1/4
1- LASER INTRODUCTION
OR FIBER LASER
2004 2012
2014
7
LASER Exemples 1/23
Confidential 8
LaMIP Schwindenhammer Patrice
Functional Decapsulation! for FA :
chemical + laser
BGA high pin count exposure
8
LASER Exemples 2/23
Confidential
Goal:
Remove nearly all EMC above or
below the chip (front side or
backside). Around 100µm
remaining EMC thickness allows
better result for SQUID or for
thermograph investigations.
Benefits:
Safe, fast, repeatable.
Very useful when chips are not
chemical compatible.
Preparation for MAGNETIC/THERMAL
9
LASER Exemples 2/23
Confidential
Power Package Goal:
Reach both dies without
separating them, without damage
on Cu wires and without acid leak.
Methodology:
1. LASER cavity.
2. Wet chemical opening at low
temperature.
Benefits:
Safe, fast, repeatable.
Allow low temp wet chemical
opening in a reasonable time.
Decrease acid consumption.
No acid leak.
10
LASER Exemples 4/23
Confidential
Glass cover removal Goal:
Remove easily a glass cover.
Methodology:
1. Vaporization of the glue with
LASER (glass cover is
1064nm fully transparent).
2. Cover removal is then very
easy with a blade.
Benefits:
Safe, fast, repeatable.
No thermal stress.
Necessary to do a sample
cleaning with acetone to remove
projected glue particles.
11
LASER Exemples 5/23
Confidential
Goal:
Observe, remove or replace a
SMD in a SIP.
Methodology:
EMC removal with LASER.
PCB and SMD are not damaged
during LASER process.
Benefits:
Safe, fast, repeatable.
No use of chemicals.
Small size opening allowed.
Passive component removal (SIP)
12
LASER Exemples 6/23
Confidential
Die paddle removal Goal:
Remove rapidly die paddle to have
a backside access.
Methodology:
1. EMC removal all around the die
paddle.
2. Die paddle removal with a
blade.
Benefits:
Safe, fast, repeatable.
No scratch on die backside.
Easy glue removal with acetone.
Not useable on thick die paddle.
13
LASER Exemples 10/23
Confidential 20 LaMIP - Schwindenhammer Patrice 2006-01-23
Cut bond wires
The ~40 µm spot laser is thin enough to precisely perform a surgical cut of a single wire without preliminary de-capsulate the package
Remark : Laminates copper tracks can be cut in a same way
14
LASER Exemples 12/23
Confidential 20 LaMIP - Schwindenhammer Patrice
Decapsulation on Module Silicon Tuner
Decapsulation on board
Decapsulation on Board is done in two steps : - chemistry + laser for packages - laser + chemistry for globe top
15
LASER Exemples 15/23
Confidential
Bond Pads
Expose Individual Bond Pads
Locate Failed Bonds
Find Broken Wires
16
LASER Exemples 17/23
Confidential
Laser Cross-Sectioning Solder Ball Cross-Sections
IC Cross-Sections
17
LASER Exemples 18/23
Confidential 18
LASER Exemples 19/23
Confidential
Gel Removal Down to the Die Remove Gel Compound in seconds rather than
hours and days with Acid Processing
Before Gel Removal After Gel Removal
19
LASER Exemples 20/23
Confidential 20
LASER Exemples 21/23
Confidential
Large Filler Mold Compounds for Copper
We’ve developed a patent pending technique for
removing the newer mold compounds associated
with copper wires and bonds
Large, Difficult to Remove Fillers Our New Process Removes them Easily
21
LASER Exemples 23/23
No other wavelength/pulse duration/pulse shape
was able to process this part
LASER before FIB
Removing Stacked Die
22
4 MAIN WAYS : 2/4
23
MECHANICAL Exemples 1/2
Confidential Pictures property : THALES / DIGIT CONCEPT
Micro-Milling machine equipped with collimator
Package
Chip Coating Die Topside
Back Side Hole
24
MECHANICAL Exemples 2/2
Confidential 25
4 MAIN WAYS : 3/4
26
ACIDES INTRODUCTION
Confidential
Auto–decapper with Peltier element for cooling to 10°C
Redesigned pump for more turbulent flow [4]
27
ACIDES Exemples 1/8
Confidential
LaMIP - Schwindenhammer Patrice
Example of a RF Module - MANUAL die exposure using Silicone polymer protector and Nitric acid (pipet) 100°C on heater.
Manual chemical decapsulation and its help/limitations
silicone
• It is hard only with chemistry to expose all bondings.
• It is also hard using only chemistries to expose BGA
28
Confidential
Part 1: Laser Pre-decapsulation
1st removed EMC above die and stitch
bonds until tops of wire loops exposed
2nd ablation step performed with
exclusion zone above die
Cannot expose die by laser, stitch bonds
no problem
Laser pre-cavity can reduce acid etch
time by half [1]
All samples in temperature study pre-
decapsulated using same auto program
[1] Klein, et.al, 2010 ISTFA Proceedings
Pre-cavity after 1st ablation step above die and stitch areas
Pre-cavity after 2nd ablation step with trench around die
ACIDES ISTFA2011 2/9
29
Confidential
Part 2: Chemical etch Low temperature (10°C – 44°C) to reduce acid attack
Use of monolithich gasket to ensure tight seal
3:1 Nitric / Sulfuric acid mixture
Etch until die and in some cases stitch bonds fully exposed
LEFT: Monolithic rubber gasket cut by laser for QFP
ACIDES ISTFA2011 3/9
30
Confidential
Acetone
Hot water (70°C – 100°C) bath for ~1 minute
Isopropanol
N2 dry
SEM photo of Cu wire cross section after decapsulation by nitric/sulfuric acid mix. In this case the hot water rinse was not used and the resulting crystal growth analyzed by EDX. Evidence of cupric sulfate on surface.
Part 3: Rinses
ACIDES ISTFA2011 4/9
31
Confidential
44°C for 40s
No observable attack or corrosion on the lead finger
coating or die pads, wires very uniform with a
measured diameter of ~19.5mm and no evidence of
thinning or pitting.
Al coating removed from lead fingers, acid attack
visible on Cu wires with measured diameters of
14-19mm.
Temperature study #1
10°C for 270s
Package type: LQFP 64 with Sumikon G631H resin
0.8 mil (~20mm) Cu wires with Al coated Cu lead fingers and Al pads
Acid etch performed at 44°C, 27°C, 15°C, and 10°C
ACIDES ISTFA2011 5/9
32
Confidential
Temperature studies: Summary
Temperature Etch time
TBGA
(8 x 8mm)
QFP
(14 x 14mm)
BGA
(23 x 23mm)
44°C 20s 40s 80s
27°C 60s 100s 120s
15°C 75s 200s 160s
10°C 90s 275s 190s
ACIDES ISTFA2011 9/9
33
ACIDES Exemples 7/8
Confidential
Optical micrograph showing a Cu wire after acid etch using a HNO3:H2SO4 mixture @ +10°C. The coating of the surface of the die (BCB) is without damage.
Chip coating Polyamide and
benzo-cyclo-butene (BCB)
low-k dielectric
Cu traces and BCB
removed
Residual organic
passivation After standard chemical etch at 80°C,
20°C using a 5:1 mixture of HNO3 and H2SO4
3:1 HNO3 to H2SO4 mixture at 10°C for 120s.
34
4 MAIN WAYS : 4/4
35
PLASMA INTRODUCTION
Confidential EDFAS USA, Golden Gate Chapter, Setember 9, 2010
SESAME Technology
incorporates Mutual
Patents / IP developed by
CNES,
DIGIT CONCEPT SA
(WO2008/090281)
36
A : LASER Ablation
• STEP 1 : Place your component on XY table
• STEP 2 : Importation of an X-ray or CSAM image
• STEP 3 : Adjust X and Y with the Table
• STEP 4 : Choose your recipe
• STEP 5 : Start LASER Ablation
PLASMA INTRODUCTION
37
B : LASER/Plasma Ablation
Only with Sesame2000 • Decrease of power LASER
• Increase of power PLASMA
• The hard coating is removed
PLASMA INTRODUCTION
38
Plasma O2/CF4
Plasma for 15mn
Blast 2 seconds
PLASMA INTRODUCTION
39
PLASMA Examples 1/4
Confidential
BCB/Cu Goal:
Reach circuitry without degradation
(BCB dielectric , Cu lines).
Methodology:
1. LASER cavity.
2. Dry chemical (PLASMA) opening.
Benefits:
No degradation due acid use.
EDFAS USA, Golden Gate Chapter, Setember 9, 2010
LAT
STM TOURS
FRANCE
40
LASER decapsulation: Plastic ablation down to passive die backside (Si).
Plasma etching (Ar then SF6): Silicon etching down to passive die backside.
Wet & dry etching, down to underfill.
Wet chemical (HNO3) underfill etching down to active die.
Xrays analysis, zoom on active die: No visible damage on 1st and 2nd interconnects.
LASER / Plasma / Acid …
2D / 3D
41
Papers about these techniques from DC
CONFIDENTIAL
• …
• EPTC 2006
• ANADEF 2006
• ESREF 2007
• ANADEF 2008
• ESREF 2008
• ISTFA 2008
• ANADEF 2010
• IPFA 2010
• ESREF 2011
• ISTFA 2011
• EPTC 2011
• IMAPS 2012
• ISTFA 2014
• …
42
One-shot acid decapsulation :
• Etch time : 1000 s
• Die is fully exposed
• Several rinse baths : acetone, hot water and alcool
• wires preserved, though wire surface is slightly attacked
2:1, 10°C
Copper wire – acid low T°
43
• Electrolysis
– 10V
– 35°C, 2:1
– 3x180s
– multi-step possible
Copper wire - Electrolysis
44
[4] Nisene, US Patent 20130082031
• Dry chemistry parameters
• Total etch time : 3 to 4 hours.
• Final acetone rinse in ultrasonic bath was needed to remove residues on die surface.
Process T (°C) %CF4 %O2
Fast 25 40 60
Slow 25 10 90
50W, 25°C
Ball-bond and wire surface are much more preserved with plasma than with acid etching.
Furthermore, the electrical functionality of the opened devices was also checked before photoemission microscopy analysis, which could be successfully performed.
45
Copper wire – Plasma
46
Ag wire decapsulation
[5] Kerisit et.al, 2014 ISTFA Proceedings
acid 2:1, 10°C plasma 100W, 80°C plasma 50W, 80°C
plasma 50W, 25°C • Total etch time of 200 minutes to completely expose the die
• Very long duration for a single sample
• Step by step procedure with a precise control over the etching.
Results Etch time Electrical influence
Cu acid low T° Good Slow None
Cu acid electrolysis
Good Fast Possible
Cu plasma Good Very slow possible
Ag acid low T° Broken wires Slow None
Ag acid electrolysis
Not tested Fast Possible
Ag plasma Good Very slow Possible
47
Summary
Thank you.
48