pcb materials behaviours - FED-Wiki

Transcription

PCB materials behaviour towards humidity and baking
impact on wettability.
Aquaboard project. Part 2-a, 2-b and 3-a
Walter Horaud, Vincent Vallat
Solectron, Design & Technology Center
Bordeaux, France
[email protected]
Sylvain Leroux
Dominique Navarro, Jean-Yves Delétage
ACB/Atlantec
Dendermonde, Belgium/Malville, France
IXL
Bordeaux, France
AFEIT, brasage 2003
September 2003
Abstract:
As Electrostatic discharge, humidity can have a bad impact on assembly quality. It requires environmental conditions and
process controls but also risks knowledge. To overcome humidity issue, parts (PCB or components) need to be baked to
remove moisture. Baking drawback is the wettability issue especially with thermal sensitive PCB finishes. This wettability
issue can introduce reliability defects. Solectron has launched on this topic a project called Aquaboard. This project mainly
deals with the PCB materials behaviours towards humidity and the impact of the baking on the solder joint reliability
according to the PCB finish, and the assembly process. Other tests have been implemented on the test vehicle such as
design for reliability. A brief sum-up of those tests can be found in the introduction. The project took place in three steps,
this paper deals with the part 2-a&b and 3-a. The parts 2-a&b are about the baking and storage efficiency. Several curves
have been drawn for most of the PCB materials (paper, E-glass, aramid, phenol, epoxy, BT, polyimide, hydrocarbon…):
baking curves (from 80° to 120°C), absorption curves (ambient atmosphere, <5% HR, dry pack storage, 85°C-85%RH).
The part 3-a is the impact of baking conditions on the wettability and solder paste spreading.
Key words:
Moisture, baking, absorption, Fick's law, diffusion coefficient, wettability spreading, PCB finish, ENIG, OSP , HASL, Im
Sn, Im Ag, PCB material, paper, E-glass, aramid, phenol, epoxy, BT, polyimide, hydrocarbon
absorption rate? Most of the PCB materials assembled at
Solectron have been studied (paper, E-glass, aramid,
ceramic, phenol, epoxy, BT, polyimide, hydrocarbon…).
For each material the absorption curve has been drawn for
the following storage conditions: dry pack (<2%RH), dry
cabinet (<5%RH), ambient air (23+/-2°C, 45+/-5%RH),
high humidity storage (85°C, 85%RH). The third study of
this part is to see the impact of humidity on assembly
quality. Humidity can be the root cause of delamination,
voids, and micro-balls... Unfortunately those defects can be
caused by other parameters such as a bad lamination, an
inappropriate reflow profile. Furthermore those defects are
difficult to identify and occasional (PPM). Too much
material investment was needed for this study so we only
focused on delamination. PCB crammed with a controlled
humidity amount went three times though reflow or wave
soldering in order to see from which level of humidity,
delamination is observed. The forth study of this part is to
finalise the baking conditions according to the percentage
of copper in the PCB (ground layer), the PCB thickness,
and the PCB finish (following the results of the third step of
the study). The fifth study is about the absorption curves
during the assembly according to the process (no-clean or
clean).
INTRODUCTION
Following bad experiences due to inappropriate baking
(delamination or poor wettability), we have decided to
launch a project to better know the material behaviours
towards humidity and to see the impact of baking on solder
joint reliability. This project called Aquaboard took place in
three steps.
l The first one is focused on literature research on the two
main topics (humidity and baking).
l The second one is dedicated to materials behaviours
towards humidity. It is mainly based on weighting PCB
during baking and storage. This second step is made of
several studies. The first one is to measure the efficiency of
the baking. What is the impact of stacking PCB? What are
the required times at different temperatures to have the
same baking efficiency? This first study was performed
mainly on E-glass epoxy and E-glass polyimide as they
represent the major part of PCB materials assembled at
Solectron. Seven PCB and five baking conditions (80°C,
90°C, 100°C, 110°C, 120°C) have been considered The
second study of this part is based on PCB storage.
According to the storage conditions, how long does need
the material to absorb water and how long is the allowed
time prior reflow? Have the baking conditions an impact on
1
Most of the results on the first and second study of this
second step will be discussed in this paper.
l The third step of the study is a test vehicle to see the
impact of baking on solder joint reliability according to the
PCB finish, the baking conditions and the assembly
processes. Daisy chain components such as 0.4 mm pitch
QFP, 1.5 mm pitch BGA and 0.5 mm CSP have been
assembled. Five PCB finishes (HASL, ENIG, Im Sn, Im
Ag, OSP), four baking conditions (from 80°C to 120°C),
three assembly processes (clean, no-clean, lead free), two
reflow atmospheres (air, nitrogen) have been studied.
Several tests have been performed on the assembled boards:
continuity test (daisy chain components) through thermal
cycling (-55°C, +125°C) and mechanical shock, shear test
performed on chips, wetting balance (raw PCB, steam aged
PCB, baked PCB…), spreading test, hole filling (wave
soldering), 3D X-rays, cross section & SEM, surface
insulation resistance. In addition to this study, design for
reliability has been carried out. On those test vehicles,
several BGA design have been evaluated (1.5 mm pitch
BGA and 0.5 mm pitch CSP): copper defined pad versus
solder mask defined pad, dog bone versus microvia in pad
design (centred, off-set and tear drop), pad shape (round,
oblong, square…). In the same way, designs for quality and
reliability have been performed on chips: 0603 and 0805
packages, resistor and capacitor components. For each case,
five PCB designs times five stencil apertures have been
tested.
The wetting balance and spreading test results will be
discussed in this paper. Only the impact of the baking will
be tackled as the study of the impact of the assembly
processes (clean, no-clean, lead free), and the reflow
atmospheres (air, nitrogen) are still on progress.
Depending on the baking temperature, we can see on this
chart that after 5 hours the baking is close to 65-80 %
efficiency of the 25 hours baking.
Forced absorption has also been studied. The following
graph compares E-Glass/FR-4 with Thermount® in an
85°C-85%RH environment.
We can see that after 225 hours at 85°C/85%RH the
Thermount® threshold is around 20000 PPM when the FR-4
threshold is close to 7000 PPM.
Fick's law8-14:
Assuming a rectangular plate is taken to be infinitely long
in the y- and z- directions, the moisture content inside the
plate varies only in the x axis. Initially the moisture
concentration ci inside the plate is uniform. The plate is
suddenly exposed to a moist environment and the exposed
faces instantaneously reach the equilibrium moisture
concentration ca, which remains constant. The moisture
uptake through the thickness of an infinite plate is given by
the following simple model Fick relationship:
Materials behaviour1-2:
Concerning the materials behaviours towards humidity, the
literature research showed that a lot of data from PCB
material suppliers are available. Unfortunately most of
those data are about raw materials and not about the PCB
itself. Obviously, it was easier to find documents about
moisture sensitive PCB material such as polyimide than
high frequency material that are usually not sensitive to
moisture (i.e. PTFE). If we focus on polyimide material
such as Polyimide/Thermount®, bake PCB for a minimum
of 4 hours at 112°C, or a maximum of 6 hours at 136°C is
mandatory. Thicker PCB or PCB with external copper
planes should be baked for 6 hours at 136°C. In general,
when a PCB contains over 50% of Thermount®
reinforcement, the maximum allowable moisture regain by
weight is 2800 PPM to assure reliable assembly. The
following chart shows the moisture amount removed from a
ten layers PCB.
The solution of this Fick equation can be calculate if the
initial concentration and the space limits are known.
This solution gives a Gauss shape curve showing the
moisture concentration diffusing through the x axis at a
given time.
0,25
C(x,t)
0,2
t1
t2
t3
0,15
C
0,1
0,05
0
-300
-200
-100
0
100
200
300
-0,05
x
At temperatures well below the Tg of the conditioned
material, water absorption of most polymers correlates with
Fick's laws. The diffusion coefficient, independent of time
and moisture concentration can be calculated from the
Fickian diffusion curve. The Fickian diffusion curve is the
percentage uptake of water by weight in function of square
root of time.
2
W − W Dry
W Dry
Intermetallics:
When solder becomes molten the immersion metals
instantly dissolve into the solder. At 232.2°C, gold
dissolves at a rate 3µms-1 and silver dissolves at 1.11µms-1.
Once copper circuitry is exposed, thin and copper begin to
form an intermetallic phase that joins the metals. Then the
copper-tin intermetallic interface is the same than the one
formed with HASL and OSP finishes.
Drawbacks:
Reliability drawbacks are well documented, while ENIG is
well known to form nickel-tin intermetallic when black line
nickel does not act as a killjoy, silver forms water soluble
salts when exposed to condensing moisture and electrical
bias, and tin is prone to whiskering. The tendency for pure
silver to migrate increases with increasing thickness, while
lower thickness are more likely to whisker. This explains
the standard thickness of silver below 0.5µm and that of tin
at about 1.0µm. Finally tin copper intermetallic is formed at
room temperature so sufficient tin thickness is necessary to
guarantee efficient storage.
Wettability/spreading:
Silver is one of the most wettable metals with eutectic
Sn/Pb soldering due to mainly the quick dissolution of
silver at soldering temperature.
Wetting balance studies with Sn/Pb alloy seemed to show
that silver remains efficient towards baking while
immersion tin suffers from thermal degradation. Spreading
studies seemed to be more profitable for ENIG. Immersion
tin and immersion silver have nearly the same spreading
properties. Solderability of immersion silver is relatively
insensitive to storage at 85°C/85%RH conditions but
depending on the type and thickness of the immersion
silver.
Wettability and spreading with Sn/Ag/Cu alloy are also
well documented and trends to show that immersion tin and
ENIG surfaces provided the best wetting results on fresh
boards follow by immersion silver and OSP. While
immersion silver can withstand multiple lead free reflow,
immersion tin cannot withstand multiple lead free reflow
without significant degradation.
= f ( t)
The diffusion coefficient D is determined from the initial
linear region of the Fickian diffusion curve using the
following relationship:
where M∞ is the equilibrium moisture concentration, M1 is
the moisture uptake after time t1, M2 is the moisture uptake
after time t2 and h is the thickness. The moisture
equilibrium concentration corresponds to the final
asymptotic value on the diffusion curve.
The rate of moisture uptake by a composite laminate is
dependent on the temperature and relative humidity of the
environment. The equilibrium moisture concentration
(saturation concentration in the material) is assumed to be
independent of temperature, depending only on the
moisture content or relative humidity of the environment.
PCB immersion finishes3-7:
HASL, OSP and ENIG are famous PCB finishes as they
represented more than 80% of the world market in 2001.
Then we will focus on immersion finishes and discuss
briefly about manufacturing, cost, intermetallics, drawbacks
and wettability/spreading.
Manufacturing:
The electromotive potential between silver and copper is
0.456V, so the reaction is instantaneous. Unfortunately,
tin's potential relative to copper is -0.480V, so the reaction
is not spontaneous. A strong complexor, typically thiourea,
is introduced into the tin solution to act as a selective
complexing agent for the cuprous cation. Additional
chemicals are used to improve the finish quality. Organic
compounds are added to the silver bath to inhibit tarnish
and to prevent electromigration. Organic surface modifiers,
inorganic grain modifiers or inorganic barrier layers are
added into tin baths. Silver and tin form a thin surface
coating due to environmental exposure to sulphides and
chlorides. These tarnishes are visible and may concern
incoming inspection. Silver oxide is not stable, and tin
oxide is not readily visible. Immersion metal chemicals
systems have been formulated to contain organic materials,
resulting in a protective film on the metal surface or within
the metal deposit. The degree of protection offered by the
film depends on such parameters as porosity, uniformity
and solubility.
Cost:
The cost associated with PCB finish depends on the cost of
the metal itself, the thickness of the metal, the cost of the
chemicals in plating bath, the cost of other chemistry and
equipment before and after metal deposition. One cost
benefit of immersion deposit is due to galvanic
displacement which is a self limiting reaction that will
usually result in a thickness of less than 0.5µm. Finally,
immersion deposits are generally less expensive than
electroless deposits because of the relative simplicity of the
immersion baths and overall processes. The electroless
nickel process cost nearly 30 times more than the nickel
metal itself and immersion silver process is at least 10 times
faster than electroless nickel. The cost of immersion
processes is about the same than HASL, and 3 times less
expensive than electroless nickel.
1. PCB MATERIALS BEHAVIOURS:
This part tackles the PCB materials behaviours towards
humidity, meaning absorption and desorption behaviours. It
is based on PCB weighting through baking and storage.
The equipment used for this study are a Heraeus UT200
oven, Sartorius LP3200D accurate scales (mg), Secasi
SLH34SP moisture oven.
1.1. Baking efficiency:
First of all the baking efficiency has been checked. We
have tried to figure out the impact of stacking the PCB
during the baking and also to estimate the time to reach the
same desorption for different temperatures and PCB
materials.
1.1.1. Inclined or stacked:
Heating:
We firstly checked the temperature of the PCB in the oven
according to if there are stacked or inclined. The following
graph shows the temperatures curves from ambient
temperature to 110°C of an inclined PCB, a five PCB stack,
and an eleven PCB stack. One thermocouple has been
attached between the first and second PCB of each stack.
3
stack. An important point has also to be noticed concerning
the baking of inclined PCB: according to the material, the
PCB thickness and PCB dimensions, inclined baking can
cause bow issue.
An other thermocouple has been placed in the middle of
each pile. Finally, one thermocouple records the
atmosphere temperature and one keeps the temperature of
the inclined PCB.
5
1.1.2. Backing conditions:
We have tried to figure out what are the required times at
different temperatures to have the same baking efficiency.
Seven PCB and five baking conditions have been studied.
The PCB have been baked 13 hours for each temperature
condition. In order to better know the real dry weight of
each PCB an additional baking at 120°C during 20 hours
was performed. The seven PCB are representative of the
PCB technology assembled at Solectron Bordeaux. The
following array shows a light description of those PCB:
6
4
1
3
2
Five PCB stack
Inclined PCB
Eleven PCB stack
Temperature curves from ambient temperature to 110°C
120
TEmperature (°C)
100
Air
80
PCB I.1
PCB I.2
PCB I.3
PCB I.4
PCB I.5
PCB I.6
PCB I.7
PCB n°1 of a stack of five PCB
60
PCB n°3 of a stack of five PCB
PCB n°1 of a stack of eleven PCB
PCB n°6 of a stack of eleven PCB
40
20
0
20
40
60
80
100
120
140
160
It can be noticed that PCB of the eleven PCB stack reach
110°C only after 80 minutes when the inclined PCB needs
only 10 minutes to reach this temperature and the five PCB
stack requires 50 minutes. Those temperature curves were
used to bake PCB and see the impact of this slow heating of
thick stacks on baking efficiency.
Desorption curve. T = 120°C.
PPM
0
Eleven PCB n°2 stack
2'
3'
4'
5'
1
2
3
4
5
6
7
8
9
10
3
4
5
6
7
8
9
10
11
12
13
1.1.2.2. Humidity amount:
This graph shows raw data meaning what is the weight lost
(mg) by each PCB through the baking. Obviously bigger is
the PCB higher is the humidity amount lost during the
baking. PCB I.6 an I.7 are the biggest (see above the
dimension array), so they loose a lot of water (above 400
45,0
1'
2
This graph roughly describes the desorption rate (PPM) for
each PCB. The weight after 13 hours of baking is not zero
as the dry weight has been recorded after 20 additional
hours of baking at 120°C. The 800 PPM line is the value
found in the IPC handbook (IPC-HDBK-001) as the
maximum value allowed for PCB assembly.
We can see that the PCB I.2, I.3 and I.5 have a higher initial
PPM (above 2400 PPM) meaning that they have absorbed
more moisture during storage. It is understandable for the
PCB I.2 as it is the E-glass/polyimide one. The PCB I.3 and
I.5 are FR-4 materials, but they have been stocked for
longer time at 50%RH 23°C. This can explain why they
have absorbed more moisture than the other FR-4 PCB.
50,0
D
1
Time (hours)
55,0
C
1600
0
60,0
B
PCB I.1
PCB I.2
PCB I.3
PCB I.4
PCB I.5
PCB I.6
PCB I.7
IPC upper limit
800
65,0
A
Length (cm) Thickness (cm)
7,5
0,148
14,2
0,182
8,5
0,182
32,3
0,209
26,5
0,212
43
0,226
43,9
0,23
2400
70,0
Five PCB n°2 stack
Width (cm)
6
7,5
4,9
19,5
22
26
26,3
1.1.2.1. Desorption curves:
Baking:
As we mentioned above it takes more time for the eleven
PCB stack to reach the final temperature. More time to
reach the final temperature means less efficient baking. For
this test the PCB have been left 4 hours at 110°C. After
weighting, the PCB have been baked 20 additional hours at
110°C in the inclined way. This allows us to have an idea of
the dry weight of each PCB and then to calculate the
efficiency of the first baking. The following graph describes
this efficiency according to if the PCB are inclined or
stacked. We can see that the PCB in the middle of the
eleven PCB stack has lost only around 53% of the moisture
when PCB on the top of the stack has lost 63 %. The PCB
on the top of the eleven PCB stack has lost nearly the same
amount of moisture as inclined PCB or PCB from the 5
PCB stack.
Inclined PCB n°2.
No stack
ANSI
FR5/23
GPY/41
FR5/23
FR5/23
FR5/23
FR5/23
FR5/23
For each PCB and for each baking condition several curves
have been drawn. Prior this study most of the PCB have
been left on shelves at ambient atmosphere (20+5/-0°C;
50+/- 10%RH) as they have been picked up from former
revision stocks. The following graphs have been drawn for
each studied temperature (80°C, 90°C, 100°C, 110°C, and
120°C).
180
Time (min)
% of desorption
Materials
E-glass + Epoxy
E-glass + Polyimide
E-glass + Epoxy
E-glass + Epoxy
E-glass + Epoxy
E-glass + Epoxy
E-glass + Epoxy
Inclined PCB
11
Designation
We made this test with different materials, and PCB having
a different thickness. We always found the same
conclusion. Above a 2.5cm stack (let say one inch), baking
is becoming less efficient for the PCB in the middle of the
4
mg). PCB I.1 and I.3 are the smallest so they loose less than
25 mg of moisture. One point that can be noticed is that the
PCB I.2 looses three times more water than the PCB I.1 and
I.3 with nearly the same dimension. It is due to the material
type (E-glass/polyimide).
Baking efficiency
100
90
80
700
%
Humidity amount lost during baking. T = 110°C.
PCB I.1
80°C
90°C
100°C
110°C
120°C
70
PCB I.2
600
PCB I.3
PCB I.4
60
PCB I.5
500
PCB I.6
PCB I.7
50
300
40
mg
400
PCB I.1
PCB I.2
PCB I.3
PCB I.4
PCB I.5
PCB I.6
PCB I.7
200
1.1.2.5. Relative desorption:
For each PCB this graph shows the relative desorption.
Relative desorption means simply that the initial weight
was considered to be equal to one for each test. This allows
to have the same starting point (0 , 1).
100
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
Time (hours)
1.1.2.3. Humidity amount versus volume:
This graph is based on the previous graph but the amount of
moisture lost during baking has been calculated per volume
unit. It is understandable that the PCB I.2 (Eglass/polyimide) is the PCB that looses more water per
volume unit. The PCB I.7 seems to loose less moisture than
the other FR-4. Basically, this PCB is full of copper (i.e.
ground layers on layer 2 and n-1).
1,2
Relative desorption curve for PCB I.1
1
PCB I.1 - 120°C
PCB I.1 - 110°C
PCB I.1 - 100°C
PCB I.1 - 90°C
PCB I.1 - 80°C
Relative PPM
0,8
0,6
0,4
5
Humidity amount lost during baking / PCB volume. T = 100°C.
4,5
0,2
PCB I.1
4
PCB I.2
PCB I.3
PCB I.5
0
PCB I.6
3
mg/cm3
0
PCB I.4
3,5
2
4
6
8
10
12
Time (hours)
PCB I.7
1,2
2,5
Relative desorption curve for PCB I.2
2
1
1,5
1
Relative PPM
0,5
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
Time (hours)
0,2
0
0
2
4
6
8
10
12
Time (hours)
100
It can be seen that the PCB I.2 chart shows tight curves
meaning E-glass-polyimide looses moisture even at lower
temperature. This point can also be seen on the previous
graph (1.1.2.4) as PCB I.2 has a baking efficiency after 13
hours above 90% from 90°C to 120°C.
Humidity percentage lost during baking. T = 80°C.
90
0,6
0,4
1.1.2.4. Humidity percentage:
This graph exposes the humidity percentage lost during
baking compared to the dry weight measured with 20
additional hours baking at 120°.
80
70
60
%
PCB I.2 - 120°C
PCB I.2 - 110°C
PCB I.2 - 100°C
PCB I.2 - 90°C
PCB I.2 - 80°C
0,8
50
Conclusion:
All the graphs (each type, each temperature, each PCB) are
available on request. The initial question was what are the
required times at different temperatures to have the same
baking efficiency? In order to answer this question, we
made the average of all the relative desorption curves
(1.1.2.5) and drew the inverse. The relative desorption
average inverse points are making lines. According to this
average, the red curve showing the 800 PPM has been
added only for information. This chart simply shows the
different rate of baking according to the temperature and
then for a fixed "y", it allows to calculate the appropriate
"x" according to the temperature.
40
PCB I.1
30
PCB I.2
PCB I.3
20
PCB I.4
PCB I.5
PCB I.6
10
PCB I.7
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
Time (hours)
The following chart is a sum-up of the previous graph. It
represents all the measurements after 13 hours of baking for
each PCB and each temperature. Obviously, 80°C baking is
less efficient than 120°C.
5
Reinforcement
9
Average relative inverse desorption curves
8
120°C
110°C
100°C
90°C
80°C
800 PPM
7
1 / (Relative PPM)
y = 0,4971x + 1,4992
2
R = 0,9826
6
5
Resin
Phenol
Epoxy
Polyester/Vinyl ester
Epoxy
Epoxy/Phenolic
Epoxy/PPO
Epoxy/Triazine
BT/Epoxy
Polyimide
APPE
Cyanate ester
Polyimide/Epoxy
PTFE
Epoxy
Cyanate ester
Polyimide
Polyimide/Epoxy
Polyimide
Cyanate ester
Hydrocarbon
PTFE
BT
Polyimide
Cellulose paper
y = 0,2983x + 1,5625
R2 = 0,997
y = 0,2336x + 1,4301
R2 = 0,9943
4
E-glass
Unidirectional or Woven
Surface or full
y = 0,166x + 1,3633
2
R = 0,9901
3
y = 0,0946x + 1,3141
2
R = 0,9876
2
1
0
0,5
2,5
4,5
6,5
8,5
10,5
Time (hours)
12,5
14,5
16,5
18,5
If we compare the baking slope, we can say what are the
required times at different temperatures to have the same
baking efficiency. For example, to obtain the same baking
efficiency between 120°C and 80°C, it requires an average
of 5.6 more times.
Aramid
Unidirectional or non woven paper or woven
Woven quartz fiber
Woven S-2 glass
Ceramic
Pure or woven glass reinforced
Expended PTFE
None
Temperatures slope comaprison
6,0
5,6
5,0
The PCB have been divided into three groups:
- The group 1: PCB absorbed less 4000 PPM in moist
conditions.
Factor
4,0
3,1
3,0
2,1
2,0
1,0
Group 1
II.3: E-glass/Epoxy//Ceramic/Hydrocarbon II.15: FR-4 full copper (L1 & Ln)
II.9: Ceramic/PTFE
II.20: FR-4 full copper (L1 & Ln)
II.10: E-glass/APPE
IB7: FR-4 Full copper (L1 & Ln)
II.12: E-glass/PTFE
1,6
1,0
0,0
120°C
110°C
100°C
90°C
80°C
- The group 2: PCB absorbed between 4000 and 9000 PPM
in moist conditions.
Temperatures
If we consider only the polyimide charts, it needs less time
to have the same efficiency (1.1.2.4) due to the material
type. But as FR-4 is the major material assembled at
Solectron Bordeaux we had to focus on it. Finally, this part
of the study seems to show that 4 hours at 120°C is
appropriate to reduce significantly the moisture in a raw
card. This condition decreases the percentage to around 400
PPM (average value) which is below the 800 PPM
recommended by the IPC. It also allows to bake in a
production environment. According to the previous graph, 4
hours at 120°C means 6.5 hours at 110°C, 8.5 hours at
100°C, 12.5 hours at 90°C, 22.5 hours at 80°C. Those
values are given for standard thickness (<2.4mm) and
storage shorter than 2 months at 50%RH and 23°C. We also
found that 8 hours at 120°C can be required to cross down
the 800PPM line with thick PCB stored for more than 2
months at 50%RH and 23°C. Once again 8 hours at 120°C
means 13 hours at 110°C, 17 hours at 100°C, 25 hours at
90°C, 45 hours at 80°C. One study not tackled in this paper
deals with the impact of the thickness and copper density
on those values.
Group 2
II.4: E-glass/Epoxy/BT
II.11: Aramid/Epoxy
II.16: FR-4
II.17: FR-4 Half copper (L2 & Ln-1)
II.18: FR-4 Ground on L2
II.19: FR-4
IB1: FR-4
IB3: FR-4
IB4: FR-4
IB6: FR-4
- The group 3: PCB absorbed more than 9000 PPM.
Group 3
II.1: None/Polyimide
II.7: Paper/E-glass/Epoxy
II.2: E-glass/Polyimide
II.8: Paper/E-glass/Phenol/Epoxy
II.5: Eglass/Epoxy//None/Polyimide
II.2: E-glass/Polyimide
II.6: Paper/Phenol
IB5: FR-4
1.2.1. Baking:
The following graphs are simply showing the baking curves
of the three groups at 110°C during 13 hours. The dry
weight was measured with 20 additional hours at 120°C.
Prior baking the PCB were left on shelves at 50%RH and
23°C.
2400
Baking curves at 110°C
2000
1.2. Materials behaviors towards humidity:
Most of the PCB materials assembled at Solectron
Bordeaux have been studied. The following array shows the
different materials available on the market. The one that are
with a coloured square have been included in this study.
PPM
1600
II.3
1200
II.9
II.10
II.15
II.20
IB7
800
400
0
0
2
4
6
Time (hour)
6
8
10
12
The PCB II.12 curve is missing. This PCB is a very thin Eglass/PTFE PCB. It probably suffered a lot during the
baking. It turned from white to yellow. Its behaviour is the
opposite of the other materials. It absorbed moisture during
baking. So the baking curve is below the x-axis.
For each group, in the following graphs you will find the
storage conditions described above but also how long did it
take for each PCB to reach 800 PPM at ambient
atmosphere.
5600
4800
4400
Dry pack storage: 3892 hrs (162 days)
Dry cabinet storage: 3074 hrs (128 days) at 23+/-2°C and less than 5% RH
How many times to reach 800 ppm? (23+/-2%C & 45+/-5%RH)
4800
Ambient atmosphere storage: 3122 hrs (130 days) at 23+/-2%C and 45+/-5%RH
4000
II.1
3600
II.2
II.5
II.7
II.8
IB2
Moist atmosphere storage: 660 hrs (28 days) at 85°C and 85%RH
4000
IB5
2000 hrs
3100 hrs
950 hrs
1600
300 hrs
2400
II.12
II.15
II.20
IB7
Never
2400
2900 hrs
PPM
2800
50 hrs
3200
3200
PPM
Initial weight before baking 33 hours at 120°C
Different storage conditions
2000
800
1600
0
1200
II.3
800
II.9
II.10
-800
400
-1600
0
0
2
4
6
8
10
PCB
12
11200
Time (hour)
10400
3600
9600
3200
8800
8000
2800
II.4
II.11
II.16
II.17
II.18
II.19
IB1
IB3
IB4
IB6
7200
2400
PPM
6400
2000
5600
II.18
250 hrs
II.17
500 hrs
II.16
475 hrs
II.11
325 hrs
165 hrs
II.4
2400
275 hrs
60 hrs
3200
475 hrs
4000
1200
70 hrs
4800
1600
160 hrs
PPM
II.19
IB1
IB3
IB4
IB6
1600
800
800
400
0
0
PCB
12
Time (hour)
38400
Baking curves will have to be drawn with the same PCB
after taking them out from forced moist atmosphere. This
will allow to have the baking curve with the starting point
being the moisture saturation.
35200
32000
28800
PPM
25600
1.2.2. Storage:
This part aims to check the storage efficiency. First PCB
have been hard baked (33 hours at 120°C) and then they
have been stored through different ways:
- Dry pack: PCB are sealed under vacuum with a desiccant.
The storage conditions can be estimated below 2%RH.
- Dry cabinet: PCB are stored in a dry atmosphere. The dry
air coming from the inlet pipe is around 2%RH and 23+/2°C. The dry cabinet has been used for production, this
means that the door was opened several times during the
day. Each time the door is opened the moisture percentage
increases to 15-20%RH and it takes around one hour to
decrease below 2%RH. That is to say that we took the
worse case to test the dry cabinet storage efficiency.
- Ambient atmosphere: PCB were left on a shelf. The
standard storage conditions are 50+/-10%RH and 20+5/0°C. During the experiment the outside air was very dry
(cold winter), then the storage observed conditions were
45+/-5%RH and 23+/-2°C.
Moist atmosphere: PCB were placed in moist conditions
85%RH and 85°C.
Of course measurements disturb measurements. For
example each time we made a measurement it took around
one hour before the moist conditions (85%RH) become
stable back. This phenomenon has not been taken into
account.
Dry cabinet storage: 3074 hrs (128 days) at 23+/-2°C and less
than 5% RH
22400
Ambient atmosphere storage: 3122 hrs (130 days) at 23+/2%C and 45+/-5%RH
19200
Moist atmosphere storage: 660 hrs (28 days) at 85°C and
85%RH
16000
6400
II.1
II.2
II.5
II.6
125 hrs
9600
15 hrs
12800
80 hrs
10
275 hrs
8
10 hrs
6
425 hrs
4
70 hrs
2
0,5 hrs
0
IB2
IB5
3200
0
II.7
II.8
PCB
We can say that more the PCB material is able to absorb a
big amount of moisture, faster it absorbs the moisture after
baking (group one versus group three). Some PCB are not
belonging to this rule. PCB II.10 (E-glass/APPE) for
example absorbs less than 3200PPM in moist conditions
but need only 10 hours to reach 800PPM after baking.
1.2.2.2. Absorption curves:
PCB materials:
All the absorption tests were made with inclined PCB.
Absorption curves of stacked PCB are on going. This
should badly decrease the absorption rate. For each material
we are able to draw the different absorption curves. If we
take for example the PCB II.4, we can draw the absorption
curves for the first 525 hours.
1.2.2.1. Storage comparison:
7
8000
17600
PCB II.4 absoption curves
Ambient atmosphere absorption curves: 23+/-2°C & 45+/-5%RH
16000
7000
II.1
14400
II.2
II.5
II.6
II.7
II.8
IB2
IB5
6000
12800
11200
4000
PPM
PPM
5000
3000
9600
8000
6400
4800
2000
3200
1000
1600
0
0
0
25
50
75 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 525
0
250
500
750
1000
1250
Time (hours)
With such a scale it seems that the moist atmosphere and
ambient atmosphere curves have started to reach their
plateau. But when we increase the scale we can see that the
ambient atmosphere curve jumps from 2000PPM to
3000PPM and that the moist atmosphere curve has only
started to bend. The moist ambient curve is still on
progress.
PCB II.4 absoption curves
7000
3200
5000
2000
2250
2500
2750
3000
2400
4000
3000
Baking 4 hours at 120°C
Different storage
&
baking conditions
PPM
1750
1.2.2.3. Absorption comparison:
For each group, the following graph are showing the initial
weight before baking 33 hours at 120°C. Following this
baking PCB are left at 45+/-5%RH and 23+/-2°C. You will
find two weights. One after 96 hours and one after 3122
hours. Then the PCB are baked 4 hours at 120°C and left
again at 45+/-5%RH and 23+/-2°C. You will find again two
weights. One after 96 hours and one after 656 hours. Then
the PCB are baked 8 hours at 120°C.
8000
6000
1500
Time (hours)
1600
PPM
2000
800
1000
0
0
0
500
1000
1500
2000
2500
3000
II.3
3500
Time (hours)
II.9
II.10
Absorption conditions:
As for the baking curve, we can draw the absorption curves
for each condition and for each PCB group. For example
the following graph are showing the ambient atmosphere
absorption curves.
4000
II.10
II.12
II.15
II.20
IB7
3200
IB7
PPM
II.9
II.20
PCB
II.3
II.15
-1600
2400
2000
II.12
-800
1600
2400
PPM
1600
1200
800
800
0
400
II.4
II.11
II.16
II.17
II.18
II.19
IB1
IB3
IB4
IB6
PCB
0
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
Time (hours)
3600
3200
II.11
II.16
II.17
II.18
II.19
IB1
IB3
IB4
IB6
PPM
2800
II.4
2400
PPM
2000
1600
1200
800
16800
16000
15200
14400
13600
12800
12000
11200
10400
9600
8800
8000
7200
6400
5600
4800
4000
3200
2400
1600
800
0
Ambient atmosphere storage: 96 hrs (4 days) at 23+/-2%C and 45+/5%RH
Ambient atmosphere storage: 3122 hrs (130 days) at 23+/-2%C and
45+/-5%RH
Ambient atmosphere storage: 96 hrs (4 days) at 23+/-2%C and 45+/5%RH
Ambient atmosphere storage: 656 hrs (27 days) at 23+/-2%C and
45+/-5%RH
II.1
400
II.2
II.5
II.6
II.7
II.8
IB2
IB5
PCB
Those graphs show that when you bake at 120°C during 4
hours, PCB left 3122 hours at 45+/-5%RH and 23+/-2°C, it
0
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
Time (hours)
8
takes around 650 hours for most of the materials to reach
again their weight prior the 4 hours baking. The third and
sixth columns of each PCB graph have the same moisture
level. The dry weight (0 PPM) has been measured after 33
hours at 120°C. It is important to notice that 4 hours at
120°C is not enough too cross down the 800PPM line after
that the PCB were left 3122 hours at 45+/-5%RH and 23+/2°C. Furthermore for some materials 8 hours at 120°C does
not decrease significantly the weight compare to 4 hours at
120°C.
3500
D2 = 6,21*10-6 mm2s-1
PCB II.10 (85%RH, 85°C)
PCB II.10 (45+/-5%RH, 23+/-2°C)
2500
D1 = 6,41*10-7 mm2s-1
PPM
2000
1500
1000
500
1.2.2.4. Materials comparison:
We can also compare the different materials behaviours
towards moisture. The first graph is the amount of moisture
per volume unit absorbed after 3122 hours at 23+/-°C and
45+/-5%RH. Of course, we find on the left the PCB
belonging to the group one and the PCB on the right are
belonging to the group three.
0
0
500
1000
1500
2500
3000
3500
4000
10000
D2 = 8,14*10-7 mm2s -1
9000
PCB II.17 absoption curves (Fick's law)
PCB II.17 (85%RH, 85°C)
PCB II.17 (45+/-5%RH, 23+/-2°C)
8000
7000
6000
Amount of absorbed humidity per volume unit after 3122 hours at 23+/-2°C & 45+/-5%RH
5000
D1 = 2,21*10-7 mm2s-1
4000
15
mg/cm3
2000
Square root of time (s0,5)
PPM
20
3000
3000
2000
10
1000
5
0
0
500
1000
2500
3000
3500
4000
II.
3:
Egl
as
s/E
p
II. oxy II.
20 // 9:
: F Ce C
R- ram era
m
4
fu ic/H ic/
II.
ll
15
co yd PTF
:F
pp roc E
er ar
R4
(L bo
fu
n
1
ll
&
IB
co
7:
pp IB1 Ln)
FR
er
:F
-4
(L RFu
1
& 4
ll
co
L
pp IB4 n)
er : F
(L R1
& 4
II. L
19 n)
II.
:F
5:
Eg
I B Rlas
3: 4
FR
s/E II.1
po 0: E IB6 -4
II. xy/ -gl : FR
18 /N as
: F on s/A -4
R- e/P PP
4
G olyi E
ro
un mid
e
d
I
II. I.1 II on L
7: 2: .1
Pa E- 6: 2
pe gl FR
r/E ass -4
II.
II.
17
/
1 -gla PT
:F
R- IB 1: A ss/ FE
4 2:
ra Ep
H
al E-g mid oxy
f c la
/
o ss Ep
II. ppe /Po oxy
4: r ( lyi
E- L2 mi
de
gl
&
as
II.
s/E Ln
8:
po -1)
Pa II.2
xy
pe : E
r/E -g IB /BT
-g las 5:
las s/P FR
s
II. /Ph olyi 4
1: en m
N ol/ ide
on E
II. e/P pox
6: ol y
Pa yim
pe
r/P ide
he
no
l
10000
9000
PCB II.2 (85%RH, 85°C)
PCB II.2 (45+/-5%RH, 23+/-2°C)
8000
7000
D2 = 8,29*10-7 mm2s-1
6000
PPM
5000
D1 = 1,02*10-7 mm2s-1
4000
3000
2000
14
1000
12
0
0
10
Days
2000
0,5
We can also compare the PCB materials on their absorption
velocity. The following graph show how many days does it
take for each PCB to reach 800 PPM after 33 hours of
baking at 120°C.
16
1500
Square root of time (s )
0
500
1000
1500
2000
2500
3000
3500
4000
0,5
Square root of time (s )
8
Some values can be compared to the moisture/water
diffusion coefficient of pure resin found in literature:
- The polyimide diffusion coefficient was found to be
nearly constant with reaction temperatures ranging from
160°C to 240°C around 2.2 to 4.8 x 10-7 mm2s-1. Moisture
diffuses through polyimide at room temperature with a
diffusion coefficient of 5 x 10-7 mm2s-1. In comparison we
found for E-glass/Polyimide 1.02-8.29 x 10-7 mm2s-1
depending on the conditions.
- The epoxy diffusion coefficient was found to be around
13 × 10-7 mm2s-1. The distilled water diffusion coefficient
was found to be 0.53 x 10-7 mm2s-1 at 22°C and 13.6 x 10-7
mm2s-1at 60°C. In comparison we found for E-glass/Epoxy
2.21-8.14 x 10-7 mm2s-1.
One point has to be noticed. The literature values are given
for pure resin. In our case there are also E-glass and
randomly copper as the tests were performed on raw PCB.
More points during the linear part of the curve would have
given a more accurate result. Furthermore the M∞ used for
the calculation is not the real M∞ as at that time the moist
test duration was about 660 hrs which seems to be not
enough to reach the asymptote for some materials.
6
4
2
II.
1:
N
on
II e/P
IB .6: oly
II.
2
17
: E Pap imi
:F
-g er/ de
P
l
RII ass he
4
H .10 /Po nol
al
f c : E- lyim
op gla
id
II. per ss/A e
2:
(
II.
E- L2 PPE
8:
gl &
a
L
Pa II
pe .11 ss/P n-1
r/E : A ol
)
yi
-g
las ram mid
e
s/P id/
he Ep
no ox
l/E y
II.
4:
po
II.
E
18 -gl IB5 xy
: F ass : F
RR- /E
p
4
4
G oxy
ro
II.
un /BT
7:
d
on
Pa
pe
L
r/E IB6 2
-g : F
las Rs/E 4
II.
p
II.
5:
12 II.1 oxy
Eg
: E 9:
las
-g FR
s/E
la
ss -4
po
/
PT
xy
//N IB FE
on 1:
e/P FR
ol -4
yi
II. mid
IB
16 e
7:
:F
F
II.
3: II.1 R-4
I B R-4
E- 5:
3:
F Fu
gl
F
as R-4 ll c
IB R-4
o
s/E
pp
4:
p ful
e
II. oxy l co r (L FR20 //C pp
4
1
&
:F
e
e
R- ram r (L Ln
4
fu ic/H 1 & )
ll
co ydr Ln)
pp oc
II. er ( arbo
9:
L
n
Ce 1 &
ra
m Ln)
ic
/P
TF
E
0
Again, we can roughly find the group three on the left and
the group one on the right.
1.2.2.5. Fick's law11-14:
Some PCB have a Fickian behaviour. For example, the
three following graphs are showing Fickian curves for one
PCB of each group described previously: PCB II.10 (Eglass/APPE), II.17 (Half copper on L2 & Ln-1 and II.2 (Eglass/Polyimide).
Conclusion:
9
Following this study we better know the PCB materials
behaviours. The three identified groups are in accordance
with PCB materials supplier data sheet. Some comments
can still be made. The PCB II.12 (E-glass/PTFE) has been
destroyed during this test (colour change and random
behaviour). Within a group, some PCB have amazing
behaviours. For example the PCB II.10 (E-glass/APPE)
absorbs less than 3200PPM in moist conditions but need
only 10 hours to reach 800PPM after baking while in the
mean time the PCB II5 has the opposite behaviour. It
absorbs more than 9600 PPM but needs 425 hours to reach
800PPM after baking.
Instrument Zero
Y
Fmax
-15 mm
t
Corrected Zero
Tw
2. BAKING IMPACT:
2.1.1. Fmax:
We have taken the highest Fmax and fixed it at 20, the
smallest Fmax was fixed to nil and then we were able to
give marks to the other Fmax between zero to 20.
This part of the paper is to check the impact of the baking
conditions on the PCB finish. Five PCB finishes were
tested: HASL, ENIG, OSP, immersion silver, and
immersion tin. Boards have been assembled following a
baking in order to see the impact of the baking on solder
paste spreading. Furthermore, several wettability coupons
have been taken out from each board at different assembly
process steps to perform the tests.
The assembly line was composed of Dek 265, Fuji CP65E,
Fuji IP3, Paragon P150. The stencil was a 150µm laser cut
for the bottom and a 110µm electroformed for the top side.
The solder paste used for the assembly is the no Clean
SMQ92J. The reflow atmosphere is air. The reflow profile
is as follow:
Relative Fmax according to the finishes and ageing conditions
25
20
80°C 10hrs
120°C 6hrs
120°C 2hrs + one reflow
100°C 5hrs + two reflow
100°C 5hrs
120°C 2hrs
Score
15
10
5
0
ENIG
OSP
Sn Im
Ag Im
HASL
Finish
2.1. Wettability:
For each finish and for each baking condition, the
wettability test has been performed after baking, after first
reflow and after second reflow. According to the finish, we
can roughly represent the different wettability curves as
follow:
Only ENIG, HASL and immersion silver are above 15.
OSP is below 5 and immersion tin is around 14. ENIG
seems to be slightly impacted with baking conditions when
immersion silver remains good whatever the baking
condition. HASL is the only finish that improves slightly its
Fmax when baking conditions are becoming more severe.
Wettability curves shapes
2.1.2. TW:
As for Fmax, we have taken the fastest Tw and fixed it at
20, the test duration was fixed as the slowest Tw means
fixed to nil and then we were able to give marks to the other
Tw between zero to 20. Unfortunately when the Tw could
not be measured (mainly OSP) because of the wettability
curve shape, we had taken into account the remaining
negative Fmax at the latest point on the curve. The negative
values are not representative of Tw, they simply indicate if
the curve was far from cross over again the corrected zero.
ENIG - HASL
OSP
ImSn
Im Ag
Relative Tw according to the finishes and ageing conditions
Immersion silver and immersion tin reach a threshold,
ENIG, OSP, HASL don't. ENIG, HASL and immersion
silver cross over the corrected zero, immersion tin reaches
with difficulties the corrected zero, OSP never crosses the
corrected zero.
According to the IPC test method it can be measured two
results: the wetting time (Tw) and the wetting force (Fmax).
28
24
80°C 10hrs
120°C 6hrs
100°C 5hrs
120°C 2hrs
20
Score
16
12
8
4
0
ENIG
OSP
Sn Im
Ag Im
HASL
-4
-8
Finish
OSP never crossed again the corrected zero meaning very
poor wettability. Immersion tin has irregular results and
faced difficulties to cross over the corrected zero. Baking
10
of each circular pad is about to be computerised with a grey
scale mode to be more accurate.
conditions seem to have a slight impact on ENIG,
immersion silver and HASL. Immersion silver and HASL
are the only finishes above 13.
20
16
2.1.3. Final score:
If we make an average whatever the baking conditions we
have a final result for Fmax and Tw. Finally, we can also
have an overall score whatever the baking condition and
measurements by making the overall average.
Circular spreading test
80°C, 10 hrs
100°C, 5 hrs
120°C, 2hrs
120°C, 6 hrs
18
14
Score
12
10
8
6
25
4
Wettability relative final score (Fmax & Tw) + overall relative final score
2
20
0
Fmax
Tw
Overall score
15
ENIG BOT ENIG TOP OSP BOT
Score
0
ENIG
Sn Im
Im Ag
TOP
HASL BOT HASL TOP
It is important to notice that the spreading test is much
better on the bottom side which is assembled first than the
top side that undergoes one additional reflow prior
assembly. In fact, this result is not better because the finish
of the top side is destroyed during the first reflow but
because the solder paste on the bottom side has two reflows
to spread. The spreading tests on the bottom side gave
nearly the same result than the one on the top side prior the
second reflow. Only OSP and HASL on the TOP side seem
to be slightly affected by the baking conditions. The HASL
spreading score on the bottom side increases when baking
conditions are severe. Only ENIG is above 15. OSP and
HASL are below 10. Immersion tin and immersion silver
are between 10 and 15.
5
HASL
Im Ag
BOT
Finish and board side
10
Ag Im
OSP TOP Im Sn BOT Im Sn TOP
OSP
-5
-10
Finish
We can see that the immersion silver has the best results,
followed by HASL. ENIG is only third because of poor Tw
result due to the wettability curve. Immersion tin is a bit
behind while OSP is far from good results.
2.2. Spreading:
We have performed two different spreading tests. One is a
standard test with rectangular apertures, the other is based
on archery target. Those two tests are present on the two
board sides in order to see the impact of the first reflow.
Finally the standard test is present twice on the top side to
see the impact of sweat on spreading. All the boards were
manipulated and assembled with appropriate gloves except
this second area where people were allowed to add their
finger print prior baking.
The two following pictures are stencil apertures designs:
2.2.2. Standard spreading test:
Like for circular pad, a mark was given to each standard
spreading test. Higher score means all the pads reach its
neighbours.
16,0
Rectangular spreading test
14,0
80°C, 10 hrs
100°C, 5 hrs
120°C, 2hrs
120°C, 6 hrs
12,0
Score
10,0
8,0
6,0
4,0
2,0
0,0
ENIG
BOT
ENIG ENIG
TOP TOP Sweat
OSP
BOT
OSP
TOP
OSP Im Sn
TOP - BOT
Sweat
Im Sn Im Sn
TOP TOP Sweat
Im Ag Im Ag Im Ag HASL HASL HASL
BOT TOP TOP - BOT TOP TOP Sweat
Sweat
Finish and board side
The three pictures are showing a circular pad after screen
printing prior reflow and the same pad after reflow for OSP
and ENIG finish.
The better spreading on the bottom side is less obvious than
with the circular test.
ENIG is around 12. HASL and immersion silver are
between 6 and 12. Immersion tin and OSP are below 6.
Concerning the impact of sweat on the spreading, if we
mainly consider the 80°C peak on the top side, OSP and
immersion silver are affected with organic contaminants.
2.2.3. Final score:
If we add all the spreading test results we can have a
relative final score that has to be compared with the
wettability final score. ENIG and immersion silver swap
their position, meaning ENIG becomes first for the
spreading test and immersion silver becomes third.
2.2.1. Circular spreading test:
It was given to each circular pad a mark like for archery
target. Higher is the mark less area on the pad is free of
solder paste. The maximum mark means full pad covered
with reflowed solder paste. The chart below represents the
average mark (6 archery target) for each side. X-ray picture
11
one hour baking at 120°C has almost the same efficiency
than 1.6 hours at 110°C, 2.1 hours at 100°C, 3.1 hours at
90°C and 5.6 hours at 80°C. We also know that 4 hours
baking at 120°C is needed to remove moisture from the
PCB thinner than 2.5mm, stored at 50%+/-10%RH for less
than two months and belonging to the group 1 (HF
material) or to the group 2 (most of the FR-4). Otherwise
for thicker PCB or PCB stored for more than 2 months or
PCB made with moisture sensitive material (group 3), 8
hours at 120°C is required instead of 4 hours. Those values
are given to decrease the moisture percentage below 800
PPM with the dry weight measured after 33 hours baking at
120°C. For example, in the literature PCB material supplier
recommend for aramid system 2800 PPM maximum for
assembly. This value allows to divide the baking times by
two. Anyway this baking has an impact on the wettability.
This impact is obvious on OSP. No wettability
improvement were observed by baking OSP finish 10 hours
at 80°C instead of 2 hours at 120°C. The wetting force is
not significantly impacted by the baking while the baking
disturbs badly the wetting time for immersion tin and
decreases significantly the wetting time for ENIG,
immersion silver and HASL. Finally, one of the two
spreading tests showed that the baking impacted all the
finishes, except HASL.
If people want to take care of moisture they have to know
that choosing a PCB technology might impose the PCB
finish. It would not be recommended to have an OSP or tin
immersion finish on a 16 layers polyimide rigid flexible.
An other example is the paper/phenol system that seems to
be one of the most sensitive material. But it is often used
with OSP for the automotive market due to their cost
effectiveness. This means that one of the most sensitive
system is used with the most temperature sensitive PCB
finish.
It just has to keep in mind that choosing a PCB technology
that requires moisture sensitive materials might reduce the
available PCB finish list due to baking constraints. The first
results of the on going delamination study show that only
few PCB are delaminating during three successive reflow
even if they are full of water (800 hours at 85°C/85%RH
meaning over 7000PPM). It might be worth assembling
moisture sensitive PCB without baking them and taking the
risk to have assembly issue than bake the PCB and destroy
the finish so badly that the PCB cannot be assembled
anymore. Anyway when a PCB finish is not affected by
temperature (i.e ENIG, immersion silver, HASL), baking is
recommended, as if delamination occurs it avoids arguing
with the PCB supplier. Anyway, prior assembly, if there is
any doubt about the storage conditions, the baking is
mandatory
Finally, after baking or good packing delivery, storage in a
dry cabinet (<5%) allows a long time storage without any
risk of moisture absorption.
20,0
20,0
Spreading overall relative score
14,4
15,0
12,7
12,1
10,0
8,8
5,0
0,0
ENIG
HASL
Im Ag
Im Sn
OSP
Conclusion:
To see the impact of the baking conditions on the PCB
finish, we can rank the finish by adding all the scores
(wettability and spreading). Three finishes have nearly the
same result: HASL, ENIG, and immersion silver. It is
important to notice that the HASL had almost the same
behaviour whatever the test (twice second in previous
ranking). But ENIG and immersion silver had once the first
place and once the third place which make an average
around the HASL result. Immersion tin is a bit behind while
OSP is very far behind. Obviously the finish that is the
most affected by severe baking conditions is OSP. ENIG,
immersion silver and immersion tin are slightly impacted
with severe baking conditions. HASL has better result with
severe baking conditions. One critical point that need to be
said is that 0.4 mm QFP and 0.5 mm CSP pitch are
assembled on the board. Furthermore 100µm SIR are also
on the outer layer. This means that hard hot air knives were
used to have a flat and regular HASL finish resulting in a
thin and weak finish. This can explain why the spreading
test and wettability test were not as good as expected with
HASL as tin-lead alloy shall spread and wet easier on tinlead finish than other metallic or organic finish.
25,0
Relative overall score : wettability + spreading
80°C, 10 hrs
100°C, 5 hrs
120°C, 2hrs
120°C, 6 hrs
20,0
15,0
10,0
5,0
0,0
ENIG
OSP
Im Sn
Im Ag
HASL
Finish
3. CONCLUSION:
Why shall we bake the PCB? On the first hand if you do not
bake the PCB you might have voids, solder balls issue or
even delamination. But on the other hand according to the
PCB material, a hard baking might be needed to remove the
moisture. Hard baking means wettability and reliability
issue if a temperature sensitive finish is used on the PCB.
Delamination or reliability issue? Who decide?
Though the parts of this study we are now able to bake
properly the PCB. We know that the PCB can be stacked as
long as the stack is not higher than 2.5cm. We know that
Acknowledgment:
We would like to thanks CIREP for giving us some PCB
samples.
REFERENCES:
1. Subhotosh Khan, Comparison of the Dielectric
Constant and Dissipation Factors of non-woven
12
Aramid/FR4 and Glass/FR4 Laminates. Downloaded from
internet.
2. Arlon, Recommendations for Storage, Drying, and
Assembly of Printed Wiring Boards Containing 85NT
Substrates. Downloaded from internet.
3. Donald Cullen, Silver & change a tale of silver, copper,
nickel and gold. Downloaded from internet.
4. Donald Cullen, Lenora Toscano, Immesion metal PWB
surface finishes: A direct comparison of selected fabrication
assembly and reliability caharacterictics. Downloaded from
internet.
5. Christ Hunt and Ling Zou, Board finish solderability
with Sn-Ag-Cu, APEX 2003.
6. Minna Arra, Dongkai Shangguan, Wetting of fresh and
aged immersion tin and silver finishes by Sn/Ag/Cu solder,
APEX 2003.
7. Robert Gordon, Susan Marr, Dongkai Shangguan,
Evaluation of immersion silver finish for automotive
electronics. Downloaded from internet.
8. Miller Brothers School of Engineering. Winona state
university. Composite materials Engineering. ENGR451:
Transport phenomena laboratory: Diffusion coefficient.
Downloaded from internet.
9. Raphaël Duboz, and al. Utiliser les modèles indviduscentré comme laboratoires virtuels pour identifier les
paramètres d'un modèle agrégé. Downloaded from internet.
10. WR Broughton and MJ Lodeiro, Techniques for
monitoring water absorption in fibre-reinforced polymer
composites. Decembre 2000. Downloaded from internet.
11. Dr. David R. Day. Moisture Monitoring at the PI-SiO2
Interface Using Microdielectric Sensors; Micromet
Instruments, Inc. Downloaded from internet.
12. David Locker. An Evaluation of Chip Scale Packages
for Missile Applications. Downloaded from internet.
13. A.S. Hussain, and al. Finite element modelling of
moisture absorption in single fibre reinforced composite
systems. Downloaded from internet.
14. J. Chin, and al. Sorption and Diffusion of Water, Salt
Water, and Concrete Pore Solution in Composite Matrices.
Journal of Applied Polymer Science, Vol. 1, 483-492, 1999.
13

pcb materials behaviours - FED-Wiki

Transcription

Similar documents

Cencorp Depaneling

Service Manual

Audio Restoration Repair Arcam AVR350 Receiver

diy- rackwah

Brochure

Fancort Solutions

Chocolate Banana Bread - Cloud 9 Specialty Bakery

Presentation

ampx user manual - Klundert Muziek

Wash Day Peach Cobbler