You are on page 1of 5

Use of The Area-of-Spread Method for

Monitoring The,Stability of Reflow Furnaces


N. N. Ekere and W. Lawson
Electronics Manufacturing Engineering Research Group
Department of Aeronautical,
Mechanical and Manufacturing Engineering
University of Salford, Manchester, M5 4WT, United Kingdom
Email : n.n.ekere @ salford. ac. uk

1. Bibliography Although the reflow soldering technique has been


around for quite a number of years, and has been used in
Ndy Ekere heads the Electronics Manufacturing the assembly of surface mount devices, there are very
Engineering Research Group at Salford University, few methods available for the on-line monitoring and
where he also holds a chair in Electronics assessment of solderability.There are a large number of
Manufacturing. He received his BEng (honours) degree off-line solderability assessment techniques, such as the
in Mechanical Engineering from the University of solder dip method, the Wetting Balance method and the
Nigeria, N s M i g e r i a [1981], the Master of Science Globule Balance method. Although these methods are
Degree in Flexible Manufacturing Systems and quite useful for process characterisation, they are not
Robotics from Loughborough University, suitable for monitoring the process stability of the
Leicestershire, UK [1984], and the Doctor of Science reflow furnace over time. This paper discusses the use
degree in Manufacturing Engineering from the of an area-of-spread experiment as a means of
University of Manchester Institute of Science and measuring the stability of the reflow process using
Technology, Manchester, UK [ 19971. His main research statistical process control (SPC) methods.
interests are in the modelling of manufacturing It is expected that this methodology will be applicable
processes, in particular the processes used for high- for process control and monitoring activities to expedite
density electronics interconnection. the development and efficient operation of reflow
Wayne Lawson currently works as a Project Engineer, soldering systems. An improved method of measuring
Hybrids with Delphi Auto in Liverpool UK. He the area-of-spread of solder paste is used in the
received the BEng (honours) degree in Electronics experiment from which quantitative result can be
Engineering [1997], and the Master of Science in produced to gain reflow oven control data.
Advanced Manufacturing [19991 from the University of 3. Introduction
Salford, UK. His main research interests are in the
Soldering technology plays an essential role in the
development of advanced tools for facilitating mass
modem electronics industry, and hence also in the
reflow soldering.
enormous range of products and disciplines that are now
2. Abstract dependent on electronics. The reliable mass production
The effectiveness of automated soldering processes used of soldered joints is thus now of paramount importance,
in the assembly of surface mount devices is very since any item of electronic equipment can only be as
dependent on the solderability of the PCB pads and the reliable as the joints that constitute its assembly. These
device leadterminations. Although there are a large concerns are emphasised by the increasing speeds of
number of factors that can impact the solderability of production and the trend towards further product
any surface, the most important consideration is known miniaturisation, which make routine inspection and
to be the wettability. The wetting characteristics of the rework more difficult. The move towards achieving
PCB pad surface by molten solder can be characterised “zero-defect’’production requires a thorough knowledge
by two key variables: the degree of wetting and the of the basic concepts of solderable surfaces, and the
speed of wetting. In this paper, we investigate the use of various techniques that can be used to test solderability
the Area Of Spread [AOS] method for the assessment of at various stages of production. There are basically two
wetting parameters which can be used as a guide to the different mass soldering methods used in the assembly
solderabilityof PCB pad surfaces in reflow soldering. In of electronic devices: the fust method involves using a
the AOS method, the wetting [and hence solderability] liquid solder bath (wave soldering), and the second
of PCB pad surface is assessed by measuring the area method involves using a pre-placed amount of solder,
over which a given mass of fluxed solder will spread on usually in form of solder paste for the reflow soldering
the sample PCB pad at a given temperature (in excess of process. This paper concerns a method for assessing the
the melting point of the solder) and in a given time. stability of the reflow soldering oven.

1999 IEEUCPMT Int‘l Electronics ManufacturingTechnology Symposium


0-7803-55024199$10.00 01999 IEEE

115

Authorized licensed use limited to: UNIVERSITY OF SALFORD. Downloaded on March 18, 2009 at 17:38 from IEEE Xplore. Restrictions apply.
The basic method currently available for monitoring the spread]. This means that wetting [and thus
reflow oven stability is thermal profiling. This involves solderability]can be improved by factors which increase
measuring the “actual” board temperatures at different &,. or by those which result in decreasing & and 4”.
stages of the reflow process and at different points on a Solderability Assessment
given substrate. The result of measuring these
Two of the most widely used techniques in indusny for
temperatures is to produce an “optimum” temperature
assessing solderability are the Solder Dip Method and
profile that meets the thermal requirements of the given
substrate material, component and solder paste type. the Wetting Balance test. These tests are based on either
the extent of wetting or the time of wetting under
Although the thermal profiling technique has the
controlled test conditions. It should be noted that
advantage of indicating the precise temperatures at the
different points on the substrate at the different stages of solderabilityis inherently dependent on the composition
the reflow process, there are two major limitations. The and condition of the surface of the base metal being
first concerns the difficulty of distinguishing between soldered and its measurement is extremely complex. It is
different soldering atmospheres and in particular the this complicated nature of solderability together with the
effect of 0 2 in inert atmosphere soldering [1,2]. The different ways in which solderability parameters can
second limitation is cost related, as the process is known interact with different soldering processes, that makes it
to be quite time consuming, tying up an expensive difficult to perform solderability tests that are fully
representativeof the real life soldering situation.
resourcesover long periods of time.
This paper presents an alternative technique, which Solder dip m e t h d
utilises the Area-Of-Spread, wettability method. The The Solder Dip Method is illustrated in Figure 3. With
method can be used as a laboratory prototyping tool and this test a small part or portion cut from a large piece of
also in production to obtain an easy and quick a non-corrosivematerial is fluxed and then dipped into a
assessment of the wettability of the base metal by the solder pot containing solder of a specified alloy at a
solder alloy, and also for assessing the efficacy of a specific temperature, for a specified period of time.
given fldvehicle system for a specific base metal and After dipping in a solder pot the test piece is removed
solder alloy combination. In its basic form, the from the solder pot, allowed to cool in air until the
technique involves heating a certain amount of solder on solder solidifies. The test piece is then examined for
a base metal in the presence of flux, and then assessing uniformity, smoothness, adherence and bedom from
the spread of the solder [both qualitative and non-wetting and dewetting. The automatic dip test was
quantitative assessment is possible]. Figure 1; give devised to eliminate the unavoidable errors in the timing
details for calculation of the Area of Spread of molten of the dipping operation. The IPC specification [3]
solder. In this way, the wettability of molten solder on a details the use of a simple cam operation device for
given substrate is used as a way of assessing providing the correct speed of entrance and withdrawal
solderability. Good wettability results in good and the correct dwell time to meet the solderability test
solderability,which means the formation of a uniform, specification for printed wiring boards.
smooth, unbroken, adherent coat of solder on the base Wetting Balance Test
metal without the need for active flux systems and the This method measures the time for the solder to wet the
associated cleaning. component termination or pad. In other words, it is
S d m e Tension and Wetting possible to measure the changes of wetting force on a
The interfacial surface forces that dictate the shape of sample as a function of time. In the wetting balance test
the molten solder are dependent on a wide range of the leaded component, after fluxing, is immersed in a
factors such as the reflow temperature, solder alloy, the bath of molten solder at 235°C. The time taken to wet
pastes flux activity and the soldering atmosphere. The the lead and the wetting force is measured and a wetting
resulting surface tensions and their effect on wetting is balance curve produced. Figure 4 demonstrates the
best illustrated by the Young’s’ relation (see Figure 2), operation of the wetting balance, which includes the
which defines the Contact Angle 8,given by: following stages.
[a]. the specimen before immersion
[b]. the specimen immediately after immersion
(upwards
force on specimen)
where:
[c]. the wetting of the specimen has reached the
K,, is the surface tension at the solid-vapour interface, point when the only force acting on the piece
& is the surface tension at the solid-liquid interface, is the buoyancy of the liquid
and [d] when the meniscus is curved upwards, wetting
equilibrium is reach on the specimen
/(Iy is the surface tension at the liquid-vapour interface.
(downward
The contact angle, Il can thus be used as a measure of force on the specimen)
wetting, with ll = 0” corresponding to perfect wetting [e&fl. the specimen has been withdrawn.
[i.e. excellent area of spread, and n > 90” signifying The test specimen is subjected to time-variant vertical
non-wetting or poor wetting [i.e. very poor area of forces made up of two main components: a buoyancy
1999 IEEWCPMT Inrl ElectronicsManufacturingTechnology Symposium
$10.00 e1999 IEEE
0-7803-5502-4/99

116

Authorized licensed use limited to: UNIVERSITY OF SALFORD. Downloaded on March 18, 2009 at 17:38 from IEEE Xplore. Restrictions apply.
force acting upwards and a force due to surface tension thermocoupleson the conveyor edge and on the smallest
acting downwards. The combined effects of these forces device.
are detected by a transducer and converted to a signal, 4. TheExperiment
which is continuously recorded to give the wetting
The experiment is based on an improved method of the
balance curve shown in Figure 4.
area-of-spread tests conducted by Muren [41, in an
Area of Spread Test attempt to monitor the effect atmospheres had on the
In this test, a uniform volume or weight of specified wetting characteristics of reflow soldering (see Figure
solder, in the form of a pallet, is placed on a test 6). The tests involved depositing solder paste onto FR4
specimen with a few drops of non-activated rosin flux. PCBs etched with copper strips (see Figure 7), placing
The part is then heated at a controlled temperature for a the sample through a reflow oven under a given
defined period in an oven. After heating, the area of atmosphere and comparing the results with the printed
spread can be measured as shown in Figure 1. The test solder paste deposit dimensions. This method has the
can also be conducted using an alternative method, by advantage of limiting the area of spread to two
printing a predetermined quantity of solder paste, using directions. However, obtaining the results from several
a stencil, onto a material (e.g. copper coupon), and samples, for statistical analysis, can be very time
placing it through a reflow oven. The reflow oven’s consuming? The test pattern used in this experiment
profile should be suitable for the given solder paste. furthers Muren’s method by increasing the number of
Although, this test does give an indication of solder deposits per copper stip at predetermined
solderability, the analysis of results can be very time deposit sizes and pitch
consuming and typically requires good vision assistance When the solder paste reaches its reflow temperature,
with precision X-Y measuring microscope. There are the solder becomes molten and wets the copper surface.
several factors that are likely to impact the results of the If the soldering conditions permit, e.g. reduced 02
AOS test, and two of the most important ones are levels or flux activity, the molten solder will spread and
discussed below. coalesce with its neighbouring molten solder deposit.
Oven repeatability Therefore, the greater the pitch covered by the molten
One of the most important considerations is the oven solder the greater the ma-of-spread the given
repeatability, which refers to the reflow oven’s conditions produce. Hence, many bridges indicate good
capability to repeat a given profile. The oven wetting and conversely, no bridging will indicate poor
repeatability is inherently dependent on the machine wetting. Producing a test specimen that bridges in N2
loading and load factor. Figure 3 and produces no bridges in air atmospheresis sufficient
Board Length (Lb) for a qualitative evaluation [5].However, if a more
Load Factor (FL)= quantitative area-of-spread evaluation is required then
& + B o a r d Gap ( G b ) certain properties of soldering must be controlled. In
this experiment a test substrate was design with different
illustrates the concept of “Load factor”. It should be widths of copper lands and stencil aperture sizes. The
noted that the higher the load factor, the more difficult it test substrate, using a stencil, is printed with a given
is for an oven to produce repeatable results. In practice, solder paste and then reflowed under a set of conditions.
reflow oven manufacturerstypically specify a maximum The test substrate was design to allow for various
loading factor of between 0.5 to 0.9. wetting levels, from influences such as atmospheresand
flux activity, and the monitoring of left and right
The issue of the loading factor becomes important when
temperature difference in a reflow oven. For statistical
sizing a reflow oven, for example determining the
stability each pattern is replicated five times, giving a
required reflow oven capacity. The non-unifopity can
maximum of fifty possible bridges per pattern.
be measured by conducting a thermal profile on a test
board, whilst loading the oven - supplying the oven The monitoring of the left and right temperature
with a regular flow of substrates, periodically differences is achieved by replicating the design on one
conducting a profile (with the test board) within the board. This replication of test patterns on the left and
oven load and comparingit with the unloaded profile. right of the test board enables the detection of thermal
differences effecting both sides of a reflow furnace. If
Oven UnifOnnizy the temperature is the same on both sides of the reflow
Another major source of concern is Oven Uniformity, oven then the values obtain for the left and right patterns
which causes non-uniform product heating. This can be should be equal. However, if the values differ then there
caused by edge effects on the heater (less heat at the may be discrepanciesin the left and right temperature of
end) or by the conveyor system sinking heat. The edge the reflow oven.
effect may be caused by insufficient airflow around the
Monitoring wetting levels is achieved by producing
edges of the machine, or by non-uniformities that are
three different land sizes in 30 thousandth of inch
inherent in the heater. The heat sinking of the conveyor
(0.03”)increments (30, 60 and 90). The aperture pitch
system is a mass effect, which may be exacerbated by
(P) distances are set at J increments of the
product mass differential.Oven uniformity and potential
corresponding land sizes multiplied by the pattern
mass differential can be measured by placing
number (1,2 or 3). For example, the pitch distances of
the 0.03” land are of 30 (IO), ’13 of 30 (20) and 3/3 of
1999 IEEUCPMT lnrl ElectronicsManufacturing Technology Symposium
0-7803-5502499 $10.00 0 1999 IEEE

117

Authorized licensed use limited to: UNIVERSITY OF SALFORD. Downloaded on March 18, 2009 at 17:38 from IEEE Xplore. Restrictions apply.
30 (30). The land size and the aperture pitch values are Type B: Low Flux activity, produce a less S value than
then multiplied to produce a coefficient value (A). The type A solder paste.
delta value corresponds to the total visible wetting area As a method of preliminarily evaluating the test four
available before reflow. This value is used to offset the boards for each solder paste type were placed through a
bridge value of each pattern reading. Hence, a pattern reflow oven at different 0 2 levels (Air - 50ppm). This
with a greater area-of-spread will signify a greater evaluation is based on hypothesis outlined above. If this
wetting influence than that of a smaller area-of-spread. stands true then the S value for type B solder paste
For example, if a test board is placed through a specific should be some degree lower than that of type A. The
reflow condition and 50 bridges are produced on a purpose of this test was to see dependence of solder
pattern with a A value of 3 (A3 = [10x30]/100) and on paste flux activity, by measuring the S factor produced
the same test board 40 bridges are produced on a pattern with a given set of conditions. The test was conducted
with a A value of 54 (h54) then, for this specific set of using a thermal profile suitable for both solder pastes in
conditions, it can be assumed that the process effects the air to 5Oppm oxygen levels in a Relm N2reflow furnace.
wetting more than A3 and less than A54. In other 4.2 Using the Test to Measure the Effect of
words, the solder wets a surface area 5400<300 Different 0,Levels
thousand of an inch. This test will produce two sets of results the fmt
Producing several patterns with different stencil aperture involves the generation of an 0 2 threshold void point
sizes and land pitches, shown in table 1, would, produce and the second test will measure and compare the
a scale of area-of-spread that could be read and used to variation in tests conducted in air and in Nz. Due to the
measure a process stability. Placing a test board through nature of operation of the reflow oven, different oxygen
a set of reflow condition, an area-of-spread value (2) levels can not be maintained. The reflow oven used was
can be produced. This parameter Z is calculated from designed to produce a control level set at 5Oppm 02.
the sum of the A values and bridge count (X)products, This problem was overcome by placing test board
shown in equation 1. through the oven at regular intervals (2 minutes) whilst
z==Amlx~l+A&m2+Arn3xrn3.. .Amnxmn switching on the N2 gas control and logging the oxygen
levels with a precision gas monitor. Admittedly, this is
.........U1 not the most accurate method of producing controlled
where m corresponds to the left or the right side pattern oxygen levels due to the fact that it can only be falsely
of the test PCB (L or R) and n is the pattern number (0- assumed that the oxygen levels recorded at the point of
9). The resultant figure from this calculation is then board entry will be the actual level at the board’s reflow
converted into a percentage, in effect producing a point. In fact, the oxygen level will increase whilst the
percentage AOS, given by: board is being conveyed through the oven.
s = -Z 100 It is a well-proven fact that Nz atmospheres increase
soldering yields by reducing the 0 2 levels in the
Zm,,
........................................ PI atmosphere. Therefore, conducting the area-of-spread
2- is the maximum value of Z; this is obtained by
experiment in various levels of O2 with different solder
pastes could provide valuable information on void
setting the X values in to 50
thresholds. For example, at what level of 0 2 can the
furnace drop before the wetting of a surface is affected
4.I Evdualing the Efficacy of the Test to a point at which voids are probable? Knowledge of
Over the years there has been much work conducted in this threshold point can benefit the manufacture with the
the area of nitrogen atmosphere for reflow processes implementation of the appropriate process control
[6,7] as a method of reducing oxide growth on metallic systems. The oxygen levels can be monitors on most
parts caused by the oxygen content in the air reflow furnaces. This can be utilised by setting the
atmosphere. Tests have been conducted [7] that threshold point on a control chart and thus preventing
demonstrate an increase in wettability with the reduction O2 levels falling below the solder pastes “voiding”
of oxygen. Hence, it can be assumed that if the test is region.
conducted in a low level of oxygen then the number of The second results produced from the data are used to
solder bridges on area-of-spread tests conducted in a N2 compare the variation in tests conducted in air and N2.
atmosphere should be greater than that of tests The comparisons were made two sets of results, each
conducted in an air atmosphere. made up of 10 runs.
As a method of measuring the efficacy of the test two 4.3 Straight Runs in Air and N2
sets of test were conducted using two types of solder The final tests involved producing two runs of test
paste, type A and type B, with known levels of flux substrates, one conducted in air and the second in N2
activity. (50ppm 02).The result of these tests will be compared
Hypothesis: to highlight the differences in 0 2 levels and variation.
Type A: High Flux activity, produce a greater S value 5. RSdtS
than type B solder pastes [where A is 50% more active The results of the experiments are summarised below.
than B].
5.1 Evaluating the solderpaste scale of S
1999 IEEWCPMT Inrl Electronics ManufacturingTechnology Symposium
0-7803-5502499$10.0001999 IEEE

118

Authorized licensed use limited to: UNIVERSITY OF SALFORD. Downloaded on March 18, 2009 at 17:38 from IEEE Xplore. Restrictions apply.
The results of four test substrates per solder paste type figure 10, there is one point that falls out of control; this
are shown in figure 8. As stated in the hypothesis, in the could be accounted to experimental error and therefore
previous section, solder paste A will produce a greater classes as an outliner. On the N 2 X-bar chart, figure 10,
area-of-spread than solder paste B. From the results three points are out of control indicating excessive
shown in figure 8 it is obvious that this hypothesis is true variation in the processes.
and thus demonstrate that flux activity affects the AOS 6. s-ry
value, S. A second point noticeable on figure 8 is the
A method of using the Area Of Spread test for
effect low levels of O2 have on the percent area-of- monitoring the stability of the reflow soldering process
spread (S) of solder paste type A. For the next tests, the
has been demonstrated. In the AOS test, the
value of the 50ppm 0 2 reading should be below SlOO
solderability of a base metal is assessed by measuring
(a value around S80, this is to allow for the variation of
the area over which a given mass of solder paste spreads
the process to be reflected on the S scale. It was for this
over the test specimen, at a given temperature in excess
reason solder paste type A was not used for further test of the melting point of the solder and in a given time.
and consequentlyonly solder paste type B was used.
The test has been adapted and used for assessing the
Using the results from this test a correlation factor could stability of the reflow soldering process. The paper also
be produced kom the gradient of the graph shown in presents results of tests on the effect of flux activity and
figure 8. This value could be used for comparing a soldering atmosphere on wettability. Our results show
solder paste’s response in 0 2 . However, this assumes that the method is quite unstable for the small sample
linearity which is not always true in practice. It is also size used for the experiments.
noticeable from the graph that there is a sigruficant To reduce this large variation, we suggest new method
difference between the left and the right hand sides of for restricting the flow of solder to only one direction on
the test’substrates, and this is likely due to the the test sample. For example, rather that placing ten
temperature differences in the reflow furnace. solder paste deposits on one copper land, we
5.2 Monitoring the effect of different 0,levels recommend using two solder paste deposits and
using the area-ofspread test restricting the flow of molten solder to one direction, as
As mention, sdder paste type A produced results that shown in figure 12. This will help to minimise the effect
were “off-scale”, possibly due to the higher activity of of the interaction between neighbouring solder deposits,
the flux content in the solder paste, and therefore only which reshcts free flow as the molten solder coalesces.
solder paste type B was used,
This test involved placing fifty test boards, stencil REFERENCES
printed, through a range of oxygen levels (Air-5Oppm). 1. Manard Pierre J, “Statistical Process Control
On placing the 12* board through the reflow oven the of Reflow Soldering”, SMT Magazine, June
N2 was switched on and switch off at the 34* board, 1996.
shown by the dotted line on the graph in figure 9. The 2. Klein de F J, ‘‘Important Issues in Reflow
graph also shows the oxygen levels at each point when Soldering”,
the board were placed in the reflow oven.
Soldering and Surface Mount Technology, Vol. 16,
As shown in the graph, figure 9, when the N2 is switched
No 2,1994
on then the S value increases, and conversely when the
N2 is switched off the S value reduces. In other word, 3. PC-S-801
the area-of-spread test indicates when there is a change 4. Muren D, “Atmosphere Influence on Reflow
in O2 levels. However, there is visually large amount of Soldering”,
variation in air compared to the test boards conducted in AGA ABB Report, REPM 94332, circa 1994.
Nz. Review of published work on inert reflow soldering Holm, Torsten “Controlled atmospheres: Soldering
5.
atmosphere show that there are a number of benefit in
of printed circuit boards”, Private Communication,
reducing 0 2 levels namely, an increased process AGA ABB,Sweden, 1993.
window and reduced variation in the level of quality.
This could aid in the explanation in the large variation 6. Dong, C C., Dawson, A S, Chen, M S K, “Effects
shown, in fi,pre 9, with test boards conducted in air of Trace Oxygen in Bulk Nitrogen on Solder
compared to the N 2 boards. Wettability”, hoc. IPC Printed Circuits Expo,
Boston, MASS. 1994
The data from the air and N2 test board has been
separated and each have been place in a X-bar R, with a 7. C, Lea, “RelationshipBetween Oxygen Level in an
subgroup of 2 that corresponds to two tests (left and Infrared Reflow Oven & Soldering Quality”,
right tests), charts shoyn below in fi,pre 10 and 11. National Physical Laboratory, 7” November, 1991
The first noticeable point about the two sets of X-bar R
charts (figure 10 & 11) is that there is more points out of
control on the N2 X-bar chart compared with that of the
air X-bar chart. On the air atmosphere X-bar chart,

1999 IEEVCPMT Int‘l Electronics Manufacturing Technology Symposium


0-7803-5502499 $10.00 01999 IEEE

119

Authorized licensed use limited to: UNIVERSITY OF SALFORD. Downloaded on March 18, 2009 at 17:38 from IEEE Xplore. Restrictions apply.

You might also like