By the time you have completed your study of this topic, you should be able to understand and explain:
You should also be able to identify some common defects associated with the reflow soldering process that are of particular importance to people working in Electronic Design Realisation.
Reflow involves heating the assembly of board plus components held by solder paste through successively higher temperatures:
Finally, the assembly is cooled, to solidify the solder joints and make it possible to handle the board.
The process can be carried out in a several ways, but volume production is normally performed on a continuous belt oven: Figure 1 shows a typical longitudinal oven in cross-section. Frequent reference is made to the ‘zones’ into which such an oven is split, but this is a word one must be rather careful about, because there are actually two ways in which it is used:
To make things more confusing, normal practice is to count only the heated zones, and not include the ‘cooling section’. Zones are normally joined together, but some designs use ‘resting zones’ (to allow temperatures to stabilise across the work). In most ovens there will not be a significant physical barrier between zones, devices such as ‘gas barriers’ and ‘air knives’ are used to produce a thermal gap between adjacent zones.
This first step involves a rapid rise in temperature, which evaporates the solvent from the paste and burns off the largest amount of contaminants. This generally occurs in the first heated zone of the oven.
The next heated zones are programmed to provide a uniform temperature, with the goal of fully preheating the assembly to a temperature generally between 100°C and 150°C. The most critical parameter is the rate of rise of temperature, which should be less than 2°C/s, to avoid solder paste spatter and minimise thermal shock on the components; multilayer ceramic chip capacitors can be vulnerable to cracking if heated too fast.
This section, also referred to as the dry-out, soak, preflow or activation zone, ensures that the solder paste is fully dried before hitting reflow temperatures, and acts as a flux activation zone for many solder pastes. Pre-heating may also cause some slumping of the paste, depending on the specification and quality of the solder paste used.
An important aim is to ensure that all joints stabilise at the dwell temperature. Smaller joints hotter than the dwell temperature will cool down, while larger joints heat up. At the end of the dwell time, all joints are intended to be at the same temperature. In general, the tighter the temperature spread is at the end of the preheat zone, the smaller it will be at the top of the reflow spike zone.
In most designs
The final heated zones are used to provide a spike (rapid rise) in temperature, when the solder paste reflows and wets the surfaces of both component and board pads. Surface tension will determine the shape of the joints and in some cases the surface tension forces during wetting can generate component movement.
Video clip: Paste reflowing and wetting a J-lead termination. Caution: this file is 1.492Mbyte
Solder reflow starts happening when the paste is taken to a temperature above the melting point of the solder, but this temperature must be exceeded by approximately 20°C to ensure quality reflow.
The length of time the solder joint is above the liquidus is referred to as the ‘wetting time’. This is normally 30–60s for most pastes. If the wetting time is excessive, intermetallics may occur in the joint, which result in brittle solder joints.
Once the product has reached the end of the heated zones, the process involves ‘ramping down’, or slowly cooling, until the assembly reaches a suitable temperature. The first stage down to the liquidus temperature of the solder is critical, and solder is weak mechanically above 150°C, so care has to be taken to avoid rapid changes of temperature, draughts, etc. Final cooling is both to reduce possible oxidation and make handling safe!
[ back to top ]
There are four heating modes involved with SMT reflow processes: conduction, latent heat, infrared radiation (IR) and convection.
Heat transfer by conduction occurs when two solid masses at different temperature are in contact, and within the same mass if a temperature differential exists within it. Reflow systems using conduction include:
All these are relatively small-scale processes, and only the last can be applied to SM assemblies on a PCB, where ‘hot bar’ soldering is occasionally used for rework or for the assembly of large complex devices.
Latent heat of evaporation is the heat which is absorbed by a fluid when it turns into vapour and released when the vapour recondenses. This mode is used by the ‘condensation soldering’ process.
Radiation: infrared (IR) radiation is emitted by all bodies and heat transfer occurs whenever two bodies at different temperatures are in sight of each other. IR is a non-contact heat transfer process that was used in the earliest designs of reflow oven, but now coexists with, and has mostly been replaced by, convection heating.
These last three methods of heat transfer can be implemented at all levels upwards from low-volume batch systems. The sections that follow concentrate on convection (the main runner for larger users) and on high-volume flow-line systems, where a tight level of process control needs to be exercised.
In what used to be called vapour phase soldering, the component and board are immersed in the vapour from a boiling liquid. As liquid condenses onto the (comparatively) cool surface of the assembly, its latent heat is transferred to the board, raising the temperature of the product to the boiling temperature of the liquid.
There are two types of fluids currently sold for condensation soldering:
Both types of material are chemically inert, and have high dielectric strength and low viscosity. They are non-toxic, and have no flash or fire point, so operator safety concerns are confined to the vapour being at high temperature. From the environmental perspective, both have zero ozone depletion potential, and are not regulated as volatile organic compounds (VOCs) as they do not contribute to ground-level smog formation. However, perfluorocarbons have high global warming potentials (GWP) and long atmospheric lifetimes, and need to carefully managed in order to minimise emissions. Although not stated by the manufacturer, perfluoropolyethers appear to have acceptably low GWP.
The boiling point of these fluids may be selected for the application: the commonly used Fluorinert FC-70 and Galden LS/215 both boil at 215°C, but perfluoropolyethers are available with boiling points up to 260°C.
Machines provide successive dwells, first in a preheating zone, where the assembly is raised to 100–160°C, and then (usually for about 1 min.) in the ‘primary vapour’. With batch machines (Figure 2), where the preheat is provided by a cooler ‘secondary vapour’ of a compatible fluid, this also performs the post-solder cooling.
Although in-line systems are available, the batch process is more common as it is considerably more economic to run: the materials used are expensive and there can be substantial vapour loss. Vapour phase soldering lost much of its popularity in the early 1990s due to its high running costs, uncertainty about achievable assembly yields, and environmental concerns; relaunched as condensation soldering in the late 1990s, the process is gaining ground, especially with the advent of lead-free solders. The technique provides good control of the maximum assembly temperature, and is particularly suited to heavily-populated multilayer boards.
All objects emit infrared to some extent: the level of energy emitted and the wavelength of the emission both depend on the absolute temperature of the object. As the temperature increases, the heat transfer output increases exponentially to the fourth power (Figure 3). Increasing the source temperature results in shorter wavelengths; decreasing the temperature results in longer wavelengths.
When operating at the low heating rates required for SMT, all types of radiant emitters (lamps, panels, tubes) operate in the medium to long wavelength part of the spectrum, from 1.5 µm–1000 µm wavelength. This minimises board discoloration and colour sensitivity.
Infrared had been used for many years for reflowing tin-lead plating. It was a familiar system and relatively inexpensive, and in the early 1980s started to be used for solder paste reflow. The board is placed on either a mesh or pin conveyor system which transports the board through the heating and cooling zones.
The general equation for heat transfer Q (W.cm2) between a source at TsK and a target at TtK is:
where K is the Stefan Boltzmann constant (5.67x1012 W.cm2.K4).
The geometric view factor V is the fraction of the energy leaving the source that hits the target. In SMT reflow, oven chamber designs yield very high view factors in the range 0.90–0.95, but this can be reduced locally if two very large components are near each other.
The source emissivity Es and the target absorptivity At factors are in the range 0.90–0.95 for most SMT applications, the board material, solder paste, and components all absorbing infrared energy quite well, although shiny gold components may be difficult to heat.
The implications of the equation are that:
If the overall emission can be regulated, infrared can provide high levels of repeatability, although in practice, due to variations in board layout, thermal mass, and reflectivity of the board and components, temperature control was always difficult to achieve. Infrared ovens are particularly sensitive to thermal loading: if the mass of products being processed varies, the process parameters will need modification. For this reason, convection has become the method of choice for most users.
Forced convection reflow was developed during the late 1980s as an alternative to infrared heating, but using a very similar constructional concept. Despite their higher cost, both to purchase and run (especially when reflowing in a nitrogen atmosphere) they became popular, because blowing hot gas at the circuit (Figure 4) provided a more controllable and consistent heating regime.
Convection heat transfer occurs when a fluid contacts a solid mass at a different temperature, the heat absorbed per unit area Q (W.cm2) being given by:
The amount of heat transferred, Q, can be modified by changing either the convective film coefficient H or the temperature difference DT.
The convective coefficient H can have a significant impact on the heat transfer, and its value is related to a number of variables, including the flow velocity and its ‘angle of attack’. Increasing the velocity of the flow increases the amount of heat transferred, but there are practical limits to this, as too high a velocity may cause components to shift.
The flow direction also has a significant impact on Q. In true parallel flow, there is a ‘boundary layer’ in which there is no fluid motion, so convective heat transfer rates are very low. It is beneficial to break up the boundary layer to allow the flow to transfer heat freely to the object, and perpendicular flow yields best heating or cooling results because there is no boundary layer.
The ability of gases to transfer heat is limited by their low specific heat, so convection systems always need a high gas flow. It would not be cost-effective to use the gas only once, and practical systems always recycle gas in a closed loop within individual zones. Depending on design, around 10–25% of the gas used will be sourced externally, with the rest recirculated.
In order to improve efficiency further and reduce the amount of process gas required, some ovens also have extended recirculation loops, where spent process gas from hot zones is used to carry out preliminary heating. However, this may contaminate the assemblies with soldering by-products, so in most systems the flow of air within the oven is from ambient to the reflow zone, and the main extract occurs there.
The major benefit of convection ovens over infrared systems is that the desired product temperature is close to the flow gas temperature and can never exceed it. This is useful when heating an assembly with a large range of component masses, as larger components will continue to heat, while the smaller components will not overheat.
Convection ovens can also tolerate changes in thermal loading. By maintaining the temperature constant and increasing the air flow, higher mass components can rapidly be elevated to the same temperature as the small devices.
Nevertheless, infrared still plays a role in the heating process: the only way to achieve a 100% convection system would be to have gas flowing over the product with no chamber walls! Practical ‘convection’ systems have convection:infrared ratios in the range of 70:30 to 85:15. Conduction also occurs to spread heat through the product.
Somewhere in the loop between gas collection from the chamber and delivery to the orifice plate it will be heated. It is a point of debate as to which position gives the most even results, and much depends on the efficiency of any in-plenum diffuser.
Heaters need to be robust, and designed to give even heat and maximum surface area (and thus access to the convective gas) in order to maximise heat transfer. This allows the oven to respond quickly to any changes in demand caused by changes in oven loading. Most are also designed to have a small turn-on response time lag (typically under 15 minutes from cold to reflow) which allows an oven to be switched off when not in use. This has implications both for the design (machine components must have a low thermal mass) and for peak power consumption.
The board transport system may use either a conveyor belt or, more usually with larger ovens, some form of edge conveyor system, often with a mesh belt underneath to catch dropped boards and components! A chain conveyor, for example, has pins protruding from the sides of the chain to provide an edge support that carries the board through the reflow system without touching the belt. This allows double-sided PCBs to be processed without disturbing the bottom-side components.
Proper design of the edge conveyor system is critical, as typically only 4 mm is in contact with the bottom-side edges of the product. If the conveyor has lateral warpage of greater than this amount, the product will fall into the oven. Thus, edge conveyor straightness is a key issue in SMT processing. In fact, some suppliers even go to the extent of building in a tensioning system to pull the rail assembly straight and counteract the tendency to grow and warp during the machine warm-up period.
Reflow soldering takes most laminates above their glass transition temperatures, so that they become less rigid. An unsupported board may therefore sag significantly during reflow, and remain bowed after soldering. This makes it difficult to mount into any enclosure, and attempts to flatten the board will stress the components.
However, supporting the centre of the board to prevent sagging during reflow means either using thermally inefficient board carriers or devising a more complex conveyor system. Unfortunately, panels are in general getting larger and heavier yet thinner. Also, as larger components and longer connectors are used, sagging can be a significant problem, leading to open-circuit joints after reflow. These factors are promoting designs of centre-support system, such as tensioned moving wires, that require a small ‘highway’ on the board to be kept clear of components.
The aim of the cooling system is to ensure that the product is below liquidus temperature before exiting the oven. An initial cooling jet between reflow zone and cooling stage can dramatically reduce the time above liquidus providing a tighter grain structure and better solder joint. The main product cooling however is usually based on a water-to-atmosphere heat exchanger, where gas convection cools the product from both top and bottom. Issues relating to this stage include:
One of your colleagues has commented that reflow ovens look like black boxes, taking in pasted assemblies, and producing soldered assemblies. Explain to him/her the process that is going on inside the oven, and how the process is normally carried out.
[ back to top ]
High quality, low defect soldering requires identifying and repeating the optimum reflow process: every solder joint on every board needs to be treated in exactly the same way. How the heat is applied does not matter as long as it is in a controlled manner. Heating and cooling ramp rates must be compatible with both solder paste and the assembly, and the time of exposure to high temperatures must be defined and maintained. Suppliers of solder paste will recommend different process conditions for their products to obtain optimum performance, particularly with low residue pastes.
The aim is to raise the temperature of the board and component terminations sufficiently far above the reflow temperature of the solder paste for the right length of time to create reliable solder joints. The process has to be
If the reflow temperature of a paste is not achieved, the paste will not become a liquid; if the paste is not at reflow temperature for a sufficient time, full wetting will not take place.
We also need to ensure that both component and board reach the reflow temperature at almost the same time in order to get satisfactory joints.
As an assembly moves through a soldering system, it is exposed to a controlled rise and fall in temperature. A plot of temperature against time is called a ‘temperature profile’. A typical target profile for reflow soldering indicating both temperature rise and fall and the time above liquidus is shown for an RMA paste in Figure 6. Figure 7 is a target profile for a reduced solids paste, which is able to tolerate less preheat.
The ideal temperature profile is usually based on three factors: peak temperature; maximum rate of change of temperature; time above liquidus.
Peak temperature. For a solder with a melting point of 179–183°C, the maximum allowable peak temperature is usually 220–230°C, and the minimum peak 195–205°C. If the board gets too hot:
If the assembly does not get hot enough, the solder paste will not reflow adequately.
The maximum slope or ‘ramp rate’ specifies how fast the assembly temperature is allowed to change. Many components, especially chip ceramic capacitors, will crack if their temperature is changed too quickly. In order to maximise throughput, the thermal profile is usually designed to have a slope just under the maximum allowable, between 1–3°C/s.
The negative slope (‘ramp down’) on cooling should theoretically be the same as the heat-up rate, but most companies have profiles with higher negative slopes, especially with ovens that have a short cooling zone. Perhaps, as seems to be the case with vapour phase systems, the ramp rate may be less critical above certain temperatures.
Most companies aim for a time above liquidus of 30–60 seconds, but 90–120 seconds is more typical of larger assemblies. This high minimum allows a margin of safety against both oven temperature drops and the chance that the coolest spot on the assembly has not been located during set-up. However, most solder experts believe that the shorter the time above liquidus the better, as the growth of tin–copper intermetallics leads to a tin-depleted and brittle solder joint.
In practice, an ‘ideal’ profile is impossible to achieve over a range of components, and a temperature profile is best displayed as a band of temperatures (Figure 8), within which all joints should be maintained. So long as no joint temperature actually falls outside the profile band, it can be assumed fairly safely that an assembly has been properly soldered.
Thermocouples are made of wires of two dissimilar materials that are welded at one end to form the monitoring junction and then fixed to the board or component under test using adhesive or high melting point solder. As its temperature is increased, the wire junction produces a voltage that is proportional to the temperature. This output can be converted to temperature and may be shown on an analogue or digital display. Connected to a chart recorder, the thermocouple can provide a true profile when it has been calibrated against time and paper speed. But because the thermocouple has to be attached at one end to the board or component, and at the other end to the monitoring equipment, the set-up can be very cumbersome.
Profile units are self-contained devices that pass through the complete soldering equipment, with multiple inputs for thermocouple cables that may be connected to the board or components under test. Profile units are designed either to record and hold or transmit temperature profiles from each thermocouple: the information can then be transferred to a computer for analysis. A thermal shield over the transmitter is intended to maintain the internal temperature of the transmitter at below 100°C.
It is unlikely that all the thermocouples on an assembly will enter the oven at the same time. On a chart recorder, this causes a separation of the resulting temperature plots against time and is somewhat confusing. The graph becomes easier to understand by plotting the same data as temperature against distance travelled, and this facility is standard on the analysis packages that accompany profile units.
A reflow system has a number of heating zones that can be independently set to allow accurate control of temperature as the board passes through the oven. However, a plot of the set-point temperature of each of the heating zones is not the same as a profile of the average temperature experience of the board. And each solder joint on the board will have a slightly different temperature experience. The reasons for this lie in thermodynamics:
It is important that you understand this last point. The rise in temperature ΔT of an object absorbing heat Q (W.cm–2) is determined by
where A is the exposed area (cm2), t is the heating dwell time (seconds), and the specific heat Cp (W.s/kg.°C). The product of the mass M and the specific heat is what is often referred to as the ‘thermal mass’.
You can see from the equation that components with greater thermal mass require more heat input to achieve a given temperature rise. Sections of an assembly populated with PLCCs, for example, will have considerably more thermal mass per unit area than sections with discrete components, and will therefore be slower to heat up.
The average profile is affected by conveyor speed, zone temperature, convection flow rate, and oven extraction. Although these are less significant than formerly, the average profile is also affected by the total thermal mass of product being processed, by its spatial distribution, and by the ambient temperature.
The profile for an individual joint will however differ from the average, sometimes substantially, because it is also affected by the mass of the associated component, the proximity and mass of neighbouring components, the size of the pads, and the amount of heat that travels through the tracks and board.
Some temperature differential is inevitable, as it will always be easier to heat an area that has no components than an area full of large components. However, in order to achieve high quality, low defect soldering results, the temperature experience of all the joints must fall within the defined reflow profile band. It is the task of initial profiling of the assembly to ensure this by adjusting the oven parameters appropriately. Thermocouples will be positioned to find the highest and lowest peak temperatures, in other words, to check the extremes of the profile. The assembly process engineer looks for specific types of non-uniformity:
The last two of these mostly reflect the particular equipment being used, and would normally be checked only as part of the initial oven proving exercise. However, each assembly is subtly different, and profiling new designs is a precaution that is nearly always taken.
Some general observations on machine settings are that:
In summary, the initial profile is crucially important because:
Continued temperature profiling is a key area for quality control, as it establishes how the temperature experience changes with time for a particular board design and a given set of furnace settings. And here the profiler software can help, by assessing results against preset limits or average actual values as an in-process quality control check, rather than just producing temperature curves and measuring peak temperatures and time above liquidus.
“One of the problems with reflow soldering is how to get the coldest soldering locations on a printed wiring board above the melting point of your solder paste, while the hottest locations do not exceed the maximum specified temperature for reflow soldering”.
Jeroen Schmits of Soltec
How do your partners in surface mount assembly ensure that all the components on your board enjoy as nearly as possible the same ‘temperature experience’, thus creating satisfactory joints?
[ back to top ]
The key issues with double-sided reflow are that:
The primary fault seen is therefore dropped components. The design should, where possible, be assembled with the heavier components, especially those with high mass joint area ratios, on the second side to be reflowed.
As far as the process is concerned, care should be taken:
One of the strategies normally considered is the use of an inert soldering atmosphere, to reduce oxidation and the demands made of the flux, and to increase the process window.
One common misconception is that independent top and bottom heat control will allow double-sided reflow without reflowing the top side twice. This does not work with most products because they are thermally too thin. Whilst independent top/bottom heating control can reduce product warpage, and does help in machines with dual conveyors, most machines have this feature, but with top and bottom heaters set at the same value!
Make a list of as many as possible of the things that can go wrong with the reflow soldering process. To which of these might your design contribute?
[ back to top ]
Author: Martin Tarr
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 2.0 Licence.
Terms and conditions apply.