Minimizing defects on board assemblies by reflow profile adjustments.

Obviously, the soldering profile created in the reflow oven is one of the most important factors in determining defect rates in the SMT manufacturing process. To gain or to maintain a good process realising the necessarily high yield, board assemblers have to look carefully and eventually optimise plenty of adjustments.

Faults on board assemblies, which are directly influenced by the reflow profile include:
component cracking, tombstoning, bridging, wicking, solder beading, cold joints, formation of excessive intermetallics, poor wetting, voiding, skewing, charring, delamination, etc.
Modifications of the currently accepted profile can minimize these defects.

	Traditionally, in the SMT industry the profile consists of: the preheat, during which the paste is heated at a rate of 2 to 4°C Is, then followed by a "soak" zone at 150 to 170°C for 60s or so, after which reflow occurs, typically with a peak temperature of about + 220°C (figure 1).
Fig 1. Conventional reflow profile

It is often supposed that the profile must have this shape because of the chemistry of solder pastes. That a fast preheat is necessary to burn off solvents, and soak is required to allow the flux to work. But this is not so.
This type of profile, which is still very widely used, came about largely because of the limitations of the method of heating, which was commonly used; namely infrared ovens.

IR reflow provided perfect satisfactory results, but this technology had certain limiting factors:

• Sensitivity towards uneven thermal mass distribution.
• Differential heating of differently coloured parts (dark components would get hot before the greenish PCB)
• Shadow effect around large components.

As a result, a considerable temperature gradient was apparent across the board.
For this reason, a fast rate of heating in preheat followed by a long soak at about +150°C became the preferred profile. Once thermal equilibrium had been reached across the board, reflow could take place.

Flux reaction usually takes place very quickly - this can be easily demonstrated by the simple wetting test.
In this test a small circle of solder paste is printed onto an aged, oxidised copper coupon. The copper coupon is then placed upon a hot plate and reflowed. Typically, the whole process of flux reaction, coalescence and wetting takes place in less than 5s. Consequently, this fast flux reaction time means that a long soak zone (at least from the paste formulator's point of view) is not required.

This traditional reflow profile has potential to cause defects, particularly because of the fast rate of heating in the preheat zone, which can be as high as 4°C/s.
The viscosity of materials with a fixed composition and chemical structure drops as the temperature rises. This is because of greater thermal agitation at the molecular level. This decrease in viscosity will naturally cause the material to spread, or slump.
Thermal agitation is a material property; it is purely temperature related and independent of time - for this reason the ramp rate (the rate of temperature increase) will have no effect.
However, with solder paste there is another factor at work, namely solvent loss. As solvents evaporate, the viscosity of the material will increase.
This will counter the effect of thermal agitation and limit slumping.
The rate of solvent evaporation is both temperature and time dependent, so this can be regulated by the rate of temperature increase in preheat.
With a slow rate of temperature increase, loss of viscosity due to thermal agitation will be counteracted by solvent loss, which will in turn tend to cause the viscosity to increase. (This is shown in figures 2 and 3.)

	Figure 2.: The effect of thermal agitation and solvent loss on viscosity as a function of temperature.
Fig 2.

	Figure 3.: Relation between ramp-up rate and viscosity due to solvent loss effect.
Fig 3.

Slump is the direct cause of many reflow-related defects, particularly solder beading.
Solder beading is caused directly by solvent outgassing in the preheat stage. If the ramp-rate is high, say 3 to 4K/s, the solvents in the paste will not gently diffuse out of the paste deposit - they will erupt out of the deposit. This outgassing force overcomes the cohesive force in the paste and isolated aggregates of paste are forced under the component.
At reflow this paste melts and coalesces into a ball at the side of the component (see figure 4 for a comparison of ramp rate versus slump).

	Figure 4.: Relation between, slump and ramp-rate. (The easiest way for the production engineer to minimize this tendency to slump is to reduce the ramp-rate. )
Fig 4.

• One is to use low-boiling point solvents; these will evaporate very quickly, thus minimizing slumping and solder beading. However, using low boiling point solvents has another effect - it reduces the stencil life of the paste, thereby increasing material costs through wastage.
• The second approach is to use fluxes with a very low activation temperature. This means the flux will clean the oxides off the powder at low temperature - this will allow the powder in the solder paste to cold-weld at low temperatures, thereby increasing the viscosity of the paste and reducing the likelihood of slump. This means that the shelf life of the product is reduced (because the activators will attack the powder at room temperature) and that the product may fail long term reliability tests due to the highly aggressive nature of the flux.

Experimental studies have shown that a ramp rate of 0.5-1.0K/s from room temperature to melting temperature is best.

Such a slow linear ramp rate, without the traditional long "soak" zone is possible due to modern oven technology.
Today, forced air convection ovens offer a fast, controllable rate of heating, which is not sensitive to variations in component colour, shadow effect, etc.
The fast even heating, which the convection oven provides, eliminates the large temperature gradient across the PCB that was so common with IR application, and therefore the reason for a long soak zone. This allows the optimum profile shown in figure 5.

	Figure 5.: Optimized profile versus conventional. (This optimum profile is also better for the long-term reliability of boards and components. There is less thermal shock due to the gentle rate of temperature change and less board stress due to lower total heat input (which equates to the area under the curve)
Fig 5.