There is much "prior art" on the subject of tombstoning and many different variables and causes. Here are a couple of articles that nicely and concisely summarize the variables and offer solutions.

From SMTnet, is an EFD article, Tombstone Troubleshooting

The one from the Circuitnet site is especially nice as it gives engineering analysis showing the forces involved which can give you greater understanding and insight as to the mechanics of tombstoning.

This can be caused by many factors:

• To rapid a rise in the preheat section of his oven.
• Solder paste deposition inconsistencies from one side to the other.
• Components not be centered between the solder pads.

Tombstoning can be designed out with Stencil/Pad layout and turning the PCB in reflow so both sides of the component receive equal heat but to be honest all these are possible but hard work.

Simplest way is start to use a Low Tombstoning paste especially if used in Vapour Phase Machine. This is the simplest answer to the question. It is generally ran on a SAC305 profile but has a mix of alloys to stop Tombstoning or greatly reduce it.

I know AIM Solders do one that works very well and probably others do as well so best to search out Low Tombstoning Paste on the Internet and give it a try.

The cause of this defect is inadequate contact surface area of the component metalized pads in the solder paste.

This could be caused by a miss alignment of the component in the printed solder paste from poor placement, poor registration of the stencil on the pad, or less than adequate component metalization surface area in the paste.

Designing the pad and stencil so that greater than 50% of both metallized leads of the chip component are making contact with the solder paste deposit is the best way to prevent tombstoning.

Use of non eutectic alloys is a widely recognized solution as well, but the placement / design is a much stronger influence over the defect.

Tombstoning can have a variety of causes but check for the following.

Are the PEC pads on both sides identical or of different sizes or, even worse, shared with another component? This will create an imbalance resulting in uneven forces on the chip during reflow.

Check that the solder paste deposits are identical in shape and volume on both sides of the chip position and the chips are being placed across the position accurately with the same amount overhanging the solder paste.

Also worth checking is the orientation of the chips through reflow. You will get best results if both solder joints reflow at the same time. This depends on the design, and whether this is feasible!

There are a few other factors but these areas are the favorites.

If you review the articles the two main, statistically significant factors are:

1. Pad Design and

2. Materials.

There are manufactures who have solder paste formulations specifically designed to minimize and/or eliminate the occurrence of tombstoning (see the APEX 2003 proceedings).

From a pad design standpoint, you want to make sure you are following well recognized design rules so that you have optimum coverage and placement of the component on the solder paste.

PCB Related:

Component Related:

• Varying degree of contamination on terminations
• Improper plating of terminations leading to varying degrees of oxidation

Paste Print Related:

• WS paste more prone than NC paste
• Tin-Lead more prone than Lead-free
• Varying amounts of paste on the pads
• Improper distribution of particles and flux between the prints on the pads

Placement Related:

• Placement offset or misalignment
• Insufficient pressure

Profile Related:

• Nitrogen reflow more prone than air reflow
• Convection flow rate
• Improper heat distribution
• Rapid ramp rate
• Short soak time
  

Tombstoning is often caused by one end of the part reaching liquidous temp just before the other end. The surface tension of the liquid solder at one end pulls the part upright off the paste of the other end.

It is important to confirm this by having a good thermal profile of the assembly with thermocouples attached at or nearthe component ends to see what the thermal nature of the assembly is in the region of the component affected.

There are several reasons why one end of a component may be reaching liquidous sooner in time then the other end:

1. The one end of the component enters the oven'szonebefore the other end due to the board geometry.
2. One end of the component is near a thermally massive component and the other not so close, making one end get hotter sooner.

Both problems can be reduced by slowing down the profile ramp rate as the parts approach liquidous and then holding the profile a little longer in the peak phase of the profile,at a slightly reduced peak value.

The goal is to allow the components to enter liquidous a little slower (reduced slope), giving more time for the entire component to heat and thus at a rate which gives both ends more time to heat the same.

With a good thermal profile of the assembly, you can make adjustments using the profiling software's prediction tools to see what oven settings can best alter the profile shape around the liquidous temp to reduce the temperature difference at each instant in time.

This may mean ramping a little faster out of the soak phase and giving the last two or three heated zones better control over the ramp rate through liquidous to peak. In other words, don't rush to peak, rather slow it down a bit and hold for a few more seconds at a little lower peak.