That sounds more like a component manufacturing issue rather than the reflow process. If your reflow is heating the component to the point of lead shoulder failure, the rest of the PCB would probably show signs of overheating as well.

This sounds like a combination of a oxidized lead surface and a low thermal profile. Check the solderability of the parts through a lab that does wetting balance testing and then have someone assess the reflow profile for the mass of the board and parts. If these areas show okay then check for a barrier film on the J-leaded QFP leads.

The black body of the part is a very good absorber of IR radiation. If you put a thermocouple on top of it, you should see how much hotter than the rest of your product it gets.

It is likely your part is overheating and the lead frame to overmold junction is being damaged due to melting and/or expansion related stresses.

Either that or you have some poorly made parts. If you get a different part with the same package type and it does not have the same problem, go back to your supplier with the problem.

Without having more detail it is difficult to diagnose the actual issue. With cracked or failed solder joints you would want to do a few things to determine the actual root cause. Is this a sudden issue?

In other words can it be related to a specific batch of components and/or PCBs? Take a look at the fracture surfaces, what do you see? Does it appear that the solder joint is just "pulling out" of the solder fillet (sometimes called "foot-in-mud"). If so it could be a plating issue on the components.

Is it fracturing at the pad surface? Could it be related to the metalization on the PCB? Those are some concerns to look into.

Since you are suspecting reflow issues, perform a profile, being sure to place a thermocouple adjacent to the solder joint in question. If this is a densely populated PCB, with components of different sizes you want to make sure you profile in multiple locations to make sure you have the proper reflow time and temperatures at each location.

This is suspicious, and a failure analysis should be undertaken. I would be guessing without examining the parts.

If you do not reflow the part to a pad but subject the part the same heating cycle, does this lead still break?

How flat is the board after reflow? The lead has to fail with stress so you need to answer where the stress is coming from.

1. Board support in placement and the placement pressure. Too much pressure with direct support under the component could stress the package body.
2. Since it is IR heating check to see if the package body is able to withstand the temperature setting for the IR panel. Possible damage occurring due to high temperature.
3. Check if the component is lead free compatible if the process is lead free.
4. Component age, handling and storage. Old components that have been in storage could have absorbed moisture over a period of time resulting in popcorn delamination. If this is the case try baking the components prior to assembly (125 deg. C for 48 hours would be good)
5. Check the reflow profile to make sure there is no thermal shocking happening to the components.
6. Also check to make sure that the cooling is not too rapid.
   

There are several contributing factors here.

A simple solution would be an evaluation of the volume of paste that is being deposited on the board.

It sounds like there could be a large deposit of solder that is working against your oven profile. If the solder deposit is extensive, and the joint cools too quickly, that could result in the crack that you are describing.

You should be able to contact your stencil vendor and work with them to determine how much paste would be required to properly solder the lead. With the proper volume and location of paste deposited, you may be able to avoid having to change the profile of the oven.