The bridged corners are likely caused by the CTE mismatch within the BGA body. Your rework profile must be too aggressive.

To solve the issue you should try to reduce your peak top heater temperature. This can be done by reducing your ramp rate and/or increasing the bottom heater temperature. If your system has the capability to inject cooling air through the pick up tube (Active Component Cooling, or Positive Flow) this may also help.

It is likely that the component will return to flat once it has been removed. If so, reballing with Winslow Automation or BEST reballing performs should do the trick. For rework, we recommend attaching the solder balls to the board rather than the bare BGA. This will reduce one thermal cycle for the BGA.

Once the array of balls is attached to the rework site (perform just like you would attach a BGA) you can place and reflow the bare BGA. Hopefully your less aggressive profile will eliminate the warp and corner bridging problem.

The component package should be considered when reballing/reworking BGA. Is the BGA a ceramic, FR4 laminate system or some other package type?

What was your root cause for corners to bridge? Sounds like "Oil Canning" of the PCB during spot reflow. Due to this, you may have scrapped assembly, not BGA.

If the preheater size and preheat the PCB prior to spot reflow, will improve your Z axis movement and probably reduce/remove the bridging result.

Moisture is often the culprit in cases like this.

Absorbed moisture and resultant hydrothermal stresses during reflow can cause excessive BGA warping; enough to induce solder ball shorting. This is why tracking a BGA’s cumulative exposure to ambient humidity is so critical.

Was this BGA stored in a sealed moisture barrier bag or 5% RH dry cabinet prior to use? Is it possible that the allowable exposure time (aka Floor Life) had been exceeded?

When in doubt the safest practice is to bake the BGA to dry it out before any reflow operation. This applies to initial assembly, as well as any subsequent rework operation.

If you have a batch of BGAs that are warping, try baking the remaining ones. If baking makes the problem disappear then you know the original culprit was moisture.

The industry standard for use and handling of moisture sensitive devices (IPC/JEDEC J-STD-033B.1) lists allowable floor lives (Table 7-1) and provides guidelines for baking components when necessary (Table 4-1).

The short answers is bga's can be reballed and replaced. We do it all the time for clients and we control the squat or fall of the bga device to insure they do not short at the corners.

The long answer requires more data because one must know if the pcb board is moving and what the degree the bga is moving --  i.e. warping or curling. The size of bga and the pcb board and the profile used for reflowing bga all play an important variable in the success of replacing bga's as well as the board prep surface wise as well as the type of flux and or paste being used.

Again there are techniques used to help compensate for any of these moving variables but we need to know more information about size of bga and board and reflow profile and if you are using localized heat or a reflow oven and wether the whole board and bga are being equalized during the reflow process

But again one can rework if one understands and controls the variables.

Bowing or warping is typically caused by thermal gradients across components, boards or both. Components are comprised mostly of fiberglass, ceramic, copper, silicon, and plastic, they expand and contract at different rates causing dissimilar forces to occur, also resulting in bowing.

Reducing heating and cooling rates while applying uniform heat across the entire board and component may relieve bowing, especially with components with thin interposers.

Once thermally optimized, persistent bowing may continue by the inherent design of the assembly. The use of temporary or permanent mechanical corner spacers has known to minimize pancaking of corner BGA spheres.

If your specific BGA is reusable, depends on its total thermal exposure and whether complies with its manufacturer’s specifications.

The warping you describe can be caused by moisture in the component or more likely too high a reflow temperature during assembly. You can remove the component, desolder the old solder balls, reball it and bake it. Then you can attempt to reflow the component.

Things you will want to do during your second attempt:

• Bake the component to remove entrapped moisture.
• Use a stencil and apply solder paste to the solder balls on the part, this will help with any coplanarity issues that you may have with this part.

When reflowing back onto the PCB make sure that you increase the subzone temperature on your rework machine and decrease the nozzle temperature. This will reduce the chances that the part will warp.

From your line of questioning, you are pointing toward the warp being purely in the BGA. The board can be contributing to this as well. This condition is usually seen around a fairly large BGA.

In our experience, this situation can be solved via the profile and not by a complicated manual use of spacer or shim to fight the effect. However, this will depend on the thermal management capability of your rework system.

Only in the case of large ceramic, or heat spreader versions, the solder may not be able to support the weight of the package when all balls are in a liquid state. In this scenario, high temperature solder balls are used for support.

Based on your description, I suspect that the profile is "aggressive" from the top side heating perspective. If this is the situation, you will likely get a non-uniform heating effect where warping of the BGA will happen and perimeter balls will reflow before the center balls, creating this bridge effect.

More bottom heat (but a very controlled one) could help to minimize the work needed from the nozzle side and create a more thermally stable situation.

Yes, the component can be saved. I would recommend to site dress this BGA and re-ball the BGA array.