Pad Cratering Under Area Array Packages
What Pad Cratering Actually Looks Like
Pad cratering is a fracture that begins inside the laminate rather than at the solder joint. Instead of the joint or the pad separating cleanly, a crack opens in the resin below the copper pad and spreads outward until the pad, the joint, and a shallow crater of dielectric material lift away from the board. From the top the failure can look like a normal open circuit. Under magnification, or after the package is removed, the crater is unmistakable: a rough dish-shaped void where the laminate used to be, with the copper pad still bonded to the solder.
The defect matters because it is a mechanical failure of the board, not of the process chemistry. Reflowing the joint again will not repair it, and rework often makes the crater larger. Because the crack path runs through the resin, the electrical open may not appear immediately. A crated pad can pass final test and still fail months later when thermal cycling or vibration completes the fracture.
Failures cluster around large area array packages where each joint carries a share of the assembly and handling load. Corner joints of a ball grid array, the outer rows of a chip scale package, and any joint near a board edge or a stiffener see the highest stress, which is why cratering is usually found on the periphery of a device rather than in the middle of the array.
Why the Failure Forms Below the Pad
The root cause is a mismatch between the stiffness of the joint and the strength of the laminate under it. The copper pad and the solder ball form a stiff column, while the dielectric beneath the pad is comparatively soft. When the assembly bends, the load concentrates at the edge of the pad where the copper ends and the resin begins, and the resin fails first.
Three conditions drive the crack. The first is mechanical loading during or after assembly, including board flexing on the conveyor, panel handling, depanel stress, and connector insertion. The second is thermal expansion mismatch between the package, the solder, and the laminate, which cycles the pad edge every time the product powers up. The third is process history, especially rework, which raises the local temperature above what the resin was qualified for and embrittles the interface.
Material choice sets the baseline. Standard FR-4 with a high filler content and a high glass transition temperature resists cratering far better than a low-Tg laminate, but the same filler that stiffens the resin also makes the board more brittle in other directions. The practical objective is not a single best material but a stackup whose resin can absorb the pad-edge stress without cracking, matched to how much bending the product will actually see.
Design Choices That Lower the Risk
Design controls the stress before any material is selected. Non-functional pads under and beside the array add stiffness and give the crack a place to stop, so keeping them, or adding them deliberately, is one of the cheapest countermeasures available. Reducing the pad diameter so that more resin remains between adjacent pads also helps, as long as the reduction does not compromise the solder joint.
Via placement matters more than most layout teams expect. A via directly in the pad creates a hole in the copper where the joint can push into the resin, so via-in-pad should be filled, capped, and planarized when it is used at all. Where possible, vias belong in a fanout pattern beside the pad rather than under it.
Keeping the package away from board edges, breakaway tabs, and stiffener transitions reduces the bending stress the joint has to survive. Where the mechanical envelope forces a package close to a high-stress feature, a local stiffener or a thicker core can shift the failure mode away from the laminate and back into the joint, where it is at least repairable.
Picking Materials That Resist Cracking
Laminate selection is usually a choice between crack resistance and everything else. High-Tg, high-filler resins with a well-controlled resin content resist pad cratering best, and many suppliers publish a cratering or pad pull resistance figure that can be compared directly. Halogen-free and low-loss materials vary widely in this respect, so a material that solves a signal integrity problem can quietly create a mechanical one.
Solder mask and surface finish contribute as well. A mask that adheres well to the laminate supports the pad edge, while a poorly adhered mask lifts and takes the pad edge with it. Surface finishes that consume copper during deposition thin the pad and reduce its stiffness, which is one reason the finish and the copper thickness should be evaluated together rather than in isolation.
Copper thickness is a double-edged parameter. Thicker copper stiffens the pad and spreads the load, but it also increases the thermal mass and makes the pad edge a sharper mechanical transition. Most suppliers settle on a moderate foil weight plus an adequate resin system rather than either extreme.
Assembly and Rework Practices
Most cratering that appears in production is introduced by handling rather than by the reflow profile. Boards should be supported during depaneling, and the support should be close to the cut so the panel cannot flex across the array. Router bits and punches need to be sharp and correctly set, because a dull tool loads the whole panel before it cuts. Conveyors, magazines, and fixtures should contact the board away from the packages, and any manual handling should follow an edge rule that keeps fingers off the array.
Rework is the single largest contributor to pad cratering in mature builds. Removing a ball grid array means heating the whole package above the reflow temperature and pulling or shearing it away, which loads the pad edge while the laminate is at its softest. A proper rework profile holds the board flat, uses bottom-side preheat, applies the correct amount of shear, and pre-bakes moisture out of the laminate. Doing it twice on the same site multiplies the damage, so rework should be a single deliberate operation rather than an iterative one.
Where rework is predicted to be frequent, the design should provide for it. Extra non-functional pads, a slightly thicker core, and a lower-Tg profile all cost something, but they cost less than a rejected assembly with a cratered array at the center.
Detecting and Analyzing Cratering
Craters hide under solder joints, so ordinary inspection cannot see them. Electrical test finds opens, but only after the crack has severed the connection, which is late. X-ray can show a crack in the pad or a lifted joint if the view is clear and the resolution is adequate, but it is far more reliable when the failure is already suspected.
The definitive method is destructive. Cross-sectioning through the row of interest reveals the crack path and shows how deep it runs into the resin. Dye-and-pry is faster and works well for screening a suspect package: the assembly is soaked in a penetrating dye, the package is pried off, and the crack surfaces are examined for dye staining. Both methods should be paired with a load-to-failure or bend test on a sample of boards, because a crater that cannot be made to fail under the specified load is a cosmetic finding rather than a reliability risk.
Root cause investigation should separate mechanical from thermal drivers. If craters appear on assemblies that were never reworked and never bent, the cause is likely thermal cycling or a material mismatch. If they appear only after rework, the cause is the rework profile. If they appear on one panel position but not others, the cause is fixture or depaneling stress.
Screening and Qualification
Because the failure is design, material, and process dependent at the same time, qualification should test the assembled product rather than the laminate alone. A standardized bend or drop test on a representative panel exercises the pad edge exactly as the field will, and a thermal cycling run to the product profile adds the expansion component. Together they give a pass or fail signal that a material datasheet cannot.
Once the design is qualified, process control keeps it qualified. Track depaneling tool wear, rework frequency, and conveyor support configuration as controlled variables, and review any crater-related field return against them. A cratering problem that appears on a mature product almost always traces back to a change in one of those three.
FAQ
Is pad cratering the same as a lifted pad? A lifted pad is a bond failure between the copper and the laminate. Pad cratering is a fracture inside the laminate below the pad. The two look similar from the top but the repair and prevention strategies are different.
Can a cratered pad be reworked? Usually not reliably. Reheating softens the resin further and often enlarges the crater, so the site may work at first and fail later. In most cases the assembly should be scrapped or downgraded.
Which packages are most at risk? Large ball grid arrays, chip scale packages, and anything with a stiff body and a high standoff, especially at corner joints and near board edges or stiffeners.
Does lead-free assembly make cratering worse? It can. Higher process temperatures soften the laminate more and increase the expansion mismatch, so the same design that was marginal with tin-lead may crater with a lead-free profile.
How do I test for it without destroying boards? There is no fully non-destructive method. The practical approach is electrical test plus a documented destructive sample per build, using cross-section or dye-and-pry on a small number of units.
Conclusion
Pad cratering is a laminate failure that lets a mechanically sound product pass test and fail later, which is exactly why it deserves attention at design review rather than at field return. Keeping non-functional pads, keeping vias out of pads, controlling depaneling and rework stress, and qualifying the assembled panel with bend and thermal cycling tests together reduce the risk to a manageable level. Teams that want a deeper look at the surrounding decisions can review our work on PCB design and layout, PCB manufacturing, quality management, and PCBA testing to see how these controls fit together in 2026.
