Head In Pillow: Design Rules and Process Limits

A head in pillow defect is a ball that has never fused with the solder on the pad. The ball sits on top of the paste, the paste melts and reflows into a fillet, and the ball remains solid on top of it, touching but not bonded. The joint can measure electrically continuous when it is new and open later, which makes it one of the hardest defects to detect and the most expensive to miss. This article explains how it forms and what has to be controlled.

How the Defect Forms

For the ball to fuse, both the ball and the paste beneath it must be molten at the same time, so that the two liquid volumes merge. Head in pillow occurs when the paste has already solidified or has not yet melted when the ball reaches its own melting point, so the two never coalesce. The ball then sits on the residue, and the geometry that results is a straight separation rather than a fillet.

The separation is often visible in a cross section as a horizontal line above the pad, with a reflowed fillet below and an unitered ball above. The oxides that form on the two surfaces during the moment of contact are part of why the bond does not form even when the temperatures later match.

Package Warpage and Its Timing

The most common root cause is package warpage. A ball grid array package changes shape as it heats, becoming convex or concave depending on the construction. If the package is convex at the moment the paste melts, the centre balls are lifted away from their pads and are the last to make contact.

The warpage is temperature dependent and the shape can reverse during the profile. The critical moment is the melting point, because that is when the paste must be in contact with the ball. Controlling the profile so that the package and the board are at similar temperatures when the paste melts reduces the gap and gives the two surfaces a chance to merge. The thermal behaviour of the assembly is described in the context of reflow oven profile verification.

Paste Volume and Collapse

The paste volume under each ball determines whether the ball can reach the paste when the package warps. A larger deposit gives more material to bridge the gap and more flux to clean the surfaces, while a smaller one leaves the ball isolated. This is one of the few defects where increasing the paste volume is a legitimate corrective action.

The volume also affects the collapse. The ball partially collapses into the molten paste, and the amount of collapse depends on the volume and on the forces. Where the volume is short, the ball may not reach the pad at all, which produces an open joint rather than a head in pillow. The volume relationships are the same ones covered in solder paste volume and stencil design.

Reflow Profile and Soak

A longer soak before the peak allows the package and the board to reach a similar temperature before the alloy melts, which reduces the differential warpage at the critical moment. A short, hot profile melts the paste before the package has heated through, which increases the gap and the risk.

The peak and the time above liquidus also matter. The joint needs enough liquid time for the ball to collapse and to merge, so a profile that spends very little time above liquidus is more likely to produce the defect on a marginal design. The trade is that more time at temperature increases intermetallic growth, so the profile has to be balanced rather than extended without limit.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/1768208893-quick-turn-pcb-assembly.webp" alt="X-ray image of a ball grid array joint with an open connection” />

Placement Force and Support

Placement force seats the ball array into the paste, and a force that is too low leaves the package resting on the highest balls. The force is limited by paste displacement and by the package itself, and it should be set per package type. A package placed with a force that suits a larger component may not seat properly.

Support beneath the board is part of the same problem. A board that sags during reflow changes the relative height of the package and the pads, which is equivalent to additional warpage. Support tooling that holds the board flat through the oven reduces the effect, and the considerations are the same as those described for PCB warpage control.

Detection Limits of X-Ray

X-ray is the standard inspection for ball grid array joints, and it is poor at detecting head in pillow. A two dimensional image shows the ball and the pad as overlapping shapes, and a joint that is touching appears connected. Only a cross section or a three dimensional scan intended for the purpose gives a clear indication.

Because detection is limited, prevention is the practical control. The process should be designed with enough paste volume, enough soak and enough support that the defect does not form, and the first article inspection should include a cross section of the array rather than only an X-ray image. The X-ray capabilities and their limits are described in BGA inspection methods.

Process Controls and Qualification

The controls are cumulative. Increase the paste volume to the maximum the stencil can support without bridging, extend the soak so that the package and board reach a similar temperature, verify the placement force per package, support the board through the oven, and confirm the profile with thermocouples on the package and on the board.

The gopcb assembly group requires a cross section of at least one area array package in the first article inspection for any product that uses them, and repeats it whenever the package, the paste or the profile changes. That single check catches the defect that X-ray misses, and it is the evidence that the process is capable before a production batch is released.

Checks Before Release

On a design of this kind, head in pillow is the item that decides how the rest of the board is arranged. Sampling is a compromise between cost and confidence, and the sample size should follow from the failure rate that has to be detected. Where a decision is made by judgement, a boundary sample makes the judgement repeatable between operators and between shifts.

Where a process is at the edge of its capability, the margin should be bought deliberately rather than discovered during production. A record that identifies the operator, the date and the settings is worth more than a record that identifies only the result.

Verification and Records

On a design of this kind, head in pillow is the item that decides how the rest of the board is arranged. The checks that matter are the ones performed on the product rather than on a sample kept for the purpose, because a coupon that travels with the panel is the only evidence about that panel. The first article confirms that the setup matches the intent, and it is the cheapest point at which a wrong setup can still be corrected.

FAQ

Can a head in pillow joint pass electrical test? Yes, and it often does when it is new. The connection is mechanical contact rather than a metallurgical bond, and it opens as the surfaces oxidise or as the assembly cycles.

Does more paste always help? Up to the point where the deposit bridges to the neighbouring ball. The volume should be set at the highest value that keeps the joints separate.

Is a longer soak always better? It reduces the warpage differential but adds intermetallic growth. The profile should be balanced against both.