Head-in-Pillow Defects: Causes, Detection, and Prevention

A head-in-pillow defect is a joint that looks complete on an X-ray image and reads as open on a continuity test. The solder ball touches the solder paste but the two never coalesced, leaving a boundary that conducts intermittently at first and fails completely later. Because the defect resists the usual inspection methods, it is one of the most expensive assembly problems in ball grid array work. This article covers the mechanism, the evidence, and the process controls that prevent it.

Anatomy of a Head-in-Pillow Joint

In a sound joint the ball collapses into the molten paste, the oxide layers break up, and the two volumes merge into a single grain structure. In a head-in-pillow joint the ball lands on paste that has already begun to cool, so the deposit forms a small dome and the ball rests on top of it. Metallurgically the interface remains two surfaces in contact rather than one continuous solid, and the contact pressure is enough to pass a bench test yet inadequate in a vibrating product.

Failure analysis usually begins with dye and pry, because the joint parts cleanly at the interface and the dye penetrates the unwetted boundary. Cross sections show a horizontal line of oxide or flux residue between the ball and the printed deposit, sometimes with a thin film of unreacted paste. The signature is distinctive: the ball is round rather than collapsed, so the standoff is greater than on the neighbouring joints.

Warpage as the Primary Cause

The dominant cause is dynamic warpage of the package during reflow. A plastic ball grid array can bow concave at the centre as it heats, lifting the ball array away from the paste at the moment the paste is molten. As the package cools it returns towards flat, so the ball descends onto a deposit that has already solidified. The cycle lasts only a few seconds, which is why the defect depends on the exact profile rather than on any visible feature of the board.

Package moisture content amplifies the effect. Absorbed moisture flashes to steam inside the mould compound above the glass transition temperature, adding internal pressure to the thermal gradient. Parts stored in a moisture barrier bag and then exposed to ambient air without a proper bake can warp tens of microns more than a dry part. The moisture sensitivity level printed on the package label states how long the part may be exposed before a bake is required.

Solder Paste Volume and Flux Depletion

Deposit volume sets the tolerance for warpage. A tall, well-formed column of solder paste gives the ball material to sink into, and its flux remains active long enough to reduce the oxide on both surfaces. A thin deposit is consumed before the ball arrives, which leaves a dry dome. Stencil aperture area ratios below 0.66, poor paste release, and long print-to-reflow intervals are common contributors when the package itself is sound.

Flux chemistry and paste freshness matter as much as volume. Paste that has stood open on the stencil for several hours loses activator through evaporation and its tack drops, so the ball cannot be held in contact with the deposit. Printing a fresh deposit every 30 to 60 minutes, controlling room humidity, and verifying deposit volume with a solder paste inspection system removes this variable from the investigation.

Package and Board Interaction

The interaction is best thought of as a gap that changes through the profile. The package bows as it heats, the board bows in its own direction, and the two movements either cancel or add together. A thick board with balanced copper resists the package lift better than a thin one, and a stiffener or a socket changes the picture entirely. Support pins placed under the package during the peak zone are a common practical fix.

Ball alloy and size also set how much collapse is available. A ball that is too small for the land pattern cannot bridge the gap, and a lead-free ball with a higher melting point collapses less than a eutectic one. Where a mixed alloy assembly is unavoidable, the profile must be adjusted so that the paste reaches full liquidus while the ball is still able to deform under the package load.

<img src="https://www.gopcba.com/wp-content/uploads/2026/06/Collaborative-Robot-Controller-PCBA.jpg" alt="BGA package warpage during reflow” />

Detection: X-ray, Dye-and-Pry, and Electrical Symptoms

Transmission X-ray reveals the gap between ball and paste only at oblique angles, so a vertical view often looks acceptable. Computed tomography or a tilted beam shows the characteristic separation. In-line automated X-ray with a dedicated head-in-pillow algorithm, which analyses the grey-level profile at the top of the ball, performs far better than an operator reading the same image, although false calls should still be calibrated against destructive analysis.

Electrical symptoms help to triage. Joints that fail only after thermal cycling, or whose resistance changes when the board is flexed, are candidates. Four-wire resistance measurement on a daisy-chained test vehicle is an efficient way to screen a profile change, because it detects small increases in joint resistance before a full open appears. Dye and pry remains the reference method for confirming what the instrument suspected.

Prevention Through Design and Profile

Prevention starts with the land pattern and the stencil. Keep the aperture area ratio above 0.66, match stencil thickness to the ball pitch, and never shrink openings to reduce bridging on a fine-pitch array. Where the package warpage specification allows a choice, select the variant with a thicker substrate, which moves less. On thin boards, add stiffening or support in the reflow fixture rather than relying on the paste alone.

Profile changes then close the remaining window. Extending the time above liquidus gives the ball more opportunity to sink into an active deposit, and reducing the ramp rate lowers the thermal gradient across the package body. A soak that brings the whole assembly to a uniform temperature before the peak is the single most effective adjustment, and each change should be validated by the same resistance or dye-and-pry method used to find the defect.

Process Monitoring in Production

Once a profile is qualified, it must be held. Log the oven recipe with every build, check that the peak temperature is measured on the package rather than on a nearby test coupon, and re-profile whenever a panel design or a paste lot changes. Incoming inspection should confirm the moisture sensitivity level and the bag seal, because an unplanned rebake can change warpage behaviour.

Deploying the same screening tool across a product family pays off. A single test vehicle with daisy-chained packages can be run at the start of every shift, and its resistance signature acts as an early warning. When the signature drifts, the cause is usually a paste lot, a stencil wipe interval, or an oven zone, all of which are cheaper to correct than a field return.

Related reading: our fabrication notes, board quality and design release notes cover the same ground.

FAQ

Can head-in-pillow be caught by normal optical inspection? No. The defect is hidden under the package. Automated X-ray with a dedicated algorithm, resistance measurement on a test vehicle, or dye and pry are the practical options.

Does baking the package always prevent it? A correct bake removes absorbed moisture and reduces warpage, which helps greatly, but a paste volume or profile problem can still create the defect on a perfectly dry part.

Why does the defect appear only after thermal cycling? The joint conducts while the two surfaces touch. Cycling oxidises and loosens that interface until resistance rises, which is why field failures appear months after a successful build.