Skip to main content

3.5 Defect Mechanisms & Fixes

Voids, tombstones, head-in-pillow, graping—what causes them and the smallest fix that actually sticks.

Defects in reflow soldering rarely appear out of nowhere—they’re usually the visible symptom of an imbalance somewhere between printing and heating. Small process quirks, like uneven paste deposits, copper heat imbalances, or a profile that’s a bit too aggressive or too timid, can cascade into visible issues such as voids, bridges, tombstones, or the dreaded head-in-pillow. Each defect type tends to have a few usual suspects, and the fastest, most reliable fixes come from targeted tweaks rather than wholesale changes. Factors like stencil design, cleaning frequency, reflow atmosphere, and the ramp/soak/peak balance all play into how well solder flows, vents volatiles, and bonds to metal. By reading both the “print story” from SPI data and the “oven story” from profiles, it’s possible to zero in on the smallest effective adjustment. When dialed in, the process becomes stable, repeatable, and almost boring—which, in manufacturing, is exactly the point.

3.5.1 How to debug (the 10-minute rhythm)

  1. Confirm the print. Check SPI on the parts that fail: volume/height/area stable? If not, fix Chapter 7 items first.
  2. Sort by family. Are fails clustered on chips, QFN thermals, BGAs, or power pads? Each has a usual culprit.
  3. Read the oven story. Pull the last profile: ramp, soak, TAL, peak, ΔT. If it disobeys the paste or parts, correct that before anything else.
  4. Change one thing. Re-run a small lot and verify with AOI/AXI. Don’t shotgun.



3.5.2 Quick map (symptom → root → first fix)

Symptom (what you see)

Likely root cause(s)

First fix (smallest move that helps most)

Tombstones / skew (chips)

Pad & copper thermal imbalance; paste imbalance; aggressive ramp

Use home-plate/inverted shapes; bias paste toward cold pad; smooth ramp (≤~2 °C/s)

Bridging (fine pitch)

Over-print; clogged/dirty stencil; fast separation

Narrow aperture 5–10%; add/shorten cleaning; slow peel off the board

QFN/LFPAK voids

Solid slug of paste can’t vent; too hot/short reflow

Window the center pad to 50–65% with small tiles; use a gentle soak; try N₂ before adding peak

BGA head-in-pillow (HIP)

Package/board warpage at melt; oxide; too short/uneven TAL

Support the board; lengthen/smooth TAL; add N₂; verify VIPPOs are filled + cap-plated

Graping (grainy beads on pads)

Starved paste on tiny apertures; long soak; oxidation

Fresh paste; slightly higher peak or N₂; improve release (nano-foil, clean more, check area ratio)

Solder balls/splatter

Too much paste; long/hot soak; tired flux

Shorter soak; trim peak slightly; reduce aperture; keep bead fresh/clean

Dull joints / poor wetting

Low peak; oxidized finish; weak flux at end of life

+5 °C peak or a touch longer TAL; consider N₂; confirm paste age/finish choice

Opens on LGA/QFN edges

Paste starvation; poor release; fast belt

Enlarge or de-reduce edge apertures; improve release (nano-coat/cleaning); slow belt a notch



3.5.3 Tombstones & skew (chip passives)

Why it happens: one pad wets first (hotter copper or more paste), pulling the part upright or sideways.
Fix order:

  1. Stencil first: move to home-plate (trim toe) or inverted (trim heel) shapes; micro-window 01005–0402.
  2. Balance copper: if one pad ties into a plane, bias paste down on that side.
  3. Profile: smooth the ramp; avoid spiky preheat that “starts the race” early.
  4. Placement: confirm centered sit-down—nozzles and board support matter.



3.5.4 QFN/LFPAK voids & float/tilt

Why it happens: big center pads trap volatiles; too much paste floats the part.
Fix order:

  1. Window-pane the thermal pad to 50–65% coverage; keep tiles 1.0–1.5 mm with 0.3–0.5 mm webs. Add a chimney slot to an edge if the pad is huge.
  2. Profile: longer, gentler soak; consider slightly lower peak so volatiles leave before freeze.
  3. Atmosphere: try N₂—often worth 5–10% fewer voids.
  4. Paste: if still stubborn, evaluate a paste tuned for low voiding on slugs.



3.5.5 BGA/CSP head-in-pillow (HIP)

Why it happens: at liquidus the ball and paste separate (warpage or oxide) and never merge.
Fix order:

  1. Mechanics: solid board support in printer and PnP; keep panels flat through the oven (edge rails/center fingers).
  2. Profile: extend and steady TAL (don’t just spike peak); aim for even heat so both sides melt together.
  3. Atmosphere: add N₂ if wetting is marginal.
  4. Upstream: ensure VIPPOs are filled/cap-plated; respect MSL/bake for the package so balls aren’t oxidized or popcorn-prone.



3.5.6 Graping (grainy “bunches” of solder)

Why it happens: small paste deposits partially dry or oxidize; at peak they bead into “grapes” instead of flowing.
Fix order:

  1. Printer hygiene: shorten cleaning interval, keep a smaller bead, and refresh paste more often.
  2. Release help: nano-coated or electroformed foils; check area/aspect ratios on the worst pads.
  3. Reflow: a touch higher peak or N₂ to kick wetting over the edge; avoid long soaks that dry flux.
  4. Materials: verify paste age/storage; consider finer powder only if area ratios demand it (then tighten handling).



3.5.7 Bridges (fine-pitch gull-wing, QFN edges)

Why it happens: too much paste or paste smeared during peel; dirty stencil.
Fix order:

  1. Apertures: narrow 5–10% on the crowded side; add tiny relief nicks at inner corners; consider stencil step-down nearby.
  2. Printer: slow separation, ensure “wipe-clean” pressure, add wet+vac cycles when SPI shows area creep.
  3. Reflow: if you must, shave peak a touch; but most bridges die in printing.



3.5.8 Solder balls & splatter

Why it happens: volatile flux cooked in the soak, dirty stencil lips, or brute-force pressure that squeezes paste onto mask.
Fix order:

  1. Shorten soak or reduce mid-zone temps; keep ramp smooth.
  2. Printer: ensure “just-enough” blade pressure and regular understencil cleaning.
  3. Stencil: remove tiny paste traps (burrs, damaged apertures); nano-coat helps the edge stay clean.
  4. Paste: refresh; don’t exceed open time.



3.5.9 Dull joints / poor wetting

Why it happens: peaks too low; oxide (finish or paste); exhausted activators.
Fix order: +5 °C peak or +5–10 s TAL; try N₂; confirm finish compatibility (e.g., aged OSP needs honest profiles), and paste shelf life.



3.5.10 Opens on LGA / starved edges on QFN

Why it happens: release issues starve the edges; belt too fast to finish wetting.
Fix order: slightly larger or less-reduced edge apertures; improve release (foil/coating/cleaning); slow belt to add a few seconds of TAL.



3.5.11 Make fixes stick (how to prove you won)

  • Before/after plots: reflow profile screenshots with TAL/peak notes.
  • SPI heat-maps: the bad geometry should drop off the Pareto.
  • AXI/AOI snapshots: BGA collapse and QFN void distributions, pre- vs post-change.
  • Recipe & drawing updates: lock the new profile name and the revised aperture notes into the Golden Recipe and stencil spec, not just an email.



3.5.12 Pocket checklist (pin this near the oven & printer)

  • SPI is stable on the failing parts (else fix 6.x first)
  • One change at a time (profile or stencil or atmosphere)
  • Small DOE if torn between two fixes (3–6 boards is enough)
  • AOI/AXI prove improvement; limits unchanged (no goal-post moving)
  • Golden Recipe / stencil drawing updated; notes explain why



Bottom line: most reflow defects are printing + profile problems wearing different hats. Balance paste forces with smart apertures, keep the stencil clean, and shape soak/TAL to give solder time to meet metal—adding N₂ only when the data asks. Do that, and your Pareto gets pleasantly boring, fast.


Back to top