The Repair LibraryRead · Learn · Master

BGA Failure Modes

The geometry and the metallurgy are built; this section catalogs how the package actually dies. The four classes the chapter has been promising arrive in order. The mechanical class first: thermal-cycling fatigue — the standoff's flexing joints slowly cracking, worst where the stress map says so — and drop fracture, the instant brittle break that runs through the intermetallic plane rather than the solder's body, read straight through the silver trade the last section taught. The stress map itself gets redrawn here: the corner-ball story the chapter opened with is the first approximation, and the die shadow refines it — the die is the package's stiffest tenant, and on the flip-chip packages this volume meets, failure maps cluster under the die's edge rather than at the array's geometric corners. The process class second: cold joints that never saw full heat and disturbed joints written by pasty-range motion — the assembly-time failures that arrive dead or die young. The wetting class third: head-in-pillow, where an oxidized or warped-away ball rests on its paste like a head on a pillow — two bodies, never one joint — and the plain unwetted open, both creatures of oxide and reflow-time warpage. And the short class last: bridges, the process-time excess the arithmetic of tight pitches invites, and electrochemical migration — dendrites growing ball-to-ball in service wherever flux residue, moisture, and bias conspire. Each class carries a timing signature — when in the package's life it strikes — and a symptom signature the bench can read: the warm-up intermittent, the drop-dead brick, the first-boot corpse, the leakage that grows by the week. The section closes by pointing every class at the chapters that answer it — and at the repair-or-not arithmetic next door.

AdvancedLow Risk23 min read

What You Will Learn

  • You will learn the mechanical class — cycling fatigue in the flexing standoff and the brittle drop fracture through the intermetallic plane — read through the alloys.
  • You will learn the redrawn stress map — why the die shadow, not the geometric corner, hosts the failures on die-dominated packages.
  • You will learn the process class — cold joints from starved heat and disturbed joints from pasty-range motion.
  • You will learn the wetting class — head-in-pillow and the unwetted open, the children of oxide and reflow warpage.
  • You will learn the short class and the signatures — bridges at process time, migration in service, and the timing and symptom reads that sort every class at the bench.

What You Will Be Able To Do

  • You will be able to name the four failure classes with their mechanisms and their alloy dependence.
  • You will be able to draw a package's stress map — array corners and die shadow — and mark its highest-risk joints.
  • You will be able to explain why drop kills through the intermetallic plane while cycling kills through the solder's body.
  • You will be able to read symptom and timing signatures into class hypotheses at the bench.
  • You will be able to state the shorts' conditions — process bridges versus in-service migration — and what prevention and inspection each demands.

Required Tools

  • The donor bin and the harvested BGA package — the failure geography is drawn on real hardware
  • The microscope and side light — edge peeks and package outlines feed the stress map
  • Package datasheets or board diagrams where they exist — die position is the map's key fact
  • A notebook — this section builds the four-class table and the symptom cards the diagnosis chapters will use
  • No hot tools — failures are cataloged and read here, never induced

When NOT to Attempt This

Do not attempt this section if any of the following apply to you:

  • You are not comfortable working with small surface-mount components.
  • You have not completed the prerequisite sections for this skill.
  • You do not have the required tools in working condition.

Section Overview

The chapter's promised catalog arrives: four classes, each with a mechanism, a geography, and a signature (what-is-bga). The mechanical class leads. Cycling fatigue cracks the flexing standoff slowly; drop kills instantly through the brittle intermetallic plane — intermetallic fracture — and the silver trade decides which death a device is prone to (bga-ball-composition-and-alloys). The stress map gets redrawn. The corner-ball story was the first approximation; the die shadow refines it — the die's stiff edge projects the true perimeter of peak strain (intermetallic-compounds-and-joint-strength). The process class arrives dead or dies young. Cold joints from starved heat, disturbed joints from pasty-range motion — assembly-time failures with first-boot signatures (why-joints-fail-metallurgical-causes). The wetting class never joined. Head-in-pillow and the unwetted open — oxide and reflow warpage leaving two bodies where one joint should be. And the short class closes. Bridges at process time; electrochemical migration growing dendrites ball-to-ball in service wherever residue, moisture, and bias conspire. Four classes, four geographies, four signatures — the catalog entire.

Why This Matters

Every dead BGA board that crosses the bench belongs to one of these classes, and the technician who can name the class has already halved the diagnosis (what-is-bga). This matters because the signatures sort the queue: warm-up intermittents, drop-dead bricks, first-boot corpses, and growing leakage each point at a different class before any instrument comes out — the calendar and the symptom story are free evidence, and this section teaches their reading (why-joints-fail-metallurgical-causes). This matters because the stress map aims the instruments: X-ray time and reball decisions are expensive, and knowing that die-dominated packages fail under the die shadow rather than at the geometric corners points every later inspection at the right joints first (intermetallic-compounds-and-joint-strength). It matters because the alloys predict the deaths: the stiff high-silver field that endures cycling but dies on the pavement, the soft leaded relic that shrugs off drops but fatigues in the heat — the last section's metallurgy becomes this section's actuarial table (bga-ball-composition-and-alloys). It matters because the repair-or-not question starts here: cracked and unwetted joints are exactly what reflow and reballing address, while a cratered pad or a cracked die is a different conversation — and the next section's arithmetic needs this section's classes as its input. And it matters because prevention is class-shaped: cleaning discipline starves migration, profile discipline prevents cold and disturbed joints, and the reflow chapters' warpage management exists because of head-in-pillow. Name the class first — everything after gets cheaper.

Required Prerequisites

  • What Is BGA? — the package anatomy, the ball's life, and the corner-ball stress story this section refines.
  • BGA Ball Composition and Alloys — the metallurgy the failures are read through: the silver trade, the pasty range, and the mixing rule.
  • Nothing consumed — the catalog is built by reading, drawing, and sorting (what-is-bga)
  • A notebook page for the four-class table — to build the sorting card every BGA diagnosis will open with (why-joints-fail-metallurgical-causes)
  • A second page for the symptom cards — to file the timing signatures beside the classes they name (bga-ball-composition-and-alloys)
  • Printouts or sketches of the donors' package outlines — to draw die shadows onto real ball fields (intermetallic-compounds-and-joint-strength)
  • The donor bin and the harvested package — to draw the stress geography on hardware instead of in the abstract (what-is-bga)
  • The microscope and movable side light — to take the edge peeks that feed the outlines (intermetallic-compounds-and-joint-strength)
  • Datasheets or board diagrams for the donors where they exist — to place each package's die honestly rather than by guess (bga-ball-composition-and-alloys)
  • A board with visible corrosion or residue history, if the bin offers one — to see migration's preconditions in the wild (why-joints-fail-metallurgical-causes)
  • The bench camera — to file the stress maps and symptom cards in the record

Real-World Applications

The catalog pays off in the sorting, and the sorting starts with symptoms. A bench handed a laptop whose screen lights only after ten warm minutes reads the signature straight into the mechanical class: a cycling crack under the GPU breathing open when cold and closing as the board expands — a warm-up intermittent pointing at the die shadow before any instrument is unpacked (why-joints-fail-metallurgical-causes). A shop pricing a phone that died the instant it hit a tile floor reads the drop class through the alloys: a stiff lead-free field passes the shock into the brittle intermetallic plane and the pads — instant, mechanical, and priced honestly against the repair arithmetic next section (bga-ball-composition-and-alloys). A refurbisher receiving a batch of boards that came back dead from a vendor's reflow recognizes the first-boot signature of the process and wetting classes — cold joints, disturbed joints, head-in-pillow from warpage — and sends questions about profile and paste before shipping anything else (what-is-bga). And an industrial repair bench facing a controller from a humid cabinet with leakage that grew for weeks names migration by its conditions — residue, moisture, bias — and knows the fix must include the cleaning, not just the short (intermetallic-compounds-and-joint-strength). The confusions this prevents: a warm-up intermittent chased as a software fault, a drop death blamed on a battery, a process corpse reflowed blindly instead of questioned, and a migration short cleared without removing the conditions that grew it.

Common Challenges

  • The classes share a symptom: dead. Every failure eventually presents as a board that does not work — the difficulty is learning that the road to dead differs: timing, history, and behavior under warmth and flex are the separators, and they cost nothing to collect (why-joints-fail-metallurgical-causes).
  • The stress map is counterintuitive. The foundation taught corner balls fail first, and now the die shadow complicates it — the difficulty is holding both: corners for package-level flex, die edge for die-dominated packages — the refinement does not repeal the rule, it redraws its map (intermetallic-compounds-and-joint-strength).
  • The invisible failures must be reasoned, not seen. Every one of these classes hides under the package — the difficulty is trusting the class logic and the signatures now, and the instruments later, instead of demanding a look the geometry will never grant (what-is-bga).

Safety Notes

Risk Level: Low. Nothing is heated and nothing is induced to fail — the section reads failures that already happened, and the standing law holds.

Professional Tips Before Starting

  • Collect the symptom story before touching any board. When did it die, what happened that day, does warmth or flex change it — the free evidence this section teaches costs only the asking, and it decays if the asking waits (why-joints-fail-metallurgical-causes).
  • Find the die before drawing conclusions. Datasheet, board diagram, or the package's own thermal shadow — the stress map is only as good as the die outline it is drawn around (intermetallic-compounds-and-joint-strength).
  • Keep the four-class card at the bench. Mechanical, process, wetting, short — the sorting habit built now is the reflex the diagnosis chapters will assume (what-is-bga).

The Catalog — Four Classes, Four Geographies, Four Signatures

Recap and Frame

The chapter built the package and its metallurgy; this section spends both on the question the bench actually asks: what killed it (what-is-bga). The foundation predicted the geography. The standoff flexes with every thermal cycle, corner balls flex hardest, and the fault classes were previewed in the geometry — this section names them properly (bga-ball-composition-and-alloys). Volume 3 taught the metallurgy of dying joints. Cold joints, voids, intermetallic growth, fatigue — the science-of-soldering chapter carried the mechanisms at single-joint scale, and the package multiplies them by hundreds and hides them (why-joints-fail-metallurgical-causes). The intermetallic layer returns as a load-bearing character. Joint strength lives and dies at the intermetallic — thin enough to bond, thick or brittle enough to break — and the drop story runs straight through it (intermetallic-compounds-and-joint-strength). What is new is the sorting. Four classes with distinct mechanisms, geographies on the ball field, and signatures in time and symptom — a catalog built for diagnosis, not just description. Hold the frame — old mechanisms, package scale, sorted for the bench — and the catalog reads fast.

The Mechanical Class — Cycling Fatigue and Drop Fracture

The mechanical class kills through stress, and it kills two ways with two different faces (bga-ball-composition-and-alloys). Cycling fatigue is the slow death. Every power cycle expands the board more than the package, the standoff's joints flex to absorb the disagreement, and solder — cycled thousands of times — cracks the way any metal does: slowly, through its body, along grain boundaries, until a working joint becomes a crack that touches closed when warm and falls open when cold. The warm-up intermittent is its voice. A device that works only after minutes of heat is describing a crack breathing with the board's expansion — the classic signature, and the reason the symptom interview asks about warmth before anything else (why-joints-fail-metallurgical-causes). The direction can invert. Some cracks pull open rather than closed as the board expands — a device that dies warm and works cold is still a temperature-dependent intermittent and still indicts the mechanical class: it is temperature dependence, not its direction, that points at the crack. Drop fracture is the fast death. Shock loads the joints faster than solder can yield, and the break runs not through the solder's ductile body but through the brittle intermetallic plane at the pad interface — the intermetallic fracture — instant, clean, and deaf to warm-up (intermetallic-compounds-and-joint-strength). The alloys pick the vulnerability. Stiff high-silver fields transmit shock into the intermetallic plane and the board's pads — with pad cratering, the board-side laminate tearing out under a pad, as the drop's board-side cousin — while soft leaded fields absorb drops and fatigue faster in cycling: the silver trade, become an actuarial table. And the stress map says where. The foundation's corner-ball rule carries the package-level story, and the die shadow — next — redraws it for the packages this volume actually meets (what-is-bga). Slow through the solder, fast through the intermetallic, sorted by alloy, mapped by stiffness — the mechanical class entire. Ask when it died and whether warmth changes it — the mechanical class answers the interview before any instrument opens.

The Stress Map Redrawn — The Die Shadow

The corner-ball story is true and incomplete, and the refinement matters because it aims every later inspection (what-is-bga). The corner rule is package-level. Treat the package as one stiff plate on a compliant board and the joints farthest from the center flex hardest — the geometric corners, exactly as the foundation taught. But the package is not uniform. The die is the package's stiffest tenant — silicon, far stiffer than substrate or overmold — and the region beneath it barely stretches while the substrate beyond its edge complies: the strain concentrates in the ring of joints under the die's edge — worst of all under the die's projected corners — the die shadow's perimeter (intermetallic-compounds-and-joint-strength). Flip-chip packages sharpen the effect. The die couples to the substrate through its own rigid bump field, and on the die-dominated consumer packages this volume meets — SoCs and GPUs with dies spanning much of the package — real failure maps cluster under the die edge rather than at the array's geometric corners (bga-ball-composition-and-alloys). The bench consequence is aim. X-ray fields of view, reball inspections, and crack hunts start at the die-shadow ring on die-dominated packages and at the corners on small-die or memory packages — the map is drawn per package, from the die outline, not from habit (why-joints-fail-metallurgical-causes). The die outline is findable. Datasheets draw it, board diagrams imply it, and the package's own thermal behavior sketches it — the map costs minutes and aims hours. Corners for the plate, die edge for the tenant — the stress map entire. The refinement does not repeal the corner rule — it tells you which rule this package obeys.

The Process and Wetting Classes — Born Dead

Two classes are written at assembly or rework time and carry first-boot signatures, and they differ in whether a joint ever formed at all (why-joints-fail-metallurgical-causes). The cold joint starved. Heat that never fully liquefied the ball leaves a dull, grainy, weak connection — formed enough to pass a first test sometimes, weak enough to die young always — the classic process failure the profile chapters exist to prevent. The disturbed joint moved. Motion during the pasty range tears the half-frozen structure — the metallurgy section's no-movement discipline, violated — and the result reads like a cold joint's twin: intermittent from birth, invisible from outside (bga-ball-composition-and-alloys). Head-in-pillow never joined. The ball rests on its melted paste like a head on a pillow — two bodies, an oxide skin or a warp-opened gap between them, never one joint — born when an oxidized ball refuses to merge or when the package warps during reflow, lifting balls off their paste at the worst moment and setting them back after the paste froze. The unwetted open is its blunter sibling. A ball or land that never wetted at all — oxide, contamination, or a starved corner of the heat — leaving an open the meters find on day one (what-is-bga). Warpage is the hidden actor. Packages bow with heat and recover with cooling; the reflow chapters manage it with profiles and support because head-in-pillow is what happens when it goes unmanaged — and the corners and edges that warp farthest are where these failures cluster (intermetallic-compounds-and-joint-strength). Starved, disturbed, pillowed, unwetted — the born-dead classes entire. A board dead or flaky from its first boot indicts the process, not the years — question the reflow before blaming the device's life.

The Short Class — Bridges at Birth, Migration in Service

The last class connects what should be separate, and its two members live at opposite ends of the package's life (what-is-bga). The bridge is a process failure. Excess paste, a smeared print, over-collapse, or a misaligned settle merges neighboring balls at reflow — the tight-pitch arithmetic's standing invitation — and the short is present from the first power-on, findable by the meters and imaged by the X-ray chapter ahead (bga-ball-composition-and-alloys). Migration grows its short in service. The electrochemical migration mechanism needs three conditions at once — ionic residue such as unclean flux, moisture, and bias between neighbors — and under them, metal dissolves at one conductor and redeposits toward the other, growing dendrites — the branching metal fingers Volume 5's corrosion work met — ball-to-ball until leakage becomes a short (why-joints-fail-metallurgical-causes). The timing signature separates the two. A short from first boot is a bridge and indicts the process; leakage that grows over weeks under power is migration and indicts the conditions — and the distinction decides the fix (intermetallic-compounds-and-joint-strength). The fix must remove the legs. Clearing a migration short without cleaning the residue and drying the environment schedules the next one — remove any leg of the triad and the growth stops, which is why the cleaning discipline the flux chapters taught is short-prevention under another name. And the class closes the catalog's map. Mechanical, process, wetting, short — each class now points at its chapters: inspection and X-ray next chapter, reballing and replacement beyond, and the repair-or-not arithmetic next section, which needs exactly this sorting as its input. Born at reflow or grown under bias — the short class entire. Ask when the short arrived — the calendar convicts either the process or the conditions, never both.

Common Mistakes

  • Chasing every dead BGA as a software or power fault first. The board reflashed and the rails probed while the symptom story goes uncollected — the free interview — when, what happened, does warmth or flex change it — sorts the class before any tool opens (why-joints-fail-metallurgical-causes).
  • Aiming inspection by habit at the geometric corners. Every package searched corner-first — die-dominated packages fail under the die shadow; the map is drawn per package from the die outline (intermetallic-compounds-and-joint-strength).
  • Reading a drop death as a fatigue story. Months of cycling blamed for a break that happened in one afternoon — drop runs through the brittle intermetallic plane and is deaf to warm-up; the histories are different and so are the repairs (bga-ball-composition-and-alloys).
  • Reflowing a first-boot corpse blindly. Heat applied to a board that arrived dead from a vendor's oven — process and wetting failures indict the profile and the paste; repeating the reflow repeats the question (what-is-bga).
  • Clearing a migration short without removing its legs. The dendrite burned away, the residue and humidity left in place — the triad regrows what was cleared; the cleaning is the repair.

Troubleshooting Guidance

The catalog troubleshoots by sorting, and the sorting has an order: interview, class, geography. If the symptom story is warm-up-dependent: mechanical class, cycling side — ask how old the device is, what it runs hot, and let the die outline aim the suspicion at the shadow ring (why-joints-fail-metallurgical-causes). If death followed an impact: mechanical class, drop side — expect an intermetallic-plane break or a cratered pad, expect no warm-up sensitivity, and expect the repair conversation to include the board, not just the package (intermetallic-compounds-and-joint-strength). If the board was born dead or flaky: process or wetting class — ask what reflow it last saw and who ran it; cold, disturbed, pillowed, and unwetted joints all indict the oven before the years (bga-ball-composition-and-alloys). If a short appeared and grew in service: migration — hunt the triad: residue from the last rework, the cabinet's humidity, the bias across the neighbors; the fix is the cleaning as much as the clearing (what-is-bga). If nothing sorts cleanly: hold multiple hypotheses honestly — the classes are distinct but boards are not obligated to fail one way at a time, and the inspection chapter ahead exists precisely to adjudicate what the interview cannot. The throughline: the class hypothesis is free and the instruments are expensive — sort first, spend second.

Verification & Testing Methods

Confirm the catalog before the repair arithmetic builds on it:

  • [ ] I can name the four failure classes — mechanical, process, wetting, short — with their mechanisms and alloy dependence.
  • [ ] I can draw a package's stress map — array corners and die shadow — and mark the highest-risk joints from the die outline.
  • [ ] I can explain why drop kills through the intermetallic fracture plane while cycling cracks the solder's body.
  • [ ] I can read timing and symptom signatures — warm-up intermittent, drop-dead, first-boot corpse, growing leakage — into class hypotheses.
  • [ ] I can state the shorts' split — process bridges versus electrochemical migration — and the triad of conditions whose removal is the migration repair.

Then try the practice exercises below — desk and observation work only; scenarios differ from the quiz.

Practice Exercises

  1. Draw two stress maps (5 minutes, donor boards, datasheets or diagrams, microscope). Pick one die-dominated package and one small-die or memory package from the donors, find each die outline honestly — datasheet, diagram, or thermal shadow — and draw both maps: die-shadow ring on the first, corner-weighted on the second, marking the highest-risk joints on each (intermetallic-compounds-and-joint-strength).
  2. Build the four-class table (5 minutes, desk). One row per class — mechanical with both its faces, process, wetting, and short with both its members — with columns for mechanism, geography on the field, alloy dependence, timing signature, and symptom signature, in your own hand as the bench's sorting card (why-joints-fail-metallurgical-causes).
  3. Sort three cold cases (5 minutes, desk). Write the class call, mechanism, and confidence for each: a tablet that boots only after five minutes against a warm charger; a handheld console dead since a stair fall, unmoved by warmth or flex; a router that shipped flaky from a repair depot's reflow and failed within the week — then note what evidence would raise each confidence (bga-ball-composition-and-alloys).
  4. Write the migration watch card (5 minutes, desk). List the triad — ionic residue, moisture, bias — with one realistic source of each at a rework bench; write the timing signature that separates migration from a process bridge, the discipline that starves each leg, and where the inspection chapter will look when the card's suspicion fires — then close with one line for the bridge side: the process discipline that prevents it and the first-boot inspection that finds it (what-is-bga).

These core steps — the stress maps, the class table, the cold-case sort, and the migration card — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

  • The mechanical class kills two ways: cycling fatigue cracks the flexing standoff slowly through the solder's body, while drop breaks instantly through the brittle plane — intermetallic fracture — and the silver trade decides each device's vulnerability (bga-ball-composition-and-alloys).
  • The stress map is drawn per package: corners for the package-level plate story, the die shadow ring for die-dominated packages — the die outline aims every later inspection (intermetallic-compounds-and-joint-strength).
  • The process and wetting classes are born at reflow — cold, disturbed, pillowed, and unwetted joints carry first-boot signatures that indict the oven, not the years (why-joints-fail-metallurgical-causes).
  • The short class splits on the calendar: bridges are process failures present from first boot, while electrochemical migration grows dendrites in service wherever residue, moisture, and bias conspire — and removing any leg stops the growth (what-is-bga).
  • The signatures are free evidence — warm-up intermittent, drop-dead brick, first-boot corpse, growing leakage — and the class hypothesis they buy halves the diagnosis before any instrument opens.

Skills Learned

  • You can now name the four failure classes with mechanisms and alloy dependence.
  • You can now draw a package's stress map from its die outline and mark the highest-risk joints.
  • You can now explain the drop-versus-cycling split — intermetallic plane versus solder body.
  • You can now read timing and symptom signatures into class hypotheses at the bench.
  • You can now state the shorts' split and the migration triad whose removal is the repair.

Glossary Additions

  • die shadow — the region of a BGA's ball field that sits beneath the package's die, and specifically the ring of joints under the die's edge where thermal-cycling strain concentrates: the die is the package's stiffest tenant — silicon far stiffer than substrate or overmold — so the field beneath it barely stretches while the substrate beyond its edge complies, and the boundary between the two regimes loads the joints under the die's perimeter hardest — worst of all under the die's projected corners. On die-dominated flip-chip packages — the SoCs and GPUs this volume meets — real failure maps cluster under the die edge rather than at the array's geometric corners, which is why the stress map is drawn per package from the die outline: corners for small-die and memory packages, the die-shadow ring for the big silicon, and the map aims X-ray fields, crack hunts, and reball inspections before any instrument spends an hour.
  • intermetallic fracture — the brittle failure that runs through a solder joint's intermetallic layer at the pad interface rather than through the solder's ductile body, and the signature break of drop and shock loading: impact loads the joint faster than solder can yield, the thin hard intermetallic plane cannot yield at all, and the fracture is instant, clean, and deaf to the warm-up behavior that betrays fatigue cracks. Stiff high-silver ball fields transmit shock into the plane hardest — with pad cratering, the board-side laminate tearing out beneath a pad, as the drop's board-side cousin — while softer alloys absorb impact at the joint; the distinction matters at the bench because a drop death and a cycling death take different repairs, and the calendar plus the symptom interview separates them for free.
  • electrochemical migration — the in-service short mechanism in which metal dissolves from one conductor and redeposits toward a neighbor under three simultaneous conditions — ionic contamination such as flux residue, moisture, and bias across the gap — growing dendrites ball-to-ball beneath a BGA until leakage becomes a hard short. Its timing signature separates it from process bridges: a bridge is present from first boot and indicts the reflow, while migration's leakage grows over days to weeks under power and indicts the conditions — and the repair is condition removal as much as short clearing, because a dendrite burned away regrows from residue and humidity left in place. Remove any leg of the triad and the growth stops, which makes cleaning discipline, moisture control, and post-rework washing short-prevention under other names.

Suggested Next Sections

Must read next:

  • BGA Repairability Assessment — Section 2.4 turns the catalog into arithmetic: which classes reflow and reballing genuinely address, which failures — cratered pads, cracked dies, torn substrates — no ball-level repair reaches, and how the bench prices the decision honestly before committing heat.

Recommended: