The Repair LibraryRead · Learn · Master

Thermal Cycling and Fatigue

The opening section of this chapter named wear-out as the rising edge of the bathtub curve, the region where aging parts reach the end of their physical life; this section takes up the single mechanism that, more than any other, drives that wear-out in real boards — the fatigue that repeated heating and cooling forces into solder joints and the parts they hold. The starting fact is that every powered board changes temperature, and not only because the room does: a board heats when it is switched on and cools when it is switched off, and the parts that dissipate the most power swing through the largest temperature changes of all, driven by their own duty cycle rather than the weather. The reason this matters is that the materials joined together on a board — the component body, the solder that holds it, the copper pad, the fiberglass board beneath — do not expand by the same amount when they warm, because each has its own coefficient of thermal expansion, and when materials that are bonded together try to grow by different amounts the difference has to go somewhere. Where it goes is into mechanical strain at the joint that ties them, and the compliant material that absorbs most of that strain, cycle after cycle, is the solder. No single warming does visible harm, but the strain repeats every time the board heats and cools, and the damage accumulates: this is thermal fatigue, the cumulative, cycle-by-cycle wearing of a joint that fails not from one overload but from the patient repetition of a stress far below the level that would break it once. The failure it produces is the fatigue crack that slowly works its way through a solder joint until the joint goes intermittent and finally open, and it falls hardest on large, hot packages, whose corner joints sit farthest from the package center and so see the greatest differential movement — which is why a ball-grid array that ran perfectly for years can develop the classic warm-it-and-it-works, cool-it-and-it-fails intermittent that a temporary reflow relieves without curing. The section teaches this as the dominant wear-out mechanism a technician meets: how the coefficient of thermal expansion mismatch and the temperature swings of power cycling combine to set the practical service life of much of what fails, how to recognize a thermal-fatigue failure by its age, its thermal history, and its temperature-dependent, large-package signature, and why a repair that merely reconnects the crack without addressing the joint and the thermal stress that made it has reset the fatigue clock to nowhere near zero, and will meet the same failure again.

AdvancedLow Risk23 min read

What You Will Learn

  • You will learn that every powered board is thermally cycled by ambient change and by its own power dissipation.
  • You will learn how a coefficient of thermal expansion mismatch forces mechanical strain into a solder joint each cycle.
  • You will learn that thermal fatigue accumulates cycle by cycle until a joint cracks, a wear-out rather than an overload.
  • You will learn why large, hot packages such as BGAs fatigue first, at their corner joints.
  • You will learn to recognize a thermal-fatigue failure and why a bare reflow does not cure it.

What You Will Be Able To Do

  • You will be able to explain how thermal cycling drives fatigue through a coefficient of thermal expansion mismatch.
  • You will be able to distinguish thermal fatigue as a cumulative wear-out from a single-overload failure.
  • You will be able to explain why power cycling stresses a part's own joints independently of ambient temperature.
  • You will be able to recognize the age, thermal, and large-package signature of a thermal-fatigue failure.
  • You will be able to explain why a repair must address the joint and the thermal cause, not just reconnect the crack.

Required Tools

  • A board with a large, hot package — a CPU, GPU, or power BGA — the classic site where thermal fatigue concentrates
  • The thermal history of a device in mind — how hot it runs and how often it is powered on and off, the drivers of thermal cycling
  • A source of gentle warming and cooling — to observe how a temperature-dependent intermittent tracks a fatigue crack opening and closing
  • The bathtub curve from the previous section — because thermal fatigue is the mechanism behind much of its rising wear-out edge

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 previous section named wear-out as the rising edge of the bathtub curve; this section takes up the mechanism that, more than any other, drives it in real boards — the fatigue that repeated heating and cooling forces into solder joints (mtbf-and-component-reliability). Every powered board changes temperature, and not only because the room does: it heats when switched on and cools when switched off, and the parts that dissipate the most power swing furthest, on their own duty cycle. The materials bonded together — component body, solder, copper pad, fiberglass — each have their own coefficient of thermal expansion, so when they warm they try to grow by different amounts, and the difference is forced as mechanical strain into the joint that ties them, absorbed cycle after cycle by the compliant solder (why-joints-fail-metallurgical-causes). No single warming does visible harm, but the strain repeats and the damage accumulates: this is thermal fatigue, a wear-out in which a crack grows cycle by cycle from a stress far below the one-time breaking level, until the joint goes intermittent and finally open. It falls hardest on large, hot packages whose corner joints sit farthest from the package center and see the greatest differential movement, which is why a BGA that ran for years develops a warm-it-works, cool-it-fails intermittent (bga-failure-modes). And because the device's own power cycling drives this as surely as the weather, a repair that only reconnects the crack, without addressing the joint and the thermal stress that made it, will meet the same failure again (capacitor-and-inductor-failure-modes).

Why This Matters

This is the section that explains why so much of what a technician sees fail, fails when it does — the mechanism on the rising edge of the wear-out curve. This matters because thermal fatigue sets the practical service life of much of what fails: the rising wear-out edge of the bathtub curve is, for a great many boards, thermal fatigue working through their solder joints, so understanding it is understanding why devices die when they die (mtbf-and-component-reliability). This matters because it explains the most maddening class of intermittent faults: a fatigue crack opens when the joint is warm and expanded and closes when it is cool, or the reverse, producing the temperature-dependent, comes-and-goes fault that defies a static test and that a momentary reflow relieves without curing (bga-failure-modes). It matters because the cause is mechanical and material, not electrical: the joint is not failing from too much current but from a coefficient of thermal expansion mismatch flexing it a little on every cycle, so the fix is mechanical and thermal, not a matter of a bigger part (why-joints-fail-metallurgical-causes). And it matters because a repair can quietly restart or worsen the clock: a reflow that reconnects a crack leaves the fatigue mechanism in place, and a repair that runs a part hotter or cools it worse shortens the fatigue life further, so a repairer who understands power cycling and fatigue fixes the joint and the heat, not just the symptom (capacitor-and-inductor-failure-modes). See the temperature swings, understand the mismatch that turns them into strain, and recognize fatigue as a cumulative wear-out — and a whole class of failures becomes legible instead of mysterious.

Required Prerequisites

Before starting this section, you should have completed:

  • MTBF and Component Reliability — the reliability view and the bathtub curve, whose wear-out region this section explains as, for many boards, thermal fatigue working through their joints.
  • Why Joints Fail — Metallurgical Causes — the metallurgy of joint failure, including CTE mismatch and solder fatigue, the material grounding this section builds into a reliability mechanism.
  • A board with a large, hot package — a CPU, GPU, or power BGA, the classic site where thermal fatigue concentrates, so the mechanism is studied where it actually bites.
  • A can of freeze spray and a source of gentle heat — to open and close a temperature-dependent fatigue crack on demand, making an intermittent fault reveal itself.
  • A notebook for a thermal history — recording how hot a device runs and how often it is cycled, because fatigue is driven by the pattern of heating over time, not a single reading.
  • A device with a temperature-dependent intermittent — one that works warm and fails cool, or the reverse, so a fatigue crack can be observed opening and closing.
  • An aging device with a hot, large package — old enough and hot enough to be on the wear-out edge, so thermal fatigue is met as a real wear-out rather than a diagram.
  • A board with an underfilled or well-cooled BGA next to a bare one — so the effect of measures that reduce strain per cycle on fatigue life can be seen side by side.

Real-World Applications

Thermal fatigue is the mechanism behind a whole family of failures a technician meets. A bench facing a device that works warm and fails cold recognizes a fatigue crack opening and closing with temperature, not a random glitch (bga-failure-modes). A repairer of an aging games console or GPU sees the corner joints of a hot, large package as the first to fatigue, and looks there (mtbf-and-component-reliability). A technician tempted by a quick reflow knows it may reconnect the crack but does not reset the fatigue clock, so plans for the joint and the heat, not just the reconnection (why-joints-fail-metallurgical-causes). And a repairer weighing a cooling upgrade or underfill understands these reduce the strain per cycle and so extend fatigue life (capacitor-and-inductor-failure-modes). The confusions this prevents: a temperature-dependent intermittent mistaken for a random fault, a fatigue failure blamed on a component rather than its joint, a reflow mistaken for a cure, and a repair that runs a part hotter and shortens the life it meant to restore.

Common Challenges

  • The fault is intermittent and temperature-dependent. It works warm and fails cool, or the reverse — because a fatigue crack opens and closes with the joint's expansion, defying a static test (bga-failure-modes).
  • The cause looks electrical but is mechanical. A joint goes open with no overcurrent — because the damage is a CTE mismatch flexing it each cycle, not a current fault (why-joints-fail-metallurgical-causes).
  • A reflow seems to fix it, then it returns. Heat reconnects the crack briefly — but the fatigue mechanism is untouched and the failure comes back (mtbf-and-component-reliability).
  • The damage hides under a large package. The failing joints are corner balls beneath a BGA — out of sight, where thermal fatigue concentrates and visual inspection cannot reach (capacitor-and-inductor-failure-modes).

Safety Notes

Risk Level: Low. This section is concepts and observation — thermal cycling, expansion mismatch, and fatigue — and the standing bench law and the soldering and rework volumes' cautions govern any heating, reflow, or hot hardware involved.

  • Hot, large packages stay hot after power-off — the processors and power devices where fatigue concentrates can burn well after shutdown; treat them as hot until confirmed cool.
  • Any reflow or hot-air work carries the rework volumes' full hazards — burn, fume, and adjacent-component risks apply the moment heat is involved; this section adds understanding, not a licence to improvise.
  • Freeze spray and diagnostic heat need care — short controlled bursts with ventilation, never onto powered high-voltage sections, and heat gentle enough not to reflow or stress a neighbor.

Professional Tips Before Starting

  • Count the temperature swings, not just the temperature. Fatigue is driven by how often and how far a joint cycles — so read a device's power-on pattern and duty cycle, not only how hot it runs (mtbf-and-component-reliability).
  • Suspect the corners of the hottest, largest package first. Differential movement is greatest farthest from the package center — so a fatigue crack usually starts at a corner ball (bga-failure-modes).
  • Treat a temperature-dependent intermittent as a fatigue signature. Works-warm-fails-cool, or the reverse, tracks a crack opening and closing — so provoke it gently with heat or cold to confirm (why-joints-fail-metallurgical-causes).
  • Never mistake a reflow for a cure. Reconnecting a crack does not reset the fatigue life — so plan for the joint and the thermal cause, not just the reconnection.
  • Do not add thermal stress in the repair. A part left hotter or cooled worse fatigues faster — so a repair should lower the thermal stress, or at least not raise it (capacitor-and-inductor-failure-modes).

How Thermal Cycling Wears a Board Out

The Expansion Mismatch — Why Heating Strains a Joint

A board is an assembly of unlike materials bonded together, and that is the root of the whole mechanism (why-joints-fail-metallurgical-causes). A component's body, the solder that holds it, the copper of the pad, and the fiberglass of the board each expand when they warm — but not by the same amount, because each has its own coefficient of thermal expansion, the measure of how much a material grows per degree of temperature rise. When these bonded materials warm together, each tries to grow to its own size, and because they are fixed to one another they cannot: the one that wants to grow more is held back by the one that wants to grow less, and the difference appears as mechanical strain concentrated at the interface between them. On a board, that interface is the solder joint. The solder is the compliant material caught between a component that expands at one rate and a board that expands at another, and it absorbs the mismatch by deforming a little each time the assembly heats and cools — a small flex on warming, a small flex back on cooling, invisibly, on every cycle. This is the quiet violence beneath a working board: not a fault, not an overload, but a relentless small strain worked into every joint by nothing more than the ordinary rise and fall of temperature, and it is the reason a joint that was sound the day it was made does not stay sound forever. The larger the temperature swing and the greater the mismatch in expansion, the more strain each cycle drives into the joint — which is why heat and dissimilar materials, the two ingredients present in every board, are together the setup for the wear-out this section describes.

Thermal Fatigue — How Repeated Strain Cracks a Joint

One flex does nothing; a hundred thousand flexes is another matter, and that is the essence of thermal fatigue (mtbf-and-component-reliability). Fatigue is failure by repetition: a material stressed once well below its breaking point takes no visible harm, but stressed the same way over and over, it accumulates microscopic damage that grows, cycle by cycle, into a crack — and eventually the crack is large enough to matter. In a solder joint, each thermal cycle works the small expansion-mismatch strain into the metal, and the damage builds where the strain concentrates: a crack initiates, usually at a high-stress corner or edge of the joint, and with each further cycle it grows a little farther across the joint. For a long time nothing is visible and the joint works perfectly — the fatigue is accumulating silently, spending a service life the datasheet never printed. Then the crack reaches the point where it interrupts the connection, and the joint goes intermittent: it conducts when the crack is squeezed closed and opens when the crack is pulled apart, which is exactly why the fault tracks temperature. Finally the crack completes and the joint is permanently open. The crucial thing to understand is that this is a wear-out, not an overload: the joint did not fail because something exceeded its rating on one bad day, but because an ordinary, in-spec stress repeated until it wore the joint through — the same distinction the bathtub curve draws between the random failures of useful life and the rising, age-driven failures of wear-out. Thermal fatigue is the mechanism that populates that rising edge for a great many boards, and recognizing it means recognizing that the joint spent its life being flexed, not being overstressed.

Power Cycling and the Large-Package Signature

Two facts turn the general mechanism into a specific, recognizable failure a technician meets again and again. The first is power cycling — the truth that a board does not need the room to change temperature to be thermally cycled, because it heats itself (capacitor-and-inductor-failure-modes). A processor or power device that draws watts warms sharply when it is working and cools when it is idle or off, swinging through a temperature change driven entirely by its own duty cycle, often larger and faster than anything the ambient does. This is why the hottest, busiest parts fatigue their own joints: their duty cycle is their fatigue clock, and a device switched on and off daily, or a chip that spikes hot under load, is cycling its joints far more than its calendar age alone would suggest. The second fact is where the damage concentrates. A large package — a ball-grid array especially — spans a wide area, and the farther a joint sits from the package's center, the more the differential expansion between package and board moves it, because the mismatch accumulates over distance (bga-failure-modes). So the corner balls of a big BGA see the greatest strain per cycle and fatigue first, which is why the classic thermal-fatigue failure is a large, hot chip that develops a corner-joint crack and, with it, the unmistakable signature: a device that worked for years, then began to fail in a temperature-dependent way — working when warm and failing when cool, or the reverse — that a press on the package or a burst of heat momentarily cures. That signature, age plus heat plus a large package plus a temperature-dependent intermittent, is thermal fatigue announcing itself, and a reflow that flows the cracked joint enough to reconnect it treats the symptom while leaving the mechanism, the heat, and the fatigue life exactly where they were.

Common Mistakes

  • Reading a temperature-dependent intermittent as random. A works-warm-fails-cool fault is called a glitch — when it is a fatigue crack opening and closing with the joint's expansion (bga-failure-modes).
  • Blaming the component instead of the joint. A chip is replaced when the fault is in its solder — because the failure was mechanical fatigue at the joint, not a dead part (why-joints-fail-metallurgical-causes).
  • Trusting a reflow as a permanent fix. Heat reconnects the crack and the device works — but the fatigue mechanism is untouched, so the failure returns on its own schedule (mtbf-and-component-reliability).
  • Ignoring the thermal cause. The joint is repaired but the part still runs as hot and cycles as often — so the same fatigue that cracked it the first time simply begins again (capacitor-and-inductor-failure-modes).
  • Overlooking the hidden corner joints. A visual pass over a BGA finds nothing — because the fatigued joints are corner balls underneath, beyond the reach of the eye.

Troubleshooting Guidance

  • A fault comes and goes with temperature — suspect a thermal-fatigue crack: a joint that conducts warm and opens cool, or the reverse, is a crack expanding and contracting, so provoke it gently with heat or freeze spray to confirm and localize (bga-failure-modes).
  • A device failed after years of good service — read it as wear-out, likely thermal fatigue: an age-driven failure on the bathtub curve's rising edge is often a joint fatigued by its thermal history, not a random event, so look at the hottest, largest packages first (mtbf-and-component-reliability).
  • A reflow fixed it but the fault returned — the crack was reconnected, not the fatigue cured: reflow does not reset the fatigue life, so a lasting repair addresses the joint itself and the heat that fatigued it (why-joints-fail-metallurgical-causes).
  • A repaired board fails the same way again — the thermal cause was left in place: if a part still runs as hot and cycles as often, its joints resume fatiguing, so reducing the thermal stress is part of the fix, not an extra (capacitor-and-inductor-failure-modes).

Verification & Testing Methods

Confirm your grasp of thermal cycling and fatigue before continuing:

  • [ ] I can explain how a coefficient of thermal expansion mismatch forces mechanical strain into a solder joint on each thermal cycle.
  • [ ] I can describe thermal fatigue as cumulative, cycle-by-cycle damage that cracks a joint, a wear-out rather than an overload.
  • [ ] I can explain how power cycling stresses a part's own joints from its duty cycle, independently of ambient temperature.
  • [ ] I can explain why large, hot packages such as BGAs fatigue first, at their corner joints.
  • [ ] I can recognize a thermal-fatigue failure by its signature and explain why a bare reflow does not cure it.

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

Practice Exercises

  1. Trace the expansion mismatch (5 minutes, a board with a large package). For a component on a board, name the bonded materials at its joints and explain, from their differing expansion, where the strain concentrates each time the assembly heats and cools (why-joints-fail-metallurgical-causes).
  2. Reason about fatigue as a wear-out (5 minutes, on paper). Explain why a joint stressed far below its one-time breaking level still fails over time, and place that failure on the bathtub curve's wear-out edge rather than among random useful-life failures (mtbf-and-component-reliability).
  3. Connect a part's own heating to its joints (5 minutes, a hot device). For a device with a power-hungry chip, describe how its duty cycle cycles its own joints independently of the room, and why that makes it fatigue faster than its calendar age suggests (capacitor-and-inductor-failure-modes).
  4. Recognize the large-package signature (5 minutes, a case). For an aging device with a temperature-dependent intermittent under a hot BGA, lay out the signature — age, heat, large package, temperature-dependent fault — and explain why a reflow would relieve it without curing it (bga-failure-modes).

These core skills — tracing the mismatch, reasoning about fatigue as a wear-out, connecting power cycling to a part's joints, and recognizing the large-package signature — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

  • Every powered board is thermally cycled — by the room and by its own heating — and because its bonded materials each have a different coefficient of thermal expansion, each cycle forces mechanical strain into the solder joints that tie them together (why-joints-fail-metallurgical-causes).
  • Thermal fatigue is the cumulative result: a joint stressed far below its one-time breaking level by an ordinary temperature swing accumulates damage cycle by cycle until a crack initiates, grows, goes intermittent, and finally opens — a wear-out, not an overload (mtbf-and-component-reliability).
  • The device's own power cycling drives this as surely as ambient change — a power-hungry chip cycles its own joints on its duty cycle, so the hottest, busiest parts fatigue their joints fastest, regardless of the calendar (capacitor-and-inductor-failure-modes).
  • The damage concentrates at the corner joints of large, hot packages, farthest from the package center where differential movement is greatest, which is why a BGA develops a temperature-dependent intermittent — the classic thermal-fatigue signature (bga-failure-modes).
  • A reflow may reconnect a fatigue crack but does not reset the fatigue life, so a lasting repair addresses the joint and the thermal stress that made it — reducing the heat and cycling, not just restoring the connection.

Skills Learned

After completing this section, you can:

  • Explain how thermal cycling drives fatigue through a coefficient of thermal expansion mismatch.
  • Distinguish thermal fatigue as a cumulative wear-out from a single-overload failure.
  • Explain why power cycling stresses a part's own joints independently of ambient temperature.
  • Recognize the age, thermal, and large-package signature of a thermal-fatigue failure.
  • Explain why a repair must address the joint and the thermal cause, not just reconnect the crack.

Glossary Additions

New terms introduced in this section:

  • coefficient of thermal expansion — the material property that measures how much a material grows in size per degree of temperature rise, usually abbreviated CTE. It is the root of thermal-cycling failure because the materials bonded together on a board — a component's body, the solder that holds it, the copper pad, the fiberglass substrate — each have a different coefficient of thermal expansion, so when they warm together they try to grow by different amounts. Because they are fixed to one another they cannot each reach their own size, and the difference is forced as mechanical strain into the interface that ties them, which on a board is the solder joint. The greater the mismatch in expansion between two bonded materials, and the larger the temperature swing, the more strain each thermal cycle drives into the joint — making the CTE mismatch, present in every board by the simple fact that unlike materials are joined, the setup for thermal fatigue.
  • thermal fatigue — the cumulative wearing-out of a solder joint — or other material — by repeated thermal cycling, in which the small expansion-mismatch strain of each heating and cooling accumulates as microscopic damage until a crack forms, grows, and finally opens the joint. Its defining feature is that it is a fatigue failure, not an overload: the joint is not broken by a single stress exceeding its rating but by an ordinary, in-spec strain repeated tens or hundreds of thousands of times, spending a service life the datasheet never prints. The failure progresses through recognizable stages — a crack initiates at a high-stress corner, grows across the joint over many cycles while the connection still works, then reaches the point of an intermittent that tracks temperature as the crack opens and closes, and finally completes to a permanent open. Thermal fatigue is the mechanism behind much of the rising wear-out edge of the bathtub curve, and a repair that reconnects the crack without addressing the joint and the thermal stress leaves the mechanism intact.
  • power cycling — the thermal cycling a device inflicts on itself through its own power dissipation, as distinct from cycling driven by ambient temperature change. A part that dissipates significant power — a processor, GPU, regulator, or power BGA — warms sharply when it is working and cools when it is idle or switched off, swinging through a temperature change set by its own duty cycle rather than the weather, and often larger and faster than the ambient ever produces. Power cycling matters for reliability because it means the hottest, busiest parts are cycling their own solder joints most heavily, so a device's fatigue life is driven by how it is used — how often it is switched on and off and how hard it is loaded — as much as by its age. It is why two identical devices of the same age can have very different remaining fatigue life, and why reducing a part's operating temperature and the severity of its cycling extends the life of its joints.

Suggested Next Sections

Must read next:

  • Design for Repairability — Section 6.3 turns at Professional depth from why parts wear out to how a design is built to be maintained: reading a product for how reliably and how easily it can be serviced, and how repair choices preserve or degrade that.

Recommended:

  • BGA Failure Modes — the package where thermal fatigue concentrates, and the corner-joint cracking and temperature-dependent faults this section explains as a reliability mechanism.
  • Capacitor and Inductor Failure Modes — other age-driven wear-out mechanisms alongside thermal fatigue, the individual ways parts reach the end of their physical life.