Electrostatic Discharge (ESD) Basics

Intuition

The shock you feel, and the one you don't

Walk across a carpet in dry weather, touch a door handle, and you get a sharp crack and a sting. That is electrostatic discharge: friction between your shoes and the carpet has scraped charge off one surface and onto the other, leaving you at a very different electrical potential from the handle. Touch it, and the imbalance equalises in a few billionths of a second.

The voltages involved are startling. Walking across a synthetic carpet in dry air can leave a person at 35 kV relative to ground. On a humid day the same walk might produce only 1.5 kV, because a thin film of moisture on every surface lets the charge leak away as fast as it builds up. Static trouble is therefore a winter problem and an air-conditioning problem.

What saves you is that almost no energy sits behind those volts, so the discharge stings and does nothing worse. A modern chip has no such margin. Its insulating layers are a few atoms thick, and its transistors are small enough to be destroyed by a current a human being would not notice at all. Worse, the threshold at which you can feel a discharge is far above the threshold at which a device is damaged. By the time you feel a spark, any sensitive part you were holding has already had a very bad day.

Your own senses are no use as a detector here. They register an event only when it is far larger than the one that ruins a part, and they stay silent for everything below that, so the only workable approach is to prevent discharges rather than to catch them.

Practitioner

Working so it does not happen

The numbers put a size on that gap. A person begins to feel a discharge at roughly 3 kV, while many current components are specified to survive only 100 V — and some RF and precision analogue parts considerably less. The entire span below your perception threshold goes unfelt and unlogged, and it is where most damage happens.

Charge builds up by triboelectric action: two materials touch and separate, and electrons transfer from one to the other. Both materials must be reasonably insulating for the charge to stay put, which is why the offenders on a bench are the ordinary plastics — bubble wrap, adhesive tape, polystyrene cups, plastic-bodied tools, ordinary zip bags and clothing made from synthetics. Metals and skin are conductive enough to equalise the moment they touch anything grounded.

Eliminating charge altogether is impossible, so prevention works on potential instead: keep every object at the same potential as anything it might touch. Everything on the bench — your body, the mat, the tools, the board, the bag — is tied to one reference point through a controlled path.

Handling a sensitive part, step by step

1. Set up the ground first. Connect a dissipative bench mat and your wrist strap to a common ground point, before any part comes out of its packaging.
2. Put the wrist strap on and check it. The strap must touch skin, and its cord must be plugged in. Test it with a strap tester if one is available; a strap with a broken cord looks exactly like a working one.
3. Clear the bench of ordinary plastics. Remove tape rolls, foam cups, bubble wrap, plastic folders and anything else that is not explicitly static-dissipative.
4. Bring the packaged part down onto the grounded mat before opening it. Let the outside of the bag equalise with the mat while the contents are still shielded.
5. Open the bag and handle the part by its body or its edges. Do not touch pins, pads or connector contacts.
6. Keep the board on the mat, not on your lap or a chair. Moving a board around picks up charge.
7. Ground any tool before it touches the board — soldering iron tip, tweezers, probe. Irons used for sensitive work should be tip-grounded and low-leakage.
8. Return the part to a shielded bag immediately when work stops, and close the bag. A part left unbagged on a shelf stays exposed for as long as it sits there.
9. Label anything suspect rather than reusing it. A part that took an unprotected knock is a candidate for latent failure, and shipping it costs far more than replacing it.

A wrist strap always carries a resistance of about a megohm in series with its cord, and the reason for it is personal safety rather than ESD. A megohm bleeds static charge away in a small fraction of a second, which is quite fast enough for the electronics. What the resistance adds is a ceiling on the current that could pass through you if you touched something live while strapped to ground.

Raising the relative humidity of a room cuts charge generation substantially, which is why ESD incidents cluster in air-conditioned offices and dry winters. Treat that as a mitigation and never as a substitute for grounding.

Packaging is a specified material. Pink antistatic bags are low-charging but do not shield; metallised shielding bags (usually silver-grey) both avoid generating charge and enclose the part in a conductive envelope. Black conductive foam holds IC pins at a common potential in transit. An ordinary clear zip bag does none of these things and is actively harmful — polythene is an excellent charge generator, so the "protection" is charging the part it contains.

Going deeper

Models, classifications and on-chip protection

The industry models ESD as three separate events, because the discharge waveforms differ enormously from one to the next.

• Human Body Model (HBM) — a charged person touching a grounded device. The model is a capacitor discharging through a resistance representing skin and body. It is slow by ESD standards, and it is the one most component-level ratings quote.
• Charged Device Model (CDM) — the device itself charges up (sliding down a shipping tube, moving through automated handling) and then discharges when one pin touches ground. The path has almost no resistance, so the current is far higher and the event far shorter — nanoseconds, with peak currents of several amperes. CDM is the dominant failure mode in modern automated assembly and the harder one to protect against.
• Machine Model (MM) — a charged piece of equipment discharging into a device. Current standards have largely superseded it with CDM.

Component ESD ratings appear on the datasheet's absolute-maximum table as HBM and CDM figures, and JEDEC classifies parts into levels from them. Both figures are routinely misread. They describe survival of a specified test waveform, which is a long way from a licence to expose the part repeatedly, and they describe the bare component. System-level immunity of the finished product is a separate standard, with much harsher waveforms applied to connectors and enclosures.

On-chip protection exists on essentially every commercial IC pin: clamp diodes to the supply rails, plus a rail clamp that turns on during an event and conducts the current harmlessly. It works, at a price. The structures add capacitance, which limits bandwidth — the reason high-speed and RF pins are the least protected on any part. They also need a supply rail to clamp to, so an unpowered board can be more vulnerable than a powered one. Their sizing assumes a test waveform rather than a cable-length transient arriving from outside the box.

That last point is why external interfaces get dedicated protection: a TVS diode across every connector line that leaves the enclosure, chosen so its clamping voltage sits below the protected device's absolute maximum and its standoff voltage above the signal's normal range. Layout matters as much as the part — a protection device with a long trace to ground has an inductance that dominates the clamp voltage during a nanosecond-scale event.

The most vulnerable structure in ordinary practice is the MOSFET gate. It is a capacitor with an extremely thin insulator, and its breakdown is destructive and permanent — see MOSFET operation and the gate-handling notes in MOSFET gate drive. Discrete MOSFETs, unlike ICs, frequently have no internal protection at all.

A formal ESD control programme (the international standard is IEC 61340-5-1, with the corresponding US document ANSI/ESD S20.20) is essentially the bench routine above written down, plus periodic verification: strap and mat resistance measured on a schedule, ionisers checked, packaging specified, and training recorded. Those standards insist on verification rather than procedure alone for the reason Layer 1 gave — nobody can feel whether the controls are working.