White crystals or sticky residue on a board after soldering will not necessarily cause failure, and cannot be judged by appearance alone. Whether flux residue causes leakage, corrosion or dendrite shorts depends on three conditions holding at the same time: mobile ions, moisture, and voltage between conductors (the smaller the spacing, the more sensitive). No-clean fluxes are designed to leave “little and mild” residue, which does not mean any operating environment can skip cleaning; whether to clean depends on the operating environment and the board’s features, and needs objective evidence behind it.

1. Why residue leaks: all three conditions are needed

The trouble with residue is that a board may measure normal when it leaves the factory and only fail once moisture gets in.

Leakage first. Ions in the residue meet moisture and dissolve into a thin conductive film spread between adjacent conductors, and insulation resistance falls. Residues from different formulations absorb moisture to different degrees — some suppress it, some promote it — so what matters is whether the residue absorbs moisture and carries ions, not just “is there residue”.

Then dendrites. With voltage between conductors, metal at the anode dissolves into ions that move toward the cathode in the electric field and deposit there, gradually growing into a dendrite (a tree-like metal crystal); when the dendrite reaches the opposite conductor, there is a short. This process is called electrochemical migration (ECM) and needs moisture, mobile ions and a large enough potential gradient at the same time (see Figure 1). The potential gradient depends on voltage and spacing: the higher the voltage and the closer the conductors, the stronger the field, so fine-pitch, high-voltage boards are sensitive.

Dendrites are very fine and may fuse open when current rises, so a board with intermittent field leakage often can’t be reproduced back in the lab.

Electromigration (EM) is not ECM. Electromigration occurs inside a metal conductor, with electrons at high current density pushing metal atoms; it is common in IC interconnects, typically results in voids, and has nothing to do with moisture. ECM occurs between conductors and results in dendrites. Leakage and shorts from flux residue are ECM. Other terms such as “electron migration” or “ion migration” are also used, so when reading sources, first check where the described process takes place.

CAF is another location. Conductive anodic filament (CAF) grows along the glass fiber–resin interface inside the laminate; it is inside the board and is also an electrochemical migration process. Distinguish first whether the leakage path is on the board surface or inside it, to tell residue from laminate problems.

2. Step 1: identify the residue on the board

What you see on the board	Common causes	Check first
White powder or a hazy white film around joints	Uncleaned flux activator salts; cleaning agent not matched to the flux; incomplete cleaning	Look at the pattern: a uniform film over the whole board points first to a mismatched cleaning agent; streaks or spots point to cleaning technique. Then sample for composition analysis
Sticky, gummy residue	Rosin or resin components; too much flux; insufficient heat, not fully activated	Compare with the flux data sheet’s recommended profile and amount to confirm whether it is normal residue for that formulation
Visible white crystals on boards from a water-soluble flux process	Insufficient cleaning; residual activator reacting with moisture	Treat it as a warning: re-clean and confirm with ionic contamination testing
Under components and around fine pitch (mostly invisible)	Residue trapped under low-standoff components, not reached in one wash	Confirm the cleaning can get in; use localized extraction, not the whole-board average
White deposits on connectors or local areas, with field leakage or corrosion	Local contamination (flux or foreign liquid); diluted away in the whole-board average	Localized extraction plus ion chromatography; microscope for dendrites or corrosion marks
No visible residue, but leakage or shorts in high humidity	Residue isn’t always visible; dendrites form briefly and fuse open, hard to reproduce	Keep the board as is; reproduce under humidity, heat and bias; localized ion chromatography
White residue appears only after cleaning, or after a material change	Cleaning agent not matched to the flux; incompletely cured solder mask can also turn white	Compare recent changes of paste, solder and cleaning agent; composition analysis to tell residue from laminate

Keep the evidence first. To find the root cause of a field failure, photograph the returned board and keep it as received before deciding whether to clean it — once washed, the evidence is gone. Residue isn’t always visible, and white residue isn’t always flux; don’t rely on your eyes alone.

A simple first screen (general guidance; individual cases vary). One supplier’s material says: put a small drop of water on the residue — if it dissolves, it is mostly ionic residue; if water doesn’t dissolve it but isopropyl alcohol (IPA) does, it is mostly organic residue. This only gives a rough split; do it on a sample or scrap board and dry it completely, and it does not replace analysis.

To identify the composition, send it for analysis. Fourier transform infrared spectroscopy (FTIR) shows the chemical structure of organic residue; ion chromatography (IC) identifies ion types and amounts and, with localized extraction, finds local contamination that the whole-board average misses.

3. Step 2: clean or not?

Operating environment / board features	General recommendation	Reason
Water-soluble flux used	Clean	Activator residue is corrosive and conductive; visible residue is already a warning
Conformal coating, potting, wire bonding or underfill afterwards	Clean first; to skip cleaning, validate the combination first	Residue affects adhesion and may trap moisture; results vary greatly with the coating–residue combination
High-voltage or high-frequency circuits	Lean toward cleaning	Residue may become a leakage path; the higher the voltage, the stronger the field
Fine pitch, low-standoff components (QFN, LGA, micro-BGA, flip chip)	Lean toward cleaning, and confirm the cleaning gets in	Residue is trapped under components; the cleaning agent must wet well to get in
High humidity, condensation, high temperature, or high-reliability fields (aerospace, automotive, telecom)	Lean toward cleaning, and verify with SIR or similar	Ionic residue absorbs moisture to form a conductive film; the harsher the environment, the less residue can be tolerated; in high-reliability fields cleaning is generally treated as a key step
Too much flux, insufficient heat, not fully activated	Clean	No-clean formulations’ low residue assumes full activation
High appearance requirements, or AOI and visual repair needed	Lean toward cleaning	Residue obscures component markings and joints and affects AOI
Components that can’t tolerate liquid or ultrasound (such as buzzers, unsealed switches, some relays, crystals), or finished assemblies already coated or potted	First assess whether this board can be cleaned; handle components that can’t be washed separately	Liquid may get into the body and not dry out, and ultrasound may damage components. See section 5, point 1
None of the above applies, the flux is fully activated, and the customer has no cleaning specification	Generally no cleaning needed, but have objective evidence	No-clean formulations are designed for low residue and low corrosion; still support this with SIR or similar evidence, not just the word “no-clean”

No-clean does not mean no validation. No-clean fluxes are designed for little residue and low corrosivity, assuming full activation during soldering; too much flux or local underheating can leave ionic residue. Generally, no-clean residue carries lower risk in mild environments, but whether it is harmless depends on the case — the key factors are humidity, voltage and spacing.

Coating is a common turning point. Coating suppliers generally recommend cleaning first: residue affects adhesion, and absorbed moisture may be released during curing and cause delamination. One technical report notes that whether a coating can go over no-clean residue depends on the coating–residue combination and must be validated case by case. Individual cases vary.

4. Step 3: prove it is clean

Cleaned doesn’t mean clean, and looking clean doesn’t mean clean. The four measurements below answer different questions; for which documents to prepare for an audit, see How do you prove a board is clean?

ROSE (IPC-TM-650 2.3.25): extracts surface ions with a solvent and deionized water mixture and measures the change in extract conductivity, converted to NaCl equivalent. It answers “is the overall ionic level out of control” and suits routine process monitoring; but it measures the whole-board average, so local contamination may be diluted, and measuring conductivity misses non-conductive residue.
Ion chromatography (IC): answers “which ions, and how much”, for root cause analysis, and can be combined with localized extraction. IC has no single pass line, and results may differ between labs; confirm the method before sending samples.
SIR (surface insulation resistance, IPC-TM-650 2.6.3.7): answers “will it actually leak under humidity, heat and bias”, by measuring insulation resistance under bias at 40 °C and 90% RH. A common pass threshold in flux qualification testing (J-STD-004) is not less than 10⁸ Ω (100 MΩ), but that is a judgement of the flux on a standard test coupon, not an acceptance value for your board.
THB (temperature-humidity-bias) testing on real boards: runs real boards or finished products under humidity, heat and voltage; some industry experts regard this as convincing evidence. Individual cases vary.

Use ROSE routinely to monitor whether the process is stable; when introducing a new process, flux or board type, use SIR or real-board testing to build evidence.

Be careful with old numbers. You may see “1.56 µg/cm² NaCl equivalent” in old specifications or customer documents. It was originally a ROSE process-monitoring reference that later came to be used as an acceptance line; current J-STD-001 no longer treats it as a general pass value, and there is no one cleanliness number for every product. If a customer still specifies it, follow the agreement; if not, the limit comes from your own process validation data.

JIAHON’s JHPB-TLZ20 ionic contamination tester: the product page states that it measures conductivity and ion concentration using a 3:1 isopropyl alcohol / deionized water extract at a constant 40 ± 1 °C; the page does not state which standard it follows, so if a customer specifies a particular method, please ask us to confirm first.

5. Once you decide to clean, check four things first

Are there components that can’t tolerate liquid or ultrasound? With water washing (including solvent washing), problems come from components whose bodies are not hermetically sealed: liquid gets in but can’t dry out. Common examples in public sources include buzzers, shielded modules, unsealed switches, humidity / gas / pressure sensors and camera lenses (one PCB and assembly service provider lists these as non-washable parts, because water trapped inside may later cause electrical problems); one piezo buzzer maker states its products are not hermetically sealed and cannot be washed; one relay maker says dust-cover and flux-protected types must not be immersion-washed, and only plastic-sealed types can be. Ultrasound is another risk: one crystal maker warns that ultrasonic cleaning may make the crystal resonate and crack; one MEMS sensor maker advises against ultrasonic cleaning of boards with MEMS. For finished assemblies already coated or potted, a technical article warns that ultrasound may damage coatings; the risk of liquid getting into potting was found only in a single industry statement, so follow that product’s specification.

With such components there are two practical routes: skip cleaning for the whole board (the service provider’s approach above), or solder the non-washable components after cleaning (often by hand, which is harder to control; single source). Either way, the basis is always the cleaning notes in that component’s data sheet; individual cases vary.

The cleaning agent must match the residue. Water-soluble, no-clean and rosin residues differ in polarity, and general-purpose solvents vary in effect, so know which flux you use before choosing. JIAHON’s stencil cleaner / flux cleaner (JH64 / JH46 / JH37): the product page states they are designed for no-clean and rosin fluxes; please give us the flux brand, model and process conditions when you ask, and an SDS is available on request.

The cleaning must get in. Fine-pitch and low-standoff boards depend on spray coverage and the cleaning agent’s wetting. The offline PCBA cleaning machine JHPBT-800P has a 610 × 600 mm cleaning area and supports both water and chemical cleaning; for water washing, plan the pure water and rinse water as well.

Verify after cleaning. Go back to step 3.

Standards referred to

Follow the current edition from each publisher.

IPC-TM-650 2.3.25, Detection and Measurement of Ionizable Surface Contaminants by Resistivity of Solvent Extract (ROSE) — IPC (Global Electronics Association)
IPC-TM-650 2.6.3.7, Surface Insulation Resistance — IPC (Global Electronics Association)
IPC-TM-650 2.6.25, Conductive Anodic Filament (CAF) Resistance Test — IPC (Global Electronics Association)
J-STD-004, Requirements for Soldering Fluxes — IPC
J-STD-001, Requirements for Soldered Electrical and Electronic Assemblies — IPC
IPC, How Clean Is Clean Enough? — what reliability risks a cleanliness test may miss (June 2026) — IPC (Global Electronics Association)

This article is general technical information. Whether to clean, the cleaning conditions and acceptance thresholds should be decided by your product’s operating environment, customer specifications and validation data; individual cases vary. Product specifications follow each product page; please ask for actual configuration and lead time.

Should Flux Residue Be Cleaned? Judging White Residue and a 3-Step Clean / No-Clean Decision