Dye penetrant inspection basics: how it works, what it finds, where it fails
A low-viscosity dye is drawn into a surface crack by capillary action, then wicked back out so you can see it. This page covers the five-step cycle, penetrant families, sizing limits, and the cases where dye penetrant inspection basics stop applying. Written for design and manufacturing engineers checking whether PT fits a part.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
What dye penetrant inspection actually detects
Dye penetrant inspection is a capillary test, not a volumetric one. A liquid carrying dye or fluorescent pigment is spread over a clean surface and left to sit. Where an opening reaches the surface, capillary pressure pulls the liquid in. After the excess is washed off, a developer draws the trapped liquid back out, and the stain marks where the defect sits.
The word to keep in mind is surface-breaking. A void 3 mm below the skin, a shrink cavity with a closed lid, or a crack that stops one grain short of the surface all stay invisible. If you need internal defects, that is ultrasonic or radiographic work, not PT.
The method is indifferent to the base metal. Aluminum, stainless, titanium, cast iron, ceramics, and many plastics all take the same test with the same five steps. What changes is the cleaning chemistry and how much the surface texture fights you.
One more boundary. PT does not measure depth. A wide shallow scratch and a tight fatigue crack can both show a line. Judging which one matters is a separate decision, and it is the point where dye penetrant inspection basics end and engineering judgment starts.
The five-step cycle and typical dwell times
Pre-clean, penetrate, remove excess, develop, inspect. Clean off afterward if the part is going to a customer or into an assembly. That is the whole cycle, and most failures trace back to step one or step three rather than to the chemistry.
Penetrant dwell commonly runs 5 to 10 minutes. Developer dwell runs 10 to 30 minutes. Both numbers assume a part temperature between 10 and 50 °C. Cold shops slow everything down. Below about 10 °C the penetrant thickens, capillary draw weakens, and you either heat the part or accept a longer dwell that the specification may not allow.
Removal is where parts get ruined. Solvent removal needs a clean lint-free wipe, not a flood of solvent poured over the surface. Water removal needs controlled pressure, typically under 2 bar, and wash water in the 20 to 40 °C range. Over-wash flushes penetrant out of shallow defects and produces a clean, confident, wrong result.
Inspection follows immediately after development. Read indications while the bleed-out is still growing, not an hour later when the developer has spread the stain into a wide pink haze that tells you nothing about size.
Penetrant types, sensitivity levels, and developer choices
Three penetrant families cover most shop work. Solvent-removable is the most forgiving and the easiest to run on a job-shop floor, but it is the least sensitive because the solvent wipe can pull penetrant back out of shallow openings. Water-washable is fast and cheap at volume, and it demands tight control of wash pressure and temperature. Post-emulsifiable sits between the two, with a separate emulsifier step that gives you control over removal but adds a variable.
Sensitivity is graded, commonly from level 1/2 (low) through level 3 (standard) to level 4 (high). Fluorescent penetrants generally outperform visible dye on tight defects because the eye picks up a glowing line in a dark booth far better than a red line on a bright bench.
Developer is not a formality. It provides contrast and, more importantly, it wicks penetrant back out of the defect. Dry powder developer gives the crispest indications and the least bleed-out, which helps on tight cracks. Aqueous and non-aqueous wet developers apply more uniformly and suit complex geometry. A film that is too thick hides indications under its own coating.
Match the level to the consequence. A cosmetic cover plate and a hydraulic manifold do not need the same sensitivity, and paying for level 4 on both is wasted money and wasted booth time.
Surface preparation: the dominant variable
Surface finish dominates results more than any other single factor. On a machined surface around Ra 1.6 μm, indications stay clean and readable. At Ra 6.3 μm or rougher, penetrant traps in tool marks and blast texture, and the resulting background of false indications swamps the real ones.
Castings and forgings often need machining, grinding, or at least a light etch before PT means anything. Blasting is the wrong preparation for a part you intend to inspect, because it folds metal over the opening you are trying to find.
Cleaning before penetrant is the step most often done badly. Oils, cutting fluid, paint, oxide, and plating all block penetrant from entering a defect. A solvent wipe handles light contamination. Vapor degreasing or an alkaline wash is better for machined parts carrying coolant residue.
The part must be dry afterward. Trapped moisture dilutes penetrant and blocks capillary action, and a wet part gives you a faint, patchy, untrustworthy result that is easy to misread as a clean part.
Defect sizing limits and false indications
The practical detection floor is a crack roughly 0.1 mm wide at the surface, and even that depends on finish and lighting. Wider shallow defects are easier to find than tight deep ones, because capillary draw competes with the difficulty of cleaning penetrant back out of a narrow opening. A tight fatigue crack may hold penetrant well but release very little, giving a faint indication that is easy to miss.
False indications come from three places. Surface texture traps dye in tool marks. Joints, rivets, and press fits hold penetrant in gaps that are not defects. And incomplete cleaning leaves residue that bleeds out slowly and looks like a large indication.
Geometry matters too. Sharp internal corners, threaded holes, and keyways all collect penetrant. Reading those areas takes experience, and a second look under different lighting is cheap insurance.
None of this makes PT unreliable. It makes PT a test whose result depends on preparation and on the person reading it. That is why the specification controls the process, not just the chemical.
When penetrant inspection is the wrong choice
PT is the wrong test for porous material. Powder-metal parts, unsealed castings, and some ceramics absorb penetrant across the whole surface, and the entire part glows or bleeds. You cannot separate a defect from the background.
It is also wrong when the defect you care about is internal or when it is subsurface. Weld root defects, internal shrinkage, and delamination belong to UT or RT. Using PT there produces a pass certificate that means less than it appears to.
Rough as-cast or as-forged surfaces are a poor fit unless you machine or etch first. So are parts with deep blind holes, heavy threading, or lap joints, where entrapment produces indications you cannot resolve.
And it is the wrong choice when throughput is the constraint. A five-step wet cycle with 10 to 30 minutes of developer dwell does not fit a high-volume line the way an eddy current or visual check might. Pick the test that matches the defect, the surface, and the takt time.
Penetrant family comparison
Dwell times assume 10–50 °C part temperature
| Family | Removal method | Relative sensitivity | Best fit |
|---|---|---|---|
| Solvent-removable | Wipe with clean solvent | Lowest | One-offs, field work, mixed alloys |
| Water-washable | Controlled water rinse | Medium | High volume, simple geometry |
| Post-emulsifiable | Emulsifier then water rinse | Highest | Tight cracks, critical parts |
| Fluorescent (any family) | Same as family | Higher than visible dye | Dark booth, tight defects |
| Visible dye | Same as family | Lower | Shop floor, no booth available |
The short version
If the defect reaches the surface and the finish is Ra 1.6 μm or better, PT is fast and reliable. If the defect is internal, the surface is rough, or the material is porous, choose another method instead of arguing with the indication.
Common questions
Can dye penetrant inspection find cracks under plating?
No. Plating, paint, and conversion coatings seal the surface opening, so penetrant never enters the defect. The part has to be stripped or inspected before finishing.
If the coating is thin and porous you may get a faint, unreliable indication. Treat that as a reason to inspect before plating, not as a valid result.
How small a crack can PT detect?
Roughly 0.1 mm wide at the surface under good conditions, with a clean machined finish and proper lighting. That is a practical floor, not a guarantee.
Tight deep cracks can hold penetrant but release very little, so the indication may be faint. Fluorescent penetrant under UV-A in a darkened booth improves the odds considerably.
Why do I get indications that turn out to be nothing?
Usually surface texture or entrapment. Tool marks, blast texture, threads, press fits, and lap joints all hold penetrant and bleed it back out during development.
Over-washing produces the opposite error: a clean surface and a missed defect. Both are process problems, not chemistry problems.
Does PT work on titanium and stainless steel?
Yes. The method does not care about the alloy. Cleaning chemistry is the part that changes, and chloride-free products matter for stainless and titanium where stress-corrosion is a concern.
Austenitic stainless and titanium both take PT well after machining, provided the surface is dry and free of coolant residue.
How long does the whole cycle take?
Plan on 30 to 60 minutes per batch including cleaning, 5 to 10 minutes of penetrant dwell, removal, and 10 to 30 minutes of developer dwell at 10 to 50 °C.
Cold parts extend every step. Below about 10 °C, warm the part or expect weaker capillary draw and longer dwells.
Send the drawing, get a process answer
We review the part, the material, and the defect you are worried about, then tell you whether PT is the right test or something else is.
12-hour quote100% inspectionDFM feedback