Lubinpla

Why powder coating adhesion failures hit only some batches

Summary

Powder coating adhesion failures on only some batches usually trace to a shared factor: substrate lot, pretreatment bath condition, oven position or shift. That holds as long as the powder, film build and test method stay constant. Put failed and passed batches side by side and count which lot, bath reading, oven zone or shift the failures share.

  • Steel mills use 25% to 95% scrap depending on mill type and product, and batch-to-batch impurity variation can make pretreatment work less well (Biller, Powder Coated Tough, 2021).

  • In one SGS investigation, spent rinse water on a failing powder coating line held 2.3% dissolved solids by weight, and salts crystallized on the steel before coating (SGS, 2021).

  • After the SGS changes, including a deionised final rinse, the line ran at a zero percent failure rate for the next 12 months of operation (SGS, 2021).

  • A chi-square comparison of counts may be poor if 20% of expected frequencies fall below 5; Fisher's exact test is the alternative (NIST Dataplot, 2007).

This answer covers adhesion loss that shows up on some production batches of powder coated metal but not on others. It does not cover a line where every batch fails, which points to a change that hit the whole process, or intercoat adhesion on recoated parts.

What do powder coating batches that fail adhesion usually share?

Powder coating batches that fail adhesion while others pass usually have one of four things in common: a substrate lot, the state of the pretreatment bath during the hour they ran, a cure condition such as oven position, or a shift and the way it handles parts. A published cause analysis lists human sweat, aging equipment, workpiece materials, degreasing materials, ceramic pretreatment materials, powders and the coating environment as causes of poor adhesion (Li, Zhang and Hu, 2024). That list tells you what can go wrong. The batch pattern tells you which one did.

Substrate or casting lot

Steel composition is not the same from one batch to the next, and the oil on hot rolled pickled and oiled steel can vary enough to change how the cleaning stage behaves (Biller, Powder Coated Tough, 2021). Castings bring a residue of their own. On die castings, release agent residues often cause problems in later painting (CHT, die cast information).

Pretreatment bath at batch time

If a bath drifts out of its range for a few hours, only the parts that went through it during those hours are affected.

Cure condition

Undercured powder is prone to chipping (Biller, 2021), and a heating rate that is too low is a listed cause of poor adhesion (Axalta Powder Coatings FAQ).

Shift and handling

Sweat from handling is one documented cause (Li, Zhang and Hu, 2024). It follows crews and glove practice, not lots.

Which records should you pull for failed and passed powder coated batches?

For powder coated batches, pull the same records for every failed batch and for an equal or larger number of passed batches from the same weeks: substrate lot or casting lot, pretreatment log entries for the hour each batch ran, rack and oven position, line speed, film thickness readings and the shift. The passed batches matter as much as the failed ones. A factor that shows up in every failed batch is a clue only if it is rare among passed batches. Biller notes that troubleshooting is extremely difficult without documentation of processes, maintenance and customer jobs (Biller, Powder Coated Tough, 2021).

The concentration table below sets the comparison axes, the factors you count failures against, before anyone looks at the counts. That keeps the investigation from chasing the first plausible cause.

Shared factor

How to check it

Points to

Substrate or casting lot

Count failed and passed batches per mill heat, coil or casting lot

Steel impurity or oil variation between lots (Biller, 2021)

Release agent or mill oil on the lot

Compare cleaning results on parts from failed and passed lots

Residue the cleaner did not remove (CHT; Biller, 2021)

Degreaser and phosphate bath

Read temperature, concentration, pH and dissolved solids logged at batch time

Pretreatment out of its range (Biller, 2021)

Final rinse

Read rinse dissolved solids or conductivity at batch time

Salt left on the surface; 2.3% dissolved solids in one case (SGS, 2021)

Oven position and line speed

Group batches by rack zone and speed; cure test parts from each zone

Undercure or slow heating (Axalta FAQ; Biller, 2021)

Film thickness

Compare thickness readings per batch

Film beyond nominal values (Axalta FAQ)

Shift and handling

Group batches by shift, crew and glove practice

Handling contamination such as sweat (Li, Zhang and Hu, 2024)

The row that matters most is the substrate or casting lot. If every failed batch shares one lot and the passed batches used others, test that lot before changing the process.

How do you tell a substrate lot problem from a pretreatment bath problem in powder coating?

A substrate lot problem follows the lot across bath conditions, while a pretreatment bath problem follows the clock across lots. To separate them, list each failed batch with both its lot and the time it went through pretreatment. If parts from one lot fail on several days with the bath in range, suspect the lot first. If parts from several lots fail within the same few hours, look at the bath readings from those hours.

The SGS case shows what a bath problem can look like. Rinse water accumulated dissolved solids, and as parts dried the salt concentrated, crystallized and formed a powdery film that the powder coating did not adhere to (SGS, 2021). A cause like that clusters wherever the rinse was most loaded, not by lot.

A lot problem looks different. Biller writes that varying impurity content in steel (nitrogen, silicon, phosphorus, sulfur and carbon) means the pretreatment "may not work as well", and that galvanized steel changes as its zinc slowly oxidizes (Biller, Powder Coated Tough, 2021).

To test a suspect lot, run retained parts from it, parts kept back from that lot, through the line on a day when the bath is confirmed in range. If they still fail while parts from another lot pass beside them, the lot is confirmed as the shared factor.

Can oven position or line speed make only some powder coated batches fail adhesion?

Oven position and line speed can make only some powder coated batches fail adhesion, because both change the time and temperature the metal actually reaches. Axalta lists insufficiently cross-linked powder and a heating rate that is too low as causes of poor adhesion, with the remedies to comply with the curing conditions and increase the heating rate (Axalta Powder Coatings FAQ). Biller adds that undercured powder coatings are prone to chipping (Biller, Powder Coated Tough, 2021).

Batch patterns that point to cure include failures on parts hung at one end of the rack or in one oven zone, failures on heavier parts in a mixed load, and failures on batches run at a faster line speed. The bake conditions on a powder technical data sheet are a time and a temperature (Biller, Powder Coated Tough, 2021). A heavier part in the same oven likely takes longer to reach that temperature, which is why mixed loads can fail unevenly.

To check, group failed and passed batches by rack zone and line speed, then run a cure test on parts from each zone. Handle the test solvent as its safety data sheet (SDS) and your site's solvent-handling and ventilation rules direct. Biller recommends random cure tests with a solvent rub test, following the bake conditions on each powder's technical data sheet (Biller, Powder Coated Tough, 2021). If only the slow-heating positions fail the cure test, the oven is the shared factor, whatever the lot.

How do you confirm the shared factor behind batch adhesion failures?

To confirm the shared factor behind batch adhesion failures, first use counts to test whether failures depend on that factor, then rerun parts with the factor changed. The standard counting tool is a contingency table: batches classified by outcome and by lot, shift or zone, used to decide whether the two classifications act independently. NIST's worked example classifies defects by type and production shift (NIST e-Handbook, section 7.4.5). With few batches, expected counts get small, and the NIST Dataplot reference notes the chi-square approximation may be poor and points to Fisher's exact test (NIST Dataplot, 2007). The same counting logic is what makes spotting repeat customer complaints across customers work.

Rate failed and passed parts with the same test and the same operator. ASTM D3359-23 says the tape test is "insensitive to all but large differences in adhesion" and that results depend largely on operator skill (ASTM D3359-23, 5.2 and 5.5). ISO 2409:2020 states that its cross-cut procedure "is not, however, a means of measuring adhesion" (ISO 2409:2020, scope). So a batch that passed on a different tape or with a different inspector is not a clean comparison.

The last step is a change. Run the suspect lot through a known good bath, or move parts out of the suspect oven zone, and see whether the failures stop.

When should you call a coatings specialist about batch adhesion failures?

Call a coatings specialist about batch adhesion failures when the concentration table shows no clear cluster, when two factors overlap so that one lot always ran on one shift, or when the failures reach customers. Counting alone cannot separate overlapping factors. They need a planned trial that changes one factor at a time. The SGS case is a reminder that the cause may be a combination: its investigators found no single cause and named the phosphate bath and rinsing as the most probable mechanism (SGS, 2021).

Bring the records that let the specialist work the case fast: the failed and passed batch list, the concentration table with counts, bath logs for the failure hours, cure test results by zone, retained parts from failed and passed lots, and the powder and pretreatment technical data sheets. Also bring the adhesion test standard, edition and method you used, since ASTM D3359-23 Method B is primarily intended for coatings under 125 µm in total thickness (ASTM D3359-23, 1.2).

Lubinpla can work a case like this from batch records you add as files. It fixes the target and the comparison axes first, flags where failures cluster in one lot or condition, and returns one cited document for your review.

Frequently asked questions

Can the powder itself cause failures on only some batches?

A powder lot can, if the failed batches share one powder lot and the passed batches used another. Li, Zhang and Hu list powders among the causes of poor adhesion (2024). Add the powder lot number as an axis in the concentration table and count it like any other factor before you blame the substrate or the bath.

How many passed batches should I compare against the failed ones?

Use at least as many passed batches as failed ones, taken from the same weeks, so that every lot, zone and shift in the failed list also has passed examples. With small counts, the NIST Dataplot reference points to Fisher's exact test as an alternative to the chi-square test (NIST Dataplot, 2007).

Does a cross-cut pass at the line rule out adhesion problems in the field?

A cross-cut pass at the line does not rule out field adhesion problems. ASTM D3359-23 only shows that some minimum bond strength was met and does not distinguish higher levels of adhesion (ASTM D3359-23, 1.3 and 5.4). Field failures can also come from handling, packing and environmental exposure (Biller, 2021).

Related posts