The polycarboxylate ether superplasticizer — PCE — is what keeps a concrete fluid at a lower water content. The molecule anchors on the cement grain; the side chain holds the next grain off. Slump rises without drowning the mix.
When the sand brings clay, and especially a smectite (sodium montmorillonite is the textbook case), that account breaks. The slump measured in the laboratory is gone at the plant or at the pump. Most of the time the admixture did not “go off.” The sand changed.
What PCE is doing in the mix
On clean cement, part of the PCE adsorbs on the grain and the rest stays in solution, still available. The polyethylene-glycol chain produces steric hindrance: the grains do not close up, the water lubricates, and the mix flows.
The dosage window is narrow. Too little, and the slump never arrives. Too much, and segregation, set retardation or excess air show up. The dosage is fixed with the materials of that mix, not copied from a table written for someone else’s sand.
Where the clay takes the admixture
Fine clay has an enormous surface. With sodium montmorillonite the PCE does not stay on the outside: the side chain enters between the mineral layers. That is intercalation. Polymer that should have dispersed the cement is locked in the clay.
Ng and Plank described the mechanism for MPEG-based polycarboxylates. In another paper the same year, about 1% sodium montmorillonite by mass of cement cut paste spread by as much as 60%, and PCE sorption on the clay was on the order of 400 mg per gram of clay — roughly a hundred times the sorption typical of Portland cement. That is a laboratory result, not a tolerance for the job. It sizes the problem: a little active clay is enough to divert the admixture.
Lei and Plank set that weakness next to the older naphthalene- and melamine-based admixtures. Those polycondensates have no ethylene-oxide chain for the clay to intercalate, and they suffer less. There are also PCEs built with side chains that do not enter the smectite layer. Changing the admixture can be a palliative. Correcting or replacing the sand is the path that does not depend on one molecule.
How it shows up on the job
The laboratory mix closes at a given PCE dosage. At the plant the same dosage misses the slump. The sand pile changed face, supplier or fines content, and nobody reopened the study.
The usual reaction has two shortcuts, and both are expensive. Raising the admixture by eye can bring the slump back and bring segregation, retardation and air with it. Adding water brings the slump back and raises the water–cement ratio: compressive strength and durability drop, while the batch ticket still says the mix was followed.
The pattern that points at the sand is the swing. One day the load passes the slump check, the next day it does not, on the same cement and the same admixture. Weather changes concrete. It does not change it with the signature of an aggregate lot.
Powder is not active clay
ABNT NBR 7211 limits material finer than 75 µm and clay lumps in aggregate for concrete. It is the Brazilian aggregate specification. In role it sits where ASTM C33 sits, and the limits are not the same. The limit matters, and it does not answer the PCE question. Crusher dust and smectite clay pass the same fine sieve and do not take admixture the same way. One fine can be nearly inert. The other is a chemical sponge.
The methylene blue test reads that activity better: the dye is adsorbed by clays, above all the smectites. The classic aggregate method is EN 933-9. In a Brazilian laboratory, confirm which document the procedure follows before comparing a figure from another country. The test does not replace a verification batch. It says the sand deserves one.
At receiving, the checks that haste skips still apply: lot origin, the look of the pile, lumps, the real moisture — so the batch water is not wrong — and a slump test to NBR 16889:2020 when the sand changes. That standard is the slump test (the Abrams cone). ASTM C143 and EN 12350-2 cover the same kind of measurement. A limit written in one of them is not a limit in the others.
What to do with the mix
Lock the sand that was used in the mix design. If the pile changed, the PCE dosage is reopened. Clay is not compensated with water.
If the contamination is one lot, washing or replacing the aggregate is usually cheaper than living with an admixture overdose.
If the region only offers sand with some clay, the study has to use that sand. Slump loss is measured over the haul time. It is not estimated. Where it fits, a more clay-tolerant PCE comes in, agreed with the producer and checked on the same lot — not a brand swap from a brochure.
Record the lot. Without sand traceability, a slump that disappears becomes a site argument instead of a control variable.
PCE is a precise tool. Clay in the sand is a sponge. The two in the same mix, untested, produce a concrete that lies on arrival: it looks like the laboratory concrete, and it is not.
References
- Ng, S.; Plank, J. Interaction mechanisms between Na montmorillonite clay and MPEG-based polycarboxylate superplasticizers. Cement and Concrete Research, v. 42, n. 6, p. 847–854, 2012. DOI: 10.1016/j.cemconres.2012.03.005.
- Ng, S.; Plank, J. Study on the Interaction of Na-montmorillonite Clay with Polycarboxylates. ACI SP-288, p. 1–15, 2012. The spread and sorption figures above are from this paper.
- Lei, L.; Plank, J. A concept for a polycarboxylate superplasticizer possessing enhanced clay tolerance. Cement and Concrete Research, v. 42, n. 10, p. 1299–1306, 2012. DOI: 10.1016/j.cemconres.2012.07.001.
- ABNT. NBR 7211 — Aggregates for concrete. Use the current edition.
- ABNT. NBR 16889:2020 — Concrete — Determination of consistency by the slump test.
- CEN. EN 933-9 — Tests for geometrical properties of aggregates. Assessment of fines. Methylene blue test. Confirm the document the laboratory actually follows.