Cement paste contains superplasticisers such as polycarboxylates to slow down hardening, and plasterboard contains water-repellent additives. Such additives are often poorly biodegradable and can lead to eutrophication when they end up in the environment. They also cost the industry billions.

Over the past three years, materials scientist Annet Baken (see photo above) has been studying how portlandite (calcium hydroxide, a key mineral in cement) and gypsum crystallise when reacting with the commonly used additives peracetic acid (PAA) and STMP. Armed with this knowledge, chemical companies could select environmentally friendly additives more effectively, rather than continuing to test random alternatives.

Baken initially conducted her research in France at the geochemical laboratory of the Earth Sciences Institute (ISTerre). ‘There, I first analysed what the ions actually do’, says Baken. During nucleation, the first phase of mineral growth, ions from a solution come together to form a stable ‘core’, the basic building block of a crystal.

‘According to the classical method, the first agglomerations already form the basic structure of the mineral that emerges from them’, says Baken. All the crystal nucleus has to do is attach ions to it. In the ‘non-classical’ nucleation method, however, no nucleus of ion pairs forms. Instead, ion clusters, dense liquid droplets or oriented aggregates grow even before nucleation takes place. The growth path differs depending on the mineral, formation conditions and additives used. 

Due to the lack of effective nanoscale observation techniques, scientists previously assumed that portlandite and gypsum grew block by block. ‘But the first particles are quite amorphous; they don’t really have a structure’, says Baken. In order to design more sustainable and high-performance additives, chemical companies need a better understanding of how conditions and additives influence non-classical nucleation.

Always a different mechanism

At ISTerre, Baken grew portlandite and gypsum from composite salt solutions that she added to deionised water at a constant rate – always in a variant with and without additives. ‘It’s a matter of continual addition’, says Baken. ‘When the mixture is supersaturated, something precipitates.’

During mixing, the solution was stirred at 500 rpm. At the same time, she monitored the mixture using pH, permeability and conductivity sensors, as well as an ion-selective electrode for Ca²⁺. ‘The sensors and electrodes were important for observing what was happening in the solution on a macro scale and for monitoring the concentrations of free ions’. When the calcium ion signal stabilised, she concluded that the liquid and solid phases were in equilibrium and that crystallisation had begun.

When Baken plotted the sensor data in the form of graphs, she discovered that bound calcium particles were forming in the unsaturated solution for both portlandite and gypsum. These particles precede nucleation, Baken concludes. This is in line with the non-classical nucleation theory.

The additives PAA and STMP slowed down the precipitation of portlandite by binding free calcium ions. ‘That was already known’, says Baken, ‘but the electrode data showed that binding alone could not explain the delayed nucleation.’ It also transpired that each additive exhibited distinct permeability in gypsum and portlandite, despite both systems containing calcium ions. ‘So there is always a different mechanism by which the additive influences nucleation’, Baken discovered.

X-ray films

‘To understand why the additives worked differently with gypsum than with portlandite, I took my setup to the ESRF synchrotron’, says Baken. At the French particle accelerator, the salt mixtures were ‘filmed’ on a nanoscale using X-rays generated by electrons circling at almost the speed of light. This radiation illuminated the solutions during mineral formation. The scattered X-rays from the formed mixtures containing additives were captured by a 2D detector, which measured the intensity and position of the signal. This allowed Baken to determine the distance between atoms and their frequency in the sample.

‘Quite fundamental’

‘We saw that the minerals follow different nucleation paths even without additives’, says Baken. ‘As a result, the ‘target stages’ at which additives have the most effect are different for each mineral.’ Moreover, portlandite forms at high pH, while gypsum forms at around neutral pH. This puts the additives in a different protonation state, influencing their interaction with prenucleation clusters. ‘For example, we suspect that, at high pH (with few protons), STMP forms an amorphous network with Ca²⁺. This may explain the observed two-step nucleation in the portlandite-STMP combination, which was absent in the gypsum-STMP combination.’

‘We don’t yet know exactly how the additives change nucleation’, says Baken. ‘But we do have a fairly good idea.’ The German chemical company BASF is keen to learn more about the practical applications. ‘However, application wasn’t necessarily the focus of my research’, Baken explains. ‘In that sense, it was actually quite fundamental.’