Assessment of the adhesion performance of a magnesium oxysulfate mortar compared to other hydraulic binders on substrates of varying roughness. Limitations of the pull-off test

Hydraulic mortars play a critical role in construction, with growing interest in sustainable alternatives to Portland cement. This study evaluates the adhesive performance of a magnesium oxysulfate (MOS) mortar compared to three other hydraulic binders on substrates of varying roughness: concrete, steel, and both sides of ceramic tiles. Pull-off tests following EN 1015–12:2016 revealed that MOS mortar uniquely bonded to the least porous and smoothest substrate—glazed ceramic tile—demonstrating superior adhesion. However, the reliability of the pull-off test was challenged by shrinkage effects, core-drilling torsion, and specimen preparation techniques. Continuous mortar application and vibration improved contact on rough surfaces, while extended curing under high humidity showed limited benefits—particularly for the MOS mortar. A clear correlation was observed between adhesion strength and microcapillary porosity, indicating that achieving a minimal yet well-distributed pore structure enhances the adhesion of mortars to smooth, low-porosity surfaces. The MOS binder consistently exhibited qualitatively better adhesion despite these uncertainties. Results also underscore critical shortcomings in current testing standards, including the lack of control over key variables like drilling method and application technique, which compromise reproducibility. Future research should be made for revised protocols to more accurately assess adhesive performance in alternative hydraulic mortars. • MOS mortar adhered uniquely to smooth glazed ceramic, outperforming all other binders. • Pull-off test reliability was affected by shrinkage, drilling torsion, and preparation. • Correlation was observed between adhesion strength and microcapillary porosity. • Continuous application with vibration improved adhesion on rougher substrates. • Current pull-off standards lack control of key variables, compromising reproducibility.