Advancing Crack Repair and Durability in Construction Materials via Enzyme-Induced Carbonate Precipitation (EICP): A Critical Review

Enzyme-induced carbonate precipitation (EICP) is an acellular biomineralisation process in which free plant-derived urease catalyses the hydrolysis of urea to precipitate calcium carbonate (CaCO3), thereby sealing cracks and densifying porous construction materials. EICP has attracted growing interest as a cell-free, shelf-stable, and field-deployable alternative. This critical review synthesises evidence on EICP for the repair of cracks and the enhancement of durability in cementitious and related construction materials. Genuine concrete/mortar crack-repair EICP constitutes the largest experimental category in the field. Across this core, narrow cracks below about 0.3 mm respond to EICP-only treatment with compressive-strength recovery of up to 98.4% and permeability reductions of two to three orders of magnitude, while additive- and fibre-assisted strategies extend the effective repair envelope to approximately 2.0 mm at 45–69% strength restoration. Soybean-derived crude urease is the overwhelmingly preferred source. CaCO3 polymorph outcome is study- and condition-dependent: calcite dominates the literature, vaterite is characteristic of rapid or narrow-crack precipitation, and additive control can shift the calcite fraction from 38% to 81%. Durability evidence is the field’s weakest dimension, with only one study reporting a coupled freeze–thaw and carbonation dataset and none exceeding roughly six months of monitoring. We discuss why EICP has not attracted the same research intensity as MICP, identify the principal barriers to translation, and outline the standardisation, durability-validation, and regulatory steps required for EICP to mature into a mainstream low-carbon repair technology. These figures are aggregated performance ranges drawn from studies with differing specimen geometries, curing regimes, and loading ages, and are therefore reported as an envelope rather than as directly normalised single values. Because the corpus was assembled principally from Scopus and Web of Science, supplemented by manual searching, relevant studies indexed elsewhere cannot be fully excluded.