Soil-biodegradable mulch film: Distinguishing between persistent microplastics and fragments released from certified soil-biodegradable products

Soil-biodegradable mulch films offer a sustainable alternative to conventional plastics in agriculture, especially where recollection or recycling are impractical. However, biodegradation of these materials must not result in the formation of persistent microplastics. This study investigates the fragmentation and biodegradation of the certified soil-biodegradable mulch film ecovio® M2351 in standardized laboratory conditions (ISO 17556). The material was incubated in agricultural soil in two different forms: cryomilled fragments and 1 cm² film pieces. Fragment formation was quantified using μ-FTIR microscopy. Biodegradation as well as fragmentation kinetics were modelled by the open-source mechanistic FRAGMENT-MNP model. Results demonstrate that fragmentation is a transient phase within the biodegradation process, with particle counts transitionally peaking, then declining as mineralization progresses. Cryomilled fragments exhibited faster biodegradation and more pronounced fragmentation than larger film pieces. When biodegradation experiments were stopped, more than 90 % of the polymeric carbon was converted into CO₂ and residual fragments showed comparable chemical composition to the original material but showed significantly reduced molar masses, indicating that the biodegradation was still progressing. This is supported by the model, predicting that particle concentrations will decrease to below one particle per gram of soil within 600-700 days for both scenarios. These findings confirm that certified soil-biodegradable polymers like ecovio® M2351 do not form persistent microplastics. The combined experimental and modelling approach provides mechanistic insights into the interplay between fragmentation and mineralization and could be further improved by additional measurements of fragments <25 μm and of the polymer mass in the dissolved phase. In future, field data can support the extrapolation of the model predictions to the real-world.