科研速览 · Science Skim继续刷下去 · Keep skimming →
◆ Lab on a chip2026-08-07

Biocementation shows a three-dimensional geometry dependence in microfluidic pore networks mapped by deep-learning segmentation.

Brooke E Filanoski, Marisa L Bobal, Kendal R Phinney, Darke R Hull, David Erickson

原始摘要(英文原文)· Original abstract
Microbially induced calcium carbonate precipitation (MICP) can transform granular media into cohesive, load-bearing materials, yet pore-scale design rules linking geometry, transport, and mineral growth remain difficult to pin down. Here we present a resin-molded microfluidic pore-network platform that mimics coarse-aggregate architectures relevant to living building materials and allows controlled variation of pore geometry. Devices loaded with Sporosarcina pasteurii were imaged using paired brightfield and transmitted-light polarization microscopy to resolve structure and quantify birefringent CaCO3. Automated image segmentation constrained analysis to the pore space, enabling high-throughput extraction of precipitation coverage, particle density, and feature size across hundreds of fields. Across the tested range, precipitation outcomes depended on channel height and porosity, whereas pillar diameter showed no measurable effect on coverage when porosity and height were fixed. Lower porosity tended to increase particle density but limit feature growth, while higher porosity and taller channels favored larger, coalescing deposits and more spatially continuous films. These patterns are consistent with transport shaping which regime occurs: channel height and porosity together appear to set diffusive replenishment and byproduct dilution, shifting mineralization between a nucleation-dominated regime of many small deposits and a coalescence-dominated regime of fewer, larger features.
读原文 · Read the paper ↗

AI 追问PRO

登录后使用 AI 追问

讨论区

登录后参与讨论

相关论文 · Related

Biocementation shows a three-dimensional geometry dependence in microfluidic pore networks mapped by deep-learning segmentation. — 科研速览 Science Skim