Spatial organization of biomass controls intrinsic permeability of porous systems

TL;DR

The study shows that biomass growth in porous media reduces permeability, but the spatial arrangement of biomass—driven by bacterial motility—controls the extent of hydraulic decline more than total biomass. Using a microfluidic porous chip under a fixed pressure gradient and time-lapse fluorescence, the authors compare motile and non-motile Pseudomonas putida and link pore-scale biomass distributions to macroscopic permeability $k_t(t)$ through a mechanistic two-pathway model. The model, characterized by a single fitting parameter $k_{bf}$, accurately predicts permeability dynamics from pore-scale data, demonstrating $k_t(t)$ can be derived from biomass distribution patterns and that $k_{bf} \approx 2.5$ Darcy best fits both strains. These findings reveal that biomass spatial organization—shaped by motility—determines clogging efficiency and have practical implications for filtration, water treatment, soil bioremediation, and oil recovery where controlling flow through biomass is essential.

Abstract

Biofilms in porous media critically influence hydraulic properties in environmental and engineered systems. However, a mechanistic understanding of how microbial life controls permeability remains elusive. By combining microfluidics, controlled pressure gradient and time-lapse microscopy, we quantify how motile and non-motile bacteria colonize a porous landscape and alter its resistance to flow. We find that while both strains achieve nearly identical total biomass, they cause drastically different permeability reductions - 78% for motile cells versus 94% for non-motile cells. This divergence stems from motility, which limits biomass spatial accumulation, whereas non-motile cells clog the entire system. We develop a mechanistic model that accurately predicts permeability dynamics from the pore-scale biomass distribution. We conclude that the spatial organization of biomass, not its total amount, is the primary factor controlling permeability.

Figures (6)

• Figure 1: Microfluidic system designed for monitoring intrinsic permeability and controlled flow dynamics. (a) Schematic of the experimental setup used to quantify the intrinsic permeability of the host porous medium. A constant pressure difference ($P_1 - P_2$, with $P_1 > P_2$) is applied across the microfluidic device while the outlet reservoir is continuously weighed to monitor flow rate. The porous medium is designed composing disks (gray) to mimic the solid matrix. Falcon tubes are used as inlet and outlet reservoirs, sealed with microfluidic adapters (black caps). (b) Cross-sectional view of the microfluidic device: A PDMS channel, (blue region) with embedded pillars (gray) of the porous structure, is plasma-bonded to a glass slide. The pore space between pillars is shown in blue. The PDMS, with the exception of the top surface, is coated with gas-impermeable NOA-81 (yellow). (c) UV-C device schematic: A 3D-printed guide with mirrors (gray) reflects UV-C light into specific regions of the chip, preventing unwanted bacterial colonization in the inlet and outlet zones. (d) Schematic illustration of a portion of the porous medium skeleton (red dashed lines), generated from the binarized image to capture connectivity of the pore space. (e) Results of a Maximum Inscribed Circle (MIC) algorithm applied to the pore structure (blue circles represent the largest inscribed circle within each pore). The centers of these circles are aligned with the skeleton (red dashed lines) and positioned where they touch the closest grain. (f) Double logarithmic plot of the probability density function (PDFs) of grain size $P_r$ (gray) and pore (MIC) diameter $P_{D_{\text{mic}}}$ (blue). The dashed lines indicates the average pore size (blue) $\bar{D}_{\text{mic}} = 47 \, \mu m$ and grain diameter (yellow) $\bar{r} = 70 \, \mu m$. To prevent redundancy, any pair of adjacent MICs with an overlap area exceeding 15% of the sum of their areas is processed by randomly removing one of the them.
• Figure 2: Dynamics of the permeability (Methods.G) and biomass for (a) $\Delta$fliC and (b) WT strains. The experimental biomass ($B_{\text{exp}}$) is assessed from the GFP signal emitted by the cells over time. The black dotted line denotes the Logistic Growth Model prediction (Methods.F; denoted as $B_{\text{th}}$).
• Figure 3: Dynamics of P. putida sp. $\Delta$fliC biomass as GFP signal (increasing light intensity going from light to dark) growing between the solid grains (gray disks) observed at various times ($t = 2.5$ h, $t = 4.5$ h, $t = 6.5$ h, $t = 11.5$ h, $t = 17.5$ h, $t = 34.5$ h, and $t = 44.5$ h).
• Figure 4: Dynamics of P. putida sp. WT biomass as GFP signal (increasing light intensity going from light to dark) growing between the solid grains (gray disks) observed at various times ($t = 2.5$ h, $t = 4.5$ h, $t = 6.5$ h, $t = 11.5$ h, $t = 17.5$ h, $t = 34.5$ h, and $t = 44.5$ h).
• Figure 5: Theoretical model and predictions. (a) Schematic view of the permeability model, conceptualizing the system as a series of (virtual) porous media with individual lengths $l_i$ and permeability $k_i$jiao2024intrinsic. (b) biomass density $\rho_{{pi}_i} = \frac{\sum I_{{\mathrm{mic}}_i}}{I_{{\mathrm{max}}} A_{{\mathrm{mic}}_i}}$ (in yellow) evaluated from fluorescence images, where $\sum I_{\mathrm{mic}_i}$ is the sum of pixel intensity within the pore $i$, $I_{\mathrm{max}}$ denotes the maximum intensity contrast, and $A_{\mathrm{mic}_i}$ is the pore area. (c) Schematic of an individual pore with biofilm. Each pore $i$ is represented as two parallel flow systems: (1) a central biofilm-free pipe with diameter $d_{1_i}$ and permeability $k_{\mathrm{d}_{1_i}} = d_{1_i}^2/32$; and (2) an annular biofilm-occupied region with thickness $d_{2_i}/2$ and permeability $k_{\mathrm{bf}}$. The equivalent local permeability is, then, given by their mean, weighted by their own thickness as $k_{e_i} = \frac{d_{1_i} k_{\mathrm{d}_{1_i}} + d_{2_i} k_{\mathrm{bf}}}{d_{1_i} + d_{2_i}}$. (d, e) Comparison of model-based ($k_{t}$; solid) and experimental ($k_{exp}$; dashed) permeability dynamics for $\Delta$fliC (blue) and WT (red) strains. The only fitting parameter is $k_{\mathrm{bf}}$.