Every molecule of urea that bacteria split to make calcite leaves behind two of ammonium. That by-product is the standing environmental objection to Ureolysis-Driven Carbonate Precipitation, and a 2019 open-access paper in Scientific Reports by Minyong Lee, Michael G. Gomez, Alexandra C. M. San Pablo, Colin M. Kolbus, Charles M. R. Graddy, Jason T. DeJong and Douglas C. Nelson is one of the few attempts to measure how much of it can be washed back out of cemented ground, and at a scale larger than a bench column. Its answer is twofold: nearly all of the dissolved ammonium came out, but a measurable share stayed stuck to the sand grains.
US patent publication; document page 3. Preserved page.
Why the by-product matters
The authors argue that the fate of ammonium (NH₄⁺) had "remained largely unaddressed" even as Microbially Induced Calcite Precipitation (MICP) was scaled up. Reaction stoichiometry predicts ammonium at about twice the applied urea concentration, and the paper notes that many earlier experiments produced more than 500 mM. For comparison, it cites U.S. EPA recommended limits for aquatic life of 17 mg/L (≈1 mM) for acute and 1.9 mg/L (≈0.1 mM) for chronic exposure, while noting that no drinking-water maximum has been set. High ammonium in surface water, the authors point out, can feed toxic algal blooms.
They set aside the alternative of converting ammonium in place through nitrification and denitrification as probably impractical at field scale, and cite earlier work showing that oxidising ammonium in the ground generates acidity that can damage the cement. That leaves rinsing: push a wash solution through the treated zone, collect what comes out, and deal with it elsewhere.
The experiment
Five hollow steel columns, each 3.7 m long with a 0.2 m by 0.2 m cross-section and holding 0.15 m³ of sand, were treated in one direction to imitate a single flow path between two wells. Three columns held an alluvial "Concrete Sand", one a marine "Delta Sand" and one an alluvial "Covelo Sand"; all were clean, poorly graded sands with fines contents of 1.1% to 1.6%.
Four columns were treated by enriching the ureolytic microbes already present in the sand, tuned to a high rate of urea breakdown (Column 1) or a low one (Columns 2, 4 and 5). Column 3 was instead augmented with cultured Sporosarcina pasteurii. Every column then received nine cementation injections with identical urea and calcium concentrations (250 mM each).
The rinse was a single 525 L pass of 200 mM calcium chloride at an initial pH of about 10.0, pumped at 750 mL/min for roughly 700 minutes — on the order of eight to eleven times each column's pore volume of 48.5 to 63.0 L. The paper says the rinse followed strategies identified in earlier centimetre-scale column experiments, which found that a higher ionic strength helped removal, attributed to calcium ions displacing ammonium from grain surfaces, while pH made no detectable difference. Only the strategy carried over, not the recipe: the earlier study had identified a rinse of pH 9.0 with 500 mM calcium chloride, whereas the meter-scale columns received 200 mM calcium chloride at an initial pH of about 10.0. Columns then sat for 12 hours with no flow before being opened and sampled.
What was measured
Dissolved ammonium fell steeply, then slowly. Water leaving the enriched columns started at 497 to 524 mM (about 342 mM in Column 3). After 200 L it was down to roughly 20 to 32 mM; a further 200 L brought it only to 3.5 to 9.0 mM; at the end of the 525 L it was 0.8 to 2.5 mM. Inside the columns, concentrations after rinsing were below 15, 6, 19, 0.4 and 4 mM at every sampling point in Columns 1 to 5 respectively, with the highest values furthest from the injection end.
Ammonium behaved as if it were sticky. A transport model calibrated on a bromide tracer predicted that an ion with no attraction to the soil would drop below 10 mM after at most 124 L; the real columns needed about 280 L. The total ammonium recovered (27.2, 27.2, 17.6, 38.6 and 28.2 moles) also corresponded to more than the 500 mM maximum that the pore water could have held in four of the five columns, which the authors read as ammonium coming off grain surfaces during the rinse.
It rebounded when the flow stopped. During the 12-hour pause, concentrations at most locations rose by 1 to 5 mM. Cumulative removal slipped from 99.1%, 99.7%, 97.9%, 99.9% and 99.6% to 98.2%, 99.2%, 97.9%, 99.8% and 99.1%.
Some stayed on the sand. Extraction of post-rinse soil samples found sorbed ammonium of 4.8 × 10⁻⁴ to 3.6 × 10⁻³ millimoles per gram of dry soil, or 1.9% to 16.1% of each sand's cation exchange capacity, rising with distance from the injection well and roughly in proportion to the ammonium in the surrounding water. Had all of it re-entered solution, the paper calculates, pore-water concentrations would have risen by 0.05 to 42.8 mM.
The cement appeared to survive. Shear wave velocity, used as a proxy for cementation, changed by no more than ±50 m/s across the rinse, which the authors translate into a calcite change of −0.2% to +0.3% by mass.
Soil and method differences
The marine Delta Sand (Column 4) rinsed cleanest, with every location above 99.2% removal, despite having the largest pore volume and the highest cation exchange capacity (4.32 meq per 100 g). The authors suggest its abundant exchangeable cations kept ammonium from engaging with grain surfaces — an interpretation, not something the experiment isolated. The augmented Column 3 did worst: its furthest sampling point reached only 90.6% removal, which the paper tentatively links to more unoccupied exchange sites and a lower pH during rinsing.
The same columns illustrate the uniformity problem described on Bio-Cementation. In Column 1, shear wave velocities ran from 1107 to 1522 m/s out to 2.35 m but fell to 382 m/s at 3.37 m. In Column 3, a reading of 1197 m/s near the injection well was followed by little detectable cementation beyond 0.31 m, which the authors attribute to bacterial cells being filtered out by the sand close to the well.
What the study does and does not show
It shows that, in clean sands and along a one-directional 3.7 m path, a large calcium-rich rinse removed at least 97.9% of dissolved ammonium without measurably dissolving the Calcium Carbonate (Calcite, CaCO₃) cement. It does not show that the result meets a water-quality standard: most final concentrations were still above the ≈1 mM acute aquatic criterion the paper itself cites, and the authors say acceptable limits will be "governed by site-specific requirements".
Several things lie outside its reach. Observation ended after a 12-hour pause, so the slow release of sorbed ammonium into passing groundwater is flagged as a concern, not measured. Only sands with very little fine material were tested. The cost in water, calcium chloride and pumping of a rinse several times the treated pore volume is acknowledged but not quantified, and the collected effluent, now ammonium-rich itself, still has to be treated somewhere else. The authors declare no competing interests; the work was supported by a U.S. National Science Foundation engineering research centre programme.
Further reading
- Is Biocement Actually Green? A 2021 Life Cycle Assessment of Six Microbial Routes to Calcium Carbonate (Porter et al.) — a 2021 life cycle assessment that counts the ammonium by-product of ureolytic MICP as an eutrophication burden but leaves the cost of treating or recycling it out of its model.
Source notes & attribution
- https://pmc.ncbi.nlm.nih.gov/articles/PMC6892930/