Bacteria-assisted crack sealing in low-carbon geopolymer concrete exposed to freeze-thaw cycling

Abstract

Bacteria-based self-healing concrete can seal service cracks through microbially induced carbonate precipitation, but most evidence comes from ordinary Portland cement matrices cured under mild conditions. Alkali-activated and geopolymer concretes offer lower clinker-related emissions, yet their high alkalinity, different pore solution chemistry, and freeze-thaw durability requirements may limit bacterial viability and healing-product stability. We report a fictional bench-scale study of bacteria-assisted crack sealing in a low-calcium fly ash/slag geopolymer concrete exposed to repeated freeze-thaw cycling. Alkali-resistant spores and calcium lactate nutrient were immobilised in lightweight expanded-clay carriers coated with a thin geopolymer shell and added at 0, 4, and 8% by coarse aggregate volume. Prismatic specimens were pre-cracked at 28 days to controlled crack widths of 0.18-0.42 mm, healed under wet-dry cycling for 56 days, and then exposed to 150 and 300 freeze-thaw cycles. The 4% carrier dosage sealed 86 +/- 7% of cracks below 0.30 mm before freeze-thaw exposure and retained 69 +/- 9% visible sealing after 300 cycles; the 8% dosage improved sealing slightly but reduced compressive strength by 11%. Water-permeability reduction reached 78 +/- 6% for 0.20 mm cracks and 43 +/- 8% for 0.40 mm cracks. Micro-CT and Raman spectroscopy indicate that the dominant healing products were calcite and vaterite mixed with alkali-aluminosilicate gel infill, rather than pure calcium carbonate. Dynamic modulus loss after 300 cycles was 8.6% for healed 4% specimens compared with 15.4% for cracked controls. The results suggest that bacteria-assisted healing can improve durability of geopolymer concrete for small service cracks, but performance depends on crack width, carrier dosage, calcium availability, and the timing of freeze-thaw exposure.

Introduction

Self-healing cementitious materials aim to restore transport resistance after service cracking rather than relying only on external repair. Bacteria-based systems are among the most studied approaches because dormant spores can survive incorporation, become active when cracks admit water and oxygen, and promote carbonate precipitation that blocks flow paths [1,2,3]. Immobilisation in carriers such as silica gel, polyurethane, lightweight aggregates, or coated granules is often needed because direct bacterial addition can reduce viability and disturb hydration or strength [4,12]. Reviews of self-healing concrete emphasise that crack width, nutrient availability, curing environment, and durability exposure all control whether laboratory sealing translates into structural benefit [5,6,7].

The mechanism is commonly described as microbially induced calcium carbonate precipitation. Ureolytic and non-ureolytic pathways can raise carbonate concentration or create nucleation sites, and the precipitation product may include calcite, vaterite, aragonite, or mixtures with cement hydrates depending on solution chemistry [8]. In ordinary Portland cement concrete, calcium availability is relatively high. In low-calcium geopolymer binders, calcium supply and pore solution chemistry differ, so the same bacterial strategy cannot be assumed to work without testing.

Geopolymer and other alkali-activated binders are attractive for low-carbon infrastructure because they can use fly ash, slag, calcined clay, or other aluminosilicate precursors while reducing dependence on Portland clinker [13,14,15]. Their environmental benefit depends on precursor sourcing, activator production, curing, and mix design, so low carbon is not automatic [16]. Durability is also binder-specific. Freeze-thaw resistance of geopolymer and alkali-activated concretes has been reported as promising in some systems but sensitive to air voids, pore structure, slag content, fibre addition, and saturation state [17,18,19,20].

This study asks whether a bacterial healing system can improve freeze-thaw durability of a fly ash/slag geopolymer concrete after controlled service cracking. The objective is intentionally modest: we test crack sealing, water-permeability recovery, bacterial viability, healing-product chemistry, and freeze-thaw performance in small specimens. The article does not claim field readiness. It identifies the mix-design and exposure conditions under which bacteria-assisted sealing is plausible in a geopolymer matrix.

Materials and mixture design

The binder consisted of 70% Class F fly ash and 30% ground granulated blast-furnace slag by mass. The alkaline activator was a sodium silicate and sodium hydroxide blend with silicate modulus 1.55 and Na2O equivalent of 6.2% of binder mass. The water-to-binder equivalent ratio, including activator water, was 0.36. Crushed basalt coarse aggregate, river sand, and a polycarboxylate-compatible alkali-activated binder admixture were used to obtain a slump flow of 520 +/- 30 mm. The reference mixture reached 44.8 +/- 2.1 MPa compressive strength at 28 days and 56.7 +/- 2.4 MPa at 90 days.

Bacterial spores were selected from an alkali-tolerant Bacillus culture maintained in the laboratory collection. We avoid naming a proprietary strain because the study is not a microbiological taxonomy paper. Spores were mixed with calcium lactate and yeast extract, then loaded into expanded-clay carriers with 1.5-3.0 mm particle size. The carriers were coated with a thin fly ash/slag geopolymer shell to delay nutrient release during mixing and to improve compatibility with the matrix. This carrier logic follows earlier immobilised-bacteria approaches in Portland cement systems and geopolymer-coated bacterial granules [4,12].

Three carrier dosages were tested: 0%, 4%, and 8% of coarse aggregate volume. A fourth control used sterile nutrient-loaded carriers at 4% to separate bacterial activity from physical reservoir and autogenous sealing effects. Carrier water absorption was measured and compensated in the mixing water. All mixtures were cast into 100 mm cubes, 100 by 100 by 400 mm prisms, and 50 mm thick permeability discs. Specimens were sealed for 24 h, demoulded, cured at 23 deg C and 95% relative humidity for 7 days, and then stored at 60% relative humidity until pre-cracking at 28 days.

The 8% carrier mixture had lower fresh density and higher entrained air than the reference mixture. This was expected because porous carriers replace dense basalt aggregate. The study therefore treats carrier dosage as a durability trade-off, not as a free healing addition.

Cracking, healing, and freeze-thaw protocol

Prisms were pre-cracked in three-point bending using a clip gauge to control crack mouth opening displacement. Target crack-width groups were 0.20, 0.30, and 0.40 mm, with accepted measured surface widths of 0.18-0.23, 0.27-0.33, and 0.36-0.42 mm. After unloading, cracks partially closed by 8-18%, so both maximum and residual crack widths were recorded. Permeability discs were split in Brazilian tension and clamped to maintain the target crack width during healing.

Healing was conducted for 56 days under wet-dry cycling: 24 h immersion in artificial rainwater followed by 48 h at 20 deg C and 65% relative humidity. Artificial rainwater contained 1.0 mM calcium chloride in half the cycles to represent mild external calcium availability; companion specimens received calcium-free water. This design tests whether the geopolymer matrix and nutrient carrier provide sufficient calcium for sealing or whether external calcium is needed.

Freeze-thaw exposure followed a rapid cycling protocol between +4 and -18 deg C in water-saturated conditions. Specimens were tested after 150 and 300 cycles. Relative dynamic modulus was measured every 30 cycles by resonant frequency, mass change was recorded, and crack-sealing retention was imaged after each inspection interval. The protocol is deliberately severe for cracked specimens, because freeze-thaw damage is driven by saturation, pore structure, and crack connectivity.

Bacterial viability was assessed by crushing sacrificial carrier particles extracted from companion paste specimens at 7, 28, 90, and 180 days and plating serial dilutions after heat activation. Colony counts are reported as order-of-magnitude viability indicators rather than exact in-situ cell numbers. The high pH and ionic strength of geopolymer pore solution make direct viability measurement uncertain, so crack-healing performance and precipitate chemistry are interpreted together with the culture data.

Characterisation methods

Crack sealing was quantified by optical image analysis using registered images taken immediately after pre-cracking, after 28 and 56 days of healing, and after freeze-thaw intervals. The sealing ratio is the fraction of initial crack area covered by visible solid material. Because surface sealing can overstate transport recovery, water permeability was measured under 30 kPa hydraulic head before healing, after healing, and after freeze-thaw cycling.

Micro-CT was performed on 20 mm diameter cores drilled across selected cracks. Voxel size was 12 micro m. Segmentation separated open crack void, dense aggregate, carrier particles, and high-density precipitates. This resolution cannot identify gel chemistry, but it can measure whether the crack is blocked through the specimen depth or merely crusted at the surface. SEM-EDS, Raman spectroscopy, and thermogravimetric analysis were used to identify precipitates scraped from crack surfaces and from internal crack bridges.

Mechanical recovery was assessed by reloading healed prisms in bending to 70% of the original peak load and then to failure. The first reload estimates stiffness recovery across the crack; the final load gives residual flexural capacity. Compressive strength and splitting tensile strength were measured on uncracked companion specimens at 28 and 90 days to quantify the cost of carrier addition. Sorptivity and rapid water absorption were measured on discs to link sealing to transport properties.

Statistical comparisons used mixed-effects models with mixture, crack-width group, calcium exposure, and freeze-thaw cycles as fixed effects and batch as a random effect. Reported uncertainties are 95% confidence intervals unless otherwise stated.

Bacterial viability and strength effects

Spores survived mixing and the geopolymer environment when protected inside coated carriers. Viable counts from extracted carriers declined from approximately 10^7 colony-forming units per gram of carrier before mixing to 10^5 at 28 days and 10^4 at 180 days. Directly added unprotected spores, tested in a screening paste series, fell below the detection limit by 28 days. This confirms that the high-alkali geopolymer pore solution is a serious biological stressor and that encapsulation is not optional.

Carrier addition reduced strength at high dosage. At 28 days, the 4% carrier mixture had compressive strength 3.8% below the reference, which is within batch scatter, while the 8% mixture was 11.2% lower. Splitting tensile strength fell by 5.1% and 13.7% for the 4% and 8% mixtures, respectively. Sterile carriers produced similar strength reductions, indicating that porous inclusion and air-void effects dominate the mechanical penalty rather than bacterial activity.

The geopolymer coating on the carriers limited early nutrient release. Isothermal calorimetry on paste showed no delayed setting at 4% dosage and a 9% reduction in 24 h heat release at 8% dosage. SEM of carrier-matrix interfaces showed a dense reaction rim after 28 days, with no obvious gap for the 4% mixture. At 8%, carrier clustering was visible in several polished sections and explains the larger strength penalty.

These results are consistent with bacteria-containing geopolymer and alkali-activated studies showing that bacterial survival and crack healing are possible but strongly dependent on immobilisation and matrix chemistry [9,10,11].

Crack sealing before freeze-thaw exposure

After 56 days of wet-dry healing, reference geopolymer specimens showed limited autogenous sealing: 28 +/- 6% of 0.20 mm crack area, 14 +/- 5% of 0.30 mm cracks, and less than 8% of 0.40 mm cracks. Sterile nutrient carriers increased visible sealing modestly, probably because calcium lactate and carrier fines promoted precipitation and gel deposition. Bacterial carriers produced substantially higher sealing for small and medium cracks.

The 4% bacterial carrier mixture sealed 86 +/- 7% of cracks below 0.30 mm and 52 +/- 9% of 0.36-0.42 mm cracks. The 8% mixture sealed 91 +/- 6% and 61 +/- 10%, respectively, but its lower strength and higher air content make that dosage less attractive. Calcium-containing wet cycles improved sealing by 12-18 percentage points, especially for 0.40 mm cracks. Calcium-free cycles still produced sealing, but the deposits were thinner and less continuous.

Water permeability followed the same trend but with lower recovery than surface images implied. For 0.20 mm cracks, the 4% bacterial mixture reduced permeability by 78 +/- 6%; for 0.30 mm cracks, by 61 +/- 7%; and for 0.40 mm cracks, by 43 +/- 8%. Surface sealing above 80% therefore does not guarantee full through-crack blocking. This agrees with the broader self-healing literature, where visual crack closure, permeability recovery, and mechanical recovery are distinct metrics [2,5,6].

Reload stiffness recovery was modest. The 4% bacterial mixture recovered 34 +/- 5% of initial flexural stiffness for 0.20 mm cracks and 18 +/- 4% for 0.40 mm cracks. Final flexural capacity increased relative to cracked unhealed controls but did not return to the uncracked baseline. The healing system is therefore best described as a durability sealing method, not a structural repair method.

Healing products and microstructure

Raman spectra from crack deposits showed peaks consistent with calcite and vaterite, while SEM-EDS indicated calcium-rich crystals embedded in sodium-aluminosilicate gel. The products were less pure than the calcium carbonate deposits often shown in Portland cement bacterial healing studies [1,2,3]. This is expected because the geopolymer matrix provides dissolved silicate and aluminate species, and the crack solution evolves during wet-dry cycling.

Micro-CT showed three sealing patterns. In small cracks, deposits bridged the full crack depth within 28-56 days. In medium cracks, deposits formed at the crack mouth and around exposed carrier particles, leaving discontinuous internal voids. In large cracks, sealing was patchy and concentrated near carriers intersected by the crack. This explains why permeability reduction fell faster with crack width than visible surface sealing.

The bacterial mixtures had more dense precipitate volume than sterile carrier controls. In calcium-containing wet cycles, precipitate volume inside 0.30 mm cracks was 2.4 times higher for bacterial carriers than for sterile carriers. In calcium-free cycles, the ratio was 1.6. The result suggests that bacterial activity contributes to carbonate precipitation, while the carrier and geopolymer matrix also support abiotic sealing.

No evidence was found for expansive reaction products around carriers. This matters because freeze-thaw durability could be compromised if nutrient release created weak, saturated pockets. The 8% mixture did show more connected entrained voids, but these were linked to carrier clustering rather than to biological precipitation.

Freeze-thaw performance

Uncracked reference geopolymer concrete retained 93.5% relative dynamic modulus after 300 freeze-thaw cycles, consistent with a moderately durable air-entrained alkali-activated mixture. Pre-cracked unhealed specimens deteriorated faster, retaining 84.6% modulus after 300 cycles and showing visible crack widening in 0.40 mm crack groups. Cracks increased local saturation and provided pathways for ice-lens damage.

Healed 4% bacterial specimens retained 91.4% relative dynamic modulus after 300 cycles for 0.20-0.30 mm cracks and 88.1% for 0.40 mm cracks. The sterile carrier control retained 87.2% and 84.9%, respectively. Mass loss after 300 cycles was 0.7% for healed 4% bacterial specimens compared with 1.4% for cracked controls. The improvement is real but not dramatic: freeze-thaw cycling still damages cracked geopolymer concrete, especially when cracks are wider than the healing system can block.

Visible sealing retention after freeze-thaw exposure was 69 +/- 9% for initially well-sealed cracks in the 4% mixture and 74 +/- 8% in the 8% mixture. Some surface deposits spalled during cycles 150-300, but internal micro-CT showed that many crack bridges remained. Water permeability increased after freeze-thaw cycling, yet remained 52 +/- 8% below the pre-healing value for 0.30 mm cracks in the 4% mixture.

The results align with freeze-thaw studies of geopolymer and alkali-activated concretes showing that durability depends on pore structure, saturation, and matrix damage rather than binder label alone [17,18,19,20]. Bacterial sealing improves transport resistance, but it does not replace mixture design for freeze-thaw durability.

Discussion

The most important finding is that bacterial healing can function in a fly ash/slag geopolymer matrix, but only when the bacteria and nutrients are protected. The geopolymer environment is hostile to unprotected spores, and calcium availability is lower than in Portland cement systems. Coated carriers provide both biological protection and a local reservoir of nutrient and calcium, which explains why healing concentrates around carrier-crack intersections.

The best dosage in this study is 4% carrier replacement. It provides substantial crack sealing with limited strength penalty and improved freeze-thaw performance. The 8% dosage gives slightly better sealing but poorer mechanical properties and more carrier clustering. This trade-off is common in self-healing concrete: adding a healing agent creates voids, inclusions, or weak interfaces, so the optimum is rarely the maximum addition rate [4,6,7].

The healing products also show why geopolymer systems should not be interpreted through Portland cement chemistry alone. Crack deposits were mixed carbonate and aluminosilicate products. That mixture may be beneficial for adhesion and pore blocking, but it complicates the simple narrative of bacteria producing calcite that seals cracks. For low-calcium geopolymers, external or carrier-based calcium is likely necessary for reliable carbonate precipitation.

Freeze-thaw exposure changes the success criterion. A crack that looks sealed after wet-dry curing may reopen or lose surface deposits during freezing. The relevant metric is retained transport recovery after cycles, not initial surface closure. In the present tests, cracks below 0.30 mm are the credible target for durable bacteria-assisted sealing. Wider cracks would require larger carriers, repeated nutrient supply, fibres to limit crack opening, or external repair.

Limitations

The study is bench-scale and uses controlled single cracks. Real geopolymer concrete in pavements, bridge decks, or facade panels develops crack networks, variable saturation, salt exposure, carbonation, and mechanical fatigue. The freeze-thaw protocol is severe but does not include de-icing salts or sustained load. Field exposure could change both bacterial activity and precipitate stability.

The bacterial culture is treated as an engineering ingredient rather than as a taxonomic subject. This is appropriate for material performance but limits biological interpretation. Future work should identify strain-level survival, metabolic pathway, and carbonate precipitation kinetics in actual geopolymer pore solution. It should also compare ureolytic and non-ureolytic pathways because ammonia generation from ureolysis may be undesirable in some applications.

The environmental assessment is limited. Geopolymer binders can reduce clinker use, but sodium silicate production, slag allocation, carrier firing, and bacterial cultivation all carry environmental burdens [16]. A fair life-cycle assessment would need regional precursor data, activator supply chains, service-life extension estimates, and repair avoidance. The present study demonstrates a durability mechanism, not a complete sustainability proof.

Finally, crack sealing does not mean structural healing. Flexural stiffness and strength recovery were partial. The technology should be used to reduce ingress and freeze-thaw damage in service cracks, not to restore load-carrying capacity after major cracking.

Conclusion

Bacteria-assisted crack sealing is feasible in a low-calcium fly ash/slag geopolymer concrete when spores and nutrients are immobilised in coated lightweight carriers. A 4% carrier dosage sealed most cracks below 0.30 mm, reduced permeability, and improved freeze-thaw durability of pre-cracked specimens with only a small strength penalty. Larger carrier dosage improved sealing only modestly and reduced mechanical performance.

The healing mechanism differs from ordinary Portland cement systems because the crack fill contains both calcium carbonate polymorphs and aluminosilicate gel. Calcium supply, carrier distribution, and crack width control performance. For cold-region geopolymer infrastructure, bacteria-assisted healing is most plausible as a durability enhancement for small service cracks, paired with air-void control and conventional freeze-thaw-resistant mixture design.

Data and code availability

Raw crack images, permeability measurements, dynamic-modulus records, strength data, bacterial viability counts, micro-CT volumes, SEM-EDS maps, Raman spectra, mix logs, and statistical scripts are included in the supplementary archive. Image processing and statistical analysis used Python 3.11, OpenCV 4.8, NumPy 1.26, SciPy 1.11, pandas 2.1, and statsmodels 0.14.