Novelty in bacteria source production and concrete binders in self-healing cementitious samples

. One of the challenges associated with creating bacterial-concrete systems capable of biomineralizing CaCO 3 to fill cracks is the high pH environment of the hydrated cement paste. In this study two approaches to address this challenge were investigated: (i) the use of calcium sulfoaluminate (CSA) cement, which develops a naturally lower pH, and (ii) the use of non-axenic bacterial cultures, which may facilitate growth of bacterial strains more resilient to harsh alkaline conditions. Axenic B. subtilis and a non-axenic bacterial system from soil were produced and utilized in ordinary portland cement (OPC) and CSA samples. The mechanical properties, water absorption, calcium carbonate precipitation capability, and survivability of bacteria were investigated. The highest B. subtilis and soil bacteria viability was obtained through use of CSA cement and may enable greater later age crack healing potential than mixtures using OPC. Incorporation of axenic bacteria resulted in increased bacteria survivability in the mortar samples when compared to non-axenic bacteria mixes. However, in both cementitious systems, use of B. subtilis and soil bacteria led to similar improvements, suggesting that non-axenic cultures may be used in concrete effectively.


Introduction
Incorporating bacteria into the concrete mixture has been proposed as a method of preventing or mitigating the negative impacts of concrete cracking. Microorganisms capable of precipitating calcium carbonate (a reaction known as "bio-mineralization") have been embedded into the concrete in order to increase the durability of concrete through self-healing of cracks or decrease porosity and permeability [1,2].
One challenge associated with the longevity of bacterial-concrete systems is the high pH environment of the cement paste. Bacteria must be resistant to high pH and capable of survival over long periods of time without nutrients, typically through forming endospores. Bacteria used successfully in concrete include Sporoscarina pasteurii, Bacillus cereus, Bacillus sphaericus, Bacillus alkalinitrilicus, and Bacillus subtilis. Axenic bacteria (a bacteria source free from non-intended strains) have been shown to be effective in filling cracks up to 0.97 mm in width, restoring flexural strength [3][4][5] and increasing durability of concrete through densification of the microstructure and permeability reductions [6,7]. Despite successes in use of axenic strains, growth of axenic microorganisms is not easy, requiring a very clean and sterile environment to prevent contamination. Non-axenic cultures, in which diverse communities of bacteria types have been curated through natural selection mechanisms, are theorized to be more resilient to harsh environments, and not as detrimentally affected * Corresponding author: acarturk.1@osu.edu by contamination. Therefore, the growth of non-axenic bacteria from local environmental materials, such as soil, which is known to harbour many of the strains that have previously been utilized in cementitious systems [8,9], may enable simplified preparation procedures, and result in lower cost and increased resiliency bacteria.
In addition, the use of lower pH cementitious systems may improve the success of bacterial concrete by preserving greater numbers of bacteria within the system over longer times, leading to better healing abilities. But while extensive research on the use of bacteria with OPC is available, limited studies explore the application of biomineralization with other cementitious systems. The use of CSA cement, which produces lower pore solution pH than OPC [10], has been shown to be beneficial for promoting bacterial growth and increasing the calcite biomineralization levels in work evaluating use of CSA as an encapsulation material to protect the bacteria. CSA-was able to preserve the bacterial activity over a long period of time and remedy cracking [11,12]. However, these systems have otherwise not been well investigated despite their ability to improve bacterial longevity and bolster crack healing.
This study evaluated the effect of CSA cement, compared to an OPC mixture, on bacteria viability and biomineralization using a more traditional axenic bacteria and a source of bacteria produced from soil, with the hopes of also promoting the use of sustainable

Materials
Calcium sulfoaluminate (CSA) cement sourced from Buzzi Unicem, ordinary portland cement (OPC) sourced from Lehigh Cement, and a reclaimed fly ash sourced from a power plant in Conesville, Ohio were used for all paste and mortar mixtures. The material oxide compositions of the OPC and CSA cement measured using X-ray fluorescence are shown in Table 1. Deionized water (18 kOhm) was used for all paste mixtures and dechlorinated Columbus tap water was used for all mortar mixtures. Dechlorinated water was obtained by allowing chlorine off-gassing for 24 hours prior to use. ASTM C109 [14] graded Ottawa sand, certified to meet the ASTM C778 [15] graded sand standard, was used as fine aggregate for all mortar cubes in this work. Lightweight aggregate (LWA, from ARCOSA, Indiana), with a fineness modulus of 3.38 was used to protect and incorporate the bacteria into the cementitious mixtures. 24 and 48-hour absorption capacities and the specific gravity were calculated according to ASTM C1761 [16] and determined as 13.5 and 15%.
For this study, soil obtained from a suburban Columbus, Ohio residence, was used to produce a nonaxenic (impure, multi-organism) bacterial system. Soil was selected because of its wide availability as an environmental source for biomineralizing bacteria. For the axenic bacteria sample a Bacillus type (Bacillus subtilis), was selected due to its known ability to biomineralize, and its soil residence [13]. Bacillus subtilis (B. subtilis) strain (type 3A1, strain 168), obtained from Bacillus Genetic Stock Center at the Ohio State University. Nutrient broth (NB) (BD Difco, dehydrated combination of peptone and beef extract) was selected as nutrient medium to grow both bacteria samples.
All cement pastes and mortars were prepared at a w/c by mass of 0.45. In order to prevent compaction and workability issues when using the rapid hardening CSA cement, 1% citric acid (Alfa Aesar, 99%) by cement weight was used as a retardant in the CSA mixes. The curing solution applied to hardened samples to promote bacterial growth and biomineralization contained nutrient broth (8 g/L), urea (Fisher Scientific, 99% purity, 10g/L) and calcium acetate (Fisher Scientific, certified, 25 g/L).

Microorganism growth, microbial community analysis, and immobilization procedures
For soil bacteria samples, 2.5 g of soil was added into 50 mL nutrient solution and incubated for 24 hours at 25 °C with 180 rpm shaking conditions. Then, frozen aliquots were prepared from soil and B. subtilis bacteria using 10% glycerol and were stored until use, then regrown with 180 rpm shaking conditions in a sterilized nutrient medium containing NB (4 g/L). pH of the medium was adjusted to 7 using NaOH unless otherwise specified. After the growth period, the solutions were centrifuged, and bacteria pellets were dried in an oven at 40 ℃ for 2-3 hours to induce sporulation, until reaching surface-dry condition. The dried bacterial spores were stored in a 4 ℃ refrigerator until just prior to immobilization into lightweight aggregate for use in samples. Mixing proportions targeted 10 7 CFU of bacterial spores/mL of mixing water.
To immobilize bacteria into the LWA, the dried B. subtilis and soil bacteria pellets were dispersed in NB solution to which dried LWA was added. The solution was incubated for 24 hours, shaking at 180 rpm. The solution was then filtered and the LWA and both bacteria types were dried in a 40 ℃ oven for 2-3 hours until reaching saturated surface dry condition. For the non-bacterial control samples, 15% tap water by weight of LWA (which corresponds to the LWA absorption capacity) was mixed with the LWA 48 hours before mixing to bring it saturated surface dry condition.
Before making samples, initial cell concentrations in the LWA were determined. Immobilized LWA was suspended in NB for 10 minutes, then sonicated at 40 Hz for 2 minutes to release bound cells into the NB solution, then allowed to settle for 10 minutes. The supernatant was extracted, and 1 mL of sample was serially diluted plated on nutrient broth -agar plates. Plates were incubated at 37 °C for 24 hours and after 24 hours viable cell counts were obtained. To enumerate spores, another 1 mL sample was pasteurized at 70 °C for 15 minutes, and the same plating process was performed. For all bacterial paste and mortar samples, initial vegetative cell concentrations were 10 8 CFU/mL and spore concentrations were 10 7 CFU/mL for B. subtilis and 10 6 CFU/mL for soil bacteria, (less due to greater difficulty in obtaining high concentrations of soil bacteria spores for mortar samples).
To investigate the microorganisms involved in concrete systems to biomineralize, DNA extracted from axenic (B. subtilis), and non-axenic (soil bacteria) cultures was sequenced using 16S rRNA gene nanopore sequencing (MinION, Oxford Nanopore Technologies) to identify the types of bacteria present.

Bacterial mortar cubes
Bacterial and non-bacterial 50 mm cement mortar samples were prepared according to ASTM C305 [17]. The sand to cement ratio was 1:2.75 and LWA replaced 10% of sand in all mixes. After placement, mortars were cured at 22-23 °C and 100% relative humidity (RH) and demolded after 24 hours. 1-day samples were tested immediately following removal from the molds. Samples were kept in the lab environment at room temperature until test day and nutrient broth, urea, and calcium acetate (NBUC) solution spray curing was used to saturate sample surfaces in days 1 -28 days. NBUC solution was used to provide nutrients, and urea and calcium acetate to enable bacterial metabolic activity, and biomineralization activities, respectively.
Compressive strengths of mortars were conducted with triplicate samples from 1 to 90 days according to the ASTM C109 [14]. Viable B. subtilis and soil bacteria cells were enumerated with duplicate tests via Most Probable Number (MPN) analysis on test days. Mortar cubes were crushed to 3-5 mm particles with a sterile ceramic mortar and pestle. 30 g of the crushed sample was then suspended in 45 mL fresh, sterilized (by autoclaving at 121 °C for 45 minutes) NB solution at pH 7 for 10 minutes. The suspension was sonicated at 40 Hz for 2 minutes to release bound cells and allowed to settle for 10 minutes. The supernatant was extracted, and triplicate serial dilutions were prepared with NB in test tubes. The tubes were incubated at 37 ℃ for 48 hours and the growth of cells in the test tubes was observed based on the solution turbidity. The positive number of tubes were entered to the EPA's MPN calculator to determine the bacterial count in the cubes [18].

Self-healing of bacterial mortar beams
To compare the self-healing ability of bacterial and nonbacterial samples, mortar beams (40 mm x 40 mm x 160 mm) were cast with OPC and CSA with no bacteria, B. subtilis, or soil bacteria. Mortars were mixed according to ASTM C305 [17] and used the same proportions of cement, water, sand, and LWA as for the mortar cubes, but 12-mm micro synthetic fibers were added to provide flexural resistance for crack initiation. The mortar bars were cured at 22-23 °C, 100% RH, and demolded after 24 hours. Spray curing with NBUC solution was applied starting from day 1. At 14 days after mixing, the samples were cracked by flexural loading using a displacementcontrolled machine ramping at 0.05 mm/sec. All samples were observed for 8 weeks. Images of the cracks were taken weekly with a digital camera (Panasonic Lumix, DC-FZ20). In order to evaluate the crack healing performance, ultrasonic pulse velocity (UPV) measurements were conducted weekly after imaging samples, according to ASTM C597 [19] using a James Instruments V-meter. Sorptivity was measured for the cracked area on each beam after the 8 weeks according to ASTM C1585 [20]. Samples were dried at 60 °C to constant mass (approximately for 48 hours). A non-absorbent coating (Conheal, CS-1800) was applied to prevent water sorption outside of a 40 x 40 mm area around the cracked zone. Initial sorptivity (water absorption up to 1 day of exposure) and secondary sorptivity (water absorption between 1 and 7 days of exposure) were calculated according to the procedures in ASTM C1585 [20].
After the completion of testing, beam samples were broken into two pieces through the cracked zone. Pieces <1 cm 3 were collected for Scanning Electron Microscope (SEM) imaging to characterize the precipitated material and to identify microstructural indications of the bacteria cells. Samples were coated with gold and imaging was performed using a scanning electron microscope (FEI Apreo) with low vacuum mode at an accelerating voltage of 5 kV and a working distance of 10±2 mm.

Results and Discussion
Growth curves of B. subtilis and soil bacteria are shown in Fig. 1. Both bacteria were grown in NB solution standardized to pH 7. To maximize growth the B. subtilis was grown in temperatures of 37 °C (a standard temperature for bacterial propagation), while the use of 25 °C, ambient temperature condition, was selected for soil in order to simplify the production of bacterial solutions in unconditioned environments without access to incubated environmental chambers. Both strains reached 10 8 -10 9 CFU/mL vegetative cell counts in 48 hours, with soil bacteria resulting in a slightly higher cell count than B. subtilis.

Fig. 2.
Relative abundance of microorganisms at Genus level (a and b represent replicates). Fig. 3 shows the bacterial viability of B. subtilis and soil cells in mortars using OPC and CSA binders. Control samples did not show any bacterial viability, and thus were not included in the graph. For most samples initial cell concentrations decreased from 1 to 3 or 7 days due to the mortality of the surface-attached spores and vegetative cells that were in direct contact with the cement matrix [24], followed by stable numbers of cells in CSA samples or an increase and regrowth of cells followed by a slow decline in the OPC mixtures. Higher bacteria survivability was observed in samples prepared with CSA cement compared to OPC, which may indicate that lower alkalinity in the CSA cement promotes bacterial growth or reduces cell death. Moreover, bacteria cells were able to remain viable at higher concentrations in the CSA samples (10 5 and 10 3 CFU/mL) even after 90 days of hydration. The use of soil bacteria resulted in overall lower cell concentrations than B. subtilis, perhaps suggesting that it had less resilience to the harsh concrete environment, although lower initial cell concentrations in the soil bacteria samples may also have contributed to lower living cell numbers over time.
Spore formation ability for the most abundant genera in soil bacteria may have influenced the viability of bacteria. Bacillus and Citrobacter are known to be spore formers, which could increase survival and subsequent activity in cementitious material [22,23,25]. However, these genera were not the most dominant types in soil bacteria (Fig. 2). The most dominant genus found in the soil bacteria, Aeromonas, may have influenced bacterial viability since it can produce biofilms to aid in toleration of more basic pH conditions ranging up to 9 [22].
The compressive strength results of the control, B. subtilis, and soil mortar cubes are shown in Fig. 4. In OPC samples the presence of the B. subtilis and soil bacteria did not significantly impact compressive strengths. It is possible that cells embedded in mortar as spores may not have regenerated in uncracked samples, despite application of nutrient solution. CSA samples had greater strength than OPC samples for all testing days regardless of their bacteria concentration. Several CSA bacterial samples had statistically significant higher strength than non-bacterial controls (soil bacteria at days 1, 28, and 90), which may be attributed to the higher retained cell count (Fig. 3) leading to increased quantities of biomineralization from the soil bacteria. Despite lower bacteria quantities CSA-S samples generally lead to higher strengths compared to CSA-B samples. This may be explained by differences in the type of bacteria present in the soil bacteria. Acinetobacter, present in the soil bacteria sample, has been shown to consolidate cracks and pores in cement mortars more efficiently than b. subtilis, and may have densified the microstructure and reduced porosity [22]. Images of the cracked beams were taken weekly to evaluate the crack healing of the cured beams. Average crack size for all samples was measured as 0.37 ± 0.04 mm but crack size was not constant throughout the beam cross-section, ranging from 0.25 to 0.60 mm. Fig. 5 shows the cracked OPC samples before (a and c) and after (b and d) crack healing. Partial crack closure was observed in both the B. subtilis and soil bacterial samples ( Fig. 5b and d).
Compared to soil bacteria samples (Fig. 5d), B. subtilis healed a greater portion of the crack (Fig. 5b). However, the closure was limited to crack widths <0.35 mm for both bacteria types. As Acinetobacter and Citrobacter have previously been shown to heal cracks in concrete through biomineralization, the crack-healing abilities of soil bacteria could be due to the presence of these two genera [22,23].
The crack images of CSA samples before and after healing are shown in Fig. 6. Full crack closure was observed in both bacterial CSA samples (Fig. 6b and d).
The precipitate on the CSA samples appeared denser, and was of a darker color, than in the OPC samples which may be attributed to different polymorph of precipitated calcium carbonate. UPV testing was conducted to provide an assessment of crack healing through the depth of the sample (Figs. 5 and 6) and, paired with the crack healing images, helps determine whether healing occurred just at the surface of the sample, or throughout the sample depth. Higher velocity values were obtained in the bacterial OPC samples when compared to the control and aligned with the crack closures observed ( Fig. 7 and Fig. 5). Although the soil bacteria sample had less visual crack healing than the B. subtilis sample, its UPV remained higher than that of its non-bacterial counterpart, suggesting that crack closure may have occurred through the depth of the crack rather than only at the mouth of the crack by penetration of nutrients through the crack depth [26]. CSA samples yielded lower velocities compared to OPC samples. However, all cracked and cured samples were able to recover the majority of the velocity loss that occurred due to the crack formation by the end of the 8week crack healing period, indicating that they returned to good quality. As the non-bacterial sample also regained most of its original velocity, it is unclear whether the improvements resulted from bacterial biomineralization and may instead be linked with continued hydration and curing, or innate self-healing in the CSA samples. A more significant difference in UPV was expected with the use of bacteria in the CSA samples since full crack closure was observed with these samples. Water absorption in the cracked area was measured after the healing period to quantify the degree of crack closure (Fig 8). Sorptivity decreased in all samples over time, but bacterial samples had lower sorptivity than the control, with 57% and 62% decreases in initial water absorption coefficient in B. subtilis and soil bacteria samples, respectively, compared to the nonbacterial sample. The differences were even larger for secondary sorptivity with 73% reductions for B. subtilis and 75% reductions for soil bacteria samples compared to the control. Significantly lower absorption compared to the control also occurred in CSA samples, with 63% and 52% reductions in both initial and secondary sorptivity values in CSA-B and CSA-S samples, respectively. The use of CSA cement led to lower initial and secondary water absorption rates than with OPC with and without bacteria. SEM images taken in the crack-healed areas of the bacterial samples are shown in Fig. 9. OPC-B and OPC-S images show similar calcium carbonate morphologies, with rhombohedral calcite observed in both samples (shown as Cc) and evidence of bacteria also identified (red circles) (Figs. 9a and b). Bacteria have previously been shown to appear as 1-3 µm rod-shaped particles or 0.5-1 µm diameter round holes [27]. The OPC-S sample showed multilayer calcite precipitation compared to the single crystals identified in the OPC-B samples. The spherical crystals present in images (Figs. 9a and b) may indicate the formation of spherical calcite or vaterite, which can result from the use of calcium acetate in the curing solution [28]. Consistent with the visual observations, UPV, and sorptivity, these images demonstrated that both microorganisms were able to precipitate calcite within the crack area.
SEM images of the CSA-B samples showed precipitation of calcite crystals and indications of bacteria in rod-shaped (with a length of 2 µm) and round holes (Figs. 9c and d). However, although round holes and calcite precipitates could be seen in the CSA-S samples in Fig. 9d, no clear indication of rod-shaped bacterial cells was found. In addition to calcite crystals, ettringite formation was observed in CSA-S samples (shown as "AFt"). In CSA samples, the presence of calcite, a phase not typically found in significant proportion in hydrated CSA systems, is evidence for the cause of the previously observed self-healing of cracks and decreased water absorption.

Conclusion
The effects of axenic and non-axenic bacteria cultures with OPC and CSA cement on bacteria survivability, compressive strength, and self-healing of cracks were investigated. The main findings showed that B. subtilis bacteria in mortar samples maintained better viability, with higher concentrations of living cells than the soil bacteria samples. Both bacterial samples showed improvements in UPV, and water absorption compared to control samples, and were able to precipitate calcium carbonate and showed indications of bacteria in SEM images. Thus, both axenic and non-axenic cultures may be used in concrete effectively. The use of B. subtilis was more effective in terms of healing at the crack mouth and through the crack depth, while the use of soil bacteria mostly resulted in internal healing (except in the CSA sample). The presence of Citrobacter and Acinetobacter in soil bacteria led to biomineralization and self-healing of cracks in soil bacteria samples. CSA samples resulted in higher concentrations of viable cells at 90 days, higher strength, and lower water absorption compared to OPC samples. SEM results of both bacteria samples with CSA provided evidence of precipitated calcite. Both bacteria types resulted in full crack closure in CSA cement systems, proving that CSA cement increases bacteria resilience and survivability due to its low system pH and both bacteria types are suitable for the use of CSA cement.