Attachment on mortar surfaces by cyanobacterium Gloeocapsa PCC 73106 and sequestration of CO2 by microbially induced calcium carbonate

Abstract Cyanobacterial carbonate precipitation induced by cells and extracellular polymeric substances (EPS) enhances mortar durability. The percentage of cell/EPS attachment regulates the effectiveness of the mortar restoration. This study investigates the cell coverage on mortar and microbially induced carbonate precipitation. Statistical analysis of results from scanning electron and fluorescence microscopy shows that the cell coverage was higher in the presence of UV‐killed cells than living cells. Cells are preferably attached to cement paste than sand grains, with a difference of one order of magnitude. The energy‐dispersive X‐ray spectroscopy analyses and Raman mapping suggest cyanobacteria used atmospheric CO2 to precipitate carbonates.

acids that would corrode steels (Bansal et al., 2016). As a result, it reduces the regular maintenance and repair of constructions that are highly labor dependent and costly (Toncheva-Panova et al., 2010).
Bio-deposition on mortar is an effective way to enhance its durability since the degradation mostly starts from the cracked surface (Stephen & Gettu, 2020).
A variety of microbes such as ureolytic bacteria, myxobacteria, and cyanobacteria have been applied to building materials (Zhu & Dittrich, 2016). Through ureolysis or ammonification, ureolytic bacteria and myxobacteria increase the pH and produce CO 2 using urea or amino acids to promote the carbonate precipitation process (Gonzalez-Munoz et al., 2010). However, ammonia (NH 3 ), a common air pollutant, is produced in both processes (Mitchell et al., 2013). Cyanobacteria, on the contrary, only require easily prepared nutrient solutions and do not generate side products during carbonate precipitation. Therefore, microbially induced carbonate precipitation using cyanobacteria is less expensive and more sustainable (Toncheva-Panova et al., 2010). Gloeocapsa PCC 73106 (Gloe. PCC 73106) is biofilm-forming bacteria that were found to live on building materials such as limestone (Carlos Rodriguez-Navarro et al., 2012). For all these reasons, the present study investigates the potential of carbonate precipitation by cyanobacteria Gloe. PCC 73106 on building material mortar. Since the pH around mortar surfaces is already high, the conditions are conducive to carbonate precipitation and the pH increase by cyanobacterial photosynthesis is not expected to be an important contributor (Zhu et al., 2017). As such, the efficiency of both living and UV-killed Gloe. PCC 73106 to mediate carbonate precipitation is investigated in this study.
One critical question that deserves further research is the attachment of cells on mortar and their detachment during the MICP application. Bacteria with negative surface charge are easily retained on positively charged or neutral surfaces, thereby favoring adhesion, and ultimately restoration of the substrates by creating a supersaturated microenvironment for heterogeneous crystallization (De Muynck et al., 2010). Several concerns may arise if the attachment of cells on mortar surfaces is low, reversible, or incomplete.
For example, cells would spread to the environment and result in uncontrolled growth. Another concern is that the protection will be less effective due to the low coverage of cells on substrates (Surabhi & Arnepalli, 2020). It is also imperative for cells to attach to the right location, such as on porous areas or in micro-cracks for successful protection.
The process of MICP requires sources of carbon, nutrients, and calcium, but the impact of the type and amount of carbon sources, including organic carbon, bicarbonate, carbonate, or CO 2 on MICP has been understudied. For example, organic carbon is used by heterotrophs to generate bicarbonate or CO 2 for the following carbonate precipitation. Sporosarcina pasteurii, a ureolytic bacteria strain, hydrolyzes urea to produce CO 2 , contributing to the gradual precipitation of calcium carbonate (Abdel-Aleem et al., 2019;Bang et al., 2001). Similarly, myxobacteria produce CO 2 through ammonification to aid carbonate precipitation (C. Rodriguez-Navarro et al., 2003). Inorganic carbon such as CO 2 or HCO 3 − can be used by photosynthetic cyanobacteria to increase pH, thereby creating a microenvironment favoring carbonate precipitation (Badger & Price, 2003;Riding, 2006). Since the types and amount of carbon sources added in the MICP applications impact the environment, the selection of carbon sources is essential. Atmospheric CO 2 can be sequestrated by cyanobacteria in stabilizing the mine tailings (McCutcheon et al., 2016), but no one has investigated the CO 2 sequestration by cyanobacteria on mortar. Previously, we reported that cyanobacterial carbonate precipitation (CCP) utilizing HCO 3 − improved the mortar performance (Zhu et al., 2015(Zhu et al., , 2018. The present study investigates CCP through atmospheric CO 2 on mortar, which has the potential to be a sustainable carbon source for carbonate precipitation. The latter process can also contribute to greenhouse-effect mitigation by decreasing the atmospheric CO 2 . In this study, one of our primary objectives is the comparison of After reaching the stationary growth stage (14 days), cells were collected by centrifuging at 3480 g for 15 min. This step was followed by three cycles of cell wash in a sterilized 0.1 M NaNO 3 solution through centrifuging. After the last wash, cells were resuspended in the deionized water. To ensure the equal amount of the cells used for different conditions (groups 2, 3, and 4), the cell suspension was equally divided into three parts. One-third of the cell suspension was exposed to UV light for 1 hour and was killed, while the rest was kept for the experiments (Figure 1a). The UV-killed cells were reinoculated in BG-11 medium in triplicates, and none of them showed the growth of cells. Just before the experiments, mortar cubes were exposed to UV light for 1 hour to kill the native microorganisms and distinguish the contribution of the cyanobacterial strain used in the carbonate precipitation ( Figure 1a). They were then divided into four groups, of which group 1 was treated abiotically in sterilized deionized water, groups 2 and 3 were immersed in living bacterial suspension (10 9 cells/ml), and group 4 was immersed in UV-killed bacterial suspen- All experiments were conducted in triplicates.

| Microscopy and spectroscopy
The bulk solution initially contained zero cells. During the experi- that is similar to the filter set used by Neu et al. (2004). Mortar treated with living and UV-killed cells were examined on days 0, 1, The spatial resolution of EDS microanalysis depends upon the beam size, nature of the matrix analyzed and the beam energy used. In this study, the spatial resolution is lower than 1 µm. Samples for FE-SEM and ESEM were carbon-coated; therefore, the EDS results did not include carbon. The elemental composites of 80 spots on the mortar surface, cells, EPS, and precipitates were analyzed by EDS. Cells attached to the mortar surface were counted manually. The total area counted was 1.5×10 6 μm 2 , and the total cells counted were between 200 and 500 for each cube.
Raman spectroscopy analyses for samples were carried out under the NTEGRA Spectra system from NT-MDT (Russia) equipped with an upright optical microscope. Raman spectra were acquired using 532 nm excitation wavelength at ca. 4 mW. The power intensity was adjusted with an ND filter. An acquisition period of 0.5 s repeated by 300 was adopted for all of the spectra. Confocal 2D Raman mapping was obtained on a mortar surface treated with living cells. The laser spot was 0.4 μm. Data were obtained across the 0˗2000 cm −1 wavenumber range with a spectral resolution of ~3 cm −1 by a cooled CCD detector.
Photomultiplier (PMT) converts optical radiation acquired from the sample into electrical signals. The laser optical scheme is designed to produce a confocal picture of a sample from laser light.
Confocal 2D pictures can be formed by scanning the sample and simultaneously recording the intensity of the reflected laser light, recorded with PMT.

| Solution chemistry
The pH of the bulk solution was measured on days 0, 1, 7, 14, and 21 with a Mettler Toledo pH meter ( Figure A5). At the same time, a 1 mL bulk solution was filtered through a 0.45 mm cellulose acetate membrane filter to analyze dissolved Ca 2+ concentration by iCE

| ANOVA analysis
We used a two-way mixed effect ANOVA model in STATISTICA to examine the statistical significance of the differences in Cell Coverage among the groups and time intervals studied. In our ANOVA model, the treatment groups were specified as a fixed effect factor and time as a random factor. The p-value was compared against a 5% level of significance to infer whether the differences among the compared means suggest rejection of the null hypothesis (Ho). To test the significance of effects in our mixed effect model, we opted for Satterthwaite's method of denominator synthesis, which finds the linear combinations of sources of random variation that serve as appropriate error terms for testing the significance of the respective effect of interest (Satterthwaite, 1946). We also conducted post hoc comparisons to identify the actual sources of statistical significance of our ANOVA results. We used the Bonferroni test to control for an excessively high Type I Error in our analysis.

| Adsorption of cells on mortar surface
All mortars were half immersed in the cell suspension with an equal concentration of cells (10 9 cells/ml). On cell-treated mortar surfaces, the majority of the cells adhered to the rough and nanometer-sized cement paste (white arrows in Figure 2a, b, and c), while few cells attached to the sand grains (black arrows in Figure 2a and b). The differentiation between sand and cement is described in the Appendices ( Figure A6). On the total counted area on mortar surfaces, the ratio between cement and sand was close to 1:1 (Figure 2d), and the cell density on the cement paste (counted manually) was 2.1 ± 0.3 × 10 3 cells/mm 2 , whereas that on sand grains was one order of magnitude lower ( Figure 2d). One limitation of the present study is that we assumed the cells form a single layer on the mortar surface, which may result in errors. The observation of the preferential attachment was similar to other studies showing that the density of calcified bacterial cells attached to the quartz grains was 1 order of magnitude lower than those observed on calcitic stone (Ferris et al., 1989;Carlos Rodriguez-Navarro et al., 2012). In cement structures, the carbonation of portlandite produces calcite, which allows a higher cell attachment (De Muynck et al., 2008). This process will subsequently contribute to a higher cohesion of the bio-deposition. Electrostatic interactions are one explanation of the bacterium-mineral adhesion on hydrophilic calcite and sand (Ozkan & Berberoglu, 2013). Bacteria with a high negative surface charge are more easily retained by the positively charged surface (Ozkan & Berberoglu, 2013). On calcitic substrates, the attractive forces easily outweigh repulsion forces; therefore, cells achieved a higher attachment rate on calcites than on sand (Carlos Rodriguez-Navarro et al., 2012). Surface roughness has also been reported to be an important parameter influencing the adhesion of cells to the hard substrate (Sekar et al., 2004). This attachment preference is significant in the real application of microbial restoration of mortar, since cement paste tends to crack due to shrinkage, leaving its nanometer-sized surface exposed.
The average initial cell coverage on mortar surfaces, calculated as the area covered by cells divided by the total area of the mortar surface counted, was 8.85 ± 2.18% in group 2, 14.75 ± 3.25% in group 3, and 18.42 ± 2.68% in group 4 analyzed by ImageJ. Cells were all intact, and cell lysis was not observed. ANOVA analysis showed that the difference among the three groups throughout the experiment duration was significant (Table 1)

| Detachment of cells from mortar surface
The density of the cells on the mortar surface in group 3 on day 1   Figure 3c and Table 2). The cell coverage on mortars in group 2 was about 8.85 ± 2.18% on day 0 and kept constant till the end of the experiment ( Figure 3c and Table 2). The difference between the detachment in group 3 and group 2 can be explained by a decrease in ionic strength of the solution in group 3 while that in group 2 remained constant (Surabhi & Arnepalli, 2020). In this study, the concentration of calcium was measured as an indicator of ionic strength.

| Composition analysis on cell surfaces, mortar, and particles
Nanometer-sized and sub-micrometer-sized particles are presented on mortar surfaces in group 1 (Figure 4a and Figure A7) as well as on cells in groups 3 and 4 (white arrows in Figure 4b and d). The EDS on particles in group 1 (black crosses 1 and 2 in Figure 4a) indicates they were a mixture of calcium silicates, calcium magnesium silicates, and calcium aluminates (Table 3). These are typical compositions of cement paste, indicating they were debris from mortar surfaces. A   *The F value for the interactive term between treatment groups (fixed effect) and time (random factor) was 5.89, which is associated with a p-value <0.001. The Bonferroni test identified the homogeneous groups in which no statistically significant differences exist. Shaded blocks of cells represent the four groups that are distinctly different with minimal overlap in terms of the combinations of treatments and time included. The results showed that group 4 on day 0 was significantly different from all other combinations of days and groups. Although group 3 on day 0 did not display a statistically significant difference with group 4 on days 1 to 21, it was significantly different from all other days in group 3. In comparison, all days of group 2 were classified into a single homogeneous cluster.
TA B L E 2 Two-way ANOVA analysis and post hoc comparison, based on the Bonferroni test, of the interactive term between the cell coverage on mortar and the time intervals in which values were recorded

F I G U R E 4 SEM images show (a)
particles on the mortar surface-treated abiotically in group 1; (b) cells (black arrow) attach to the mortar surface in group 3 after 1 day, and few particles (white arrow) adsorbed on cells; (c) EPS (black arrow) around cells in group 4 after 1 day were much thicker than those in group 2, and wrapping the cells; (d) the enlarged area of the white square in (c) showing cells adsorbing particles surfaces in group 1, might be debris from the mortar that was adsorbed by cells. Cells in group 4 increased their EPS thickness as a stress response to UV exposure (Wingender et al., 1999), but were killed due to prolonged UV treatment. As a result, UV-killed cells in group 4 were covered by a thicker EPS layer compared to cells in group 3 (Figure 4c). This might be the reason more cell coverage was observed after UV pretreatment (Figure 3c). EPS are found to aid cells in adhering to substrates (Xu et al., 2020). The UV-killed cells and the particles around them had the highest amount of Cl (black crosses 6 and 7 in Figure 4d, Table 3). The higher amount of Cl might be retained by a higher content of organic matter as a thicker EPS layer was produced. This indicates that EPS and cells have the potential of absorbing chlorine ions, which would otherwise cause corrosion and deterioration of concrete or mortar over time (Achal & Mukherjee, 2015). Similarly, particles on cells could be newly formed precipitates or debris from mortar surfaces. In addition, the slope of the linear regression in Figure 5

TA B L E 3
Composition of particles and cells in group 1 (abiotic condition, sites 1, 2, and 3), group 3 (living cells with calcium, sites 4 and 5), and group 4 (UV-killed cells with calcium, sites 6 and 7) corresponding to the site numbers in Figure 4 F I G U R E 5 Linear regressions between Cl/Ca and Cl/O on cells, EPS, particles, and mortar surfaces in different groups significant (F = 134.06, p = 4.73 × 10 −10 ). In this experiment, additional carbon sources besides atmospheric CO 2 were not provided, so the formation of calcium carbonate confirms that cyanobacteria are capable of using dissolved atmospheric CO 2 to nucleate CaCO 3 (Obst et al., 2009 points in Figure 5. One of the two cases displayed an excessively high amount of oxygen while the other point is associated with an extremely high percentage of carbon, which made the concentration of other elements closer to the detection limit, resulting in a less accurate ratio. Another factor that needs to be taken into consideration is the spatial resolution (<1 μm) of the EDS beam. For particles with a much smaller size, the beam picked up the surrounding signals as well.

| Calcium carbonates distribution around cells
Gloe. PCC 73106 cells with EPS in the middle and right bottom corner of the image square were captured with PMT ( Figure 6a). Both cells and EPS were highly reflective due to autofluorescence. The reflection on the left top corner was much lower, corresponding to the surface of the mortar. Raman mapping was collected on the same area (Figure 6b). The green area in the mapping showed peaks at 213, 1088, 1171, and 1527 cm −1 (the green curve in Figure 6c), indicating a mixture of carbonates and organic matter (Tables 4   and 5). The green area overlapped with Gloeocapsa cells that contain pigments scytonemin and β carotene . Carotenoids show strong peaks in the 1400 -1600 cm −1 region indicating ν 1 (C=C) stretching vibration (Withnall et al., 2003). The correlation between ν 1 (C=C) and the number of double bonds indicates the chain length of the carotenoids (Merlin, 1985;Withnall et al., 2003). In this study, the peak at 1527 cm −1 suggests the presence of β carotene (Kleinteich et al., 2017;de Oliveira et al., 2015). Other signature peaks at 1156 cm −1 and 1007 cm −1 are very weak, due to the interference of autofluorescence. The peak at 1171 cm −1 indicates scytonemin (Edwards et al., 2000), while other peaks at 1590, 1549, 1444, and 1323 cm −1 are missing due to impurities. The peaks at 213 and 1088 cm −1 indicated aragonite.
The blue layer exhibited peaks at 213, 745, and 1088 cm −1 , revealing that vaterites and aragonites resided in it (Table 4). This blue Ca ratio is a governing factor of CaCO 3 polymorphs (Davis et al., 2000). Both cells and EPS were able to nucleate carbonates in the absence of bicarbonates or organic carbon source. The nucleation is essential for the subsequent carbonate precipitation, serving as the template for crystal growth, which in turn could lead to CO 2 sequestration as well.
On UV-killed samples, Raman spectra were carried out on individual points and provided similar results to those for living cells.
This resemblance might be due to the same functional groups residing in UV-killed and living cells that induced calcium carbonate precipitation. This leads to the conclusion that functional groups in EPS and cells were the main driving factor for carbonate precipitation. Similar conclusions about the crucial role of cell surfaces in calcium carbonate nucleation were proposed for other cyanobacterial strains Synechococcus elongatus (Obst et al., 2009 to cementitious grouts (Cuthbert et al., 2013;Phillips et al., 2016).

| CON CLUS IONS
We investigated the attachment of cyanobacteria Gloe. PCC 73106 on mortar surfaces and their carbonate precipitation in the absence of inorganic or organic carbon sources. Based on the following summary, this study provided evidence that the attachment of Gloe. PCC 73106 cells on mortar surfaces was enhanced by the UV pretreatment and carbonate precipitation using dissolved atmospheric CO 2 occurred in the presence of cells and EPS.
1. The initial coverages of living and UV-killed cells on mortar surfaces were 14.75% and 18.42%, respectively. The UV pretreatment on cells helped them to produce more EPS, which in turn enhanced the cell attachment on mortar.
2. Both living and UV-killed cells were preferably attached to cement paste than to sand grains, and the difference was one order of magnitude.
3. The detachment of living and UV-killed cells on the first day was the highest at 26-31% and was close to 0 during the rest of the experiments.
4. In the absence of additional inorganic and organic carbon sources, both living and UV-killed cells were able to promote carbonate precipitation using atmospheric CO 2 that was dissolved in water as HCO 3 − . The carbonate signal on cells or EPS was much higher than on mortar surfaces.
5. By decreasing the environmental impacts including mitigating the greenhouse effect, the restoration of mortar by CCP is a sustainable application.

E TH I C S S TATEM ENT
None required.

ACK N OWLED G M ENTS
We acknowledge the generous help from the Microscopy Center at

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are provided in full in this paper except for the cell coverage summary, which is available in the figshare repository at https://doi.  , 1997;Rumyantsev et al., 2017). The destruction of the autofluorescence compounds was a result of thermal denaturation (Sheng et al., 2011). In this study, the UV-killed cells were reinoculated in BG-11 medium in triplicates, and none of them showed the growth of cells. Neu et al. (2004) detected autofluorescence of cyanobacteria using a filter set (excitation 567 nm, emission 600 +/-30 nm, channel red), and our study adopted a similar wavelength using a TRITC filter set (exciter, 546 nm, emitter, 575-640 nm, and dichroic beam splitter, 560 nm). The empirical experience shows that cells exhibit autofluorescence in both blue and red channels ( Figure A2). However, since the concrete surface showed scattered blue fluorescence ( Figure A1), a red channel was used to inspect cyanobacteria autofluorescence. Both living and UV-killed cells exhibit autofluorescence in the red channel ( Figure A3). The intensity of autofluorescence was preserved for at least 7 days under our experimental conditions ( Figure A4). This intensity slightly decreased after 14 days but did not impact the coverage estimation with the selected threshold.

Solution chemistry
The calcium concentrations in groups 1, 3, and 4 decreased to 91.26 mM, 92.04 mM, and 87.95 mM, respectively, after 6 hours on day 0 ( Figure A5), due to high pH caused by the concrete. The pH of the bulk solution in group 2 (concrete cubes immersed in 15 ml deionized water without calcium) was 10.41. The pH in other groups decreased on day 0 due to the precipitation of calcium carbonates.
The pH decreased to 9.10 in group 1 (the abiotic condition with calcium) and 9.11 in group 3 (fresh cells with calcium). A larger decrease in pH to 8.52 was observed in group 4 (UV-killed cells with calcium), as a result of a larger amount of calcium carbonate precipitation. As the bulk solution absorbed the CO 2 from the atmosphere, the pH of the bulk solution decreased. As the pH decreased, the precipitated calcium carbonate became unstable, and the calcium concentration in the bulk solution increased. By the end of the experiment (day 21), the calcium concentration in group 1 (the abiotic condition) was significantly higher than in groups 3 (with fresh cells) and 4 (with UV-killed cells). Since the calcium was not added in group 2, no calcium precipitation was observed, and the pH was the highest at the beginning. The changes were monitored in the bulk solution at the macroscopic scale, while the microenvironment can be very different around the interface between the mortar and cells. Other cations were below the detection limits.

Sand and cement differentiation
The SEM images show the morphological differences, and the EDS show the composition differences between sand and cement ( Figure A6). The sand (curser 2 in Figure A6b) is composed of Si and O ( Figure A6c), while the cement (curser 7 in Figure A6b  F I G U R E A 7 SEM image and EDS on mortar treated abiotically on day 1.