Materials and Methods

Bacteria and culture medium

This study used the ureolytic bacterium Sporosarcina pasteurii(ATCC 11859, from the Guangdong culture collection centre of China), was used. The culture medium contained yeast extract 20.0 g/L, polypeptone 10.0 g/L, (NH4)2SO4 4 g/L, NaCl 8.0 g/L, and distilled water.S. pasteurii was grown in the medium. The medium was prepared with deionized water and was autoclaved at 121 °C for 20 min. The initial pH of the medium was 7.0 and the urea concentration was 20 g/L. Media and inoculants were placed on a shaker table at 30 °C for 48 h of growth.
Previous research indicated that absorbance and urease activity were often used to represent the growth characteristic and urealysis ability of bacteria (Sun et al., 2018a; Sun et al., 2018b; Stocks-Fischer et al., 1999 ). Changes in cell density were assessed by monitoring the absorbance (optical density (OD)) of the suspension at a wavelength of 600 nm (OD600 ) (Fredrickson et al., 2001 ). The urease activity was measured as the rate of conductivity increase because of the very strong positive linkage between the increase of conductivity and urea hydrolysis (Chin and Kroontje, 1962; Omoregie et al., 2017 ). A bacterial suspension with OD600 of 1.205 and urease activity of 0.712 mM urea hydrolysed/min was used for all subsequent tests.

Loess

Loess from the Loess hilly area of Ningxia province in China was used for all tests. The specific gravity of collapsible loess was 2.721. To obtain the gradation of loess, particle analysis tests were performed. The particle size distribution curve is shown in Fig. 1 . The collapsible loess was mainly medium silty soil and coarse silty soil, with a clay content of about 20%. The plastic limit was 21.48%, the liquid limit was 31.34%, and the plasticity index was 9.86. The loess used here contained calcite (CaCO3), dolomite (CaMg(CO3)2), Sodium chloride (NaCl), magnesium sulfate (MgSO4), calcium sulfate (CaSO4), and other ions.

Experimental setup

The tests to determine loess slope erosion and the application of treatment solutions were all conducted in a cube-shaped container (\(0.18\ m\times 0.24\ m\times 0.04\ m\)), which consisted of polymethyl methacrylate. The slope angle was fixed at 30° and a collection vessel was used to collect the lost soil after rainfall simulation. A schematic view of the setup is shown in Fig. 2 . Over the course of the tests, pictures were taken by a digital camera to observe the surface erosion pattern of the loess slopes at 1, 2, 3, 5, 7, 10, 15, 20, 30, 40, and 50 min. All model slopes were prepared under identical conditions so that the testing results were comparable; therefore, the boundary effect was ignored in the tests. During rainfall simulation, the rainfall intensity was 3 mm/min (180 mm/h) and the pH of the utilized water was about 7.8. During tests, the water was sprayed uniformly by a sprayer head on the loess-slope surface.

Treating slope with different methods

Microbially induced carbonate precipitation

For convenient comparison, the initial dry density of model slopes was identical at 1.52 g/cm3. Then, the mixed bacteria solution and cementation solution (0.75 M of Ca(Ac)2-urea solution) was uniformly sprayed on the artificial loess-slope surfaces. Several researchers used CaCl2 for application of MICP (Mahawish et al., 2018; Sun et al., 2019a ). Calcium carbonate produced from calcium chloride is mainly calcite, but might also be aragonite with calcium acetate, the bonding effect of which is better than that of calcite (Tittelboom et al., 2010 ). The amount of bacterial solution used here was less, therefore, calcium acetate was used to increase the bonding effect. Khan et al., (2015) reported that cell densities of bacterial suspension extremely affect the production rates for precipitation. Apart from the OD600 values, urease activity of bacteria also affects the cementation effect (Sun et al., 2018b ). Therefore, bacterial suspensions with similar OD600 and urease activity were used for treatment.
Six identical model slope samples were prepared and were divided into three groups (A, B, and C) according to different volumes of the mixed solution. Each group had two samples (A1, A2; B1, B2, and C1, C2). For the subsequent application and promotion of practical engineering, the amount of the utilized mixed solution was expressed as the spraying amount per unit area; therefore, the amounts of the mixed solution for A1, B1, and C1 were 2 L/m2, 4 L/m2, and 6 L/m2, respectively. With regard to the control sample in each group, the same amount of distilled water was sprayed instead. Then, the slopes were dried at 25 °C for 6 days before subsequent rainfall simulation tests.

Microbially induced carbonate precipitation and Polyvinyl acetate

To further improve loess slope stability, PVAc (CAS: 9003-20-7) emulsion was added to the cementation solution. PVAc is a synthetic resin, prepared by the polymerization of vinyl acetate (Gordon et al., 2019 ). PVAc emulsion is a type of water-based and environmentally friendly adhesive (Zhang et al., 2018 ), which does not negatively impact the environment. Consequently, PVAc was added for better slope erosion mitigation. The addition of PVAc to cementation solution might affect the bacterial activity or the MICP process. Consequently, the effect of the addition of PVAc on bacterial activity and the production rates of calcium carbonate were investigated. The amounts of PVAc added to the cementation solution were 20 g/L, 40 g/L, and 60 g/L.
According to the calculation method proposed by Whiffin, (2004) , 6 ml of bacterial suspension was mixed with 54 ml urea solution (1 mol/L) and the electrical conductivity was measured every 5 min. The average change in conductivity per minute (\(ms/cm\bullet min\)) was calculated and can be converted to the amount of urease hydrolysis per unit time. Eventually, the rate of hydrolysis of urea per minute (\(m\text{M\ urea\ hydrolysed}\bullet\min^{-1}\)) was obtained by multiplying by the dilution factor of 10, which represented enzyme activity. This method was applied to compare the enzyme activity of the bacterial suspension after addition of PVAc. The total volume of the solution was 60 mL; therefore, the amounts of PVAc were 1.2 g, 2.4 g, and 3.6 g.
To investigate the effect of PVAc addition on the production rates of calcium carbonate, PVAc was added to the cementation solution (a mixture of 0.75 M of urea and 0.75 M of calcium) at various amounts (0 g/L, 20 g/L, 40 g/L, and 60 g/L). In this study, the precipitated calcium carbonate was directly evaluated in transparent polypropylene (PP) tubes at a temperature of 30 °C. 20 mL of bacterial suspension of S. pasteurii with an OD600 of 1.205 was mixed with 20 mL of cementation solution. In a sterile environment, 12 groups of samples were used, which were divided into four groups according to the amount of PVAc used. Every group consisted of three parallel samples, all with an original pH of 7.0. The evaluation criterion was the precipitation rate of calcium carbonate, i.e., the ratio of the actually produced amount of calcium carbonate to the theoretically calculated total amount, which was obtained after a 48-h reaction process. The measuring method of the actually produced amount of calcium carbonate has been described in Sun et al. (2018a) .
During this process, dried samples were weighed and washed with 0.1 mol/L of HCl several times until air bubbles no longer appeared. The lost dry weight caused by acid leaching was evaluated and was assumed to be the weight of the actually precipitated CaCO3. The theoretical total mass of CaCO3 was evaluated by\(C\times V\times M\times t\), where C represents the concentration of calcium ions in the cementation solution in mol/L, V represents the solution volume in L, M represents the molar mass of CaCO3 (100.087 g/mol), and t represents the solidifying time in days.
To compare the curing effect of PVAc addition on loess slopes, various amounts of PVAc were added to cementation solution. Five samples (P1, P2, P3, P4, and P5) were prepared. The amount of mixed solution for samples was 6 L/m2. The sample P3, P4, and P5 were treated with MICP, and various amounts of PVAc (20 g/L, 40 g/L, and 60 g/L) were added. P2 was used as MICP control sample and was only treated with MICP. Sample P1 was a blank control sample, which sprayed the same amount of distilled water instead. Similarly, the slopes were dried at 25 °C for 6 days before subsequent rainfall simulation tests.

Rainfall Simulation Test

During the rainfall simulation test, the soil that was washed out from the model loess-slopes was collected by a collection vessel and the weight of the loess within the collection vessel was measured at 1, 2, 3, 5, 7, 10, 15, 20, 30, 40, and 50 min.

Surface strength test

Repeated experiments were conducted with the same samples. These samples had not experienced rainfall erosion test, but surface strength test. A soil penetrometer (type: WISO-750-1, Juchuang company, China) was used to measure the surface strength of samples. This soil penetrometer could directly obtain the penetrated depth and surface strength of the soil via insertion of a probe into the soil (Ulusay and Erguler, 2012; Miao et al., 2019 ). The depth of measuring points for surface strengths were all about 0.02 m. The inserting direction was perpendicular to the soil surface. Six measuring points were randomly chosen to measure surface strengths.