State-of-the-Art Review of Microbial-Induced Calcite Precipitation and Its Sustainability in Engineering Applications

Abstract

Microbial-induced calcite precipitation (MICP) is a promising new technology in the area of Civil Engineering with potential to become a cost-effective, environmentally friendly and sustainable solution to many problems such as ground improvement, liquefaction remediation, enhancing properties of concrete and so forth. This paper reviews the research and developments over the past 25 years since the first reported application of MICP in 1995. Historical developments in the area, the biological processes involved, the behaviour of improved soils, developments in modelling the behaviour of treated soil and the challenges associated are discussed with a focus on the geotechnical aspects of the problem. The paper also presents an assessment of cost and environmental benefits tied with three application scenarios in pavement construction. It is understood for some applications that at this stage, MICP may not be a cost-effective or even environmentally friendly solution; however, following the latest developments, MICP has the potential to become one.

1. Introduction

With increasing population and civil infrastructure demands worldwide, the availability of suitable soil sites for construction continues to decrease and ground improvement is now an integral part of modern development. The most common methods to strengthen soils use either one or a combination of several mechanisms such as compaction, preloading, vibration and chemical grouting. These techniques have been proven to have different degrees of effectiveness in improving soil strength and other properties. However, they come at a cost of consumption of a substantial amount of energy either in their application or production of the grouting materials or both.

Microbial-induced calcite precipitation (MICP) uses naturally occurring bacteria to bind soil particles together through calcium carbonate (CaCO₃) precipitation as shown in Figure 1, thereby increasing the strength. The expected life of MICP-treated soil is more than 50 years, which is compatible with the expected service life of many geotechnical structures [1]. Therefore, biogeochemical processes in MICP offer the potential for solving many engineering issues related to ground improvement. MICP also offers advantages over other common approaches as it uses natural processes and it has the potential of being a comparatively inexpensive technique. The effectiveness of MICP in cementing soil depends on the types of bacteria used, the methods of growing these to the required concentration, the pH and temperature during urea hydrolysis, the concentration and flow rate of cementation solution (e.g., calcium concentration and input flow rate), the soil properties (e.g., the availability of nucleation sites, degree of saturation, soil gradation, particle size, pore throat size) and so forth [2,3,4,5,6]. It is noteworthy to report that there are a number of review articles published in the past focusing on a particular aspect of MICP, e.g., optimizing protocols [7], mitigating liquefaction [8,9], stabilization [7,9,10,11], construction [12] and other aspects [13,14,15,16,17,18,19,20]. A comprehensive review of the current state of knowledge on overall MICP processes, application and sustainability has not yet been undertaken. Therefore, this article presents a comprehensive review of overall MICP processes including the historical development of the technique, reactions and biological sources involved, optimization of the process, the behaviour of MICP-treated soils, modelling of improved soil (including empirical, constitutive and discrete element method models) for geotechnical engineering applications, various application areas, together with a quantitative assessment of associated costs as well as energy and CO₂ footprints (for three hypothetical construction scenarios).

2. Historical Development of MICP

The significant contributions towards the development of MICP techniques are presented in Figure 2 in chronological order. Arguably, MICP was first applied for geotechnical applications in 1995 by Gollapudi et al. [21] for controlling leaching of groundwater contaminants in highly permeable channels by packing a mixture of bacteria (Bacillus pasteurii) with sand. Note Bacillus pasteurii is an outmoded taxonomic name for Sporosarcina pasteurii, which is a gram positive bacterium with the ability to precipitate calcite. Gollapudi et al. [21] observed a maximum of 75% reduction in permeability, which was achieved after 95 h. Stocks-Fischer et al. [22] further investigated the physical and biochemical properties of CaCO₃ precipitation via MICP using B. pasteurii. They observed that the microorganisms used in the MICP process create an alkaline environment and provide nucleation sites favourable for CaCO₃ precipitation. They hypothesized that, at high biomass concentrations, the rate of CaCO₃ precipitation decreases possibly due to accumulation of calcium ions (Ca²⁺) over the surface of the bacteria cells. Later, Fujita et al. [23], while studying the application of MICP as a remediation technique for divalent radionuclides and other contaminants, e.g., strontium⁹⁰ (⁹⁰Sr²⁺) in groundwater, suggested that the rate of CaCO₃ precipitation is directly related to the rate of urea hydrolysis in an alkaline environment. MICP was also used by Hammes et al. [2] to facilitate the removal of soluble calcium (Ca²⁺) from industrial wastewater as an alternative to the use of Na₂CO₃. Their results showed that about 85–90% of Ca²⁺ was removed by using 8.3 mM of urea.

The application of MICP in the improvement of strength and stiffness for soils was first studied in 2002 by Ismail et al. [24]. They found an increase in the unconfined compressive strength (UCS) of MICP-treated soil after 5 h of treatment. Several factors were identified that influenced the UCS including the strength of individual grains, soil grain size and shape, density of soil and the presence of pre-existing calcite crystals.

Despite earlier applications in soil improvement, little was known about the performance of urease produced by bacterial cells during the MICP process. Whiffin [25] was the first to develop a model called the “Whiffin’s conductivity method” to predict the changes in urea hydrolysis during CaCO₃ precipitation by measuring the electrical conductivity of the MICP solution. The shear strength and stiffness of MICP-treated soil were later evaluated by DeJong et al. [3] using a series of consolidated isotropically undrained compression (CIUC) triaxial tests. Their results showed that MICP-treated soil specimens exhibited a non-collapse/strain-softening behaviour with a higher initial shear stiffness and ultimate shear capacity compared to untreated loose specimens.

In MICP, biocementation (generation of particle bonding) and bio-clogging (production of pore-filling materials) have notably been explored [26]. Over the last decade, the MICP technique has been applied to enhance the engineering properties (strength and stiffness) of different soils [27,28], reduce the liquefaction potential of loose sand [29,30], heal cracks [31,32,33,34], control erosion [35,36] and reduce soil permeability via bio-clogging [37].

Other studies have attempted to investigate the different factors affecting the MICP treatment process [29,38,39,40]. Despite numerous laboratory-scale studies, there have been a lot of uncertainties regarding the overall effectiveness of the technique in large scale applications. Van Paassen et al. [41] reported a large-scale MICP experiment with a 100 m³ sand specimen. Their results showed that MICP treatment significantly improved the strength and stiffness of granular soils in large-scale applications. However, the uniform distribution of cementation, especially in large-scale experiments, still proves to be a significant challenge [42,43,44,45].

Constitutive models are being developed and implemented for the prediction of the soil-cement and stress-strain behaviour of MICP-treated soils [29,38,39,40,46]. Discrete element method (DEM) simulations of MICP-cemented sand have been used to evaluate the micro-scale behaviour of the sand-CaCO₃ bond [47,48,49,50].

In recent years, research efforts have grown significantly in terms of investigating various aspects of the MICP process. Figure 3a presents a snapshot of the number of publications starting from 1995 to date and shows exponential growth in the number of publications reflecting the growing interest and increased research intensity in the field. While the majority of the research (about 70%) has been conducted in the US and China (Figure 3b), researchers from Iran, Australia, Canada, Brazil, India, middle eastern countries, Russia, Europe, south-east Asia and Africa have also stepped in to contribute to the knowledge base.

3. Biogeochemical Processes

The effectiveness of the MICP process depends on several physical and biogeochemical factors such as the bacterial genus used, the reaction environment and the approaches used. These factors are carefully reviewed and discussed in the subsequent sections below.

3.1. Sources of Bacteria Used in MICP

The most commonly used urease bacteria in past studies are S. pasteurii, Spoloactobacilus, Clostridium and Desulfotomaculum. Of these, S. pasteurii, an alkalophilic non-pathogenic bacterium with highly active urease enzymes, has been found to be one of the most effective and efficient and has been widely used [5,28,41,42,43] even though contradictory evidence can be found in the literature [52,53].

Venda Oliveira et al. [54] assessed the performances of two types of bacteria (S. pasteurii and I. insulisalsae) on the strength and stiffness of sandy soil using UCS and splitting tensile strength (STS) tests. The results from their study show that the bacterium I. insulisalsae was more efficient than S. pasteurii in strengthening the soil. However, the optimum environmental conditions for the growth of each bacterium was not considered in their study, which could possibly have affected their findings.

Screening of bacteria for self-healing of concrete cracks was investigated by Zhang et al. [55]. In their study, the calcium precipitation activity (CPA) of bacterial strains was evaluated using genomic 16S rDNA sequencing and phylogenetic tree analysis. The results from their study show that the Bacillus species (which was designated as the H4 strain) showed the highest CPA of 94.8%. By assessing the influence of other factors on the performance of the H4 strain, they observed that lactate and nitrate were the best carbon and nitrogen sources for the H4 bacterial strain, with optimal concentrations of approximately 25 and 18 mM, respectively, at an optimum pH range of 9.5–11.0.

Liu et al. [56] assessed the interaction between S. pasteurii and an indigenous bacterium during MICP soil treatment. In their study, two soil samples, i.e., natural soil extract (with indigenous bacteria present) and artificial soil extract (with no microorganisms) were treated with an MICP treatment solution containing S. pasteurii. The results from their study show that S. pasteurii exhibited a significantly higher growth rate with maximum CaCO₃ precipitation in the artificial soil extract compared to the natural soil extract. They postulated that the deceleration of the MICP process in the natural soil could be due to the competition for nutrients between S. pasteurii and indigenous bacteria.

Nayanthara et al. [57] assessed the performances of isolated local ureolytic bacterium for the MICP process and observed that among other sources of bacteria such as Bacillus, Halomonas, Oceanobacillus and Sulfitobacter pasteurii, S. pasteurii had the maximum growth, urease activity and performance at 25 °C under moderately alkaline conditions. Figure 4 summarises their observations.

3.2. Biogeochemical Mechanisms in MICP

Biological processes involved in MICP can be broadly categorised in two groups, i.e., biostimulation and bioaugmentation. In biostimulation, indigenous microbes of the soil are stimulated with external nutrient medium, thereby inducing growth. Bioaugmentation occurs when external microbes are either injected or percolated into the soil along with nutrient medium to help their growth.

Gomez et al. [44] assessed the performance of biostimulated treatment solutions to stimulate native ureolytic bacteria in a variety of soils using soil columns (10.2 cm height × 5.1 cm diameter). The results from their study showed strength improvement and a significant reduction in the permeability of the treated soils. A maximum UCS of 5.3 MPa with an average calcite content of 13.2% was achieved in their study. A large-scale biostimulation experiment was conducted by Gomez et al. [58] using two identical 1.7 m diameter and 0.3 m thick soil layers in a tank. Cone penetration tests were conducted after treatment and they observed that biostimulation may provide good cementation improvement at that scale. Chen and Achal [59] demonstrated that a biostimulation process can be used to precipitate calcite inside a soil matrix and at the same time can effectively remediate Cu contamination by precipitating carbonates of Cu. Feng and Achal [60] used a small quantity of cement and biostimulation to improve the strength of rammed earth materials.

Despite the success of implementing the biostimulation approach in some studies, this approach has its drawbacks, such as homogeneity of treatment and a longer time requirement for the stimulation and growth of microbes.

Even though biostimulation is considered to be more suitable than bioaugmentation due to many reasons such as the external addition of microbes, increased cost, non-uniform distribution over depth and the probability of a decrease in microbe numbers if the conditions are not conducive, MICP techniques using bioaugmentation approaches have been studied extensively at the laboratory scale [29,61,62,63].

The MICP process can be achieved through many processes including urea hydrolysis, denitrification, iron reduction, sulphate reduction and other pathways [43,52,53]. Among these, urea hydrolysis has been most widely used due to its high CaCO₃ precipitation efficiency. On the other hand, denitrification, iron reduction and sulphate reduction have been paid less attention due to the low solubility of oxidizing substrates and as a result, require a large amount of substrate solution to obtain sufficient precipitation [64]. These processes are discussed below and are also summarised in

Table 1. A summary of the literature showing the relative advantages and limitations of major biochemical mechanisms involved in the MICP process.

Microbial Pathway	Relative Advantage	Limitation
Urea hydrolysis	(1) high chemical conversion efficiency up to 90% [65] and CaCO3 precipitation [71], (2) straightforward and easy process to control [66,72].	(1) releases undesirable ${\mathrm{NH}}_4^{+}$ [53], (2) requires aerobic conditions for bacteria growth [52].
Denitrification	(1) generates a large amount of N2 gas for liquefaction mitigation via desaturation of pore water [8,73,74], (2) oxygen independent (anaerobic) [53], (3) requires a lower concentration of substrate [53].	(1) has a slower rate of CaCO3 precipitation [75], (2) generates a high initial ${\mathrm{NO}}_3^{-}$ concentration which inhibits bacteria growth [53].
Sulfate reduction	(1) oxygen independent (anaerobic) [53]	(1) releases H2S which is odorous and highly toxic [53] (2) requires a large quantity of substrate [53].

.

3.2.1. Urea Hydrolysis

In urea hydrolysis, the major ingredients involved are urease enzyme, urea $\left(\mathrm{CO}{\left({\mathrm{NH}}_2\right)}_2\right)$ and calcium chloride $\left({\mathrm{CaCl}}_2\right)$. In general, ${\mathrm{CaCO}}_3$ precipitation via urea hydrolysis can be divided into three main stages: (1) hydrolysis of urea into ${\mathrm{NH}}_4^{+}$ and carbonate ions $\left({\mathrm{CO}}_3^{2-}\right)$; (2) dissociation of ${\mathrm{CaCl}}_2$ into calcium ions $\left({\mathrm{Ca}}^{2+}\right)$; and (3) ${\mathrm{CaCO}}_3$ precipitation. The chemical reactions involved are presented in Equations (1) to (3).

$$\mathrm{CO}{\left({\mathrm{NH}}_2\right)}_2+2{\mathrm{H}}_2\mathrm{O}\stackrel{ureahydrolysis}{\to }2{\mathrm{NH}}_4^{+}+{\mathrm{CO}}_3^{2-}$$

(1)

$${\mathrm{CaCl}}_2\stackrel{{\mathrm{H}}_2\mathrm{O}}{\to }{\mathrm{Ca}}^{2+}+2{\mathrm{Cl}}^{-}$$

(2)

$${\mathrm{Ca}}^{2+}+{\mathrm{CO}}_3^{2-}\stackrel{precipitation}{\to }{\mathrm{CaCO}}_3$$

(3)

Compared to other microbial pathways, ${\mathrm{CaCO}}_3$ precipitation via urea hydrolysis provides many advantages including a high chemical conversion efficiency up to 90% [65] and ease of control of the process [66]. A disadvantage of this process can be the release of undesirable ${\mathrm{NH}}_4^{+}$, which is a major cause of water pollution and potent oxygen demand [53].

3.2.2. Denitrification

Many studies have investigated the potential of soil improvement using the denitrification process [53,67,68]. In addition to improving soil strength via CaCO₃ precipitation, denitrification processes have often been used to desaturate soils (i.e., through the release of insoluble N₂) for liquefaction mitigation [69,70]. MICP by denitrification refers to the dissimilatory reduction of nitrate $\left({\mathrm{NO}}_3^{-}\right)$ to generate nitrogen gas (N₂), inorganic carbon (CO₂), and alkalinity $\left({\mathrm{OH}}^{-}\right)$ using denitrifying bacteria (e.g., Pseudomonas denitrificans) under anaerobic conditions. The generation of CO₂ raises the carbonate content of the solution, while the consumption of ${\mathrm{NO}}_3^{-}$ increases the pH. The production of alkalinity favours precipitation of CaCO₃ in the presence of ${\mathrm{Ca}}^{2+}$. The chemical reactions are presented in Equations (4) and (5).

$${\mathrm{NO}}_3^{-}+1.25{\mathrm{CH}}_2\mathrm{O}\stackrel{denitrification}{\to }0.5{\mathrm{N}}_2+1.25{\mathrm{CO}}_2+0.75{\mathrm{H}}_2\mathrm{O}+{\mathrm{OH}}^{-}$$

(4)

$${\mathrm{Ca}}^{2+}+{\mathrm{CO}}_2+2{\mathrm{OH}}^{-}\stackrel{precipitation}{\to }{\mathrm{CaCO}}_3+{\mathrm{H}}_2\mathrm{O}$$

(5)

Compared to urea hydrolysis, denitrification requires a lower concentration of the substrate to induce CaCO₃ precipitation; however, the rate of CaCO₃ precipitation is considerably lower, possibly due to the accumulation of intermediate ${\mathrm{NO}}_3^{-}$. A high initial ${\mathrm{NO}}_3^{-}$ concentration may inhibit bacterial growth by altering the pH across the cell membrane [53,67,68].

3.2.3. Sulphate Reduction

In MICP by sulphate reduction (similar to iron reduction), sulphate-reducing bacteria (e.g., Desulfovibrio and Desulfotomaculum) oxidize sulphates under anaerobic conditions to produce hydrogen sulphide (H₂S), CO₂ and increased alkalinity. pH due to increased alkalinity favours CaCO₃ precipitation. The sulphate reduction process is presented in Equation (6).

$${\mathrm{SO}}_4^{2-}+2{\mathrm{CH}}_2\mathrm{O}\stackrel{sulphatereduction}{\to }{\mathrm{H}}_2\mathrm{S}+2{\mathrm{CO}}_2+2{\mathrm{OH}}^{-}$$

(6)

The release of CO₂ in the presence of ${\mathrm{Ca}}^{2+}$ fosters CaCO₃ precipitation (see Equation (5)). Anaerobic oxidation via sulphate reduction requires a large substrate quantity due to low solubility. The mechanism also results in the production of H₂S, which is an odorous and highly toxic gas even at low concentrations. Table 1 shows a summary of the relative advantages and limitations of these biochemical mechanisms.

4. Engineering Properties of Treated Soil

The calcite precipitation modifies or enhances the strength of the sample and different studies have investigated this influence through various laboratory tests namely, UCS tests [43,76,77,78,79,80,81], STS tests [54,79,80,82,83,84,85], direct simple shear tests [30], triaxial tests [28,79,86,87,88,89,90,91], cyclic triaxial tests [92] and cone penetration tests [62]. Some of the observations are summarised below.

4.1. Unconfined Compressive Strength

Several studies conducted UCS tests on many different types of MICP-treated soil to quantify the improvement of strength. A summary of the studies is presented in

Table 2. Summary of studies using UCS tests to evaluate strength improvement in MICP-treated soils.

Sand Type	MICP Process			D10 (mm)	D50 (mm)	CuS	CcS	Testing Method
	BTM	DoS (%)	TM					Size H × D (mm)	Ave. UCS (MPa)	Ave. CC (%)	k (m/s × 10−5)
Plaster sand [44]	BS	100	Gf	0.18	-	7	0.7	102 × 51	3.25	6.6	0.1–100
Concrete sand [44]	BS	100	Gf	0.18	-	10.1	0.6	102 × 51	3.64	9.9	0.1–100
Cushion sand [44]	BS	100	Gf	0.09	-	3.6	1.3	102 × 51	3.95	9.8	0.1–100
Russian River [44]	BS	100	Gf	0.24	-	8.7	0.7	102 × 51	1.22	6.5	0.1–100
Folsom Lake [44]	BS	100	Gf	0.24	-	6.9	0.7	102 × 51	1.07	7.4	0.1–100
Napa Bay [44]	BS	100	Gf	0.18	-	1.6	0.8	102 × 51	5.34	13.2	0.1–100
Cemex Fill [44]	BS	100	Gf	0.38	-	8.4	1.2	102 × 51	2.67	6.0	0.1–100
Granite sand [44]	BS	100	Gf	0.22	-	7.7	0.6	102 × 51	2.73	7.5	0.1–100
Coarse sand [78]	BA	30–100	I	0.54	0.70	1.27	0.1	110 × 55	0.1–2.4	4–14	8.0–42
Silica sand [77]	BA	100	I	0.25	-	-	-	1000 × 45	9–20	-	-
Pure silica sand [43]	BA	Unsat	SP	0.23	-	-	-	1000 × 45	19.61	-	-
Silica sand [81]	BA	Unsat	SP	0.35	-	-	-	2000 × 55	0.065	-	-
Ottawa silica [76]	BA	100	Sb	0.19	0.30	1.8	1.1	102 × 51	1.9–15	1.8–14	-
Fine sand [93]	BA	100	I	0.10	0.19	2.1	0.9	100 × 50	0.6–2.5	4–9	-
Medium sand [93]	BA	100	I	0.26	0.39	1.6	0.9	100 × 50	0.5–12	3–11	-
Coarse sand [94]	BA	100	I	1.45	1.60	1.4	1.0	102 × 51	0.5–15	2–23	-
Coarse sand [95]	BA	100	SP	0.61	0.72	1.2	1.0	100 × 50	0.2–2.3	3–16	0.1–66
Ottawa 20–30 [79]	BA	100	I	-	0.72	1.2	1.0	75 × 25	2.5	6.5	-
Ottawa 50–70 [79]	BA	100	I	-	0.22	1.4	0.9	75 × 25	3.04	10.8	-
Navada [79]	BA	100	I	-	0.12	1.7	1.2	75 × 25	2.6	13.9	-
Ottawa 20–30 [84]	BA	100	I	0.65	-	1.2	1.0	100 × 50	0.2–1.8	8–12	2.0–6.0
Beach sand [96]	BA	100	I	0.50	0.70	-	-	100 × 50	0.4–10	10–29	-
Ottawa 20–30 [97]	BA	100	I	0.65	-	1.2	1.0	100 × 50	0.8–1.1	6.4–7.6	4.0–6.5
Mizunami sand [98]	BS	100	Gf	-	1.50	-	-	60 × 30	0.9–10	10–32	-
Silica sand [99]	BA	100	I	0.5–0.1	-	1.2–6.3	0.9–1.1	90 × 45	0.1–2.0	1.4–10	-
Silica sand [100]	BA	100	I	-	-	-	-	180 × 50	0–0.31	1.0–4.1	4.0–10
Ottawa sand [101]	BA	100	Gf	-	0.42	-	-	100 × 50	0.2–0.5	5.2–7.7	0.1–0.6
14 soil types [102]	BA	100	I	0.07–0.38	-	2.3–10.1	0.6–1.4	-	0.1–5.4	2.6–14	-
Standard sand [103]	BA	100	Sb	-	0.42	-	-	-	0.03–2	1.4–10	-
Residual soil [104]	BA	100	I	-	-	-	-	170 × 50	0–0.2	0.6–2.8	-
Ottawa sand [76]	BA	100	Sb	0.19	0.30	-	-	102 × 51	0–2.3	1.8–15	-
Itterbeck fine [41]	BA	100	I	-	0.17	1.64	-	-	0.7–13	12–28	-
Silica sand [38]	BA	100	I	-	0.17	-	-	100/250 × 35/100	0–3.0	2.6–10	0–13

BTM—Bio-treatment approach, BA—Bioaugmentation; BS—Biostimulation; DoS—Degree of Saturation; Dx—Particle size at ‘x’% finner; Cus—Uniformity coefficient of sand particles; Ccs—Coefficient of curvature of sand particles; TM—Treatment method; Gf—Gravity fed; I—Injection; Sb—Submerged; SP—Surface Percolation; k—Soil permeability.

. A brief overview of the literature is presented below.

Al-Thawadi [77] applied MICP on a 1 m long column to investigate strength enhancement and achieved strengths of up to 30 MPa. Cheng and Cord-Ruwisch [43] applied MICP through surface percolation for in-situ placement of bacteria and a cementation solution to improve the mechanical properties of soil.

Gomez et al. [44] applied MICP to a variety of soils and found significant improvement in engineering properties. Other studies have also investigated the UCS of different types of soil and also analysed the CaCO₃ content [38,41,76,78,81,84,93,95,96,97,98,99,100,101,102,103,104]. In some studies, the influence of CaCO₃ precipitation via MICP treatment on soil permeability has been assessed [38,78,80,84,95,97,100,101].

Here, experimental data from the literature on geotechnical properties such as UCS and permeability are compiled and critically analysed to determine any possible correlations. Figure 5a plots UCS against Cc for the compiled data, where Cc is the calcium carbonate content as a percentage (%). It is worth noting that the UCS of MICP-treated soils ranges from 0.03 to 14.5 MPa, while Cc ranges from 0.6 to 32.0%. It can be seen that UCS exponentially increases with the increase of Cc (Figure 5a), which is consistent with previous results reported by van Paassen et al. [41] and Cheng et al. [78]. However, the trend in Figure 5a shows high variability at Cc > 5%, possibly due to the variation in soil particle size and shape. Other factors such as the concentration of chemical constituents (urea, CaCl₂, etc.), urease activity of bacteria and treatment method directly influence the amount of precipitated CaCO₃, which is considered in this study in terms of Cc.

Figure 5b shows a plot of permeability against Cc for data compiled from the literature. The permeability of MICP-treated soils ranges from 0.01 to 66 (m/s × 10⁻⁵) with Cc ranging from 0.002 to 14.8%. The permeability of MICP-treated soil decreases exponentially with the increase of Cc (Figure 5b), possibly due to the reduction in void spaces by the precipitated CaCO₃. Again, the trend of the data as shown in Figure 5b displays some degree of variation (less than that observed for UCS), possibly due to the influential factors discussed earlier.

4.2. Indirect (Splitting) Tensile Strength

Similar to UCS, a series of STS tests on MICP-treated soils have been reported in recent literature [54,79,80,82,83,84,85]. In these studies, STS values have been assessed and compared with other parameters such as the CaCO₃ content [79,80,82], different soil types [54,82], and reinforced fibre content [83,84,85]. A summary of the studies can be found in

Table 3. A summary of the literature reporting STS test results on MICP-treated soils.

Sand Type	MICP Process		D10 (mm)	D60 (mm)	CuS	CcS	Testing Method
	EM	TM					Size H × D (mm)	Method	Ave. STS (MPa)	Ave. CC (%)
Calcarous sand [79]	-	I	0.15	0.35	2.33	1.01	20 × 40	STS	0.04–0.36	-
Ottawa 20–30 [82]	-	I	0.65	0.80	1.17	1.02	75 × 25	STS	0.51	4.83
Ottawa 50–70 [82]	-	I	0.18	0.25	1.40	0.90	75 × 25	STS	0.52	8.55
Nevada [82]	-	I	0.08	0.13	1.70	1.24	75 × 25	STS	0.39	14
Sand [54]	Clay (0%)	MC	0.53	1.35	2.55	0.99	20 × 70	STS	0.01	-
Sand + Kaolin [54]	Clay (28%)	MC	-	-	-	-	20 × 70	STS	0.03	-
Ottawa 20–30 [97]	-	I	0.65	-	1.2	1.0	100 × 50	STS	0.15	6.9
Ottawa 20–30 [84]	FR (0%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.15	7.57
Ottawa 20–30 [84]	FR (0.4%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.29	8.50
Ottawa 20–30 [84]	FR (0.8%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.44	8.93
Ottawa 20–30 [83]	FR (0%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.05	4.03
Ottawa 20–30 [83]	FR (0.2%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.07	4.10
Ottawa 20–30 [83]	FR (0.4%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.07	4.30
Ottawa 20–30 [83]	FR (0.6%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.08	4.30
Ottawa 20–30 [83]	FR (0.8%)	I	0.65	0.80	1.17	1.02	100 × 50	STS	0.09	4.50
Silica sand [85]	BFR (0%)	I	0.21	0.28	1.3	0.98	100 × 50	STS	0.09	10.08
Silica sand [85]	BFR (0.4%)	I	0.21	0.28	1.3	0.98	100 × 50	STS	0.32	14.08
Silica sand [85]	BFR (0.6%)	I	0.21	0.28	1.3	0.98	100 × 50	STS	0.37	15.94
Silica sand [85]	BFR (0.8%)	I	0.21	0.28	1.3	0.98	100 × 50	STS	0.44	17.14
Silica sand [85]	BFR (1.0%)	I	0.21	0.28	1.3	0.98	100 × 50	STS	0.45	17.58
Silica sand [85]	BFR (1.2%)	I	0.21	0.28	1.3	0.98	100 × 50	STS	0.33	14.87

EM—Enhancement method; I—Injection; TM—Treatment method; FR—Fibre reinforcement; BFR—Basalt fibre reinforcement; MC—Mixed and compacted; STS—Splitting tensile strength; E^s₅₀—secant elastic modulus at 50% of the peak tensile stress.

.

To evaluate the influence of different factors such as soil type, Cc and fibre content on the STS of MICP-treated soils, data from previous studies were compiled and critically analysed. Figure 6a shows a plot of STS versus Cc for different soil types (represented by D₁₀). The STS of MICP-treated soils ranges from 0.04 to 1.06 MPa, C_C ranges from 3.8 to 31.0%, while the fibre content ranges from 0 to 1.2%. As shown in Figure 6a, STS increases exponentially with increasing C_C; however, three distinct trends were observed possibly due to the effect of the particle size distribution not being properly captured by D₁₀ as well as the grain shape.

In Figure 6b, STS was plotted against C_C for different fibre contents. A good correlation was observed between STS and C_C, with two distinct trends. This is an indication that both Cc and the fibre content have a combined effect on the STS of MICP-treated soils. For example, considering similar soil types (same D₁₀) and fibre content, a slight variation in STS was observed possibly due to differences in C_C.

4.3. Triaxial Tests

Many researchers have conducted triaxial testing on bio-cemented soils under drained and undrained conditions [3,28,42,105,106,107]. Ismail et al. [106] studied the mechanical behaviour of cemented soil under CIUC triaxial testing. Their study showed that soil samples treated with Portland cement exhibited ductile yield whereas samples treated with gypsum or MICP showed brittle behaviour. Similar observations were reported by DeJong et al. (2006) where MICP- and gypsum-cemented samples exhibited brittle behaviour compared to uncemented samples.

Lin et al. [107] conducted a series of tests on MICP-treated soil under consolidated isotropically drained compression (CIDC) triaxial testing. In their study, the effect of different substrate (specifically CaCl₂) concentrations on the mechanical behaviour of MICP-treated soils was evaluated. Their observations showed that MICP-cemented sand specimens exhibited an increase in peak deviatoric stress with an increase in the CaCl₂ concentration. Cui et al. [105] studied the effect of different cementation levels on the mechanical behaviour of MICP-treated sand through a series of CIDC triaxial tests. Their study showed that a higher degree of cementation resulted in increased strength. A summary of the studies can be found in

Table 4. A summary of the literature on MICP and triaxial tests.

Sand type	* MICP Process		D50 (mm)	Cus	Ccs	Testing Method
	Dr (%)	NTC				Size H × D (mm)	Method	${\mathbf{p}}_0^{\prime }$ (kPa)	Peak c′ (kPa)	Peak ϕ (°)	NL
Ottawa 20–30 [88]	36–42	6–29	0.72	1.17	1.02	144 × 72	CIDC	10–400	109	41	-
Ottawa 50–70 [88]	36–45	6–35	0.22	1.40	0.90	144 × 72	CIDC	10–400	108	39	-
Ottawa 20–30 [91]	37	10	0.72	1.17	1.02	144 × 72	CIDC	100	-	-	-
Ottawa 50–70 [91]	36	13	0.22	1.40	0.90	144 × 72	CIDC	100	-	-	-
Nevada [88]	36–42	10–46	0.12	1.70	1.24	144 × 72	CIDC	10–400	126	38	-
Itterbeck [108]	-	8–16	0.19	2.10	0.86	100 × 50	CIUC	100–400	253	44	-
CHINA ISO [105]	30/80	2–16	0.52	5.08	1.05	80 × 39.1	CIUC	100–300	450	45	-
Ottawa 50–70 [109]	40–75	10–40	0.22	1.40	0.90	145 × 72	CIDC	100–400	59	41	-
Ottawa 20–30 [107]	39	-	0.71	1.17	1.02	145 × 72	CIDC	25–100	-	-	-
Ottawa 50–70 [107]	41	-	0.33	1.43	1.01	145 × 72	CIDC	25–100	-	-	-
Ottawa 50–70 [28]	40	6–16	0.22	1.40	0.90	144 × 72	CIDC/CIUC	100	-	43.7	-
Calcareous sand [79]	40	-	0.36	2.36	1.01	80 × 40	CIDC	100–400	425.6	43.2	-
Calcareous sand [92]	10–80	3–6	0.38	2.26	1.03	80 × 39	Cyclic	100	-	-	8–380

* All MICP specimens were treated using the injection method; D_r—Relative density; N_TC—Number of treatment cycles; ${\mathrm{p}}_0^{\prime }$—Initial effective confining stress; C’—Cohesion; ϕ—Frictional angle; N_L—Number of cycles to liquefaction.

.

4.3.1. Undrained Behaviour

In general, the behaviour of MICP-treated soils is affected by the initial relative density (D_r), confinement ($p_0^{\prime }$) and cementation levels in term of C_C. Figure 7 compares typical undrained shearing responses of MICP-treated and untreated sandy soils. At a constant $p_0^{\prime }$ = 100 kPa and D_r, the MICP-treated soils under undrained shearing show breakage of bonds (brittleness) at different axial strains (Figure 7a). The MICP-treated soils also exhibit higher peak and residual strengths and increased initial shear stiffness compared to the untreated state when plotted in $q$–$p^{\prime }$ space (Figure 7b).

It is worth noting that the undrained triaxial behaviour of both MICP-treated and untreated soils, as shown in Figure 7, were for the same initial $p_0^{\prime }$ and D_r. Therefore, the difference in behaviour was due to the difference in their C_C. MICP-treated soils with high C_C exhibited higher strength (peak deviatoric stress, $q_{peak}$). For example, the MICP-treated specimen (with C_C = 11.86%) exhibited approximately five times higher strength compared to the untreated soil (Figure 7a). Also, the specimen with C_C exhibited higher brittleness compared to the ductile nature of low cemented MICP-treated soils [3,28,105]. Therefore, a further detailed analysis was performed to correlate C_C with other index properties such as the brittleness index (BI), peak deviatoric stress ($q_{peak}$) and residual stress ratio (${\eta }_{Res}=q/p\prime$) at the end of the tests.

The ductile or brittle failure was also evaluated using the brittleness index (BI) proposed by Consoli et al. [110], and later, many adopted a similar approach to evaluate soil behaviour [111,112,113]. The BI can be expressed as:

$$BI=\left(q_{peak}/q_{Res}\right)-1$$

(7)

where $q_{Res}$ is the peak residual strength. The results of the variation of C_C with $q_{peak}$, BI and ${\eta }_{Res}$ are plotted in Figure 8. As shown in Figure 8a, $q_{peak}$ increases exponentially with C_C. A good correlation was obtained for $q_{peak}$ versus C_C with a coefficient of determination (R²) of 0.99. A plot of BI versus C_C is shown in Figure 8b; here, the BI of MICP-treated soil increased exponentially with C_C (with R² of 0.75). However, it is worth noting that the rate of increase of BI was significantly higher only after C_C > 3%, i.e., there may be a threshold C_C of approximately 3% for these tests. Figure 8c shows a plot of ${\eta }_{Res}$ versus C_C. It is evident that ${\eta }_{Res}$ ranges from 1.4 to 1.8, with no observed relation with C_C, i.e., ${\eta }_{Res}$ did not change significantly for both untreated and MICP-treated soils.

4.3.2. Drained Behaviour

In this study, data from drained triaxial tests are compiled from the literature [28,88,91,107,109,114]. To reduce the effect of soil particle size and geometry, two distinct soil types (Ottawa 50–70 in Figure 9a,b and Ottawa 20–30 in Figure 9c,d) were considered. Figure 9 shows the typical drained behaviour of MICP-treated and untreated sand at $p_0^{\prime }$ = 100–400 kPa. Plots of deviatoric stress ($q$) versus axial strain for Ottawa 50–70 and 20–30 are shown in Figure 9a,c, respectively. A comparison of untreated and MICP-treated soils shows that for the same $p_0^{\prime }$ and and similar D_r (40 ± 10%), the MICP-treated specimens exhibit approximately three times higher strength compared to the untreated soils. For example, at $p_0^{\prime }$ = 100, the MICP-treated soils show peak $q$ = 318 kPa at C_C = 0.6% [28], $q$ = 375 kPa at C_C = 2.8% [91], $q$ = 378 kPa at C_C = 1.5% [107], $q$ = 780 kPa at C_C = 4.3% [109] and $q$ = 1059 kPa at C_C = 5.0% [88] compared with the untreated Ottawa 50–70 sand which shows peak $q$ = 228 kPa [28] (Figure 9a). This also indicates that the peak strength and the residual strength both increased with an increasing cementation level of the MICP-treated soil.

Plots of volumetric and axial strain for Ottawa 50–70 and 20–30 are presented in Figure 9b,d, respectively. For the same $p_0^{\prime }$, the volumetric strain response was found to be more dilative for MICP-treated soils compared to the untreated sand. For example, the MICP-treated soils exhibit less contraction (with no contraction observed in some cases) at small axial strains (less than 3%) followed by more dilation at large axial strains compared to the untreated sand specimen. Similar to the observed behaviour in the stress-strain plot, the dilative behaviour of the MICP-treated soils increased with increasing C_C.

To quantitatively assess the influence of C_C on the drained shear strength behaviour of MICP-treated Ottawa sand, $q_{peak}$ was obtained from Figure 9 and plotted with C_C in Figure 10a. Similar to the results for undrained behaviour, $q_{peak}$ increases exponentially with C_C as shown in Figure 10a with an R² value of 0.99. The dilatancy (d), which is defined as the rate of plastic volume change ($d{\epsilon }_v^p$) with the plastic deviatoric strain ($d{\epsilon }_q^p$), is an important characteristic behaviour of drained soil [114,115,116] and has been used in prediction/constitutive modelling of soil behaviour [117,118]. The dilatancy (d) can be calculated using Equation (8). Soil and slightly cemented/treated soil exhibit negligible elastic behaviour under large strain; therefore, the total volumetric and deviatoric strains were assumed to be plastic [117,119].

$$d=\left(d{\epsilon }_v^p/d{\epsilon }_q^p\right)\approx \left(d{\epsilon }_v/d{\epsilon }_q\right)$$

(8)

As in Figure 10b, d also varies exponentially with C_C with two different trends for the two soils. Although limited datasets were used for this analysis, it is envisaged that this behaviour may provide a pathway for future studies on the drained triaxial response of MICP-treated soils and a unified constitutive model.

4.4. Modelling of MICP-Cemented Soils

4.4.1. Empirical Correlations

The strength and stiffness of MICP-treated soils have been correlated with parameters such as the CaCO₃ content [76,103,120], shear wave velocity (Vs) [121], frictional angle (ϕ) [27], curing time [76] and soil types [39,122]. There have many attempts to empirically model the strength and stiffness of MICP-treated soils. These attempts can be broadly classified into three groups, namely, as a function of Cc, as a function of porosity, the volume of cementation solution and Cc and as a function of soil particle size and Cc.

UCS as a Function of C_C

Several empirical relationships or equations have been proposed to correlate UCS with C_C. For example, Harkes et al. [120] developed the following relationship (Equation (9)) for MICP-treated Itterbeck fine sand (D₅₀ = 0.17) to correlate UCS with C_C (R² = 0.91). However, there were inconsistencies with the physical meaning of the equation terms. For example, at 0% C_C, the equation gives a UCS value of 128.11kPa, which is not true for coarse-grained soils such as sands.

$$UCS=128.11\times e^{0.0112\times C_C}$$

(9)

Li et al. [103] proposed a parabolic function to correlate UCS and C_C for a sand (D₅₀ = 0.42). They imposed a limiting condition of UCS = 0 when C_C is less than the intercept to avoid inconsistencies for UCS at Cc = 0.

$$UCS={(a\times C_C+b)}^2-c\ge 0;\hspace{1em}\hspace{1em}Intercept=\frac{\sqrt{c}-b}{a}$$

(10)

where a, b and c are constant parameters that can be obtained via regression analysis.

Another non-linear empirical function was proposed by Choi et al. [19] to correlate the UCS and C_C of MICP-treated soils based on compiled data from the literature for a range of soil types. Their function was expressed as:

$$UCS={\alpha }_{UCS}\times {\left(C_C/1\%\right)}^{{\beta }_{UCS}},for{C}_C<30\%$$

(11)

where ${\alpha }_{UCS}$ and ${\beta }_{UCS}$ are fitting parameters.

Vs as a Function of C_C

Qabany et al. [121] proposed an empirical correlation (Equation (11)) between shear wave velocity (V_s) from bender element testing and the C_C of MICP-treated Ottawa 50–70 sand (D₅₀ = 0.22) and proposed that this could be used as an indirect indicator or measure of C_C.

$$V_s=9.7\left(C_c\right)+147$$

(12)

UCS or STS as a Function of Porosity, Volume of Cementation Solution and C_C

Xiao et al. [85] adopted the empirical relationship proposed by Consoli et al. [123] to predict the UCS or STS of BFR-MICP cemented sand. They observed in their study that the UCS and STS were primarily affected by the specimen porosity (n) and cementing-agent volumetric content (C_iv). Their equation is presented below.

$$UCSorSTS=A{\left(I_{cf}\right)}^{-\alpha }=A{[n/{\left(C_{iv}\right)}^{\beta }]}^{-\alpha }$$

(13)

where $I_{cf}$ = cementing factor, and A, $\beta$ and $\alpha$ are fitting parameters.

Liu et al. [79] proposed a sample volume ratio (R_C) parameter, which was expressed as the volume of the injected cementation solution (V_C) divided by the total volume of the sample (V). Here, the molarity of the cementation solution (C) was normalized by the standard concentration (C_a). R_C is defined by the following equation:

$$R_C=\frac{V_c\left(C/C_a\right)}{V}$$

(14)

The R_C parameter gave a reasonably accurate and convenient determination of C_C, and this avoided the determination of the CaCO₃ content by the conventional acid-washing technique. Their study also suggested that a good correlation could be achieved between UCS or STS and R_C using the following empirical equation:

$$UCSorSTS=m\times exp\left(n\times R_C\right)$$

(15)

where m and n are fitting parameters.

In order to evaluate the performance of the parameter R_C, a detailed assessment was undertaken in the current study by correlating UCS or STS and R_C for data compiled from previous studies (Figure 11). Figure 11a shows a plot of UCS versus R_C for different soil types (specifically D₁₀). It is worth noting that R_C shows an exponential relationship with UCS for a few datasets especially data from the authors [79], who proposed the R_C parameter. However, for other sets of data from different authors, R_C did not have a good correlation with UCS. A similar behaviour (trend) was observed for STS versus R_C as shown in Figure 11b.

It is evident that the R_C parameter can only capture the volume of the injected cementation solution per sample volume for a given chemical constituent concentration, which may not have a direct relationship with the strength of MICP-treated soils. For a given MICP treatment, the strength (UCS or STS) may be influenced by several factors including the treatment method, urease activity of bacteria, the molar ratio of chemical constituents, curing time, soil type and geometry. It is generally known that most of these factors control the amount of precipitated CaCO₃ within the soil. Therefore, further research is warranted to develop a model that can correlate the strength of MICP-treated soils with C_C and soil type.

UCS or STS as a Function of Soil Particle Size and Cc

A study by Rahman and Hora [124] correlated soil grading (mainly D₁₀) and C_C with the UCS of MICP-treated soils. The study used data from previous studies and selected tests with soil specimens of diameter 50 mm and height 100 mm under saturated conditions. A statistical analysis tool was used to analyse these data points to understand the variation of UCS with D₁₀ and C_C. A 3D contour plot, as shown in Figure 12a, was generated with this method, which shows a significant influence of D₁₀ on the UCS strength gains. The study also performed a regression analysis on the same data points and presented a best-fit equation between these three parameters as:

$$\frac{UCS}{UC{S}_{10\%C_C}}=0.689-3.85D_{10}+0.073C_C+2.75D_{10}{}^2+0.196D_{10}{C}_C-0.0017{\left(C_C\right)}^2$$

(16)

The UCS values were normalized by the UCS values at 10% CaCO₃ of corresponding soils, and Figure 12b presents the 3D plot with the above equation as a mesh surface.

The analysis reported by Rahman and Hora [124] indicates a significant influence of D₁₀ on UCS, but the analysis was performed on limited data sets through some screening criteria. Therefore, this is a preliminary study that can provide a pathway for developing better models in the future.

4.4.2. Constitutive Modelling

Despite numerous studies conducted, limited attention has been paid to the constitutive behaviour of MICP-treated soils. Several constitutive models can be found in the literature that have been developed for cemented soils [46,125,126,127,128,129,130,131,132,133,134]. However, only in a very limited number of cases have the models been either developed or validated specifically for MICP-treated soils. El Kortbawi et al. [135] evaluated the capability of a bounding surface plasticity model developed for granular materials [136] to capture different behavioural aspects of an MICP-treated soil (see Figure 13 for a comparison of experimental data and model prediction). They concluded that accounting for degradation of cementation, differences in the location of CSL (because of broken cementation in the final material) may improve the model capability.

Nweke and Pestana [134] proposed modification to an existing constitutive model developed for cemented soils [137] to better capture the behaviour of MICP-treated soils. Their model allowed the degradation of strength and stiffness with the degradation of CaCO₃ bonds. The following equation was proposed to achieve a stable friction angle at lower stresses which degraded towards a critical state value at larger confining pressures.

$${\varphi }_{peak}-{\varphi }_{cs}=\alpha \left(\frac{{\varphi }_{cs}}{2}\right)D_r\mathit{exp}\left[\left(\frac{-0.24\beta }{D_r}\right){\left(\frac{P_f}{P_{atm}}+P_t\right)}^m\right]$$

(17)

where ${\varphi }_{cs}$ is the friction angle at a critical state, $D_r$ is the initial relative density, $P_f$ is the mean effective stress at failure, $P_t$ represents the effect of cementation, $P_{atm}$ is the atmospheric pressure, α and β represent the effect of grain shape and size and m is a constant.

The maximum shear modulus ($G_{max}$) of treated soils was estimated by incorporating the degradation cementation using a “damage” model as expressed in the following equations:

$$\frac{G_{max}}{P_{atm}}=\frac{G_b}{e_0{}^{1.3}}{\left(\frac{P_f}{P_{atm}}+P_t\right)}^{0.5}$$

with

$$P_t=\frac{P_{to}}{1+a{\gamma }^b}$$

(18)

where $G_b$ is a constant depending on the soil type, e is the initial void ratio, $P_{to}$ is the initial $P_t$, $\gamma$ is the shear strain and a and b are the degradation model parameters.

Despite the capability of the above model in predicting the behaviour of lightly cemented soils (see Figure 14), it, however, does not incorporate the effect of cementation on the void ratio of soils after MICP treatment. Therefore, the model may not be able to predict the behaviour of heavily cemented (MICP) soils under high confining stresses.

Nweke and Pestana [138] proposed a modified form of the above model by introducing a cementation parameter to capture the degree of cementation. The modified equations are presented as:

$${\varphi }_{peak}-{\varphi }_{cs}=\alpha \left(\frac{{\varphi }_{cs}}{2}\right){\left(1+\frac{P_t}{P_{atm}}\right)}^m{D}_R\mathit{exp}\left[\left(\frac{-0.24\beta }{D_R}\right){\left(\frac{P_f}{P_{atm}}\right)}^m\right]$$

(19)

$$P_t=Z\left(\mathit{exp}\left(r{C}_C\right)-1\right)$$

(20)

Here, C_C is the cement content (%), Z and r are parameters that control the scale and rate of degradation.

An elastoplastic constitutive model for MICP-cemented soils was recently developed by Gai and Sánchez [46]. The mechanism for MICP soil enhancement was captured using a critical state yield surface and sub-loading concepts, while the bonding degradation effects under shear were captured using an evolution law. A detailed discussion of the main components of their model is discussed here.

Yield surface: Based on the hierarchical single surface (HISS) framework [139], three yield surfaces (YS), i.e., (1) the critical state soil mechanics (CSSM) YS, (2) the MICP-enhanced YS and (3) the sub-loading YS were considered. The CSSM YS was used to describes the mechanical response of the untreated soil and is represented by the equation below:

$$F_{CSSM}=\frac{\alpha }{3M^2}{q}^2-9\gamma p^{\prime 2}+9\gamma p^{\prime n}{p}_c^{2-n}$$

(21)

Here, $p^{\prime }$ and $q$ are the mean effective and deviatoric stresses, respectively; M is the slope of the critical state line in the $q$–$p^{\prime }$ space; $p_c^{\prime }$ is the effective pre-consolidation pressure; n controls the transition from compressive to dilative volume change behaviours; a and c are constants that define the shape of the yield surface.

The MICP-enhanced yield surface considers the mechanical influence of CaCO₃ precipitation and is represented by the equation below:

$$F_{MICP}=\frac{q^2}{M^2}-p^{\prime 2}-p^{\prime }\left(p_c+p_b\right)$$

(22)

To account for the effect of bonding effect via CaCO₃ precipitation, a linear relationship was assumed to relate the CaCO₃ content and the mechanical hardening parameter ($p_b$) as described in the equation below:

$$p_b=a\left(X{m}_c\right)$$

(23)

Here, $m_c$ represents the CaCO₃ content and a and X are the scaling constant and the damage factor, respectively, which account for the degradation of the cementation during loading.

The sub-loading YS as part of the CSSM framework was also resented by the equation below:

$$F_{SUB}=\frac{q^2}{M^2}-p^{\prime 2}-R{p}^{\prime }\left(p_c+p_b\right)$$

(24)

where R is the sub-loading ratio (0 < R < 1).

Elastic moduli: The elastic bulk modulus (K) and shear modulus (G) were presented as a function of volumetric state, confinement and Poisson’s ratio as shown below,

$$G=\frac{3\left(1-2v\right)}{2\left(1+v\right)}K$$

(25)

where v is the Poisson’s ratio and K is a constant parameter expressed as:

$$K=\frac{1+e}{k}{p}^{\prime }$$

(26)

Here, e represents the void ratio and k is the slope of the unloading/reloading line in the e–log(p’) space.

Hardening law and flow rule: The proposed model incorporated three different strain hardening parameters: $p_c$, R and $p_b$. In the elastoplastic model, the hardening/softening behaviour of soils depends on the plastic volumetric strains. However, here the (hardening) pre-consolidated pressure depended on both $d{\epsilon }_v^p$ and $d{\epsilon }_q^p$, as expressed as:

$$\frac{d{p}_c}{p_c}=\frac{1+e}{\lambda -K}d{\epsilon }_v^p+D_s\frac{1+e}{\lambda -K}d{\epsilon }_q^p$$

(27)

where λ represents the slope of the normal compression line in e–log(p’) space, and $D_s$ is an exponential parameter associated with soil dilation at failure.

Stress-strain relationship: The stress-strain relationship used in their proposed elastoplastic model can be expressed as:

$$d{\sigma }^{\prime }=D^e\left(d\epsilon -d{\epsilon }^p\right)=Dd\epsilon +D_{mc}d{m}_c$$

(28)

where $d\epsilon$ is the total strain increament, D is the elastoplastic stiffness matrix and $D_{mc}$ is associated with changes in the CaCO₃ content.

The experimental data (from Cui et al. [105]) and model prediction (from Gai and Sánchez [46]) of the stress-strain and volumetric responses for the MICP-treated and untreated specimens are presented in Figure 15a,b respectively. The overall response of both untreated and treated soils shows a dominant dilative volume change behaviour, which becomes more pronounced with an increase in C_C. By comparing both experimental and modelling results, it is evident that the constitutive model developed by Gai and Sánchez [46] was able to capture the main features of both untreated and treated soil behaviours, as follows: (1) increase in strength and stiffness with an increasing C_C (Figure 15a) and (2) post-peak stress trends for different C_C values. However, the overall dilative volumetric behaviour for different C_C values was not accurately captured (Figure 15b).

4.4.3. DEM Modelling

In general, DEM modelling of chemically cemented soils (e.g., Portland cement) has been investigated by several researchers [140,141,142,143,144,145,146]. Some studies have attempted to also model the micro-scale behaviour of MICP-cemented soils using DEM [47,48,49,50]. A summary of these studies is presented in

Table 5. A summary of DEM studies on cemented MICP and triaxial tests.

Cementation Approach	Object Dimension	No. of Particles	Particle Shape	Testing Method	Software	Contact Bond Model
MICP [151]	Rectangular (3D)	1000	Spherical	CIUC	YADE	CBM
MICP [152]	Rectangular (3D)	1000	Spherical	CIDC	YADE	CBM
MICP [47]	Cubic (3D)	-	Spherical	-	PFC	CRBM
MICP [48]	Cubic (3D)	53,000	Spherical	-	PFC	CRBM
MICP [49]	Cylindrical (3D)	13,000	DPS	CIDC	PFC	PBM
MICP [50]	Cylindrical (3D)	7000	Spherical	CIDC	PFC	PBM
OPC [149]	Cylindrical (3D)	52,000	Spherical	CIDC	PFC	PBM
OPC [144]	Cylindrical (3D)	3350	Spherical	CIDC	PFC	PBM
OPC [150]	Cylindrical (3D)	49,864	Spherical	CIDC	PFC	CRBM
OPC [145]	Rectangular (3D)	20,000	Spherical	CIDC	PFC	PBM and SBM
OPC [142,143]	Rectangular (2D)	8000	-	CIDC	PFC	CRBM + PBM
OPC [146]	Cubic (3D)	1400	Spherical	CIDC	PFC	BM
BSMT [148]	Cylindrical (3D)	19,600	Spherical	CIDC	PFC	PBM

OPC—Ordinary Portland cement; BSMT—Bioplymer stabilized mine tailings; DPS—Different particle shape; PFC—Particle flow code.

. In DEM, a contact bond model represents the cementing material binding adjacent soil particles that is considered to behave as a glue [147]. Different contact bond models have been proposed in an attempt to simulate the behaviour of the bond that exists between the cemented material and the soil particles. Examples of contact bond models used to simulate the behaviour of MICP-cemented soils include the parallel bond model (PBM) [49,50,142,143,144,148,149], the beam model (BM) [146], the cement ring bond model (CRBM) [47,48,142,143,150], the cohesive bond model (CBM) [151,152] and the serial bond model (SBM) [145].

Parallel bond model: A parallel bond is a cylindrical material that can be used to represent the force-displacement behaviour of cemented material between two particles in contact, as shown in Figure 16. This arrangement is similar to the bond between sand and CaCO₃ in MICP, as shown by the scanning electron microscopy (SEM) image in Figure 16a.

A parallel bond can be envisaged as a set of elastic springs with normal and shear stiffness acting in parallel (Figure 16b). The force and moment acting on the bonded particles can be linked to the maximum shear and normal stresses acting around the bond (Figure 16c). In DEM simulation, the parallel bond is defined by the following parameters: normal stiffness (${\overline{K}}^n$), shear stiffness (${\overline{K}}^s$), normal strength (${\overline{\sigma }}_c$), shear strength (${\overline{\tau }}_c$) and parallel bond radius ($\overline{R}$). Similar to a cemented physical specimen, the parallel bond breaks when the shear or normal forces exceed their shear or normal strengths. The total force (${\overline{F}}_i$) in a parallel bond can be resolved into normal (${\overline{F}}_i^n$) and shear (${\overline{F}}_i^s$) components and can be expressed as:

$${\overline{F}}_i={\overline{F}}_i^n+{\overline{F}}_i^s$$

(29)

In general, the parallel bond model has been widely used in DEM to simulate the behaviour of cemented soils. However, the model has the following limitations:

• Parallel bonds in a DEM simulation usually underestimate the residual strength of cemented soils since the bond shows progressive breakage during loading when the shear or normal forces exceed their shear or normal strengths.
• Simulation using parallel bonds cannot capture both cohesive failure (bond breakage within the calcite phase) and adhesive failure (bond breakage between the calcite minerals and sand grains).
• The model assumes uniformity for the size and distribution of the cementation material, which is usually unrealistic compared to a physical experiment.

Cement ring bond model: Evans et al. [47] proposed a biocementation model in which progressive failure of a bond at the contact point between two particles can be modelled. In this model, two sand particles are joined by a ring that consists of several smaller cement particles (Figure 17a). This model was used to capture the progressive failure (ductile behaviour) of bio-cemented sand since sand-cement bonds do not necessarily fail at the same time. The overall bond in this model prevents sliding or separation and can also transmit moments. Figure 17b shows the geometry of two overlapping particles (a and b) with radii of $r_1$ and $r_2$ connected through a cement chain. Here, the cement particles have a radius of $r_3$, while the cement chain has a radius of $\rho$. The centre to centre distance of the sand particles is R.

Despite the ability of this model to capture the residual strength, nonlinearity and softening of cemented soils in DEM, the model utilises large amounts of particles, which can be computationally demanding and may increase the simulation time.

Beam bond model: Obermayr et al. (2013) performed a numerical simulation of cemented sand using DEM. The study approximated the soil grains as a spherical particle bonded together by a beam of cementitious material as shown in Figure 18a. The bond element obeys a linear elastic material law with finite values for displacements and rotations. In this bond model, two particles (1 and 2) are assumed to be in contact and bonded by a beam element that connects the centres of the spheres (Figure 18b). The bond element is represented by a beam with circular cross-section of radius $r_b$ which can be expressed by the following equation:

$$r_b={\overline{r}}_b\left(r_1+r_2\right)/2$$

(30)

where ${\overline{r}}_b$ is the relative radius of the bonded element, and $r_1$ and $r_2$ are the radii of the bonded particles.

5. Key Engineering Applications of MICP

5.1. MICP as Binders

Le Metayer-Levrel et al. [153] suggested using the bacterial ability of MICP for producing superficial protective coatings for limestones buildings, monuments and statuary. Another study by Webster and May [154] also suggested bioremediation as an additional technology for restoring stone surfaces in heritage buildings.

MICP being a non-toxic and eco-friendly process has advantages over commonly used methods for binding soil particles, such as chemical grouting. Ivanov and Chu [26] evaluated the cost of raw materials for chemical grouting to be in the range of $2–$72 per m³ of soil whereas for microbial grouting was in the range of $0.5–$9 per m³ of soil when waste materials are used as a carbon source for microbial growth.

Ramachandran et al. [155] concluded through microscopy investigation that MICP is an effective method for crack remediation in concrete. Jonkers et al. [156] established that MICP is effective as a self-healing agent to activate the process of autonomous repair of freshly formed cracks. Achal et al. [157] suggested MICP as an alternative high-quality concrete sealant and crack remediation method which demonstrated a 36% increase in compressive strength of cement mortar as well as six times lower water absorption in the treated samples. Amidi and Wang [158] proposed a new surface treatment method for treating concrete and similar absorbent materials to enhance their resilience and mechanical properties and achieved a 36% increase in compressive strength due to MICP.

5.2. Soil Strengthening and Stabilisation

Multiple studies have applied MICP to different types of soil and tested these under various conditions for strength enhancement and soil stability. DeJong et al. [3] applied MICP to improve the engineering properties of sands such as shear strength and stiffness. The results showed that the ultimate shear capacity and initial shear stiffness were both higher for treated samples compared to untreated loose specimens. Whiffin et al. [4] applied MICP successfully using a 5 m long sand column for ground improvement which was achieved with relatively low flow rates. This study also mentioned that balancing the rate of urea hydrolysis with the delivery of reactants aided in the uniform distribution of CaCO₃.

A study by Harkes et al. [5] evaluated MICP as in situ soil strengthening technique in fine-grained sand. The study reported that for a homogenous distribution of bacteria in large sand bodies, a low ionic strength solution promoted bacterial transport over longer distances. Many researchers have evaluated the strength of biotreated sands and have demonstrated improved strength, increased stiffness, liquefaction resistance and enhanced dynamic properties of the treated specimens [30,31,87,100].

5.3. MICP in Bricks

Bricks constitute a significant part of construction materials and are known for their durability and sustainability. However, bricks are also prone to deterioration over time due to the presence of voids and pores resulting in cracking. MICP has proved to be a novel method of treating these cracks or strengthening bricks [159]. Raut et al. [160] demonstrated MICP in bricks and studied the effect of the method on compressive strength and water absorption capacity. Bricks treated with MICP showed 83.9% improvement in compressive strength and 48.9% lower water absorption capacity after 28 days as compared to the control specimen. Lambert and Randall [161] evaluated the process of MICP to produce bio-bricks using the urea from stabilized human urine. Results demonstrated higher compressive strength with an increase in the number of treatments with the highest compressive strength of 2.7 MPa.

5.4. Remediation of Contaminants from the Environment

Rapid industrial development poses a major threat in the form of heavy metals and other contaminants as a by-product of these industries which impacts our environment. In the past, conventional treatments were used to remove heavy metals from contaminated environments. However, these methods are ineffective, expensive and consume high amounts of chemicals and energy [162]. Therefore, alternative methods such as MICP are needed to effectively remove heavy metals without having much impact on the environment. Several researchers [2,23,163] have reported the capability of MICP for heavy metal remediation in the environment.

6. Sustainability Analysis

The cost of MICP treatment depends on the specific processes and ingredients used. Also, with MICP being a relatively new technology with very few reports of large-scale engineering implementation [41,58,164,165,166,167], the actual cost of treatment can be difficult to estimate. A wide range of costs has been reported in the literature. For example, Ivanov and Chu [26] estimated the material cost for MICP to be as low as $5/m³ of treated soil; while Esnault-Filet et al. [168] estimated the cost of materials, labour and equipment to be as high as $500/m³ of treated soil. In this section, a quantitative sustainability assessment has been conducted for a simplified hypothetical engineering construction scenario. The cost of treatment, together with its environmental benefits in terms of reducing the CO₂ and energy footprints, has been estimated and compared with conventional construction methods and materials.

The engineering application chosen here is concrete block pavement (CBP). In general, CBP consists of a layer of rigid blocks laid on top of a sand or gravel bedding course underlain by granular or other materials depending on traffic conditions and the mechanical properties of the subgrade. CBPs are often treated as flexible pavement and are commonly used in road or industrial applications [169]. Common configurations of CBPs are presented in Figure 19. This particular application is chosen for two reasons. Firstly, CBP, similar to any other pavement structure, comprises several different material layers and there is a potential for using MICP treatment in more than one layer of a CBP structure. Secondly, the construction of roads/pavements consumes a large volume of materials and thus could magnify the costs or benefits in terms of CO₂ and energy footprints.

Three alternative design scenarios have been considered for the cost and environmental benefit analysis. In the first case, only the concrete pavers are treated with MICP. In the second case, both the concrete paver and the sub-base layer are improved with MICP. The third case involves MICP treatment of paver and the subgrade soil. In all three scenarios, comparisons have been made with conventional construction methods. The CBP design for each case has been conducted with a traffic load of 10 × 10⁶ expressed in terms of equivalent standard axle (ESA) repetition together with a subgrade soil elastic modulus of 35 MPa. Other related properties are discussed in the following sections. The calculations are presented for a 1 km long and 7 m wide road section.

6.1. Scenario 1—Only Paver Blocks Treated with MICP

A layer configuration similar to Figure 19b has been chosen. The properties of the different materials used are listed in

Table 6. Mechanical properties of the different materials used in designing a pavement structure for Scenario 1 and the volume required for construction of a 1 km long and 7 m wide road.

Layers	Thickness (mm)	Elastic Modulus * (MPa)	Poisson’s Ratio	Vol. Required for 1 km Length of Road (m3)
Paver	80	3200	0.30	560
Bedding sand	20	200	0.35	140
Base course	100	350	0.35	700
Sub-base course	445	250	0.35	3150
Subgrade	-	35	0.40	-

* a range of material properties can be found in the literature [169,175,176,177,178].

and these are typical values used in current design/construction practice. With MICP being a new technology, its field implementation needs to be progressed in a phased manner, and Porter et al. [170] advised that partial replacement of cement in concrete or other stabilized material would be a logical first step. Multiple studies [157,171,172] have reported an increase of compressive strength in the range of 17 to 36% in cementitious materials as a result of microbial treatment. Following from Achal et al. [157], in this study it is assumed that MICP using the treatment solution presented in

Table 7. Estimated cost of materials for MICP treatment.

Material	Quantity (g/L of Treatment Solution)	Unit Price ($/kg)	Cost/L ($)
Nutrient broth	8 *,#	-	-
NaCl	5 *	0.5 ^	0.0025
Urea	20 *	0.2 ^	0.004
CaCl2	2.77 *	0.2 ^	0.000555
		Total	0.00705

* values from Achal et al. [157] and Achal et al. [179]; # assumed to be sourced from waste material [157]; ^ approximate current market price (industrial grade)*.

inoculated with microbes will improve the compressive strength of the paver brick by 25% and thus allow 25% replacement of cement in paver concrete (approximated based on the relationship between the water-cement ratio and the compressive strength [173,174]).

The required compressive strength of the concrete to meet the elastic modulus requirements of the paver blocks in Table 6 can be estimated using correlations [180] and in this case, it is found to be 40 MPa. A concrete mix design calculation [174] indicates that to produce concrete of 40 MPa strength with a nominal aggregate size of 12.5 mm and a target slump of 50 mm (assumed based on current industry practice), approximately 485 kg/m³ of cement will be required. Thus, MICP treatment will allow replacement of approximately 121 kg of cement per m³ of concrete.

Table 7 shows the estimated quantity and cost (in Australian $) of chemicals to produce 1 L of treatment solution per the mix design and Achal et al. [157]. Approximately 600 L (200 L in mix design + 400 L for curing) of treatment solution will be required to treat 1 m³ of concrete, leading to a treatment cost of approximately $4.5/m³. It should be noted that the calculated cost is only approximate. Prices for industrial-grade chemicals have been used even though most reported work used laboratory grade chemicals. However, it is expected that with the progression of research, especially in the area of using recycled wastes in biocementation [179,181,182] and other developments, the cost will come down to that level. Also, to be noted here is that the labour costs have been assumed to be equal to the conventional method of the production of concrete (with or without MICP treatment).

The thickness of the different layers has been calculated using the traffic data and subgrade properties mentioned earlier with the aid of a computer program DesignPave [183], and these are shown in Table 6.

Figure 20 shows a comparison of the CO₂ footprint and energy consumption for treated and untreated scenarios. The process life cycles of conventional and MICP-treated concrete pavers considering the materials and processes involved, the differences in energy consumption and CO₂ emissions, and costs were estimated for the pavement system. As the same aggregate would be applied in both the cases, its contribution towards cost and energy is neutral. The energy consumption for manufacturing and transport is also assumed to be the same. For simplicity, rather than considering the total embedded energy, including the embedded energy in raw materials, for the cement and chemicals, only the energy intensity of the manufacturing processes was considered in this research.

As can be seen from Figure 21, MICP treatment of pavers yields potentially significant economic and environmental benefits over using conventional concrete pavers. The total economic and environmental costs (in the form of CO₂ and assuming the estimated unit prices are representative) for a 1 km road using both categories of bricks are provided in Figure 20, which shows MICP treatment contributes to a 19% environmental benefit (about 47,000 kg of CO₂) and 25% economic benefit. As both processes ensure the quality of the system, the risk factor is neutralized.

6.2. Scenario 2—Pavers and Sub-Base Layer Treated with MICP

Porter et al. [170] indicated that the treatment of granular pavement materials with MICP can lead to a significant improvement of the unconfined compressive strength (close to 1 MPa). The composition of the growth media and treatment solution per Porter et al. [170] is presented in

Table 8. Composition of growth media and cementation solution.

	Material	Quantity (g/L)	Unit Price $/kg	Cost/L ($)
Growth media
1.	Yeast extract	20	18 *	0.36
2.	Ammonium sulphate	10	0.15 ^	0.0015
3.	Nickel Chloride	0.0002592	10 ^	0.0000026
4.	Tris buffer (pH 9)	4.6575	150	0.0047
			Total	0.366
Cementation solution
1.	Urea	30.03	0.2 ^	0.006
2.	CaCl2	55.49	0.2 ^	0.0111
3.	Yeast extract	0.5	18 *	0.009
			Total	0.0261

* Estimated cost after [181]; ^approximate current market price (industrial grade).

. The MICP treatment included the use of 11% by mass of bacterial solution (one-off) followed by 20 treatments of cementation solution (1 pore volume = approximately 250 L each time) leading to an approximate treatment cost of $215 per m³ of stabilized materials. The costs (in Australian $) in Table 8 indicate that replacing yeast extract with a cheaper alternative such as recycled waste [179] and recycling a portion of the treatment solution together with the partial replacement with waste materials [179,182] can significantly reduce the cost of treatment, possibly to a market competitive level.

In this scenario, it is assumed that the sub-base layer has been treated with MICP and now acts as a stabilized sub-base with a UCS of 0.9 MPa. The elastic modulus of this layer can be estimated to be 3500 MPa [174]. For the same traffic and subgrade conditions as for Scenario 1,

Table 9. Mechanical properties of different materials used in designing a pavement structure for Scenario 2 including the volume required for construction of a 1 km long and 7 m wide road.

Layers	Thickness (mm)	Elastic Modulus (MPa)	Poisson’s Ratio	Vol. Required for 1 km Length of Road (m3)
Paver	80	3200	0.3	560
Bedding sand	20	200	0.35	140
Base course	100	350	0.35	700
Stabilised sub-base	330	3500	0.2	2310
Subgrade	-	35	0.4	-

presents the calculated design thicknesses of the stabilized sub-base layer. Assuming that the same level of performance (elastic modulus of 3500 MPa) can be achieved by treating the sub-base layer with 3–4% (by mass) cement [191], a cost and environmental benefit analysis is conducted as described below. The thickness of the stabilized sub-base layer has been calculated using the computer program DesignPave [183] by considering the traffic data and subgrade properties mentioned earlier.

Figure 22 shows a comparison of the CO₂ footprint and energy consumption for MICP-treated and conventionally (cement) treated scenarios for a stabilized sub-base. The aggregate quantity and characteristics for both cases as well as other components of the pavement are considered the same and thus are neutralized.

As can be seen from Figure 23, MICP treatment for stabilizing sub-bases requires significant economic and environmental costs compared to the conventional concrete cement stabilization. The difference occurs mainly due to the high volume of chemical solution requirements (about 5000 L/m³ of concrete) and the associated embedded energy (about 994 MJ additional energy required in MICP treatment per m³ of concrete compared to the treatment with cement). The total economic and environmental costs (in the form of CO₂) for a 1 km road using both categories of stabilization are provided in Figure 23, which shows that MICP treatment leads to a 3.4 times environmental impact and about 1.6 times the cost in materials compared to conventional cement treatment.

6.3. Scenario 3—Pavers and Subgrade Treated with MICP

Van Paassen et al. [41] treated a material volume of approximately 100 m³ and demonstrated the applicability of MICP to treat large volumes of soil. The quantities of various chemicals used in their study are presented in

Table 10. Composition of different chemical solutions used by van Paassen et al. [41].

	Material	Quantity (g/L)	Unit Price ($/kg)	Cost/L ($)
Nutrient broth
1.	Yeast extract	20	18 *	0.36
2.	Ammonium chloride	10	0.15 ^	0.0015
3.	Nickel Chloride	1.296	10 ^	0.0000026
			Total	0.366
0.05M CaCl2 solution
1.	CaCl2	5.55	0.2	0.00111
Reagent solution
1.	Urea	60.06	0.2 ^	0.022
2.	CaCl2	110.98	0.2 ^	0.0111
			Total	0.033

* Estimated cost after [181]; ^ approximate current market price (industrial grade).

. In total, 5 m³ of nutrient broth was inoculated with bacteria and after incubating, this broth was injected into the soil followed by 5 m³ of 0.05M CaCl₂ solution. This was followed by an injection of approximately 96 m³ of reagent solution in 10 batches (9.6 m³/batch). In their experiment, large variability in the treated strength of the soil was observed with a maximum achieved UCS value of 12.4 MPa. A volume of 0.05 m³ of nutrient broth along with 0.05 m³ of 0.05M CaCl₂ solution and 0.96 m³ of reagent solutions would be required to treat 1 m³ of subgrade soil. The cost of this treatment would be approximately $60/m³ as estimated in Table 10. Similar to the case stated in Scenario 2, it is possible to significantly reduce the cost of the chemical constituents to a level that can be competitive in the market.

With the UCS values in the reported range, it is possible to use a subgrade elastic modulus of 150 MPa [191]. For the same traffic condition but improved subgrade condition, the thickness of the sub-base layer can be estimated. The results are presented in

Table 11. Mechanical properties of various materials used in designing the pavement structure for Scenario 3 and the volume required for construction of a 1 km long and 7 m wide road.

Layers	Thickness (mm)	Elastic Modulus (MPa)	Poisson’s Ratio	Vol. Required for 1 km Length of Road (m3)
Paver	80	3200	0.3	560
Bedding sand	20	200	0.35	140
Base course	100	350	0.35	700
Granular sub-base	260	250	0.35	1820
Subgrade	-	150	0.4	-

.

For this scenario, it is assumed that cement treatment of the top 500 mm of the soil layer can achieve the same level of performance. Usual cement treatments in such scenarios are undertaken with approximately 3–4% of cement (by mass). Assuming a typical unit weight of the soil of 17 kN/m³, approximately 70 kg of cement would be required to treat 1 m³ of subgrade.

The process life cycles of cement-treated and MICP-treated subgrade stabilization are assessed for a pavement system and the differences in energy consumption, CO₂ emissions and costs are shown in Figure 24. Similar to the previous two scenarios, the aggregate quantity and characteristics for both cases are considered to be the same, and the costs for pavers also remain the same.

While the MICP treatment for sub-base stabilization is relatively costly, Figure 25 shows that MICP treatment for stabilizing subgrade yields relatively lower economic and environmental costs compared to conventional concrete cement stabilization. Again, the amount of chemicals inoculated into the system (about 1060 L of chemicals per m³ of concrete) contributes mainly to the cost. The calculation shows that MICP treatment leads to a 9% economic loss and about a 19% environmental loss (in the form of CO₂) when MICP treatment is applied to both subgrade stabilization and the pavers (Figure 25). The total estimated economic and environmental costs for a 1 km road using both types of stabilization and pavers are also provided in Figure 25.

7. Conclusions

This paper has reviewed developments over the last 25 years in microbial-induced calcite precipitation (MICP), which has significant potential to become an economic and environmentally sustainable solution to engineering problems such as soil and concrete strengthening, liquefaction and remediation. The findings of detailed studies on the biogeochemical mechanisms of MICP, such as urea hydrolysis, denitrification and sulphate reduction, have been summarised. The engineering properties of MICP-treated soils have also been described in terms of both unconfined compressive strength and indirect tensile strength. Methods of modelling MICP-cemented soils have also been reviewed, including empirical correlations, constitutive modelling and modelling using the Discrete Element Method. Based on these findings, a sustainability analysis was undertaken using three hypothetical scenarios of application of MICP in pavement construction. The major conclusions from the study are summarised below.

• Among different MICP processes, urea hydrolysis showed the maximum chemical conversion efficiency (up to 90%) and CaCO₃ precipitation. Sporosarcina pasturii was found to be one of the most effective, efficient and widely used bacteria in urea hydrolysis.
• Application of MICP treatment has so far been focused on different types of sand, and unconfined compressive strengths of up to 19.6 MPa have been reported. This strength increases with the CaCO₃ content but has also been found to depend on the morphology of the CaCO₃, the location of precipitation, soil particle size and its distribution.
• The increases in the CaCO₃ content observed in a wide range of soils have generally resulted in reductions in soil permeability.
• Evaluation of previous triaxial tests on MICP-treated soils revealed that the dilative tendency along with the peak strength of soil significantly increased with increasing CaCO₃ content. At the same time, the brittleness of the soil samples also increased.
• A range of models of various levels of complexity have been proposed in the literature to capture the strength gain in soils due to CaCO₃ precipitation. Most of these models attempted to predict the unconfined compressive strength of soils as a function of the CaCO₃ content. However, the strength gain is understood to be influenced by other factors such as the soil’s particle size distribution, location of precipitation and morphology of precipitates. This limits the applicability of most models to the soil for which it was developed and possibly the treatment type used. Only one study attempted to take soil properties into account, but with limited success.
• Studies on constitutive modelling of the behaviour of MICP-treated soils have been limited to a handful of cases. A bounding surface plasticity model was shown to be able to capture some of the behavioural characteristics of cemented soils. Other studies incorporated the degradation of strength and stiffness with the degradation of CaCO₃ bonds and captured many of the features of cemented soil behaviour. However, there is scope for improvement.
• In the sustainability analysis, two of the three MICP treatment scenarios cost more and had a larger CO₂ footprint and thus may not be economically viable or environmentally sustainable technology, at least at the current time. Alternative sources of nutrients might lower the cost and environmental footprint, and a number of alternative solutions are currently being explored. For example, Achal et al. [179] showed that a by-product from the milk industry “lactose mother liquor” can be used as an alternative source of nutrients. Fang et al. [182] showed the potential of using waste products from the tofu industry. With further research, this technology can potentially be cost-effective as well as environmentally friendly, particularly since it uses natural processes. The use of biological and other waste effluents in the process will further enhance the environmental benefits in terms of reducing CO₂ and energy footprints.