Life Cycle Sustainability Assessment of Microbially Induced Calcium Carbonate Precipitation (MICP) Soil Improvement Techniques

Abstract

Microbially induced calcium carbonate precipitation (MICP) is a biomediated ground improvement technology that uses ureolytic bacteria to precipitate calcium carbonate minerals to improve the strength and stiffness of soils. MICP can be mediated by either augmented non-native or stimulated indigenous microorganisms, resulting in biocemented soils and generated aqueous ammonium (NH₄⁺) byproducts. Although the process has been extensively investigated, the fate and transport of generated NH₄⁺ byproducts has posed an environmental challenge and to date, their associated environmental impacts have remained poorly understood. In an effort to better quantify process impacts, a large-scale experiment was conducted involving three 3.7 m long soil columns, wherein three different ureolytic biocementation treatment approaches were employed. A life cycle sustainability assessment (LCSA) was performed to compare the environmental impacts and costs of these different MICP treatment approaches as well as evaluate the potential environmental benefits of NH₄⁺ byproduct removal using post-treatment rinsing. The objective of this paper is to present the results of the LCSA study. LCSA results suggest that when treatments are consistent with those performed in this study, stimulation can be more sustainable than augmentation, and the use of lower ureolytic rates can further reduce process environmental impacts by achieving greater spatial uniformity and extent of biocementation. The LCSA outcomes also illustrate tension between the environmental benefits afforded by NH₄⁺ byproduct removal and the life cycle impacts and costs associated with this removal. For the specific testing conditions, the injection of 1.8 pore volumes of rinse solutions to remove generated NH₄⁺ byproducts following biocementation was found to minimize environmental impacts; however, further refinement of such approaches will likely result from future field-scale applications.

1. Introduction

Unprecedented societal and environmental challenges such as those related to population growth, urbanization, poverty, climate change, and natural resource depletion have increased the need for creative and sustainable solutions for challenging civil and geotechnical infrastructure problems. Microbially induced calcium carbonate precipitation (MICP), or biocementation, is a biomediated technology that can improve the geotechnical engineering properties of soils by precipitating calcium carbonate (CaCO₃) on soil particle surfaces and at particle–particle contacts. Biocementation increases the shear strength and stiffness of soils with limited reductions in soil porosity and permeability [1,2,3,4]. MICP has been proposed for many different geotechnical and geoenvironmental applications, including soil improvement, the mitigation of soil liquefaction during earthquakes, erosion and scour prevention, rock fracture sealing, and contaminant immobilization, among other use cases [4,5,6,7,8,9].

MICP is facilitated by microorganisms that contain urease enzymes, which catalyze a hydrolysis reaction in the presence of urea, resulting in the generation of ammonia (NH₃) and carbonic acid (H₂CO₃) (Equation (1)). Generated ammonia participates in an equilibrium reaction with water and can produce ammonium (NH₄⁺) and hydroxide (OH⁻) in accordance with the surrounding solution pH (Equation (2)). The resulting hydroxide production encourages generated carbonic acid to deprotonate to become bicarbonate (HCO₃⁻) and carbonate (CO₃²⁻) species (Equation (3)). The addition of calcium (Ca²⁺) ions can then supersaturate aqueous solutions with respect to calcium carbonate minerals and initiate precipitation (Equation (4)).

$${\left({\mathrm{NH}}_2\right)}_2\mathrm{CO}+2{\mathrm{H}}_2\mathrm{O} \to 2{\mathrm{NH}}_3+{\mathrm{H}}_2\mathrm{C}{\mathrm{O}}_3$$

(1)

$${\mathrm{NH}}_3+{\mathrm{H}}_2\mathrm{O} \rightleftarrows {\mathrm{NH}}_4^{+}+{\mathrm{OH}}^{-}$$

(2)

$${\mathrm{H}}_2\mathrm{C}{\mathrm{O}}_3 \rightleftarrows {\mathrm{HCO}}_3^{-}+{\mathrm{H}}^{+} \rightleftarrows {\mathrm{CO}}_3^{-2}+{2\mathrm{H}}^{+}$$

(3)

$${\mathrm{Ca}}^{+2}+{\mathrm{CO}}_3^{-2} \rightleftarrows {\mathrm{CaCO}}_{3\left(\mathit{solid}\right)}$$

(4)

Establishing ureolytic activity in soils for MICP can occur either by augmentation of non-native bacteria or by the enrichment (i.e., stimulation) of native ureolytic microorganisms. Both approaches increase the soil’s capacity to hydrolyze urea for the MICP process by increasing ureolytic cell densities and/or ureolytic microbial activity. To date, augmentation has been primarily used to generate ureolytic activity [4,10,11,12]. However, researchers have recently shown that stimulation is effective in a diverse range of environments [13,14,15,16,17,18] and can improve the spatial extent and uniformity of MICP by stimulating in situ ureolytic communities already present in the soil pore space, thereby eliminating issues related to cell injections and filtration in porous media [19]. Stimulation may also offer significant advantages over augmentation through the elimination of costs and environmental impacts associated with ex situ bacterial cultivation and the introduction of non-native bacterial species.

For both stimulation and augmentation treatment approaches, the fate of produced NH₄⁺ byproducts remains a key environmental consideration for the deployment of MICP at field sites. If left untreated, generated NH₄⁺ concentrations may cause adverse environmental and human health impacts, thereby compromising other environmental benefits of the process. For example, when present in surface water, high NH₄⁺ concentrations can promote bacterial and algal bloom growth, thereby decreasing dissolved oxygen concentrations and producing toxins that can endanger aquatic life and other flora and fauna [20]. While the U.S. Environmental Protection Agency (EPA) has not established NH₄⁺ concentration limits for drinking water or freshwater aquifers, their recommendations for aquatic life state that NH₄⁺ concentrations should not exceed ≈1 mM and ≈0.1 mM for acute (1 h) and chronic (30 day) exposure, respectively [21]. Other recommended limits for total NH₄⁺ in surface water vary throughout the U.S. depending on location and governance; however, these limits generally range between ≈0.01 mM and ≈1.8 mM [22]. The European Union has also set a maximum total NH₄⁺ concentration for drinking water at ≈0.03 mM [23]. While acceptable NH₄⁺ concentrations following MICP will likely be site-specific, concentrations produced during biocementation will likely exceed such recommendations by several orders of magnitude, therefore necessitating removal and/or in situ remediation in order to comply with water quality requirements [24].

In past laboratory experiments and field studies, rinse solutions have been applied to treated soils to remove NH₄⁺ byproducts following MICP [19,24,25,26]. Although rinsing may require large quantities of materials and energy to remove NH₄⁺ byproducts at a field scale, it can be effective for NH₄⁺ byproduct removal in the absence of other proven remediation technologies. With the appropriate management of NH₄⁺ byproducts, the sustainability benefits afforded by MICP may be maximized and the process may become increasingly competitive relative to conventional geotechnical ground improvement technologies, which primarily use high mechanical energy and energy- and emissions-intensive materials, like Portland cement.

To date, the impacts associated with stimulated MICP treatment techniques and post-treatment rinsing, including environmental, social, and economic impacts, have not been quantified, although recent LCSAs have been performed for other biomediated soil improvement technologies such as enzymatically induced carbonate precipitation [27,28,29]. In an effort to bridge this gap in knowledge, this study conducts a life cycle sustainability assessment (LCSA) that compares the environmental (i.e., resource depletion, acidification, eutrophication, ecotoxicity, human health impacts, ozone depletion, and smog formation), social, and economic impacts of different ureolytic MICP treatment approaches applied at a meter-scale and evaluates the potential benefits of post-treatment NH₄⁺ byproduct removal using rinsing. The employed MICP treatment techniques included the augmentation of soils with Sporosarcina pasteurii (S. pasteurii), a well-studied ureolytic bacterium, and the stimulation of indigenous ureolytic microorganisms. The methods, models, and data used in the study were informed by the results of meter-scale biocementation experiments on poorly graded natural sands [19,24]. Using the obtained LCSA results, the objective of this paper is to discuss the implications for field-scale applications of MICP and present sustainability-oriented guidance for future research and development.

2. Experimental Setup

Three 3.7 m long horizontal soil columns (Figure 1) were improved using different ureolytic biocementation treatment approaches to investigate the spatial uniformity and extent of MICP, as well as the success of post-treatment NH₄⁺ byproduct removal as a function of different soil materials and treatment techniques. Although additional experimental details are provided in [19,24]’s works, a summary of all relevant details is provided here. All columns contained concrete sand, a clean, poorly graded alluvial sand (D₁₀ = 0.23 mm, D₃₀ = 0.54 mm, D₆₀ = 1.54 mm, fine content = 1.1%, e_min = 0.35, e_max = 0.60) that has been used extensively in past MICP experiments [17,19,24]. Each column contained 0.14 m³ of concrete sand that was prepared to similar initial porosities (between 0.30 and 0.32), relative densities (between 56% and 67%), and had pore volumes (PVs) between 48.5 L and 50.7 L (average value of 49.9 L). Columns received different biological treatment injections intended to either enrich soils for native ureolytic microorganisms (Columns 1, 2) or augment soils with S. pasteurii bacteria (Column 3) uniformly across column lengths. To examine the effect of urea hydrolysis rates on the spatial uniformity and extent of biocementation, different treatment techniques were applied to either achieve a high ureolytic rate with stimulation (Column 1) or augmentation (Column 3) or achieve a low ureolytic rate using stimulation (Column 2). Differences in ureolytic rates were achieved by either controlling the augmented cell densities (Column 3) or varying the applied yeast extract concentrations during enrichment injections (Columns 1, 2), which altered enriched ureolytic cell densities and reaction rates. Supplementary Table S1 presents the treatment schemes, including solution chemical constituents and concentrations, numbers of injections, and injection volumes and durations. All treatment solutions were injected using small electric pumps (Wayne Inc., Harrison, OH, USA, 0.1 HP). In summary, all columns received treatment solutions over three treatment phases, namely (1) stimulation or augmentation, (2) cementation, and (3) NH₄⁺ rinsing. Following enrichment treatments in Columns 1 and 2, however, a flush solution was applied immediately before the first cementation treatment to remove aqueous carbonate species expected after enrichment. Once ureolytic activity was established, nine cementation injections containing identical calcium and urea concentrations (250 mM) were applied to the columns over nine days in the high-rate columns (Column 1, 3) and over eighteen days (one injection every two days) in the low-rate column (Column 2). Shear wave velocity (V_s) measurements were obtained using bender element sensors placed at four different locations that were 0.31 m (V_s1), 1.33 m (V_s2), 2.35 m (V_s3), or 3.37 m (V_s4) from the injection source (Figure 1). Vs measurements were conducted at all locations before and after all cementation injections and allowed for the non-destructive monitoring of biocementation progression in time and with distance from the injection location. Initial Vs values at the four different measurement locations in all three columns had an average value of 204 m/s (standard deviation of 37.4 m/s) prior to starting all treatments. Following all cementation injections, 10.5 PV (525 L) of a high-pH and high-ionic-strength rinse solution was applied to each column to remove produced NH₄⁺. During rinse injections, aqueous NH₄⁺ concentrations were measured at four sampling port locations in each column (labeled P₁ through P₄ in Figure 1) to evaluate NH₄⁺ byproduct removal with the effects of rinsing on biocementation integrity, also assessed using Vs measurements. After rinsing, physical soil samples were collected at various locations along columns to estimate sorbed NH₄⁺ masses remaining on soil particle surfaces [19,24].

3. Methods

A life cycle sustainability assessment (LCSA) is a method to evaluate the environmental, economic, and social impacts/benefits of a product/system over its entire life cycle (i.e., from “cradle to grave”). An LCSA typically integrates an environmental life cycle assessment, life cycle costing, and social life cycle assessment. An environmental life cycle assessment (LCA) characterizes, quantifies, and interprets a product’s/system’s environmental impacts across the entire supply chain, from the extraction of raw materials to the end of life [30]. Life cycle cost analysis (LCCA) and social life cycle assessment (S-LCA) complement the environmental LCA by evaluating the relevant costs and social/socioeconomic impacts to stakeholders that are incurred over the equivalent life cycle defined in the LCA. LCA methods have been standardized in the International Organization for Standardization (ISO) standards [31,32,33,34]. Similarly, many standards and guidelines exist for the LCCA of civil infrastructure and the built environment [34,35]. The guidelines for the S-LCA of products are less developed and not as widely used [36]. In place of a formal S-LCA, this study estimates social damage costs associated with greenhouse gas (GHG) emissions (e.g., carbon dioxide, methane, and nitrous oxide) to communicate one aspect of the potential social impacts of MICP.

3.1. Goal and Scope Definition

The primary goals of this study were twofold, namely (1) to compare the life cycle impacts and costs of the different MICP treatment techniques investigated during the meter-scale experiment (i.e., high ureolytic rate stimulation, low ureolytic rate stimulation, and high ureolytic rate augmentation) and (2) to evaluate the potential environmental benefits of post-treatment NH₄⁺ removal using rinsing. The scope includes the environmental flows (inputs and outputs) associated with bacterial cultivation (for augmentation only), enrichment/stimulation or augmentation treatments, pre-cementation flushing (for enrichment only), cementation treatments, and NH₄⁺ rinsing. The study excludes any subsequent remediation of collected effluent that would be completed ex situ at water reclamation facilities. Figure 2 illustrates the process flow diagrams for MICP via stimulation and augmentation.

To conduct a meaningful and robust comparative assessment of the MICP treatment approaches, the systems under comparison must be functional equivalents (i.e., they must have the same functionalities and technical characteristics) [31,32,37,38]. In an LCSA, a “functional unit” is used to describe and quantify the identified function(s) and performance characteristics of the studied system or systems. The functional unit is a reference unit for the LCSA results and provides a basis to compare different systems [31,32]. For this study, the functional units considered were the improvement of 0.14 m³ of soil through biocementation to achieve specific shear wave velocity increases at two different treatment distances (i.e., ΔV_s = 150 m/s at locations 2 and 4). The LCSA results are evaluated at multiple discrete performance criteria (i.e., vs. values) to further inform comparisons between approaches.

3.2. Life Cycle Inventory Development

A life cycle inventory (LCI) of the environmental flows, including inputs (e.g., raw materials and energy) and outputs (e.g., solid wastes, coproducts, or other releases, as well as emissions to air, water, and soil), was developed for each treatment approach over its life cycle [31,32].

In an LCSA, a product system can be divided into two subsystems, the foreground system and the background system. In the context of MICP research and development, the foreground system consists of those processes which are under the direct control of the researchers (e.g., solution constituents and treatment volumes, which in turn affect energy consumption during pumping). The background system consists of the supply chains on which the foreground system relies but over which researchers do not have influence (e.g., production and delivery of electricity). The primary distinction lies in the way the LCI data are compiled. When possible, the foreground system is described by primary data based on the actual processes and their operating conditions, while the background activities are typically characterized using reference LCI datasets and often represent average industry data [39].

3.2.1. Foreground Data

Primary data for modeling each treatment approach were collected during the experiment, and secondary data obtained from the published literature were supplemented where necessary. Data were measured or estimated to model the life cycle impacts of (1) materials and energy use during the culturing of S. pasteurii, (2) the use of chemical reagents, water, and other materials during stimulation/augmentation, flushing (in stimulated Columns 1 and 2 only), cementation, and NH₄⁺ rinsing treatments, (3) pumping time and energy use during each treatment phase, and (4) MICP byproducts (e.g., aqueous NH₄⁺, unhydrolyzed urea, and sorbed NH₄⁺) as a function of injected rinse solution volume. The input data are summarized in

Table 1. Abridged LCI results.

	Column 1: Stimulation (High Ureolytic Rate)		Column 2: Stimulation (Low Ureolytic Rate)		Column 3: Augmentation
	C1	C1 *	C2	C2 *	C3	C3 *
Inputs
Materials
Calcium chloride (kg)	1.05 × 10	1.24 × 10	7.42	9.41	1.76 × 10	1.96 × 10
Urea (kg)	7.02	7.02	5.38	5.38	9.55	9.55
Ammonium chloride (kg)	2.73	2.73	2.66	2.66	4.25 × 10−1	4.25 × 10−1
Ammonium sulfate (kg)	--	--	--	--	4.57 × 10−2	4.57 × 10−2
Sodium acetate (kg)	5.24	5.24	4.61	4.61	3.68	3.68
Sodium chloride (kg)	--	--	--	--	4.79	4.79
Sodium hydroxide (kg)	5.81 × 10−1	5.81 × 10−1	5.81 × 10−1	5.81 × 10−1	--	--
Tris base (kg)	--	--	--	--	7.20 × 10−2	7.20 × 10−2
Yeast extract (kg)	1.66 × 10−1	1.66 × 10−1	2.35 × 10−2	2.35 × 10−2	9.14 × 10−2	9.14 × 10−2
S. pasteurii (cells)	--	--	--	--	7.75 × 1012	7.75 × 1012
Water (L)	9.06 × 102	9.96 × 102	7.96 × 102	8.86 × 102	1.17 × 103	1.26 × 103
Energy
Electricity (MJ)	3.08 × 10	3.24 × 10	2.70 × 10	2.87 × 10	5.29 × 10	5.45 × 10
Outputs
Inorganic emissions to air (kg)
Ammonia (NH3)	1.52 × 10−2	1.77 × 10−2	1.14 × 10−2	1.38 × 10−2	2.23 × 10−2	2.47 × 10−2
Carbon dioxide (CO2)	3.66 × 10	3.82 × 10	3.03 × 10	3.19 × 10	4.28 × 10	4.44 × 10
Carbon monoxide (CO)	2.71 × 10−2	2.93 × 10−2	2.20 × 10−2	2.42 × 10−2	3.37 × 10−2	3.59 × 10−2
Methane (CH4)	1.23 × 10−1	1.27 × 10−1	1.01 × 10−1	1.05 × 10−1	1.39 × 10−1	1.42 × 10−1
Nitrogen monoxide (NO)	6.95 × 10−3	1.78 × 10−3	6.67 × 10−3	1.06 × 10−3	6.94 × 10−3	2.66 × 10−3
Nitrogen oxides (NOx)	5.37 × 10−2	5.70 × 10−2	4.45 × 10−2	4.78 × 10−2	6.44 × 10−2	6.77 × 10−2
Nitrous oxide (N2O)	1.52 × 10−2	5.63 × 10−3	1.42 × 10−2	3.80 × 10−3	1.58 × 10−2	7.95 × 10−3
Sulfur dioxide (SO2)	7.48 × 10−2	8.06 × 10−2	6.28 × 10−2	6.86 × 10−2	8.19 × 10−2	8.77 × 10−2
Particulates to air (kg)
Dust (PM2.5)	1.71 × 10−2	1.84 × 10−2	1.48 × 10−2	1.60 × 10−2	1.44 × 10−2	1.57 × 10−2
Dust (PM10)	1.25 × 10−2	1.36 × 10−2	1.05 × 10−2	1.17 × 10−2	1.23 × 10−2	1.34 × 10−2
Inorganic emissions to water (kg)
Ammonia (NH3)	6.87 × 10−2	6.87 × 10−2	6.22 × 10−2	6.23 × 10−2	3.73 × 10−2	3.74 × 10−2
Ammonium (NH4+)	4.12 × 10−1	1.03 × 10−1	3.96 × 10−1	6.09 × 10−2	4.11 × 10−1	1.56 × 10−1
Nitrate (NO3−)	4.00 × 10−1	1.09 × 10−1	3.82 × 10−1	6.64 × 10−2	4.01 × 10−1	1.60 × 10−1
Nitrogen (N)	1.30 × 10−3	1.53 × 10−3	9.56 × 10−4	1.18 × 10−3	2.02 × 10−3	2.24 × 10−3
Phosphate (PO43−)	1.83 × 10−2	2.02 × 10−2	1.51 × 10−2	1.70 × 10−2	1.85 × 10−2	2.04 × 10−2

* Results include optimal rinsing (i.e., injection of 1.8 PV = 90 L).

, as well as in Supplementary Table S1.

3.2.2. Background Data

Secondary LCI data were identified and collected to model the environmental flows associated with a variety of background processes, including the production of chemical reagents (e.g., calcium chloride and urea), water, and electricity (Supplementary Table S2). No LCI data were available for sodium acetate production; therefore, an LCI dataset was created from reference datasets for sodium hydroxide, acetic acid, and deionized water using mass-based allocation. Similarly, no LCI data were available for Tris base (NH₂C(CH₂OH)₃) and yeast extract production; however, impacts from these materials were assumed negligible due to the small quantities used (Table 1). U.S.-based LCI data were used when available and supplemented with European datasets when necessary. For example, no data were available to model calcium chloride production in the U.S., which is typically accomplished by refining natural brines. Instead, calcium chloride derived from the Solvay process (i.e., the major industrial process for the production of soda ash) was used. This European dataset is expected to moderately overestimate impacts from calcium chloride production. All reference LCI datasets were collected from the GaBi Professional database (2018) and the ecoinvent database (2018), accessed through GaBi ts 8 software [40,41].

3.2.3. MICP Process Emissions

MICP has direct and indirect process emissions associated with its application to soils and produced byproducts, including aqueous NH₄⁺, unhydrolyzed urea, and sorbed NH₄⁺. MICP byproducts and associated process emissions were estimated from the obtained experimental data (see “Estimating MICP By-products” section in the Supplementary Materials) [19,24] and following the recommendations of the other published literature [42,43]. Estimates of in situ MICP byproduct masses as a function of injected rinse solution volumes and rinsing time (rinse injection flow rate = 0.75 L/min) are provided in Figure 3 and tabulated in Supplementary Table S3.

Following MICP treatments, generated in situ nitrogen (N) remaining within soils (in the form of aqueous NH₄⁺, unhydrolyzed urea, and sorbed NH₄⁺) may contribute to air and water pollution through various environmental mechanisms or impact pathways. These include the potential for (1) the volatilization of produced nitrogen species as ammonia gas (NH₃), (2) the generation of nitrous oxide (N₂O), nitric oxide (NO), and nitrogen gas (N₂) emissions resulting from the potential aerobic oxidization of NH₄⁺ to nitrate (NO₃⁻) and subsequent anaerobic reduction of NO₃⁻, and (3) aqueous transport and runoff/leaching of N, mainly as NH₄⁺ and NO₃⁻ [43]. Although the fate and transport of NH₄⁺ byproducts following MICP remain uncertain and are likely site-specific, assumptions were made to approximate these emissions and their environmental impacts. For conservatism, it was assumed that the mass of total MICP byproducts (Figure 3b) would eventually become available as aqueous NH₄⁺. Thus, it was assumed that all remaining urea (when present) would eventually be hydrolyzed to form NH₄⁺ and all sorbed NH₄⁺ would desorb and exist in free soil solution. Thirty percent of this aqueous NH₄⁺ was assumed to nitrify to produce NO₃⁻ [43]. Direct N₂O and NO emissions due to the denitrification of this NO₃⁻ were assumed to be 1.1% and 0.7% of in situ N, respectively [42]. Direct N₂ emissions were assumed to be 8.9% of in situ N, following studies which found that the N₂O:N₂ production ratio during denitrification in natural soils is approximately 0.12 [44]. The remaining NO₃⁻ that was not denitrified was assumed to leach into groundwater. All remaining NH₄⁺ (approximately 59.3% of in situ N) was assumed to remain in the soil pore space or leach into groundwater. Indirect N₂O emissions from N losses through leaching and/or runoff were assumed to be 0.75% of leached NO₃⁻ and NH₄⁺ [43]. It was assumed that no emissions were released via NH₃ volatilization as MICP was applied to saturated subsurface soils and post-treatment pH values were near 7.5, resulting in a minimal speciation of total NH₄⁺ as NH₃. These allocations are best estimate values based upon current knowledge, but the assumed fate of generated N species may be further refined following future chemical measurements during and after MICP treatments.

3.3. Life Cycle Impact Assessment

A life cycle impact assessment (LCIA) was performed to translate the LCI data into indicators of the environmental and human health impacts associated with each treatment approach [31,32]. The impact categories considered in this study include resource depletion (e.g., primary energy and water consumption), climate change (e.g., global warming potential), acidification, eutrophication, ecotoxicity, human health impacts (e.g., from respiratory inorganics, carcinogens, and noncarcinogens), ozone depletion, and smog formation. Global warming potential was modeled using the 100-year global warming potential values published in the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [45]. All other indicator values were modeled using the factors in version 2.1 of the tool for the reduction and assessment of chemical and other environmental impacts (TRACI), which is a midpoint-level LCIA method developed by the U.S. EPA [46].

3.4. Life Cycle Costing

A life cycle cost analysis (LCCA) was performed to evaluate the project costs involved with each treatment approach. The study includes the purchase costs associated with materials and energy use over the life cycle illustrated in Figure 2 and excludes any costs associated with the use and/or maintenance of laboratory equipment. The results also exclude other costs that would occur in a field-scale implementation of MICP (e.g., material transportation, equipment mobilization and use, construction consumables, effluent management, and labor costs). However, these additional project costs are expected to be comparable between all of the MICP treatment approaches employed.

The LCCA uses unit costs for laboratory-grade chemicals and does not reflect the economies of scale that will be achieved at field scale when using large quantities of commercially sourced industrial grade chemicals. The tracking of materials and energy used during future field trials is needed to more accurately estimate these efficiencies for future field applications; however, at present, a detailed analysis of the data obtained from the performed meter-scale columns provides the best resource for obtaining an initial evaluation and establishing a baseline scenario. Cost data for chemical reagents and other miscellaneous materials (e.g., yeast extract) were obtained from Avantor Performance Materials, Inc. (Radnor, PA, USA) and Thermo Fisher Scientific, Inc. (Waltham, MA, USA). The average cost of water and electricity in the United States was assumed to be USD 0.002 per L and USD 0.03 per MJ, respectively [47,48]. The unit costs used in this study are listed in Supplementary Table S4.

3.5. Social Cost of Greenhouse Gas Emissions

The social cost of GHG emissions was quantified to represent the potential social impacts of each MICP treatment approach. The social costs of carbon (SC-CO₂), methane (SC-CH₄), and nitrous oxide (SC-N₂O), which are monetized damage estimates associated with incremental increases in emissions in a given year, were computed and summed using the methods and models employed by the U.S. EPA and other federal agencies [49,50]. The integrated assessment models (IAMs) used to develop the social cost estimates combine climate processes, economic growth, and feedback between the two to translate climate impacts to economic damages. To account for uncertainty within the IAMs, the Interagency Working Group provides four estimates based on different discount rates for the social cost of each pollutant in any given discount year (i.e., the year pollutants are emitted) [49]. In this study, the social costs for an average discount rate of 3% in 2020 were used. These estimates are not comprehensive (i.e., they do not currently include all of the important physical, ecological, and economic impacts of climate change); however, they are a useful measure to assess the climate impacts of GHG emissions in terms of socioeconomic damages.

4. Results

The LCSA results are first presented without NH₄⁺ rinsing to compare the impacts of the three considered MICP treatment approaches at different magnitudes of biocementation improvement, as captured by shear wave velocity values. Following this comparison, the impacts and benefits afforded by NH₄⁺ byproduct removal using rinsing are examined. Finally, the LCSA results corresponding to specific improvement levels are presented, considering both the presence and absence of NH₄⁺ byproduct removal injections in order to examine the life cycle impacts and costs of each MICP treatment approach in more detail. Unless otherwise specified, all results presented reflect the life cycle impacts and costs associated with the improvement of the 0.14 m³ volume of soil that was present within a single 3.7 m long column.

4.1. Effects of Increasing Cementation Treatments

Figure 4 illustrates the increases in global warming potential associated with obtaining higher Vs values and Vs increases (ΔV_s) in all three columns. Soil Vs increases were measured during the application of successive cementation treatments, which ranged from 0 to 9 injections during the experiment. Results are shown for Vs improvements measured at locations 2 (V_s2, Figure 4a) and 4 (V_s4, Figure 4b), which were located 1.33 m and 3.37 m from the injection source, respectively, as shown in Figure 1. All columns received treatment injections in a single direction in this experiment and were expected to be representative of a single injection-to-outlet well alignment. It should be mentioned, however, that during field-scale applications, treatment wells will likely be placed in more complex two-dimensional patterns and treatments may also be applied bidirectionally, thereby affording the potential for larger well spacings to be used. An example of a possible two-dimensional triangular well pattern with the same injection-to-outlet well distances present between the injection source and Vs locations 2 and 4 in the meter-scale column experiments is shown in Supplementary Figure S1.

As evident in the experimental results, increasing the number of applied cementation treatments increased Vs due to the higher magnitudes of biocementation that were achieved. However, the level of improvement varied between treatment approaches and with distance from the injection well. The high-ureolytic-rate stimulation approach (Column 1) obtained more rapid and localized improvement near the injection well, resulting in large increases in V_s2 (greater than 1000 m/s) but minimal increases in V_s4 (only 168 m/s) after nine cementation injections. The low-ureolytic-rate stimulation approach (Column 2) achieved smaller Vs improvements than Column 1 near the injection source but achieved the greatest uniformity and extent of biocementation when compared to the other two approaches, increasing V_s2 and V_s4 values by 698 m/s and 406 m/s, respectively. In the augmented column (Column 3), small increases in V_s2 and V_s4 values of 72 m/s and 166 m/s were measured, respectively, reflective of more localized improvement near the injection well and poor ureolytic activity and biocementation at larger treatment distances [19].

Increasing the number of cementation treatments required more material and energy consumption and thus increased environmental impacts and cost. While global warming potential is the only impact indicator presented in Figure 4, all other indicators exhibited similar trends with increasing cementation treatments. For all columns, each additional cementation injection increased total global warming potential linearly by about 5 kg CO₂ equivalent (eq.) but had nonlinear effects on the realized engineering improvements, as indicated by V_s.

To appropriately compare the life cycle impacts and costs of the three MICP treatment approaches, each column must achieve the same level of biocementation. In this study, the LCSA results were compared for two different performance criteria, namely obtaining a ΔV_s increase of 150 m/s at locations 2 and 4. While this change in shear wave velocity is relatively small in comparison to the Vs improvements observed at shorter treatment distances (i.e., location 1), recent centrifuge modeling and laboratory element testing have shown that biocementation-induced Vs increases near 100 m/s can significantly increase liquefaction triggering resistances. Since ΔV_s increases of exactly 150 m/s were not measured during the treatment program, the number of cementation treatments required for each column to meet this performance criteria at the two considered locations was estimated using linear interpolation. In order to obtain a ΔV_s of 150 m/s at location 2, Columns 1, 2, and 3 required 0.76, 1.39, and greater than 9 cementation injections, respectively. However, when considering this same improvement level (ΔV_s = 150 m/s) for location 4 at the larger treatment distance, Columns 1, 2, and 3 required 4.98, 3.54, and 8.41 cementation injections, respectively. Though not necessarily feasible nor practical during field implementation, partial cementation injections were modeled in the LCSA to impose functional equivalence of the systems under comparison. The LCSA results for each MICP treatment approach were reevaluated using the partial cementation injection values above with the following outcomes: (1) When ΔV_s2 = 150 m/s, the global warming potentials of Columns 1 and 2 are near 24 and 27 kg CO₂ eq., respectively. Column 3 failed to meet this performance criterion even after nine cementation injections. However, by extrapolation, the global warming potential for Column 3 would be expected to exceed 55 kg CO₂ eq (over twice that of Columns 1 and 2). (2) When ΔV_s4 = 150 m/s, the global warming potentials of Columns 1, 2, and 3 are approximately 44, 37, and 51 kg CO₂ eq., respectively. These results suggest that when improvements at smaller injection-to-outlet well spacings are targeted, MICP using the high ureolytic rate stimulation approach may be the more sustainable method, while the low ureolytic rate stimulation approach may be more sustainable for larger injection-to-outlet well spacings. In both cases, augmented MICP is the least sustainable approach due to limited improvement overall. This outcome may be in part due to the specific augmentation procedure employed by [19] and merits further investigation. For field-scale applications, larger injection-to-outlet well spacings are generally preferred over smaller well spacings, as process life cycle impacts and costs will increase with quantities of required treatment wells. However, when large increases in Vs (greater than 400 m/s) are required and/or soils have lower initial hydraulic conductivities, smaller well spacings may be necessary to ensure the effective delivery of reactants.

4.2. Impacts and Benefits of Ammonium Rinsing

In the experiment, alkaline calcium-rich rinse solutions were injected in columns following biocementation to remove MICP byproducts. However, this post-treatment rinsing technique requires significant energy and materials, namely water and calcium chloride. This implies a tradeoff between the environmental benefits of NH₄⁺ by-product removal and the life cycle impacts and costs of the rinsing technique.

To investigate this tradeoff, the LCSA results were evaluated as a function of the byproduct removal that occurs with increasing rinsing effort (i.e., increasing injected rinse solution volume and rinsing injection time). Figure 5 shows the changes in indicator values as a result of rinsing relative to their values at the end of cementation (i.e., before rinsing). As shown, high masses of in situ nitrogen byproducts (near 450 g of N) are present immediately following biocementation, with reductions in these values occurring in all columns with increases in applied rinse injections. With increasing rinsing effort, all environmental impacts and cost indicators followed one of the following three observed trends: (1) global warming potential, smog formation potential, and social cost decreased slightly and reached a minimum value after injecting between 0.9 and 1.8 PV (45 to 90 L) of rinse solution but subsequently increased as rinsing became less effective at removing byproducts at low aqueous NH₄⁺ concentrations; (2) eutrophication potential progressively decreased and reached a minimum value after injecting nearly 5.4 PV (270 L) of rinse solutions; or (3) all other indicators increased minimally at first and then increased more substantially as rinsing became less effective after injecting about 5.4 PV (270 L) of rinse solution.

As with any tradeoff, an optimal solution is one that best satisfies the objective functions. Although reductions in nitrogen byproducts from initial values ≈450 g N to ≈20 g N in all columns resulted in near linear reductions in eutrophication potential, global warming potential illustrated a more complex behavior, achieving a minimum as nitrogen byproducts decreased to between ≈50 g N and ≈100 g N but with large increases below ≈20 g N. This behavior illustrated the tension between the benefits afforded by rinsing, which can effectively remove large nitrogen byproduct masses at the start of rinsing, and the drawbacks of rinsing observed at larger injection volumes (lower nitrogen byproduct concentrations), wherein a more limited removal of byproducts occur. At these low nitrogen concentrations, large increases in costs and impacts are observed, despite likely being practically necessary to meet site water quality requirements. Considering all impact categories, these trends collectively suggest that injecting between 1.8 and 3.6 PV (90 L to 180 L) of rinse solution, thereby reducing nitrogen byproducts to between 4% (18 g N) and 38% (171 g N) of their initial values in each column, can minimize the life cycle impacts and costs associated with post-treatment rinsing. In all further comparisons considered in this study, the injection of 1.8 PV (90 L) of rinse solution will be considered the optimal scenario, which minimizes process impacts.

However, as mentioned previously, site regulations may necessitate further removal to achieve nitrogen byproduct concentrations below what can be obtained with this optimal strategy. At these larger injection volumes, results from the meter-scale experiments demonstrate that injecting 10.5 PV (525 L) of rinse solution can reduce total in situ N by over 97.5% in all columns, resulting in average aqueous NH₄⁺ concentrations of 3.5 mM, 1.5 mM, and 3.8 mM NH₄⁺ for Columns 1 through 3, respectively, albeit at the expense of substantially larger process impacts. These post-rinsing concentrations are slightly higher but on the same order of magnitude as those which may be required at many sites throughout the U.S. [22]. However, if site-specific regulations require further reductions in NH₄⁺ concentrations, recent studies have also shown that aqueous NH₄⁺ concentrations can be further reduced to values near and below 1 mM through modifications to rinsing techniques including the use of stop-flow rinse injections and the inclusion of potassium ions [51] as opposed to the continuous calcium-based rinse injections considered in this study. Furthermore, future studies considering alternative remediation methods such as in situ microbial nitrogen transformations, selective absorption, and electrokinetic transport may provide new viable methods to achieve more effective and efficient nitrogen byproduct removal [52,53], with results from this study providing a baseline scenario against which the success of alternative strategies could be assessed.

4.3. Comparison of MICP Treatment Approaches

In this section, the LCSA results for each MICP treatment approach are further examined for the case when ΔV_s4 = 150 m/s, indicative of improvement at a distance of 3.37 m from the injection well. Results are presented in Table 1 and Figure 6 and consider two scenarios, namely (1) the performance of MICP followed by no NH₄⁺ rinsing and (2) the performance of MICP followed the use of the optimal rinsing technique identified earlier, namely the injection of 1.8 PV of rinse solutions. When computing the results with the optimal rinsing strategy, it was assumed that the mass of in situ MICP byproducts (Figure 3b) would be independent of the number of cementation treatments applied and that removal depends only on the injected rinse solution volume [51]. This assumption is reasonable because the sum of aqueous NH₄⁺ and unhydrolyzed urea remaining in the soil pore space after cementation depends only on the concentration of applied urea, which was held constant at 250 mM for all cementation injections. Though the mass of sorbed NH₄⁺ was based upon measurements obtained after nine cementation injections, given the high aqueous NH₄⁺ concentrations present after all treatments (≈500 mM) and the relatively small cation exchange capacity (CEC) of concrete sand (CEC = 2.58 meq/100 g of soil), it is expected that these sorbed NH₄⁺ values are reasonable estimates for all previous cementation injections.

Using the LCSA results in Table 1 and Figure 6, the MICP treatment approaches can be compared on the basis of environmental impact and cost. The potential impacts associated with the low-ureolytic-rate stimulation approach (Column 2) are less than those of the high-ureolytic-rate stimulation approach (Column 1) by an average of 16% across all environmental impact and cost indicators, excluding human toxicity and ozone depletion potentials. The human toxicity (i.e., noncancerous and cancerous) and ozone depletion potentials are very small and may be considered negligible for all columns. Compared to the augmented approach (Column 3), the potential impacts of the low-ureolytic-rate stimulation approach (Column 2) are lower by about 20% on average. The potential impacts of the high-rate augmented approach (Column 3) exceed those of the high-ureolytic-rate stimulation approach (Column 1) for all impact categories by about 5% on average, except water consumption, acidification potential, and particulate potential. Without rinsing, Columns 1, 2, and 3 cost approximately USD 1760, USD 1470, and USD 1800, respectively. These costs (per 0.14 m³ of treated soil) reflect the use of laboratory-grade chemicals and do not consider the economies of scale that would be achieved at a field scale when using larger volumes of industrial-grade chemicals. Hence, these monetary costs are expected to be substantially refined following data obtained from future field trials, while the environmental impact observations will likely change only modestly. The social cost for each column is approximately 0.14% of the calculated life cycle cost and ranges from USD 1.87 to USD 2.71.

The impacts and benefits associated with using the optimal rinsing technique were also similar for all columns. Eutrophication potential was greatly reduced from 50% to 67% after injecting 1.8 PV (90 L) of rinse solution. Rinsing also decreased global warming potential, smog formation potential, and social cost by an average of 2%, 3%, and 4%, respectively. All other environmental indicators (excluding human toxicity and ozone depletion potentials) were increased by 6% on average. Treatment costs increased by only USD 12 (<1%) due to the additional materials and energy required for rinsing.

The LCSA results were also examined to identify environmental and economic “hotspots” (i.e., life cycle stage(s), process(es), and flow(s) with large contributions to one or several impact categories) [54]. Figure 6 shows that for Columns 1 and 2, the stimulation and cementation phases of MICP treatment are the primary contributors to impacts for all indicators, except eutrophication potential. Across these impact categories (and excluding eutrophication), the stimulation phase represents approximately 41% of total impacts for Column 1 and 48% of total impacts for Column 2. Similarly, the cementation phase is responsible for about 52% and 44% of total impacts for Column 1 and 2, respectively. For Column 3, most impacts stem from the cementation phase, which accounts for approximately 88% of total impacts. This is due to the large number of cementation injections required for the augmented column to meet the defined performance criterion (ΔV_s4 = 150 m/s). The bacterial cultivation and augmentation phases together represent between 0.1% and 15% of the total impacts.

For all columns, the production of materials and energy is responsible for about 80% and 15% of total global warming potential, respectively. The remaining 5% of the global warming potential results from process emissions related to the generation of in situ nitrogen byproducts. Materials production accounts for over 99% of ecotoxicity, human toxicity, and ozone depletion potentials and over 73% for all other indicators, excluding eutrophication potential and life cycle cost. Eutrophication impacts for all three columns can be primarily attributed to process emissions, which represent approximately 79% to 82% of the total impact prior to rinsing and 37% to 61% after rinsing, with the balance largely stemming from materials production. Less than 1% of the eutrophication potential is caused by electricity consumption during pumping. Regarding economic impacts, materials account for 99.9% of the cost, while electricity use represents only 0.1%.

The most impactful (and costly) materials are urea, calcium chloride, ammonium chloride, and sodium acetate, in descending order of impact. Urea production accounts for approximately 34% of total global warming potential and when considered along with MICP, process emissions is responsible for nearly 40% of the total global warming potential. Calcium chloride production represents about 20% of the total global warming potential due to the large quantities used. Though smaller quantities of ammonium chloride and ammonium sulfate are needed for MICP, the production of these chemicals is resource-intensive and environmentally impactful, requiring 38 MJ and 26 MJ of primary energy and emitting 2.4 kg and 1.9 kg CO₂ eq. per kilogram. Together, these materials comprise 85% to 91% of the total cost. The cost of urea alone accounts for 38% to 56% of the total cost; however, future assessments can use commercially sourced industrial grade chemicals, likely enabling further refinement of these estimates.

While these results do not consider the impacts of effluent management, the volume of generated effluent (and thus impact/cost of effluent management) would likely be greatest for the augmented column, which required more cementation treatments to meet the defined performance criterion.

5. Discussion

5.1. Implications for Field-Scale Implementation

The LCSA results presented above, which model the impacts of meter-scale column testing performed in the laboratory, were upscaled to estimate the environmental impacts of MICP for two field-scale implementation scenarios (see “Methods for Upscaling LCSA Results” Section in the Supplementary Materials). The upscaled scenarios assume that solution transport in these treatment volumes would be similar to that observed in the performed one-dimensional column experiments. While this may not be strictly true, the scenarios provide important practical comparisons through which the relative differences between the studied treatment approaches can be better understood.

Scenario 1 considers the impacts of MICP treatment at a hypothetical project site consisting of loose saturated sandy soils that require the mitigation of earthquake-induced soil liquefaction. The zone of improvement is approximately 30 m × 30 m × 11 m deep (100 ft × 100 ft × 35 ft). The functional unit is defined as the improvement of approximately 9900 m³ of soil such that shear wave velocity is increased by 150 m/s (i.e., ΔV_s = 150 m/s). Scenario 1 assumes a two-dimensional triangular well configuration for all three MICP treatment approaches, with an injection-to-outlet well spacing of 3.5 m, similar to the treatment distance present at Vs location 4 (Supplementary Figure S1). In this scenario, MICP treatment is followed by the use of the optimal rinsing technique described earlier (i.e., injection of 1.8 PV of rinse solution). The results for primary energy consumption and global warming potential are shown in

Table 2. LCSA results for Scenarios 1 and 2.

	Column 1: Stimulation (High Ureolytic Rate)	Column 2: Stimulation (Low Ureolytic Rate)	Column 3: Augmentation
Scenario 1 (ΔVs = 150 m/s)
Primary Energy (MJ)	4.52 × 107	3.74 × 107	4.76 × 107
Global Warming Potential (kg CO2 eq.)	2.61 × 106	2.14 × 106	2.84 × 106
Scenario 2 (ΔVs = 400 m/s)
Primary Energy (MJ)	7.82 × 107	6.63 × 107	--
Global Warming Potential (kg CO2 eq.)	4.55 × 106	3.79 × 106	--

. For this scenario, the low-ureolytic-rate stimulation approach (Column 2) is the most sustainable. Relative to low-ureolytic-rate stimulation, the impacts of high-ureolytic-rate stimulation and augmentation are about 21% and 30% greater on average, respectively. One limitation of these findings is the limited scope of analysis; results consider impacts from materials use and process emissions but exclude impacts associated with energy consumption from pumping and several other life cycle phases, including equipment mobilization, materials transportation, well construction, and effluent management. It is expected that the impacts from these latter three categories would significantly decrease for the spacing of 3.5 m (Scenario 2) compared to the spacing of 1.3 m (Scenario 1), showing a further benefit of Scenario 2. These exclusions are the result of a lack of data and cannot be estimated by simply scaling laboratory experimental processes and results (e.g., pumps used during field applications may be very different than those used in the laboratory setting) and will be addressed in future work.

Scenario 2 considers the same project site and well layout as considered in Scenario 1. However, in this case, the functional unit is the improvement of 9900 m³ of soil such that Vs is increased by 400 m/s. Because the augmented column (Column 3) did not achieve this performance criterion during the laboratory experiment, only the high-ureolytic-rate stimulation (Column 1) and low-ureolytic-rate stimulation (Column 2) approaches were considered as viable treatment options. It was assumed that in order to achieve these improvements, however, injection-to-outlet well distances would need to be modified depending on the treatment technique with the high- and low-rate stimulation approaches requiring injection-to-outlet well spacings of 1.3 m and 3.5 m (Supplementary Figure S1), respectively, which were comparable to the injection distances present at Vs locations 2 and 4 in the experiments. The results, presented in Table 2, illustrate that even when large increases in Vs are desired, the low-ureolytic-rate stimulation approach is more sustainable than the high-ureolytic-rate stimulation approach (by about 19% on average). This is due to the smaller injection well spacing required by the latter approach. In Scenario 2, the use of low-ureolytic-rate stimulation over high-ureolytic-rate stimulation would be anticipated to decrease GHG emissions by about 760,000 kg CO₂ eq.

5.2. Study Limitations and Recommendations for Future Work

The results of this LCSA are specific to the described scope and are not necessarily representative of other applications of MICP, as the engineering performance and predicted impacts of MICP treatment, including process emissions, may vary for different soil types or treatment volumes. Some factors, such as life cycle costs, are expected to vary widely between projects and geographic regions. Other modeling choices, however, may have little influence on the LCSA results. For example, the sensitivity of the results to the reference LCI dataset selected for modeling electricity production is expected to be small, as energy use impacts were minimal relative to materials production impacts for MICP. In fact, when the U.S. average electricity grid mix is substituted with the electricity grid mix for California (i.e., the CAMX subregion), which relies less on burning coal and more on natural gas and renewables (e.g., hydro, wind, and solar power), total global warming potential is reduced by less than 5% for all columns.

Results obtained from this study may help prioritize areas for future research and development efforts in order to improve the sustainability of MICP for geotechnical and geoenvironmental applications [55]. Based on the LCSA results obtained, it is recommended that the following research areas be prioritized in future studies: (1) the optimization/minimization of materials identified as environmental/economic hotspots (i.e., urea, calcium chloride, ammonium chloride, and sodium acetate); (2) the optimization of MICP engineering performance per unit mass of CaCO₃ (i.e., reducing the mass of CaCO₃ needed to achieve a specific engineering performance metric); (3) the reduction in produced nitrogen byproducts near the end of MICP treatments through treatment strategy modifications; (4) a further investigation of the fate and transport of produced nitrogen species resulting from MICP; (5) the evaluation of the relevant impacts and costs that would occur during MICP field-scale implementation, including materials transportation, equipment mobilization and use, construction consumables, effluent management, and labor costs; and (6) a comparative LCSA of MICP against other conventional soil improvement techniques to identify applications where MICP may offer large sustainability gains (e.g., liquefaction mitigation under existing structures where many ground improvement methods may not be effective or would be cost-prohibitive).

6. Conclusions

The results from this study provide valuable insight into the environmental, social, and economic impacts of three different MICP treatment approaches. Through meter-scale soil–column experiments and subsequent analysis, it is shown that the low-ureolytic-rate stimulation treatment approach achieved more uniform biocementation improvement and generally resulted in lower environmental impacts and costs than the high-ureolytic-rate stimulation and augmentation approaches. Additionally, this study highlights tensions between the benefits afforded by NH₄⁺ byproduct removal using rinsing and the environmental impacts and costs associated with these rinsing processes. An optimal rinsing strategy was found to consist of the injection of approximately 1.8 PV of rinse solution, which reduced in situ N masses by between 62% and 95% and subsequently reduced eutrophication potential, global warming potential, smog formation potential, and social cost. The life cycle thinking and quantitative LCSA methods presented herein may help identify environmental and economic hotspots and inform decision making in academia (e.g., during the research and development of emerging technologies) and industry (e.g., during the consideration of alternative technologies or designs) alike to ultimately realize more sustainable practices and solutions for society.