Soil biostabilisation and interaction with compaction processes for earthen engineering structures production

2.1 Materials

2.1.3. Soils

 ⌅

Four commercial soils were studied; their particle size distributions are presented in Figure 1. Small particle sand (SS), coarse sand (CS), clayey sand (M1), and sandy clay (M2) were considered to cover a wide range of realistic soils for stabilisation. In Figure 1, it is observed that approximately 50% of the mass of CS was attributed to particles greater than 1 mm in size. Approximately 50% of the mass of SS and M1 was attributed to particles greater than 0.2 mm in size. Approximately 50% of the mass of M2 was attributed to particles greater than 0.074 mm in size, indicating 50% silt-clay content. All soils were supplied by a local producer, Sorres i graves Egara S.A. Soil was not aseptic, and no treatment was applied before stabilisation.

Figure 1.  Particle size distribution of the used soils. SS Small particle sand. CS Coarse particle sand. M1 clayley sand. M2 Sandy clay.

2.2. Testing methods and specimens

Different specimen definitions and testing methodologies were used for the four distinct studies conducted in this research. Details for each campaign are provided in the following paragraphs.

2.2.1. Study of precipitation kinetics

 ⌅

The aim of this study was to set the optimum conditions for calcium carbonate precipitation in a liquid environment promoted by Sporosarcina pasteurii MICP. The influences of the culture temperature, the calcium source, the calcium concentration, the bacterial concentration, and the combined effect of increasing bacterial and calcium concentrations were studied through six experiments.

For each experiment, 11 Erlenmeyer flasks with the same bacterial suspension volume were introduced into a water bath prepared with a shaking system (see Figure 2); the temperature was maintained constant. The corresponding calcium and urea solutions were added to all flasks to reach the compound concentrations listed in Table 1. The water bath maintained the desired temperature and the flasks were continuously shaken. The first flask was removed immediately after the addition of calcium and urea. The quantity of precipitated calcium carbonate was determined by filtering the liquid through a previously weighed filtering paper, drying it for 24 h, and weighing it again. One flask was removed every hour for the following 10 h for calcium carbonate quantification. In addition, the theoretically consumed urea was calculated by stoichiometry from the measured calcium carbonate precipitation. The precipitated particles were observed under a microscope to determine their morphology.

Figure 2.  Water bath with shaking system for the study of precipitation kinetics (left) and testing configuration for the comparative load-bearing experiments (right).

Table 1.  Sporosarcina pasteurii MICP kinetic experiments.

Experiment	Temperature (ºC)	[Bacteria] (cfu/ml)	[CH4N2O] (mmol/l)	Calcium source	[Ca2+] (mmol/l)
K1	37	5.1e6	333	CaCl2·2H2O	25
K2	25	5.1e6	333	C6H10CaO6	25
K3	25	5.1e6	333	CaCl2·2H2O	25
K4	25	5.1e6	333	CaCl2·2H2O	250
K5	25	5.1e8	333	CaCl2·2H2O	25
K6	25	5.1e8	333	CaCl2·2H2O	250

2.2.2. Study of treatment procedure for non-compacted soils

 ⌅

The kinetics of calcium carbonate precipitation can be altered by a non-aseptic solid soil environment. Eighty experiments were performed to determine the treatment procedure resulting in the highest compressive strength for each soil, without compaction.

These experiments studied the influences of the bacterial presence, the bacterial concentration, performing a single treatment or periodic treatments, the time between periodic additions of urea and calcium, and including a second bacterial inoculation.

The testing procedure (see Figure 3) began by placing the dry soil into a 45 mm internal diameter transparent PVC pipe with a 75 μm sieve sheet covering the bottom end to allow excessive fluid to be drained. The pipes were filled with soil (100 mm in height) and 55 ml of Bang medium with suspended Sporosarcina pasteurii (T3-T10 in Table 2) or 55 ml of distilled water (T1 and T2 in Table 2) was added to SS, M1, and M2 soil specimens. A volume of 45 ml was used for CS specimens. These quantities (55 ml or 45 ml) were experimentally determined to ensure complete soil wetting without liquid flowing out of the sample. The liquid was added to the top of the specimens. Bang medium (4 ml) was added 4 h after the first bacterial inoculation with Sporosarcina pasteurii (T3-T10 in Table 2). Treatments were started 24 h after the first liquid addition; 1.4 ml of Bang medium was added with urea (20 g/l) and calcium (36.75 g/l) every 2 h, five times per day for 5 d, or every 4 h, three times per day for 5 d. The calcium ion source was calcium chloride dihydrate for all 80 experiments. The environmental temperature was 25°C for all tests.

Figure 3.  Procedure for the study of the treatment method and the influence of soil compaction time. (a) Filling the PVC pipe with soil. Sieve sheet in the bottom ending; (b) initial soil wetting; (c) compacting (only for corresponding cases; compaction was set after treatment in some cases); (d) treatment application; (e) unmolding and drying; (f) unconfined uniaxial compressive testing.

Table 2.  Soil stabilisation experiments.

Experiment	[Bacteria] (cfu/ml)	2nd inoculation	Treatment period (h)
T1	-	N	-
T2	-	N	2
T3	1E7	N	-
T4	1E7	N	S
T5	1E7	N	2
T6	1E9	N	-
T7	1E9	N	S
T8	1E9	N	2
T9	1E9	N	4
T10	1E9	Y	2

Some experiments (T3 and T6 in Table 2) were performed with a single treatment addition 24 h after the first liquid addition. One experiment (T10) included a second bacterial inoculation 24 h after the first inoculation; treatment started 24 h later in this case.

The experimental list and the corresponding specifications are presented in Table 2. Each test was repeated twice for each type of soil.

Soil samples were unmoulded two days after finishing treatment and dried in indoor environmental conditions for three weeks before performing compressive tests. The top surfaces of the samples were polished using abrasive paper to obtain a plain surface parallel to the bottom surface to ensure convenient contact with the testing machinery. Specimens were measured (height and diameter) and tested with an unconfined uniaxial compression configuration at a loading rate of 1 mm/min. The load cell measured the applied load at 50 Hz. The compressive strength value was corrected by a geometric factor according to (3434. ASTM C42 / C42M-20, Standard test method for obtaining and testing drilled cores and sawed beams of concrete, ASTM International, West Conshohocken, PA. 2020. https://doi.org/10.1520/C0042_C0042M-20.
). The average value was calculated.

3.1. Kinetics experiments

Charts summarising the evolution of the remaining urea and the precipitated calcium carbonate for the kinetics experiments (Table 1) are presented in Figure 4. The morphology of the particles from the kinetics tests is shown in Figure 5.

Regarding the testing temperature, it was observed that increasing the temperature from 25°C to 37°C (K1 experiment vs. K3 experiment in Figure 4) did not increase the amount of calcium carbonate produced, as reported by other researchers (3939. DeJong, J.T.; Fritzges, M.B.; Nüsslein, K. (2006) Microbially induced cementation to control sand response to undrained shear. J. Geotech. Geoenvironmental Eng. 132 [11], 1381-1392. https://doi.org/10.1061/(ASCE)1090-0241(2006)132:11(1381).
) In fact, urea consumption and calcium carbonate production rates (slope of curves in Figure 4) were higher at lower temperatures (25°C, K3, control experiment, Figure 4). It was only during the last period (from 9 h on) that urea consumption and the calcium carbonate precipitation rate increased in the higher temperature environment. The observed particles were qualitatively larger at higher temperatures (comparing Figure 5 (a) for higher temperature and 5(c) for control temperature), although more small calcium carbonate particles were expected. This contradiction may result from a large number of the smallest particles passing through the filtering paper for the K1 (higher temperature) test; the resulting sample for observation may have missed the smallest calcium carbonate particles. Thus, it cannot be concluded that calcium carbonate production is greater at 25°C than at 37°C, because 37°C is the recommended culture temperature in the literature (3939. DeJong, J.T.; Fritzges, M.B.; Nüsslein, K. (2006) Microbially induced cementation to control sand response to undrained shear. J. Geotech. Geoenvironmental Eng. 132 [11], 1381-1392. https://doi.org/10.1061/(ASCE)1090-0241(2006)132:11(1381).
). The results seem to indicate that calcium carbonate particles produced at higher temperatures may be smaller. Further testing, including the repetition of kinetic tests reducing the potential loss of the smallest particles and acquiring pH data is a must for future developments to confirm this justification for the obtained evidences.

Regarding the influence of the calcium source (comparing K3 calcium chloride dihydrate and K2 calcium lactate experiments in Figure 4), it was observed that using calcium lactate was not as efficient as using calcium chloride dihydrate. Calcium chloride (K3) produced over three times greater weight of particles than calcium lactate (K2). In addition, particles produced with calcium lactate did not exhibit the typical calcium carbonate shape and these were significantly larger (100 μm) than any other particle observed in the kinetics tests. It is concluded that the particles observed in K2 were not calcium carbonate. That may be explained by the lower solubility of calcium lactate (50 g/100 ml at 60°C, see (4040. Kubantseva, N.; Hartel, R.W. (2002) Solubility of calcium lactate in aqueous solution. Food Rev. Int. 18 [2-3], 135-149. https://doi.org/10.1081/FRI-120014355.
)) compared with calcium chloride (134.5 g/100 ml at 60°C). The lower solubility of calcium lactate produced a smaller number of free calcium ions to combine with carbonates, resulting in lower calcium carbonate production when calcium lactate was used as the source of calcium ions. Hence, calcium chloride dihydrate was chosen for all tests after as the source of calcium ions.

Increasing (multiplying by 10) the calcium concentration (K3 vs. K4 in Figure 4) produced an initial reduction in the calcium carbonate precipitation rate; urea hydrolysis decreased accordingly. However, from 6 h after the test start and on the calcium carbonate production continuously increased up to more than four times the reference value (K3 experiment in Figure 4). The shapes of the particles obtained in experiment K4 were qualitatively more aggregated (Figure 5 (d)) than the reference experiment (K3, Figure 5 (c)). This behaviour may be explained by the initial blocking effect of the precipitation of calcium carbonate due to the high calcium chloride concentration and the resulting reduced pH. After 6 h of bacterial hydrolysis of urea, pH overpassed the recommended threshold value (pH > 9) to promote and accelerate calcium carbonate precipitation (K4). The availability of carbonate and calcium ions resulted in a faster production, leading to a qualitatively more aggregated morphology of the calcium carbonate.

Increasing (by 100 times) the concentration of Sporosarcina pasteurii (comparing experiments K3 and K5 in Figure 4) produced a ten-fold increase in calcium carbonate production; urea consumption was similar to that in the reference experiment (K3). An increase in the particle density and a decrease in the particle size are qualitatively observed by comparing Figure 5 (c) and Figure 5 (e). This response may be explained by the presence of more nucleation points (cells) with greater bacterial concentration (K5). Calcium carbonate precipitated around these nucleation points, generating more particles of smaller size than with lower bacterial concentration (K3).

The combination of increasing calcium (by ten times) and bacterial concentrations (by 100 times) (K6 vs. K3 in Figure 4) increased production of calcium carbonate (by 50 times). In addition, the production ratio was time-stable. The produced particles were small in size and highly aggregated, as shown in Figure 5 (f). This was the result of the combination of two independently observed effects: a large number of nucleation points (high concentration of bacteria) producing a large number of particles, and the availability of free calcium ions (high calcium chloride concentration), causing aggregation of particles through faster production.

Summarizing the qualitative analysis of calcium carbonate precipitation kinetics from MICP, increasing the bacterial concentration increases the number of nucleation points; increasing the calcium concentration increases the aggregation of precipitated calcium carbonate.

3.2. Treatment influence

The average compressive strength of the two specimens tested in each experiment defined in Table 2 is summarised in Figure 6; the variation of each result is also included, with the range indicated by arrows. These results are compared pair by pair for each soil type to examine the influence of the different treatment alternatives.

Periodically adding urea and calcium dissolved in Bang medium to soils without adding Sporosarcina pasteurii (T1 vs. T2 in Figure 6) resulted in increased compressive strength for all soils, by a factor of up to five (SS soil; Figure 6 (a)). Calcium carbonate may have precipitated from the direct reaction between urea and calcium chloride, even without the presence of Sporosarcina pasteurii. Hence, bacteria acted as a catalyser of calcium carbonate precipitation reaction.

Bacterial inoculation with no further treatment (T3 vs. T1 in Figure 6) increased the compressive strength of SS soil (up to 6 times) but decreased the compressive strength of CS and M1 soils, proving that the voids left by bacterial sporulation reduced the compressive strength in most of the cases. The results for M2 soils were analogous for the control (T1) and with bacteria only (T3) cases. For SS, with few original ordered voids, spores may contribute to filling the existing space more than generating new voids after bacteria sporulation, resulting in the reported strength increase. Nevertheless, further research including microscopic observation of the treated soils is necessary for confirmation of this justification.

Adding bacteria and performing periodic treatment (T2 (only treatment) vs. T5 (bacteria + treatment) in Figure 6) increased the compressive strength of clayey M1 and M2 soils, proving that bacterial activity was maintained under periodic irrigation conditions in clayey soils, whereas bacteria may have been partially removed from sandy soils (SS and CS) due to their interconnected porosity when irrigated with periodic treatments. This inferred statement is confirmed by analysing the cases with a single irrigation.

Performing a single irrigation process on soils that were previously inoculated with bacteria (T4 and T7 for low and high bacterial concentrations, respectively) reduced the compressive strength compared with only adding bacteria (T3 and T6 for low and high bacterial concentrations, respectively) for all soils (Figure 6). This effect was more evident in sandy soils (SS and CS) due to interconnected porosity. Thus, it is possible to conclude that bacterial removal by irrigation processes has been proven.

Repeating treatment irrigation (T5 and T8, irrigation every 2 h for low and high bacterial concentrations, respectively, vs. T4 and T7, corresponding single irrigation, in Figure 6) increased the compressive strength for all soils, proving that the remaining bacteria after the first irrigation removal required urea and calcium supply to produce calcium carbonate.

Spacing irrigation time up to 4 h (T9 (4 h) vs. T8 (2 h) in Figure 6) decreased the compressive strength of SS and M2 soils, but increased the compressive strength of CS and M1 soils approximately 25%. A slower treatment rhythm was favourable for more porous soils (CS and M1) even with smaller amounts of treatment compounds (urea and calcium chloride). Preventing bacterial removal is the key issue in porous soils. In contrast, soils that retained bacteria (SS and M2) achieved greater compressive strength with increased irrigation repetitions, increasing urea and calcium chloride input and enhancing MICP. Hence, it is only recommendable to increase the treatment rhythm for those cases whose bacterial removal is little like clayey soils and small particle sandy soils.

Increasing (by 100 times) bacterial concentration (T3 vs. T6, T4 vs. T7, and T5 vs. T8 in Figure 6) increased the compressive strength for almost all soils and treatment rhythms, proving that Sporosarcina pasteurii was effective in promoting the MICP process. CS soil compressive strength was increased up to six times with increased bacterial concentration and no treatment (T3 vs. T6), suggesting that repeated irrigation was especially detrimental for coarse soil due to bacterial removal. Two exceptions were observed: M2 soil with single treatment showed no clear difference (T4 vs. T7), and M1 soil with 2 h periodic treatment exhibited a compressive strength reduction (of approximately 1/3) with increased bacterial concentration (T5 vs. T8 in Figure 6). A hypothesis to explain this result is that increasing the volume of the MICP process with a greater bacterial concentration and extending it over time owing to the greater volume of urea and calcium (2 h irrigation treatment) may have promoted calcium carbonate formation even on dried M1 soil specimens. This late MICP process may have broken typical clay bonds, which were thought to be weaker in the M1 soil than in the M2 soil, which had a higher clay content. This hypothesis requires additional microscopic observation to be confirmed but it points out the idea that there is a counterproductive interaction between clay drying process and MICP that requires extensive research.

Adding a second bacterial inoculation (T10 vs. T8 in Figure 6) increased the compressive strength in all cases. The effect was more significant for SS and M1 soils, which doubled their respective compressive strengths. A second bacterial inoculation increased the bacteria amount and also allowed the first bacterial inoculation to rest an additional 24 h before starting irrigation treatments. The bacteria had more time to set, and less bacterial removal caused by irrigation was likely. From a practical point of view, it would be of major interest to research on the possibility of spacing the starting time of the treatment respect from the initial bacterial inoculation, especially for those soils which showed greater bacterial removal due to treatment liquid flux.

3.3. Influence of compaction time

The moisture content at compaction time of the samples compacted after treatment, the optimum moisture content, and the difference between these two values are presented in Table 5 to better understand the compressive strength results in Table 6. The results of the study on the influence of applying compaction before or after periodic treatments and the study on the influence of replacing the Bang medium with distilled water are presented in terms of the average value of three tests of the compressive strength for each combination (Table 6) of the tests defined in Table 3 and Table 4.These results showed that sandy soils reached greater compressive strength (13% and 87% for SS and CS, respectively) when compacted before urea and calcium chloride treatment irrigation. In contrast, clayey soils reached greater compressive strength (19% for M2, no significant change for M1, which was compacted with insufficient moisture content after treatment, see Table 5) when compacted after urea and calcium chloride treatment irrigation. The expected size of the voids is consistent with these results. The compaction process closed the larger voids in sandy soils, reducing bacterial removal due to irrigation flow. Thus, compacting before irrigation resulted in higher compressive strength for SS and CS soils because more bacteria remained in the soil to accelerate MICP. In contrast, clayey soils had few and poorly interconnected voids. The plugging effect reported in (1414. Cheng, L.; Cord-Ruwisch, R. (2014) Upscaling effects of soil improvement by microbially induced calcite precipitation by surface percolation. Geomicrobiol. J. 31 [5], 396-406. https://doi.org/10.1080/01490451.2013.836579.
) was the main issue for these soils, rather than bacterial removal. Plugging the top surface voids prevented treatment irrigation from reaching all samples’ depth, resulting in a heterogeneous specimen whose compressive strength was limited by the weaker lower part of the specimen. This was confirmed by the failure mode observed in the unconfined compressive strength tests. Greater compressive strength was reached when clayey soil voids were kept open until the MICP process was completed. MICP generated small particles that filled the original voids when compacted after treatment. Thus, it is important to compact sandy soils before treatment irrigation and do the opposite for clayey soil to get the best performance of the combined compaction-MICP process.

3.4. Replacing Bang medium with distilled water

Replacing the Bang medium with distilled water decreased the compressive strength in almost all cases (CBB and CAB vs. CBD and CAD in Table 6) because the nutrients in the Bang medium were required for bacterial deployment, and especially for bacterial growth. Two exceptions were found: CS soil compacted before treatment, which showed no clear influence (see CBB and CBD tests for CS soil in Table 6) and M1 soil compacted after treatment (see CAB and CAD tests for M1 soil in Table 6). This exception was due to the anomalous low compressive strength of M1_CAB specimens resulting from an insufficient moisture content at compaction (24.4% less than optimum, Table 5). Thus, the moisture content during the compaction process was more influential than the Bang vs. distilled water selection in the MICP process. This fact proves that a suitable compaction process brought greater mechanical strength results than MICP treatment.

3.5. Load bearing capacity and compressive stiffness

The stress-deflection plots of the load-bearing experiments are presented in Figure 7. The real moisture content at compaction for the M2_ Biostabilised sample was 14.94% (optimum was 16.42%). The moisture content in the load-bearing experiment was not the same for all samples. The moisture content was 3.80%, 4.28%, 6.67%, and 0.87% for SS_Control, SS_ Biostabilised, M2_Control, and M2_ Biostabilised, respectively. The ballast coefficient was obtained from the previous stress-deformation curves at different deflection ranges. The results are summarised in Table 7.

It was observed that the biostabilised soil specimens were initially stiffer (averaging a 45% increase) than the control specimens, but the mechanical response was the opposite for deflections over 5 mm (see the slope of the curves in Figure 7 and the data in Table 7), proving that the top surfaces of biostabilised specimens were stiffer than the rest of the soil. Irrigation treatments were applied to the top surfaces, which were also directly exposed to the air. It was concluded that the MICP process preferentially occurred at the surface of the specimens because of (i) the plugging effect, which blocked calcium ions from deep penetration into the specimen, especially in M2 soil, and (ii) greater oxygen availability in the top surface, which increased bacterial activity, especially in SS soil. Although oxygen is not strictly required, Sporosarcina pasteurii activity increases in aerobic environments (3939. DeJong, J.T.; Fritzges, M.B.; Nüsslein, K. (2006) Microbially induced cementation to control sand response to undrained shear. J. Geotech. Geoenvironmental Eng. 132 [11], 1381-1392. https://doi.org/10.1061/(ASCE)1090-0241(2006)132:11(1381).
)treatment uniformity, cost, environmental impact, site accessibility requirements, etc. All of these methods have benefits and drawbacks, and there continues to be a need to explore new possibilities of soil improvement, particularly as suitable land for development becomes more scarce. This paper presents the results of a study in which natural microbial biological processes were used to engineer a cemented soil matrix within initially loose, collapsible sand. Microbially induced calcite precipitation (MICP. This fact reported in literature was confirmed from a practical point of view.

The final compressive strength of the control and biostabilised specimens was the same because the MICP process was locally developed at the surface, which did not change the overall response of the tested blocks; this conclusion must be verified through future diffusion experiments. The compressive strength of the M2 control specimen was abnormally lower than that of the biostabilised specimen because of its higher moisture content at the time of testing (6.67% for control M2 vs. 0.87% for biostabilised M2). Thus, the observed increase in the compressive strength of the biostabilised M2 specimen (Figure 5) was not realistic, and this data was discarded from the analysis.

From a practical point of view, load-bearing tests showed that MICP is not effective at enhancing mechanical properties of soils if only superficial treatment was used. Hence, studying the possibility of in-depth injections is required in future researches.

3.6. Optimal solution per soil

The best general stabilisation procedure was to supply the maximum concentration of Sporosarcina pasteurii (including a second inoculation 24 h after the first one) and the maximum concentration of calcium chloride, using the Bang medium for solving urea and calcium chloride during treatment irrigation. There were minor procedural differences depending on the soil to be stabilised: (i) CS did not benefit from a second bacterial inoculation because of its larger pore size; (ii) SS and M2 soils benefitted from greater irrigation frequency (2 h); (iii) sandy soils should be compacted before treatment and clayey soils after treatment. The optimal stabilisation procedure for each soil is summarised in Table 8.

The experimental results (Table 6) demonstrated that Sporosarcina pasteurii MICP did not increase the compressive strength of compacted CS and M1 soils. CS soils were sensitive to bacterial removal during irrigation cycles. Hence, supplying all required resources at the initial soil mixing should be studied in future development of CS soil stabilisation, setting the research line for this specific type of soil. In contrast, the M1 soil particle size distribution was adequate for optimum compaction results. In this case, the benefits of MICP did not compensate for its potential issues: void formation at bacteria sporulation, fine particle removal during irrigation treatments, and breaking of clay bonds due to late calcium carbonate precipitation. In conclusion, MICP is beneficial for those cases which cannot be properly compacted because of a non-suitable particle size distribution or excessive moisture content that cannot be fixed.

3.7. Practical implications of the obtained results

The compressive strength results are not complete enough to design stabilisation solutions for practical applications in a generic way; further research is necessary to determine the deformability in drying-wetting cycles, to analyse the water resistance and freeze-thaw resistance, and to study the cohesive-frictional response of biostabilised soils. Only a hypothesis on how these properties may be affected by the proposed MICP can be considered from the results of the current research. Considering that bacterial sporulation is likely to produce a significant void volume, it is expected that the structure may be more stable in drying-wetting cycle deformability and freezing/thawing conditions. The additional porosity may lead to minor water resistance. All these hypothesis are worth for future research.

Obtaining aggregated calcium carbonate requires high bacterial and calcium concentrations, which are not likely to be provided in field applications. Thus, soils biostabilised with the proposed methodology are not expected to exhibit increased cohesive behaviour, but rather a frictional response from the production of small disaggregated particles due to lower bacteria and calcium concentrations. As a practical conclusion, MICP in civil engineering and architectural applications has not to be aimed to increase the cohesion of the soils.

The time required for the complete biostabilisation process (up to three weeks in some cases) is similar to the full hydration time of cement. However, cement may be sufficiently effective in 2-4 days, whereas the minimum required MICP treatment time is 10-14 days, according to the current results. MICP is not time saving respect common binders stabilisation.

The average cost of the performed treatments is approximately 50 times higher than the cost of Portland cement alternatives. Thus, the technology is not economically competitive at the current state of development compared to other available alternatives, and may only be justifiable if common stabilisers are forbidden due to sustainability reasons.